University of Southern California Dissertations and Theses

TREM2 and C1q signaling regulates immunoproteostasis in Alzheimer's disease

(USC Thesis Other)

TREM2 and C1q signaling regulates immunoproteostasis in Alzheimer's disease

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

TREM2 and C1q signaling regulates immunoproteostasis
in Alzheimer's disease.

by

Brian Pak Yan Leung

A dissertation presented to the

Faculty of the University of Southern California Graduate School

In Partial Fulfillment of the Requirements for the Degree

DOCTOR OF PHILOSOPHY

in

NEUROSCIENCE

August 2018

Advisory Committee:
Terrence C. Town Ph.D.
Caleb Finch Ph.D.
Pat Levitt Ph.D.
Karen T Chang Ph.D.
John T. Tower Ph.D.

2

© 2018
Brian Pak Yan Leung
ALL RIGHTS RESERVED
3

“If I have seen further it is by standing on the shoulders of Giants…”

Sir Isaac Newton
February 5, 1675

4

This dissertation is dedicated to

both my grandparents, my mother, and father.

My grandparents made that initial decision to

escape pre-war-torn China and the pre-cultural revolution turmoil

to start anew in Hong Kong.

Above all, this dedication is for my mother and father,

who immigrated from Hong Kong to the United States,

to chase their American Dream,

in hopes to provide me with every opportunity

to chase mine.

5
Abstract

Alzheimer’s disease (AD) now afflicts more than 5 million Americans, and an effective
treatment or cure does not exist. Risk for late onset AD (LOAD) is likely determined by
complex interplay between environmental and genetic factors. Multiple genome-wide
association studies and gene network analyses have implicated two major innate immune
pathways as key risk factors for late-onset Alzheimer’s disease: Triggering Receptor
Expressed on Myeloid cells 2 (TREM2) and protein complement component C1q. Further,
a more recent GWAS specifically identified peripheral macrophage genes as AD risk
factors. We show that both TREM2 and C1q proteins are dysregulated in AD mouse
brains and in LOAD brains vs. non-demented controls. While classically thought to
regulate different types of immune responses, our data raise the tantalizing possibility
that the TREM2 and C1q pathways intersect in AD. Specifically, we have made the novel
observation that C1q, TREM2 and Aβ physically interact, forming a heteromeric complex
on innate immune cells. We further show that small Aβ assemblies preferentially bind to
TREM2, while C1q more avidly associates with Aβ aggregates. In the APP/PS1 mouse
model, compound genetic loss of TREM2 and C1q abrogates extracellular regulated
kinase (ERK)1/2 innate immune signaling. Peripheral mononuclear phagocyte
experiments demonstrate that phagocytosis of C1q opsonized Aβ is both Trem2 and
pERK1/2 dependent, whereas phagocytosis of Aβ alone requires Trem2 and p-p38
mitogen activated protein kinase (MAPK). For the first time, these results shed light on
the biological significance of the TREM2-C1q-Aβ neuroimmune complex in transition from
the healthy CNS to the AD brain, and have key implications for myeloid Aβ phagocytosis,
clearance, and immunoproteostasis.
6
Forward

Throughout my PhD, I learned more than just neuroscience or immunology. I learned
something valuable my cultural roots and identity. Living abroad was the catalyst that
forced me to understand my cultural upbringing. Being first generation as an American
immigrant, I struggled with my cultural identity because, often times, my Hong Kong
identity wasn’t culturally accepted. Take this example about Confucian Philosophy
regarding social hierarchy. In Western culture, people with more senior titles are above
me, but often times, to seem more personable, the relationship does not feel stratified. In
Eastern culture, Confucianism and hierarchy is paramount and elders and status must be
respected and there cannot be a breach of formality. Understanding who abides by which
principle or even some combinatorial pattern of the two is difficult to assess. When I was
in Europe, these themes were brought out and I had more difficulty than usual when
navigating these social norms.

Throughout this dissertation, every chapter will have a quote that will simultaneously
acknowledge my cultural background and to the chapter as a whole. When I first heard
the lines by Sun Tzu in The Art of War, each one these quotes immediately resonated
with me because they felt very similar to my academic tensions navigating the social
norms during my PhD. None of these are meant to be offensive to any group/person.
These are merely how I associate these quotes to a feeling within the chapter. Oddly
enough, manipulating the immune system in Alzheimer’s disease is like playing with a
molecular army, after all I am just studying the Art of War.

7
Acknowledgements

"Omwana ni wa bhone,"

“Regardless of a child's biological parent(s), its upbringing belongs to the community.”

-Kijitan (Wajita) proverb

Throughout this entire journey, I have to thank my two mentors, Dr. Terrence Town and
Dr. Burkhard Becher. While have given me the freedom to learn, study, and grow as a
PhD student, Terrence has set the bar very high, ensuring that I reach that point as a
scientist. Although Terrence’s tactics are unconventional, he has made me a better
scientist, team-player, and advocate. Without his guidance, I wouldn’t have been able to
receive the National Science Foundation, Graduate Research Fellowship Program award
and the other accolades that are come with this fellowship. Being trained in his lab has
given me the confidence to stand on my own in the scientific community. I cannot stress
how thankful I am for the freedom that has come from being in his lab, not just project
freedom, but also intellectual freedom that lead me to Switzerland for over a year. For my
time in Zurich, Switzerland, I truly have to thank Dr. Burkhard Becher for welcoming me
into his lab at the University of Zurich as well teaching me the fundamentals of classical
immunology that I could not receive at USC. Additionally, the amount of time Burkhard
spent training the lab on how to use R and code is a skillset I am truly thankful for, because
it has given me a slight advantage in science and the work I have done in this dissertation.

I have to thank each and every one of my committee members who has been with me
from the beginning of my PhD. I truly appreciated each suggestion and comment to make
8
my dissertation stronger, even if they were difficult to hear. I first want to thank my
committee chair, Dr. Caleb Finch, who has provided me with great suggestions and
feedback through my dissertation work, and as a faculty member who I could turn to when
times to difficult. I especially have to thank Dr. Karen Chang who has provided not only
with advice and scientific input, but also the financial/resource/mental support to carry out
many of my experiments in this dissertation. These not only include fly lines, but also
molecular biology and biochemistry reagents for my experiments. As an Asian American
in academia, watching Dr. Chang ascend from an assistant professor to tenured,
associate professor is not only inspirational, but also comforting. During my PhD, I
watched Dr. Chang lead by example, and it has given me the courage and strength to
push through as a PhD student. Additionally, I’d like to thank Dr. John Tower, who is not
only provided resources for my dissertation, but has he been one few faculty members I
could turn to during my most difficult times. Most importantly, as my last committee
member, Dr. Pat Levitt, who has not only provided the platform and foundation to develop
professionally as a scientist, but also provided advice and support during my most difficult
times during my PhD. Outside of my dissertation, Pat has taught me how to become an
advocate, manage a team, and a lead. There were many difficult moments during my
PhD that were not scientific in nature but has happened at USC. During these times I
learned by watching his lead and other times I learned directly from him. In summary, to
all my committee members, mentors and neuroscience faculty, thank you for molding me
into the scientist that I am today.

9
In addition to the faculty who supported me throughout this PhD, I have to thank the
people in these labs, especially the Town and Becher lab. To all the post docs in the
Town lab, Thank you. To Dr. Marie-Victoire Gulliot-Sestier, thank you for proving the
emotional support throughout my PhD and for training me to become a better scientist
and independent thinker. To Dr. Kevin Doty, thank you for taking me under your wing the
first few years of my PhD and for being the support system throughout this journey. To
Dr. Tara Weitz, thank you for being my sense of reason and providing the emotional
support when times were getting tough. To Drs. Rodrigo Lopez Leal and Juan Carlos
Biancottti for teaching me many useful scientific skills early on in my PhD. To Dr. David
Gate, thanks for the advice, support, and for setting the expectations so stupidly high. I
had a great time working alongside you during your time as a student at USC.

To the current graduate students in the lab, thank you so much for providing one of my
main support networks during this PhD. Chris Im, I am truly thankful for the discussions,
advice, and conversations throughout this PhD. Whether it was over dinner or late into
the night until the early morning hours in the lab, Thank you. Thank you, Alicia Quihuis,
for being one of my emotional pillars in the lab. Working alongside you has helped me
grow scientifically and personally. Mariana Uchoa, thank you so much for all the scientific
discussions throughout my entire PhD because you’ve forced me to think outside of the
box and pushed to become a better scientist. Muito obrigado! Rachel Oseas, I am grateful
that you joined our lab and immediately became part of our support system. Thank you,
Javier Rodriguez Jr., for keeping the lab together and for being part of my support system
in the lab. Last but not least, to the two laboratory technicians and undergraduate students
10
in our lab, Cole Miller and Alex Vesling, for neither one I could do without during my PhD.
The two of you have been the two people who I’ve lean upon so much throughout my
PhD, I don’t know what I would have done without the two of you there. While I can say
that some people helped in some aspects of my dissertation, the two of you have been
involved in almost every data chapter in this dissertation. I am incredibly grateful that you
two decided to join the Town lab and for dealing with me the first few years as I was
struggling through my PhD.

To the members of the Becher lab, thank you so much for brining including me into the
Becher Lab family. All of you have made the cultural immersion a wonderful experience.
From all the jokes, laughs, and intellectual discussions, thank you. I am truly grateful. For
all the scientific support and social support in the lab, thank Melanie Greter, Vinko
Tosevski, Denis Shapiro, Dr. Felix Hartman, Dr. Isabel Ohs, Natalie Karkoski, Dilay
Cansever, Tom Hartwig, Bithi Chatterjee, Dr. Edoardo Galli, Dr. Ekaterina Terskikh,
Juliana Komuczi, Dr. Caterina Raposo, Laura Ducimetiere, Jennier Jaberg, Dr. Sabine
Spath, Dr. Margit Lanxinger, Dr. Xueyang Yu, Dr. Andrew Croxford, Dr. Sonia Tuges, Dr.
Anne Buttgereit, Nikola Misljencevic, Pascale Eede, and Sandra Konieczka for being an
part of my scientific development and being that support network at the Univeristy of
Zurich. Additionally, I especially want to thank Dr. Paulina Kulig for reaching out that extra
hand to teach me immunology and for being part of my support system at UZH. I next
want to thank Dr. Kathrin Nussbaum and Sebastian Utz for being my running buddies and
for integrating me into the lab. Thank you Dunja Mjrden for having me under your wing
the first few months in Zurich; thank you for teaching me the basics of autoimmunity and
11
for being part of my support system in Zurich. Thank you to Dr. Raul Catena and Dr.
Florian Mair for teaching me the basics of CyTOF, immunology, and for being part of my
support system at UZH. Thank you, Melissa Vrohlings, for teaching me a lot about
neuroimmune cancers and for being such an integral part of my Zuirch experience. Last
but not least, Ana Amorim and Claudia Haftman. You two have taught me so much in the
past few years about myself by helping me grow scientifically and personally. From all the
discussions, laughs, cultural-political conversations, you’ve made me feel like Zurich was
my home. I cannot thank you enough. Vielen dank für alles!

My PhD experience wouldn’t have been what it is today without the help of many
organizations who’ve supported me throughout my training, scientifically and
extracurricular. I can’t thank the staff at the Society for Neuroscience, Alzheimer’s
Association, Alzheimer’s Greater Los Angeles staff, and staff at the Graduate School. I
especially want to thank Barbra McLendon, Ruth Gay, and Kelly Honda for giving me the
teaching me and giving me the opportunity to advocate on the national stage. You’ve
taught me more than what primary literature can teach me, the social side and dig into
my emotional side Alzheimer’s disease. From these experiences, I have become a better
advocate and scientist. I especially want to thank Dr. Meredith Drake Reitan from the
graduate school, for without I don’t think I could have achieved the accomplishments that
I currently have without her support, tirelessly helping the next generation of scientists
succeed. You have become my mentor outside of my field, lending a hand into the policy
realm. Special thanks to Kate Tegmeyer for managing all my fellowships at USC and for
being my support throughout my PhD.
12

None of these experiences at USC would have been possible without my funding agency
am truly grateful, beyond grateful to be a National Science Foundation, Graduate
Research Fellowship Program. The opportunities provided for me have given me so much
academic freedom and privilege. With this privilege, it has always reminded me to give
back to the community and help underrepresented minorities and students, K-12, to go
towards the sciences. Often times, this fellowship has given me the academic credibility
that I wouldn’t have received based off of initial preconceived academic prejudices.
Because of these unfortunate biases, I have done my very best to break these prejudices,
one interaction at a time.

In addition to this NSF, I want to thank our collaborators, Dr. Andrea Tenner and Dr.
Marco Colonna for being so incredibly helpful whenever I needed help during my PhD.
Whether it was in the US or internationally, these two always reached out to me, and lent
a hand. They were unafraid to ask the hardest questions and for that, I am truly grateful.
I am also thankful for the USC Flow Cytometry core, Jeff Boyd and Bernadette Masinsin
for assisting in many of my experiments and being that emotional support when
experiments to awry. To the biochemistry core Dr. Shuing Li for teaching me more about
spectroscopy and for providing the support whenever I needed it during all my
biochemistry experiments. The support staff in the neuroscience graduate program,
Mallory Redel, Vanessa Clark, Deanna Solorzano, Dawn Burke, and Gloria Wan.

13
My graduate career would not have been possible if it wasn’t for two very important people
in my life, Dr. Aleister Saunders and Dr. Daniel Marenda. Both of whom were instrumental
in getting me where I am today. I didn’t see myself as a scientist until Junior/Senior year
of college. I had the support of the Saunders lab who inspired me to enter neuroscience.
From by beginnings as a chemistry major, I never thought I’d end up in neuroscience or
immunology. I remember my first neuroscience course taught by Dr. Daniel Marenda, and
at that point, I knew, I wanted to study neuroscience because it was way too cool. Dr.
Aleister Saunders helped carry that momentum forward, but without the two of them
inspiring me, I wouldn’t be where I am.

Lastly, to the entire family, both immediate and distance relatives. Thank you so much for
believing in me and giving me motivation to keep pressing on. To my brother Benson
Leung who has been with me from the beginning and who has always been that emotional
support. Thank you. To all my friends, not just at USC or in the Neuroscience Graduate
Program, but all around the world. I really could not have done it without your love and
support at every step of the way. Special thanks to Dr. Ashley Landicho, Dr. Max
Henderson, Bryant Wu, Chloe Fink, Michael Sin, Victoria Xie, Patrick Wong, Dr. Sara
Ansaloni, Dr. Colleen Hoffman, Dr. Maliha Ahmed, Dr. Sherif Madkour, Dr Nicholas
Goeden, Dr. Panthea Heydari, Angel Martinez, Dr. Gabriella Garcia, D.J. Kast, Jennifer
Leung, Dr. Anna Vorbyeva, Dr. Sean Miller, Alexander Nelson, Grayson Stamps, Candice
Nguyen, Julie Guerin, Krystal Lin, Dr. Jillian Shaw, Dr. Joo Yeun Lee, Dr. Hanke Heun-
Johnson, Melanie Sweeney, Kirsten Lynch, Laura Badertscher, Dr. Candace Lewis,
Adonis Bou Chakra, Aixin Deng, Malte Stiel, Adam Lundquist, Anakha Ajayan, Adam
14
Marentes, Jennifer Tribble, Liana Engie, AJ Cooper, Alexis Gorin, Samson King, Dr. Alex
Moser, Brandy Riedel, Brian Zingg, Rachel Yuan, Dr. Ted Hsu, Dr. Radhika Palkar, and
Dr. Chiara Mazzasette.

15
Table of Contents for Figures

Chapter 1

Figure 1.1: Evolution of the innate immune and adaptive immune system 40

Figure 1.2: Evolution of the Blood Brain Barrier 41

Figure 1.3: Embryonic origins of innate immune cells 42

Figure 1.4: Comparing Cognitively normal and AD brains 43

Figure 1.5: Innate immune response in a balance 44

Figure 1.6: Trem2 and C1q are transcriptomically linked 45

Figure 1.7: Structure of TREM2 and DAP12: 46

Figure 1.8: TREM2 stimulates phagocytic pathways without inflammation 47

Figure 1.9: Evolution of Complement signaling 48

Figure 1.10: C1q activation initiates the Classical Complement cascade 49

Figure 1.11: Structure of C1q 50

Figure 1.12: The Central premise of this dissertation 51

Chapter 2

Figure 2.1: Homologous Signaling pathways Toll, IMD and Upd 77

Figure 2.2. Paralleled olfactory circuitry 79

Figure 2.3. D. melanogaster cytokines could polarize glia to perform
different functions
80

Figure 2.4: Survival curve of Aβ expressing flies 81

Figure 2.5: Relish activation occurs in Neuronally expressing Aβ and
AβArctic Flies
82

Figure 2.6: Neuronal Aβ and AβArtic Flies upregulate Toll pathways genes 83

Figure 2.7: Phagocytic machinery is upregulated in Neuronal AβArctic flies 84

Figure 2.8: D. melanogaster hemocytes are able to recognize and engulf Aβ 85

Figure 2.9: C3-like Complement genes are increased in AβArctic flies 86

Chapter 3

Figure 3.1: Immunoproteostasis vs proteostasis 97

Figure 3.2: Mouse breeding paradigm: 98

Figure 3.3: Trem2 and C1q deficiency impacts fibrillar Aβ plaque formation 99

16
Figure 3.4: Trem2 and C1q deficiency impacts fibrillar Aβ plaque formation 100

Figure 3.5: Trem2 deficiency increases soluble Aβ 38 formation 101

Figure 3.6: Trem2 deficiency increases soluble Aβ 40 formation 102

Figure 3.6: Trem2 deficiency increases soluble Aβ 42 formation 103

Figure 3.8: C1q increases oligomeric Aβ 104

Figure 3.9: Trem2 and C1q deficiency differentially affects Aβ 105

Figure 3.10: Trem2 and C1q deficiency independently affects Aβ
proteostasis
106

Figure 3.10: Trem2 and C1q affect Aβ proteostasis 107

Chapter 4

Figure 4. 1: Increased GFAP and Iba 1 expression in 6 months APP/PS1 124

Figure 4. 2: Reduced GFAP and increased Iba 1 expression in APP/PS1
Trem2 and/or C1q deficient mice
125

Figure 4. 3: Increased Trem2 and C1q gene expression in APP/PS1 mouse
brains
126

Figure 4. 4: APP/PS1 mice deficient in Trem2 and/or C1q alter Trem2 and
C1a protein expression
127

Figure 4. 5: APP/PS1 mice deficient in Trem2 and/or C1q may not alter
Cleaved Caspase 3 levels
128

Figure 4. 6: APP/PS1 mice deficient in Trem2 and/or C1q increase
Synaptophysin levels
129

Figure 4. 7: APP/PS1 mice deficient in Trem2 and/or C1q may impact NF-
κB expression
130

Figure 4. 8: APP/PS1 mice deficient in Trem2 and/or C1q may modulate
GSK-3 activity
131

Figure 4. 9: APP/PS1 mice deficient in Trem2 and/or C1q mostly reduces
pAKT activity
132

Figure 4. 10: APP/PS1 mice deficient in Trem2 and/or C1q may modestly
affects p38 signaling
133

Figure 4. 11: APP/PS1 mice deficient in Trem2 and/or C1q mostly reduces
pERK activity
134

Figure 4. 12: APP/PS1 mice deficient in Trem2 and/or C1q mostly reduces
C3
135

17
Figure 4. 13: APP/PS1 mice deficient in Trem2 and/or C1q show some
Anxiety like behavior impairment
136

Figure 4. 14: APP/PS1 mice deficient in Trem2 and/or C1q show some
hyperactivity (distance)
137

Figure 4. 15: APP/PS1 mice deficient in Trem2 and/or C1q show some
hyperactivity (velocity)
138

Figure 4. 16: APP/PS1 mice deficient in Trem2 and/or C1q show no
potential aggressive-like behavior
139

Figure 4. 17: APP/PS1 mice deficient in Trem2 and/or C1q show no
impairments in working memory
140

Figure 4. 18: APP/PS1 mice deficient in Trem2 and/or C1q show no
differences in during NOR training
141

Figure 4. 19: APP/PS1 mice deficient in Trem2 and/or C1q show no
differences in during NOR recall
142

Figure 4. 20: TREM2 and C1q protein expression LOAD brains 143

Figure 4. 21: TREM2 and C1q protein expression LOAD brains 144

Figure 4. 22: Adult mouse brains have even C1q expression throughout the
brain, whereas shows Trem2 expression
145

Chapter 5

Figure 5.1: Microglial origins 158

Figure 5.2: Pre-programmed Migration vs Microenvironmental driven
maturation
159

Figure 5.3: Transcriptional changes depend on microenvironmental changes
macrophages
160

Figure 5.4: Human LOAD brains suggest microglial heterogeneity 161

Figure 5.5: Workflow of the following experiment 162

Figure 5.6: NP-C are macrophage specific and enter the brain 163

Figure 5.7: peripheral macrophages express microglia and contain NP-C
+

vesicles
164

Figure 5.8: NP
+
leukocytes are brain resident macrophages 165

Figure 5.9: NP
+
brain resident macrophages express Trem2 and C1q 166

Figure 5.10: APP/PS1 animals Leukocyte numbers do not change among
APP/PS1 mice deficient in Trem2 or C1q
167

18
Figure 5.11: Peripheral macrophages can engage in Trem2 mediated C1q
-
opsonized Aβ phagocytosis
168

Figure 5.12: Trem2 deficiency impairs Trem2-C1q-Aβ complex formation 169

Figure 5.13: TgF344 AD Rats injected with NP show NP
+
, Iba1
+
, Trem2
+

macrophages
170

Figure 5.14: Summary of microglial provenance 171

Chapter 6

Figure 6.1: Phagocytosis and Immunoproteostasis 186

Figure 6.2: Phagocytosis paradigm workflow 187

Figure 6.3: Peripheral macrophage gating strategy 188

Figure 6.4: Trem2 deficiency reduces pERK activity 189

Figure 6.5: Peripheral macrophage phagocytic capacity fluctuates 190

Figure 6.6: pERK inhibition dose curves 191

Figure 6.7: C1q opsonized Aβ phagolysosomal degradation depends on
pERK
192

Figure 6.8: pERK inhibition impairs C1q opsonized Aβ but not Aβ in
peripheral macrophages
193

Figure 6.9: pERK inhibition impairs C1q opsonized Aβ but not Aβ in
microglia
194

Figure 6.10: Inhibition drug control experiments 195

Figure 6.11: Images inhibiting the Syk-MAPK axis in peripheral
macrophages
196

Figure 6.12: C1q-Aβ quantification of inhibitory phagocytic experiments on
the Syk-MAPK axis in peripheral macrophages
197

Figure 6.13: Aβ quantification of inhibitory phagocytic experiments on the
Syk-MAPK axis in peripheral macrophages
198

Figure 6.14: Quantification of inhibitory phagocytic experiments on the Syk-
MAPK axis in microglia
199

Figure 6.15: Inhibitory phagocytic experiments on the Syk-AKT-MAPK axis
in peripheral macrophages and microglia
200

Figure 6.16: TGF-β inhibition rescues the Trem2 phagocytic impairment 201

Figure 6.17: Trem2-C1q-Aβ signaling axis working hypothesis 202

19
Chapter 7
Figure 7.1: Trained immunity vs immune tolerance 222

Figure 7.2: Human microglia express more TREM2 in response to Aβ in
more acute settings
223

Figure 7.3: Human microglia signal through the TREM2-pSYK axis in more
acute settings
224

Figure 7.4: Paradigm for recreating immune tolerance in AD 225

Figure 7.5: Chronic Aβ Tolerance inducing paradigm reduced TREM2
protein expression in human microglia
226

Figure 7.6: Chronic Aβ tolerance-inducing paradigm impacts TREM2
expression
227

Figure 7.7: Chronic Aβ tolerance-inducing paradigm impacts cytokine
transcript expression
228

Figure 7.8: Chronic Aβ tolerance-inducing paradigm impacts C1q transcript
expression
229

Figure 7.9: Under chronic Aβ tolerance-inducing paradigm, macrophages
respond with an increase in Trem2 expression
230

Figure 7.10: Chronic Aβ tolerance-inducing paradigm impacts Trem2
expression in macrophages
231

Figure 7.11: Chronic Aβ tolerance-inducing paradigm impacts Cytokine
transcript expression in macrophages
232

Figure 7.12: Trem2-dependent macrophage trained immunity vs immune
tolerance
233

Figure 7.13: Chronic Aβ tolerance-inducing paradigm impacts IL-10
transcript expression in macrophages
234

Figure 7.14: Chronic Aβ tolerance-inducing paradigm impacts TGF-β
transcript expression in macrophages
235

Figure 7.15: Chronic Aβ tolerance-inducing paradigm may promote a TGF-β
feedforward mechanism
237

Figure 7.16: Chronic Aβ tolerance-inducing paradigm impacts immune
functions in macrophages
238

Figure 7.17: Incorporating a C1q-Aβ into the training-inducing paradigm 239

Figure 7.18: Chronic Aβ training-inducing paradigm impacts Trem2 surface
expression
240

20
Figure 7.19: Chronic C1q-Aβ tolerance-inducing paradigm impacts Trem2
low

populations
241

Figure 7.20: Chronic C1q-Aβ treatment may sustains pSyk Signaling in
Trem2
low
populations
242

Figure 7.21: pERK Signaling impacts C1q
-/-
macrophages but does not
impact macrophage training
243

Figure 7.22: Chronic C1q-Aβ treatment may sustains p38 Signaling in
Trem2
low
populations
244

Figure 7.23: Chronic C1q-Aβ treatment impacts the pSyk-p38 signaling axis
in Trem2
-/-
macrophages
245

Figure 7.24: TREM2 posttranslational regulation 246

Figure 7.25: Trem2-dependent macrophage trained immunity vs immune
tolerance
247

Chapter 8

Figure 8.1: Working hypothesis for TREM2-C1q-Aβ interaction 264

Figure 8.2: Size Exclusion Chromatography-Multi Angle Light Scattering
Spectrometry
265

Figure 8.3: Circular Dichroism Spectrometry 266

Figure 8.4: TREM2 interacts with Aβ 267

Figure 8.5: TREM2 interacts with higher concentrations of Aβ 268

Figure 8.6: C1q interreacts with Aβ 269

Figure 8.7: Trem2 and C1q do not independently interact 270

Figure 8.8: Trem2 binds to C1q opsonized Aβ 271

Figure 8.9: Trem2-C1q-Aβ formation is at an equilibrium 272

Figure 8.10: Trem2-C1q-Aβ complex is heavier than C1q-Aβ 273

Figure 8.11: Trem2-C1q-Aβ formation is Aβ species specific 274

Figure 8.12: Aβ can be Aggregated at 37ºC for 24 hours 275

Figure 8.13: Trem2-C1q-Aβ formation is Aβ species specific 276

Figure 8.14: Assessing the formation of TREM2-Aβ agg 277

Figure 8.15: TREM2-Aβ agg may be oligomeric Aβ 278

Figure 8.16: High Aβ agg may break the Trem2-Aβ interaction, forming
oligomers
279

21
Figure 8.17: High Aβ agg may break the Trem2-Aβ interaction from an
equilibrium point of view
280

Figure 8.18: High concentrations and Aβ agg endorses the C1q-Aβ
interaction, forming oligomers
281

Figure 8.19: Low Aβ agg may form the TREM2-C1q-Aβ interaction 282

Figure 8.20: High concentrations and Aβ agg endorses the C1q-Aβ
interaction, over the trimeric complex
283

Figure 8.21: High concentrations and Aβ agg negatively impacts the stability
of TREM2-C1q-Aβ
284

Figure 8.22: High concentrations and Aβ agg endorses the C1q-Aβ interaction
and offsets the equilibrium
285

Figure 8.23: TREM2 posttranslational regulation 286

Chapter 9

Figure 9.1: CyTOF captures the high-dimensional data sets and includes
structural information
296

Figure 9.2: CyTOF workflow 297

Figure 9.3: Neuro-architecture comparisons 298

Figure 9.4: imCyTOF comparison with immunofluorescence 299

Figure 9.5: Expression patterns of MBP and GFAP in 2 ischemic brain
regions using iMC
300

Figure 9.6: Image Processing Workflow 301

Figure 9.7: High Dimensional Imaging leads to complex comparisons 302

Figure 9.8: Dimensional reduction shows distinct myeloid populations
between regions
303

Figure 9.9: Mononuclear cellular expression intensities are regionally
restricted
304

Figure 9.10: Heterogeneous myeloid populations are found between regions 305

Figure 9.11: Myeloid expression clusters are localized in specific ischemic
stroke subregions
306

Figure 9.12: CD16
+
and Iba1
+
cells colocalize in the penumbra but not in the
necrotic regions
307

Figure 9.13: More p22PhOX reactive Iba1
+
cells are located within the
penumbra
308

22
Figure 9.14: CD16
+
and CD68
+
cells located more in the necrotic regions 309

Chapter 10

Figure 10.1: Trem2 and C1q affect Aβ immunoproteostasis 326

Figure 10.2: NK cell activation 327

Figure 10.3: Evolution of Trem2 and Natural Killer cells 328

Figure 10.4: pSyk signaling axis is a highly conserved mechanism that
determines activation from inhibition
329

Figure 10.5: Defining Resistance vs Tolerance paradigm 330

Figure 10.6: Tolerance vs Immune privilege 331

Figure 10.7: Bridging Trem2, C1q, and adaptive immunity 332

Figure 10.8: Trem2-dependent adaptive immune signaling 333

Figure 10.9: TREM2-C1q-Aβ regulates a fundamental signaling axis in brain
filtrating peripheral macrophages in AD
334

23
Table of Contents for Tables

Chapter 2
Table 2.1 Comparison between Reticular/Astrocyte-like Glia and
Ensheathing Glia:
78
Chapter 7
Table 7.1: Assessing potential feedforward/feedback mechanisms: 236
Materials and Methods
Table of All Primers 349
Table of All Reagents 350
Table of All CyTOF Antibodies 352

24
Table of Contents

Abstract ............................................................................................................................. 5
Forward ............................................................................................................................. 6
Acknowledgements ........................................................................................................... 7
Table of Contents for Figures .......................................................................................... 15
Table of Contents for Tables ........................................................................................... 23

Chapter 1: Introduction
Introduction...................................................................................................................... 29
Immune system evolution ................................................................................................ 29
Brain immune evolution ................................................................................................... 31
Alzheimer’s disease......................................................................................................... 33
Risk Factors for Alzheimer’s disease ............................................................................... 35
Triggering Receptor expressed on Myeloid Cells 2 (TREM2) .......................................... 36
Complement Pathway ..................................................................................................... 37
Trem2 and C1q in AD ...................................................................................................... 39

Chapter 2: Cerebral Innate Immunity in Drosophila Melanogaster
Abstract: .......................................................................................................................... 52
Introduction...................................................................................................................... 53
D. melanogaster as a model for vertebrate cerebral innate immunity ......................... 54
D. melanogaster CNS architecture ............................................................................ 56
Glial cell populations .................................................................................................. 58
Pericyte-like cells: surface, perineural, and cortex glia ............................................... 59
Astroglia .................................................................................................................... 59
Microglial homologs: ensheathing glia and reticular glia............................................. 60
Conservation of ensheathing vs. reticular glia activation states .................................. 62
Neuroinflammatory pathway conservation in D. melanogaster: cytokines .................. 64
Beyond innate immunity: adaptive immunity .............................................................. 67
D. melanogaster glia in human disease models ......................................................... 69
Ensheathing Glia resemble microglial-like cells ......................................................... 71
Results ............................................................................................................................ 72
Discussion: ...................................................................................................................... 74
Chapter 3: Cerebral Amyloidosis in Mus Musculus
Introduction:..................................................................................................................... 87
25
Amyloid-β: A Balance ............................................................................................... 87
Immunoproteostasis ................................................................................................. 88
Trem2 and C1q independently affect immunoproteostasis ....................................... 88
Results: ........................................................................................................................... 90
Initial APP/PS1 Trem2 and C1q deficient mice characterization ................................ 90
Trem2-C1q impact on immunoproteostasis in vivo..................................................... 91
Discussion: ...................................................................................................................... 93
Trem2 immunoproteostasis ....................................................................................... 93
The inflection point..................................................................................................... 95

Chapter 4: Characterization of Trem2 and/or C1q deficient APP/PS1 mice
Introduction.................................................................................................................... 108
Results: ......................................................................................................................... 109
Gliosis: ..................................................................................................................... 109
Identifying the Trem2-C1q signaling nexus .............................................................. 111
Establishing the Complement impact ....................................................................... 114
Mouse Behavior ....................................................................................................... 114
Human LOAD .......................................................................................................... 116
Discussion: .................................................................................................................... 117
Reconciling Trem2 and C1q deficiency in the APP/PS1 Mouse model .................... 117
Reconciling Trem2 and C1q dysregulation in Human LOAD .................................... 121

Chapter 5: Microglial Provenance
Introduction:................................................................................................................... 146
Microglial origins: ..................................................................................................... 146
Microglial development: Pre-programmed or the Tissue-Microenvironment-dependent
................................................................................................................................ 147
Results: ......................................................................................................................... 149
CNS-infiltrating hematogenous macrophages express Trem2 ................................. 149
Establishing microglial provenance .......................................................................... 150
Paradigm shift– Peripheral macrophages ................................................................ 152
Discussion: .................................................................................................................... 153
Identifying adult microglial ontogeny ........................................................................ 153

Chapter 6: Regulating Trem2 Phagocytosis
Introduction.................................................................................................................... 172
Phagocytosis ........................................................................................................... 172
26
Extracellular signal-Regulated Kinase (ERK) ........................................................... 174
Results: ......................................................................................................................... 174
Modeling phagocytosis ............................................................................................ 174
Trem2 signaling bifurcation ...................................................................................... 177
Microglial C1q opsonized Aβ phagocytosis .............................................................. 178
Impact of AKT on C1q opsonized Aβ phagocytosis.................................................. 179
Breaking tolerance– rescuing Trem2 phagocytosis impairments .............................. 179
Discussion: .................................................................................................................... 180
Syk-AKT-MAPK axis ................................................................................................ 183

Chapter 7: Innate Immune Tolerance
Introduction:................................................................................................................... 203
Innate immune training and tolerance: ..................................................................... 203
Results: ......................................................................................................................... 204
Establishing an immune training and tolerance paradigm in vitro ............................. 204
Chronic inflammatory settings induce tolerance in microglia. ................................... 206
Chronic inflammatory settings induce immune training in peripheral macrophages .. 208
Characterizing the innate immune training response to Aβ in peripheral macrophages
................................................................................................................................ 209
Assessing C1q-Aβ macrophage training paradigms................................................. 213
Discussion: .................................................................................................................... 215
Evolutionary advantage for immune tolerance in the brain ....................................... 215
TREM2 biology still left unanswered ........................................................................ 217

Chapter 8: Resolving the Neuroimmune Complex
Introduction:................................................................................................................... 248
Size Exclusion Chromatography, Multi Angle Light Scattering: ................................ 248
Circular Dichroism ................................................................................................... 250
Results: ......................................................................................................................... 251
TREM2 binds to Aβ; C1q opsonizes Aβ ................................................................... 251
TREM2, C1q and Aβ physically interact in vitro ....................................................... 252
Aβ can oligomerize and fibrilize in vitro .................................................................... 255
Aβ oligomers and prefibrils differentially impact Trem2-Aβ and C1q-Aβ stability ...... 256
Aggregated Aβ destabilizes the formation of TREM2-C1q-Aβ in vitro ...................... 258
Discussion: .................................................................................................................... 260

27
Chapter 9: imCyTOF
Introduction:................................................................................................................... 287
Imaging Cytometric Time of Flight (imCyTOF) ......................................................... 287
Imaging CyTOF Rationale ....................................................................................... 289
Results: ......................................................................................................................... 290
Comparing imaging techniques................................................................................ 290
Resolving the High Dimensional landscape of an ischemic stroke brain .................. 291
Discussion: .................................................................................................................... 295

Chapter 10: Discussion
Dissertation Discussion ................................................................................................. 310
Reflecting on Trem2’s evolutionary past .................................................................. 311
TREM2 the macrophage NCR (Trem1 vs Trem2) .................................................... 314
Not all TGF-β and IL10btolerogenic environments are immunoproteostatic
environments ........................................................................................................... 317
T-cell-mediated Tolerogenic environments .............................................................. 320
What about us? ....................................................................................................... 322

Materials and Methods
Protein isolation and detection ....................................................................................... 336
Histology ....................................................................................................................... 337
Immunocytochemistry .................................................................................................... 340
Circular Dichrosim ......................................................................................................... 341
Peripheral macrophage isolation ................................................................................... 343
Microglial isolation ......................................................................................................... 344
Flow Cytometry ............................................................................................................. 344
Confocal Microscopy & Imaris Bitplane .......................................................................... 345
ELISA ............................................................................................................................ 346

References
References .................................................................................................................... 354

28

29
Chapter 1: Introduction

“ 凡戰者，以正合，以奇勝。”

“In all fighting, the direct method maybe used for joining battle, but indirect methods will
be needed in order to secure victory.”

Ch5. Energy, Sun Tzu – The Art of War

Immune system evolution

During the Cambrian explosion, many species were created, and many died. In this
competitive ecosystem, the few that survived, acquired advantageous traits, one of which,
is the ability to defend itself from pathogens(Kimbrell and Beutler, 2001; Medzhitov and
Janeway, 2002; Fujita et al., 2004; Nahmias et al., 2011; Flajnik, 2014). When attacked,
the organism depends on the immune system to discriminate between self and non-self.
It then decides the mode of attack and elimination(Kimbrell and Beutler, 2001; Medzhitov
and Janeway, 2002). The immune system is not restricted to Kingdom Animalia, but
rather across all domains: Archea, Bacteria, and Eukarya(Kimbrell and Beutler, 2001;
Fujita et al., 2004; Vestergaard et al., 2014; Jones et al., 2016). For example, Domain
Archea utilize RNAi or CRISPR to eliminate viral infection(Vestergaard et al., 2014;
Koonin and Krupovic, 2015). Within Domain Eukarya, Kingdom Plantae adopts regionally
secreted antimicrobial factors and cytokines without a mobile army(Jones et al., 2016).
Kingdom Animalia is the beginning of what is commonly described as the immune
system(Kimbrell and Beutler, 2001; Fujita, 2002; Medzhitov and Janeway, 2002; Fujita et
al., 2004; Koonin and Krupovic, 2015).

30
The evolution of the innate immune system can be traced to Kingdom Porifera (Figure
1.1, orange line), sponges, with their pluripotent cells with macrophage-like abilities called
archeocytes (Wiens et al., 2005). As we move up the evolutionary cladogram,
Protostomes and Deuterostomes develop more specialized phagocytes called
coelomocytes that primarily clear dead cell debris(Hibino et al., 2006; Kim and Mylonakis,
2012). Upon entering Phylum Arthropoda, we develop Hemocytes and the homologous
chordate version, macrophages(Kimbrell and Beutler, 2001; Boehm, 2012; Leung et al.,
2015). By this point, these hemocytes and macrophages can do more than clear dead
cell debris, but also sense pathogens, eliminate threats, and undergo innate immune
training(Streilein, 1995; Rast et al., 2006; Pham et al., 2007; Cheng et al., 2014;
Schneider and Tate, 2016).

The more commonly known adaptive immune system with immunological memory
developed during Class Agnatha,” jawless” fish, which showed the earliest evidence of
adaptive immune-like cells(Flajnik, 2014) (Figure 1.1, Teal line). One line of evidence
suggests the development of the adaptive immune system originated from CRISPR-Cas
systems, developing into transposon-like elements, which then formed Major
Histocompatibility Complexes (MHCs), an essential system for the innate immune system
to identify self from non-self(Koonin and Krupovic, 2015). Spurred by gene duplication,
these transposons evolved into Recombining Activating Genes (RAG) and created
Variable Diversity Joining regions, V(D)J(Leavy, 2013; Kapitonov and Koonin, 2015;
Koonin and Krupovic, 2015). The Agnathans had primitive V(D)J and RAG system, that
developed into orthologous adaptive-like cells (Stet et al., 2005; Flajnik, 2014). By the
31
development of Class Chondrichthyes (sharks) and Actinopterygii (Teleost fish), T-Cells
and B-Cells were more homologous with the T-Cells and B-Cells(Stet et al., 2005; Boehm,
2012; Flajnik, 2014). During this time point, the emergence of cell types and genes
bridged the divide between adaptive and innate immunity, blurring the distinction between
cells involved in immune memory and pathogen elimination(Rast et al., 2006). Allele-and-
location-restricted expression, cellular proliferation, and expansion are some
characteristics that are classically attributed to the adaptive immune system; however,
these traits are now acquired by the innate immune system(Rast et al., 2006), begging
the question why and how?

Brain immune evolution

As these chordates moved from aquatic to terrestrial life, the necessity to eliminate
immune threats without causing self-harm became ever increasingly important(Fujita,
2002; Nahmias et al., 2011; Flajnik, 2014; Vivier et al., 2016). The working hypothesis
behind cellular specificity within a tissue specific microenvironment gained more traction,
thus came the rise of immune privileged organs(Streilein, 1995; Carson et al., 2006;
Banerjee and Bhat, 2007; Bundgaard and Abbott, 2008; Stein-Streilein and Caspi, 2014;
Sankowski et al., 2015). In Figure 1.2, we can observe when the modern-day blood brain
barrier (BBB) first developed. The development of the glial-sheath-like barrier evolved
independently and in parallel with the rise of coelomocytes(Bundgaard and Abbott, 2008).
Whether there is an extant or extinct common ancestor with both coelomocytes and a
glial-like barrier before the Proto-Deutero bifurcation, has not yet been discovered. While
32
these structural developments are coined functionally analogous or transcriptomically
homologous(Bainton et al., 2005; Mayer et al., 2009; DeSalvo et al., 2014), the arthropod
BBB only highlights the evolutionary importance of the BBB, giving rise to tissue-specific
immune systems, e.g. immune privileged organs(Streilein, 1995; Bundgaard and Abbott,
2008; Benhar et al., 2012).

The parallel evolution of Arthropod BBB-like tight-junction structures were identified in a
Drosophila melanogaster Moody mutant(Bainton et al., 2005), and later transcriptomics
of D. melanogaster surface glia suggested that mammalian vertebrate pericytes had a
common ancestor(DeSalvo et al., 2014). Regardless of its evolutionary derivation, the
hemolymph-brain barrier (HBB) still emphasizes the importance of a tissue specific
microenvironment that has acquired restricted immune function(Rast et al., 2006;
Banerjee and Bhat, 2007). As tissues obtained more specific functions, the immune
system symbiotically co-adopted more features, a reoccurring theme that is found
throughout this entire dissertation. These specific themes include immunoprotoeostasis
(Chapter 2), tissue specific immunity (Chapter 4), and immune tolerance(Chapter 8).

The co-evolution of the brain, and tissue-specific-macrophages were not evolved by
happenstance. The neuroimmune system depends on brain-resident macrophages,
microglia, to eliminate pathogens, to surveil the brain alongside astrocytes, and to
maintain brain homeostasis(Hickman et al., 2013; Ginhoux and Prinz, 2015; Crotti and
Ransohoff, 2016). Since the discovery of microglia by Pio Hortega in the early 1900s,
microglial ontogeny is understood to as hemopoietic in origin, and myeloid precursors
33
enter the brain at embryonic day 7.5(Ginhoux et al., 2010, 2016; Amit et al., 2015;
Matcovitch-Natan et al., 2016). Once brain the BBB forms at embryonic day 9.5, the brain
is sealed off from the environment, creating embryonic microglia(Ginhoux and Jung,
2014; Amit et al., 2015). Figure 1.3 illustrates microglial ontogeny, migration, and
development. Figure 1.3 also illustrates the development of tissue specific macrophages
in other organs, such as the lung, and skin. Additionally, these tissue specific
macrophages, are repopulated in the adult by the hemopoietic stem cells from the bone
marrow(Amit et al., 2015), yet microglial repopulation and survival is still
controversial(Elmore et al., 2014a; Huang et al., 2018). Whether macrophages enter the
brain or microglia proliferate in response to neuroinflammation or disease is still an active
area of research(Town et al., 2008; Mosher and Wyss-Coray, 2014; Jay et al., 2015).

While understanding microglial dysfunction through an evolutionary angle is valuable, this
dissertation will further address microglial ontogeny and microglial response in
Alzheimer’s disease.

Alzheimer’s disease

Alzheimer’s disease (AD) is the most common form of dementia, where a patient suffers
mood disturbances, progressive memory loss, and death. The World Health Organization
asserts that AD is the 6th leading cause of death, for which there is no cure, nor
preventative strategies. Age is an established AD risk factor, making everyone
susceptible to this neurodegenerative disease. Early onset AD can be linked to mutations
34
in three genes–amyloid precursor protein (APP), Presenilin (PSEN) 1, and PSEN2–
making up 1-6% of all Alzheimer’s cases(Selkoe, 2001; Selkoe and Hardy, 2016).
Alternatively, late onset AD (LOAD) is thought to arise from both genetic and/or
environmental risk factors(Zhang et al., 2013; Li et al., 2015). Of the many LOAD genes,
Apolipoprotein ε4 has been heavily documented(Perrin et al., 2009), however some
environmental risk factors and behaviors have been associated LOAD(Cacciottolo et al.,
2016, 2017; Moser and Pike, 2017).

Regardless of the etiology, the pathological hallmarks of the neurodegeneration seen in
AD are Amyloid β (Aβ) plaques, neurofibrillary tau tangles, and neuroinflammation
demonstrated by glial activation (gliosis)(Alzheimer, 1907; Selkoe, 2001; Heneka and
Obanion, 2007; Heneka et al., 2015; Selkoe and Hardy, 2016)(Figure 1.4). According to
the amyloid hypothesis model, neurotoxic Aβ is produced and released into the
extracellular space via the proteolytic cleavage of APP by β- secretase (BACE) and γ–
secretase (composed of Presenilins, Nicastrin, Aph1 and Pen2)(Selkoe and Hardy,
2016). While alternative Aβ species do exist, Aβ42 is thought to be the most neurotoxic
form when it aggregates and forms fibrils{Paresce:1996im, Szaruga:2017kn}, which
become a major contributing factor to innate immune activation in AD. Regarding the
strength of the immune response towards Aβ, understanding regulatory signaling
pathways that push microglia to engage in either Aβ clearance or tolerance are not well
understood(Figure 1.5). In other words, when do the Aβ adaptive and maladaptive
pathways engage, and what determines these shifts? Unfortunately, we still do not have
an answer in the context of AD.
35
Risk Factors for Alzheimer’s disease

The most common form of the disease, “sporadic” or LOAD, has a complex etiology that
includes genetic, environmental, and lifestyle risk factors. While the strongest genetic
susceptibility factor is a double polymorphism in the apolipoprotein E gene (APOE), a
protein involved in lipid transport(Corder et al., 1993), recent GWAS have identified a
cluster of AD risk genes related to innate immunity(Medway and Morgan, 2014; Gjoneska
et al., 2015). These findings have been a watershed, and include TREM2(Guerreiro et
al., 2013a; Jonsson et al., 2013) and the protein complement system(Reitz et al., 2013).
A common phenotype linking TREM2 and other innate immune gene risk alleles (e.g.,
CR1, CD33, MS4A4, MS4A6A, CD2AP, EPHA1) is microglial phagocytosis(Hazrati et al.,
2012; Griciuc et al., 2013; Reitz et al., 2013). Additionally, molecular imaging of AD patient
brains has revealed widespread microglial activation(Edison et al., 2008), and clinico-
pathological studies show a strong association between microglial abundance and
disease severity(McGeer et al., 1987; Arends et al., 2000; Bornemann et al., 2001;
Vehmas et al., 2003; Cagnin et al., 2006; Bolmont et al., 2008; Okello et al., 2009). Recent
transcriptomic evidence has suggested the convergence of two genes associated with
LOAD, TREM2 and C1q(Hazrati et al., 2012; Forabosco et al., 2013; Jonsson et al., 2013;
Reitz et al., 2013; Zhang et al., 2013; Matarin et al., 2015b) (Figure 1.6). Mechanisms for
how these two pathways intersect have not been studied in Alzheimer’s disease. This
dissertation will tackle this, head on.

36
Triggering Receptor expressed on Myeloid Cells 2 (TREM2)

Recent integrated genome wide association studies have identified a single nucleotide
variant in TREM2 as a significant risk factor associated with AD pathology(Jonsson et al.,
2013; Reitz et al., 2013; Zhang et al., 2013; Li et al., 2015; Matarin et al., 2015a). Previous
reports demonstrate that TREM2 loss-of-function confers risk for Nasu-Hakola disease,
a fatal disorder characterized by bone cysts, CNS lesions, and pre-senile
dementia(Turnbull et al., 2006; Colonna et al., 2007; Humphrey and Nakamura, 2015),
and GWAS have identified multiple TREM2 single nucleotide variants as significant risk
factors for AD(Forabosco et al., 2013; Jonsson et al., 2013; Zhang et al., 2013; Li et al.,
2015; Matarin et al., 2015b). Under homeostatic conditions, TREM2 promotes the
maturation of monocytes and dendritic cells(Bouchon et al., 2000; Sharif and Knapp,
2008; Ito and Hamerman, 2012; Hall and Agrawal, 2017).

TREM2 is a type 1 trans-membrane receptor with large extracellular iGg domain and a
small intracellular domain, which activates its adapter protein TYROBP, DAP12, to
engage any downstream activity(Klesney-Tait et al., 2006; Turnbull et al., 2006; Kober et
al., 2016), (Figure 1.7). A general consensus emerging in the field is AD risk-incurring
TREM2 variants are loss-of-function alleles(Kleinberger et al., 2014; Haas et al., 2016;
Song et al., 2016; Schlepckow et al., 2017). TREM2 activation among cells of myeloid
lineage dampens toll-like receptor (TLR) mediated pro-inflammatory signaling, shifting the
cell to perform anti-inflammatory efferocytosis (Klesney-Tait et al., 2006; Colonna et al.,
2007; Sharif and Knapp, 2008; Hsieh et al., 2009; Kober et al., 2016)(Figure 1.8).
37
Furthermore, TREM2 acts as a damage-associated molecular pattern (DAMP) receptor
that recognizes negatively-charged, lipids on apoptotic cells, and proteins including
Aβ(Wang et al., 2015a; Song et al., 2016). Trem2 overexpression in mouse models of AD
doesn’t alter AD-like pathology(Jiang et al., 2014), whereas Trem2 deficiency accelerates
AD-like pathology in two reports(Jay et al., 2015; Wang et al., 2015a) and attenuates it in
other reports(Jiang et al., 2016, 2017a; Leyns et al., 2017).

Complement Pathway

The protein complement pathway is a major component of the humoral innate immune
response; characterized by germ-line encoded, plasma soluble factors that bind to and
neutralize noxious proteins(Merle et al., 2015). This highly conserved pathway has three
major arms, listed oldest to more recent: the Alternative pathway, Mannose Binding
pathway, and the Classical pathway(Fujita, 2002; Fujita et al., 2004; Hibino et al., 2006;
Rast et al., 2006; Boehm, 2012) (Figure 1.9). All three pathways are involved in cytolysis,
inflammatory signaling, apoptotic cell clearance, chemotaxis, and opsonization(Merle et
al., 2015). Activation of each of these pathways converge upon Complement component
3, C3(Merle et al., 2015). The alternative pathway is thought to be a self-activating,
secreted, ancient immune defense and have evolutionary similarity with a family of thio-
esterase proteins(Levashina et al., 2001; Fujita et al., 2004; Flajnik, 2014; Merle et al.,
2015). The manose-binding lectin pathway made its mark after the bifurcation, where the
mannose-binding complex resembles the Complement component 1 (C1q)(Fujita, 2002;
Fujita et al., 2004).
38

C1q is a multimeric complex that binds pathogens and initiates the complement pathway
though opsonization and proteolysis of downstream complement proteins(Merle et al.,
2015; Thielens et al., 2017). The activation of bound C1q will drive a molecular cascade
creating a cytokines (e.g. C3a) and proteolytic factors (e.g. C3 convertase, leading up to
the creation of a membrane attack complex (MAC, C5b-C9) to lyse pathogens(Kishore
and Reid, 2000; Merle et al., 2015; Thielens et al., 2017). Taking a closer look at C1q
(Figure 1.11), C1q structure contains 3 major domains, a collagen stalk holding the 6
subunits together, a collagen stalk, and globular recognition domains (McGeer et al.,
1987; Rogers et al., 1996; Kishore and Reid, 2000; Gaboriaud et al., 2004, 2012; Ressl
et al., 2015). C1q activation in mononuclear phagocytes provokes an “anti-inflammatory
response,” promotes blood vessel dilation, recruits peripheral leukocytes, and produces
the membrane attack complex, initiating cytolysis(Ausubel, 2005; Arandjelovic and
Ravichandran, 2015; Merle et al., 2015; Shah et al., 2015; Thielens et al., 2017). C1q
deficiency in humans has been associated with systemic lupus erythematosus and C1q
-
/-
mice exhibit autoimmune-like pathology(Merle et al., 2015; Vasek et al., 2016). In mouse
models of cerebral amyloidosis, C1q colocalizes with Aβ plaques(Fonseca et al., 2004;
Hong et al., 2016).

39
Trem2 and C1q in AD

It is striking that two different innate immune genes, TREM2 and C1q, are linked to
increased LOAD risk by GWAS/integrative genomic studies(Forabosco et al., 2013;
Guerreiro et al., 2013a; Jonsson et al., 2013; Reitz et al., 2013; Zhang et al., 2013; Li et
al., 2015; Matarin et al., 2015b; Merle et al., 2015; Hong et al., 2016; Vasek et al., 2016)
(Figure 1.6). Deficiencies in TREM2 signaling and function are associated with white
matter loss and mild cognitive impairment (MCI)(Guerreiro et al., 2013a; Jonsson et al.,
2013; Reitz et al., 2013; Luis et al., 2014; Humphrey and Nakamura, 2015), and
neurological deficiencies related to complement signaling include Schizophrenia, Autism
Spectrum Disorder, and MCI(Perrin et al., 2009; Severance et al., 2014; Hong et al., 2016;
Vasek et al., 2016). While classically regarded as regulators of distinct immunological
responses, I will show that genetic loss of Trem2/C1q interrupts the formation of
hypothesized Trem2-C1q-Aβ neuroimmune complex (Figure 1.12), to engage Aβ immune
clearance, preventing AD-like pathology. This dissertation will demonstrate how TREM2
and C1q regulate Aβ immunoproteostasis in AD.

40

Figure 1.1: Evolution of the innate immune and adaptive immune system: Innate immune
systems evolved early in Kingdom Animalia, whereas adaptive immunity evolved within Phylum
Chordata. Cladogram depicts the evolutionary trajectory of different Phyla in Kingdom Animalia.
Black represents the Phylum/Class name, grey lists an example species, and red provides the
known name of macrophage-like cells of that phylum. Orange box represents all of phylum
Chordata.

41

Figure 1.2: Evolution of the Blood Brain Barrier: Tight-Junction-like barriers and Glial-like
sheaths evolved analogously in Kingdom Animalia. Cladogram depicts the evolutionary
trajectory of different Phyla in Kingdom Animalia. Black represents the Phylum/Class name,
grey lists an example species, and red provides the known name of macrophage-like cells of
that phylum. Orange box represents all of phylum Chordata.

42

Figure 1.3: Embryonic origins of innate immune cells: Embryonic myeloid cells migrate into
the blood stream and into their tissues of residence. Complete maturation and immune training
occurs in their tissue-specific residence.

43

Figure 1.4: Comparing Cognitively normal and AD brains: Gross brain drawings compares
between patients living with AD and cognitively normal. In addition to neuronal loss, insert
shows a cartoon representing the neurofibrillary tangles, gliosis and Aβ plaque deposits.

44

Figure 1.5: Innate immune response in a balance: Innate immune cells are trained to
respond to a pathogen and clear it. In certain circumstances, tolerance mechanisms may
engage and become mal-adaptive. With respect to the Amyloid Hypothesis, are these microglial
functions adaptive or maladaptive?

45

Figure 1.6: Trem2 and C1q are transcriptomically linked: Transcriptomic evidence from
Matarin and colleagues show that Trem2 and C1q are in the same transcriptomic network in the
hippocampus and cingulate cortex in mouse models of AD. This evidence also mirrors the
human disease. (Data from Matarin and colleagues Cell Reports 2015)

46

Figure 1.7: Structure of TREM2 and DAP12: Biochemistry data from the literature are
summarized and illustrated in this structural cartoon figure of TREM2 and DAP12.

47
Figure 1.8: TREM2 stimulates phagocytic pathways without inflammation: Upon TLR4
activation, TREM2-ligand interaction can inhibit NF-κB signaling, driving “anti-inflammation.”

48

Figure 1.9: Evolution of Complement signaling: C3-like activation evolved early in Kingdom
Animalia and is highly conserved across Kingdoms. MBL and C1q related proteins evolved
much later in the Deuterostomes. Cladogram depicts the evolutionary trajectory of different
Phyla in Kingdom Animalia. Black represents the Phylum/Class name, grey lists an example
species, and red provides the known name of macrophage-like cells of that phylum. Orange box
represents all of phylum Chordata.
49

Figure 1.10: C1q activation initiates the Classical Complement cascade: Bound C1q have
proteases that cleave C2 and C4 which creates C3 convertase, beginning the unstoppable
cascade. Arrows coming off the pathway are cytokines that elicit an inflammatory response to
nearby tissues/other immune cells. These pathways ultimately converge and create the
membrane attack complex (MAC) to lyse pathogens.

50

Figure 1.11: Structure of C1q: Biochemistry data from the literature are summarized and
illustrated in this cartoon figure of C1q.

51

Figure 1.12: The Central premise of this dissertation: A schematic diagram depicting the
hypothesized interaction of TREM2, C1q opsonized Aβ.

52
Chapter 2: Cerebral innate immunity in Drosophila melanogaster

“ 合于利而動，不合于利而止”

“If it is to your advantage, make a forward move; if not stay where you are. “
Chapter 12 Attack by fire
Art of War, Sun Tzu

Manuscript submitted to: AIMS Neuroscience

Authors: Brian P. Leung, Kevin R. Doty, and Terrence Town *

Abstract:

Modeling innate immunity in Drosophila melanogaster has a rich history that includes
ground-breaking discoveries in pathogen detection and signaling. These studies revealed
the evolutionary conservation of innate immune pathways and mechanisms of pathogen
detection, resulting in an explosion of findings in the innate immunity field. In D.
melanogaster, studies have focused primarily on responses driven by the larval fat body
and hemocytes, analogs to vertebrate liver and macrophages, respectively. Aside from
pathogen detection, many recent mammalian studies associate innate immune pathways
with development and disease pathogenesis. Importantly, these studies stress that the
innate immune response is integral to maintain central nervous system (CNS) health.
Microglia, which are the vertebrate CNS mononuclear phagocytes, drive vertebrate
cerebral innate immunity. The invertebrate CNS contains microglial-like cells –
53
ensheathing glia and reticular glia – that could be used to answer basic questions
regarding the evolutionarily conserved innate immune processes in CNS development
and health. A deeper understanding of the relationship between D. melanogaster
phagocytic microglial-like cells and vertebrate microglia will be key to answering basic
and translational questions related to cerebral innate immunity.
Introduction

The discovery of Toll receptors in Drosophila melanogaster provided a molecular context
to understand pathogen recognition by the innate immune system
1
. This finding opened
the door to the discovery of mammalian Toll-like receptors (TLRs) and other pattern
recognition receptors (PRRs), which in turn has illustrated the broad variety of pathogen
detection mechanisms, signaling components, immune modulating factors, and innate-
adaptive immune cross-talk, which has revolutionized the field of immunology(Meister et
al., 1997; Janeway CA Jr1, 2002; O’Neill et al., 2013). These initial studies in D.
melanogaster, as well as subsequent invertebrate studies, provided mechanistic insight
into peripheral immune responses driven by hemocytes and fat body cells(Hoffmann and
Reichhart, 2002; Williams, 2007).

More recent work in the invertebrate central nervous system (CNS) has revealed that
dysregulation of cerebral innate immune signaling in glial cells can lead to neuronal
dysfunction and degeneration(Petersen et al., 2012, 2013; Cao et al., 2013). While much
more remains to be learned regarding the immune-specific properties and function of
invertebrate glia in CNS health, glial biology studies in D. melanogaster have identified
54
specific glial subtypes that are hypothesized to perform functions similar to vertebrate
glia, ranging from phagocytosis to neurotrophic support, signifying the important role that
these enigmatic cells have within the CNS(Freeman and Doherty, 2006; Eroglu and
Barres, 2010). This review will provide anatomical, cellular, and molecular support for the
glial analogs found in the D. melanogaster CNS and address the promise this field holds
for modeling cerebral innate immunity.
D. melanogaster as a model for vertebrate cerebral innate immunity

Innate immune signaling is highly conserved throughout evolution(Levashina et al., 2001).
In D. melanogaster, innate immunity is largely carried out by hemocytes and the fat body,
analogous to vertebrate macrophages and liver(Hoffmann and Reichhart, 2002; Williams,
2007). Pathogen recognition pathways are initiated through genome encoded PRRs,
typified by the Toll receptor(Lemaitre et al., 1996; Poltorak et al., 1998). The high degree
of conservation between Toll signaling in fly hemocytes and TLR signaling in vertebrate
macrophages enables Drosophila to be used as a model system in mechanistic studies
of TLR signaling (Figure 2.1).

In mammalian tissues, there are numerous populations of resident, tissue specific,
phagocytic cells that are front-line responders to pathogens and injury. In the mammalian
CNS, microglia play this role and stand poised to survey the local environment and to
mediate innate immune responses. In D. melanogaster, glial cells are present within the
mushroom body and several studies have characterized these cell populations(Awasaki
et al., 2008; Hartenstein, 2011). Transciptome analysis between wild type and glial cell
missing (Gcm) mutants identified 45 glial-specific genes with human conservation
55
hovering around 80%
14
. This surprising finding hints at conservation of innate immune
roles extending beyond peripheral hemocytes and into the CNS. Since the innate immune
system is highly conserved in evolution and is the sole immune system of invertebrates,
it allows for the dissection of these molecular pathways that uniquely drive innate
immunity. Furthermore, the Gal4-UAS system provides a tractable D. melanogaster
system, allowing spatial and temporal manipulation of gene expression(Jenett et al.,
2012). Taken together, D. melanogaster represents an excellent in vivo model to
understand basic glial innate immune function that can be applied to vertebrate systems.

In the mammalian brain, innate immune activation and neuroinflammatory pathways are
thought to be major players in neurodegeneration(Ransohoff and Brown, 2012). Several
recent studies in D. melanogaster have examined the relationship between glia and
neurodegeneration. In one paradigm, activation of either of two innate immune pathways
leads to neurodegeneration. Specifically, loss of defense receptor 1 (Dnr1), a negative
regulator of the IMD innate immune response pathway, resulted in shortened lifespan and
age-dependent neuropathology(Petersen et al., 2012; Cao et al., 2013). In Dnr1 deficient
flies, whole brains exhibited pathologic vacuole formation throughout the neuropil,
indicative of neurodegeneration(Petersen et al., 2012). Furthermore, Dnr1 deficient flies
had reduced life span, motor impairment, and increased anti-microbial gene expression.
Glial knockdown of Relish – a nuclear factor-kappaB (NF-κB) analog and a key regulatory
gene in the IMD pathway – in Dnr1 deficient flies halted neurodegeneration, restored
lifespan and improved motor function(Cao et al., 2013). Moreover, similar experiments
found that bacterial infection in the CNS led to increased anti-microbial gene expression
56
and hastened onset of neurodegeneration. Similarly, knock down of Relish in glia
prevented neurodegeneration. These experiments suggest that glial cells can mediate a
detrimental form of innate immunity that endorses neuron loss.

D. melanogaster CNS architecture

While few would argue that the anatomy of the fly brain mimics that of the mammalian
brain, there remain key parallels between both structures(Kohl and Jefferis, 2011). The
D. melanogaster adult CNS is comprised of four, fused major ganglia: the subesophageal
ganglion, the protocerebrum, the deutocerebrum, and the tritocerebrum(Butler and
Hodos, 2005). The largest ganglion is the protocerebrum, which contains the majority of
the known brain regions and acts as a hub for environmental inputs from antennae and
omatidia, analogous to the vertebrate cerebrum. For the olfactory system, input from the
antennae synapse onto the antenna lobe, which then relays parallel projections to the
mushroom body (the vertebrate hippocampus analog), and the lateral horn. By
comparison, the vertebrate olfactory bulb sends projections, directly or indirectly, to the
thalamus and the hippocampus(Davis, 2004; Sakano, 2010). Summarized in Figure 2.2,
this olfactory circuitry parallelism demonstrates why the protocerebrum is an ideal brain
region to model glial influence on cerebral processing of environmental stimuli, including
learning and memory(Yu et al., 2013).

In the vertebrate CNS, the most abundant cell type is astroglia; however, in flies, neurons
outnumber all other brain cells(Rowitch and Kriegstein, 2010). Using cell proportions as
a basic metric for inter-species comparison, one may conclude different roles for glia.
57
However, across species, glia consistently play critical roles in CNS architecture,
neuronal maintenance, axon guidance, debris clearance, and brain barrier
formation(Freeman and Doherty, 2006; Mayer et al., 2009; Edwards and Meinertzhagen,
2010; Hakim et al., 2014; Ou et al., 2014). Therefore, invertebrate glial do, at least
somewhat, mirror the greater complexity in the vertebrate CNS. Glial cell populations in
the vertebrate brain consist of microglia and macroglia (astroglia and
oligodendrocytes)(Rowitch and Kriegstein, 2010). Although flies do not have
oligodendrocytes, they do possess a single population that has microglial and astroglial
functions. Despite the lower proportion of cells, glia in the adult D. melanogaster are
widely distributed, consistent with performing specialized functions.

Glial architecture in the protocerebrum provides evidence that these cells are positioned
to support neuronal function in this higher-order brain region that mediates behavioral
changes. Within the mushroom body, glial cells are distributed in heterogeneous clusters
that ensheathe the mushroom body somas, divide the mushroom body into
compartments, and interlace the neuropil in an unorganized mesh-like network(Ito et al.,
1997; Crittenden et al., 1998; Leiss et al., 2009; Edwards and Meinertzhagen, 2010).
Although studies have not functionally classified any particular glial subtype in the
mushroom body, this relative distribution parallels microglial heterogeneity found in the
murine CNS. Early work that enumerated microglial distribution in the mouse CNS found
a higher density of microglia in the hippocampus, olfactory telencephalon, basal ganglia,
and substantia nigara(Lawson et al., 1990; Tremblay et al., 2015). Within the microglia-
dense mouse hippocampus, there are greater numbers of microglia within the CA1 and
58
CA3 subregions(Jinno et al., 2007; Jinno and Kosaka, 2008). These studies suggest that
microglia have a role in learning and memory. However, associations between neuronal
activity and microglial density remain inconclusive(Lawson et al., 1990; Tremblay et al.,
2015). Since the fly glial populations around the mushroom body are largely
uncharacterized, their roles in olfactory learning and memory are not well understood.
Nevertheless, functional characterization of analogous structures within the D.
melanogaster brain can provide valuable information related to vertebrate biology and
function(Kohl and Jefferis, 2011).
Glial cell populations

As previously mentioned, glial cell numbers within the D. melanogaster CNS differ
substantially from vertebrates. In flies, 10-20% of the cells are glia, while glia make up at
least 50% of the vertebrate CNS(Rowitch and Kriegstein, 2010). In the vertebrate CNS,
glial cells are further subdivided into several classes, including microglia, astrocytes,
oligodendrocytes, and pericytes, amongst others(Freeman and Doherty, 2006; Rowitch
and Kriegstein, 2010). D. melanogaster glia show remarkable morphological and
functional similarity to vertebrates; however, functional conservation of specific glial
subtypes in flies has been more difficult to determine(Coutinho-Budd and Freeman,
2013). Although these differences suggest evolutionarily complex neural function, an
important concept to consider is that both anatomical and genetic approaches have
suggested at least some degree of interspecies functional overlap. This parallelism
provides a unique opportunity to decipher specific glial subtypes in the fly CNS that may
inform vertebrate microglia and macroglia function(Rowitch and Kriegstein, 2010). A
comprehensive review on invertebrate glial subtypes can be found elsewhere(Edwards
59
and Meinertzhagen, 2010). For the purpose of this review, we will cover three main cells:
pericyte-like cells, astrocyte-like cells, and microglial-like cells.
Pericyte-like cells: surface, perineural, and cortex glia

In the mammalian brain, pericytes are endothelial cells that comprise the blood-brain-
barrier (BBB), a physical and metabolic barrier that regulates cerebral blood flow(Winkler
et al., 2011). Although flies do not have a closed circulatory system, they do have pericyte
analogs that encase the periphery the brain, thereby separating the CNS from the
hemolymph, the fluid that acts as the primary distributer of oxygen, water, proteins, fats,
and sugars(Awasaki et al., 2008; Edwards and Meinertzhagen, 2010; Stork et al., 2012).
The Drosophila pericyte analogs – sub-divided into surface, perinurial, and cortex glia –
create a hemolymph-brain barrier, which is analogous to the vertebrate BBB(Mayer et al.,
2009; Bell et al., 2010; Stork et al., 2012). Evidence for the role of surface, perineurial,
and cortex glia in D. melanogaster comes from hemolymph dye injection experiments.
Specifically, Moody mutants have dye penetration into the CNS, whereas wild-type flies
do not. In Moody mutant flies, Moody knock-in animals expressing the protein on surface,
perineural, and cortex glia fully restore dye blockade into the CNS(Mayer et al., 2009).
Although this review will not consider hemolymph-brain-barrier glia in detail, these dye
penetration experiments highlight the key homologous properties of surface, perineural,
and cortex glia with vertebrate pericytes.
Astroglia

While surface, perineural, and cortex glia are analogous to pericytes, D. melanogaster
astrocytes are homologous to the same cells in vertebrates. In the mammalian CNS,
60
astrocytes are the most abundant glial cell type, and perform a wide array of
responsibilities ranging from nutrient transport, metabolism, and maintenance to
development, axon guidance, and synaptic function(Barres, 2008; Liu et al., 2010;
Stephan et al., 2013; Hakim et al., 2014; Hall et al., 2014; Ou et al., 2014; Tasdemir-
Yilmaz and Freeman, 2014). Astrocytes are easily identifiable in the fly brain and perform
similar roles as their vertebrate counterparts. While it has been shown that astrocytes
provide neurotrophic support in the fly eye(Edwards and Meinertzhagen, 2010), we will
focus on astrocytes within the protocerebrum. In the protocerebrum, astrocytes provide
neurotrophic support for dopaminergic neurons during development. As evidence of this,
Palgi and colleagues ablated D. melanogaster mesencephalic astrocyte-derived
neurotrophic factor (MANF), in astrocytes during development and observed
degenerating dopaminergic axons. When MANF was reintroduced as a knock-in, the
phenotype was reversed(Palgi et al., 2009). These experiments illustrate how astrocytes
promote neuronal survival and metabolic support.
Microglial homologs: ensheathing glia and reticular glia

Despite being outnumbered by other glia subtypes, microglia receive the majority of the
attention in studies of neuroinflammation and neurodegeneration in mammalian systems.
Microglia are the CNS resident mononuclear phagocyte in vertebrates, and play a vital
role in pathogen clearance, neuronal phagocytosis, and leukocyte recruitment into the
brain(Gate et al., 2010; Guillot-Sestier et al., 2015b). The last major glial cell type in the
Drosophila CNS are ensheathing glia. Recent studies demonstrate that these cells are
capable of microglial-like functions found in vertebrates. During axonal injury to olfactory
neurons, ensheathing glia express the Draper (Dpr) receptor and engage in phagocytic
61
clearance of neuronal debris, while astrocytes and other glia lack Dpr expression and do
not have a phagocytic function. To confirm the cell type and temporal specificity of this
phenotype, flies with Dpr RNAi knockdown in astrocytes maintained the ability to remove
neuronal debris, whereas flies with Dpr RNAi knockdown in ensheathing glia were unable
to remove neuronal debris(Freeman and Doherty, 2006; Doherty et al., 2009). Therefore,
these experiments illustrate that ensheathing glia are the main glial cell type that clears
neuronal debris. Although more work is needed to further characterize ensheathing glia
in D. melanogaster, we can look to other vertebrate studies and draw parallels between
vertebrates and invertebrates to understand how ensheathing glia may have immune-like
properties in vivo.

Vertebrate microglia morphology is plastic – alternating between a ramified state with
extended mobile processes and an activated, amoeboid shape(Town et al., 2005; Aguzzi
et al., 2013). Interestingly, both Awasaki and Hartenstein have described two potential
subtypes of adult fly ensheathing glia based on morphology: cells with flat bodies with
either small extensions or highly ramified processes(Awasaki et al., 2008; Hartenstein,
2011). Doherty and colleagues observed that glia with highly ramified extensions do not
express phagocytic genes such as Dpr, and therefore described them to be more
astrocyte-like(Doherty et al., 2009). However, recent evidence illustrates that microglia
can play a neurotropic and neuroprotective role, similar to classical astrocytes. For
example, live imaging experiments in the optic tectum of larval zebra fish illustrate that
resting microglia can perform astrocyte-like roles by extending their processes to reduce
neuronal activity(Li et al., 2012). Importantly, astrocyte-like glia are capable of expressing
62
Dpr to engage axonal phagocytosis during metamorphosis(Hakim et al., 2014; Tasdemir-
Yilmaz and Freeman, 2014), and therefore should not be confused with canonical
astrocytes. To minimize this confusion, Hartenstein asserts that this highly ramified,
astrocyte-like glial morphology should be renamed as reticular glia(Hartenstein, 2011).
Taken together, although ensheathing glia and reticular glia have defined roles within the
D. melanogaster CNS, their roles seem to overlap with vertebrate microglial and thus,
both glial subtypes should be considered as microglial homologs.
Conservation of ensheathing vs. reticular glia activation states

Vertebrate microglia morphologies are generally categorized into three phenotypes:
rounded, extended processes, and “bristled” (or highly ramified). The rounded
morphology is not commonly found in the healthy adult CNS, whereas the extended
processes and “bristled” phenotypes abound. Microglia with few extended processes are
found along neural tracts and highly ramified are commonly found interspersed within the
neuropil, surrounding cell bodies and synapses(Lawson et al., 1990). It is thought that
these structural differences can be attributed to microglial polarization states(Li et al.,
2012). Using common macrophage polarization terminology, the rounded morphology is
generally classified as M1 microglia. On the other hand, ramified cells are largely
classified as M2 microglia(Town et al., 2005; Aguzzi et al., 2013; Colton, 2013; Eggen et
al., 2013). A more extensive explanation of the M1/M2 dichotomy can be found
elsewhere(Colton, 2013). However, more recent studies assert that M1/M2 represent only
the extremes of a spectrum of polarization states influenced by the local environment,
neurons, and other microglia. Additionally, different microglial phenotypes can be
simultaneously present in any one brain region(Town et al., 2005; Breunig et al., 2013;
63
Doty et al., 2015), and these dynamic phenotypes offers explanation for the well-
recognized heterogeneity amongst microglia(Town et al., 2005; Tremblay et al., 2015).

Because microglial phenotype polarization is a key functional aspect of these innate
immune cells in the vertebrate CNS in both health and disease, it is important to consider
parallels in the fly. To this end, several studies have illustrated that both the three cell
morphologies and two polarization states exist in flies. First, reticular glia are capable of
engulfing mushroom body axonal processes during metamorphosis though a Dpr-
mediated process(Tasdemir-Yilmaz and Freeman, 2014). Although this contradicts a
previous study that describes ensheathing glia as the primary cell type to engage in Dpr-
mediated phagocytosis(Doherty et al., 2009), it is possible that both ensheathing and
reticular glia perform this role under different conditions in the adult fly CNS. Importantly,
Dpr expressing ensheathing glia have an aggressive phagocytic state characterized by
flattened cell bodies and small extensions, structurally mimicking M1 microglia in mice.
On the other hand, the generally anti-inflammatory and phagocytic M2 reticular glia have
longer, more ramified extensions, thus morphologically mirroring microglia found in the
healthy vertebrate CNS(Streit, 2006; Awasaki et al., 2008; Colton, 2013). Table 2.1
summarizes the comparison between ensheathing glia and reticular glia. It is likely that,
just like for vertebrate microglia, the polarization phenotype not only impacts
phagocytosis, but also expression of inflammatory mediators.

64
Neuroinflammatory pathway conservation in D. melanogaster: cytokines

Cytokines are secreted factors that instruct surrounding cells respond and play an integral
role in modulating immune responses in both vertebrates and invertebrates. Classical
vertebrate innate immune cytokines, such as tumor necrosis factor-alpha (TNF-α) and
interferons, have not been found in the fly(Foley and O’Farrell, 2003). However,
homologous innate immune pathways do still exist and consist of the immune modulators:
nitric oxide (NO), ATP, and transforming growth factor-beta (TGF-β)/Unpaired
(Upd)(Brown et al., 2001; Foley and O’Farrell, 2003; Johansson et al., 2005; Novakova
and Dolezal, 2011). These immune modulators are capable of recruiting and polarizing
hemocytes, and also ensheathing and reticular glia.

NO, a small molecule with a diverse set of roles in physiology, neurobiology and
immunology, plays beneficial roles as an anti-microbial and immunoregulatory cytokine.
Additional functions include vasodilation and cytotoxicity. Studies in flies have shown that
nitric oxide synthase (NOS) mutants infected with gram-negative bacteria at both the
larval and adult stages are more susceptible to infection compared to wild-type flies.
Those authors also demonstrated that feeding NOS inhibitors to wild-type flies resulted
in greater susceptibility to infection(Foley and O’Farrell, 2003). While NO is a potent
immunomodulator, it also causes collateral tissue damage, often associated with
dysregulated M1 responses in macrophages(Colton, 2013). NO cytotoxicity was
demonstrated in flies by exposing NOS mutant and wild-type flies to an airborne fungal
compound. Under those conditions, NOS mutants lived longer and were protected from
neurodegeneration. Similarly, feeding NOS inhibitors to wild-type flies protected them
65
from neurodegeneration and extended longevity(Inamdar and Bennett, 2014). These
results indicate that NO is a conserved immunoregulatory molecule capable of M1
polarization.

Thus far, peripheral hemocytes have been the major immune cell type studied in flies.
Similar to vertebrate macrophages, D. melanogaster hemocytes respond to
immunological challenge. In response to injury, hemocytes acutely express ADGF-A, a
homolog of adenosine deaminase 2. ATP is further converted to adenosine, amplifying
the inflammatory signal(Novakova and Dolezal, 2011). The origin of extracellular
adenosine is owed either to ATP released by neurons or degenerating axons. Studies in
vertebrates show that microglia respond to extracellular ATP via purinergic receptors and
initiate an innate immune response(Kettenmann et al., 2011). Although more research is
needed to show homologous CNS innate immune signaling in flies, ensheathing glia and
reticular glia are well-positioned to respond to ATP and participate in the inflammatory
cascade. Importantly, these experiments further illustrate how immune modulators
activate hemocytes from surveying their environment to responding to pathogens(Wood
and Jacinto, 2007).

Cytokines are known to play a critical role in synapse maturation and circuitry formation
during development and disease(Garay and McAllister, 2010). For example,
neurodevelopmental experiments in mice have found that chronic TNF-α exposure to
neurons rapidly matures synapses(Kaneko et al., 2008). There is no TNF-α analog or
homolog in D. melanogaster. However, the TGF-β homolog Upd has been shown to play
66
a dual role in infection and in olfactory learning and memory(Royet et al., 2005; Copf et
al., 2011). In the mushroom body, Upd modulates long-term memory by signaling though
Dome and short-term memory by modulating the Hop/Stat92E pathway(Brown et al.,
2001). Studies from invertebrates show that TGF-β signals though the Smad2/3
pathway(Copf et al., 2011). Upd is evolutionarily conserved and is also found in
Caenorhabditis elegans. In C. elegans, Dbl-1, the Upd homolog, is found to be necessary
for aversive olfactory learning and memory(Royet et al., 2005; Zugasti and Ewbank, 2009;
Zhang and Zhang, 2012). Since the Dbl-1 and Upd pathways are highly conserved across
species, this homology in worms and flies corroborates classical Pavlovian conditioning
experiments for learning and memory in mammals, supporting that expression of Upd at
specific regions along the olfactory circuit strengthen local mushroom body
synapses(Royet et al., 2005).

Moving from invertebrate TGF-β homologs to the vertebrate gene, studies from our group
have demonstrated that blocking TGF-β signaling in murine CNS-infiltrating peripheral
macrophages alleviates learning and memory deficits found in mice with Alzheimer’s
disease-like pathology(Town et al., 2008). Furthermore, CNS pathogen injection studies
show that microglial TGF-β secretion promotes neuronal survival(Mitchell et al., 2014).
Additionally, in visual circuitry formation, TGF-β is necessary for synaptic pruning and
development(Bialas and Stevens, 2013)(Bialas and Stevens, 2013). Drawing upon the
results in vertebrates, ensheathing and reticular glial-dependent secretion of TGF-β may
be necessary for learning and memory. Therefore, the study from Town and coworkers
67
illustrates a potential role for ensheathing or reticular glial Upd signaling within the
mushroom body, affecting learning and memory.

Since Pavlovian olfactory-dependent learning experiments are associated with
mushroom body function in flies, one study proposed that certain types of learning and
memory (e.g., short-term learning, long-term memory, and anesthesia resistant learning)
promote characteristic proteomic shifts. Using a proteomics approach, one study found
two immune-related genes that were differentially regulated: hemolectin and immune-
induced peptide 4 (also known as Dim4)(Zhang et al., 2014). In one study, Dim4 knock-
down forced hemocytes into a more amoeboid shape, suggesting M1
polarization(D’Ambrosio and Vale, 2010). Similarly, hemolectin is expressed by
hemocytes and modifies the dendritic tree during development. Because M2 is the
macrophage polarization phenotype that prunes synapses, these results are likely similar
to the Dim4 knock-down study(Williams, 2005). Additionally, hemolectin is a cytokine that
has been found in the hemolymph during bacterial infection(Lesch et al., 2007; Williams,
2007; Wood and Jacinto, 2007). This proteomics study further illustrates how Dim4 and
hemolectin could influence ensheathing glia and reticular glia polarity in response to CNS
immune challenge. Figure 2.3 proposes a putative role for cytokines in the D.
melanogaster CNS: how Dim4 could suppress M1/ensheathing glial polarization and how
hemolectin promotes M2/reticular glial polarization.
Beyond innate immunity: adaptive immunity

Adaptive immunity uniquely tailors the immune response to a pathogen through
genetically-encoded immunological memory. It has been suggested that at least some
68
form of adaptive immunity has been in existence throughout evolution, originating from
prokaryotes or early eukaryotes(Janeway CA Jr1, 2002; Pradeu and Cooper, 2012;
Westra et al., 2014; Barrangou, 2015). Primitive forms of adaptive immunity include
clustered, regularly interspaced short palindromic repeats (CRISPR) associated proteins
(i.e., CRISPR-Cas; found in prokaryotes) and RNA interference (present in eukaryotes).
Both of these enable pathogen recognition through integration of short nucleic acid
sequences into the genome, thereby providing “immunological memory”(Westra et al.,
2014). These intracellular defense systems primarily fight against viruses, while
evolutionarily recent adaptive immune mechanisms in vertebrates combat a more diverse
array of pathogens(Sunyer, 2013). Jawed vertebrates (beginning with teleost fish) were
the first to develop analogs to mammalian T and B cells, suggesting that these types of
adaptive immune responses are evolutionarily restricted, and not present in
invertebrates(Ting and Davis, 2005; Lange et al., 2011; Uribe et al., 2011; Sunyer, 2013).

While most agree that Drosophila melanogaster does not possess a mammalian-like
adaptive immune system, studies have nonetheless found evidence of fly immunological
memory. For example, priming flies with a sub-lethal dose of Streptococcus pneumoniae
improved survival in response to a second, lethal injection of the bacterium(Pham et al.,
2007). This elegant experiment demonstrates a type of immune memory(Schneider,
2007; Ayres and Schneider, 2012; Chambers and Schneider, 2012). Subsequent
experiments in the mosquito showed that the AgDscam gene contains a hypervariable
region. Furthermore, activation of the Toll and IMD pathways activated splicing factors
that modified the AgDscam transcript, increasing receptor avidity towards a pathogen–
69
potentially enabling a form of immune memory(Dong et al., 2012). In vitro experiments
using the Drosophila hemocyte S2 cell line have shown Dscam-mediated bacterial
surface recognition, resulting in pathogen elimination via phagocytosis(Schmucker and
Chen, 2009; Kounatidis and Ligoxygakis, 2012). Therefore, it seems that while D.
melanogaster has at least some form of adaptive immunity via a hypervariable gene
region, the fly eliminates pathogens via phagocytic, innate immune pathways.
D. melanogaster glia in human disease models

The presence of functional microglia-like cells supporting neuron health and inflammatory
processes in the fly CNS suggest that flies could be used to model neurodegenerative
disorders. Because glia evolved alongside neurons in both flies and vertebrates, it is
hypothesized that these two cell types have conserved mechanisms and analogous glial
responses to neurodegeneration(Kraft-Terry et al., 2009; Oikonomou and Shaham, 2011;
Marsh and May, 2012). By capitalizing on CNS specific immune activation in D.
melanogaster, one study modeled the neurodegenerative disease, Ataxia-
telangiectasia(Petersen et al., 2013). Ataxia-telangiectasia (A-T) is a multi-system
disease, characterized by radiation sensitivity and predisposition to cancer, caused by a
mutation in the A-T mutated (ATM) kinase, which ensures genomic integrity in response
to DNA damage(Bakkenist and Kastan, 2003; Derheimer and Kastan, 2010). In the
human CNS, A-T is characterized by significant neuronal loss(Yang, 2005). In flies, the
use of temperature-sensitive ATM mutant flies has revealed the presence of vacuoles
and wide-spread neurodegeneration in the CNS. Additionally, these flies exhibited
reduced mobility and an increase in Relish-dependent neurodegenerative immune
responses (Petersen et al., 2013). Furthermore, Relish knock-down experiments in glia
70
reduced CNS neurodegeneration, suggesting a neurotoxic innate immune role. These
relish-depleted flies also showed increased mobility and restored longevity(Petersen et
al., 2012). These studies show how glia have a central role in Drosophila CNS innate
immune responses, providing evidence that glia could drive this form of immune-
dependent neurodegeneration, and illustrating the relationship between immune cell
activation and neurodegeneration.

From these A-T experiments, we can infer that glial immune activation can cause learning
and motor impairment, an important step towards demonstrating the glia-dependent
innate immune activation seen in neurodegenerative pathologies such as Alzheimer’s
disease(Pandey and Nichols, 2011; Crotti et al., 2014), which is an age-related
neurodegenerative disorder that is characterized by memory loss, the cellular deposition
of neurotoxic peptides, hyper phosphorylated neurofilaments, gliosis, and
neurodegeneration(Holtzman et al., 2012; Selkoe, 2012). According to the amyloid
cascade hypothesis, amyloidogenic processing of amyloid precursor protein releases
amyloid-β peptide that is thought to drive Alzheimer’s disease pathogenesis(Selkoe,
2012). D. melanogaster Alzheimer’s disease-like models exhibit axonal transport deficits,
neurodegeneration, Aβ aggregate formation, and behavioral and motor impairment(Guo
et al., 2003; Iijima et al., 2004; Pandey and Nichols, 2011; Mhatre et al., 2013; Shaw and
Chang, 2013). These established Alzheimer’s disease-like flies use the fly eye to model
neurodegeneration and demonstrate neuronal pathology; however, the fly eye does not
correlate with key vertebrate brain regions. Therefore, future studies may go on to utilize
the protocerebrum and mushroom body to examine neuron-glial interactions. Such
71
studies should enable further clarification of the relationship between neurodegeneration
and memory loss driven by ensheathing and reticular glia. The studies from A-T fly model
stress the role of glial cells in driving neurodegeneration. Investigating the cell type,
phenotype, and activation state(s) that drive these pathological outcomes could suggest
a novel role for ensheathing/reticular glia in fly neurodegeneration. Furthermore, these
experiments would have the potential to establish an innate immune driven response.
Ensheathing Glia resemble microglial-like cells

Microglia are brain-resident myeloid cells that have macrophage-like qualities(Hickman
et al., 2013). Studies have functionally and genetically demonstrated that ensheathing
glia and reticular glia can parallel microglia-like behavior. Furthermore, observations have
structurally demonstrated that ensheathing glia more closely resemble the phagocytic
cells in the D. melanogaster CNS, and that reticular glia may also have a hand in
phagocytosis. From an anatomical perspective, there seems to be heterogenic
distribution of glia surrounding different regions of the mushroom body. These differences
may be attributed to the dynamic nature of microglia. Both ensheathing and reticular glia
respond to extracellular cytokines and inflammatory signals released into the extracellular
space. Current models of neurodegeneration demonstrate the increasingly important role
glia may have during the onset of neurodegeneration. Moving forward, we should
continue to utilize hemocytes to understand the role of ensheathing and reticular glia
within the D. melanogaster CNS. Although more research is needed to fully characterize
the diverse immune roles that these enigmatic glia play, homologous and analogous
comparisons between invertebrate and vertebrate models illustrate significant structural
and functional overlap.
72
Results

This previous section introduced the current evidence for a neuroimmune system in the
fly. To experimentally show the presence of a functional neuroimmune system in
Drosophila melanogaster. We took advantage of D. melanogaster genetics by using the
GAL4/UAS system, GAL4 transcription factor binds do the UAS(upstream activating
sequence) to drive the expression of a downstream gene. Since one of the hallmarks of
Alzheimer’s disease is the pan neuronal expression of Aβ, we used the GAL4/UAS
system to drive the expression of Aβ. Therefore, we used the promoter ELAV-Gal4, a pan
neuronal promoter, a W1118 (WT), UAS–Aβ42, and UAS–Aβ42
Arctic
.

To determine whether Aβ is neurotoxic to the fly, we crossed ELAV–GAL4 with W
1118
,
UAS–Aβ42, and UAS–Aβ42
Arctic
. These crosses produces offspring that are ELAV;W
1118

(herein known as W
1118
), ELAV;Aβ42 (herein known as Aβ42), and ELAV;Aβ42
Arctic
(herein
known as Aβ42
Arctic
). Using the survival assay as a measurement of Aβ neurotoxicity, we
found that that W
1118
live longer than Aβ42 and Aβ42
Arctic
flies (Figure 2.4). Since pan-
neuronally expressing W
1118
and Aβ42 flies can live beyond 100 days, pan-neuronal
expressing Aβ42
Arctic
flies can live up to 50 days, suggesting that Aβ42
Arctic
pan-neuronal
expression is more neurotoxic than Aβ42. When Aβ42
Arctic
flies are group by sex, we found
that Aβ42
Arctic
male die earlier than female flies.

To untangle why these pan-neuronally expressing Aβ42
Arctic
flies are more susceptible to
death, we used flies at two time points, 5 day old flies and 40 day old flies and quantified
the expression levels of NF-κB analog, Relish and Dif (Figure 2.5). We found that Relish
73
mRNA expression was much higher in Aβ42
Arctic
than Aβ42 flies at 5 days. This was
phenocopied at 40 days as well. When we assessed Relish at the protein level, we found
that this effect was phenocopied. Since there a parallel NF-κB analog pathway that
signals through Catus/Dorsal/Dif, we found that Cactus and Dif expression are increased
in the brains of flies expressing Aβ42
Arctic
(Figure 2.6) at 5 days of age, but not 40 days.
Furthermore, between W
1118
and Aβ42
Arctic
flies, downstream signaling gene Drosomycin
is also increased at 5 days. When evaluating W
1118
and Aβ42
Arctic
flies at 40 days of age,
we found a reduction in Cactus and Dif expression, however, the increase found in
Drosomycin was not observed. While Cactus/Dif expression does not change between
W
1118
and Aβ42 flies at either time point, it does change with Drosomycin expression.
These results suggest that Aβ and Aβ
Arctic
expressing flies could induce different and
immune-specific pathways.

Since flies predominantly have the innate immune system, we asked whether the D.
melanogaster phagocytic machinery was altered in vivo. We found that pathways
governing the phagocytic machinery was increased mostly in Aβ42
Arctic
flies (Figure 1.7);
these phagocytic genes in pan-neuronal Aβ42
Arctic
flies, Shark and Lamp1, are increased
at 5 days, when compared to W
1118
and Aβ42 flies. However, this increase in phagocytic
gene expression was abolished at 40 days, when comparing all genotypes. When
compared to the aged pan neuronally expressing W
1118
flies, flies that best model age-
dependent inflammation (inflammaging), the non-significant phagocytic gene expression
in these pan neuronally expressing Aβ42
Arctic
flies suggest that phagocytic activity is
reduced. These results suggest that (1) Aβ42 does not engage innate immune
74
phagocytosis when compared to Aβ42
Arctic
and aging reduces the engagement of
phagocytic pathways, suggesting that these effects could be related to immune tolerance.
While Aβ is a human-specific protein, whether the Drosophila innate immune system can
recognize Aβ and engage in phagocytosis has not been well documented. To
demonstrate that Drosophila immune system was capable of clearing Aβ (Figure 1.8), we
isolated hemocytes (macrophage-like cells) from W
1118
larvae and found that when
hemocytes were treated with aggregated Aβ42, Aβ42-treated hemocytes created Actin
+

vesicle around Aβ aggregates, suggesting the Drosophila immune system is capable of
clearing Aβ. Lastly, we observe whether Complement-like activation and phagocytosis
through Thio-ester peptide (TEP), Complement C3 homolog, activation occurs in Aβ42
Arctic
expressing flies (Figure 1.9). We found that mRNA expression in the two isoforms, TEP1
and TEP4, were both increased, however TEP4 was increased in 40-day old flies. These
results suggest complement-like activity could exists in our D. melanogaster model of AD.
However, the mechanism for how TEP proteins drive phagocytosis or Complement
activation remains poorly understood.
Discussion:

Throughout this chapter, I tried to lay the foundation for future studies in the neuroimmune
system of AD-like models in the D. melanogaster. I first found that innate immune
inflammatory pathways are upregulated in vivo, and these pathways are Aβ species
specific, with Aβ42
Arcitc
as the potentially more neurotoxic peptide. I also showed that
hemocytes phagocytose aggregated Aβ. The brain of these Aβ42
Arctic
flies express more
phagocytic transcripts and more TEPs, a Complement-like homologue for C3, than W
1118

and Aβ42 flies. While previous reports suggest that the D. melanogaster contains a
75
neuroimmune system using models of TBI(Cao et al., 2013), establishing whether the
microglial-like cell homolog (enshething glia) that engulfs Aβ in flies still unanswered. One
disadvantage of this chapter is that I did not identify the role of enshething glia or highlight
glial activity in vivo. Rather I measured mRNA transcripts and documented hemocyte
phagocytosis in vitro, suggesting the role of innate immune cells in the brain. While these
are whole brain lysate, whether neuronal cells are the main cell types expressing innate
immune inflammatory genes (Dorsal, Dif, and Relish) remains unanswered. However,
these data still does not exclude the possibility of innate immune involvement in vivo.

Aside from neuroimmune activation in these flies, these data also indicate Aβ does not
elicit a strong neuroimmune response compared to Aβ42
Arctic
. To reconcile this
discrepancy, Aβ alone may exist as monomers and re more readily cleared by the
neuroimmune system in the D. melanogaster, whereas in Aβ4
Arctic
flies, the Aβ aggregates
into oligomers more rapidly, eliciting the immune reaction. One interesting facet from
these data, is the potential effect of a weakened immune response toward Aβ42
Arctic
in
these AD-like flies at 40 days of age. Whether this is an evolutionary driven, last-ditch
effort to protect the organism from dying is not known. However, it is could be a form of
immune training as observed with a bacterial challenge(Pham et al., 2007). Whether this
immune weakening is impacted by the adaptive-like mechanisms in the fly are also not
known(Leung et al., 2015; Schneider and Tate, 2016). Furthermore, in addition to the
reduced phagocytic mRNA expression in 40-day old flies, we also saw this effect in
Dorsal/Dif expression and Relish expression. This reduced immune response is a
76
phenomenon known as immune training or immune tolerance in the innate immune
system, and this will be described in more detail in Chapter 6.

Overall, these data suggest that the Aβ has the potential to activate both Relish and
Dorsal/Dif pathways. However, these data also indicate that Aβ42
Arctic
may activate Relish
pathways over Dorsal/Dif pathways. As mentioned earlier, measuring mRNA levels can
be informative, one experiment that would resolve this cell type and immune specific
pathway activation would be using a Relish mutant, Drosal mutant, or Dif mutant flies.
These data would reassure the Aβ-dependent pathways that are Relish dependent or
Dorsal/Dif-dependent in the neuroimmune system of the fly. Even though these limitations
do exist, it is likely that the neuroimmune system plays an integral role in an Aβ42
overexpression fly model of AD-like pathology.

77

Figure 2.1: Homologous Signaling pathways Toll, IMD and Upd: The Innate immune
signaling pathway between vertebrates and invertebrates is highly conserved. Additionally, the
mechanisms by which these proteins promote transcription are very similar. Although the
downstream pathways for each pathway are different, they main focus is on the initial immune
response that is elicited by the pathogen.

78

Astrocyte-like/Reticular Glia Ensheathing Glia
Phenotype Larger Cell Body Flattened and smaller cell body
Process Length Long Short
Polarity M2 M1
Functions Phagocytosis of unneeded axons Phagocytose damaged axons
Draper Expression During Metamorphosis, pruning
Mushroom body axons
During axonal damage

Table 2.1. Comparison between Reticular/Astrocyte-like Glia and Ensheathing Glia: The
function and morphology of these glia are summarized given what is known about these cell
types. Additionally, hypothesized comparisons between vertebrate macrophage nomenclature
of M1 and M2 are used to describe the similarities found in Drosophila glia.

79

Figure 2.2. Paralleled olfactory circuitry: The olfactory circuitry in invertebrates and
vertebrates are very similar. Although the regional names and structures are different, the
functions for each of these regions mirror one another.

80

Figure 2.3. D. melanogaster cytokines could polarize glia to perform different functions:
Drawing up parallels from hemocyte function, since reticular/astrocytic glia are similar to M2
microglia/macrophages, they could modify the dendritic tree through the expression of
Hemolectin or Upd. Alternatively, reticular/astrocytic glia could express DIM4 and could
suppress the cytotoxic effect of activating ensheathing glia. On the other hand, ensheathing glia
could express NOS and ADGF in response to neuronal damage or pathogens, initiating neuron
death.

81

Figure 2.4: Survival curve of Aβ expressing flies: Survival curve using the D. melanogaster
(n=300) shows how long each genotype lived for. Each of these genotypes shown are ELAV
crossed flies. Flies were separated by sex. Below

82

Figure 2.5: Relish activation occurs in Neuronally expressing Aβ and Aβ
Arctic
Flies: Relish
mRNA was measured from individual fly brains. Below, immunoblot of normalized 49/110 relish
densitometry. N=40-50 flies. For all data represented in this figure, experiments were
repeated 3 times followed by a subsequent two-tailed ANOVA. Statistical annotations
are as follows: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

83

Figure 2.6: Neuronal Aβ and Aβ
Artic
Flies upregulate Toll pathways genes: Downstream Toll
pathway genes were isolated from individual fly brains, n=6. For all data represented,
experiments were repeated 3 times followed by a subsequent two-tailed ANOVA.
Statistical annotations are as follows: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

84

Figure 2.7: Phagocytic machinery is upregulated in Neuronal Aβ
Arctic
flies: Shark and Lamp
show sequence homology to mammalian phagocytic genes. mRNA was isolated from individual
fly brains, n=6. For all data represented, experiments were repeated 2 times followed by
a subsequent two-tailed ANOVA. Statistical annotations are as follows: † p<0.1; *
p<0.05 ; ** p<0.01; *** p<0.001.

85

Figure 2.8: D. melanogaster hemocytes are able to recognize and engulf Aβ: WT
drosophila hemocytes were isolated and treated with aggregated Aβ. Arrow denotes the
overlay of Aβ intensity and Actin phagosome.

86

Figure 2.9: C3-like Complement genes are increased in Aβ
Arctic
flies: TEP, Thio-ester
proteins show sequence homology to C3-like genes. mRNA was isolated from individual fly
brains, n=6. For all data represented in this figure, experiments were repeated 2 times
followed by a subsequent two-tailed ANOVA. Statistical annotations are as follows: †
p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

87
Chapter 3: Cerebral Amyloidosis in Mus Musculus:

“ 勢如張弩，節如機發。”

“Energy maybe linked to the bending of a crossbow; decision, to the releasing of the
trigger.”
Chapter 5 Energy
Art of War, Sun Tzu

Introduction:
Amyloid-β: A Balance

According to the amyloid hypothesis, amyloid-β (Aβ) species are produced and released
into the extracellular space via the proteolytic cleavage of APP(Hardy and Selkoe, 2002;
Selkoe and Hardy, 2016). While alternative Aβ species do exist, Aβ42 is thought to be the
most neurotoxic when found in aggregated form or in fibrillary form(Paresce et al., 1996;
Szaruga et al., 2017). These forms become the major contributing factor to drive innate
immune activation in Alzheimer’s disease (AD)(Hardy and Selkoe, 2002). In non-
pathological settings, Aβ is cleared by pericytes, astrocytes, or microglia, minimizing the
amount of pathological forms of Aβ(El Khoury et al., 2007; Guillot-Sestier et al., 2015b;
Heppner et al., 2015).

While the mechanisms of Aβ proteolytic processing and clearance are still an active area
of investigation, this chapter will not elaborate on APP metabolism and instead focus on
the receptor-ligand equilibrium. Specifically, when disequilibrium occurs among Trem2-
Aβ, C1q-Aβ, and Trem2-C1q-Aβ, I will describe how these each interaction alters
immunoproteostasis.

88
Immunoproteostasis

Since this dissertation predominantly focuses on the innate immune clearance, namely
microglia and macrophages, the balance between neurotoxic Aβ aggregate formation and
innate immune clearance is immunoproteostasis(Ebstein et al., 2013; Chakrabarty et al.,
2015; Kimura et al., 2015). This concept implicates that protein formation and extracellular
pathogen clearance is maintained at an equilibrium and that these clearance mechanisms
engages non-cytotoxic (and cytotoxic) phagocytosis (Ebstein et al., 2013). Furthermore,
immunoproteostasis also considers the canonical proteostasis mechanisms that occurs
within all cells and the additional roles immune cells must do to sustain an immune
response(Ebstein et al., 2013; López-Otín et al., 2013; Díaz-Villanueva et al., 2015).
Thus, while immune cells have an additional layer to standard proteostatsis, they must
also degrade the pathogen within the phagolysosomes and to recycle both membranes
and lysosomes for additional rounds of phagocytosis(Ebstein et al., 2013). Figure 3.1
illustrates this concept to capture the extracellular and intracellular mechanisms that
regulate immunoproteostasis.
Trem2 and C1q independently affect immunoproteostasis

Studies have shown how the innate immune pathways that regulate Trem2 or C1q activity
can independently affect immunoproteostasis. Trem2 has been described to bind Aβ and
Apoe for Trem2-dependent clearance(Wang et al., 2015b; Yeh et al., 2016; Lee et al.,
2018a) (red box). Simultaneously, Trem2 function can be terminated by the proteolytic
processing of metalloproteases, forming sTrem2(Kleinberger et al., 2014; Schlepckow et
al., 2017) . Intracellular functions of Trem2 maturation from ER glycosylation to cell
89
surface expression is also dependent on the turnover of this receptor to the surface(Kober
et al., 2016; Song et al., 2016; Park et al., 2017). Each level of regulation determines the
spatial and functional control that is needed to ensure Trem2 activates when necessary.
While there is an overwhelming amount of evidence pointing to the loss of Trem2
impairing signaling exacerbating disease pathology, Trem2 overexpression does not
simply improve disease pathology or pathogen clearance(Zhu et al., 2014; Jiang et al.,
2017b).

On the other hand, C1q overexpression can induce dysregulated apoptosis or C1q
deficiency can drive the onset of Systemic Lupus Erythematosus (SLE) (van
Schaarenburg et al., 2016; Song et al., 2017). In addition to expression levels of C1q,
spatial expression of C1q expression in the tumor microenvironment can either
encourage cancer immune evasion(Bulla et al., 2016), promote wound healing and
angiogenesis(Bossi et al., 2014), or suppress axonal growth post spinal cord
injury(Galvan et al., 2012). While these studies show the benefits and detrimental effects
of C1q expression, we can conclude that tightly regulating C1q expression and quantity
within a tissue-microenvironment fundamentally shows how C1q subtilty modulates
immune activity. While this dissertation does not investigate the additional downstream
mechanisms of C1q, we know that Complement activation can affect immune
homeostasis via the classical, lectin, and alternative pathways(Merle et al., 2015). When
these pathways are activated, each proteolytic activation will create a cytokine and propel
the complement cascade. Cytokine release (pink) and the activated cascade (orange) will
affect immunoproteostasis.
90

In this section, I will interrogate one facet of immunoproteostasis (orange box) that
incorporates pathogen sensing (red box), phagolysosome recycling (red box), and to
autophagy (green box). The results from this section will explain how the deletion of
Trem2 and C1q in vivo leads to changes in Aβ deposition in the brain of our mouse model
of AD.

Results:

Initial APP/PS1 Trem2 and C1q deficient mice characterization

Previous reports that study the loss of Trem2 or C1q in vivo show that complete loss of
Trem2 exacerbates amyloid deposition(Wang et al., 2015b; Ulland et al., 2017), whereas
complete loss of C1q does not exacerbate amyloid deposition(Fonseca et al., 2011).
Using the classic mouse model of cerebral amyloidosis, APPSwedish Presenilin1∆E9 (herein
known as APP/PS1)(Jankowsky et al., 2001, 2004), we crossed Trem2 deficient animals
with C1q deficient animals and obtain 10 primary genotypes for all main in vivo
experiments. This combinatorial deletion of Trem2 and/or C1q using this cerebral
amyloidosis model should show differences in amyloidosis that can be attributed to either
Trem2-dependent effects, C1q-dependent effects, or a synergistic effect of Trem2 and
C1q dependent effects. Figure 3.2 illustrates one of our many breeding strategies to
obtain these mice.
91
Trem2-C1q impact on immunoproteostasis in vivo

To examine the impact of cerebral amyloidosis in Trem2 or C1q deficient APP/PS1mice,
these mice revealed genotype-specific alterations in cerebral amyloidosis. Using 6E10
immunostaining in the cingulate cortex, entorhinal cortex, and hippocampus (Figure 3.3),
Trem2 deficiency affects diffuse plaque staining, whereas the C1q deficiency shows a
minimal difference. Additionally, the compound loss of both Trem2 and C1q reveal a
significant increase in diffuse 6E10 positive plaque staining. The compound loss of Trem2
and C1q are more pronounced in the cingulate cortex and entorhinal cortex (Figure 3.3A,
B), suggesting that plaque deposition occurs in these regions of the brain before affecting
the hippocampus.

To assess the denser Aβ plaque cores, Thioflavin S staining was used. Unlike with 6E10
staining, genetic loss of Trem2 in APP/PS1 transgenic mouse displayed minimal
increases in both cingulate and entorhinal cortex brain regions (Figure 3.4A, B), whereas
hippocampal stains do not show an increase (Figure 3.4C). Conversely, the loss of a
single C1q allele potentiates dense plaque formation. Moreover, compound loss of C1q
and/or Trem2 on dense plaque formation appears more dependent on C1q than Trem2
(Figure 3.4), where a single loss of C1q can severely exacerbate the nucleation of dense
core plaque formation.

Combining the immunohistochemistry results from both diffuse and dense Aβ, we find
that impairments in Trem2 could affect soluble forms of Aβ, while C1q deficiencies could
92
affect the more insoluble forms. These differences suggest differences in receptor-
pathogen binding affinities.

To corroborate the immunohistochemistry results, we used electrochemiluminescence
assays (multiplex of Aβ1-38, Aβ1-40, and Aβ1-42) to quantify the amount of Aβ species. An
increase in soluble Aβ levels was associated with the absence of Trem2 (Figure 3.5-7),
whereas the guanidine Aβ levels were related to C1q (Figure 3.7). Additionally, these
soluble oligomeric Aβ levels were more dependent on the loss of C1q (Figure 3.8).

While bar graphs show data in one-dimensions, we plotted these measurements on a
scatter plot by Aβ species, and Aβ solubility (Figure 3.9). We observed that in each
scenario, these transformed measurements could segregate into two groups: between
animals with a complete Trem2 (magenta arrow) or C1q (green arrow) deficiency. These
initial scatterplots further corroborate the ELISA and IHC data that we observed earlier,
suggesting that Trem2 and C1q operate differently to impact Aβ cerebral amyloid
deposition in the brain. Since each parameter is plotted in pairs or 2 dimensions, we are
unable to interpret Aβ states as a whole. Therefore, we reduced the dimensionality from
seven parameters (each of the MSD/ELISA measurements) down to two, using the
principal components analysis (Figure 3.10). We found that these 5 genotypes are
grouped into three clusters. We noticed a few stark differences: Cluster 2 (green) grouped
mice associated with C1q deficiencies, whereas Cluster 3 (pink) grouped mice associated
with Trem2 deficiencies. Although we observed that Cluster 1 contained mostly APP/PS1
mice, there were a few C1q deficient animals in this cluster, suggesting that these mice
93
are in the process of falling towards Cluster 3. Taking all these data together, we find that
that both Trem2 and C1q independently affect Aβ deposition in vivo, and deficiencies in
either gene result in impaired Aβ immunoproteostasis.
Discussion:

Trem2 immunoproteostasis

Data from this chapter indicates that the loss of Trem2 or C1q can exacerbate Aβ species
formation and its deposition (Figure 3.3-3.10). These in vivo data also illustrate that any
impairments in Trem2 function could impact the more soluble Aβ compartment, whereas
C1q would impact the oligomeric Aβ fraction. Although the loss of both genes affects the
insoluble forms of Aβ, the biochemistry for how insoluble Aβ forms become dense core
plaques are subtlety different. As an additional variable, C1q deficiency has the ability to
induce protein nucleation when bound to a protein, promoting the initial Aβ peptide
seeding(Webster et al., 1995). Whereas, Trem2 deficiency via shedding (proteolytic
cleavage of Trem2 to form sTrem2) or misfolding could drive Aβ plaque formation by
reducing the Aβ phagocytosis mechanisms that are Trem2-dependent. To restate,
shedding would terminate all receptor-pathogen interaction because the transmembrane
domain cannot activate downstream signaling pathways(Wu et al., 2015a; Zhong et al.,
2017), whereas misfolding though improper glycosylation could alter Trem2 function,
impeding downstream signaling(Kober et al., 2016; Song et al., 2016; Park et al., 2017).

While sTrem2 has been described to promote macrophage and microglial survival,
whether it sustains the Aβ response is not well understood. Measuring levels of sTrem2
94
in these mice would describe the activation state of these mononuclear phagocytes,
however the sample preparation and the sample quality are not properly prepared for
sTrem2 detection. Furthermore, in this chapter, most of these histological quantifications
are qualitative without enough quantitative measures to measure each facet of
immunoproteostasis. For example, this chapter focuses on how the loss of Trem2 or C1q
impairs the detection and sensing of Aβ in the extracellular environment, which is
summarized by the red box in Figure 3.1. Aβ immune detection is important for
immunoproteostasis because it is, by definition, the main process that deviates standard
proteostasis from immunoproteostasis. Downstream degradation, clearance, and
recycling of phagolysosomes further differentiate immunoproteostasis from standard
proteostasis. This later point was not addressed in this chapter, but it will be characterized
by in vitro experiments in Chapter 6.

In addition to pathogen sensing, measuring the signaling switch between
immunoproteostasis and standard proteostasis was not addressed in this chapter and in
this dissertation. Recent evidence has shown that immune specific expression of Low-
Molecular-Mass Protein (LMP7), a subunit within the immunoproteasome, determines the
regulatory switch between immunoproteostasis-specific pathways and standard
proteostasis pathways(Ebstein et al., 2013; Kimura et al., 2015). Because LMP7 was not
measured in vivo, it is difficult to address intracellular immunoproteostasis pathways or
extracellular pathways that are Trem2 and/or C1q dependent. The central hypothesis of
this dissertation indicates that Trem2 interacts with C1q opsonized Aβ in the extracellular
95
environment to initiate phagocytosis. Therefore, we moved away from intracellular
immunoproteostasis pathways.

The inflection point

If we address the extracellular immunoproteostasis effects, we must to address the
equilibrium constants between Trem2-Aβ, C1q-Aβ, and the hypothesized trimeric
complex, Trem2-C1q-Aβ. Showing quantitative data for equilibrium constants or binding
affinities would show which species interacts more strongly with Aβ. In this chapter, I was
able to show some evidence for an equilibrium. To recapitulate our definition for
equilibrium, equilibrium implicates a balance point between two or more partners
producing an inflection point holds balance. In this chapter, I did not measure the
equilibrium constants among Trem2, C1q, and Aβ. However, the data in this chapter
implicate that the loss of Trem2 or C1q in APP/PS1 mice produces more soluble Aβ that
is Trem2 deficiency-dependent and oligomerized Aβ that is C1q deficiency dependent.
Additionally, with respect to the equilibrium, the PCA analysis in Figure 3.9 shows
APP/PS1 values between Trem2 and C1q deficient values. To highlight the importance
of this figure, the PCA analysis groups animals with similar Aβ expression levels tougher.
Therefore, when Trem2 and C1q deficient mice reside in different quadrants in relation to
APP/PS1, it means they operate in independent ways to modulate the Aβ species in the
extracellular environment. When we address compound deficient mice, APP/PS1 Trem2
-
/-
C1q
+/-
or APP/PS1 Trem2
+/-
C1q
-/-
mice, these data show these mice in a “transition
zone” between complete Trem2 deficiency or C1q deficiency, where losing an additional
gene “pulls” the values towards the APP/PS1 midline.
96

These data further corroborate the notion that, if these two proteins do intersect to alter
Aβ immunoproteostasis, their method of interaction would be through Trem2-C1q-Aβ
phagocytosis at an inflection point that shifts microglial and macrophage function in AD
(Figure 3.11). This inflection point could also be sensitive to Aβ conformational state and
Aβ concentration. This inflection point supports the previous studies in mouse models of
AD that suggest a Trem2-dependent transition between activation states within
mononuclear phagocytes (Keren-Shaul et al., 2017; Krasemann et al., 2017). While these
experiments in this chapter do not provide the complete evidence for an inflection point
in vivo, the following chapters will provide a better picture of this bifurcation and inflection
point in relation to clearance (Chapter 5) and interaction(Chapter 8).

97

Figure 3.1: Immunoproteostasis vs proteostasis: Schematic diagram compares the classical
proteostasis pathway in the context of immunology. The immune system accounts for pathogen
sensing, clearance, and coordinated pathogen eradication, which is summarized by the pink
and red pathways. In this dissertation, I focus on the pathogen sensing and clearance (red box)
as it may potentially relate to the other processes, green (stress response), blue (protein
degradation), and yellow (protein synthesis).

98

Figure 3.2: Mouse breeding paradigm: Breeding flow chart do depict some of the crosses
used to capture all the mice needed for this study.

99

Figure 3.3: Trem2 and C1q deficiency impacts fibrillar Aβ plaque formation: 6-month-old
APP/PS1 mice (n=5 per genotype) were stained with 6e10. Representative micrographs are
shown for only CC, adjacent to these images are quantified percent areas for CC. Scale bar
represents 100um (CC, cingulate cortex; EC, entorhinal cortex; HC, hippocampus). Statistics
annotations are compared to APP/PS1: * p<0.05 ; ** p<0.01; *** p<0.001.
100

Figure 3.4: Trem2 and C1q deficiency impacts fibrillar Aβ plaque formation: 6-month-old
APP/PS1 mice (n=5 per genotype) were stained with ThioflavinS. Representative micrographs
are shown for only CC, adjacent to these images are quantified percent areas for CC. Scale bar
represents 100um (CC, cingulate cortex; EC, entorhinal cortex; HC, hippocampus). Statistics
annotations are compared to APP/PS1: * p<0.05; ** p<0.01; *** p<0.001.

101

Figure 3.5: Trem2 deficiency increases soluble Aβ 38 formation: Aβ 38 species were
measured using the MSD platform. (Left) Detergent soluble fraction and (Right) Guanidine-HCl
extracted fraction were measured. Raw values were normalized to mg of cerebral cortex.
Statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

102

Figure 3.6: Trem2 deficiency increases soluble Aβ 40 formation: Aβ 40 species were
measured using the MSD platform. (Left) Detergent soluble fraction and (Right) Guanidine-HCl
extracted fraction were measured. Raw values were normalized to mg of cerebral cortex.
Statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

103

Figure 3.6: Trem2 deficiency increases soluble Aβ 42 formation: Aβ 42 species were
measured using the MSD platform. (Left) Detergent soluble fraction and (Right) Guanidine-HCl
extracted fraction were measured. Raw values were normalized to mg of cerebral cortex.
Statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

104

Figure 3.8: C1q increases oligomeric Aβ: 8E21 oligomeric ELISA was measured using the
detergent soluble fraction. Statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05 ;
** p<0.01; *** p<0.001.

105

Figure 3.9: Trem2 and C1q deficiency differentially affects Aβ: Log 10-based ELISA values
were plotted two-dimensionally by (Upper Left) Triton Soluble by Guanidine soluble Aβ 40, (Upper
Right) Triton Soluble by Guanidine soluble Aβ 42, (Lower Left) Triton Soluble Aβ 42 by Oligomeric
Aβ, (Lower Right) Guanidine Soluble Aβ 42 by Oligomeric Aβ. Green Arrows depicts a C1q
dominant effect, whereas magenta arrows depict Trem2 dependent effects.

106

Figure 3.10: Trem2 and C1q deficiency independently affects Aβ proteostasis: Principal
Component analysis was performed on the 7 Aβ ELISA measurements (Top). Clustering
algorithms isolate Trem2 deficient animals apart from C1q deficient animals and are drawn on
the graph. K-means clustering algorithms indicate the number of animals in in each cluster
(Bottom).

107

Figure 3.10: Trem2 and C1q affect Aβ proteostasis: Summary of this chapter is illustrated
above. If the three molecules interact, they should interact somewhere between small Aβ
assemblies and Aβ oligomers.

108
Chapter 4: Characterization of Trem2 and/or C1q deficient APP/PS1 mice:

“ 紛紛紜紜，鬥亂，而不可亂也。渾渾沌沌，形圓，而不可敗也。”

“Amid the turmoil and tumult of battle, there may be seeming disorder and yet no real
disorder at all; amid confusion and chaos, your array may be without head or tail, yet it
will be proof against defeat. “

Chapter 5 Energy
Art of War, Sun Tzu

Introduction

The motivation behind this chapter is to understand how two disparate innate immune
pathways intersect in Alzheimer’s disease. To echo the model described in Chapter 2, we
utilized mice deficient in Trem2 and/or C1q in a wild type background, as well as mice
with cerebral amyloidosis, APPswe PS1∆E9 (APP/PS1) mice, a classic mouse model to
study Alzheimer’s disease. While the pathways for Trem2 or C1q generally been studied
independently, one study investigates the role of both genes in response to pneumonia
at molecular level(Sharif et al., 2014a). While this study was restricted to the lung, we
have taken this further to understand how the deficiency of either Trem2 or C1q impairs
signaling pathways and cognitive function in AD. This chapter will characterize the in vivo
signaling pathways that are mediated by the loss of Trem2 or C1q in APP/PS1 mice and
assess the behavioral impairments in these Trem2 and/or C1q deficient mice.

109
Results:

Gliosis:

While the previous chapter described Aβ immunoproteostasis in the APP/PS1 mice, we
sought to investigate how the Trem2 and C1q deficient mice affected the glial response
to Aβ deposition. We chose to focus on the two main types of glial cells in the brain
because astrocytes and microglia support neurons and have been described to clear Aβ.
To assess glial activation at 6 months of age, we first assessed whether we had regional
differences between wildtype B6 and APP/PS1 mice (Figure 4.1). We used Glial Fibrillary
Acidic Protein (GFAP) to visualize the percent area of GFAP in a given region of the
mouse brain. Among the three regions of the brain that we assessed, GFAP expression
is consistently increased, suggesting that the glia are activated in response to the
neurotoxic peptide Aβ. Whereas, Ionized calcium binding adaptor protein 1 (Iba1) a
marker for microglia, show regional-specific increases in Iba1 expression.

When we assess how Trem2 deficiency and/or C1q deficiency affects gliosis (Figure 4.2),
we find that there are regional specific differences in GFAP and microgliosis. In the
cingulate, we find a drop in GFAP intensity among all genotypes, whereas Iba1 staining
increases among almost all genotypes, suggesting a microgliosis in the cingulate cortex.
When we assess the entorhinal and hippocampal cortices, we find that GFAP levels show
that Trem2 deficiency drives the GFAP response, while C1q slightly reduces GFAP
activation. However, in compound deficient animals there is no change in GFAP
activation. This genotype specific trend in glial activation suggests that a single deletion
110
of Trem2 drives reduced astroglial activation. When we assess microgliosis in the
entorhinal and hippocampal regions, we find that Trem2 and C1q deficiency slightly
increases microgliosis in AD, however this effect is abolished in compound deficient
animals. Taken together, these genotype-specific and regional differences between
Trem2 and C1q deficient animals could be inherently attributed to the brain’s regional-
specific differences or the inherent heterogeneity that exists among all innate immune
cells of mononuclear lineage, microglia/macrophages.

Since the overall trend is a reduction in astrogliosis and an increase in microgliosis, we
next asked whether Trem2 and C1q was increased in APP/PS1 mice (Figure 4.3). We
found that at 6-months APP/PS1 mouse brains all presented an increase in both Trem2
and C1q, protein and mRNA. These results corroborate the transcriptomic studies found
in in APP/PS1 mice. We next asked if the loss of Trem2 and/or C1q affected Trem2
expression. We found that a single loss in Trem2 slightly increased Trem2 expression
(Figure 4.4), while the single loss of C1q reduced Trem2 protein expression. However, in
compound deficient animals, APP/PS1 Trem2
+/-
C1q
-/-
, we found a further increase in
Trem2 protein expression. When we assess C1q expression in these animals, we find
that they are all reduced. These results show that C1q expression is greatly reduced when
we remove a single copy of Trem2 and C1q in APP/PS1 mice. These immunoblot
quantification results suggests that there is a relationship between Trem2 and C1q, where
a reduction in C1q expression will increase Trem2 expression in APP/PS1Trem2
+/-
C1q
-/-

mice. These results could indicate a compensatory effect in the Trem2-C1q signaling axis.

111
Identifying the Trem2-C1q signaling nexus

Previous reports have characterized that Trem2 sustains microglial activation and
survival, therefore we assessed whether there was an increase in cleaved caspase 3
activity in these mice and we found that there were no significant changes in cell death in
these mice at 6 months of age (Figure 4.5). On the other hand, C1q deficiency in AD-like
mouse models has been shown to have synaptic impairments (Figure 4.6). We found that
synaptophysin levels increase in Trem2 or C1q deficient mice. While, the most striking
effect is in the APP/PS1 Trem2
+/-
C1q
-/-
mice, the complete loss of C1q with a single loss
of Trem2 could suggest the lack of synaptic pruning in these mice. However, differences
in synaptophysin levels do not necessitate synaptic impairments, therefore further
experiments are needed to determine whether synaptic pruning is impaired in these mice.

Since the absence of Trem2 and C1q predominantly affects mononuclear phagocytes,
we switched gears to assess neuroinflammation, in particular NF-𝜅β signaling. To activate
the canonical NF-𝜅β pathway, translocation of p50 and p65 subunits of the NF-𝜅B
complex depends on the phosphorylation of IKK (I𝜅β). Once phosphorylated, this gene is
marked for ubiquitination and degradation. Therefore, we assessed whether I𝜅β is
activated our Trem2 and C1q deficient animals (Figure 4.7). We found that Trem2
deficiency increases I𝜅β phosphorylation, however, it does not increase in C1q deficient
animals, suggesting that the NF-𝜅β pathway is uninhibited when Trem2 is deleted. In
other words, Trem2 activation may inhibit NF-𝜅β signaling and Trem2 NF-𝜅β is relatively
normal in C1q deficient animals. Additionally, these data indicate Trem2 or C1q deficient
112
mice do not converge upon NF-𝜅β because there is not synergistic interaction between
the two genes.

To link the activation of I𝜅β with the lack of synaptic pruning, we turned to previous reports
that connected inflammation to synaptic pruning. We found that dysregulated Glycogen
Synthase Kinase 3 (GSK3) links both synaptic pruning impairments and inflammation.
Furthermore, GSK3s act as a signal amplifier and that Trem2 activation has been shown
to signal through the AKT/GSK3-β pathway to prevent microglial apoptosis. Since this
was described as a global Trem2-dependent phenomoneon, we decided to assess GSK3
𝛼 and β activity using immunoblotting in whole brain lysates (Figure 4.8). We did not find
any additional increase in GSK-3β activity in most of these transgenic mice (Figure 4.8A),
except between APP/PS1 and APP/PS1 Trem2
-/-
C1q
+/-
mice. While pGSK-3𝛼 activity can
be inversely related to pGSK-3β activity, we found that some groups showed an inverse
relationship when compared to GSK-3β, namely APP/PS1 Trem2
+/-
mice (Figure 4.8B).
To reconcile these differences, we took the ratio of GSK-3β/𝛼 (Figure 4.8C), which
showed an uncoupling of GSK-3β/𝛼 activity. These trending results indicate that GSK-3β
could have some impact on microglial survival. However, if we posit that C1q activation
signals through Trem2, then we should expect that compound deletion should have a
synergistic effect.

Working upstream of GSK3 activation, we turned to AKT (Figure 4.9). Using whole brain
lysates, we probed for phosphorylated pAKT and found an increase from WT to APP/PS1
mice. This result suggests that pAKT/pGSK activity may be increased in APP/PS1 mice.
113
When we probe for pAKT in the Trem2/C1q deficient mice, we found that the loss of
Trem2 mostly affects pAKT signaling, whereas the loss of C1q does not reduce pAKT
signaling as drastically. If the Trem2 and C1q converge on pAKT signaling pathways,
then compound loss of Trem2 and C1q would further reduce pAKT signaling. Thererfore,
the lack of a synergistic effect indicates that both Trem2 and C1q do not affect pAKT
signaling.

Since AKT signaling influences many downstream signaling partners such as the
Mitogen-activated protein kinase (MAPK) family, we decided to assess whether the
reduction in AKT impaired MAPK activity. There are two major MAPKs, p38 and p40/42
(Extracellular signal related kinase, ERK). Both have similar roles to amplify a response,
but their downstream responses lead to different outcomes. Therefore, we immunoblotted
for phosphorylated p38 and found that p-p38 was not greatly affected in these mice at 6
months of age (Figure 4.10), however C1q deficient animals are more affected by p38
phosphorylation. When we compare the response of p38 with pERK, we find that pERK
phosphorylation affected both genotypes independently, but compound deficiency in both
Trem2 and C1q greatly decreased pERK phosphorylation (Figure 4.10). When we
compare these responses to WT mice, we find that the compound deficient animals had
pERK phosphorylation levels akin to WT. In Trem2 deficient mice on the APP/PS1
background, we observed a 59.36% pERK1/2 decrease, whereas C1q deficient animals
pERK1/2 have a 61.35% decrease. Compound loss of Trem2 and C1q further reduces
pERK1/2 by 90.80%, signifying a synergistic relationship between the Trem2/C1q
deficiency and ERK1/2 phosphorylation. Taken together, the stepwise decrease in pERK
114
signaling leading up to a synergistic affect in the compound deficient mice, indicates that
both these pathways converge upon pERK signaling in our cerebral amyloidosis mouse
model.
Establishing the Complement impact

While most of this chapter has focused on the impact of downstream of Trem2, another
facet of this Trem2-C1q interaction that has not been characterized is the downstream
complement activation. As described earlier, the complement system is activated via the
classical pathway, mannose-binding-lectin, or alternative activation. Regardless of the
activation system, these three pathways converge upon C3 and the production of C3
convertase. Therefore, we performed an ELISA for activated C3 (Figure 4.12) Comparing
between WT and APP/PS1 mice, we found that C3 activity is increased in APP/PS1 mice.
We next asked whether C3 levels were the same in APP/PS1 mice and we found that C3
remain mostly unchanged, while Trem2 deficient animals showed a slight reduction. The
most notable result is the increase in C3 activity in APP/PS1 C1q
-/-
mice, which correlates
with studies showing that other Complement pathways were activated to compensate for
C1q deficiency. Taken together, these results suggest that the interaction of Trem2 and/or
C1q remain largely upstream of C3 activation.
Mouse Behavior

Previous data has focused on the mechanism downstream of Trem2 and C1q activity.
While studies indicate that the loss of TREM2 or C1q drive AD onset by observing post
mortem LOAD brains, we wanted to assess whether the loss of Trem2 or C1q would
affect mouse behavior at 6 months of age. APP/PS1 mice are reported to have behavioral
115
impairments at 6 months. We assessed these mice for their anxiety-like behavior (Figure
4.13), hyperactivity (Figure 4.14-15), aggression-like behavior (Figure 4.16), episodic
memory (Figure 4.17), and short-term memory (Figure 4.18-19).

To characterize anxiety-like behavior, we utilized the open field maze assay. Under
aversive illumination, we monitored mice for 30 min and tracked their location, distance
traveled, and velocity. When tracking location, we found that APP/PS1 exhibited more
thigmotaxic behavior compared to their wildtype controls (Figure 4.13). However, when
we compare this behavior across all animals with their respective controls, we did not find
any significance, except for a few trending comparisons from APP/PS1 to APP/PS1 C1q
-
/-
and APP/PS1 to APP Trem2
-/-
C1q
-/-
. When we tracked the distance traveled by these
mice (Figure 4.14), we found that APP/PS1 mice traveled more than their wildtype
controls. Furthermore, we observed that APP/PS1 C1q deficient animals traveled more
than their genetic controls, suggesting hyperactivity in these mice could be C1q
dependent. When we assessed how fast these mice moved in the open field (Figure 4.15),
we found that 6-month-old APP/PS1 C1q
-/-
mice could be inherently more hyperactive
than their controls. The last measurement we recorded using the open field paradigm is
rotational behavior (Figure 4.16). Studies have claimed that rotational behavior suggests
potential aggressive-like behavior(Stein and Filosa, 1964). Although we found some
potential trends in these mice, APP/PS1 C1q
-/-
mice are the only ones that exhibited
significance in rotational (circling) behavior. These initial open-field dependent results
show that there are no significant differences between Trem2 and C1q deficient mice.
116

We next wanted to characterize the episodic and short-term memory in these mice. To
characterize episodic memory (Figure 4.17), we used the 5-min Y-maze paradigm that
counts the number of consecutive arm entries that mice undergo as they navigate through
the maze. APP/PS1 mice performed equally well when compared to their wildtype
controls. Furthermore, across all mice, there were no differences in spontaneous
alternation, suggesting that Trem2 deficiency and C1q deficiency does not affect episodic
memory in 6-month-old mice. To assess short-term memory (Figure 4.18-19), we utilized
the novel-object paradigm. During the training phase, none of these genetic lines
familiarize themselves with two similar objects. We find that during this phase, neither
mouse, when compared to their respective controls, explores the mouse more than
another (Figure 4.18), suggesting that the mice do not have a preference for one object
over another. After 20 minutes, one of the two objects are replaced with a novel object.
Murine instinctual behavior is to familiarize themselves with the more novel object. When
we assess their short-term memory (Figure 4.19), we find that WT mice are able to
perform this task better than APP/PS1 mice. Among all the other genotypes, we find that
all mice do not perform better than random chance. These results indicate that Trem2
and C1q deficient mice inherently have short-term memory impairments at 6 months of
age that are not amyloid dependent.
Human LOAD

For most of this chapter, we have characterized how the loss of Trem2 and/or C1q
affected signaling and behavior in APP/PS1 mice. Next, we wanted to correlate our
finding in mice to Human LOAD patients. We found an increase in TREM2 and C1q
117
expression (Figure 4.20A). Furthermore, when we correlate TREM2 or C1q protein
densitometry to neuropathological severity, Braak score, we find that these genes
correlate with the disease severity (Figure 4.20B). Since the LOAD data mirror the mouse
data, we continued to assess downstream signaling of Trem2 and C1q, in particular the
AKT/ERK/p38 axis (Figure 4.21). We found that LOAD AKT expression levels were
reduced when compared to cognitively normal controls. When we assessed pERK and p-
p38 expression, we found no difference in MAPK activity.
Discussion:
Reconciling Trem2 and C1q deficiency in the APP/PS1 Mouse model

In comparison to single knock out studies, characterizing double knockouts are not
common. In this chapter, we started characterizing gliosis, downstream Trem2-
dependent signaling pathways, mouse behavior, and LOAD patients. The majority of
these initial experiments are dependent upon immunohistochemistry, immunoblotting,
ELISAs, and mouse behavior, there are still many limitations to these experiments. One
limitation is getting to a cellular resolution because these data were utilizing whole brain
protein lysates. While comparisons could have been made to cerebellum, a region
associated with the lack of amyloid pathology, the amount of time and resources used to
characterize both regions, cerebrum and cerebellum, may be beyond the scope of this
dissertation.

Our initial finding used APP/PS1 mice deficient in Trem2 and/or C1q, which showed an
inverse relationship between astrogliosis and microgliosis in the cingulate cortex. While
118
these differences may not be enough to drive worsen or improve behavioral outcomes in
these mice, the loss of Trem2 or C1q seems to impact cellular function and
immunoproteostasis at the earlier stages of the disease. Whether Trem2 and/or C1q
deficiency leads to a behavioral impairment later in the disease, significantly older mice
would be required to characterize a future impairment.

In relation to other Trem2 and C1q studies in mouse models of AD, our data agreed with
previous studies on single transgenic knockouts(Fonseca et al., 2004; Wang et al.,
2015b), we can reconcile this phenomenon by comparing the differences that are inherent
in the cytoarchecture and regional brain function(Grabert et al., 2016). Evidence from the
Allen Brain atlas illustrates that within the different brain regions Trem2 expression varies
greatly between 2-month-old wildtype mice (Figure 4.22), while C1q does not. Since the
interaction and signaling between Trem2 and C1q is dependent on whether Trem2 is
expressed on the cell surface, the amount of Trem2 expression should determine the
strength of the microglial response(Kleinberger et al., 2014; Schlepckow et al., 2017).
While the data generated from the Allen Brain Alas only provides gross transcript
expression, what we would need is cell type specific expression to resolve which cells
have the Trem2 transcript and engage in Aβ clearance, thereby reducing
neuroinflammation.

Regarding Trem2 compensatory changes, we find that Trem2 and C1q gene expression
is greatly increased in APP/PS1 mice, yet as we remove Trem2 or C1q, one allele at a
time, we find that Trem2 protein expression peaks in animals without C1q, suggesting a
119
potential feed forward response, corroborating previous regulatory signaling between
Trem2 and C1q(Sharif et al., 2014a). However, more research is needed to demonstrate
a response loop exists in AD, outside of the lung(Sharif et al., 2014a). What will make this
difficult to study in the brain is the binding competition that will occur among C1q
opsonizing agents, especially in the case of mouse models of AD and other
neurodegenerative disorders(Severance et al., 2014; Sekar et al., 2016). C1q can mark
synapses for clearance, initiate efferocytosis, or create a C1q pathogen shield, or engage
in Aβ phagocytosis(Boulanger and Shatz, 2004; Stevens et al., 2007; Stephan et al.,
2012; Severance et al., 2014). Thus, we chose an earlier time point from most AD papers
because were interested in how these two molecules would interact with one another, at
a time point between the end of neurodevelopment (P45) and before the start of
neurodegeneration(9M)(Halima et al., 2007; Wirths and Bayer, 2010).

Aside from difficulties associated with studying the competition that exists among all C1q
opsonizing agents, the downstream signaling pathways associated with downstream
Trem2 activation are unusual. Unlike some receptors that Trem2 depends on DAP12 and
pSyk to activate other downstream targets. Syk can halt all Trem2 downstream activation
if Syk is not present at the surface of the cell membrane, and over activation can lead to
the formation of stress granules(Mócsai et al., 2010; Ghosh and Geahlen, 2015). When
Syk is functioning properly, it will activate other signal amplifiers(Mócsai et al., 2010;
Kremer, 2011). GSK-3β is one of these signal amplifiers that lead to a host of responses
such as amplifying NF-𝜅B responses or survival. While AKT has the possibility of
phosphorylating GSK-3β, it can also induce calcium signaling and alter the
120
immunometabolism of mononuclear phagocytes(Guha and Mackman, 2002; Covarrubias
et al., 2015; Lin et al., 2016). Since AKT has the potential to interact with p38 MAPK and
pERK(Singh et al., 1999a; McGuire et al., 2013; Lee et al., 2018b), identifying which one
these is the likely target the intersection of Trem2 and C1q will be a question of interest
in a later chapter.

In this study, one of our biggest concerns in our paper is the potential for downstream
C1q activity. If there were any Complement related activity, downstream of C1q activation,
we would have to account for alternative factors impacting the Trem2-pAKT-pERK
axis(Singh et al., 1999a; ZHANG and LIU, 2002; McGuire et al., 2013; Zhao et al., 2018).
Although APP/PS1 C1q
-/-
mice have an increase in activated C3, whether it affects the
physical interaction of Trem2 and C1q is not known. However, from the pERK
experiments (Figure 4.11), we find that there’s no difference between Trem2 deficient
mice and C1q deficient mice, suggesting any the involvement of C3 activation may not
affect this pathway in vivo. Although the APP/PS1 Trem2
+/-
C1q
+/-
mice showed an
increase in C3 activity, this suggests downstream complement activation and pERK
activation(Zhou et al., 2008; Merle et al., 2015; Shi et al., 2017). However, we do not see
any additional pERK activation in this genotype, suggestion that downstream C3
activation, may not influence pERK signaling.

While mice exhibit a host of pathological hallmarks ranging from microgliosis, synaptic
impairments, and dysregulated Trem2 signal transduction, we did not observe any
behavioral deficits among all our animals. This result indicates that the pathological
121
changes that are identified in these past few chapters arise before any cognitive changes.
Although these mice could have long term memory impairments, these paradigms were
not tested on these mice. This would have been a significant finding that would resolve a
diagnostic marker for the onset of TREM2-C1q dependent cognitive changes.

Reconciling Trem2 and C1q dysregulation in Human LOAD

These data generated from the human experiments illustrates how the upregulation of
Trem2 and C1q mirror the phenomenon seen human patients. This corroborates the
findings from many studies that suggest the potential co-regulation of TREM2 and C1q
in GWAS and LOAD transcriptomic network studies. The novelty in these data are that
we found a potential correlation between disease severity (Braak score) and the
expression of TREM2 or C1q in post-mortem LOAD patients. Whether this is a
phenomenon is common among LOAD patients, this experiment would require a larger
sample size. One aspect that I did not consider was the potential for different
expression patterns for TREM2 or C1q among patients in relation to their pathology.
Regardless, the number of patients used is unable to answer these questions and
requires a more diverse set of patients.

The microenvironment within each brain region in relation to the pathological severity
highlights the need to resolve the underlying heterogeneity that exists in each
microenvironment in the brain or the chronological time point for each signaling marker
that is activated during the disease(Sun, 2003). In other words, targeting TREM2 on all
monocytes may not be wise because TREM2 may be necessary to elicit an immune
122
response to clear Aβ. Removing TREM2 from the equation may block all coordinated
responses to clearing Aβ and maintaining cerebral immune homeostasis. Therefore, this
may be one potential reason for why targeting TREM2 is not wise. From our mouse data,
we found that the loss of Trem2 or C1q affected many signaling pathways. There may be
a signaling pathway that is pathological and Trem2 dependent, but it remains
uncharacterized.

Form our human data, we have found that reduction of pAKT in the brains of AD patients,
but this doesn’t hold for pERK or p38. To reconcile this result, we can speculate that the
lack of significance is due to inflammaging, cell specificity, or the terminal state of LOAD.
Since MAPKs integrate the signal relays and directs the cell towards a particular function,
the rationale for inflammaging hinges upon how background inflammation during the
aging process is already higher than “normal.” Therefore, measuring the real response
signal against the aging noise, would make it difficult to respond. Furthermore, these
samples were lysed from whole human brains. This impacts the cellular resolution
because neurons and other glia express MAPKs. Lastly, the patient brains are the
terminal result of LOAD patients and it represents a snapshot of the signaling pathways
at the end of the disease. Since these human data shows that that TREM2 and C1q
expression correlates with disease pathology and previous reports showed that Trem2
and C1q respond via feedback(Sharif et al., 2014b), then the pathology seen in humans
represents an overcompensation of the pathways that are not normal classical response
to a pathogen (Aβ). I.E., if 6 month-old-APP/PS1 mouse pathology mirrors a 50-60-year-
old late onset disease patient in this hypothetical example, then at this time point, we
123
could show that pERK signaling in TREM2 microglia is responding to C1q to sustain an
immune response to clear Aβ in AD. While isolating human patient brains at 50-60 years
of age is an unethical experimental design, we can counter this using computational
means to find a spectrum of activation states that will be described in Chapter 9. Current
means in immunology and neuroscience, or biology as a whole, are that we are limited
by the number of genes to correlate to a given spatial region. This is the Heisenberg-like
property in biology. Chapter 9 will explain this concept in more detail as it pertains to
TREM2-C1q in AD.

124

Figure 4. 1: Increased GFAP and Iba 1 expression in 6 months APP/PS1: Glial Fibrillary
Acidic Protein (GFAP) and Ionized calcium-Binding Adapter (Iba1) were histologically stained in
C57BL/6J (WT, blue) mice and APP/PS1 (red) mice. (n=5) Statistical annotations are compared
to APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

125

Figure 4. 2: Reduced GFAP and increased Iba 1 expression in APP/PS1 Trem2 and/or
C1q deficient mice.: Glial Fibrillary Acidic Protein (GFAP) and Ionized calcium-Binding Adapter
(Iba1) were histologically stained in APP/PS1 mice with/without Trem2 and/or C1q (n=4-8) Two-
tailed ANOVA was performed and statistical annotations are compared to APP/PS1: † p<0.1; *
p<0.05 ; ** p<0.01; *** p<0.001.
126

Figure 4. 3: Increased Trem2 and C1q gene expression in APP/PS1 mouse brains:
Cerebral whole brain lysate from C57BL/6J (WT, blue) mice and APP/PS1 (red) mice were
stained with Trem2 (ab, Colona) and C1q (ab, Tenner) (n=5) Immunoblot quantifications are
normalized to APP/PS1 and mRNA levels are compared to β-Actin. Statistical annotations are
compared to APP/PS1, † p<0.1; * p<.05.

127

Figure 4. 4: APP/PS1 mice deficient in Trem2 and/or C1q alter Trem2 and C1a protein
expression: Cerebral whole brain lysate from APP/PS1 mice with/without Trem2 and/or C1q
(n=3-5). Trem2 and C1q Immunoblot quantifications are compared to APP/PS1. Two-tailed
ANOVA was performed and statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05
; ** p<0.01; *** p<0.001.

128

Figure 4. 5: APP/PS1 mice deficient in Trem2 and/or C1q may not alter Cleaved Caspase
3 levels: Cerebral whole brain lysate from APP/PS1 mice with/without Trem2 and/or C1q (n=2-
3). Cleaved caspase immunoblot quantifications are compared to APP/PS1. Two-tailed ANOVA
was performed and statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05 ; **
p<0.01; *** p<0.001.

129

Figure 4. 6: APP/PS1 mice deficient in Trem2 and/or C1q increase Synaptophysin levels:
Cerebral whole brain lysate from APP/PS1 mice with/without Trem2 and/or C1q (n=3-5).
Synaptophysin immunoblot quantifications are compared to APP/PS1. Two-tailed ANOVA was
performed and statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01;
*** p<0.001.

130

Figure 4. 7: APP/PS1 mice deficient in Trem2 and/or C1q may impact NF-κB expression:
Cerebral whole brain lysate from APP/PS1 mice with/without Trem2 and/or C1q (n=3-5). Indirect
measurements were made on NF-κB activity using Ikβ protein immunoblot quantifications, which
are compare to APP/PS1. Two-tailed ANOVA was performed and statistical annotations are
compared to APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

131

Figure 4. 8: APP/PS1 mice deficient in Trem2 and/or C1q may modulate GSK-3 activity:
Cerebral whole brain lysate from APP/PS1 mice with/without Trem2 and/or C1q (n=3-5). GSK-
3α and GSK-3β gene immunoblot quantifications are compared to APP/PS1. GSK-3β
normalized to GSK-3α show differences in Trem2 and C1q deficient animals. Two-tailed
ANOVA was performed and statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05
; ** p<0.01; *** p<0.001.
132

Figure 4. 9: APP/PS1 mice deficient in Trem2 and/or C1q mostly reduces pAKT activity:
FACS immuno-stained beads (CST) from cerebrum whole brain lysates from APP/PS1 mice
with/without Trem2 and/or C1q (n=3-5). % pAKT cell quantifications were compared to
APP/PS1. Two-tailed ANOVA was performed and statistical annotations are compared to
APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

133

Figure 4. 10: APP/PS1 mice deficient in Trem2 and/or C1q may modestly affects p38
signaling: Cerebral whole brain lysate from APP/PS1 mice with/without Trem2 and/or C1q
(n=3-5). Phospho p38 immunoblot quantifications are compared to APP/PS1. Two-tailed
ANOVA was performed and statistical annotations are compared to APP/PS1: † p<0.1; * p<0.05
; ** p<0.01; *** p<0.001.

134

Figure 4. 11: APP/PS1 mice deficient in Trem2 and/or C1q mostly reduces pERK activity:
FACS immuno-stained beads (CST) from cerebrum whole brain lysates from APP/PS1 mice
with/without Trem2 and/or C1q (n=3-5). % pERKcell quantifications were compared to
APP/PS1. Two-tailed ANOVA was performed and statistical annotations are compared to
APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.

135

Figure 4. 12: APP/PS1 mice deficient in Trem2 and/or C1q mostly reduces C3: Activated
C3 ELISA results were generated using cerebrum whole brain lysates from APP/PS1 mice
with/without Trem2 and/or C1q (n=3-5). C3 quantifications were normalized to brain mass. Two-
tailed ANOVA was performed and statistical annotations are compared to APP/PS1: † p<0.1; *
p<0.05 ; ** p<0.01; *** p<0.001.

136

Figure 4. 13: APP/PS1 mice deficient in Trem2 and/or C1q show some Anxiety like
behavior impairment: APP/PS1 mice with/without Trem2 and/or C1q (n=6-17) with their AD
non-transgenic controls were tested for thigmotaxic behavior in the open field maze. Two-tailed
ANOVA was performed and statistical annotations are compared to non-transgenic control (top),
B6 (bottom left) and APP/PS1 (bottom right): † p<0.1; * p<0.05.
137

Figure 4. 14: APP/PS1 mice deficient in Trem2 and/or C1q show some hyperactivity
(distance): APP/PS1 mice with/without Trem2 and/or C1q (n=6-17) with their AD non-
transgenic controls were tested hyperactivity by measuring total distance traveled in the open
field maze. Two-tailed ANOVA was performed, and statistical annotations are compared to non-
transgenic control (top), B6 (bottom left) and APP/PS1 (bottom right): † p<0.1; * p<0.05.

138

Figure 4. 15: APP/PS1 mice deficient in Trem2 and/or C1q show some hyperactivity
(velocity): APP/PS1 mice with/without Trem2 and/or C1q (n=6-17) with their AD non-transgenic
controls were tested hyperactivity by measuring how distance traveled/second in the open field
maze. Two-tailed ANOVA was performed, and statistical annotations are compared to non-
transgenic control (top), B6 (bottom left) and APP/PS1 (bottom right): † p<0.1; * p<0.05.

139

Figure 4. 16: APP/PS1 mice deficient in Trem2 and/or C1q show no potential aggressive-
like behavior: APP/PS1 mice with/without Trem2 and/or C1q (n=6-17) with their AD non-
transgenic controls were tested for rotational movements from the Y-maze paradigm. Two-tailed
ANOVA was performed and statistical annotations are compared to non-transgenic control (top),
B6 (bottom left) and APP/PS1 (bottom right): † p<0.1; * p<0.05.
140

Figure 4. 17: APP/PS1 mice deficient in Trem2 and/or C1q show no impairments in
working memory: APP/PS1 mice with/without Trem2 and/or C1q (n=6-17) with their AD non-
transgenic controls were tested for working memory using the Y-maze paradigm. Two-tailed
ANOVA was performed and statistical annotations are compared to non-transgenic control (top),
B6 (bottom left) and APP/PS1 (bottom right): † p<0.1; * p<0.05.
141

Figure 4. 18: APP/PS1 mice deficient in Trem2 and/or C1q show no differences in during
NOR training: APP/PS1 mice with/without Trem2 and/or C1q (n=6-17) with their AD non-
transgenic controls were first trained prior to memory recall in the NOR paradigm. Two-tailed
ANOVA was performed and statistical annotations are compared to non-transgenic control (top),
B6 (bottom left) and APP/PS1 (bottom right): † p<0.1; * p<0.05.

142

Figure 4. 19: APP/PS1 mice deficient in Trem2 and/or C1q show no differences in during
NOR recall: APP/PS1 mice with/without Trem2 and/or C1q (n=6-17) with their AD non-
transgenic controls were first trained prior to memory recall in the NOR paradigm. Two-tailed
ANOVA was performed, and statistical annotations are compared to non-transgenic control
(top), B6 (bottom left) and APP/PS1 (bottom right): † p<0.1; * p<0.05.

143

Figure 4. 20: TREM2 and C1q protein expression LOAD brains: (Top Left) Blot showing
human TREM2 (2B5) and C1q (DAKO) expression from age matched, non-demented controls
(n=10) and LOAD brains (n=10). (Top Right) Quantification of immunoblot with normalized
values. Statistical comparison are vs age-matched controls. Statistical annotations are
compared to APP/PS1. * p<.05. (Bottom) Densitometry values were plotted against Braak
scores, with Pearson’s R
2
below.

144

Figure 4. 21: TREM2 and C1q protein expression LOAD brains: (Left) pAKT, (Middle)
pERK, and (Right) p-p38 activity from age matched, non-demented controls (n=10) and LOAD
brains (n=10). pAKT and pERK are from FACS beads quantifications (% phosphorylation from
parent gate) and p38 is from immunoblot (normalized to β actin). Statistical comparison are vs
age-matched controls,* p<.05.

145

Figure 4. 22: Adult mouse brains have even C1q expression throughout the brain,
whereas shows Trem2 expression: Representative micrographs were taken from the Allen
brain atlas showing no major regional differences in C1q expression(A), whereas Trem2 shows
differences between the CC, HC, and EC (B).
146
Chapter 5: Microglial Provenance

“ 故善用兵者，譬如率然；率然者，常山之蛇也，擊其首，則尾至，擊其尾，則首至，擊
其中，則首尾俱至。”

“The skillful tactician may be likened to the shuairan. Now the shuairan is a snake that is
found in the Ch’ang mountains. Strike at its head, and you will be attacked by its tail;
strike at its tail, and you will be attacked by its head; strike it at its middle, and you will
be attacked by head and tail both. “
Chapter 11 The nine situations
Art of War, Sun Tzu

Introduction:
Microglial origins:

Since the discovery of microglia by Pio de Rio Hortega, microglia were cells that were
often ignored because they were not the main cells in the brain, but rather part of the
supporting cells in the brain(Ginhoux and Prinz, 2015; Tremblay et al., 2015; Sierra et al.,
2016). Later these cells were found to regulate neuronal activity and shape the neuronal
arborization during development, health, aging, and disease(Stevens et al., 2007; Aguzzi
et al., 2013; Bialas and Stevens, 2013; Greter and Merad, 2013; Tremblay et al., 2015).
While every cell type in the brain are derived from the same embryonic pool, microglia
are mesodermal in origin(Amit et al., 2015; Goldmann et al., 2016). As the brain develops
during the early embryonic time points, neuronal progenitors divide, differentiate, and
migrate; yet, microglia are still not present. By embryonic day 9.5, the blood brain barrier
is formed and seals off the brain from the outside world(Amit et al., 2015). By this time
point, microglia are scattered across the neuropil, activated and ready to engage(Amit et
al., 2015; Ginhoux and Prinz, 2015).

147
Previous evidence found that microglia enter the brain at an earlier time point, embryonic
day 7.5 and stop at embryonic day 9.5(Ginhoux and Prinz, 2015). Furthermore studies
found that microglia came from a fetal liver population and they express a monocyte-
specific gene, CX3CR1(Jung et al., 2000; Fuhrmann et al., 2010; Goldmann et al., 2016).
Using a florescent tracer (GFP) to identify cells expressing CX3CR1, they traced from the
fetal liver to their final organ (Figure 5.1)(Kierdorf et al., 2013; Amit et al., 2015; Goldmann
et al., 2016). They found that these progenitor cells enter all organs, but they mature into
different macrophage populations in a tissue micro-environment-dependent fashion. For
example, cells from the fetal liver enter the lung and express a unique transcription factor
PPAR-ɣ to become alveolar macrophages(Amit et al., 2015; Yu et al., 2017). Meanwhile
in the skin, these cells enter dermal layers, and upregulate Ahr or RunX3 and mature to
become Langerhans cells. Whereas, in the brain, these CX3CR1
+
cells express
transcription factors Irf8 and Sall1 become microglia (Figure 5.1)(Amit et al., 2015;
Buttgereit et al., 2016; Matcovitch-Natan et al., 2016). However, even though these
common monocyte progenitor (CMP) cells exit the fetal liver to become mature tissue-
specific macrophages, these cells were either pre-programmed to migrate into specific
tissues or arbitrarily migrate into tissues to mature was still controversial in the field.

Microglial development: Pre-programmed or the Tissue-Microenvironment-
dependent

Several studies identified specific transcription factors that were lineage-specific in
different immune cells(Saijo and Glass, 2011). These combinatorial patterns are thought
to drive the expression of pattern of surface receptors that created tissue-specific
leukocytes (Figure 5.2, left). However alternative theory of leukocyte maturation was the
148
idea of prototype macrophages or CMPs that roam the blood and eventually reside in
tissues (Figure 5.2, right)(Amit et al., 2015; Matcovitch-Natan et al., 2016). Once these
macrophages enter the tissue, the specific microenvironment will drive a specific
transcription factor in these tissue-invading macrophages to make them into fully tissue
resident macrophages(Amit et al., 2015).

This hypothesis was backed by the idea that a specific myeloid progenitor would have a
transcription factor that is found prior to entry of the tissue. When this myeloid progenitor
enters the spleen, the hemoglobin would be sensed by the myeloid progenitor and drive
the programming for a splenic macrophage (Figure 4.3)(Amit et al., 2015). Whereas in
the peritoneum, the retinoic acid rich micro-environment would prime these myeloid
progenitors to upregulate transcription factor, GATA-6, to drive the peritoneal
macrophage transcriptomic profile(Wang and Kubes, 2016).

In my dissertation, I am most interested in the microglial origins, survival, and function in
Alzheimer’s disease. Several independent studies have tried to characterize microglial
origins. This chapter will trace the origin of these cells in vivo regarding the intersection
of TREM2 and C1q in Alzheimer’s disease. Furthermore, this chapter will first interrogate
the cellular heterogeneity of TREM2 expressing cells in post mortem human tissue and
assess the Trem2 heterogeneity in the cerebral amyloidosis mouse model that I use in
my dissertation. I will also show that this effect is not just found in our mouse model, but
also found in a rat model for Alzheimer’s disease-like pathology expressing APPswe and
Presenilin 1𝚫E9 (herein known as TgF344-AD rats)(Cohen et al., 2013). Lastly, I will show
149
the provenance of TREM2 expressing cells that also interact with C1q, specifically in
hematogenous macrophages. Overall, this chapter will examine one of the most
controversial concepts in neuroimmunology– microglial provenance.

Results:
CNS-infiltrating hematogenous macrophages express Trem2

Since TREM2 has been hypothesized to facilitate Aβ phagocytosis and microglia are the
primary phagocytes in the brain that express TREM2, we assessed TREM2 and IBA1
expression using confocal microcopy in LOAD patient brains (Figure 5.4). We found that
TREM2 did not localize on all IBA1 cells, showing a heterogeneous pattern; we see that
TREM2 colocalizes on some IBA1
+
cells, while some IBA1
+
cells do not express TREM2
(Figure 5.4A). Additionally, in cognitively aged matched LOAD patients we see some cells
with colocalized TREM2 and IBA1; however, this may be fewer in number(Figure 5.4B).

To further interrogate this mononuclear phagocyte heterogeneity on TREM2 expressing
brain phagocytes, we injected Coumarin(C6)-labeled nanoparticles (NP-C) into the tail
vein of 6-month APP/PS1 mice (Figure 5.5). Previous studies indicate that peripheral
macrophages engulf these nanoparticles, enter the CNS while maintaining an intact blood
brain barrier(Ragheb et al., 2013; Strohbehn et al., 2014). We assess whether peripheral
macrophages can engulf NPs compared to their mononuclear cousin, microglia (Figure
5.6 A,B). To determine if they enter the CNS, Xenogen IVISTM imaging of murine brains
showed that these nanoparticles enter the CNS (Figure 5.6C). After 3 weeks of weekly
injections, mice were sacrificed, and isolated brains were (1) fixed for
150
immunohistochemistry, (2) sorted using flow cytometry for CD45
+
, C6

expression, and (3)
RNA was extracted from these C6 populations (Figure 5.5).

Histological analysis of Iba1
+
and Trem2
+
expression in the brain APP/PS1 mice show a
heterogeneous pattern, mirroring the LOAD patient brains (Figure 5.7A). We also found
that Iba1
+
cells contained NP-C
+
puncta and Trem2
+
cells were also NP-C
+
(Figure 5.7A).
When merged, we find that some cells express both Trem2 and Iba1 with NP-C inside
them, suggesting that these Iba1
+
, Trem2
+
, and NP-C
+
cells are hematogenous in origin
(Figure 5.7A). Using q3Dism(Guillot-Sestier et al., 2016), a voxel based quantification
method, we find that the cells that are Iba1
+
and NP-C
+
are mostly expressing Trem2,
where cells only expressing Iba1
+
, express very little Trem2 (Figure 5.7B).
Establishing microglial provenance

To further address the provenance of these Trem2 expressing cells, we sorted CD45
int
and CD45
hi
populations using flow cytometry(Becher et al., 2014; Greter et al., 2015;
Mrdjen et al., 2018a) and found that Trem2 is predominantly expressed within the CD45
int

population (Figure 5.8A-C), suggesting that Trem2 is microglial in origin. Interestingly,
when we gated on all CD45
+
and further gated cells on C6
+
fluorescence (Figure 5.8D),
found that the cells that were more C6
-
(cyan) were situated in the very low population
that is CD45
int
, unlike the CD45
hi
(green) population that consists of all microglia (Figure
5.8E-F). From these CD45
+
, C6
+
cells, we isolated mRNA and found that the C6
-
fraction
expressed bona fide microglial markers (Figure 5.8G), which consist of Sall1, Maf-B, Irf8,
and MyB(Glass et al., 2010; Saijo and Glass, 2011; Buttgereit et al., 2016; Matcovitch-
151
Natan et al., 2016; Mrdjen et al., 2018a). These results suggest that cells isolated from
the C6
-
fraction are microglial in origin compared to those that are C6
+
, which are
macrophages of hematogenous origin. To determine whether these C6
+
cells were
macrophage in origin, we found that these cells are F4/80
+
and MHCII
+
, suggesting that
these are from the periphery (Figure 5.8H). Assessing other macrophage markers, we
found that neither of these populations expressed more or less CD11c or Ly6C (Figure
5.8I). We have also assessed the developmental transcriptional factor, PU.1, however
the mRNA transcript for either population was not detected.

Within the C6
+
and C6
-
fraction, we found that Trem2 is more likely to be expressed on
C6
+
cells (Figure 5.9). Interestingly, we find that relative Trem2 mRNA levels between
these two populations are the same, suggesting that Trem2 expression is differentially
regulated between bona fide microglia and macrophages (Figure 5.9). We then asked
whether these infiltrating peripheral macrophages expressed C1q and found that C1q
mRNA is predominantly expressed by infiltrating peripheral macrophages (Figure 5.9).
Previous reports have indicated that microglia can enter a neurodgenerative state
(MGnD) that is Apoe dependent(Krasemann et al., 2017), of which we find are expressed
mostly by C6
+
macrophages. While some reports indicate that these C6
+
cells should self-
proliferative and come from a Nestin
+
pool(Elmore et al., 2014a), we found that both cell
populations express CsfR1, while Nestin was undetectable. Recent reports indicate that
IL-10 are expressed by microglia and come from the periphery to induce innate
immunological memory(Guillot-Sestier et al., 2015a; Wendeln et al., 2018), we find that
Suppressor of cytokine signaling (SOCS) is unchanged between populations, but IL-10
152
is predominantly expressed in the C6
+
population. These changes IL-10 dependent
changes in immunological memory are thought to alter energetic and cytokine
signaling(Wendeln et al., 2018), however, we find no changes in IL-6, Diglyceride
acetyltransferase (DGAT) and Lactic acid dehydrogenase (LDH) signaling. Since C1q
has been reported to enhance homing related responses(Jalili et al., 2010), we assessed
whether there would be increases in CD45
+
cells in 6-month WT, APP/PS1, APP/PS1
Trem2
-/-
, and APP/PS1 C1q
-/-
mice, we found no differences in the percentage of
infiltrating brain CD45
+
cells in the brain (Figure 5.10).
Paradigm shift– Peripheral macrophages

Since Trem2 and C1q are expressed by infiltrating peripheral macrophages, we asked
whether hematogenous peripheral macrophages could interact with C1q. We isolated
hematogenous peripheral macrophages from 6-month old mice, treated with C1q
opsonized Aβ (Figure 5.11). We found that triple co-localization of Trem2-C1q-Aβ,
forming this novel neuroimmune complex. In our Aβ treated and C1q heat-inactivated
opsonized Aβ treated controls, we do not see the formation of the complex, rather the
Trem2-Aβ interaction. In summary, these results demonstrate that peripheral,
hematogenous macrophages enter the brain and predominately express Trem2. Upon
brain entry, these cells adopt a microglial-like morphology and engaging C1q opsonized
Aβ phagocytosis, of which could be Trem2-dependent.

To corroborate these experiments performed in APP/PS1 mice, we also utilized the rat
model of Alzheimer’s disease, TgF344-AD, which expresses human APPswe and PSN1
delta E9 mutation under the prion promoter(Cohen et al., 2013). From our workflow
153
(Figure 5.12A), we injected 10-month-old rats for 6 months in the footpad with these
nanoparticles. At 16months, we isolated brains for immunohistochemistry. Looking into
the hippocampal region after 6 months of NP-C injections, we found a similar result as
APP/PS1 mice (Figure 5.12B). Both experiments found NP-C
+
, Trem2
+
, and Iba1
+
, cells.
However, when quantified, the results indicate that over 90% of the Trem2
+
cells were
NP-C
+
and Iba1
+
, compared to less than 5% found in Iba1
+
alone. While these results
could indicate differences in rodent models of AD, these experiments suggest that
peripheral macrophages migrate into the CNS to repopulate the CNS & sustain
neurological function.
Discussion:
Identifying adult microglial ontogeny

The data presented in this chapter inserts itself into one of the most hotly debated topics
in the field of neuroimmunology, the provenance of microglia. The results of this chapter
show that these PGL-A coated nanoparticles are negatively charged and are specific for
macrophages, not microglia. They enter the brain while maintaining an intact blood-brain-
barrier. As they enter the TGF-β microenvironment of the brain, they adopt a microglial
morphology by upregulating Iba1 and reducing their CD45 expression from high to
intermediate (Figure 5.8)(Butovsky et al., 2013; Hickman et al., 2013; Buttgereit et al.,
2016). We characterized these cells by looking at the transcriptomic profile using genes
that mark the transition from embryonic/young to juvenile/adult microglia MAF-B, IRF-8,
and Sall1 (Figure 5.8-9) (Buttgereit et al., 2016; Matcovitch-Natan et al., 2016).

154
These results suggest that the microglia shown in this chapter are more likely brain-
resident macrophages that adopt a microglial-like morphology. This statement asserts
that the provenance-specific connotation associated with using the term “adult microglia”
is contradictory because this term assumes that the brain remains immune privileged and
stands against the overwhelming amount of evidence supporting immune cells surveilling
the brain(Schwartz et al., 1999; Becher et al., 2000, 2014; Kipnis et al., 2002; Carson et
al., 2006; Stein-Streilein and Caspi, 2014; Baruch et al., 2015a; Croxford et al., 2015).
Furthermore the argument of tissue specific microenvironment determines the cellular
function and maturation is fundamental concept in biology supporting this notion; it
especially holds in immunology where the tissue microenvironment impacts cancer
progression(Nussbaum et al., 2017), in tissue-specific immune cell development(Saijo
and Glass, 2011; Butovsky et al., 2013; Buttgereit et al., 2016; Yu et al., 2017), and in
neuroinflammation(Takahashi et al., 2005; Town et al., 2008; Dey et al., 2015; Yuan et
al., 2016). These findings in this chapter may seem novel, however macrophages
adopting microglial-like phenotypes builds upon the work showing peripheral
macrophages enter the brain in AD(Town et al., 2008; Gate et al., 2010; Jay et al., 2015),
a Trem2-dependent-activational-state shift(Keren-Shaul et al., 2017; Krasemann et al.,
2017; Ulland et al., 2017), and Trem2 impairing immunoproteostasis(Jiang et al., 2014,
2017a; Wang et al., 2015b; Yeh et al., 2016; Lee et al., 2018a).

Therefore, by accepting that the brain is not at as immune privileged as it once was, we
can learn to appreciate the heterogeneity that exists among all brain-resident immune
cells that carry out different functions during disease, aging, neuroinflammation, and
155
neurodegeneration(Mrdjen et al., 2018a). Therefore, to summarize this concept,
hematopoetic stem cells from the bone marrow, become myeloid precursor cells, which
would then enter the brain to become brain-resident macrophages with a microglial like
morphology (Figure 5.14).

It is important to note that the brain-infiltrating macrophages will never be the same as
microglia, nor would they. For example, these mature microglia are experienced, and
these experience-trained, mature microglia are marked more than just by telomere
shortening(Frenck et al., 1998; López-Otín et al., 2013; Pawlas et al., 2015), “newer”
brain-infiltrating macrophages have different methylation patterns(Sen et al., 2008;
Burchfield et al., 2015), and a different set of environmental factors that educate their
baseline to develop a homeostatic function(Ben-Shaanan et al., 2016; Piirainen et al.,
2017; Wendeln et al., 2018). For example, we can speculate that these new brain-
infiltrating macrophages could have the higher Tmem17 or P2ry12 (homeostatic
microglial genes) because from the perspective of these macrophages, the newer
environment is categorized as baseline in an APP/PS1 brain. As an APP/PS1 brain ages
and becomes more proinflammatory, these once newer macrophages become trained,
reduce their homeostatic gene expression as a new wave of macrophages enter the
brain. This introduces a new perspective on microglial homeostatic genes, rather these
more closely resemble microglial allostatic genes.

Supporting this speculation for these Tmem17 and P2ry12 brain-resident mononuclear
phagocytes, Butovsky and colleagues found that when TGF-β signaling was ablated in
156
microglia, P2ry12 staining was reduced(Butovsky et al., 2013). This loss in a homeostatic
gene corroborates aging epigenetic data and where immune activation returns to a
homeostatic set point. However, a counter ideology to immune homeostasis is the idea
that the homeostatic equilibrium will also be more activated with each response, resetting
the cell to a new allostatic set point(Tomiyama et al., 2012). Unfortunately, this chapter
does not address immune allostasis, however it is addressed in Chapter 7.

While we identified a self-proliferative Nestin
+
pool of microglia(Elmore et al., 2014b), our
results suggests that these results are all Nestin negative and that there are not any early
myeloid progenitors.(Glass et al., 2010; Saijo and Glass, 2011; Elmore et al., 2014a);
however, whether this macrophage population is described as self-proliferative, our data
Our results corroborate the results by Huang and Colleagues in that there is not a Nestin
+

pool, however, our results differ in that they did not find any peripheral infiltrates and that
they found transient-Nestin expression in vivo(Wendeln et al., 2018). They assessed
whether peripheral infiltrates entered from the periphery using a GFP-WT parabiosis
model and did not find any GFP+ cells in the brain using flow cytometry. Furthermore,
Huang’s study focused mainly on steady-state brains in non-disease-based models, while
our model is in an APP/PS1 mouse model. Furthermore, none of these studies
investigated whether leukocyte diapedesis, marked by ICAM, VCAM, or Vimentin, occurs
in the brain. Although not identifying diapedesis markers is a limitation to this dissertation
work, the most parsimonious explanation for leukocyte entry into the CNS, while
maintaining an intact blood brain barrier, is that these cells are peripheral in origin.
157
Chapter 9 will also address the ontogeny of these brain-resident macrophages in the brain
using imaging cytometric time of flight, imCyTOF.

158

Figure 5.1: Microglial origins: Schematic diagram that describes microglial origins from fetal
liver to myeloid progenitor cell, just before entry into the tissue of residence. Transcription
factors indicated in these figures show the transcription factors necessary to gain the necessary
microglial phenotype, alveolar phenotype, or skin macrophage phenotype.

159

Figure 5.2: Pre-programmed Migration vs Microenvironmental driven maturation:
Schematic diagram summarizes the two opposing theories behind leukocyte development from
progenitor to mature leukocyte. Abbreviations are as follows: Hemopoietic Stem Cell (HSC),
Common Myeloid Progenitor (CMP), Common Lymphocyte Progenitor (CLP), Natural Killer
(NK), Granulocyte-myeloid progenitor (GMP), Megakaryocytic-erythroid progenitor (MEP), Red
Blodo Cell (RBC), Macrophage (Mϕ), Monocyte (Mo), Granulocyte (Gn).

160

Figure 5.3: Transcriptional changes depend on microenvironmental changes
macrophages: Schematic diagram illustrates how a prototype macrophage can respond to
tissue specific stimuli to initiate the transcriptional profile to create splenic macrophages,
peritoneal macrophages, or microglia. For example, the prototype macrophage senses the
Heme complex in the extracellular space and initiates the transcriptional programming pathways
to develop into mature splenic macrophages.

161

Figure 5.4: Human LOAD brains suggest microglial heterogeneity: Human post mortem
LOAD (A, B) and cognitively age matched brains (C) are histologically stained with TREM2 and
Iba1. B shows an Imaris bitplane rendered image of A. Scale bars represent 100microns.
Arrows represent overlay of TREM2 and Iba1.

162

Figure 5.5: Workflow of the following experiment: APP/PS1 mice were injected with NP-C
for 3 weeks prior to sacrifice. Brains were isolated for IHC and flow cytometry with subsequent
RNA isolation.

163

Figure 5.6: NP-C are macrophage specific and enter the brain: (A) Microglia and (B)
peripheral macrophages were isolated from 6-month-old mice. They were treated with
aggregated Aβ-Cy3 and NP-C for 2 hours. Scale bar represents 3µM. (C) Xenogen IVISTM
imaging of treated rodent brains show entry of NP-C into the brain.

164

Figure 5.7: peripheral macrophages express microglia and contain NP-C
+
vesicles: (A)
Immunohistochemistry Staining of in the hippocampus of 6-month-old APP/PS1 mice show the
triple coloclization of Trem2
+
, Iba1
+
, and NP-C
+
cells. 3D Imaris renderings as insets. Bars
represent 10µM. (B) Using Imaris Bitplane, % volume of Trem2 was quantified in double positive
cells using q3Dism. Statistical test used, Students T-test, ** p<.01.

165

Figure 5.8: NP
+
leukocytes are brain resident macrophages: (A) The lymphocyte population
gated from forward and side scatter, followed by a single cell gate. (B) Cells were then either
gated for analysis (C, pink and orange) or sorting (D, purple). Cells were analyzed by drawing
gates around CD45
int
or CD45
hi
populations, each were measured against Trem2 expression.
(D) Sorted cells are all taken from the CD45
+
gate (larger drawn polygonal gate) and further
sorted upon C6 expression (Green vs Cyan). (E) From these gates, ancestry of these gates
were plotted on a CD45 by FSC scatterplot to determine where these C
+
and C
-
populations
originated from. (F) These same gates were plotted against CD11b by FSC. (G)RNA isolated
from C
+
and C
-
gates were assessed for bona fide microglial markers and (H) macrophage
markers. (I) Other monocyte markers were also used to characterize these populations.
Statistics student t-test, † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.
166

Figure 5.9: NP
+
brain resident macrophages express Trem2 and C1q: (A)Cells from CD45+,
C+ or C- gates were assessed for Trem2 expression. (B) Trem2 and C1q mRNA expression
were measured from C+ and C- fractions. (C) Apoe, (D) Csf-R1, NEstin, (E) SOCS, IL-10, (F)
Il6, DGAT, and LDH mRNA levels were also measured using C
+
and C
-
fractions. Statistics
student t-test, † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001. (ND, not detectable)

167

Figure 5.10: APP/PS1 animals Leukocyte numbers do not change among APP/PS1 mice
deficient in Trem2 or C1q: (A)Cells gated upon CD45 show no changes in cell influx numbers
at 6months between B6, APP/PS1, APP/PS1 Trem2
-/-
and APP/PS1 C1q
-/-
mice. N=3 per
genotype. Two-way ANOVA followed by post hoc students T-test was performed. Statistical
annotations are compared to APP/PS1: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001.\

168

Figure 5.11: Peripheral macrophages can engage in Trem2 mediated C1q-opsonized Aβ
phagocytosis: 6-month-old peripheral macrophages were treated C1q-opsonized Aβ, Aβ, heat
inactivated C1q opsonized Aβ, C1q opsonized Aβ and C1q alone. Representative micrographs
show the raw confocal image, followed by Imaris bitplane images and magnifications of insert
areas to show complex Trem2-C1q-Aβ interaction in vitro. Scale bars represent 2µM.
169

Figure 5.12: Trem2 deficiency impairs Trem2-C1q-Aβ complex formation: 6-month-old
peripheral macrophages from WT and Trem2
-/-
mice treated with C1q opsonized Aβ. Scale bar
represents 2µm. Colors are indicated below the representative Imaris Bitplane generated
images.

170

Figure 5.13: TgF344 AD Rats injected with NP show NP
+
, Iba1
+
, Trem2
+
macrophages: (A)
TgF344-AD rats were given weekly footpad injections at 10 months of age for 6 months. Upon
sacrifice, Rats brains were isolated for IHC according to workflow. (B) IHC of the HC shows NP
+

iba1
+
brain invading macrophages are mostly Trem2
+
.

171

Figure 5.14: Summary of microglial provenance: Hemopoietic stem cells from the bone
marrow exit the bone marrow to become myeloid progenitors and peripheral macrophages.
These peripheral macrophages pick up nanoparticles in the periphery and enter the CNS. In the
CNS microenvironment, they adopt a microglial like phenotype.

172
Chapter 6: Regulating Trem2 Phagocytosis

Phagocytosis
“ 故善戰者，求之于勢，不責于人，故能擇人任勢。”

“The clever combatant looks to the effect of the combined energy and does not require
too much from individuals. Hence his ability to pick out the right men and utilize
combined energy.”
Chapter 5, Energy
Art of War, Sun Tzu

Introduction

Chapter 4 showed differences in signaling pathways downstream of Trem2 activation in
APP/PS1 mice. While we saw differences in pAKT, we found that pERK1/2 signaling is
the most impaired in mice deficient in Trem2 and/or C1q. Additionally pERK signaling
followed a synergistic drop in phosphorylation, which is the main motivation for tackling
pERK signaling. The introduction will be broken down in two ways. First, I will discuss
the mechanisms of phagocytosis and what this process requires. Second, I will detail
the motivation for pERK signaling in C1q opsonize Aβ phagocytosis.

Phagocytosis

Phagocytosis is a cellular process where the word is derived from 3 Greek terms
“phago-” for destroy/eat, “-cyto-” for cells, and “-sis” for process/state. The combination of
these three terms means the process for eating and destroying cells. This is not to be
confused with pinocytosis, where pino, also Greek, means “drink.” While there are many
types of endocytic pathways, phagocytosis in relation the immunobiology inside
173
macrophages can execute innate or adaptive immune functions. Figure 6.1 illustrates the
general cascade (1-5), and in relation to this dissertation, time points 5-8 depicts
immunoproteostasis in action.

During rest, macrophages (Mϕ) are more ramified (1) and as it recognizes a pathogen via
its receptors, (2) the Mϕ adapts a more ameboid morphology to initiate phagocytosis of
the pathogen, (3) forming a phagosome. Upon phagosome formation, NADPH oxidase
incorporates into the phagosome membrane and pumps Hydrogen ions, dropping the pH
down to 6.0 (4)(Geisow et al., 1981; Stuart and Ezekowitz, 2005; Fairn and Grinstein,
2012; Richards and Endres, 2014). Upon fusion of the lysosome, the phagosome
becomes a mature phagolysosome and the pH drops to ~4.0 (5)(Geisow et al., 1981). As
the pathogen degrades, at this point, the macrophage will either incorporate the
pathogen’s antigens into the antigen presenting complex for an adaptive immune
response, prepare for another round of phagocytosis, nutrient cycling, degranulate
(release cytokines), or release opsonizes (Complement)(Stuart and Ezekowitz, 2005;
Ebstein et al., 2013; Kimura et al., 2015).

To illustrate how the immunoproteasome can alter intracellular pathways and extracellular
pathogens, step 6 illustrates how the lysosome is ready to form another phagolysosome
and releases opsonins (C1q)(Ebstein et al., 2013; Sharif et al., 2014b; Merle et al., 2015;
Thielens et al., 2017). As a phagolysosome is formed to degrade a pathogen (7), C1q
binds the bacterium starting the Classical complement cascade to initiate cell lysis and
clear the extracellular space. As the Mϕ prepares for another round of phagocytosis (8),
174
it has cleared almost all pathogens by optimizing the immune tactics(Stuart and
Ezekowitz, 2005; Ebstein et al., 2013; Merle et al., 2015). These phagocytic mechanisms
are not limited to bacteria, but also dying cells and Aβ. In this chapter, phagosomes are
marked with CD68, Macrosialin, and phagolysosomes are marked by pHrodo. The
novelty in this chapter is that I will show data suggesting the unification of 2 independent
phagocytic pathways as one pathway.

Extracellular signal-Regulated Kinase (ERK)

Previous studies show that microglial and macrophage survival depends on Trem2
expression via p38 phosphorylation(Chue et al., 2004; Klesney-Tait et al., 2006; Turnbull
et al., 2006; Ulland et al., 2017; Zhong et al., 2017), whereas Trem2 expressing
macrophages activate pERK signaling (Chue et al., 2004; Wu et al., 2015b). Furthermore,
other studies also demonstrate that apoptotic clearance via Trem2 dependent
phagocytosis in macrophages requires ERK signaling(Takahashi et al., 2005) and that
C1q opsonizes apoptotic cells for clearance could depend on ERK1/2(Fraser et al., 2010;
Willmann et al., 2017). The purpose of this chapter is to interrogate whether peripherally
infiltrating peripheral macrophage will engage in pERK signaling upon Trem2 activation
by C1q.

Results:
Modeling phagocytosis

To model our in vivo observations in vitro, we first injected our 6-8-month-old mice with
thioglycolate to induce the maturation of peritoneal macrophages, of which are
175
hematogenous in origin (Figure 6.2). After 1 week, we isolated CD45
hi
, CD68
hi
, peripheral
macrophages and brain-resident mononuclear phagocytes from 6-8-month-old mice (as
described in the workflow Figure 6.1). Our gating strategy are to isolate all lymphocytes
and isolate CD45
+
, CD68
+
cells, which are described as peripheral macrophages, similar
to the Trem2 expressing cells shown in vivo (Figure 6.3)(Town et al., 2008; Jay et al.,
2015; Ulland et al., 2017). Isolated Trem2
-/-
peripheral macrophages treated with vehicle,
Aβ and C1q opsonized Aβ all showed decreased ERK1/2 phosphorylation compared to
treated wildtype peripheral macrophages (Figure 6.4, gating strategy 6.3). These in vitro
results phenocopy the in vivo results, therefore verifying the validity of our model.

We questioned whether modulating ERK1/2 phosphorylation levels would impair C1q
opsonized Aβ phagocytosis in wildtype peripheral macrophages. Because the interaction
of Trem2 and C1q could be offsetting the ligand-receptor equilibrium, thereby affecting
immunoproteostasis, we needed to capture first peak of Aβ phagocytosis in peripheral
macrophages. We performed an Aβ phagocytosis kinetics experiment by measuring the
amount of Aβ inside phagolysosomes using confocal microscopy (0.5hr, 1 hr, 2,hr, 3 hr,
4 hr and 5 hr.; Figure 6.5) and found that macrophage phagocytosis is a cyclical process
with peaks and troughs. From these data, we time point for these phagocytosis
experiments were 1.5 hours. To pharmacologically modulate ERK1/2 phosphorylation,
we used an established pERK1/2 inhibitor U0126 (Chue et al., 2004). U0126 treatment
on peripheral macrophages inhibited ERK1/2 phosphorylation, even when the peripheral
macrophages were treated with Aβ or C1q opsonized Aβ(Park-Min et al., 2009; Zhong et
al., 2017) (Figure 6.6).
176

Since U0126 can modulate ERK1/2 phosphorylation levels in wildtype peripheral
macrophages, we wanted next asked whether C1q opsonized Aβ encapsulated within
phagolysosomes. To test this experiment, we utilized LysoTraker and pHrodo dyes, both
of which are pH sensitive fluorescent probes. These probes increase in fluorescence
when the pH drops below 7.0 and plateaus around 2.0; therefore, CD68+ vesicles with
Aβ or C1q should contain the pH sensitive dye, signifying the development of a
phagolysosome. When we cultured peripheral macrophages with the pH-sensitive dye
(pHrodo)(Figure 6.7), we found that C1q opsonized Aβ is found within pHrodo
+
, CD68+
phagolysosomes. When we add U0126, pERK1/2 inhibitor, we found the loss of pHrodo
vesicle staining and C1q-Aβ engulfment. These results suggest that ERK signaling
regulates C1q-Aβ phagocytosis.

Previous reports found that Trem2 deficiency reduces Aβ phagocytosis. We asked
whether Trem2 deficiency also impaired C1q opsonized Aβ phagocytosis. Therefore, we
isolated and cultured peripheral macrophages from wildtype and Trem2
-/-
mice.
Afterwards, we treated them C1q opsonized Aβ and Aβ, with or without U0126 (Figure
6.8). Using q3DISM to quantify the amount of C1q opsonized Aβ in phagolysosomes, we
found that WT peripheral macrophages volumetrically engulf C1q opsonized Aβ up to
45.36%. However, with the addition of U0126, C1q opsonized Aβ phagocytosis is reduced
to only 3.95%. In Trem2 deficient peripheral macrophages, we find that C1q opsonized
Aβ only minimally localized inside phagolysosomes (4.20%). With U0126 treatment,
Trem2
-/-
peripheral macrophages also do not phagocytose C1q opsonized Aβ, 0.00%.
177
Comparing the Aβ paradigm, Aβ specific phagocytosis is largely unaffected by U1026
inhibition in WT and Trem2
-/-
peripheral macrophages (Figure 6.8 bottom, right). In
Chapter 5, we found that peripheral macrophages enter the brain and adopt a microglial
like morphology. To test this phenomenon, we also isolated brain resident macrophages,
microglia, at 6-8 months of age from wildtype and Trem2
-/-
mice (Figure 6.9). We found
that these brain-macrophages, microglia, mirror the same phenotype in peripheral
macrophages. Taken together, these results indicate that Trem2 dependent phagocytosis
of C1q opsonized Aβ signals through pERK1/2, whereas phagocytosis of Aβ alone does
not. Furthermore, C1q opsonized Aβ phagocytosis is not cell type specific phenomenon,
but rather an effect that occurs in mononuclear phagocytes.
Trem2 signaling bifurcation

To demonstrate that C1q opsonized Aβ signals through Trem2-pSyk-pERK, we used
several drug inhibitors. To differentiate pERK1/2 signaling from other MAPK family
members, we used a p38 and pERK1/2 selective inhibitor (SB220025), a p38–specific
phosphorylation inhibitor (SML05430 affecting only p38 α/β phosphorylation), and a pSyk
selective inhibitor (SYKVi) (Figure 6.10). Images of cultured macrophages (Figure 6.11)
and microglia (5.14) show differences in C1q-opsonized Aβ phagocytosis. When we
assess C1q opsonized Aβ phagocytosis in these microglia, we observed that pSyk
inhibition reduced C1q opsonized Aβ phagocytosis (Figure 6.12A). When we use a pan-
MAPK inhibitor, we find an reduction in C1q-Aβ phagocytosis; however, with a p38 α/β
inhibitor, we do not find a reduction in Aβ phagocytosis (Figure 6.12B). When we isolate
peripheral macrophages from C1q
-/-
mice and treat them with C1q opsonized Aβ, we find
that in DMSO treated wildtype controls compared to C1q
-/-
macrophages, C1q-Aβ volume
178
in phagolysosomes decreases significantly. Whereas, in drug treated C1q-/-
macrophages, we find that there’s no change in C1q-Aβ volume within phagolysosomes
when we compare wildtype controls to C1q
-/-
macrophages. When we observe the
peripheral macrophages treated with Aβ, we find that Syk inhibition reduces Aβ
phagocytosis (Figure 6.13). With our pan-MAPK inhibitor (SB220025), we find a decrease
in Aβ phagocytosis, whereas p38 α/β selective inhibition greatly reduces increases Aβ
phagocytosis. When we treat C1q
-/-
with Aβ, we observed a decrease in Aβ phagocytosis
between wildtype controls and C1q
-/-
peripheral macrophages, whereas no significant
differences were observed in the other drug treatments. These initial results in peripheral
macrophages illustrates that C1q opsonized Aβ phagocytosis is Trem2 dependent and
signals through the pSyk-pERK axis.

Microglial C1q opsonized Aβ phagocytosis

In Chapter 4, we made the distinct difference between peripheral macrophages and brain
resident macrophages, microglia. Therefore, these experiments were replicated in brain
resident macrophages (Figure 6.14). Micrographs for C1q opsonized Aβ treatments in
brain resident macrophages show similar differences as observed in peripheral
macrophages (Figure 6.14A). When we quantify the C1q-Aβ volume within
phagolysosomes, we find that these results phenocopy the peripheral macrophage
results in (Figure 6.12). Whereas in our Aβ phagocytosis paradigm, we find an increase
in Aβ phagocytosis in all drug treatments except for U0126, the pERK1/2 inhibitor. These
results indicate that brain-resident macrophages can engage in C1q opsonized Aβ
179
phagocytosis like peripheral macrophages, however, Aβ phagocytosis pathways may not
be Trem2 dependent.

Impact of AKT on C1q opsonized Aβ phagocytosis

Since several Trem2 studies indicate that Trem2-dependent cell survival and proliferation
requires AKT signaling, we explored whether AKT is involved C1q opsonized Aβ
phagocytosis and Aβ phagocytosis alone (Figure 6.15). We used 2 se inhibitors for AKT,
A6730 (AKT1/2/3 inhibitor) and GSK (GSK690693, pan-AKT inhibitor, PI3K inhibitor, and
Gsk-3β inhibitor). On our first glance, we found that the selective AKT inhibitor (A6730)
increases Aβ phagocytosis but not C1q opsonized Aβ phagocytosis. Furthermore, GSK
inhibition slightly reduces C1q-Aβ phagocytosis in peripheral macrophages and induces
cell death brain-resident macrophages (microglia). While AKT inhibition increase Aβ
phagocytosis in both cell types, whether AKT is necessary or indirectly affecting C1q-Aβ
or Aβ phagocytosis remains inconclusive.

Breaking tolerance– rescuing Trem2 phagocytosis impairments

Adopting the nanotechnology used in Chapter 4, we used nano-particle encapsulated
with TGF-β selective inhibitor, SB505124. We isolated macrophages and treated these
macrophages with NP-BL (blank), NP-C (Coumarin), and NP-SB (Coumarin + SB50512).
When we inhibit one of these major cytokines in the brain that maintains microglial-like
morphology(Town et al., 2008; Butovsky et al., 2013; Buttgereit et al., 2016), using this
nanoparticle technology, we find an increase in Aβ phagocytosis in WT macrophages
180
when co-treated with the TGF-β inhibitor(Figure 6.16). Previously, in Trem2 knockouts,
we found that a deletion of Trem2 impairs Aβ phagocytosis. However, when we treated
Trem2 deficient macrophages with Aβ and NP-SB, we found a restoration in alternative
phagocytic pathways that are Trem2-independent. These results suggest that TGF-β
inhibition reverses the phagocytic pathology seen in Trem2
-/-
macrophages.

Discussion:

This chapter solely focused on how C1q opsonized Aβ converges on Trem2 to engage in
Trem2-dependent phagocytosis. Previously, Trem2 has been shown to interact with
bacteria, dead cells, Aβ, APOE, and, in this chapter, C1q opsonized Aβ(Klesney-Tait et
al., 2006; Colonna et al., 2007; Wang et al., 2015b; Yeh et al., 2016). It is not clear why
and how Trem2 can have multiple ligands leading to differences in downstream signaling.
While one common denominator is the possibility for negatively charged ligands(Wang et
al., 2015b), whether Trem2 binding affinity is dependent on glycosylation is an area of
research that is not well understood. What is known is that Trem2 must be glycosylated
prior to cell surface expression, and any impairments in glycosylation will impair TREM2
function(Kleinberger et al., 2014; Kober et al., 2016; Park et al., 2017). One question that
remains is how different binding interactions could yield different glycosylation patterns.
Modifications in glycosylation are particularly sensitive to reactive oxygen species (ROS),
namely O-GlcNAcy-lation and N-glycosylation. In particular, nutrient excess pathways
(not nutrient sensing or mTOR-related pathways) are found to reduce the number of
proper glycosylation patterns on proteins, thereby reducing their function, leading to
metabolic diseases and cancer(Wellen and Thompson, 2010). Since TREM2 has been
181
associated with cellular energetics(Keren-Shaul et al., 2017; Ulland et al., 2017), the
correlation among glycosylation, nutrient excess, and immune function may actually be
more than just an association, but a causal relationship that drives a feedforward loop
driving the disease.

To summarize the findings of this chapter, we have identified a potential Trem2
dependent-signaling axis that affects macrophage activation in Alzheimer’s disease.
While this may seem novel, Trem2 has already been identified to regulate macrophage
polarization and dendritic cell maturation in vivo(Bouchon et al., 2000, 2001; Bachmann,
2002; Chue et al., 2004; Ito and Hamerman, 2012; Wu et al., 2015b; Zheng et al., 2016).
What is novel in this chapter is that we have identified a cell type specific activation state
driving C1q-Aβ phagocytosis or Aβ phagocytosis (Figure 6.17). In this chapter we have
seen how deletion of Trem2 reduces Aβ and C1q-Aβ phagocytosis. Furthermore, we have
also seen how selective inhibitors for the Trem2-Syk-ERK axis alter phagocytic states in
AD.

Another novel facet of this chapter is that we integrate pervious genomic studies that
cross examine the Trem2-Aβ axis in AD and mouse models of AD. Many studies have
found that Trem2 deficiency impairs microglial immunometabolism(Amit et al., 2015;
Keren-Shaul et al., 2017; Krasemann et al., 2017; Ulland et al., 2017; Lee et al., 2018a).
Furthermore, single-cell RNAseq studies 8 month old mouse models of AD from Keren-
Shaul and colleagues found that the impairments converge on the activation of
Lipoprotein lipase (Lpl) and a Trem2 dependent state restriction in microglia(Keren-Shaul
182
et al., 2017). They classified this stage as damage associated microglia (DAM), DAM1 vs
DAM2, with DAM2 being the latter stage and upregulation of Lpl. Experiments from Ulland
and colleagues corroborated upon these studies and found that this transitional stage is
dependent upon p38 in 8-month-old mouse models of AD, deficient in Trem2(Ulland et
al., 2015, 2017). While our studies don’t investigate Lpl activity in peripheral macrophages
treated with C1q opsonized Aβ or Aβ alone, previous studies have shown that
phosphorylation of p38 MAPK can lead a unilateral Lpl activation in diabetes(Singh et al.,
1999a; Kim et al., 2008; Gao et al., 2017), further supporting an energetics-centric view
of p38 MAPK signaling.

To reconcile the increase in p38 phagocytosis in our phagocytosis experiments, we can
take a competition centric view for the MAPK signaling. Previous reports have shown that
p38 and pERK have different functions even with common upstream signaling partners
(Koistinaho and Koistinaho, 2002; Sun, 2003; Chue et al., 2004; Lee et al., 2018b). In
relation to energetics and phagocytosis, whether there is a shift in MAPK signaling from
p38 to pERK isn’t well established in CNS-resident mononuclear phagocytes. Studies in
immortalized hepatocytes found an energetic shift in cellular function from p38-dependent
induction of Low density lipoprotein (LDL) receptor, which then licenses subsequent
pERK-dependent LDL receptor supported expression(Singh et al., 1999b). This energetic
shift is supported in a human immortalized macrophage cell line, where LDL supported
p38 activation and increased phagocytic genes(Mei et al., 2012).These two studies
indicate that p38 regulates more cholesterol/fatty acid energetics, while pERK could
modulates more phagocytic pathways.
183

Taking a step back, while some reports indicate that AKT is upstream of MAPK signaling,
our AKT inhibition results from Figure 6.15, show that blocking AKT activity results in
increased Aβ phagocytosis, suggesting a shunt in the AKT pathway to promote more
inflammatory, phagocytic pathways(Guha and Mackman, 2002; Vergadi et al., 2017). In
this particular case in peripheral macrophages, we observed an increase in Aβ
phagocytosis, which corroborates previous studies. These data also support previous
studies done that found that Trem2 activation is dependent on AKT for microglial survival
(Park et al., 2015; Zhao et al., 2018), in which we observe in Figure 6.15. Aside from Syk,
my dissertation doesn’t address upstream factors of AKT, therefore, whether any other
molecule upstream of AKT can interfere with AKT signaling remains unexplored.

Syk-AKT-MAPK axis

Previously described, Trem2 is a recent gene that shares close homology to natural killer
(NK) cell gene Nkp44 with gene orthology traced to teleost fish (Phylum Chordata Class
Actinopterygii)(Allcock et al., 2003; Stet et al., 2005), suggesting a potential homologous
function. While Trem2 expression is restricted to mononuclear phagocytes and Nkp44 is
restricted to NK cells, they both converge upon ITAMs (e.g. Dap12) as its intracellular
signaling partner(Lanier and Bakker, 2000; Allcock et al., 2003; Turnbull et al., 2006;
Ivashkiv, 2011; Ito and Hamerman, 2012). The Syk-dependent signaling pathways within
NK cells are well studied especially the class of Natural Cytotoxicity Triggering Receptors
(NCRs) including NKp44(Jiang et al., 2002; Allcock et al., 2003; Kruse et al., 2014). In NK
184
cells, NKp44 activation leads to Syk activation and the activation of the PI3K-AKT-MAPK
pathway. While NK cells may have a different signaling biology and the result in different
activation states, numerous studies show that the Syk-MAPK signaling axis propels the
cell to either target a pathogen or enter a more homeostatic profile(Trotta et al., 2000;
Kruse et al., 2014).

Whether this signaling axis requires a feedback or forward response to engage a
pathogen is an area that is not well understood and often overlooked. However, to
reconcile the drug inhibitory experiments (Figure 6.10) showing increased Syk
phosphorylation upon p38 α/β inhibition, evidence has shown that in p38 or Akt inhibition
in macrophages has led to increased Syk phosphorylation(McGuire et al., 2013). This
study suggests that that blocking downstream factors (p38 or Akt) could lead to an
increase in upstream factors (Syk) in macrophages, suggesting a potential feedback
stress response(Trotta et al., 2000; McGuire et al., 2013). Applying this concept to the
Trem2-C1q-Aβ signaling axis, whether Aβ or C1q opsonized Aβ potentiates a Trem2-
dependent energetic shift with p38 as its signaling nexus is currently an ongoing
investigation.

Furthermore, whether this Trem2-dependent energetic shift promotes the formation of
trained macrophages or tolerized macrophages is followed up in Chapter 7, Trem2-
dependent macrophage tolerance. Data from the nanoparticle experiments in Figure 6.16,
show that inhibiting TGF-β signaling in peripheral macrophages show an increase in Aβ
phagocytosis. Whether this is breaking through tolerance mechanisms or a compensatory
185
phagocytic mechanism is not known. However, Chapter 7 will go further into these data,
as well as Chapter 10, the discussion.

186

Figure 6.1: Phagocytosis and Immunoproteostasis: Phagocytosis is summarized in sections
1-5. Phagocytosis is shown in relation to immunoproteostasis in sections 5-8. (1) Resting Mϕ
sense pathogens and (II) engage in chemotaxis. (III) Upon phagosome formation, (IV) NADPH
oxidase inserts into the membrane to drop the pH. (V) When lysosomes must fuse to form a
phagolysosome to degrade the pathogen. (VI) As this cycle repeats, initiation of phagocytosis
will engage parallel pathways to release cytokines or opsoninogens (C1q). As the extracellular
pathway is reshaped by pathogen lysis from the opsins (VII-VIII), the macrophage can focus on
clearing up pathogens.

187

Figure 6.2: Phagocytosis paradigm workflow: 6-month-old mice were injected with
Thioglycolate and peripheral macrophages were isolated in parallel with microglia. After
overnight serum starvation, cells were treated with DMSO, Aβ, or C1q opsonized Aβ.

188

Figure 6.3: Peripheral macrophage gating strategy: Leukocytes were gated by forward and
side scatter, followed by CD45 and CD68+ cells. Phosphorylation of pERK and other signaling
partners were then gated by CD45 and pERK positivity.

F S C
S S C p E R K
C D 4 5
C D 6 8
C D 4 5
189

Figure 6.4: Trem2 deficiency reduces pERK activity: Peripheral macrophages treated with
DMOS, Aβ, or C1q opsonized Aβ exhibit reduced pERK activity in Trem2
-/-
macrophages. For
these groups, standard Students T-test was performed. Statistical annotations are as follows: †
p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001

190

Figure 6.5: Peripheral macrophage phagocytic capacity fluctuates: Peripheral macrophage
phagocytosis kinetics were assessed at 0min, 30min, 1hr, 2hr, 3hr, 4hr, and 5hrs.

0 2 4 6
0.00
0.05
0.10
0.15
0.20
0.25
pMACs
Time (h)
%Volume of Aβ
in Phagolysosomes
191

Figure 6.6: pERK inhibition dose curves: peripheral macrophages were treated for 1.5 hours
with increasing concentrations of U0126, pERK selective inhibitor 3 treatment conditions: PBS,
Aβ and C1q opsonized Aβ. Statistics two-way ANOVA with post-hoc students T-test. n=4.
Statistical annotations are as follows: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001

192

Figure 6.7: C1q opsonized Aβ phagolysosomal degradation depends on pERK: Treated
peripheral macrophages with C1q opsonized Aβ were imaged. pHrodo was found localized
inside CD68 phagolysosomes. This effect was lost in U0126 treated conditions.

193

Figure 6.8: pERK inhibition impairs C1q opsonized Aβ but not Aβ in peripheral
macrophages: Treated WT and Trem2
-/-
peripheral macrophages with C1q opsonized Aβ and
Aβ were imaged. Representative images are located above with confocal image, Bitplane Imaris
renderings, rotation and magnification. Scale bars represent 2µm. Volume of C1q opsonized Aβ
was measured using q3Dism (Bottom Left) and Aβ lone (Bottom Right). Quantified plots show
the plots from 3 experiments that show the inhibitor’s effect on each phosphorylated protein.
Statistics two-way ANOVA with post-hoc students T-test, n=10-15 cells/image. Statistical
annotations are as follows: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001

194

Figure 6.9: pERK inhibition impairs C1q opsonized Aβ but not Aβ in microglia: Treated
WT and Trem2
-/-
microglia with C1q opsonized Aβ and Aβ were imaged. Representative images
are located above with confocal image, Bitplane Imaris renderings, rotation and magnification.
Scale bars represent 2µm. Volume of C1q opsonized Aβ was measured using q3Dism.
Statistics two-way ANOVA with post-hoc students T-test, n=10-15 cells/image. Quantified plots
show the plots from 3 experiments that show the inhibitor’s effect on each phosphorylated
protein. Two-tailed ANOVA was performed, followed by a student T-Test. Statistical annotations
are as follows: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001
195

Figure 6.10: Inhibition drug control experiments: Peripheral macrophages were isolated and
gated on the strategy mentioned previously. Drug inhibitors were used to selectively inhibit
signaling genes downstream of Trem2: SykIV (pSyk inhibitor), SML0543 (p38 inhibitor), and
SB220025 (p38 and pERK inhibitor), and U0126 (pERK). Quantified plots show the plots from 3
experiments that show the inhibitor’s effect on each phosphorylated protein. Two-tailed ANOVA
was performed, followed by a student T-Test. Statistical annotations are as follows: † p<0.1; *
p<0.05 ; ** p<0.01; *** p<0.001
196

Figure 6.11: Images inhibiting the Syk-MAPK axis in peripheral macrophages: C1q
opsonized Aβ phagocytosis is inhibited by pERK1/2 drugs: Peripheral macrophages treated with
SykIV (pSyk inhibitor), SML0543 (p38 inhibitor), and SB220025 (p38 and pERK inhibitor) impact
C1q opsonized Aβ phagocytosis are rendered using Imaris. Scale Bar represents 3µm.

197

Figure 6.12: C1q-Aβ quantification of inhibitory phagocytic experiments on the Syk-
MAPK axis in peripheral macrophages: C1q opsonized Aβ phagocytosis is inhibited by
pERK1/2 drugs: Peripheral macrophages treated with SykIV (pSyk inhibitor), SML0543 (p38
inhibitor), and SB220025 (p38 and pERK inhibitor) impact C1q opsonized Aβ phagocytosis are
quantified using q3Dism on the Imaris. These experiments were performed on WT and C1q
-/-
(CKO) macrophages. Scale bar represents 3µm Quantified plots show the plots from 3
experiments that show the inhibitor’s effect on each phosphorylated protein. Two-tailed ANOVA
was performed, followed by a student T-Test. Statistical annotations are as follows: † p<0.1; *
p<0.05 ; ** p<0.01; *** p<0.001
198

Figure 6.13: Aβ quantification of inhibitory phagocytic experiments on the Syk-MAPK
axis in peripheral macrophages: Aβ phagocytosis is inhibited by pERK1/2 drugs: Peripheral
macrophages treated with SykIV (pSyk inhibitor), SML0543 (p38 inhibitor), and SB220025 (p38
and pERK inhibitor) impact Aβ phagocytosis are quantified using q3Dism on the Imaris. These
experiments were performed on WT and C1q
-/-
(CKO) macrophages. Scale bar represents 3µm.
Quantified plots show the plots from 3 experiments that show the inhibitor’s effect on each
phosphorylated protein. Two-tailed ANOVA was performed, followed by a student T-Test.
Statistical annotations are as follows: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001

199

Figure 6.14: Quantification of inhibitory phagocytic experiments on the Syk-MAPK axis in
microglia: C1q opsonized Aβ and Aβ phagocytosis are inhibited by pERK1/2 drugs: WT
microglia were treated with SykIV (pSyk inhibitor), SML0543 (p38 inhibitor), and SB220025 (p38
and pERK inhibitor) impact C1q opsonized Aβ and Aβ phagocytosis are shown in representative
micrographs (Left) and quantified using q3Dism on the Imaris (Right). These experiments were
performed on WT and C1q
-/-
(CKO) macrophages Scale bar represents 3µm. Quantified plots
show the plots from 3 experiments that show the inhibitor’s effect on each phosphorylated
protein. Two-tailed ANOVA was performed, followed by a student T-Test. Statistical annotations
are as follows: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001
200

Figure 6.15: Inhibitory phagocytic experiments on the Syk-AKT-MAPK axis in peripheral
macrophages and microglia: C1q opsonized Aβ or Aβ phagocytosis are inhibited by pERK1/2
drugs: WT peripheral macrophages and microglia were treated with SykIV (pSyk inhibitor),
SML0543 (p38 inhibitor), SB220025 (p38 and pERK inhibitor), A6730 (pAKT inhibitor), and GSK
(pAKT, GSK-β, PI3K inhibitor) impact C1q opsonized Aβ or phagocytosis are quantified using
q3Dism on the Imaris. These experiments were performed on WT and C1q
-/-
(CKO)
macrophages Scale Bar represents 3µm. Quantified graphs are from 3 experiments that show
the inhibitor’s effect on each phosphorylated protein. Two-tailed ANOVA was performed,
followed by a student T-Test. Statistical annotations are as follows: † p<0.1; * p<0.05 ; **
p<0.01; *** p<0.001

201

Figure 6.16: TGF-β inhibition rescues the Trem2 phagocytic impairment: Peripheral
macrophages were treated with Aβ and nanoparticles (NP) encapsulated nothing (NP-BL),
coumarin (NP-C), or coumarin+TGF-β inhibitor (SB50512, NP-SB). Micrographs (Left) are
representative images show the phagocytic capacity of Aβ loaded pHrodo vesicles inside p-
macrophages. Quatifications (Right) are for WT and Trem2
-/-
for all NP conditions with Aβ. Scale
bar represents 3µm. Two-tailed ANOVA was performed, followed by a student T-Test. Statistical
annotations are as follows: † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001

202

Figure 6.17: Trem2-C1q-Aβ signaling axis working hypothesis: Peripheral macrophage and
microglial experiments suggest that the Trem2-pSyk-pERK signaling axis regulates
phagocytosis 6 months of age.

203
Chapter 7: Innate Immune Tolerance

“ 故兵以詐立，以利動，以分合為變者也，故其疾如風，其徐如林，侵掠如火，不動如山
，難知如陰，動如雷霆。”

“In War, practice dissimulation and you will succeed. Move only if there is a real
advantage to be gained. Whether to concentrate or divide your troops, it must be
decided by circumstance. Let your rapidity be that of the wind, your compactness that of
the forest. In raiding and plundering be like fire, in immovability like a mountain. Let your
plans be dark and impenetrable as night, and when you move, fall like a thunderbolt. “
Chapter 7 Maneuvering
Art of War, Sun Tzu

Introduction:
Innate immune training and tolerance:

The overused clichéd phrase, “with great power, comes great responsibility,” defines
immune system’s role in disrupting or maintaining homeostasis. How the immune system
chooses to respond to a pathogen and to what degree is an area that is still an active
area of research. Especially when incorporating the network of other leukocytes into this
network dramatically escalates this complexity. In the adaptive immune system,
immunological memory is a novel evolutionary adaptation; however, innate immune
memory is highly conserved across species. Innate immune cells, predominantly those
of myeloid origin, are capable of learning when to attack, train, or tolerate a pathogen.
For example, the Clostridium tetani vaccine or more commonly known as the Tetanus
vaccine, is an example of innate immune training, while Gastrointestinal-bacterial
symbiosis is more associated with tolerance(Alvarez-Errico et al., 2015; Schneider and
Tate, 2016).

204
Since the latter two concepts, trained immunity and immune tolerance, are intimately tied
to one another, Figure 7.1 illustrates how a myeloid cell will show an immediate response
to a pathogen (Aβ or LPS)(Alvarez-Errico et al., 2015). The bifurcated response is post-
activation. Trained immunity returns to baseline and will elicit a stronger immune
response to the same pathogen; whereas, immune tolerance will not respond.
Understanding how this bifurcation occurs is still an active area of research. Recent
evidence has shown that immune tolerance can reduce the severity of a disease(Wendeln
et al., 2018).

While the terminology between what Wendeln and colleagues call immune tolerance and
training contradicts the classical terminology for macrophage tolerance and
training(Alvarez-Errico et al., 2015; Wendeln et al., 2018), the phenomenon is the same.
Some hypotheses include the epigenetic shifts, cellular aging, or microenvironmental
priming. However, as described in a previous chapter, TREM2 is an unusual receptor, in
that it sits at the nexus between cellular activation or maintaining innate immune
homeostasis. For this chapter, I will present the evidence that links TREM2 macrophage
expression to innate immune training and tolerance in AD.
Results:
Establishing an immune training and tolerance paradigm in vitro

Innate immune training and tolerance is experimentally simplified to 2 repeated
pathogenic hits. To understand whether TREM2 plays an integral role in tolerance, we
treated human microglia (SV40) for 3 hours with LPS (Lipopolysaccharide) and with Aβ
to determine whether TREM2 expression differed between stimuli. From my initial
205
preliminary experiments, we found that TREM2 protein expression increased with one
treatment of Aβ or LPS (Figure 7.2). To model tolerance, we decided to treat the human
microglia for 24 hours, followed by 3 hours. We found that after 26 hours, TREM2 levels
showed no change when compared to unstimulated cells, which mirrors the phenomena
seen in LPS treated microglia. This preliminary evidence hints at the potential for TREM2-
dependent tolerance effects because these microglia do not elicit a sustained innate
immune response. If these data showed an increase in TREM2 protein expression post
26 hours, these preliminary results would suggest trained innate immunity.

Next, we asked whether this phenomenon could be TREM2-dependent. Since TREM2
requires the scaffolding protein TYROBP, DAP12 and subsequent phosphorylation of
SYK to elicit any intracellular activity, we next tested whether SYK was phosphorylated in
this preliminary tolerance paradigm (Figure 7.3). Preliminary evidence at three hours
shows that Aβ treatment could drive more pSYK activity. When compared to this 26-hour
tolerance paradigm, we found that pSYK activity is abolished, the same as unstimulated
microglia. Albeit preliminary, Aβ phenocopies LPS to potentially induce tolerance.

Although these preliminary data hint that TREM2 may have immune tolerance-inducing
properties, we further established a paradigm with more controls that show the connection
between TREM2 and immune tolerance (Figure 7.4). This chronic stimulation setup
contains a high concentration of Aβ, followed by a low concentration of Aβ. Further
experiments in this chapter will utilize this specific tolerance-inducing paradigm that
206
controls for Aβ/C1q-Aβ treatment time and concentration in future biochemical
experiments (Chapter 8).

Chronic inflammatory settings induce tolerance in microglia.

With this new tolerance-inducing paradigm, we treated microglia for 24 hours and 12
hours to induce tolerance (Figure 7.5). Chronically stimulated microglia (B) expressed
less TREM2 protein than untreated microglia. This was significantly lower than Aβ
High
(C)
and Aβ
Low
(D) conditions. These results indicate that TREM2 protein expression and
function is reduced in a chronic stimulated paradigm. To ensure mRNA levels are
consistent with protein levels, mRNA was isolated 3 hours into the Aβ
Low
treatment. In
Figure 7.6a, we found that TREM2 mRNA was significantly increased all Aβ treated
paradigms (B, C, and D). However, in the double Aβ treatment, tolerance-inducing
paradigm (B), mRNA expression was significantly increased when compared to
untreated. While this contradicts the protein expression data (Figure 7.5), this suggests
the potential for post transcriptional regulation by mRNA or by other means. To show that
this is effect was temporally restricted to this immediate Aβ
Low
treatment, we isolated
mRNA from the 12-hour time point (Figure 7.6b). we found that TREM2 mRNA does not
change among unstimulated (A), the tolerance-inducing condition (B), and the single
Aβ
High
treatment condition (C). While the Aβ
Low
treatment (D) was the only condition that
reduced TREM2 mRNA expression, these results suggest that TREM2 mRNA expression
is capable of returning to the homeostatic baseline post-Aβ treatment (12 hours) and that
microglial Aβ tolerance could be regulated post transcriptionally by reducing TREM2
protein expression.
207

While measuring TREM2 expression explains one facet of this tolerance story, measuring
cytokine production will describe the microglial function in response to Aβ. To determine
if cytokines were being produced in these tolerance-induced settings, mRNA expression
of proinflammatory cytokine, Tumor Necrosis Factor-𝛼, TNF-𝛼, was quantified at 3 hours
(Figure 7.7A) and 12 hours (Figure 7.7B). At three hours, there was a significant reduction
in TNF-𝛼 mRNA expression among all Aβ treated samples. However, at 12 hours, this
was returned to the homeostatic baseline. These data indicate that TNF-𝛼 mRNA
expression is reduced in chronic Aβ stimulated microglia. On the other hand, Interleukin-
6, IL-6, another cytokine with pro-inflammatory and inflammatory properties, does not
phenocopy TNF-𝛼. IL-6 mRNA expression is reduced at three hours in the tolerance-
inducing condition (B) compared to the untreated control (Figure 7.7C). However, at 12
hours, these cells return IL-6 mRNA expression to the homeostatic baseline (Figure
7.7D). Taken together, these cytokine results indicate that chronic Aβ exposure induces
tolerance in microglia by reducing TREM2 expression and cytokine mRNA expression.
The reduction of TREM2 in this tolerance paradigm can greatly impair mononuclear
phagocyte activation, or more specifically, Aβ clearance.

While C1q and TREM2 are intimately interact and impact microglial and peripheral
macrophage microglial phagocytosis. We sought to test whether C1q expression changes
in tolerance-inducing situations at three hours and 12 hours after the Aβ
High
stimulation
(Figure 7.8). We found that C1q expression does not change across all paradigms and
time points, suggesting that C1q may not be linked to microglial tolerance.
208

Chronic inflammatory settings induce immune training in peripheral
macrophages

As previously described in Chapter 4, microglia are derived from the fetal liver, whereas
macrophages develop from the bone marrow. Microglia enter the brain and mature into
full-functioning microglia given the tissue-specific environmental cues in the brain.
Although this working hypothesis is highly controversial, monocyte-cell fate dependent
on the microenvironment is a common reoccurring theme found among many immune
cell types(Höglund and Brodin, 2010; Amit et al., 2015; Thomas, 2015; Wang et al.,
2016; Nussbaum et al., 2017; Yu et al., 2017; Mrdjen et al., 2018b). While microglia
have a unique function compared to other macrophages, they are derived from the
same embryonic progenitor cell pool, yet require the slightly different
microenvironmental cues to mature. This fundamental difference in their developmental
origin makes brain-invading peripheral macrophages similar, yet different from
microglia. Anything otherwise would not seem biologically parsimonious.

To determine whether peripheral macrophages phenocopy microglia in my tolerance
paradigm, we isolated peripheral macrophages and followed the same tolerance inducing
paradigm as previously described (Figure 7.4). Isolating peripheral macrophages after 12
hours, we quantified Trem2 protein expression and found that Trem2 is significantly
upregulated above all conditions (Figure 7.9). This difference is drastically different from
the microglial response in the previous section. Using this same paradigm, mRNA was
isolated after three hours (Figure 7.10) and Trem2 mRNA expression was significantly
209
increased in all Aβ treated conditions when compared to the untreated condition in
C57JBL/6 (B6) animals. Using Trem2 deficient animals, we find a step-wise decrease in
Trem2 mRNA expression (Figure 7.10). To determine whether proinflammatory cytokine
mRNA expression is increased we measured TNF-𝛼 expression and found that TNF-𝛼
produced a step-wise increase in mRNA fold expression as a copy of Trem2 was removed
(Figure 7.11A). Furthermore, while IL-6 production remains mostly unchanged in wildtype
animals (Figure 7.11B), IL1β mRNA levels modestly increase in wildtype animals (Figure
7.11C). Taken together, since Trem2 expression increased and proinflammatory mRNA
expression is mostly increased, these results suggest that peripheral macrophages do
not undergo innate immune tolerance, but rather innate immune training (Figure 7.12).

Characterizing the innate immune training response to Aβ in peripheral
macrophages

This cell-type specific response to Aβ further illustrates that microglia and brain-infiltrating,
microglial-like peripheral macrophages do not share the same properties. From Chapter
4, we show that these peripheral macrophages enter the brain. This previous data section
suggests that peripheral macrophages adopt an immune training-like response by
eliciting a more robust Aβ inflammatory response than microglia, which adopt a more
immune tolerance-like phenotype. Since microglia do not express C1q when in this
tolerance-inducing paradigm, we sought to ask whether these peripheral macrophages,
at baseline, require C1q during their innate immune training phase. We found that during
the immune training condition, C1q mRNA expression was slightly increased in Trem2
deficient cells compared to wildtype macrophages. These results suggest that C1q could
210
have a role in immune training to optimize trained microglia to engage in more efficient
phagocytosis of Aβ.

If trained macrophages produce more C1q to optimize their phagocytic capacity, one
question that remains is, does the microenvironment in this training condition produce the
necessary cytokines to support this behavior. Recent work from our lab has shown that
inhibition or deletion of two major cytokines, Transforming Growth Factor-β (TGF-β) or
Interleukin-10 (IL-10), increase monocyte phagocytosis. Thus, measuring IL-10 in Trem2
deficient these peripheral macrophages (Figure 7.13), IL-10 mRNA expression levels
drop in the training paradigm (B) compared to the single Aβ
Low
treatment (D). Whereas in
C1q deficient macrophages, IL-10 mRNA expression levels are reversed. Since neither
of these conditions produce more IL-10 mRNA transcript than the untreated paradigm(A),
the results still remain inconclusive as to whether Trem2-dependent macrophage training
impacts IL-10 production.

On the other hand, measuring TGF-β mRNA expression in the macrophage-training
paradigm (Figure 7.14A), we found that TGF-β mRNA expression is increased in these
trained, wildtype macrophages. Furthermore, Trem2 deficient and C1q deficient
macrophages show no difference between trained and untrained conditions. These
results suggest that TGF-β could play a role in maintaining macrophage Aβ-training. If
trained macrophages produce more TGF-β, we wanted to know whether these
macrophages could respond to TGF-β. Therefore, we measured TGF-βR1 mRNA
expression (Figure 7.14B) and found a decrease in TGF-βR1 mRNA expression in the
211
wildtype and Trem2 deficient macrophages. These results suggested that there could be
a cytokine-receptor communication occurring in these trained macrophages.

To determine if these trained macrophages utilize TGF-β/TGF-βR1 signaling to sustain
immune training or an Aβ-dependent immune response, we calculated the mRNA
expression ratio for these two cytokines. We hypothesized that if this ratio was high then
this should be a feed-forward signaling, whereas a smaller ratio could suggest feed-back
(Table 6.2.1). Comparing TGF-β/TGF-βR1 ratios, we found that trained wildtype
macrophages had a higher TGF-β/TGF-βR1 ratio (Figure 7.15A), suggesting that this
could be a feed-forward mechanism. However, to determine if this effect was Trem2-
dependent, we plotted these ratios against the respective Trem2 genotype (Figure 7.15B)
and took the first derivative of this regression (Figure 7.15C). This regression measures
the rate of this change with respect to the Trem2 gene (R
2
=.80, p=0.1). Taken together,
these results could suggest that macrophage training could be dependent on the
intersection of Trem2-TGF-β/TGF-βR signaling.

Trained macrophages are undoubtedly activated, but the term “activation” does not
capture the full meaning of how these trained macrophages are responding to Aβ.
Although the conventional term “alternatively activated” could characterize these
macrophages, this still doesn’t capture the macrophage’s function in this Aβ environment.
If immune cells are migrating, they will upregulate C-C motif chemokine ligand 2 (CCL2).
Therefore, we measured CCL2 mRNA expression in wildtype, Trem2
-/-
, and C1q
-/-

macrophages (Figure 7.16A). Only the Aβ
Low
treatment (D) induced CCL2 expression in
212
wildtype and Trem2 deficient macrophages. Whereas, CCL2 expression was reduced in
C1q deficient macrophages. These results implicate that CCL2 may be associated with
Trem2 and C1q upon Aβ
Low
treatment, but not during macrophage training.

Another method to characterize macrophage activation states is to use a metabolic gene,
such as Arginase1 (Arg1). This gene classically defines a metabolic switch in leukocytes,
implicating a non-cytotoxic pathogen clearance. In Figure 7.16B, we found that Arg1
mRNA expression is slightly reduced in trained macrophages (B), compared to the
untreated. This effect is also phenocopied in the single-dose, Aβ
High
treated cells (C), but
absent in the single-dose, Aβ
Low
condition (D). This effect is lost in the Trem2 and C1q
deficient macrophages (Figure 7.16B). These preliminary results suggest that there could
be a slight metabolic difference in these trained macrophages, but this is not the defining
characteristic.

In addition to characterizing metabolic changes and monocyte migration, we were
interested in the potential for macrophages to engage in antigen presentation. The
primary cell type that can engage in antigen presentation are dendritic cells, which more
commonly express CD11c. In this macrophage training paradigm, we found that CD11c
expression does not increase in the trained macrophage condition (Figure 7.16C).
Although C1q deficient macrophages slightly increases CD11c expression in the trained
paradigm, the potential for antigen presentation is still inconclusive. Furthermore, antigen
presentation requires costimulatory receptors as well as cytokines. Because cytokines
were not measured, we hypothesized that if antigen presentation should occur, then
213
another co-stimulatory receptor Major Histocompatibility Complex II (MHCII) should
increase in mRNA expression. we found that MHCII mRNA expression in our trained
macrophages does not change from the untreated condition (Figure 7.16D). These
preliminary results indicate that Aβ trained macrophages may not carry antigen
presentation like capabilities.

Assessing C1q-Aβ macrophage training paradigms

To further characterize these trained macrophages, we adopted flow cytometry in
wildtype mice. In addition to assessing the Aβ training paradigm, we created a novel
paradigm to assess peripheral macrophage training in vitro (Figure 7.17). Peripheral
macrophages were isolated and cultured in vitro as described earlier and were trained as
described above. We found no differences in the percentage of Trem2 macrophages.
However, we did find differences in the number of Trem2
hi
and Trem2
lo
macrophages
(Figure 7.18A). Since previous evidence suggests that Trem2 deficiency and C1q
promotes a feedforward mechanism(Sharif et al., 2014a), we assessed if C1q affected
the high and low populations. We found that the number of Trem
hi
and Treme
lo
cells were
reversed in the C1q
-/-
(Figure 7.18B) These results suggests that C1q endogenous
expression may impact the transcriptomics and macrophage priming.

To determine if the loss of C1q affected Trem2 downstream signaling between training
conditions, we found that in the WT conditions, C1q opsonized Aβ produced more Trem2
lo

macrophages to sustain training (Figure 7.19). However, in C1q deficient populations, this
214
effect was lost. When we assessed downstream signaling parameters of training, we
looked into pSyk and found that there was no difference between Trem2
hi
or Trem2
lo

populations in all conditions (Figure 7.20A). However, when we assessed pSyk in C1q
-/-
macrophages, we found that C1q opsonized Aβ sustained the pSyk phosphorylation in
the Trem2
lo
cells (Figure 7.20B). Furthermore, in these C1q-/- macrophages, we found
that low Aβ stimulation also promoted pSyk activation.

Since Chapter 6 found that Trem2-dependent phagocytosis of C1q-Aβ is dependent on
pERK, we assessed whether macrophage training was dependent on pERK (Figure
7.21). We found that in WT macrophages, the percentage of Trem2
hi
or Trem2
lo
cells do
not vary much among the different treatment paradigms. However, pERK activation varies
with C1q opsonized Aβ and low concentration Aβ modulates, suggesting that pERK could
be modulated by these ligands. In C1q deficient macrophages, there is a higher
percentage of Trem2
hi
macrophages than Trem2
lo
, however among the different
treatment conditions within each cell type remains unchanged. Therefore, whether pERK
plays a role in macrophage training remains inconclusive.

In comparison to pERK, we also assessed p38 phosphorylation in these cells and we
found that in the WT conditions (Figure 7.22A), we did not find differences between each
cell type populations and treatment condition. However, when we assessed p38
phosphorylation in C1q
-/-
macrophages, we found a mirror image of our results found in
our pSyk experiments. To determine whether this is truly Trem2 dependent, we assessed
pSyk, p38, and pERK in a Trem2
-/-
macrophages (Figure 7.23). We found that pSyk
215
phosphorylation is reduced in these macrophages treated with C1q opsonized Aβ (Figure
7.23A). While we did not observe any signaling in pERK in any of our samples, we found
that p38 was activated in these macrophages. However, in conditions that were treated
with C1q opsonized Aβ and low Aβ concentrations, we found a reduction in p38
phosphorylation(Figure 7.23B). Since results are an inverted phenocopy of C1q
-/-

macrophages, these experiments suggest that C1q opsonized Aβ training may potentially
signal through the pSyk-p38 axis in trained macrophages. Whether these macrophages
depend on C1q to prime their function by acting as a cytokine in macrophage training
remains unexplored.

Discussion:
Evolutionary advantage for immune tolerance in the brain

The adaptive immune system is not the only branch of the immune system that is capable
of immunological memory(Pham et al., 2007; Rodrigues et al., 2010; Yoshida et al., 2015;
Schneider and Tate, 2016; Wendeln et al., 2018). Trained innate immunity is an innate
immune response that is capable of providing the proper immune response to clear a
pathogen without collateral damage(Pham et al., 2007; Wendeln et al., 2018). It has been
conserved through evolution, In Caenorhabditis elegans, innate immunological memory
is known as immune conditioning(Kim and Mylonakis, 2012), while in Drosophila
melanogaster, it is known as innate immune priming(Pham et al., 2007; Christofi and
Apidianakis, 2013; Conteras-Garduño et al., 2016). This evolutionary conserved
mechanism is more studied in bacterial and fungal infections, however little is known in
216
the context of AD or using Aβ as the pathogen. In this chapter, I showed the how different
monocytes respond to Aβ and exhibit innate immune tolerance or training.

As species become more physiologically complex, innate immunological tolerance and
training become essential because tolerance becomes a necessary mechanism for
survival(Pham et al., 2007; Vincent and Sharp, 2014; Wendeln et al., 2018). In addition
to symbiotic relationships between anaerobic bacterium in the gut and immune tolerant
cells, the creation of immune privileged organs is another specialized result of innate
immune tolerance(Benhar et al., 2012; Stein-Streilein and Caspi, 2014; Louveau et al.,
2015). While the brain remains the most commonly known immune privileged organ, the
eyes, testes and ovaries also fall into this same category(Streilein, 1995; Stein-Streilein
and Caspi, 2014). Each of these organs have innate immune cells that could enter an
activated state like microglia, where they have the potential to clear a pathogen/cell debris
without the secretion of pro-inflammatory cytokines(Streilein, 1995; Stein-Streilein and
Caspi, 2014).

In this chapter, I showed how innate immune tolerance to Aβ could be Trem2 dependent
in microglia, whereas innate immune training to Aβ could be Trem2 dependent in
macrophages. The implications of these results extend beyond how the innate immune
system responds to pathogens in the brain because these in vitro results provide a
potential mechanism for why brain microglia become unresponsive to Aβ in post-mortem
LOAD patient brains. It is still not well understood whether microglial tolerance to Aβ is
developmental in origin, or if it is dependent on the brain microenvironment altering
217
microglial physiology. In other words, is it pre-programed or micro-environmentally driven
to be epigenetic? There are several lines of evidence that shows tolerance is dependent
on the epigenetic reprogramming in macrophages(Yoshida et al., 2015; Wendeln et al.,
2018). However, whether this phenomenon affects Trem2 or C1q expression is not well
understood.

TREM2 biology still left unanswered

Among all the data shown in the chapter, the most fascinating aspects of this
training/tolerance phenotype is the Trem2 specific expression differences and the
complex post-translational feedback/feedforward networks. Interestingly, recent evidence
from Wendeln and colleagues have found that immune training in the periphery could be
driven by IL-10 and TGF-β(Wendeln et al., 2018), both of which we see in this chapter.
Whether there is a cytokine-mediated or Trem2-dependent feed forward or feedback
mechanism involved in tolerance is an area that remains unexplored. While previous
evidence does point a TGF-β microenvironment acting in a feed forward mechanism
promote Trem2 expression and C1q expression(Sharif et al., 2014a; Yu et al., 2017),
whether this occurs in the brain is not known.

The primary way TREM2 expression has been measured in this chapter has been via
western blot, rt-PCR, and flow cytometry. These two methods do not show where Trem2
is expressed because Trem2 remains functional only when brought to the cell surface. If
Trem2 remains improperly folded, it will also remain in the ER and will not engage in
phagocytosis or pathogen sensing(Kleinberger et al., 2014; Kober et al., 2016; Song et
218
al., 2016; Schlepckow et al., 2017), thereby impairing immunoproteostasis. Furthermore,
while flow cytometry can implicate cell surface expression and provide some location-
specific expression for Trem2, internalized Trem2 expression can either mean the cell is
engaging in Trem2-dependent phagocytosis or dysregulated Trem2 expression.

In addition to TREM2 having location-specific signaling effects, SYK is prone to
intracellular aggregation, which prevents any downstream signal transduction
cascades(Ghosh and Geahlen, 2015). SYK can aggregate in stress granules, impairing
mononuclear cells from engaging in phagocytosis or pathogen sensing(Mócsai et al.,
2010; Ghosh and Geahlen, 2015). Whether this is a mechanism for extinguishing TREM2
signaling in AD has not been fully explored. In this chapter, the western blot data showed
that mostly Trem2 peripheral macrophages exhibited two forms of Trem2 protein, a
mature form and immature form. Recently studies pointed to Trem2 as having N and O
glycosylation sites that alters its receptor-binding abilities(Kleinberger et al., 2014; Song
et al., 2016; Park et al., 2017; Schlepckow et al., 2017). These concepts are summarized
in Figure 7.24.

In addition, TREM2 has been described to undergo proteolytic processing like TREM1,
creating a soluble TREM2 fragment (sTREM2)(Klesney-Tait et al., 2006; Turnbull et al.,
2006; Sharif and Knapp, 2008; Roe et al., 2014). This fragment is cleaved by an 𝛼-
secretase, metalloprotease(Kleinberger et al., 2014; Wu et al., 2015a; Schlepckow et al.,
2017). While these proteolytic products are speculated to alter extracellular or intracellular
signaling, reports have indicated that sTREM2 can opsonize neurons to mark them for
219
neuronal clearance, simultaneously sTREM2 in the extracellular space has been shown
to extinguish Trem2 function.

Whether any, a combination, or all of these factors impair TREM2 expression in tolerance
or training in vitro has still not been explained. There are several limitations to these
experiments. First, these data do not provide concrete evidence for tolerance or training
in an AD brain, nor is microglial tolerance dependent on TREM2. For all microglial work,
immortalized human microglia (SV40) were used to assess microglial TREM2 function.
Immortalization alters the cellular physiology and may not provide accurate results to
correlate the in vivo phenomenon.

In these experiments, we used immortalized cell lines to measure the microglial response
to Aβ in these tolerance-inducing paradigms. However, immortalized cell lines can not
replace primary microglia because the cellular metabolism is extremely different and how
they respond to pathogens are also very different. Two major experimental limitations to
using primary microglia from 6-month-old brain isolations are that we will not be able to
isolate enough cells to show whether microglia enter Aβ-dependent tolerance phenotype
in vitro. Furthermore, maintaining what is considered the microglial phenotype in vitro is
not possible because we lose that microenvironment that is needed to educate microglia
in a specific manner. While this is an active area of study, what’s known is that the TGF-
β microenvironment and Tregs may drive this “immune-privileged” tissue-
microenvironment(Saijo and Glass, 2011; Butovsky et al., 2013; Stein-Streilein and
Caspi, 2014; Baruch et al., 2015b). Therefore using immortalized cell lines quickly
220
answers whether microglia become tolerized in vitro. Additional experiments that use
combinatorial paradigms using TGF-β or Tregs to create this microenvironment for
inducing tolerance is beyond the scope for this dissertation. However, one future direction
is to potentially use a single cell proteomics approach to show a shift in the proteome
towards a reduction in proinflammatory gene expression and Trem2 expression post
stimulation. Although proteomics would capture the heterogeneity that exists in myeloid
cells, large-scale computational experiments are not usually the cleanest experiments
that validate microglial tolerance. These concerns will be answered in Chapter 9 - CyTOF,
where I will provide some hope to link the high dimensional “omics” data with a spatial
location in vivo.

Overall, this chapter has introduced a novel concept of macrophage training and
microglial tolerance in the context of Alzheimer’s disease by studying how these
mononuclear cells respond to Aβ and C1q opsonized Aβ. This concept can be
summarized in Figure 7.25, which is a modified version of the first Figure in this chapter.
One important concept in these last few figures are that these are preliminary experiments
show may show macrophage training, however, there is one critical control that was left
out in the final signaling experiments, and that was the C1q-Aβ low concentration control.
This control would demonstrate the concentration aspect of Trem2-C1q-Aβ
immunoproteostasis. While this is a major limitation, Chapter 8 dives into concentration
dependent mechanisms impact the stability of this neuroimmune complex using multiple
biochemistry experiments.

221
Coming back to this chapter, what we have found is that Trem2 may modulate Aβ
macrophage training in brain-filtrating peripheral macrophage that adopt a microglial
morphology. We found that under normal circumstances, there the impairments in the
pSyk-p38 axis remain intact and they may signal thorough Trem2
lo
expressing cells.
When we gain a deficiency in either C1q or Trem2, the signaling pathway may become
compromised and may shift to sustain the energetics of the cell, to sustain normal immune
function. While the other downstream signaling proteins of Trem2 were not examined, it
is likely that macrophage training activates GSK3-β or pAKT upon pSyk activation.
Whether C1q-Aβ or low Aβ concentrations both endorse the GSK3-β or pAKT axis
remains an active area of investigation. While there are some Trem2 studies that asses
Trem2 microglial survival being dependent on the GSK3-β-pAKT axis(Park et al., 2015;
Zhao et al., 2018), whether C1q-Aβ or low Aβ concentrations have anything to do with
this pathway also remains unexplored. Additionally, this chapter doesn’t explain how
macrophage training or microglial tolerance can affect immunoproteostasis in AD. This
chapter skirts at the idea. Chapter 10 will dive further into the concept of tolerance and
provide a wholistic view of the Trem2-C1q-Aβ inflection point in immunoproteostasis
regulation.

222

Figure 7.1: Trained immunity vs immune tolerance: Schematic diagram showing the
differential responses of mononuclear cell upon LPS, or in this dissertation Aβ. Red circles
indicate cellular degranulation, or cytokine release.

223

Figure 7.2: Human microglia express more TREM2 in response to Aβ in more acute
settings: Immortalized human microglia were treated with PBS, LPS, and Aβ to compare
TREM2 expression during acute vs chronic inflammatory settings. Mature TREM2 was
immunoblotted and quantification is presented above.

Unstimulated
LPS
Aβ
0.0
0.5
1.0
1.5
Trem2 Mature/Immature
TREM2 DENSITOMETRY (A.U.)
3 hr 24 + 3 hr
224

Figure 7.3: Human microglia signal through the TREM2-pSYK axis in more acute settings:
Immortalized human microglia were treated with PBS, LPS, and Aβ to compare pSYK
expression during acute vs chronic inflammatory settings. Mature pSYK was immunoblotted and
quantification is presented above.

Unstimulated
LPS
Aβ
0.0
0.5
1.0
1.5
P-Syk/Syk
P-Syk/Syk Densitometry (A.U.)
3 hr 24 + 3 hr
225

Figure 7.4: Paradigm for recreating immune tolerance in AD: 4 treatment conditions were
designed to recreate immune tolerance. High dose Aβ for 24 hours, followed by a 30 min
PBS/media wash, and a low Aβ treatment.

226

Figure 7.5: Chronic Aβ Tolerance inducing paradigm reduced TREM2 protein expression
in human microglia: Immortalized human microglia were treated using the Aβ tolerance
inducing paradigm. After treatment, samples were processed for immunoblot after 12 hours.
Experiment was repeated three times. Two-way ANOVA followed by a student’s T-test was
performed. Statistical annotations are compared to PBS untreated: † p<0.1; * p<0.05; ** p<0.01;
*** p<0.001.

227

Figure 7.6: Chronic Aβ tolerance-inducing paradigm impacts TREM2 expression:
Immortalized human microglia were treated using the Aβ tolerance inducing paradigm. After
treatment, samples were processed for RT-PCR. Yellow boxes indicate the two timepoints
used. Experiment was repeated three times. Two-way ANOVA followed by a student’s T-test
was performed. Statistical annotations are compared to PBS untreated: † p<0.1; * p<0.05; **
p<0.01; *** p<0.001.

228

Figure 7.7: Chronic Aβ tolerance-inducing paradigm impacts cytokine transcript
expression: Immortalized human microglia were treated using the Aβ tolerance inducing
paradigm and assessed for TNF-α and IL-6 expression. After treatment, samples were
processed for RT-PCR. Yellow boxes indicate the two timepoints used. Experiment was
repeated three times. Two-way ANOVA followed by a student’s T-test was performed. Statistical
annotations are compared to PBS untreated: † p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

229

Figure 7.8: Chronic Aβ tolerance-inducing paradigm impacts C1q transcript expression:
Immortalized human microglia were treated using the Aβ tolerance inducing paradigm and
assessed for C1q expression. After treatment, samples were processed for RT-PCR. Yellow
boxes indicate the two timepoints used. Experiment was repeated three times. Two-way
ANOVA followed by a student’s T-test was performed. Statistical annotations are compared to
PBS untreated: † p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

230

Figure 7.9: Under chronic Aβ tolerance-inducing paradigm, macrophages respond with
an increase in Trem2 expression: Peripheral macrophages were treated using the Aβ
tolerance inducing paradigm and assessed for Trem2 expression via immunoblot at the 12-hour
timepoint. Experiment was repeated three times. Two-way ANOVA followed by a student’s T-
test was performed. Statistical annotations are compared to PBS untreated: † p<0.1; * p<0.05;
** p<0.01; *** p<0.001.

231

Figure 7.10: Chronic Aβ tolerance-inducing paradigm impacts Trem2 expression in
macrophages: Peripheral macrophages were treated using the Aβ tolerance inducing paradigm
and assessed for Trem2 expression via immunoblot at the 12-hour timepoint. Experiment was
repeated three times. Two-way ANOVA followed by a student’s T-test was performed. Statistical
annotations are compared to PBS untreated: † p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

232

Figure 7.11: Chronic Aβ tolerance-inducing paradigm impacts Cytokine transcript
expression in macrophages: Peripheral macrophages were treated using the Aβ tolerance
inducing paradigm and assessed for TNF-α, IL-1β, and IL-6 transcript expression at the 3-hour
timepoint. Experiment was repeated three times. Two-way ANOVA followed by a student’s T-
test was performed. Statistical annotations are compared to PBS untreated: † p<0.1; * p<0.05;
** p<0.01; *** p<0.001.

233

Figure 7.12: Trem2-dependent macrophage trained immunity vs immune tolerance:
Schematic diagram summarizing this entire chapter, the differential responses macrophages
and microglia upon Aβ treatment. Red circles indicate cytokine degranulation.

234

Figure 7.13: Chronic Aβ tolerance-inducing paradigm impacts IL-10 transcript expression
in macrophages: Peripheral macrophages were treated using the Aβ tolerance inducing
paradigm and assessed for IL-10 transcript expression at the 3-hour timepoint. Experiment was
repeated three times. Two-way ANOVA followed by a student’s T-test was performed. Statistical
annotations are compared to PBS untreated: † p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

235

Figure 7.14: Chronic Aβ tolerance-inducing paradigm impacts TGF-β transcript
expression in macrophages: Peripheral macrophages were treated using the Aβ tolerance
inducing paradigm and assessed for TGF-β and TGF-βR1 transcript expression at the 3-hour
timepoint. Experiment was repeated three times. Two-way ANOVA followed by a student’s T-
test was performed. Statistical annotations are compared to PBS untreated: † p<0.1; * p<0.05;
** p<0.01; *** p<0.001.

236

TGF-β TGF-βR1 Ratio

↑ ↓ ↑ feedforward
↑ ↑ ? ?
↓ ↓ ? ?
↓ ↑ ↓ feedback
1 1 1 ?

Table 7.1: Assessing potential feedforward/feedback mechanisms: Table represents
whether a signaling process represents a potential feedback or feedforward mechanism

237

Figure 7.15: Chronic Aβ tolerance-inducing paradigm may promote a TGF-β feedforward
mechanism: Using the tolerance inducing paradigm, TGF-β/TGF-βRI ratios were created to
assess feedforward, feedback mechanisms. Regression plot shows an upward trend for the
chronic Aβ paradigm. First derivative of the regression plot shows the potential dependent
relationship between Trem2 and TGF-β/TGF-βR1 signaling.

238

Figure 7.16: Chronic Aβ tolerance-inducing paradigm impacts immune functions in
macrophages: Peripheral macrophages were treated using the Aβ tolerance inducing paradigm
and assessed for CCL2, Arg1, CD11c, and MHCII at the 3-hour timepoint. Experiment was
repeated three times. Two-way ANOVA followed by a student’s T-test was performed. Statistical
annotations are compared to PBS untreated: † p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

239

Figure 7.17: Incorporating a C1q-Aβ into the training-inducing paradigm: Schematic
diagram that shows the experimental paradigm that includes C1q-Aβ into the training paradigm.

240

Figure 7.18: Chronic Aβ training-inducing paradigm impacts Trem2 surface expression:
Peripheral macrophages were treated using the Aβ tolerance inducing paradigm and assessed
for Trem2 using flow cytometry. Cells were gated upon CD45
+
and CD68
+
cells. From these
gates, Cells were assessed for Trem2 expression in WT cells (Top) and C1q
-/-
cells (Bottom).
Experiment was repeated three times and compared to WT. Two-way ANOVA followed by a
student’s T-test was performed. Statistical annotations are compared to PBS untreated: †
p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

241

Figure 7.19: Chronic C1q-Aβ tolerance-inducing paradigm impacts Trem2
low
populations:
Peripheral macrophages were treated using the C1q-Aβ training inducing paradigm and
assessed for Trem2 using flow cytometry. Cells were gated upon CD45
+
and CD68
+
cells. From
these gates, macrophages were further gated for Trem2
high
and Trem2
low
expression in WT cells
(A) and C1q
-/-
cells (B). Experiment was repeated three times and compared to WT. Two-way
ANOVA followed by a student’s T-test was performed. Statistical annotations are compared to
PBS untreated: † p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

242

Figure 7.20: Chronic C1q-Aβ treatment may sustains pSyk Signaling in Trem2
low

populations: Peripheral macrophages were treated using the C1q-Aβ training inducing
paradigm and assessed for Trem2 using flow cytometry. Cells were gated upon CD45
+
and
CD68
+
cells. From these gates, macrophages were further gated for Trem2
high
and Trem2
low

expression to assess pSyk in WT cells (A) and C1q
-/-
cells (B). Two-way ANOVA followed by a
student’s T-test was performed. Statistical annotations are compared to PBS untreated: †
p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

243

Figure 7.21: pERK Signaling impacts C1q
-/-
macrophages but does not impact
macrophage training: Peripheral macrophages were treated using the Aβ tolerance inducing
paradigm and assessed for Trem2 using flow cytometry. Cells were gated upon CD45
+
and
CD68
+
cells. From these gates, macrophages were further gated for Trem2
high
and Trem2
low

expression to assess pERK in WT cells (A) and C1q
-/-
cells (B). Two-way ANOVA followed by a
student’s T-test was performed. Statistical annotations are compared to PBS untreated: †
p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

244

Figure 7.22: Chronic C1q-Aβ treatment may sustains p38 Signaling in Trem2
low

populations: Peripheral macrophages were treated using the C1q-Aβ training inducing
paradigm and assessed for Trem2 using flow cytometry. Cells were gated upon CD45
+
and
CD68
+
cells. From these gates, macrophages were further gated for Trem2
high
and Trem2
low

expression to assess p38 phosphorylation in WT cells (A) and C1q
-/-
cells (B). Two-way ANOVA
followed by a student’s T-test was performed. Statistical annotations are compared to PBS
untreated: † p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

245

Figure 7.23: Chronic C1q-Aβ treatment impacts the pSyk-p38 signaling axis in Trem2
-/-

macrophages: Peripheral macrophages were treated using the C1q-Aβ training inducing
paradigm and assessed for Trem2 using flow cytometry. Cells were gated upon CD45
+
and
CD68
+
cells. From these gates, macrophages were further gated for Trem2
high
and Trem2
low

expression to assess pSyk-p38 signaling axis in Trem2
-/-
macrophages. Two-way ANOVA
followed by a student’s T-test was performed. Statistical annotations are compared to PBS
untreated: † p<0.1; * p<0.05; ** p<0.01; *** p<0.001.

246

Figure 7.24: TREM2 posttranslational regulation: Mechanisms that can affect Trem2
response in tolerance: (1) matured glycosylation, (2) misfolding, (3) metalloprotease cleavage,
(4) improper glycosylation, (5) Syk stress granules blocking downstream signaling, (6) TREM2
ER-localization.

247

Figure 7.25: Trem2-dependent macrophage trained immunity vs immune tolerance:
Schematic diagram summarizing this entire chapter, the differential responses macrophages
and microglia upon Aβ and C1q-Aβ treatment. Red circles indicate cytokine degranulation.
Downstream Trem2 dependent signaling mechanisms are presented to the right.

248
Chapter 8: Resolving the Neuroimmune Complex

“ 辭卑而益備者，進也。辭強而進驅者，退也。”

“Humble words and increased preparations are signs that the enemy is about to
advance. Violent language and driving forward as if to attack are signs that he will
retreat. “
Chapter 9 Army on the March
Art of War, Sun Tzu

Introduction:

In previous chapters, the interaction of Trem2 and C1q impacts immunoproteostasis,
specifically the interaction among (1) phagolysosomes to Trem2, (2) Trem2 to C1q
opsonized Aβ, and (3) Trem2 to the different conformational states of Aβ. To briefly
recapitulate the central thesis of this entire dissertation, Trem2 should interact with C1q
opsonized Aβ (molecular weights, Figure 8.1). To interrogate whether this interaction can
assemble into a neuroimmune complex in the brain, SEC-MALS (size exclusion
chromatography, multi angle light scattering spectroscopy, herein I will also use SEC-
MALS as MALS, Figure 8.2) was adopted to interrogate whether Trem2 will interact with
C1q and Aβ, forming the neuroimmune complex (Figure 8.1). This chapter will focus on
the biochemistry techniques that were employed to characterize this neuroimmune
complex.
Size Exclusion Chromatography, Multi Angle Light Scattering:

MALS is a form of light spectroscopy that first separates biomolecules by size using High
Performance Liquid Chromatography (HPLC) columns to separate the proteins out of the
column(Berg et al., 2007). Chromatography is the study of separation using two phases,
249
stationary and mobile phase. The HPLC column is the stationary phase, packed with
TSKgel, which traps biomolecules in the gelatinous matrix. The mobile phase consists of
water or saline, which moves the larger biomolecules through the column faster than
smaller ones. To reconcile this phenomenon, the porous gel traps smaller proteins and
they meander through the column slower than the larger biomolecules(Berg et al., 2007;
Skoog, 2007).

In addition to the molecular distance that is required for the biomolecule to travel through
the column, Van der Waal’s forces can affect how the gel interacts with the biomolecule.
If one imagines these biomolecules are spheres, larger molecules will be “heavier” and
thus have lower surface area to mass ratio than a smaller molecule, which would have a
higher surface area to mass ratio. This larger surface area to mass ratio (smaller
biomolecules) will be more affected by the Van der Waals than smaller surface area to
mass ratio (larger biomolecules). This can be interpreted as more drag or resistance for
smaller biomolecules, which will affect when they come off the column(Skoog, 2007).

As the protein flows through the column, it will come upon a columnar and
monochromatic wavelength of light (laser beam) and the protein will scatter this light in
all directions (Figure 8.2). The scattered light is detected by the 20 photomultiplier tubes
in the instrument(Skoog, 2007). The instrument will then take the measured light photons
to back-calculate the molecular mass of the protein at a given time during the flow.

250
The calculation of molecules and proteins in the MALS is based upon the physical
property called Rayleigh scattering(Berg et al., 2007; Skoog, 2007). This same physical
property explains why the sky is blue. To start, when light approaches an object it can be
absorbed, refracted, diffracted, or phase shifted. In this case, the incident beam
approaches the biomolecule and refracts the light into detectors. In the case of the blue
sky, it’s the sunlight hitting a molecule of Nitrogen gas, scattering the incident mean of
sunlight into blue light that we see in our eyes)(Skoog, 2007).

In this chapter, my goal is to explain my rationale for using MALS to answer the
biochemical interaction to form this trimeric complex we have seen in previous chapters.

Circular Dichroism

Circular dichroism (CD)is a spectroscopy technique that typically used to characterize
secondary structures and tertiary structures in biomolecules. Biomolecules can absorb
light; however, certain atoms, Carbon and Nitrogen, contain the ability to form chiral
centers(Berg et al., 2007; Skoog, 2007). These chiral centers produce molecules of the
same molecular structure, but they are mirror images carrying out vastly different
functions. For example, Glucose comes in two forms, D and L, of which only D is
metabolized, and L remains unaltered. For polypeptides, the a helix and β pleated sheets
of contain many chiralic centers and these centers are particularly sensitive to the light
emitted by CD(Berg et al., 2007; Skoog, 2007).

251
CD a biochemistry technique that is based upon the Beer’s law. Circular polarized light is
the primary light source that is emitted by in CD spectroscopy. Figure 8.3 illustrates how
Left and Right circularly polarized light are emitted from the light source at a specific
wavelength to produce a forward photon beam(Skoog, 2007). As the left and right
circularly polarized light exits the light source, both sides will hit the protein with equal
strength. Certain wavelengths are absorbed differently by certain chiral centers and the
peptide will preferentially absorb one rotational polarized light direction over the other.
This average imbalance in left or right circularly polarized light is received by the detector
which displays the average ellipticity of the peptide(Skoog, 2007). In this dissertation, CD
was will be used to identify which Aβ conformation interacts with Trem2 and/or C1q, a
necessary control to interrogate Trem2-C1q immunoproteostasis.

Results:
TREM2 binds to Aβ; C1q opsonizes Aβ

We adopted Size Exclusion Chromatography Multi Angle Light Scattering (SEC-MALS)
technology, a technique that separates proteins via a chromatography column and back-
calculates the molecular weight of each protein based on the light scattering pattern as it
exits the column. TREM2 binding to C1q opsonized Aβ is illustrated in a cartoon (Figure
8.1). To confirm previously reported interactions(Wang et al., 2015a; Lee et al., 2018a),
we first determined whether TREM2 could interact with soluble Aβ at a 1:1 nM ratio
(Figure 8.4). We found a molecular weight shift from 47.7kDa to 72.1kDa, suggesting
TREM2-Aβ interaction. When we add 100-fold more Aβ to TREM2, from 1:1 nM (blue)
and 1:100 nM (green) (Figure 8.5, Top), we found that TREM2 comes off the column
252
independently of Aβ with a molecular weight that could implicate TREM2-Aβ interaction,
suggesting that TREM2–Aβ interaction may still exist. However, a molecular weight of
80.6±34.6kDa could suggest multiple binding interactions between Aβ monomers,
dimers, and larger oligomeric species that interact with TREM2 (Figure 8.5 below,
illustrated summary). Therefore, TREM2 complexation is more likely to be specific for
solubilized Aβ at lower equimolar concentrations, rather than at high concentrations.

To further validate the biochemical experiments that illustrate whether C1q is capable of
opsonizing Aβ(Terai et al., 1997), we first tested that C1q and Aβ formed a complex
regardless of Aβ concentration (Figure 8.6 top). We found that the molecular weight does
not change, signifying that C1q will opsonize Aβ at any concentration. However,
remaining Aβ that is not bound by C1q will aggregate to form larger Aβ oligomers. These
data support the previous studies that show C1q-dependent Aβ nucleation, in this case,
formation of larger beta oligomers (Figure 8.6 bottom). Taken together, these data
demonstrate that both Trem2 and C1q are capable of interacting with Aβ independently
yet altering the equilibrium by increasing the Aβ concentration shows differences in the
stability between TREM2-Aβ or C1q-Aβ in vitro.

TREM2, C1q and Aβ physically interact in vitro

Since we previously established that TREM2 and C1q could interact with Aβ
independently. We assessed whether TREM2 and C1q could interact without the
influence of Aβ (Figure 8.7). We found that both proteins independently come off the
column at the molecular weights 453.5kDa (C1q) and 41.7kDa (TREM2). While there is
253
one larger recording at 1023.4kDa, this has been reported to be a dimerization of C1q, a
non-biological effect that arises when C1q cannot opsonize a pathogen. From this control
experiment (summarized in Figure 8.7, bottom), I can conclude that TREM2 and C1q do
not physically interact when alone.

To determine whether this heteromeric complex forms in vitro in the presence of Aβ, we
added each component at a 1:1:1 nM ratio. At a molecular weight of 559.2kDa, TREM2,
C1q, and Aβ (purple) forms a more stable complex than C1q and Aβ (blue) alone at
503.7kDa (Figure 8.8), signifying that TREM2 (47.5kDa) has bound to C1q opsonized Aβ.
Additionally, these data also indicate that the interaction of TREM2-C1q-Aβ at 1:1:1 ratio
is more molecularly stable as evidence by the inset (Figure 8.8).

One question that still remains is, if MALS contains a chromatography component (size
exclusion chromatography, SEC), then each of the individual peptides added into the
reaction should exit the column. Comparing the addition of TREM2, Aβ, and C1q (purple)
to other controls (Figure 8.9A), we find that the other proteins come off the column at
different times. Figure 8.9B illustrates each of these color-coordinated components from
the MALS spectrograph above. One unusual feature of this trimeric equilibrium is the
potential TREM2-Aβ interaction or large Aβ oligomers (purple) that may have fallen out
of the complex. The characterization of the proteins at this molecular weight will be
described within a later section of this chapter. To further characterize the interaction of
Trem2-C1q-Aβ, we used 3-month-old APP/PS1 mice to perform an IP experiment to show
the interaction of all three partners in vivo (Figure 8.9C). We found that when we
254
independently probed for Trem2, Aβ or C1q, we found that Trem2 and C1q would also
be found in the flow through, suggesting that that the trimeric compound is found in vivo.

To test the stability of the complex at higher concentrations, we increased the soluble Aβ
concentration 100-fold and incubated this with TREM2-C1q (green, Figure 8.10). I see
that the TREM2 interaction between C1q opsonized Aβ at a 1:1:1 ratio holds at ~560kDa
(purple), whereas the 100-fold increase in Aβ (green) offsets this equilibrium, dropping
the molecular weight to C1q opsonized Aβ molecular weights (blue and red) (Figure 8.10).
This result suggests that TREM2-C1q-Aβ interaction preferentially exists when Aβ is
soluble at equimolar concentrations and that C1q preferentially could opsonize Aβ at
higher concentrations.

We next wanted to determine whether this trimeric complex forms in the presence of Aβ1-
40 and the reverse peptides (Figure 8.11). Incubating Aβ1-40 with TREM2 and C1q (green),
we found that the tri-meric complex does not form because the molecular weight holds
steady at 499.96 kDa (Figure 8.11). When comparing between the reverse peptides
incubated with TREM2 and C1q (Figure 8.11B, turquoise and brown), we found that all
three proteins, Aβ, TREM2, and C1q, independently fall off the column. Taken together,
these results suggest that the formation of this trimeric complex exists in vitro and it is Aβ
concentration specific.

255
Aβ can oligomerize and fibrilize in vitro

Previous experiments were performed on freshly solubilized Aβ. Since the Aβ found in
the later stages of AD are considered more aggregated, we aggregated Aβ according to
a previously set protocol where Aβ is incubated in a 37ºC humidified chamber for 24 hours
(Paresce et al., 1996). To confirm that the Aβ is oligomgerized or prefibrilized, we used
Circular Dichroism to tease apart the secondary and tertiary structures found when Aβ
aggregates into its more complex forms (Figure 8.12). Many studies have characterized
the formation of Aβ species as it autonomously aggregates or with a catalyst(Jiang et al.,
2012). Since Aβ tertiary structure is best resolved between 190nm and 200nm, we find a
that the reverse peptides (green and purple) show negative molar ellipticity values within
this wavelength. These values are characteristically described to be more monomeric
peptides than their monomeric isoforms. When we look at the freshly solubilized Aβ1-42
(purple), we find that the molar ellipticity values between 190 and 200nm are not entirely
monomeric, rather these Aβ species seem to be a heterogeneous mixture of monomeric,
small aggregates, or soluble oligomers. In the aggregated Aβ1-42 spectrograph, the molar
ellipticity curve suggests that the Aβ peptides range between oligomers and fibrils. From
this circular dichroism spectrograph, we can conclude that our aggregation paradigm is
successful. Although there is a heterogeneous mixture of Aβ states the overall main Aβ
species in the population between freshly soluble and aggregated are different.

To confirm the results provided by the CD, we use SEC-MALS and found that when we
compared aggregated Aβ1-42 and freshly solubilized Aβ1-42, the molecular weights (Figure
8.13), at first glance, remarkably different, even though the refractive index curves are
256
similar. The higher molecular weights for the aggregated sample (turquoise) suggests
that the Aβ contains more aggregated species at higher molecular weights. However,
there is a noticeable peak at the 6mL mark that shows a very high molecular weight,
suggesting that these forms are mostly likely prefibrilized Aβ. When we compared these
peaks to the freshly solubilized Aβ, the prefibrilized peak does not exist, but oligomeric
forms and monomeric forms are detected. These SEC-MALS results further indicate that
the Aβ used in these experiments are generally more aggregated.

Aβ oligomers and prefibrils differentially impact Trem2-Aβ and C1q-Aβ stability

Since aggregated Aβ consists of larger oligomers and Aβ, we wanted to see whether
TREM2 can interact with aggregated Aβ (magenta, Figure 8.14). We found that when
TREM2 and Aβ are added at equimolar ratios, freshly solubilized Aβ (blue) can interact
with TREM2 as described previously, but with aggregated Aβ, there is no difference. This
suggests that TREM2 and aggregated Aβ may interact. However, closer examination into
the spectrograph, we see that the individual measurements, are not consistent,
suggesting TREM2-Aβ complex instability. When comparing between an aggregated Aβ
control (black, Figure 8.15) and TREM2+Aβagg(magenta), we found that this variability
could also be oligomerized Aβ. While these data suggest that the 64.5kDa complex is the
interaction between TREM2-Aβ, these results are inconclusive because aggregated Aβ
yields a similar yet overlapping molecular weight.

Next, we wanted to assess whether increasing the concentration of aggregated Aβ alters
TREM2-Aβ interaction, we find when we compare equimolar concentrations of TREM2
257
and Aβ (magenta, Figure 7.16) to 100x more aggregated Aβ (cyan), we find that the
potential TREM2-Aβ (1:100) interaction increase to 91.1kDa. This increase in molecular
weight mirrors the same phenomenon as the TREM2-Aβ equimolar concentration, but
with a 19kDa increase (72.1 kDa, monomeric Aβ at 1:1 molar; 91.1kDa, oligomeric Aβ at
1:100 molar). This increase more strongly associates with larger Aβ oligomers than a
potential TREM2-Aβ interaction (Figure 8.17). Therefore, incubating aggregated Aβ with
TREM2 in our SEC-MALS experiments may show the loss of a TREM2-Aβ interaction.

While aggregated Aβ impairs the TREM2-Aβ binding, we wanted to assess how well C1q
opsonizes aggregated Aβ and at different concentrations (Figure 8.18). Placing C1q with
aggregated Aβ at high concentrations (turquoise), we see that C1q opsonizes aggregated
Aβ, yielding a higher molecular weight (880.9±112kDa) than C1q and soluble Aβ (red,
506±7.6kDa), suggesting that C1q could preferentially opsonize aggregated Aβ over
soluble Aβ. The other peaks with C1q aggregated Aβ (turquoise), 376.9±23kDa and
156.8±kDa, are considered to be larger Aβ aggregates, compared to aggregated to Aβ
alone (grey). This molecular weight shift (turquoise, 376kDa and 156kDa) is thought to
be a property of C1q, which has nucleating properties. These nucleating properties
catalyze the Aβ aggregation state to much larger species, providing the same refractive
index curve as aggregated Aβ (comparing turquoise to grey). These data indicate that the
addition of aggregated Aβ improves C1q-Aβ opsonization.

258
Aggregated Aβ destabilizes the formation of TREM2-C1q-Aβ in vitro

Establishing that Aβ aggregation impairs TREM2-Aβ binding, but improves C1q-Aβ
opsonization, we sought to assess the whether the trimeric complex forms in the presence
of aggregated Aβ and at higher concentrations of aggregated Aβ (Figure 8.19-23). We
first asked what would happen if we added aggregated Aβ at equimolar concentrations
(Figure 8.19). In the left insert, we find that the addition of aggregated Aβ increases the
molecular mass of the TREM2-C1q-Aβ complex (magenta, 559kDa; cyan, 603.9kDa).
However, we find that the measurements made are highly variable suggesting some
trimeric complex instability. If there is instability in the complex equilibrium, TREM2 should
independently come off the column apart from the complex. In the right insert, we find
that all 4 molecular weight measurements are similar in molecular weights, suggesting
the molecular interactions formed in this region are more likely to be oligomerized Aβ.

When we increase the concentration of aggregated Aβ (red, Figure 8.20), we find that
multiple complexes are produced from the reaction. We find that there’s a large molecular
complex that is formed in the concentrated, aggregated Aβ sample with a molecular
weight of 951kDa (red). At first glance, this molecule could form the large trimeric complex
TREM2-C1q-Aβ, but when comparing this sample (red) to the concentrated C1q
opsonized aggregated Aβ sample (turquoise), we find that the molecular weights are
similar, but the red, trimeric complex sample, has more variable measurements,
suggesting molecular instability. This indicates that the molecule formed is more likely to
be a large C1q opsonized Aβ molecule, compared to the trimeric complex. In the right
panel insert, we find that the concentrated, aggregated Aβ with TREM2-C1q sample (red)
259
produces multiple fractions at the 8mL time point (Figure 8.20). Comparing this sample
among the other controls, we find that it’s difficult to discern what the species best
describe each of the red species. What we can conclude from this Figure 8.20 is that the
23.0±4kDa complex is most likely the oligomerize. For the other complexes,
767±314kDaand 41.9kDa, Figures 8.21-22 will further explain the rationale behind these
molecular complexes.

To describe the last two complexes, the complex with the molecular weight of
767.2±314kDa is compared among all TREM2-C1q-Aβ complexes (Figure 8.21). In the
insert, this complex (red) shows a similar MALS scatter to the equimolar TREM2-C1q-
Aβagg sample (cyan). While it has a molecular weight that could indicate the formation of
TREM2-C1q-Aβ, the stability of this interaction is more likely to be C1q-Aβ opsonizing
large Aβ aggregates or protein aggregates. When assessing the other complexes that
detected by the instrument, we compare the concentrated, aggregated TREM2-C1q-Aβ
sample among other controls (Figure 8.22). We find that the 41.9±14kDa complex (red)
is mirrors the molecular weight found in the TREM2 alone sample 41.7±7.5kDa (brown),
suggesting that this complex is likely TREM2 that falls out of the trimeric complex
equilibrium with C1q-Aβ, both the 767.2kDa complex and the large 951Da complex.
Furthermore, this result indicates that the formation of this trimolecular complex when we
disturb the equilibrium by altering Aβ, both aggregation state and concentration, which
promotes the formation of the C1q-Aβ complex and forces TREM2 to elute independently.
As a whole, these MALS results illustrate that the TREM2-C1q-Aβ more likely forms
around soluble.
260
Discussion:

The SEC-MALS experiments illustrated in this chapter points to an equilibrium that occurs
with TREM2, C1q, and Aβ. The interaction of TREM2 with Aβ is more stable and defined
when Aβ is soluble and at equimolar concentrations. As we increase the soluble Aβ, we
still find the complex remains. As we alter the aggregation state of Aβ, we find that
TREM2-Aβ MALS scatter becomes more unstable, resembling larger Aβ oligomers.
These TREM2 studies indicate that Aβ interacts with TREM2 at lower concentrations,
rather than higher. When we assessed the equilibrium of C1q and Aβ, we find that the
complex forms, regardless of concentration; however, C1q prefers larger Aβ aggregates
than Aβ. What is interesting here is that we, again, have a molecular indication that there
is an inflection point like the in vivo experiments found in Chapter 2, where we may have
a strong interaction at a particular inflection point. When we examine the MALS that
contains the three components, we find similar results showing how the complex is stable
in the presence of soluble Aβ over aggregated Aβ (Figure 8.21). This phenomenon also
alludes to immunoproteostasis. Because of experimental limitations, these experiments
explain one facet of immunoproteostasis described in Chapter2, the interaction between
receptor and pathogen. In vitro peripheral macrophage isolations and in vivo mouse
studies have alluded to the interaction of macrophages with Aβ plaques that is Trem2 and
C1q dependent in previous chapters. In this chapter, the evidence suggests that the
interaction between TREM2 and C1q is concentration dependent and pathogen
dependent.

261
Although the potential conclusions from this chapter introduce a novel phenomenon in
TREM2 biology, there are limitations to MALS and the current biochemical experiments
thus far. For one, SEC MALS doesn’t label proteins with high accuracy, rather we infer
the molecular weight from what comes off the column from our controls. For example, the
molecular weight of TREM2 is 47.5kDa; however, when TREM2 interacts with Aβ, we
don’t know how many Aβ monomers interact with TREM2. The ratio impacts the
molecular weight, but if the interaction between TREM2-Aβ molecular weight is
sufficiently higher than molecular weight of TREM2 alone, then it would suggest an
interaction. This same logic was applied for interrogating the interaction of C1q
opsonization with Aβ and the trimeric complex formation.

The identification of peptides and complexes that are identified in SEC MALS becomes
significantly more difficult when aggregated Aβ is used as the input measurement. That
is because Aβ oligomers have been characterized to span 20kDa to 100kDa. This is
further complicated because the chromatography portion of SEC-MALS isolates these
oligomers around the same molecular weight as TREM2. For example, in these SEC-
MALS experiments, we can only infer whether this 80.6±34.6kDa protein is in relation to
a single loaded control of TREM2 vs a single loaded control of Aβ. The (*) notation in the
text denotes the variability in identifying the specific peptide at the time point. When one
looks at the Figures, many of these * labeled proteins are paired with other control runs
to identify the complex in question. In parallel to a direct molecular weight comparison is
the MALS scatter that identifies the recording from 20 detectors at a given time point. The
262
larger the scatter, the more unstable the protein is, suggesting the formation of Aβ
oligomers or protein aggregates.

While one solution for this problem is to perform Circular Dichroism on this entire complex
with TREM2, C1q, and Aβ. However, this is not possible because CD is best used for
interrogating smaller peptides(Skoog, 2007). If the proteins become too large, the chiral
centers will be obstructed, and the measurements and recordings will not yield an
informative spectral curve. Other biochemistry techniques were considered, Surface
Plasmon Resonance, Atomic Force Microscopy, and X-ray Crystallography; however,
each of these techniques requires more protein than MALS. One of my biggest concerns
throughout this entire SEC-MALS was the difference between glycosylated TREM2 and
un-glycosylated TREM2. The post translational changes in TREM2 biology are not well
studied and the two peptides offered in the market are either made in bacteria or
eukaryotic cells. Studies have indicated that TREM2 glycosylation is necessary for
TREM2 function and it promotes the interaction of its ligands(Kleinberger et al., 2014;
Kober et al., 2016; Song et al., 2016; Schlepckow et al., 2017). Others have shown that
TREM2 processing, downstream pSyk localization, and ER-localized-TREM2 all play a
large role in the regulation of Trem2 (Figure 8.23). Among the studies that characterize
TREM2 mutations, the main and more detrimental mutations are speculated to reside
around TREM2 binding domain and impair glycosylation. Whether aging impairs
glycosylation and protein maturation, specifically TREM2 maturation, has not been
investigated.

263
While this chapter answers the fundamental premise of this dissertation, its opens up
more questions regarding the innate immune balance, trimeric complex equilibrium, and
function in relation to immunoproteostasis. All of which, require more biochemical
experiments that can resolve this trimeric complex at a higher resolution.

264

Figure 8.1: Working hypothesis for TREM2-C1q-Aβ interaction: Schematic diagram shows
the hypothesized interaction for TREM2-C1q-Aβ, neuroimmune complex. Molecular weights
were taken from the manufacture’s MSDS.

265

Figure 8.2: Size Exclusion Chromatography-Multi Angle Light Scattering Spectrometry:
Schematic diagram shows how the protein capillary column carries the separated protein
through an interrogation point where the incident beam scatters the light in multiple directions,
with readings on the detectors.

266

Figure 8.3: Circular Dichroism Spectrometry: Schematic diagram shows how polarized
circular light beams can intersect and form a specified interrogation point, which is at the
cuvette. Protein in the cuvette absorbs the one circular polarized light over the other. Absorption
differences is measured by the detector and produces the values as molar ellipticity.

267

Figure 8.4: TREM2 interacts with Aβ: SEC-MALS spectrum showing equimolar concentrations
of TREM2+Aβ(blue). Summary diagram is shown below.

268

Figure 8.5: TREM2 interacts with higher concentrations of Aβ: SEC-MALS spectrum
showing equimolar concentrations of TREM2+Aβ and 1:100 molar ratio o TREM2+Aβ.
Summary diagram is shown below.

269

Figure 8.6: C1q interacts with Aβ: MALS spectrum showing the SEC-MALS spectrum with
equimolar concentrations of C1q+Aβ. Summary diagram is shown below.

270

Figure 8.7: Trem2 and C1q do not independently interact: SEC-MALS spectrum showing
equimolar concentrations of TREM2+C1q. Summary diagram is shown below.

271

Figure 8.8: Trem2 binds to C1q opsonized Aβ: (A) SEC-MALS spectrum showing the
equimolar concentrations of TREM2+C1q+Aβ. Insert shows the magnification of the interaction
to suggest molecular stability of the complex. (B)Summary diagram is shown below. (C)
Immunoprecipitation experiments were performed on 3-month-old mice APP/PS1 mice.
Samples were either Immunoprecipitated for TREM2, Aβ or C1q. Columns represent
Immunoblots detecting Trem2 or C1q.

272

Figure 8.9: Trem2-C1q-Aβ formation is at an equilibrium: SEC-MALS spectrum showing
with equimolar concentrations of TREM2+C1q (Gold), C1q-Aβ (Dark Blue), TREM2-C1q-Aβ
(Magenta), and Trem2-Aβ (blue). Insert shows the complex stability in comparison with other
molecules at the same weight. Summary diagram is shown below.

273

Figure 8.10: Trem2-C1q-Aβ complex is heavier than C1q-Aβ: SEC-MALS spectrum showing
the 1:1:100 molar concentration ratios of TREM2+C1q+Aβ (Green) and C1q+Aβ (Red). This is
also overlaid with equimolar concentrations of C1q-Aβ (Dark Blue), TREM2-C1q-Aβ (Magenta).
Insert shows the complex stability in comparison with other molecules at the same weight.
Summary diagram is shown below.

274

Figure 8.11: Trem2-C1q-Aβ formation is Aβ species specific: SEC-MALS spectrum showing
C1qAbeta with different species of Aβ. (A) Aβ1-42 and Aβ1-40 with TREM2 and C1q are overlaid.
(B) the reverse sequences for Aβ 42-1 and Aβ 40-1 are overlaid with forward sequences.

275

Figure 8.12: Aβ can be Aggregated at 37ºC for 24 hours: Circular dichroism of different Aβ
species show that Aβ1-42 can be aggregated to form oligomers and it is very different from the
other Aβ species. Ellipticity values have been normalized to the concentration of the peptides.
Experimental measurements, n=5/sample.
276

Figure 8.13: Trem2-C1q-Aβ formation is Aβ species specific: SEC-MALS spectrum showing
equimolar concentrations of aggregated and non-aggregated Aβ1-42. Below shows the
spectrograph is a diagram showing the aggregation protocol and the species that is identified
from the SEC-MALS spectrum.

277

Figure 8.14: Assessing the formation of TREM2-Aβ agg: SEC-MALS spectrum shows the
difference between aggregated and non-aggregated Aβ species with TREM2.

278

Figure 8.15: TREM2-Aβ agg may be oligomeric Aβ: SEC-MALS spectrum shows the difference
between aggregated and non-aggregated Aβ species with TREM2. These spectra are overlaid
with Aggregated Aβ1-42 to compare molecular weights. Below the spectrograph shows the
summary of this figure.
279

Figure 8.16: High Aβ agg may break the Trem2-Aβ interaction, forming oligomers: SEC-
MALS spectrum compares the difference between high and low concentrations of aggregated
Aβ with TREM2. Each condition is overlaid on the MALS spectrograph and compared to
aggregated Aβ alone. Below this spectrograph, a schematic diagram that correlates to each
species
280

Figure 8.17: High Aβ agg may break the Trem2-Aβ interaction, from an equilibrium point of
view: SEC-MALS spectrum shows the difference between high and low concentrations of Aβ in
2 different forms, aggregated and non-aggregated Aβ, with TREM2. Each condition is overlaid
on the MALS spectrograph. Below this spectrograph, a schematic diagram that correlates to
each species

281

Figure 8.18: High concentrations and Aβ agg endorses the C1q-Aβ interaction, forming
oligomers: SEC-MALS spectrum shows the difference between high and low concentrations of
Aβ in 2 different forms, aggregated and non-aggregated Aβ, with C1q. Each condition is
overlaid on the MALS spectrograph. Below this spectrograph, a schematic diagram that
correlates to each species

282

Figure 8.19: Low Aβ agg may form the TREM2-C1q-Aβ interaction: SEC-MALS spectrum
shows the overlay among three Aβ conditions, high, low, and low aggregated, all with TREM2
and C1q. Inserts show the size comparison for molecular species around the same molecular
weight as Trem2-C1q-Aβ and Trem2-Aβ. Below shows a spectrograph, a schematic diagram
correlates to each species.
283

Figure 8.20: High concentrations and Aβ agg endorses the C1q-Aβ interaction, over the
trimeric complex: SEC-MALS spectrum shows the overlay among three Aβ conditions, high,
low, and High aggregated, all with Trem2 and C1q. Inserts show the size comparison for
molecular species around the same molecular weight as Trem2-C1q-Aβ and Trem2-Aβ. Below
shows a spectrograph, a schematic diagram correlates to each species.
284

Figure 8.21: High concentrations and Aβ agg negatively impacts the stability of TREM2-
C1q-Aβ: SEC-MALS spectrum shows the overlays Aβ in all conditions, high, low, aggregated
and non-aggregated Aβ, with Trem2 and C1q. Inserts show the size comparison for molecular
species around the same molecular weight as Trem2-C1q-Aβ. Below shows a spectrograph, a
schematic diagram correlates to each species.

285

Figure 8.22: High concentrations and Aβ agg endorses the C1q-Aβ interaction and offsets
the equilibrium: SEC-MALS spectrum shows the overlays the high concentration of Aβ agg with
Trem2 and C1q with other controls, high concentrated Aβ agg with C1q and Trem2-C1q alone.
Inserts show the size comparison for molecular species around the same molecular weight as
C1q-Aβ and Trem2 alone. Emphasis for these inserts show the fall out of Trem2 from the
trimeric complex. Below shows a spectrograph, a schematic diagram correlates to each
species.
286

Figure 8.23: TREM2 posttranslational regulation: SEC-MALS does not have the ability to
answer the post-modifications of Trem2. Above shows the ways TREM2 can be regulated: (1)
mature glycosylation, (2) Misfolding, (3) Metalloprotease cleavage, (4) no glycosylation, (5) Syk
stress granules preventing downstream signaling, (6) TREM2 ER-localization.
287
Chapter 9: imCyTOF

“ 故以火佐攻者明，以水佐攻者強，水可以絕，不可以奪。”

“Hence those who use fire as an aid to the attack show intelligence; those who use
water as an aid to the attack gain an accession of strength. By means of water, an
enemy may be intercepted, but not robbed of all his belongings.”

Chapter 12, Attack by fire
Art of War, Sun Tzu

Introduction:

While previous chapters focused on the interaction between Trem2 and C1q in
Alzheimer’s disease, this chapter will center around imaging mass cytometry (iMC), or
imaging Cytometric Time of Flight (imCyTOF) and its adaptation from cancer immunology
into neuroimmunology and neuroscience. Additionally, this chapter will not have novel
data regarding the interaction between Trem2 and C1q but provide the framework for a
future methods publication with implications for using this technology in brains of post
mortem patients with Alzheimer’s disease, Multiple Sclerosis, or other neurological
disorder. The motivation for adopting and setting the foundation for imaging CyTOF was
to assess microglial provenance in a human brain. Thus, this chapter will describe the
technological workflow, followed by the computational analysis of these data, and the
technological limitations.

Imaging Cytometric Time of Flight (imCyTOF)

While the Heisenberg uncertainty principle in physics states the inability to calculate an
electron’s location and momentum simultaneously, the biological correlate of the
288
Heisenberg principle is simultaneously resolving high dimensional gene expression along
with high structural resolution (Figure 9.1). For example, while traditional confocal
microscopy achieves structural resolution, because only 4-5 antigens can be imaged at
once, there is limited gene expression information on a per-cell basis. Many techniques
follow this same trend, especially Florescence Automated Cell Sorting (FACS). Although
FACS allows rapid and facile detection of multiple antigens, like fluorescence microscopy,
a major drawback of this technology is spectral overlap between fluorescent probes that
limits number of analytes(Mair et al., 2016; Melchiotti et al., 2016). Moreover, auto-
fluorescence in leukocytes occurs at 500-600nm, limiting the use of fluorochromes at
these wavelengths(Becher et al., 2014). Mass-cytometry Time of Flight (CyTOF)
addresses this limitation by substantially increasing antigen detection up to 100 metal
probes, without the problem of spectral or mass overlap. This allows for a rich multi-
dimensional dataset, providing high resolution cellular expression necessary to
interrogate immune cell populations. However, this comes at a cost of losing spatial
information. The solution to this challenge is combining confocal microscopy with
CyTOF—a technique known as imaging CyTOF (imCyTOF) (Figure 9.1).

ImCyTOF is a revolutionary technique that combines traditional confocal microscopy with
standard mass-cytometry cytometric time of flight. Because antibodies are conjugated to
rare earth metals (mostly lanthanide and actinide isotopes), our number of antigen
analytes dramatically increases from conventional confocal microscopy (4-5) up to 50
(Figure 8.2A). The imCyTOF technique works by vaporizing a 1µmx1µmx5µm “cyto-pixel”
and passing this ion cloud to an argon plasma chamber (Figure 9.2B,C). As it arrives in
the mass spectrophotometer detector (Figure 9.2D), a software algorithm (Cell Profiler or
289
HistoCat) reconstructs the image from precisely timed and aligned pulses from each
pixel(Bodenmiller et al., 2017). Using advanced non-linear dimensionality reduction and
clustering approaches(Becher et al., 2014; Bodenmiller et al., 2017) (Figure 8.2 E,F), we
can compare up to 100 different probes at a single cell level in the cortex of post mortem
human brains. Imaging CyTOF will afford unprecedented gene (protein, antibodies;
mRNA, in situ) expression resolution that is unparalleled by any other technique(Becher
et al., 2014; Mair et al., 2016; Newell and Cheng, 2016). While single cell seqFISH
techniques rival imCyTOF in spatial and transcriptome resolution(Shah et al., 2016),
single cell seqFISH is limited to 250 mRNA transcripts, whereas imCyTOF can tag to 100
markers, both protein and mRNA.

Imaging CyTOF Rationale

Importantly, traditional approaches (i.e., confocal microscopy and flow cytometry) used
to identify and phenotype innate immune cells do not capture the degree of complexity
within the myeloid system. What’s novel about this technology is that it can tease apart
CNS resident microglia from peripheral mononuclear phagocytes. In this dissertation, this
has been difficult to tease apart brain-resident microglia from infiltrating hematogenous
macrophages because they share many of the same markers. For example, IBA1 is
accepted by the field as a structural marker of reactive microglia. Yet, studies indicate
that infiltrating peripheral macrophages can enter the brain and adopt a microglial-like
morphology including expression of IBA1. As seen in Chapter 4, IBA1 immunoreactivity
does not guarantee microglial in origin. Furthermore, while CD11c is expressed by certain
subpopulations of microglia, this antigen is well-known to be expressed by peripheral
290
macrophages and dendritic cells. For example, macrophages canonically express CD11b
and F4/80; however alveolar macrophages do not express CD11b. Therefore, if we define
macrophages only expressing CD11b and F4/80, we risk losing the potential role of
alveolar macrophages in our analysis(Becher et al., 2014). Therefore, using imCyTOF,
we can assess the heterogeneous population of macrophages in a post-mortem brain
with regard to spatial location.

Results:

To optimize imCyTOF for neurological histology, we required a model with immune
infiltrates that has a stark difference between peripheral immune cells and brain-resident
immune cells. Although Multiple Sclerosis and Alzheimer’s disease brains have a
heterogeneous population of brain immune cells, where these immune cells come into
the brain and how develop a microglial-like morphology is not well understood. Therefore,
using ischemic stroke brains, we can use the penumbra, a region where the “shadow” of
cell death spreads, to identify immune cell populations within one image.

Comparing imaging techniques

Histological stains were performed on 5µm fixed serial sections from biopsied brain tissue
from stroke patients acquired from the Universität Göttingen Medizin. To determine
whether imCyTOF can produce images that are comparable to conventional H&E and
immunohistochemistry techniques (Figure 9.3), imCyTOF brain sections were stained
using MBP and we see the same characteristic fading of MBP between H&E, IHC, and
291
iMC. These initial results indicate that the gradual fading in signal intensity correlates to
conventional techniques, validating the signal that is generated by imCyTOF. Adding
more parameters to the sample complicates the analysis, therefore we wanted to assess
how imCyTOF images compared to immunofluorescence stains (Figure 9.4). We find that
using three colors, we can get the neurofillament/MBP gradient and cellular shape
(GFAP) found using immunofluorescence compared to imCyTOF, demonstrating the
accuracy that imCyTOF can provide. To ensure we can see regional differences in an
ischemic stroke biopsy, we imaged the penumbra and the necrotic region (Figure 9.5).
Furthermore, from the necrotic region, we observed a minimal amount of non-specific
antibody or auto-fluorescence associated with necrotic tissue that interferes with the
antigen signal and image resolution. Taken together, these figures illustrate the quality of
images generated by the imCyTOF using human ischemic brain tissue.

Resolving the High Dimensional landscape of an ischemic stroke brain

Since we demonstrated that the imCyTOF can produce images of comparable quality to
a confocal microscope at 20x magnification, we scaled up our antibody panel and imaged
in two regions of the brain, penumbra and necrotic regions. We can categorize these
antigens into 6 main categories: T-Cell (cyan), B-Cell (yellow), Monocyte (magenta),
Astrocyte (orange), Neuronal (green), and nuclear (Histone) (Figure 9.7). These
micrographs represent the average signal recorded per antigen. While we can see gross
differences in signal intensity between each of the signals, one method for analyzing
these images are to overlay several antigens at a time. However, between each region,
each overlapping image combination would generate an incomprehensible number of
292
images that would be difficult to reconcile a potential biological effect. Therefore, we
proceeded to process these images using computational means. Since each cell has a
nucleus, we decided to set a threshold to identify all nuclei above a given threshold
(Figure 9.6A) and size (Figure 9.6B). Once nuclei were established, positive nuclei are
color coordinated and given a number (Figure 9.6C). To estimate an immune cell size,
we estimated a 2-micron border (purple) around the nuclei (green) (Figure 9.6D). Cell
area was filled-in and given a number (Figure 9.6E). This template was applied to each
image and the intensity of each antigen was calculated per cell and intensity. These raw
intensities were then summed and created into FCS files.

With 20 parameters, it is difficult to assess what combinatorial pattern is useful for drawing
conclusions and to even see patterns within a data set. To reduce the dimensionality
using an supervised, machine learning algorithm, I used a t-stochastic neighbor
embedding (t-SNE) to generate plots placed similar expressing cells together, grouping
them by expression similarities (Figure 9.8). Since these t-SNE scatterplots were able to
separate the penumbra from the necrotic, this indicates that the both images contained
unique cell populations, restricted to certain regions of ischemic stroke. Furthermore, this
scatterplot also demonstrates that there are some cells of the penumbra overlap with the
necrotic region, suggesting a common cell-type found in both regions. Since we were
interested in cells of myeloid origin, we found that certain myeloid genes were restricted
to specific regions of the t-SNE scatterplots (Figure 9.9). For example, p22Phox intensity
was correlated with CD68, while CD16 overlapped with Vimentin, but not IBA1.
Furthermore, these gene expression patterns were restricted to either the penumbra or
293
necrotic regions of stroke. These results indicate that there are certain genes may be
restricted to mononuclear phagocytes and not IBA1+ cells. If there are some monocytes
that do not express IBA1 and are found in the brain, then the cell type could be utilized to
differentiate non-brain resident cells.

While t-SNE scatterplots can provide correlations between cell expression populations
with some regional specificities, we adopted CITRUS, another clustering algorithm that
depends on unsupervised, unbiased learning by using biological relevant metrics (Figure
9.10). In other words, this algorithm identifies clusters of cell populations that are
significantly different from another population. Comparing the penumbra and necrotic
regions, we generated a cladogram with clusters cell populations that are all myeloid in
origin. Of all IBA1
+
cells (black), we find that some cells do not express high levels of
CD68, p22PHOX, or S100A9, suggesting that these microglia could be homeostati
(purple). Furthermore, while there are IBA1
+
cells that express high levels of CD68 and
p22PHOX, there are some within these two populations that are Ki67
+
. These phagocytic,
KI67
+
, and IBA1
+
cells are proliferating, activated microglia. Lastly, there are cells that are
express Vimentin, a gene associated with activated, migrating monocytes. Additionally,
there are cells that are Vimentin
+
and CD16
+
, that are not Iba1
+
(green), suggesting these
are peripheral in origin.

Taken these data from CITRUS, we can categorize certain populations and identify where
these innate immune cells are located within the penumbra and necrotic regions. Taking
a few of these CITRUS populations and overlaying them onto imCyTOF reconstructions
294
(Figure 9.11), we find that the penumbra contains a larger area of phagocytic, proliferating
microglia (red) and a region of phagocytic HLA-DR microglia. Interestingly in the necrotic
region, we find the same proliferating, phagocytic microglia (red) in the necrotic region.
But one novel cluster that is not found in the Necrotic region are the heterogeneous
macrophages (green), that are not IBA1
+
.

Taking this a step further to reconcile the heterogeneity found in these cellular
populations, we find that within the penumbra we asked which populations within the
phagocytic, proliferating populations were CD16
+
and how many of these cells co-
localized with IBA1. We found that these unique cells were restricted to the penumbra
and not the necrotic regions (Figure 9.12). Further characterizing these cells that are
proliferating and phagocytic, we asked which cells were triple positive for CD16, IBA1 and
p22PHOX (Figure 9.13). We found that a majority of these cells were in the penumbra,
but a few were located in the necrotic tissue. Lastly, we wanted to ask whether we could
find cells that were not IBA1
+
that are Vimentin
+
and CD16
+
(Figure 9.14). Within these
heterogeneous macrophage populations (green), we found cells some of these cells are
located in the penumbra, however, a majority of these cells are located in the necrotic
regions. As a whole, these data illustrate that that imCyTOF is capable of differentiating
cells that are from the periphery by identifying innate immune mononuclear phagocytes
that are not IBA1
+
. However, there are better genes to identify microglia from tissue
invading macrophages, such as SALL1. However, this is a step forward to understanding
the heterogeneity of the immune system in the CNS.

295
Discussion:

The results presented in this chapter highlight a potential role for implementing imCyTOF
in the field of neuroscience. We find that data generated from the imCyTOF can be treated
as flow cytometry data, CyTOF data, or standard IHC. What is important to note in this
chapter is how the data was used, specifically, how CyTOF analysis methods were used
to elucidate populations in the imCyTOF reconstructed images. How imCyTOF can be
used to answer larger biological questions will take time because there were many
technological barriers before reaching this point. As described before, this technique tries
to tackle the Biological Heisenberg, for we will not be able to identify cell expression with
high fidelity, simultaneously with high resolution spatial information. For this discussion, I
will further describe the what the future direction could have been for this project and
where it could go.

From Chapter 5, we found that the human brain histological stains illustrated a
heterogeneous pattern. Additionally, associations from whole brain lysates described
part of the story regarding the intracellular signaling processes. However, we cannot
identify which cell type is p38
+
or pERK
+
. Therefore, the purpose of imCyTOF, would
have broken through the barriers that impede our understanding of the cellular
heterogeneity that exists in the diseased and cognitively normal brain. In other words,
combining the cell-type expression and spatial localization of signaling markers,
correlated with epidemiological studies, would allow the field to leap forwards. As of
writing this dissertation, this would be the first time, imCyTOF technology was be
applied on a human brain.
296

Figure 9.1: CyTOF captures the high-dimensional data sets and includes structural
information: Schematic diagram describes how imaging CyTOF technology incorporates the
two spectrums of biology, providing a large number of genetic expression data and high
definition structural resolution data.

297

Figure 9.2: CyTOF workflow: Schematic diagram describes the standard workflow for
imCyTOF. (A) antibodies are labeled with metal chelators. (B) tissue with stained with these
metal antibodies are ablated, one “cytopixel” at a time by a laser and (C) aerosolized in an
Argon-Helium plasma chamber. (D) as they enter the quadrupole to sort out organic matter, the
sample hits the detector. (E) As cytopixels enter, they are digitally reconstructed, and they are
computationally processed, and data is treated normal CyTOF data.

298

Figure 9.3: Neuro-architecture comparisons: Serial sections were stained using Hematoxylin
and Eosin, DAB staining for MBP, or imaging CyTOF. Dotted yellow lines indicates the
penumbra’s shadow. Scale bar represents 100µm.

299

Figure 9.4: imCyTOF comparison with immunofluorescence: Representative micrographs of
two serial sections of IHC and imCyTOF stained sections show the comparisons between the
image quality. Dotted white line represents the penumbra shadow. Scale bar represents 200µm.

300

Figure 9.5: Expression patterns of MBP and GFAP in 2 ischemic brain regions using iMC:
Micrographs from one tissue section in two regions, imCyTOF stained sections can resolve
differences between tissues using MBP and GFAP. Dotted white line represents the penumbra
shadow. Scale bar represents 100µm.

301

Figure 9.6: Image Processing Workflow: Cell Profiler generated images show the workflow of
(A) histone threshold, (B) histone size exclusion, (C) nuclear outline, (D) projected cell area, and
(E) generation of a cell template. Each cell color is an individual cell and data points are
recorded within that one cell. These data points are further converted to flow cytometry files for
further downstream processing. Insets are in the lower right-hand corner enlarging the area that
is marked in white.

302

Figure 9.7: High Dimensional Imaging leads to complex comparisons: Figure shows all the
parameters used on this tissue section. Colors are grouped by cell type: T-cells (Cyan), Myeloid
(Magenta), Glial (orange), Cellular structures (green), and nuclear (black).

303

Figure 9.8: Dimensional reduction shows distinct myeloid populations between regions:
t-SNE scatterplots shows the 20-parameter reduction of these two brain regions down to 2
dimensions. Similar groups of cells are clustered together. Gross patterns are seen between
penumbra and necrotic regions.

304

Figure 9.9: Mononuclear cellular expression intensities are regionally restricted: t-SNE
scatterplots overlay expression patterns of each point in this scatterplot for each mononuclear
cell parameter. These scatterplots do not provide any information on cellular location.

305

Figure 9.10: Heterogeneous myeloid populations are found between regions: CITRUS
generated cladogram identifies several main populations with heterogeneous patterns in each
cluster. Major clusters are drawn and labeled by color. Each colored circle is correlated with an
expression intensity for Iba1.

306

Figure 9.11: Myeloid expression clusters are localized in specific ischemic stroke
subregions: CITRUS generated cladogram clusters are overlaid on top of penumbra region
and necrotic regions to identify regions that contain the three clusters mentioned above.

307

Figure 9.12: CD16
+
and Iba1
+
cells colocalize in the penumbra but not in the necrotic
regions. CD16
+
and Iba1
+
cells are circled in white and both penumbra and necrotic regions are
compared. Scale bar resents 100µm.

308

Figure 9.13: More p22PhOX reactive Iba1
+
cells are located within the penumbra.
p22Phox
+
and Iba1
+
cells are circled in white and both penumbra and necrotic regions are
compared. Scale bar resents 100µm.

309

Figure 9.14: CD16
+
and CD68
+
cells located more in the necrotic regions. CD16
+
and
CD68
+
cells are circled in white and both penumbra and necrotic regions are compared. Scale
bar resents 100µm.
310
Chapter 10: Discussion

“ 兵者，詭道也。”

“All warfare is based on deception.”
Chapter 1 Laying Plans
Art of War, Sun Tzu

Dissertation Discussion

From all 8 chapters of this dissertation, one reoccurring theme behind the intersection of
Trem2 and C1q is that the immune response is not binary– it’s a spectrum. Among all the
immune responses that a macrophage an elicit, there is a coordination of one response
that out competes the others. In Chapter 2, I demonstrated that Drosophila melanogaster
ensheathing glia and hemocytes can evoke an Aβ-dependent immune response and that
the D. melanogaster brain shows evidence for immune tolerance. In Chapter 3, I used a
cerebral amyloidosis mouse model of Alzheimer’s disease (AD) and mice deficient in
Trem2 and/or C1q to show the shift in Aβ immunoproteostasis. In Chapter 4, I continued
characterizing these mice and found that the loss of Trem2 and C1q synergistically affects
pERK signaling in mononuclear phagocytes. In Chapter 5, I identified that Trem2
+

hematogenous macrophages enter the brain from periphery and acquire a microglial-like
phenotype. Chapter 6 describes a pSyk-pERK signaling axis that is C1q opsonized Aβ
dependent in peripheral macrophages. In Chapter 7, I show the difference between
Trem2-dependent tolerance in microglia and Trem-dependent training in macrophages.
Chapter 8 focuses on the biochemical interaction of Trem2 bound to C1q opsonized Aβ
and how this neuroimmune complex is dependent on Aβ aggregation state and
concentration. Lastly, Chapter 9 validates the use of imaging CyTOF in human ischemic
311
stroke brains, with implications for LOAD post mortem brains. Using multiple animal
models, these chapters are summarized in Figure 10.1, illustrating how Trem2 and C1q
can affect Aβ immunoproteostasis in AD, and how C1q opsonized Aβ signals through the
pSYK-pERK axis on peripheral macrophages.

The purpose of this final chapter/discussion will explore (1) the evolutionary framework
for Trem2 and C1q in immunoproteostasis, (2) Trem2 as an immune regulatory gene, (3)
immune tolerogenic environments, (4) upstream mediators of Trem2-C1q signaling, and
(5) future directions. These five sections should shed some light on future experiments
and hypotheses that are beyond the framework of this dissertation, beyond
immunoproteostasis. Additionally, this section provides evidence to support these
speculative hypotheses for how Trem2 and C1q can alter innate and adaptive immunity
in the brain.

Reflecting on Trem2’s evolutionary past

Unlike the evolution of the innate immune system, the TREM2 did not arise early in
Kingdom Animalia, rather later in the Phylum Chordata, class Actinopterygii(Stet et al.,
2005). Genomic sequencing of Teleost fish found a Trem2 homolog, with Major
Histocompatibility (MHC) class 1-like homology, NKp44(Allcock et al., 2003; Stet et al.,
2005). NKp44 is a receptor found on Natural Killer (NK) cells, which are innate lymphoid
cells that kill somatic cells, especially those that have been infected with viruses or
mycobacteria(Lanier, 2008; Carrillo-Bustamante et al., 2016; Spits et al., 2016). During a
viral infection, somatic cells downregulate a “self” identifying gene, MHCI and NK cells
312
sense the MHCI downregulation via NK receptors, thereby targeting these infected cells
for lysis or apoptosis(Lanier, 2008; Spits et al., 2016).

NKp44 belongs to a family of receptors known as Natural Killer receptor Complexes
(NKCs), which either induce an activating response (kill) or inhibitory response (don’t kill)
when encountering a cell (Figure 10.2). The mechanism behind how an NK cell attacks
is quantitatively determined from the number of “don’t-eat-me” verses “eat-me”
signals(Campbell et al., 2004; Spits et al., 2016). This mechanism indicates that the
number of receptors to sense both “eat and don’t-eat” signals should be numerous with
complex regulatory checkpoints(Carrillo-Bustamante et al., 2016). Furthermore,
educating NK cells prior to maturation is necessary to prevent mis-activation, and the NK
microenvironment shapes their responses to determine how an NK cell should responds
to a cell-type-specific and dependent manner(Höglund and Brodin, 2010; Pegram et al.,
2011; Stojanovic et al., 2013; Thomas, 2015).

While the divergence of Trem2 from NKp44 occurred further up in phylum Chordata, the
earliest genomic evidence for NKC was recorded in subphylum Urochordata with NK-like
cells and NK receptors (Figure 10.3)(Khalturin et al., 2003). These early NK-like cells
were the first cells that identified “self from non-self” through a series of MHCI-like
interactions through a highly-conserved signaling axis(Khalturin et al., 2003; Lanier, 2008;
Carrillo-Bustamante et al., 2016). As MHCI-like genes evolved, NK cell receptors followed
suit, giving a diversity of receptor ligands(Carrillo-Bustamante et al., 2016). While the
receptors may be different, the conserved, downstream signaling axis depends on
313
intracellular accessory proteins, such as DAP10 or DAP12 to initiate any action(Lanier
and Bakker, 2000; Jiang et al., 2002; Campbell et al., 2004; Stet et al., 2005; Barrow and
Trowsdale, 2006; Ivashkiv, 2011)(Figure 10.4). In comparison to DAP12 mediated
signaling in peripheral macrophages, Trem2 also signals through the DAP12-pSyk-MAPK
signaling axis. Like NK cells, this not only suggests a complex regulatory system
downstream of Trem2 signaling, but it also suggests that regulation through Trem2
mediated activation and inhibition is through this DAP12-pSyk axis.

Additionally, NKC and Trem2 evolutionary data suggesting these homologous receptors
regulate a functional inflection point, but the microenvironment is necessary, especially
for Trem2 expressing macrophages to engage in phagocytosis or tolerance. Mentioned
earlier, NK education first occurs in the bone marrow via structural MHCI exposure
(Thomas, 2015). As NKs mature and enter their tissue-specific residence, they become
re-educated to their environment, learning the normal levels of MHCI or “don’t eat me”
signals in their tissue-microenvironment(Höglund and Brodin, 2010; Pegram et al., 2011;
Thomas, 2015). NK education is an example of innate immune memory, which adopts a
similar mechanism to macrophage microenvironment-dependent maturation and
macrophage response training(Höglund and Brodin, 2010; Ayres and Schneider, 2012).
In this dissertation, the presented data supports this conserved signaling nexus
downstream of pSyk (Chapter 7), and macrophage training paradigm (Chapter 8) akin to
NK education. For example, inhibiting downstream regulators of Trem2 either increased
or decreased C1q opsonized phagocytosis. Inhibiting one arm (p38 MAPK
phosphorylation), provoked a feedforward response to increase pSyk activation,
314
promoting peripheral macrophages Aβ phagocytosis. This macrophage feedforward
signaling mechanism is also observed in NK cells, when an activated “eat-me” signal is
confirmed bi-directionally, licensing the NK to kill(Höglund and Brodin, 2010; Pegram et
al., 2011; Thomas, 2015; Spits et al., 2016).

Unfortunately, this section and this dissertation does not address the other ligands that
interact with Trem2, nor does it address the downstream signaling of these other Trem2-
ligand binding interactions. Whether this speculation is a phenomenon that’s infection-
specific or microenvironment-dependent, is not known. Additionally, TREM2 dysfunction
in LOAD that are not single nucleotide variants in LOAD are due to protein misfolding or
glycosylation(Kober et al., 2016; Song et al., 2016; Park et al., 2017). Whether TREM2
varied glycosylation is uniquely a macrophage-specific phenomenon compared to the
plethora of NK receptors has not been explored.

TREM2 the macrophage NCR (Trem1 vs Trem2)

Recent evidence has shown that macrophages can be trained(Wendeln et al., 2018).
Taking into consideration evolutionary entanglement of Trem2 and NCRs, does Trem2
become trained to specific ligands that are regionally specific, or are they more sensitive
to some ligands over others? While there is some diversity in Trem2 receptors(Allcock et
al., 2003; Klesney-Tait et al., 2006; Zheng et al., 2016), Trem2 has been shown to bind
dead cells(Hsieh et al., 2009), Apolipoprotein (Apoe)(Yeh et al., 2016), bacteria(N’Diaye
et al., 2009), Aβ(Wang et al., 2015b), and in this dissertation, C1q opsonized Aβ. One
hypothesis is how Trem2 could have differences in glycosylation and how maturation
315
could affect the avidity between Trem2 and its ligand(Kleinberger et al., 2014; Kober et
al., 2016; Schlepckow et al., 2017).

Unfortunately, this educationally-trained hypothesis is more difficult to experimentally
address. Another more parsimonious explanation is the competition between Trem1 and
Trem2 on macrophages(Sharif and Knapp, 2008). Trem1, like Trem2, is also derived from
NKp44 with a very similar function. While both require DAP12 as its intracellular
accessory protein, Trem1 depends on Immunoreceptor Tyrosine-based Activating Motifs
(ITAM), whereas Trem2 signals through an ITIM (Immunoreceptor Tyrosine-based
Activating Motifs)(Bouchon et al., 2000; Allcock et al., 2003; Colonna, 2003; Sharif and
Knapp, 2008; Roe et al., 2014). During an immune infection, Trem1 is found to potentiate
the inflammatory response, while Trem2 pushes for homeostasis(Bouchon et al., 2000;
Colonna, 2003; Sharif and Knapp, 2008; Colonna and Wang, 2016). Both proteins are
present in all mononuclear cells (e.g. dendritic cells and macrophages), however, TREM1
has not been identified on bona fide microglia, only TREM2. Interestingly, both TREM1
and TREM2 signal through Syk, but how wildtype macrophages calculate the integration
of their intracellular responses still remains as elusive as quantum entanglement.

Currently, the model used to study the competition between Trem1 and Trem2 are in
inflammatory bowel disease (IBD) and pneumonia(Genua et al., 2014). This is because
mononuclear cells express both Trem1 and Trem2 in both the lung and gut. While no
study has performed a double knockout of Trem1 and 2 in either of these models, Trem1
deficiency improves disease outcome, whereas Trem2 deficiency worsens(Bouchon et
316
al., 2000; Turnbull et al., 2006; Genua et al., 2014; Roe et al., 2014; Sharif et al., 2014b).
Regardless of how Trem1 or Trem2 interact with Dap12 in mononuclear cells, the immune
response, especially the macrophage response, is not binary– it’s a spectrum(Town et
al., 2005; Genua et al., 2014). Like NK activation and education depending on a
quantitative number of “kill vs don’t kill” signals(Pegram et al., 2011; Thomas, 2015), the
number ligands bound to Trem1 or Trem2 should determine which signaling pathways
dominate inside a macrophage to reset homeostasis, induce tolerance, or initiate
phagocytosis. The number of ligands should depend on the leukocyte-pathogen proximity
and the microenvironment (pathogen concentration and cytokine stimulation).

In this dissertation, I demonstrated how the peripheral macrophages enter the brain and
adopt a microglial-like morphology (Chapter 5). I also showed how the interaction of Aβ
and C1q-Aβ differ between microglia and macrophages in vitro (Chapter 6). In these two
chapters, Trem1 expression was not assessed because these were brain-specific cells
that don’t express Trem1. Additionally, Trem2 deficiency dramatically impaired C1q
opsonized Aβ phagocytosis in vitro, suggesting this Trem2-dependent. However, in
Chapter 6, I showed the how Aβ concentration and treatment duration affected Trem2
dependent macrophage training and tolerance in vitro. In these experiments, I saw how
Trem2 deficiency exacerbated TNF-α release and other pro-inflammatory genes. These
data hint at the potential for a Trem1 compensatory effect, for which data was not
presented in this dissertation. Lastly, in this Chapter 8, we also saw how tolerance shows
small changes in IL-10 and feedforward dependency on TGF-β. Interestingly, recent
317
studies on innate immune training showed that these two pathways needed for
training(Wendeln et al., 2018).

Not all TGF-β and IL10btolerogenic environments are immunoproteostatic
environments

With large handwaving generalizations, transforming growth factor-β (TGF-β) and
Interleukin 10 (IL-10) are canonically known as the anti-inflammatory cytokines that
regulate homeostasis. These cytokines are found in high concentrations in not only
immune privileged organs like the brain, testes, eyes, and ovaries, but also in non-
privileged organs, such as the lung and gut(Streilein, 1995; Matzinger and Kamala, 2011;
Benhar et al., 2012). Studies that block TGF-β signaling in the lung impair alveolar
macrophage (AM) survival, whereas the overexpression of TGF-β improves AM
clearance (Cui et al., 2003; Yu et al., 2017). Regards to IL-10, Macrophage IL-10
expression protects the gut in the experimental colitis, whereas the loss of IL-10
expression badly damages the gut (Hayashi et al., 2013; Dann et al., 2014). When we
examine the brain, we see that inhibition of IL-10 or TGF-β reduces pathology in AD
mouse models(Town et al., 2008; Chakrabarty et al., 2015; Guillot-Sestier et al., 2015a),
including microglial survival(Butovsky et al., 2013; Buttgereit et al., 2016); these results
mirror the studies performed in the gut or lung.

In relation to Trem2 and C1q in pneumonia, experimental evidence has shown that Trem2
deficient mice create a feedforward loop to produce C1q and that blocking pathways that
promote homeostasis, produces more C1q(Sharif et al., 2014b). Interestingly, Trem2
deficient mice also enhanced their bacterial phagocytic ability and enhanced C1q
318
expression(Sharif et al., 2014b). While these results contradict the conclusion in this
dissertation, Trem2 and C1q engage a feedforward response to regulate pathogen load.
To reconcile these differences found between Sharif and colleagues, and this
dissertation, are narrowed down to 3 reasons: tissue-microenvironment-dependent
macrophage maturation, Trem2 vs Trem1, and the blood-brain barrier(BBB). As
described in the Introduction and Chapter 5, myeloid progenitors migrate to their tissue
of residence and mature. AMs are different than brain-resident macrophages, because
AMs do not express CD11b, whereas brain-resident macrophages do(Becher et al.,
2014). Yet, by nomenclature, they are both macrophages and derive from the same
embryonic and hemopoietic stem cell pool(Amit et al., 2015). AMs express high levels of
Trem1, whereas brain-resident macrophages do not(Colonna, 2003; Sharif and Knapp,
2008). Lastly, the BBB isolates and restricts certain cell types from entering the
brain(Streilein, 1995; Matzinger and Kamala, 2011; Baruch et al., 2015a), thus creating
an immune tolerant environment.

While it’s important to note, not all immune tolerant micro-environments are immune
privileged, TGF-β and IL-10 are known to regulate immune tolerance(Matzinger and
Kamala, 2011; Ayres and Schneider, 2012; Stein-Streilein and Caspi, 2014). While
immune tolerance and privilege have been used interchangeably, their functional
difference is drastically different. Immunological ignorance or evasion can be confused
with tolerance; ignorance is an artifact of isolated immune systems in vitro and evasion is
an evolutionary advantage given to bypass the immune detection(Streilein, 1995; Ayres
and Schneider, 2012; Carrillo-Bustamante et al., 2016). These are important distinctions
319
because the immune system was designed respond to pathogens, making ignorance
artificial, whereas immunological tolerance implicates a symbiotic relationship. Immune
privilege goes a step further to protect vital organs with baseline immune tolerance
mechanisms.

To restate the definition for immune tolerance from Chapter 8, immune tolerance is the
execution of the immune response upon measuring pathogen load and the host’s
health(Streilein, 1995; Matzinger and Kamala, 2011; Ayres and Schneider, 2012). In the
spectrum of resistance to tolerance (Figure 10.5), the region shaded in blue is considered
tolerance, a measurement of an organism’s health to pathogen load (Ayres and
Schneider, 2012). As the pathogen load increases (moving yellow circle right), the health
of the individual stay the same, then the effect is considered a tolerogenic response. If
the load decreases and the host heath remain constant, then the effect is considered
resistance.

In the case for tissue-specific microenvironments that have specialized immune functions
(Figure 10.6), the tissue’s microenvironment defines how the immune system can
responds to pathogen at the moment of entry and according to the tissue stage (I, II, III)
that can occur (Streilein, 1995; Matzinger and Kamala, 2011). The first stage recruits an
educated innate immune response, second invites microenvironment-tailored effector
lymphocytes, and third licenses the immune system to kill(Streilein et al., 1997; Schwartz
et al., 1999; Matzinger and Kamala, 2011; Baruch et al., 2015a; Doty et al., 2015; Leung
et al., 2015). This is analogous to how frostbite works: cold extremities dilate capillaries,
320
long exposure to cold constricts capillaries, and this leads to extremity rejection to
preserve core temperature(Matzinger and Kamala, 2011). Since the interaction between
Trem2 and C1q may regulate more of the innate components within Stage I, evidence
does show implications for supporting adaptive immune functions(Bouchon et al., 2001;
Matzinger and Kamala, 2011; Hall and Agrawal, 2017). Each receptor Trem2 and C1q
independently has shown some evidence for adaptive immune interactions.
Acknowledging evolution, adaptive immunity arose after C1q-like proteins came into play,
but before Trem2 proteins came into play (Figure 10.7). Whether Trem2 evolved because
of necessity to enhance the bride between innate-adaptive immune functions, isn’t known,
but it is a good speculative thought experiment.

T-cell-mediated Tolerogenic environments

In this section, I will address possible links to adaptive immunity that are Trem2 and C1q
dependent during Stage II and Stage III(Matzinger and Kamala, 2011). From these
studies, I will try to speculate a potential mechanism. From this dissertation, I show how
Trem2 and C1q opsonized Aβ interact to produce a neuroimmune signaling complex on
brain-resident hematogenous macrophages. Beyond C1q-osponized Aβ phagocytosis
and degradation, it’s not entirely clear what happens to these degraded proteins, and if
Aβ is antigen presented to T cells.

A recent asthma study found that, upon allergen exposure using an OVA model, Trem2
+
,
Cd11c
+
dendritic cells (DC) enter the lung, engage with the OVA protein, and migrate to
the lymph nodes to generate T-Cells, TH2 and TH17(Hall and Agrawal, 2017). To reconcile
321
these findings, in this dissertation, we found that our brain-resident macrophages of
hematogenous origin adopt microglial-like characteristics, CD11c
+
and Trem2
+
, in the
TGF-β microenvironment of the brain. Also, we know that CD11c is classically recognized
as a dendritic cell gene, that is found in the brain(Bouchon et al., 2001; Town et al., 2008;
Keren-Shaul et al., 2017; Mrdjen et al., 2018a). Whether Trem2
+
, CD11c
+
cells are found
in the brain with TH2 an TH17 cells has not been reported.

Peripheral immunization studies in mice with Aβ found that immunization generated TH1
and TH2 cells(Cribbs et al., 2003). Furthermore, using more brain-derived immunization
studies, Aβ immunized mixed glial populations produced TH1, TH17, and TH2 T-
cells(McQuillan et al., 2010). In T-cell competition experiments, these mixed populations
of TH1 and TH17 were in fact regulated by TH2 cells, suggesting TH2 exerts a more
dominant role in their in vitro studies(McQuillan et al., 2010). Corroborating these TH2-
dependent results, studies in uveitis showed that inside immune privileged organs,
CD95L or FasL receptors line the epithelium to induce apoptosis in any cell that interacts
with it(Ferguson and Griffith, 2006; Matzinger and Kamala, 2011). A recent report
identified TH1s express more Fas than TH2, suggesting TH1 entry into the eye undergo
apoptosis more that TH2s(Fang et al., 2010; Matzinger and Kamala, 2011). These studies
indicate that Trem2
+
CD11c
+
brain macrophages could induce the formation of TH2s and
play a more dominant role in these immune privileged organs.

While Trem2
+
CD11c
+
cells can induce TH1, TH2, and TH17s, what about C1q? Recent
reports from the Tenner lab showed that when human DCs were exposed to C1q
322
opsonized apoptotic cells, DCs secrete IL-10, thereby diminishing TH1 and TH17 numbers
in co-culture experiments(Clarke et al., 2015). In the context of AD, if C1q is bound to Aβ
in the brain or in the periphery, would it produce TH2s or Tregs(Matzinger and Kamala,
2011; Baruch et al., 2015a)? More broadly, in the context of immune privileged organs,
are these T-cells the Stage 2 lymphocytes observed during immune tolerance? While
many of these experiments were conducted in vitro, we currently know that Tregs could
have a potential role in the pathogenesis of AD. Summarized in Figure 10.8, whether
Trem2 or C1q deficiency affecting the immunoproteasome and downstream adaptive
immune properties is an area that could impact the pathogenesis of LOAD.

What about us?

Figure 10.9 summarizes the work in this dissertation where I used 3 different animal
models to ask whether the interaction of both TREM2 and C1q correlate to the human
disease. I used the D. melanogaster model to characterize the immune functions that are
highly conserved. Using the rat model of Alzheimer’s disease-like pathology and cerebral
amyloidosis mouse model of AD-like pathology, we found that hematogenous
macrophages infiltrate the brain and then adopt a microglial-like morphology. Using
human post-mortem tissue, we found the same heterogenous pattern in the brains of the
transgenic rats and mice. This heterogenous pattern is important because in relation to
the human disease, this dissertation shows that suggests that TREM2
+
hematogenous
macrophages enters the brain and acquire microglial-like characteristics, that can lead to
tolerization in the brain microenvironment.

323
In addition to human brain histology, I used whole brain frontal cortex lysates from these
LOAD patients. I did not find any changes in p38 phosphorylation or pERK, yet I found
differences in pAKT between cognitively normal controls and AD. To reconcile this
discrepancy, we turned to the literature. While, there are some studies that found elevated
phospho-p38 and pERK signaling during later Braak stages(Sun, 2003; Swatton et al.,
2004), we did not find this correlation. When we assessed pAKT levels, our data matched
their signaling data where Liu and colleagues found a decreased pAKT in LOAD (Liu et
al., 2011). Interestingly, these patients that Liu and colleagues were assessing were more
diabetic leaning. Large genomic and epidemiological studies that assessed diabetes
within Latino and African American women found that among their sample populations,
they found an association of TREM2 and complement-related genes(Reiner et al., 2012).
Although these women were not given a cognitive assessment, studies already suggest
that there is an association between diabetes and cognitive decline/Alzheimer’s disease
(Medinas and Hetz, 2013; Munshi, 2017). Furthermore, the association between obesity,
inflammation, neuroinflammation, and proteostasis are becoming more
apparent(Medinas and Hetz, 2013; Park et al., 2015; Gao et al., 2017; Moser and Pike,
2017).

These epidemiological studies bring out an additional layer of complexity in this tripartite
relationship between Trem2, diabetes, and AD; it is the ethnic background of the sample
population being questioned. The breakthrough landmark studies for Trem2 were first
assessed in northern Europeans and Americans, of mostly Caucasian decent(Forabosco
et al., 2013; Guerreiro et al., 2013b; Jonsson et al., 2013; Pottier et al., 2013; Reitz et al.,
324
2013; Zhang et al., 2013; Li et al., 2015). When assessing other cultural groups, In the
Han Chinese population, we find that TREM2 variants for LOAD are not associated with
LOAD(Yu et al., 2014; Li et al., 2016; Tan et al., 2016; Wang et al., 2017), whereas
Spanish population is at risk for TREM2(Benitez et al., 2013). Furthermore, among all
East Asian populations, TREM2 variants that were highly associated with Caucasian
penetrance, showed no association among all East Asian Populations(Huang et al.,
2015). When comparing TREM2 across African Americans, TREM2 variants were
associated with LOAD risk, but these variants were different than the variants associated
with Caucasian penetrance(Jin et al., 2015). These epidemiological studies suggest that
TREM2 does not have a binary outcome for LOAD onset, rather it’s multifactorial.

For my final thoughts, we, as Homo sapiens, can be colloquially be described as “walking
knockouts,” with risk factors that in a unique combinatorial pattern that can be protective.
Alternatively, in the case of wild immunology or the study of human immunology, our
bodies have entered a state of systemic tolerance(Ayres and Schneider, 2012), especially
what we are environmentally exposed to everyday or how we behave, think, or feel(Ben-
Shaanan et al., 2016). In the case of combinatorial genetics, where overstepping an
allostatic threshold or equilibrium to predispose one to a disease(Thompson et al., 2009),
the conclusions from this study illustrates the need to understand how the relationship
between two deficient genetic risk factors intersect, not just one gene. In other words, risk
factors for AD are conventionally characterized independently, not simultaneously. In
LOAD, compound deficiency of two or more genes indicates the need to understand that
multiple pathways may depend on one another because biological deficiencies do not
325
function or exist in a vacuum. Moreover, these data also lay the foundation for future
therapeutic interventions that target two pathways by identifying a common signaling
pathway, further emphasizing that pharmaceutical interventions could have a more
beneficial effect if they target the one intersecting molecule. Although we assess the
interaction of Trem2 and C1q on brain-resident peripheral macrophages in Alzheimer’s
disease, the neuroimmune complex of Trem2-C1q-Aβ is not restricted to AD, as it could
have implications outside the brain. Thus, further investigation into this TREM2-C1q-Aβ
axis could reveal a fundamental mechanism that is found in synaptic pruning, neural
development, mononuclear heterogeneity, and global immune function.

326

Figure 10.1: Trem2 and C1q affect Aβ immunoproteostasis, Summary diagram illustrates
how deficiencies in Trem2 impact more soluble Aβ, whereas C1q impacts more insoluble. The
intersection (demarcated in red) illustrates a possible inflection point that occurs on
macrophages that converges on the Trem2-Syk axis to regulate C1q opsonized Aβ
phagocytosis.

327

Figure 10.2: NK cell activation: Schematic diagram shows an oversimplified diagram of how
NK cells sense when cells die or live based on “eat me” signals (green) or “don’t eat me” signals
(red). Top represents inhibition, indicating the target cell survives; bottom represents activation,
indicating the target cell dies.

328

Figure 10.3: Evolution of Trem2 and Natural Killer cells: NK cells evolved early during the
deuterostome bifurcation, whereas Trem2 evolved within Phylum Chordata around Class
Actinopterygii. Cladogram depicts the evolutionary trajectory of different Phyla in Kingdom
Animalia. Black represents the Phylum/Class name, grey lists an example species, and red
provides the known name of macrophage-like cells of that phylum. Orange box represents all of
phylum Chordata.
329

Figure 10.4: pSyk signaling axis is a highly conserved mechanism that determines
activation from inhibition: Trem2 is homologous to NKp44, a NCR that also signals through
Syk, ITIMs, and MAPKs. While the end results are slightly different, the signaling similarity is
very different than NKG2D receptors.

330

Figure 10.5: Defining Resistance vs Tolerance paradigm: Blue represents normal immune
function in response to a pathogen. Yellow circle denotes how an organism can respond to a
pathogen/Aβ and what the process will be defined as. Yellow regions denote the blurred lines
between resistance and tolerance.

331

Figure 10.6: Tolerance vs Immune privilege: Figure is modified from previous depictions of
resistance vs tolerance. Green circle represents “start.” Stages are colored as (1) blue, (2)
green and (3)white. Death is considered the darkened region below all circles. Red circles
denote fading health and it’s a one-way process to death if tolerance is unregulated. Green
arrows indicate a natural progression of disease severity.

332

Figure 10.7: Bridging Trem2, C1q, and adaptive immunity: Evolutionary comparison
between Trem2, C1q and adaptive immune systems. Cladogram depicts the evolutionary
trajectory of different Phyla in Kingdom Animalia. Black represents the Phylum/Class name,
grey lists an example species, and red provides the known name of macrophage-like cells of
that phylum. Orange box represents all of phylum Chordata.

333

Figure 10.8: Trem2-dependent adaptive immune signaling: A Trem2 macrophage is shown
in the center. As Trem2 macrophages become antigen trained, they can become APCs that
promote the genesis of TH1, TH2 and TH17s. C1q opsonized signaling meditates IL-10 secretion
and TH1/TH17 suppression. This promotes TH2 and TReg survival.

334

Figure 10.9: TREM2-C1q-Aβ regulates a fundamental signaling axis in brain-infiltrating
peripheral macrophages in AD: This dissertation is summarized in this final figure. (A) First
insert from the brain shows how brain resident macrophages that express TREM2 and have
mutations that cannot impair glycosylation or impair TREM2 protein structure/location. These
mutations can be culturally specific. (B) Secondly, hematogenous macrophages enter the brain
and develop into brain-resident macrophages, microglial-like cells. (C) Deficiency in TREM2 or
C1q signaling can impair Aβ immunoproteostasis and this impairment can be dependent on the
TREM2-pSYK-MAPK signaling axis. This dissertation has extensively characterized the novel
interaction of TREM2-C1q-Aβ upon the pSYK-pERK axis in macrophages.
335

336
Materials and Methods

Protein isolation and detection

Mouse or Human brain tissue samples were homogenized using 1x Cell Lysis Buffer (Cell
Signaling, #9803S) supplemented with protease and phosphatase inhibitors (Sigma
P5726-1mL, P8340-mL). Homogenized samples were centrifuged. Both fractions, pellet
and supernatant, were saved. Supernatant was then used as the primary protein source
for the BCA assay (Pierce, Thermo Fisher #23225). For tissue culture (human and
mouse), cells were homogenized and protein was detected in the same manner. Equal
amounts of protein was added into each lane, 10-20ug for cell culture and 50-100ug for
tissue homogenates. These samples were run on an SDS poly acrylamide gel and semi-
dry transferred onto nitrocellulose membranes (BioRad #1704158). Blocking for non-
phosphorylated antibodies was performed using 5% milk in PBS + .1% Tween and for
phosphorylated antibodies, 5% milk in TBS + 1% Tween. Primary antibodies incubated
overnight in 5% milk + PBS + .1% Tween or 5% BSA + TBS + .1% Tween for non-
phosphorylated antibodies and phosphorylated antibodies respectively. Blots were
washed with PBS or TBS. Secondary antibodies were incubated in 5% milk + PBS + .1%
Tween or 5% BSA + TBS + .1% Tween respectively. Bands were exposed using Clarity
Western ECL Blotting Substrate from BioRad according to the manufacture’s protocol
(#1705061). Antibodies used: a-Tubulin (1:3000, Sigma Aldrich, B-5-1-2), Trem2 (1:1000,
Abcam, RM0139-5J46), TREM2 (1:1000, R&D Systems, AF1828), TREM2 (1:100, Novus
2B5), Amyloid Precursor Protein/Aβ (1:3000, Covance, 6E10), P-Syk(1:1000, Cell
337
Signaling, C87C1), Syk (1:1000, SCBT, 4D10). Donkey anti goat IgG-HRP (1:3000,
SCBT, sc2020), Amersham Sheep anti mouse IgG-HRP (1:5000, VWR, 95017-324),
Amersham Sheep anti rabbit IgG-HRP (1:5000, GE, NA934).

Histology

Mice were sacrificed at 6 months of age post behavioral assessment. Mice were
sacrificed under isoflurane and perfused using 1x ice cold PBS. Brains were quartered
and one quarter was fixed in 4% PFA (Electron Microscopy Sciences, 15714-S) overnight
in 4ºC. Fixed quartered brains were then transferred into PBS and washed overnight at
4ºC. Finally, fixed and washed brains were dehydrated in 70% ethanol in preparation for
paraffin embedding. Sections were cut on the microtome at 10uM sections, mounted on
slides using a water bath, and dried overnight at 37ºC. Dried slides were deparaffinized
for 5 min in 100% Xylene (Sigma), 100% Xylene, 100% Ethanol (Fisher Scientific), 95%
Ethanol (KOPTEC USP), 80% Ethanol (KOPTEC USP), 70% Ethanol (KOPTEC USP),
and 50% Ethanol (KOPTEC USP). Slides were then washed 3 times in 1x PBS.

Slides that underwent Thioflavin S staining were incubated in a 1%w/v solution of
ThioflavinS in ddH2O. Slides were then washed three times in 70% ethanol for 10 min
each, followed by one ddH2O wash. Slides were then allowed to dry overnight. DAPI
mounting media from Prolong Gold antifade reagent (LifeTechnologies, P36935) were
added on each tissue sample, mounted with a cover slip, and dried for confocal imaging.

338
Slides that required antibody incubation underwent an antigen retrieval protocol in a 6.0
pH citrate solution (DAKO) for 30 min at 90-95ºC. Slides were then washed in PBS 3
times before blocking for 30 min with the blocking buffer: 10%NDS (Jackson Immuno), 1x
PBS, and .3% Trition. Primary antibodies were added into the blocking buffer and
incubated overnight at 4ºC, washed 3 times with PBS + .3% Trition, incubated in
florescent 2º antibodies (Jackson Immuno), and washed 3 times in 1x PBS + .3% Triton.
DAPI mounting media from Prolong Gold antifade reagent (LifeTechnologies, P36935)
were added on each tissue sample, mounted with a cover slip, and dried for confocal
imaging.

Nanoparticle Injection

5-6 month old APP/PS1 mice were injected with Coumarin encapsulated PGL-A
Nanoparticles (NP-C) that were generated and donated by Tarek Fahmy’s lab. Mice were
given weekly injections with 100uL of NP-C via the tail vein for 3 weeks. After 3 weeks,
mice were sacrificed and thoroughly perfused with ice cold PBS. Brains were processed
for lymphocytes using Percoll based separation. Cells were stained with CD45 and
Trem2. Cells were sorted for CD45+ and C6
+
or C6
-
cells using the AriaII. RNA was
isolated using the RNeasy Qiagen Micro Kit (74004) as per manufacturer’s instructions.
cDNA was isolated using 5x iSCRIPT (1708891) as per manufacturer’s instructions.
Sybr-Green (1725151) was used to for RTPCR experiments and primer sequences are
listed in the supplementary figure table 1.

339
Cell Culture

In vitro studies performed on immortalized cell lines were kept at 37ºC in a humidified
incubator with 5%CO2. C1q dissolved in glycerol (CompTech, A100) was added on to
cells at a concentration of 4µg/mL and Aβ was added onto the cells at .2µg/mL. For
conditions, as with controls heat inactivated C1q, C1q was incubated in the same glycerol
solution and incubated for 20 min at 56ºC. All samples were stored in the -80ºC and
thawed on ice prior to use. For conditions where C1q opsonizes Aβ, we incubated the
complex for 20 min as described by previous studies(Webster et al., 1995; Galvan et al.,
2012; Bohlson et al., 2014). This 20-min incubation also applies for the heat inactivated
C1q opsonized Aβ prior to cell treatment. For drug treatments, 10x the IC50 was used as
the best concentration for all inhibition experiments. (All drugs were from Sigma Aldrich:
Syk inhibitor V, 574715; p38 MAPK inhibitor, SML0543; U0126 monoethanolate, U120;
SB220025 trihydrochloride, S9070). Prior to all cell signaling experiments, cells were
serum starved in OptiMEM (Gibco) overnight. All treatments for all experiments were
performed in OptiMEM and added onto the cells. Incubation times varied depending on
protocol. For SV40 microglia, 5 hour incubations were performed as recommended by
previous studies(Webster et al., 2000; Galvan et al., 2012; Bohlson et al., 2014).
However, for the signaling experiments performed in vitro, performed a kinetics
experiment (data not shown), which showed optimal phagocytosis between 1-2 hours.
Thus, all signaling experiments were performed at 2 hours. For downstream analysis,
FACS or ICC, protocol as described in their respective methods section.

340
Immunocytochemistry

Primary cells and immortalized cell lines were initially seeded onto cover slips. Post
treatment, they were removed from the 37ºC Incubator (5%CO2) and washed with 37ºC
sterile PBS. Ice Cold 4% PFA + 1x PBS was added to the cells and incubated on ice for
10 min. Fixed cells were washed 3 times with 1x PBS and subsequently blocked using
our blocking buffer mentioned above for 1 hour. Primary antibodies were then added into
1xPBS + .3% Triton and incubated overnight at 4ºC in the dark. Afterwards, the primary
antibodies were removed, slides were washed three times in 1x PBS + .3% Triton and
incubated in secondary antibodies for an hour at room temperature on an orbital shaker
using the same buffer. The coverslips were then washed with the wash buffer and rinsed
in PBS. Prior to mounting using the Prolong Gold anti-fade reagent, slides were lightly
dried. Slides were then dried overnight in the dark and imaged on the confocal microscope
the following day. C1q (1:100, Abcam, 4.8), CD68 (1:100, Abcam, FA-11 and KP1),
P2Y12 (1:100, Abcam, EPR18611)

Size Exclusion Chromatography - Multi Angle Light Scattering (SEC-MALS)

Samples were prepared in 100uL volumes using equimolar concentrations. Samples
were prepared in ddH2O on ice. Samples were then incubated for 24 hours on ice and
then ran on the HPLC-MALS. Protein samples used in these experiments are TREM2
(Abcam, ab153366), Aβ42-1 (AnaSpec EGT, AS-27276), Aβ40-1 (AnaSpec EGT, AS-
341
22817), Aβ1-42 (AnaSpec EGT, AS-24224), Aβ1-40 (AnaSpec EGT, AS-24235), and C1q
(CompTech, A099). Aβ was aggregated by incubating at 37ºC for 24 hours.

Protein samples were loaded into the Shodex HLPC Column (KW-802.5) and were
transferred into the HPLC (Agilent Technologies 1200 series) under the flow rate of
1mL/min in 1xPBS + Azide, under an Iso Pump Pressure between 15-30Bars. Post HPLC,
samples were transferred into the MALS instrument from Wyatt Technology, Dawn Helios
and Optilab. Samples were recorded under a duration of 25mL and measurements were
recorded from the scattered the light, the UV spectral data, and the differential refractive
index. Data was analyzed using ASTRA software 6.0, developed by Wyatt Technology.
Control reference protein used in this study was BSA (Sigma Aldrich). Columns, filters,
and capillaries were washed in 1x PBS+ Azide, ddH2O, and 20% Ethanol.

Circular Dichrosim

Aggregated Aβ samples used for MALS were subsequently subjected to the Jasco J815.
Samples were diluted in DPBS (10nM) and placed into a .1mm cuvette. Cuvette was
washed with a detergent, ethanol, and water for a few times in succession. Once dried,
sample was added. Instrument was flooded with N2 gas prior to running the samples with
a flow rate of 10 psi. All samples were normalized to the DPBS buffer. Results displayed
are an average of 5 consecutive runs and plotted on molar elipticity 𝜃 [deg•cm
2
•dmol
-1
]
by wavelength [nm].

342
Behavioral Testing

Mice were raised and housed according to IACUC protocol, i.e. ad lib food and 12/12
light-dark cycle. Mice were raised up to 5.5-5.75 months of age for behavioral testing and
sacrificed at 6 months. For clarification, APP/PS1 mice are from Jankowsky and Borchelt.

At 5.5-5.75 months, mice were first given neurological tests to ensure no motor or sensory
impairments. Mice that failed any of these initial neurological tests were removed from all
behavioral analysis. Mice that passed these tests were then subjected to the open filed
maze, Y-maze, and novel object, with one test per day. Prior to any test, mice were
habituated into the testing room for an hour.

Open field maze consisted of a 30-min trial in an 70% ethanol-cleaned chamber and
ethanol was evaporated prior to the start of the trial. Mice were placed in the center of the
maze in a brightly lit room. The tester left the room during every trial. Tracking software
used was Ethovision and area of the perimeter was determined as only 10% of the area
at the perimeter. Manual corrections were made to ensure tracking was correct and
reflections were also corrected.

Y-maze was performed in a chamber that where the corridors are placed at 120º angles.
Prior to every test, the chamber was 70% ethanol cleaned and allowed to evaporate. Mice
were placed in the same corridor and recorded for a total of 5 min. No restrictions were
343
made on the number of arms. However, animals that did not pass 12 arms were removed
from the study.

Novel Object experiments were performed in brightly-lit, 70% ethanol cleaned chambers
and ethanol was evaporated prior to the start of the test. Two identical items, made of the
same material, were used to pre-expose the mouse during the ‘training’ phase for 3 min.
After 3 min, chambers were cleaned and a novel object, of a different material and shape,
was placed into the chamber in the same location as one of the ‘older’ objects. After 20
min, mice were re-introduced into the chamber and was recorded for 3 min. During the
entire recording phase, the behavioral tester was in the room. For analysis, Ethovision
recorded the mice and tracked nose, body, and tail. If a perimeter was made around the
objects and an outer area expanded to 150% of the object. If nose or body entered object
center or object outer area, time was recorded.

Peripheral macrophage isolation

Mice without the APP/PS1 transgene were raised up to 5.75 months of age before
injecting with 1mL of a .5M sterile thioglyolate solution I.P. under isoflurane anesthesia.
At 6 months, mice were sacrificed according to IACUC protocol. To isolate peripheral
macrophages, the abdominal cavity was opened while keeping the peritoneum intact.
10mLs of sterile, 0ºC DBPS(GIBCO) was injected into the peritoneal cavity and shaken
for 1 min to isolate the macrophages. Macrophages were centrifuged at 500gs for 5 min,
washed with cold DPBS, and counted for plating in 10% FBS, 1% Penicillin/Streptomycin,
344
1% L-Glutamine, and high glucose DMEM. Cells were initially starved using OPTIMEM
for 24hrs and treated with florescent aggregated Aβ, or C1q opsonized florescent
aggregated Aβ(Paresce et al., 1996) (ratios were 1nM:1nM) for 2 hours. For P-MAPK
signaling experiments cells were treated with non-fluorescent Aβ, or C1q opsonized non-
fluorescent Aβ (ratios were 1nM:1nM) for 2 hours.

Microglial isolation

Mice were scarified and perfused at 5.5-6 months. Brains were harvested and quickly
processed using the neural dissociation kit (Miltyni) according to the manufacturer’s
instructions. Cells then washed with 20% FBS, 1% Penicillin/Streptomycin, 1% L-
Glutamine, and high glucose DMEM. Cells were counted and plated for in vitro
experiments. Microglia were then used after 5-7 days after harvesting with a daily media
change. Cells were initially starved using OPTIMEM for 24hrs and treated with florescent
Aβ, or C1q opsonized florescent aggregated Aβ(Paresce et al., 1996) (ratios were
1nM:1nM) for 2 hours.

Flow Cytometry

Peripheral macrophages were scraped from tissue culture plates, fixed, and
permeableized (BD GolgiStop). Cells were washed and stained using our solution of .1%
Saponin, PBS, 2% BSA, and .1% azide. All signaling antibodies were labeled with Rapid
conjugation kits from Abcam. (pERK, p-p38, and pSyk, were purchased from CST.
345
moCD45, moCD11b, and CD68 FA-11, Biolegend). For in vivo p-ERK experiments 2uLs
of whole cerebrum protein lysate was used for CST’s cell proliferation assay (CST).

Confocal Microscopy & Imaris Bitplane

Confocal images were taken on in vitro experiments with peripheral macrophages or
microglia. Using a Nikon 1A confocal microscope for the images, cells were imaged at
60x in stacks of .125microns apart. Fluorescent levels were standardized and saved using
the Nikon’s Image Acquisition platform. Images were taken to Imaris for 3D
reconstruction. Thresholds were standardized, recorded, and % voxel overlap were
acquired following Imaris’ q3Dism platform. Values acquired were then used for
downstream statistical analysis in Graphpad Prism. Further description of q3Dism can be
found in Gulliot-Sestier, JOVE 2015(Guillot-Sestier et al., 2016).

Statistical Analysis

Graphpad Prism 7.0 was used to generate graphs and data. One-way ANOVA with
Bonferroni correction and outlier (Q-test) removal was used to analyze data. Unless
mentioned, Two-way ANOVA was the standard for all experimental statistics, followed by
T-Test with multiple comparisons were used if 3 or more variables were being compared,
otherwise standard T-test (two-tailed) was used for 2 variables. For all two-way ANOVAs
with multiple corrections, Tukey was used, unless otherwise stated. All statistically
significant values are detonated by † p<0.1; * p<0.05 ; ** p<0.01; *** p<0.001
346

Cell Profiler Analysis

IHC analysis on mouse brains were calculated using freeware, Cell Profiler. % area was
calculated using a learning algorithm with a threshold setting from 1.00 to 0.550. Size
exclusion for the % area calculation was set at 15pixels. Integrated % area values were
used to analyze all the images with at least 3 images per animal and from 4-8 animals
per genotype (sex-matched). Calculations were done for ThioflavinS, 6E10, GFAP, and
Iba1. % Area was also validated by hand to ensure computational accuracy and data
matching.

ELISA

MSD Aβ ELISA kit was used to measure the amount of Aβ species present in the whole
cerebrum lysates from the transgenic mice. Triton soluble Aβ was diluted 1:10 and
guanidine soluble fraction was diluted 1:1000. Protocol and analysis was measured
according to manufacturer’s instructions. To measure Aβ Oligomers, 82E1 ELISA kit was
used as per manufacturer’s instructions (IBL). Triton Soluble cell lysate samples for the
Oligomeric Elisa were diluted 1:10 prior to use. For the C3 mouse EILSA, brain lysates
were diluted 1:4 and ELISA kits directions were followed as described by Abcam.

347
D. melanogaster experiments

All flies are on the WT
1118
background and fly lines were obtained by the Chang Lab,
Tower Labs, or Bloomington Fly Stock. Flies were on a cornmeal-molasses diet. Flies
were bred with 1 male and 3 virgin females. Longevity experiments were mixed with males
and females, mixed together and tubes were changed every other day. For all western
immunoblot experiments, 10 fly brains (sex matched) were homogenized with a mortar
and pistol in lysis buffer. For RNA isolations, single fly brains were homogenized with
mortar and pistol in Trizol. RNA isolations were performed as described above. Relish
antibody was acquired from Prof. Yiva Engström (Stockholm University). Drosophila
phagocytosis experiments were extracted from W1118 larvae and lysed in sterilized
Schneider’s media with preaggregated Aβ. Cells were fixed and stained according to the
ICC protocol mentioned above.

Imaging CyTOF

Ischemic Stroke brains were obtained from the Universität Medizin Göttingen with
approved IRB approval. Brains were fixed and sectioned and stained with antibodies.
These antibodies were labeled with Fluidigm labeling kits Table listed below. After
sections were stained with the antibody, tissue sections were taken go the Bodenmiller
lab and were counterstained with a Rubidium dye. Images were acquired with the Helios
imCyTOF and data was analyzed as described in Giesen and colleagues. Data obtained
was then processed using the software IMCReader. imCyTOF images are converted into
TIFFs and processed using Cell Profiler. Histone was used as the template by identifying
348
the nucleus and threshold was set at (.55). Size exclusion selected as larger than 5 units
and smaller than 20. Cells boundaries were established by adding 2 extra units around
the circumference. Cells were labeled and overlaid on all images. Cell identifiers were
assigned and produced an CSV file, which was converted into an FCS file. Rstudio was
used to process the FCS using the Flowcore package. CITRUS and Wonderlust
packages were also acquired from github for this analysis.

349
Table of All Primers

Gene Forward Primer Reverse Primer
Sall 1 GACATCCCCAGTTCTGCTCC ACCTCGCCGCTAGATCCTTC
Maf B TGAGCATGGGGCAAGAGCTG CCATCCAGTACAGGTCCTCG
Irf-8 CGTGGAAGACGAGGTTACGCTG GCTGAATGGTGTGTGTCATAGGC
MyB TGTCCTCAAAGCCTTTACCG TTCACGTATTTCCGAGCCG
Trem2 CTATGACTCCATGAAGCA GATTCCGCAGCGTAATGG
C1q a CGGAATTCGACAAGGTCCTCACCAACCAG CGGGATCCGGGGTCCTTCTCGATCC
APP CAC CAC AGA GTC TGT GGA AG AGG TGT CTC GAG ATA CTT GT
Actin TCCCTGGAGAAGAGCTACGA AGGAAGGAAGGCTGGAAGAG
mCd11b(Itgam) ATGGACGCTGATGGCAATACC TCCCCATTCACGTCTCCCA
mC1qa AAAGGCAATCCAGGCAATATCA TGGTTCTGGTATGGACTCTCC
mApoE CTGACAGGATGCCTAGCCG CGCAGGTAATCCCAGAAGC
dTEP1 AGTCCCATAAAGGCCGACTGA CACCTGCATCAAAGCCATATTG
dTEP2 TGTTCTGCACCAACAGCGATAC CTGGCGATCCATCAACATTCTT
dTEP4 GCTGCAGAACCAGATCGAAATC ATGACTTTGGCGACGTCTTGAT
dTEP6 CGCCTTCCTGAACGAAACAA GAGGCTTATCGGTCTGCACAA
dSHARK TTGATTGGCATTGCGAAGGG TGTGACGCCCAAACTTTCAG
dLamp1 ATCACTGCACTCGTGTTCAG ATCGCAATCGTGTCCAACTG
dRelish ACCAAATGGCTTGCCAAACC TTGCGGAAATCTTCGGAAGC
dCactus AGGCAACAGCAACAAGTGAC TTGTTAATGGCGGCCGTTTG
dHop ATTTAGGATCGTGCGCAAGG TGAATGATGTGCGCCACTTC
dMarelle TGCTTGAACAGGTGCAGAAC TTGTTGTGAATGCCCAGCTG
dDefensin ATCGCTTTTGCTCTGCTTGC ATCCTCATGCACCAGGACATG
dDiptericin TTCATTGGACTGGCTTGTGC AGGGGCACATCAAAATTGGG
dDrosomycin AGTACTTGTTCGCCCTCTTCG ACCCTTGTATCTTCCGGACAG
dUnpaired2 ACCTCGAAAACTTGCGGAAC TTAGCTTCACCGCACTTGTG
moIL6 CTCTGGGAAATCGTGGAAAT CCAGTTTGGTAGCATCCATC
mo_CD11c CTGGATAGCCTTTCTTCTGCTG GCACACTGTGTCCGAACTC
Sall-1 mouse GAC ATC CCC AGT TCT GCT CC ACC TCG CCG CTA GAT CCT TC
Ms_LPL GGAGAAGCCATCCGTGTGAT CTCAGGCAGAGCCCTTTCTC
Ms_FATP1 GGCGTTCTGTGTGTACGTGG CGAATCAGAACAGAGAGGCCA
Ms_DGAT1 TAGAAGAGGACGAGGTGCGA GTCTTTGTCCCGGGTATGGG

350
Table of All Regents

Reagent Name Company
Reference
Number
Notes
Clone
Number
Cell Lysis Buffer
Cell Signaling
Technology
#9803S 1x

Pierce BCA Protein Assay Kit Thermo Fisher #23225

Trans-Blot Turbo Mini
Nitrocellulose Transfer Packs
BioRad #1704158

Clarity Western ECL Blotting
Substrate
BioRad #1705061

alpha-Tubulin Antibody
Sigma Aldrich - EMB
Millipore
B-5-1-2 1:3000

TREM2 Antibody R&D Systems AF1828 1:1000 2B5
TREM2 Antibody Novus 2B5 1:100

Amyloid Precursor Protein/Aβ
Antibody (6E10)
Covance SIG-39300 1:3000

P-Syk Antibody
Cell Signaling
Technology
C87C1 1:1000

Syk Antibody (4D10) SCBT sc-1240 1:1000

Synaptophysin Antibody Abcam ab23754 1:1000

donkey anti goat IgG-HRP
Antibody
SCBT sc2020 1:3000

Amersham Sheep anti mouse
IgG-HRP Antibody
VWR 95017-324 1:5000

Amersham Sheep anti rabbit
IgG-HRP Antibody
GE NA934 1:5000

Paraformaldehyde (PFA)
Electron Microscopy
Sciences
15714-S 4%

Xylenes
Sigma Aldrich - EMB
Millipore
534056-4L 100%

Ethanol KOPTEC USP V1001 100%

Prolong Gold antifade reagent LifeTechnologies P36935 10%

Normal Donkey Serum Jackson Immuno

10%

C1q Antibody Abcam ab182451 1:100 4.8
CD68 Antibody AbCam ab955 1:100 KP 1
CD68 Antibody Abcam ab53444 1:100 FA-11
P2Y12 Abcam EPR18611 1:100

TREM2 Abcam ab153366 1:100

Aβ42-1 AnaSpec EGT AS-27276

Aβ40-1 AnaSpec EGT AS-22817

Aβ1-42 AnaSpec EGT AS-24224

Aβ40-1 AnaSpec EGT AS-24235

Beta Amyloid 1-42 HiLyte
Fluor 555
AnaSpec EGT AS-6048001

C1Q CompTech A099

Neural Tissue Dissociation Kit Miltenyi Biotec 130-092-628

mouse CD45 Antibody Biolegend 147711

I3/2.3
351
CD68 Antibody Biolegend 137001

FA-11
Cell proliferation Assay
Cell Signalling
Technology
#6813

82E1 ELISA kit IBL 27725

Opti-MEM Gibco 11058-021

DMEM Gibco 11965-092

ThioflavinS
Sigma Aldrich - EMB
Millipore
T1892

Target Retreival Solution,
Citrate pH 6
DAKO #S236984

p38 Phospho CST #9211 1:100

pERK1/2
Cell Signalling
Technology
#9101 1:100

ERK1/2
Cell Signalling
Technology
#9102 1:100

p38
Cell Signalling
Technology
#9212 1:100

Thioglycolate
Sigma Aldrich - EMB
Millipore
211716

Fetal Bovine Serum
Sigma Aldrich - EMB
Millipore

Penicillin/Streptomycin Gibco 15140122 1%

L-Glutamine Gibco 25030081 1%

high glucose DMEM Gibco 11965092

Iba1 WAKO 019-19741

NCNP24
GFAP DAKO M0761

6F2
Dimethyl Sulphoxide Hybri-
Max
Sigma Aldrich - EMB
Millipore
D2650

BD Cytofix/cytoperm Plus BD Biosciences 554715

SuperSignal West Femto
Maximum
Thermo Scientific 34096

Cell Strainer 70uL Falcon 352350

Syk Inhibitor V - Calbiochem
Sigma Aldrich - EMB
Millipore
574715

p38 MAP Kinase Inhibitor IV
Sigma Aldrich - EMB
Millipore
SML0543

U0126 monoethanolate
Sigma Aldrich - EMB
Millipore
U120

SB 220025 trihydrochloride
Sigma Aldrich - EMB
Millipore
S9070

Akt1/2 kinase inhibitor
Sigma Aldrich - EMB
Millipore
A6730-5MG

GSK690693
Sigma Aldrich - EMB
Millipore
SML0428-5MG

Protease Inhibitor (100x)
Sigma Aldrich - EMB
Millipore
P8340-1ML

Phosphatase Inhibitor (100x)
Sigma Aldrich - EMB
Millipore
P5726-1ML

352
Table of all CyTOF Antibodies

Metal Tag Target Antibody Clone Dilution
Y89 CD45 HI30 1:200
Nd148 CD16 3G8 1:200
In113 Histone 9715 1:200
Pr141 Iba1 019-19741 1:200
Nd143 Olig2 18953 1:200
Nd144 Amyloid Beta 6E10 1:200
Nd145 GFAP Z0334 1:200
Sm149 MPO 59A5 1:200
Nd150 S100A9 S36.48 1:200
Eu151 MPO 59A5 1:200
Sm154 iNOS NB300-605 1:200
Gd155 p22Phox FL-195 1:200
Tb159 TREM2 AF1187 1:200
Dy161 Glutamine Synthetase EPR16661 1:200
Dy162 Ki67 MIB-1 1:200
Dy163 Vimentin 3B4 1:200
Ho165 ALDH1L1 ab87117 1:200
Er166 CD138 MI15 1:200
Er167 S100 Z0311 1:200
Er170 CD3 OKT3 1:200
Tm169 CD33 WM53 1:200
Yb171 CD68 PG-M1 1:200
Yb173 MBP A0623 1:200
Yb174 HLADR CR3/43 1:200
Lu175 CD235ab HIR2 1:200

353

354
References

Aguzzi A, Barres BA, Bennett ML (2013) Microglia: scapegoat, saboteur, or something
else? 339:156–161.
Allcock RJN, Barrow AD, Forbes S, Beck S, Trowsdale J (2003) The human TREM
gene cluster at 6p21.1 encodes both activating and inhibitory single lgV domain
receptors and includes NKp44. Eur J Immunol 33:567–577.
Alvarez-Errico D, Vento-Tormo R, Sieweke M (2015) Epigenetic control of myeloid cell
differentiation, identity and function. Nat Rev … 15:7–17.
Alzheimer A (1907) Über eine eigenartige Erkrankung der Hirnrinde. Allg Zeitschrift fur
Psychiatr und Psych Medizin 64:146–148.
Amit I, Winter DR, Jung S (2015) The role of the local environment and epigenetics in
shaping macrophage identity and their effect on tissue homeostasis. Nat Immunol
17:18–25.
Arandjelovic S, Ravichandran KS (2015) Phagocytosis of apoptotic cells in
homeostasis. Nat Immunol 16:907–917.
Arends YM, Duyckaerts C, Rozemuller JM, Eikelenboom P, Hauw JJ (2000) Microglia,
amyloid and dementia in Alzheimer disease: A correlative study. Neurobiol Aging
21:39–47.
Ausubel FM (2005) Are innate immune signaling pathways in plants and animals
conserved? Nat Immunol 6:973–979.
Awasaki T, Lai S-L, Ito K, Lee T (2008) Organization and Postembryonic Development
of Glial Cells in the Adult Central Brain of Drosophila. J Neurosci 28:13742–13753
Available at: http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.4844-08.2008.
Ayres JS, Schneider DS (2012) Tolerance of Infections. Annu Rev Immunol 30:271–294
Available at: http://www.annualreviews.org/doi/10.1146/annurev-immunol-020711-
075030.
Bachmann MF (2002) TREM-2: a novel link between dendritic-cell maturation and T-cell
priming. Trends Immunol 23:10 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S147149060102124X [Accessed April 16,
2018].
Bainton RJ, Tsai LTY, Schwabe T, DeSalvo M, Gaul U, Heberlein U (2005) moody
encodes two GPCRs that regulate cocaine behaviors and blood-brain barrier
permeability in Drosophila. Cell 123:145–156.
Bakkenist CJ, Kastan MB (2003) DNA damage activates ATM through intermolecular
autophosphorylation and dimer dissociation. Nature 421:499–506.
Banerjee S, Bhat MA (2007) Neuron-Glial Interactions in Blood-Brain Barrier Formation.
Annu Rev Neurosci 30:235–258 Available at:
http://www.annualreviews.org/doi/10.1146/annurev.neuro.30.051606.094345
[Accessed April 22, 2018].
Barrangou R (2015) The roles of CRISPR-Cas systems in adaptive immunity and
beyond. Curr Opin Immunol 32:36–41.
Barres BA (2008) The Mystery and Magic of Glia: A Perspective on Their Roles in
Health and Disease. Neuron 60:430–440.
Barrow AD, Trowsdale J (2006) You say ITAM and I say ITIM, let's call the whole
thing off: the ambiguity of immunoreceptor signalling. Eur J Immunol 36:1646–
355
1653.
Baruch K, Rosenzweig N, Kertser A, Deczkowska A, Sharif AM, Spinrad A, Tsitsou-
Kampeli A, Sarel A, Cahalon L, Schwartz M (2015a) Breaking immune tolerance by
targeting Foxp3+ regulatory T cells mitigates Alzheimer’s disease pathology. Nat
Commun 6:7967 Available at: http://www.nature.com/articles/ncomms8967
[Accessed April 17, 2018].
Baruch K, Rosenzweig N, Kertser A, Deczkowska A, Sharif AM, Spinrad A, Tsitsou-
Kampeli A, Sarel A, Cahalon L, Schwartz M (2015b) Breaking immune tolerance by
targeting Foxp3+ regulatory T cells mitigates Alzheimer/’s disease pathology. 6 SP-
:7967.
Becher B, Prat A, Antel JP (2000) Brain-immune connection: Immuno-regulatory
properties of CNS-resident cells. Glia 29:293–304 Available at:
http://doi.wiley.com/10.1002/%28SICI%291098-
1136%2820000215%2929%3A4%3C293%3A%3AAID-GLIA1%3E3.0.CO%3B2-A
[Accessed April 17, 2018].
Becher B, Schlitzer A, Chen J, Mair F, Sumatoh HR, Teng KWW, Low D, Ruedl C,
Riccardi-Castagnoli P, Poidinger M, Greter M, Ginhoux F, Newell EW (2014) High-
dimensional analysis of the murine myeloid cell system. Nat Immunol 15:1181–
1189.
Bell RD, Winkler EA, Sagare AP, Singh I, LaRue B, Deane R, Zlokovic B V. (2010)
Pericytes Control Key Neurovascular Functions and Neuronal Phenotype in the
Adult Brain and during Brain Aging. Neuron 68:409–427.
Ben-Shaanan TL, Azulay-Debby H, Dubovik T, Starosvetsky E, Korin B, Schiller M,
Green NL, Admon Y, Hakim F, Shen-Orr SS, Rolls A (2016) Activation of the
reward system boosts innate and adaptive immunity. Nat Med 22:940–944
Available at: http://www.nature.com/articles/nm.4133 [Accessed April 17, 2018].
Benhar I, London A, Schwartz M (2012) The privileged immunity of immune privileged
organs: the case of the eye. Front Immunol 3:296 Available at:
http://journal.frontiersin.org/article/10.3389/fimmu.2012.00296/abstract [Accessed
April 19, 2018].
Benitez BA, Cooper B, Pastor P, Jin S-C, Lorenzo E, Cervantes S, Cruchaga C (2013)
TREM2 is associated with the risk of Alzheimer’s disease in Spanish population.
Neurobiol Aging 34:1711.e15-1711.e17 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S0197458012006495.
Berg J, Tymoczko J, Stryer L (2007) Biochemistry, 6th ed. W.H. Freman.
Bialas AR, Stevens B (2013) TGF-β signaling regulates neuronal C1q expression and
developmental synaptic refinement. Nat Neurosci 16:1773–1782.
Bodenmiller B, Giesen C, Schulz D, Schapiro D, Jackson HW, Fischer JR, Catena R,
Raghuraman S, Zanotelli VRT, Varga Z (2017) histoCAT: analysis of cell
phenotypes and interactions in multiplex image cytometry data. Nat Methods
14:873–876.
Boehm T (2012) Evolution of vertebrate immunity. Curr Biol 22:R722–R732 Available at:
http://dx.doi.org/10.1016/j.cub.2012.07.003.
Bohlson SS, O\textquoterightConner SD, Hulsebus HJ, Ho M-M, Fraser DA (2014)
Complement, C1q, and C1q-Related Molecules Regulate Macrophage Polarization.
Front Immunol 5:113.
356
Bolmont T, Haiss F, Eicke D, Radde R, Mathis CA, Klunk WE, Kohsaka S, Jucker M,
Calhoun ME (2008) Dynamics of the microglial/amyloid interaction indicate a role in
plaque maintenance. J Neurosci 28:4283–4292 Available at:
http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.4814-07.2008.
Bornemann KD, Wiederhold KH, Pauli C, Ermini F, Stalder M, Schnell L, Sommer B,
Jucker M, Staufenbiel M (2001) Abeta-induced inflammatory processes in microglia
cells of APP23 transgenic mice. Am J Pathol 158:63–73.
Bossi F, Tripodo C, Rizzi L, Bulla R, Agostinis C, Guarnotta C, Munaut C, Baldassarre
G, Papa G, Zorzet S, Ghebrehiwet B, Ling GS, Botto M, Tedesco F (2014) C1q as
a unique player in angiogenesis with therapeutic implication in wound healing. Proc
Natl Acad Sci U S A 111:4209–4214 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/24591625 [Accessed April 18, 2018].
Bouchon A, Dietrich J, Colonna M (2000) Cutting edge: inflammatory responses can be
triggered by TREM-1, a novel receptor expressed on neutrophils and monocytes. J
Immunol 164:4991–4995.
Bouchon A, Hernández-Munain C, Cella M, Colonna M (2001) A DAP12-mediated
pathway regulates expression of CC chemokine receptor 7 and maturation of
human dendritic cells. J Exp Med 194:1111–1122.
Boulanger LM, Shatz CJ (2004) Immune signalling in neural development, synaptic
plasticity and disease. Nat Rev Neurosci 5:521–531 Available at:
http://www.nature.com/articles/nrn1428 [Accessed April 19, 2018].
Breunig JJ, Guillot-Sestier MV, Town T (2013) Brain injury, neuroinflammation and
Alzheimer’s disease. Front Aging Neurosci 5.
Brown S, Hu N, Hombría JCG (2001) Identification of the first invertebrate interleukin
JAK/STAT receptor, the Drosophila gene domeless. Curr Biol 11:1700–1705.
Bulla R, Tripodo C, Rami D, Ling GS, Agostinis C, Guarnotta C, Zorzet S, Durigutto P,
Botto M, Tedesco F (2016) C1q acts in the tumour microenvironment as a cancer-
promoting factor independently of complement activation. Nat Commun 7:10346
Available at: http://www.nature.com/doifinder/10.1038/ncomms10346 [Accessed
April 8, 2018].
Bundgaard M, Abbott NJ (2008) All vertebrates started out with a glial blood-brain
barrier 4-500 million years ago. Glia 56:699–708 Available at:
http://doi.wiley.com/10.1002/glia.20642 [Accessed April 22, 2018].
Burchfield JS, Li Q, Wang HY, Wang R-F (2015) JMJD3 as an epigenetic regulator in
development and disease. Int J Biochem Cell Biol 67:148–157 Available at:
https://www.sciencedirect.com/science/article/pii/S1357272515001910?via%3Dihu
b [Accessed April 17, 2018].
Butler AB, Hodos W (2005) Comparative Vertebrate Neuroanatomy: Evolution and
Adaptation: Second Edition.
Butovsky O, Jedrychowski MP, Moore CS, Cialic R, Lanser AJ, Gabriely G,
Koeglsperger T, Dake B, Wu PM, Doykan CE, Fanek Z, Liu L, Chen Z, Rothstein
JD, Ransohoff RM, Gygi SP, Antel JP, Weiner HL (2013) Identification of a unique
TGF-[beta]-dependent molecular and functional signature in microglia. Nat
Neurosci 17:131–143.
Buttgereit A, Lelios I, Yu X, Vrohlings M, Krakoski NR, Gautier EL, Nishinakamura R,
Becher B, Greter M (2016) Sall1 is a transcriptional regulator defining microglia
357
identity and function. Nat Immunol 17:1397–1406.
Cacciottolo M, Christensen A, Moser A, Liu J, Pike CJ, Smith C, LaDu MJ, Sullivan PM,
Morgan TE, Dolzhenko E, Charidimou A, Wahlund L-O, Wiberg MK, Shams S,
Chiang GC-Y, Finch CE (2016) The APOE4 allele shows opposite sex bias in
microbleeds and Alzheimer’s disease of humans and mice. Neurobiol Aging 37:47–
57.
Cacciottolo M, Wang X, Driscoll I, Woodward N, Saffari A, Reyes J, Serre ML, Vizuete
W, Sioutas C, Morgan TE, Gatz M, Chui HC, Shumaker SA, Resnick SM, Espeland
MA, Finch CE, Chen JC (2017) Particulate air pollutants, APOE alleles and their
contributions to cognitive impairment in older women and to amyloidogenesis in
experimental models. Transl Psychiatry 7:e1022.
Cagnin A, Kassiou M, Meikle SR, Banati RB (2006) In vivo evidence for microglial
activation in neurodegenerative dementia. Acta Neurol Scand Suppl 185:107–114.
Campbell KS, Yusa S -i., Kikuchi-Maki A, Catina TL (2004) NKp44 Triggers NK Cell
Activation through DAP12 Association That Is Not Influenced by a Putative
Cytoplasmic Inhibitory Sequence. J Immunol 172:899–906 Available at:
http://www.jimmunol.org/cgi/doi/10.4049/jimmunol.172.2.899.
Cao Y, Chtarbanova S, Petersen AJ, Ganetzky B (2013) Dnr1 mutations cause
neurodegeneration in Drosophila by activating the innate immune response in the
brain. Proc Natl Acad Sci U S A 110:E1752-60 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/23613578 [Accessed April 20, 2018].
Carrillo-Bustamante P, Keşmir C, de Boer RJ (2016) The evolution of natural killer cell
receptors. Immunogenetics 68:3–18.
Carson MJ, Doose JM, Melchior B, Schmid CD, Ploix CC (2006) CNS immune privilege:
hiding in plain sight. Immunol Rev 213:48–65.
Chakrabarty P, Li A, Ceballos-Diaz C, Eddy JA, Funk CC, Moore B, DiNunno N,
Rosario AM, Cruz PE, Verbeeck C, Sacino A, Nix S, Janus C, Price ND, Das P,
Golde TE (2015) IL-10 Alters Immunoproteostasis in APP Mice, Increasing Plaque
Burden and Worsening Cognitive Behavior. Neuron 85:519–533.
Chambers MC, Schneider DS (2012) Pioneering immunology: Insect style. Curr Opin
Immunol 24:10–14.
Cheng S-C et al. (2014) mTOR-and HIF-1a–mediated aerobic glycolysis as metabolic
basis for trained immunity. Science (80- ) 345:1250684.
Christofi T, Apidianakis Y (2013) Drosophila immune priming against Pseudomonas
aeruginosa is short-lasting and depends on cellular and humoral immunity.
F1000Research 2 Available at: http://f1000research.com/articles/2-76/v1 [Accessed
April 19, 2018].
Chue SC, Seow CJ, Duan W, Yeo KSLL, Koh AHMM, Wong WSFF (2004) Inhibitor of
p42/44 mitogen-activated protein kinase, but not p38 MAPK, attenuated antigen
challenge of guinea pig airways in vitro. Int Immunopharmacol 4:1089–1098.
Clarke E V., Weist BM, Walsh CM, Tenner AJ (2015) Complement protein C1q bound to
apoptotic cells suppresses human macrophage and dendritic cell-mediated Th17
and Th1 T cell subset proliferation. J Leukoc Biol 97:147–160 Available at:
http://www.jleukbio.org/cgi/doi/10.1189/jlb.3A0614-278R.
Cohen RM, Rezai-Zadeh K, Weitz TM, Rentsendorj A, Gate D, Spivak I, Bholat Y,
Vasilevko V, Glabe CG, Breunig JJ, Rakic P, Davtyan H, Agadjanyan MG, Kepe V,
358
Barrio JR, Bannykh S, Szekely CA, Pechnick RN, Town T (2013) A Transgenic
Alzheimer Rat with Plaques, Tau Pathology, Behavioral Impairment, Oligomeric
A$β$, and Frank Neuronal Loss. J Neurosci 33:6245–6256.
Colonna M (2003) Trems in the immune system and beyond. Nat Rev Immunol 3:445–
453.
Colonna M, Turnbull I, Klesney-Tait J (2007) The Enigmatic Function of TREM-2 in
Osteoclastogenesis. In, pp 97–105. Springer, Boston, MA. Available at:
http://link.springer.com/10.1007/978-0-387-72009-8_13 [Accessed April 23, 2018].
Colonna M, Wang Y (2016) TREM2 variants: New keys to decipher Alzheimer disease
pathogenesis. Nat Rev Neurosci 17:201–207.
Colton CA (2013) Immune heterogeneity in neuroinflammation: Dendritic cells in the
brain. J Neuroimmune Pharmacol 8:145–162.
Conteras-Garduño J, Lanz-Mendoza H, Franco B, Nava A, Pedraza-Reyes M, Canales-
Lazcano J (2016) Insect immune priming: ecology and experimental evidences.
Ecol Entomol 41:351–366 Available at: http://doi.wiley.com/10.1111/een.12300
[Accessed April 19, 2018].
Copf T, Goguel V, Lampin-Saint-Amaux A, Scaplehorn N, Preat T (2011) Cytokine
signaling through the JAK/STAT pathway is required for long-term memory in
Drosophila. Proc Natl Acad Sci U S A 108:8059–8064.
Corder EH, Saunders AM, Strittmatter WJ, Schmechel DE, Gaskell PC, Small GW,
Roses AD, Haines JL, Pericak-Vance MA (1993) Gene dose of apolipoprotein E
type 4 allele and the risk of Alzheimer’s disease in late onset families. Science
261:921–923.
Coutinho-Budd J, Freeman MR (2013) Probing the enigma: Unraveling glial cell biology
in invertebrates. Curr Opin Neurobiol 23:1073–1079.
Covarrubias AJ, Aksoylar HI, Horng T (2015) Control of macrophage metabolism and
activation by mTOR and Akt signaling. Semin Immunol 27:286–296 Available at:
https://www.sciencedirect.com/science/article/pii/S1044532315000639?via%3Dihu
b [Accessed April 16, 2018].
Cribbs DH, Ghochikyan A, Vasilevko V, Tran M, Petrushina I, Sadzikava N, Babikyan D,
Kesslak P, Kieber-Emmons T, Cotman CW, Agadjanyan MG (2003) Adjuvant-
dependent modulation of Th1 and Th2 responses to immunization with β-amyloid.
Int Immunol 15:505–514.
Crittenden JR, Skoulakis EMC, Han K-A, Kalderon D, Davis RL (1998) Tripartite
Mushroom Body Architecture Revealed by Antigenic Markers. Learn Mem 5:38–51.
Crotti A, Benner C, Kerman BE, Gosselin D, Lagier-Tourenne C, Zuccato C, Cattaneo
E, Gage FH, Cleveland DW, Glass CK (2014) Mutant Huntingtin promotes
autonomous microglia activation via myeloid lineage-determining factors. Nat
Neurosci 17:513–521.
Crotti A, Ransohoff RM (2016) Microglial Physiology and Pathophysiology: Insights from
Genome-wide Transcriptional Profiling. Immunity 44:505–515.
Croxford AL, Lanzinger M, Hartmann FJ, Schreiner B, Mair F, Pelczar P, Clausen BE,
Jung S, Greter M, Becher B (2015) The Cytokine GM-CSF Drives the Inflammatory
Signature of CCR2+ Monocytes and Licenses Autoimmunity. Immunity 43:502–514
Available at: http://www.ncbi.nlm.nih.gov/pubmed/26341401 [Accessed April 17,
2018].
359
Cui X, Zeni F, Vodovitz Y, Correa-de-Araujo R, Quezado M, Roberts A, Wahl S, Danner
RL, Banks SM, Gerstenberger E, Fitz Y, Natanson C, Eichacker PQ (2003) TGF-β1
increases microbial clearance but worsens lung injury during Escherichia coli
pneumonia in rats. Cytokine 24:115–127.
D’Ambrosio M V., Vale RD (2010) A whole genome RNAi screen of Drosophila S2 cell
spreading performed using automated computational image analysis. J Cell Biol
191:471–478.
Dann SM, Le C, Choudhury BK, Liu H, Saldarriaga O, Hanson EM, Cong Y, Eckmann L
(2014) Attenuation of intestinal inflammation in interleukin-10-deficient mice
infected with Citrobacter rodentium. Infect Immun 82:1949–1958.
Davis RL (2004) Olfactory learning. Neuron 44:31–48.
Derheimer FA, Kastan MB (2010) Multiple roles of ATM in monitoring and maintaining
DNA integrity. FEBS Lett 584:3675–3681.
DeSalvo MK, Hindle SJ, Rusan ZM, Orng S, Eddison M, Halliwill K, Bainton RJ (2014)
The Drosophila surface glia transcriptome: evolutionary conserved blood-brain
barrier processes. Front Neurosci 8:346 Available at:
http://journal.frontiersin.org/article/10.3389/fnins.2014.00346/abstract [Accessed
April 23, 2018].
Dey A, Allen J, Hankey-Giblin PA (2015) Ontogeny and Polarization of Macrophages in
Inflammation: Blood Monocytes Versus Tissue Macrophages. Front Immunol 5:683
Available at: http://journal.frontiersin.org/article/10.3389/fimmu.2014.00683/abstract
[Accessed December 7, 2017].
Díaz-Villanueva JF, Díaz-Molina R, García-González V (2015) Protein folding and
mechanisms of proteostasis. Int J Mol Sci 16:17193–17230.
Doherty J, Logan MA, Tasdemir OE, Freeman MR (2009) Ensheathing Glia Function as
Phagocytes in the Adult Drosophila Brain. J Neurosci 29:4768–4781 Available at:
http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.5951-08.2009.
Dong Y, Cirimotich CM, Pike A, Chandra R, Dimopoulos G (2012) Anopheles NF-κB-
regulated splicing factors direct pathogen-specific repertoires of the hypervariable
pattern recognition receptor AgDscam. Cell Host Microbe 12:521–530.
Doty KR, Guillot-Sestier M-V, Town T (2015) The role of the immune system in
neurodegenerative disorders: Adaptive or maladaptive? Brain Res 1617:155–173
Available at: http://linkinghub.elsevier.com/retrieve/pii/S0006899314011950.
Ebstein F, Voigt A, Lange N, Warnatsch A, Schröter F, Prozorovski T, Kuckelkorn U,
Aktas O, Seifert U, Kloetzel PM, Krüger E (2013) Immunoproteasomes are
important for proteostasis in immune responses. Cell 152:935–937.
Edison P, Archer HA, Gerhard A, Hinz R, Pavese N, Turkheimer FE, Hammers A, Tai
YF, Fox N, Kennedy A, Rossor M, Brooks DJ (2008) Microglia, amyloid, and
cognition in Alzheimer’s disease: An [11C](R)PK11195-PET and [11C]PIB-PET
study. Neurobiol Dis 32:412–419.
Edwards TN, Meinertzhagen IA (2010) The functional organisation of glia in the adult
brain of Drosophila and other insects. Prog Neurobiol 90:471–497.
Eggen BJL, Raj D, Hanisch UK, Boddeke HWGM (2013) Microglial phenotype and
adaptation. J Neuroimmune Pharmacol 8:807–823.
El Khoury J, Toft M, Hickman SE, Means TK, Terada K, Geula C, Luster AD (2007)
Ccr2 deficiency impairs microglial accumulation and accelerates progression of
360
Alzheimer-like disease. Nat Med 13:432–438.
Elmore MRP, Najafi AR, Koike MA, Dagher NN, Spangenberg EE, Rice RA, Kitazawa
M, Matusow B, Nguyen H, West BL, Green KN (2014a) Colony-Stimulating Factor
1 Receptor Signaling Is Necessary for Microglia Viability, Unmasking a Microglia
Progenitor Cell in the Adult Brain. Neuron 82:380–397 Available at:
https://www.sciencedirect.com/science/article/pii/S0896627314001718?via%3Dihu
b [Accessed April 17, 2018].
Elmore MRP, Najafi AR, Koike MA, Dagher NN, Spangenberg EE, Rice RA, Kitazawa
M, Matusow B, Nguyen H, West BL, Green KN (2014b) Colony-Stimulating Factor
1 Receptor Signaling Is Necessary for Microglia Viability, Unmasking a Microglia
Progenitor Cell in the Adult Brain. Neuron 82:380–397 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S0896627314001718 [Accessed April 7,
2018].
Eroglu C, Barres BA (2010) Regulation of synaptic connectivity by glia. Nature 468:223–
231.
Fairn GD, Grinstein S (2012) How nascent phagosomes mature to become
phagolysosomes. Trends Immunol 33:397–405.
Fang Y, Yu S, Ellis JS, Sharav T, Braley-mullen H (2010) Comparison of sensitivity of
Th1, Th2, and Th17 cells to Fas-mediated apoptosis. J Leukoc Biol 87:1019–1028.
Ferguson TA, Griffith TS (2006) A vision of cell death: Fas ligand and immune privilege
10 years later. Immunol Rev 213:228–238.
Flajnik MF (2014) Re-evaluation of the Immunological Big Bang. Curr Biol 24:R1060–
R1065 Available at:
https://www.sciencedirect.com/science/article/pii/S0960982214012664 [Accessed
April 21, 2018].
Foley E, O’Farrell PH (2003) Nitric oxide contributes to induction of innate immune
responses to gram-negative bacteria in Drosophila. Genes Dev 17:115–125.
Fonseca MI, Chu S-H, Berci AM, Benoit ME, Peters DG, Kimura Y, Tenner AJ (2011)
Contribution of complement activation pathways to neuropathology differs among
mouse models of Alzheimer’s disease. J Neuroinflammation 8:4.
Fonseca MI, Zhou J, Botto M, Tenner AJ (2004) Absence of C1q Leads to Less
Neuropathology in Transgenic Mouse Models of Alzheimer9s Disease. J Neurosci
24:6457–6465.
Forabosco P, Ramasamy A, Trabzuni D, Walker R, Smith C, Bras J, Levine AP, Hardy
J, Pocock JM, Guerreiro R, Weale ME, Ryten M (2013) Insights into TREM2
biology by network analysis of human brain gene expression data. Neurobiol Aging
34:2699–2714.
Fraser DA, Pisalyaput K, Tenner AJ (2010) C1q enhances microglial clearance of
apoptotic neurons and neuronal blebs, and modulates subsequent inflammatory
cytokine production. J Neurochem 112:733–743.
Freeman MR, Doherty J (2006) Glial cell biology in Drosophila and vertebrates. Trends
Neurosci 29:82–90.
Frenck RW, Blackburn EH, Shannon KM, Dhabhar FS, Adler NE, Morrow JD, Cawthon
RM (1998) The rate of telomere sequence loss in human leukocytes varies with
age. Proc Natl Acad Sci U S A 95:5607–5610 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/9576930 [Accessed April 17, 2018].
361
Fuhrmann M, Bittner T, Jung CKE, Burgold S, Page RM, Mitteregger G, Haass C,
Laferla FM, Kretzschmar H, Herms J (2010) Microglial Cx3cr1 knockout prevents
neuron loss in a mouse model of Alzheimer’s disease. Nat Neurosci 13:411–413.
Fujita T (2002) EVOLUTION OF THE LECTIN – COMPLEMENT PATHWAY AND ITS
ROLE IN INNATE IMMUNITY. 2:346–353.
Fujita T, Matsushita M, Endo Y (2004) The lectin-complement pathway - Its role in
innate immunity and evolution. Immunol Rev 198:185–202.
Gaboriaud C, Frachet P, Thielens NM, Arlaud GJ (2012) The human C1q globular
domain: Structure and recognition of non-immune self ligands. Front Immunol 2.
Gaboriaud C, Thielens NM, Gregory LA, Rossi V, Fontecilla-Camps JC, Arlaud GJ
(2004) Structure and activation of the C1 complex of complement: unraveling the
puzzle. Trends Immunol 25:368–373 Available at:
https://www.sciencedirect.com/science/article/pii/S1471490604001279 [Accessed
March 21, 2018].
Galvan MD, Foreman DB, Zeng E, Tan JC, Bohlson SS (2012) Complement
Component C1q Regulates Macrophage Expression of Mer Tyrosine Kinase To
Promote Clearance of Apoptotic Cells. J Immunol 188:3716–3723.
Gao Y, Vidal-Itriago A, Kalsbeek MJ, Layritz C, García-Cáceres C, Tom RZ, Eichmann
TO, Vaz FM, Houtkooper RH, van der Wel N, Verhoeven AJ, Yan J, Kalsbeek A,
Eckel RH, Hofmann SM, Yi C-X (2017) Lipoprotein Lipase Maintains Microglial
Innate Immunity in Obesity. Cell Rep 20:3034–3042 Available at:
https://www.sciencedirect.com/science/article/pii/S221112471731269X?via%3Dihu
b [Accessed March 20, 2018].
Garay PA, McAllister AK (2010) Novel roles for immune molecules in neural
development: Implications for neurodevelopmental disorders. Front Synaptic
Neurosci.
Gate D, Rezai-Zadeh K, Jodry D, Rentsendorj A, Town T (2010) Macrophages in
Alzheimer’s disease: The blood-borne identity. J Neural Transm 117:961–970.
Geisow MJ, D’Arcy Hart P, Young MR (1981) Temporal changes of lysosome and
phagosome pH during phagolysosome formation in macrophages: Studies by
fluorescence spectroscopy. J Cell Biol 89:645–652.
Genua M, Rutella S, Correale C, Danese S (2014) The triggering receptor expressed on
myeloid cells (TREM) in inflammatory bowel disease pathogenesis. J Transl Med
12:293.
Ghosh S, Geahlen RL (2015) Stress Granules Modulate SYK to Cause Microglial Cell
Dysfunction in Alzheimer’s Disease. EBioMedicine 0:1785–1798.
Ginhoux F, Greter M, Leboeuf M, Nandi S, See P, Gokhan S, Mehler MF, Conway SJ,
Ng LG, Stanley ER, Samokhvalov IM, Merad M (2010) Fate Mapping Analysis
Reveals That Adult Microglia Derive from Primitive Macrophages. Science (80- )
330:841–845 Available at:
http://www.sciencemag.org/cgi/doi/10.1126/science.1194637 [Accessed April 7,
2018].
Ginhoux F, Jung S (2014) Monocytes and macrophages: developmental pathways and
tissue homeostasis. 14:392–404.
Ginhoux F, Prinz M (2015) Origin of microglia: current concepts and past controversies.
Cold Spring Harb Perspect Biol 7:a020537.
362
Ginhoux F, Schultze JL, Murray PJ, Ochando J, Biswas SK (2016) New insights into the
multidimensional concept of macrophage ontogeny, activation and function. Nat
Immunol 17:34–40.
Gjoneska E, Pfenning AR, Mathys H, Quon G, Kundaje A, Tsai L-H, Kellis M (2015)
Conserved epigenomic signals in mice and humans reveal immune basis of
Alzheimer’s disease. Nature 518:365–369.
Glass CK, Saijo K, Winner B, Marchetto MC, Gage FH (2010) Mechanisms Underlying
Inflammation in Neurodegeneration. Cell 140:918–934.
Goldmann T et al. (2016) Origin, fate and dynamics of macrophages at central nervous
system interfaces. Nat Immunol 17:797–805.
Grabert K, Michoel T, Karavolos MH, Clohisey S, Kenneth Baillie J, Stevens MP,
Freeman TC, Summers KM, McColl BW (2016) Microglial brain regionâ ’dependent
diversity and selective regional sensitivities to aging. Nat Neurosci 19:504–516.
Greter M, Lelios I, Croxford AL (2015) Microglia Versus Myeloid Cell Nomenclature
during Brain Inflammation. Front Immunol 6:249 Available at:
http://www.frontiersin.org/Antigen_Presenting_Cell_Biology/10.3389/fimmu.2015.00
249/abstract [Accessed March 20, 2018].
Greter M, Merad M (2013) Regulation of microglia development and homeostasis. Glia.
Griciuc A, Serrano-Pozo A, Parrado AR, Lesinski AN, Asselin CN, Mullin K, Hooli B,
Choi SH, Hyman BT, Tanzi RE (2013) Alzheimer’s disease risk gene cd33 inhibits
microglial uptake of amyloid beta. Neuron 78:631–643.
Guerreiro R et al. (2013a) TREM2Variants in Alzheimer’s Disease. N Engl J Med
368:117–127.
Guerreiro R et al. (2013b) TREM2 variants in Alzheimer’s disease. N Engl J Med
368:117–127 Available at:
http://www.nejm.org/doi/abs/10.1056/NEJMoa1211851%5Cnhttp://www.ncbi.nlm.ni
h.gov/pubmed/23150934%5Cnhttp://www.pubmedcentral.nih.gov/articlerender.fcgi
?artid=PMC3631573.
Guha M, Mackman N (2002) The phosphatidylinositol 3-kinase-Akt pathway limits
lipopolysaccharide activation of signaling pathways and expression of inflammatory
mediators in human monocytic cells. J Biol Chem 277:32124–32132 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/12052830 [Accessed April 16, 2018].
Guillot-Sestier M-V, Doty KR, Gate D, Rodriguez J, Leung BP, Rezai-Zadeh K, Town T
(2015a) Il10 deficiency rebalances innate immunity to mitigate Alzheimer-like
pathology. Neuron 85:534–548.
Guillot-Sestier M-V, Doty KR, Town T (2015b) Innate Immunity Fights Alzheimer’s
Disease. Trends Neurosci 38:674–681.
Guillot-Sestier M-V, Weitz TM, Town T (2016) Quantitative 3D In Silico Modeling
(q3DISM) of Cerebral Amyloid-beta Phagocytosis in Rodent Models of Alzheimer’s
Disease. J Vis Exp.
Guo M, Hong EJ, Fernandes J, Zipursky SL, Hay BA (2003) A reporter for amyloid
precursor protein gamma-secretase activity in Drosophila. Hum Mol Genet
12:2669–2678 Available at: http://hmg.oxfordjournals.org.ezp-
prod1.hul.harvard.edu/content/12/20/2669.long.
Haas MJ, Jafri M, Wehmeier KR, Onstead-Haas LM, Mooradian AD (2016) Inhibition of
endoplasmic reticulum stress and oxidative stress by vitamin D in endothelial cells.
363
Free Radic Biol Med 99:1–10.
Hakim Y, Yaniv SP, Schuldiner O (2014) Astrocytes play a key role in Drosophila
mushroom body axon pruning. PLoS One 9.
Halima C, Laura RM, William BS, Xiaoxian M, L. RD, Fahmeed H (2007)
Neurodevelopment of C57B/L6 mouse brain assessed by in vivo diffusion tensor
imaging. NMR Biomed 20:375–382 Available at: https://doi.org/10.1002/nbm.1130.
Hall CN, Reynell C, Gesslein B, Hamilton NB, Mishra A, Sutherland BA, Oâ Farrell FM,
Buchan AM, Lauritzen M, Attwell D (2014) Capillary pericytes regulate cerebral
blood flow in health and disease. Nature 508:55–60.
Hall SC, Agrawal DK (2017) Increased TREM-2 expression on the subsets of CD11c+
cells in the lungs and lymph nodes during allergic airway inflammation. Sci Rep
7:11853 Available at: http://www.nature.com/articles/s41598-017-12330-6
[Accessed April 23, 2018].
Hardy J, Selkoe DJ (2002) The Amyloid Hypothesis of Alzheimer’s Disease: Progress
and Problems on the Road to Therapeutics. 297:353–356.
Hartenstein V (2011) Morphological diversity and development of glia in Drosophila.
Glia 59:1237–1252.
Hayashi A, Sato T, Kamada N, Mikami Y, Matsuoka K, Hisamatsu T, Hibi T, Roers A,
Yagita H, Ohteki T, Yoshimura A, Kanai T (2013) A single strain of Clostridium
butyricum induces intestinal IL-10-producing macrophages to suppress acute
experimental colitis in mice. Cell Host Microbe 13:711–722.
Hazrati L-N, Van Cauwenberghe C, Brooks PL, Brouwers N, Ghani M, Sato C, Cruts M,
Sleegers K, St. George-Hyslop P, Van Broeckhoven C, Rogaeva E (2012) Genetic
association of CR1 with Alzheimer’s disease: A tentative disease mechanism.
Neurobiol Aging 33:2949.e5-2949.e12 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S0197458012003764 [Accessed
December 4, 2017].
Heneka MT et al. (2015) Neuroinflammation in Alzheimer’s disease. Lancet Neurol
14:388–405.
Heneka MT, Obanion M (2007) Inflammatory processes in Alzheimer’s disease. J
Neuroimmunol 184:69–91.
Heppner FL, Ransohoff RM, Becher B (2015) Immune attack: the role of inflammation in
Alzheimer disease. Nat Publ Gr 16:358–372.
Hibino T, Loza-Coll M, Messier C, Majeske AJ, Cohen AH, Terwilliger DP, Buckley KM,
Brockton V, Nair S V., Berney K, Fugmann SD, Anderson MK, Pancer Z, Cameron
RA, Smith LC, Rast JP (2006) The immune gene repertoire encoded in the purple
sea urchin genome. Dev Biol 300:349–365 Available at:
https://www.sciencedirect.com/science/article/pii/S0012160606011572 [Accessed
April 22, 2018].
Hickman SE, Kingery ND, Ohsumi TK, Borowsky ML, Wang L, Means TK, El Khoury J
(2013) The microglial sensome revealed by direct RNA sequencing. Nat Neurosci
16:1896–1905 Available at: http://www.nature.com/articles/nn.3554 [Accessed April
17, 2018].
Hoffmann JA, Reichhart JM (2002) Drosophila innate immunity: An evolutionary
perspective. Nat Immunol 3:121–126.
Höglund P, Brodin P (2010) Current perspectives of natural killer cell education by MHC
364
class i molecules. Nat Rev Immunol 10:724–734.
Holtzman DM, Mandelkow E, Selkoe DJ (2012) Alzheimer disease in 2020. Cold Spring
Harb Perspect Med 2.
Hong S, Beja-Glasser VF, Nfonoyim BM, Frouin A, Li S, Ramakrishnan S, Merry KM,
Shi Q, Rosenthal A, Barres BA, Lemere CA, Selkoe DJ, Stevens B (2016)
Complement and microglia mediate early synapse loss in Alzheimer mouse
models. 352:712–716.
Hsieh CL, Koike M, Spusta SC, Niemi EC, Yenari M, Nakamura MC, Seaman WE
(2009) A role for TREM2 ligands in the phagocytosis of apoptotic neuronal cells by
microglia. J Neurochem 109:1144–1156.
Huang M, Wang D, Xu Z, Xu Y, Xu X, Ma Y, Xia Z (2015) Lack of genetic association
between TREM2 and Alzheimer’s disease in East Asian population. Am J
Alzheimers Dis Other Demen 30:541–546.
Huang Y, Xu Z, Xiong S, Sun F, Qin G, Hu G, Wang J, Zhao L, Liang Y-X, Wu T, Lu Z,
Humayun MS, So K-F, Pan Y, Li N, Yuan T-F, Rao Y, Peng B (2018) Repopulated
microglia are solely derived from the proliferation of residual microglia after acute
depletion. Nat Neurosci 21:530 Available at:
http://www.nature.com/articles/s41593-018-0090-8 [Accessed April 7, 2018].
Humphrey MB, Nakamura MC (2015) A Comprehensive Review of Immunoreceptor
Regulation of Osteoclasts. Clin Rev Allergy Immunol:1–11.
Iijima K, Liu H-P, Chiang A-S, Hearn S a, Konsolaki M, Zhong Y (2004) Dissecting the
pathological effects of human Abeta40 and Abeta42 in Drosophila: a potential
model for Alzheimer’s disease. Proc Natl Acad Sci U S A 101:6623–6628.
Inamdar AA, Bennett JW (2014) A common fungal volatile organic compound induces a
nitric oxide mediated inflammatory response in Drosophila melanogaster. Sci Rep
4.
Ito H, Hamerman JA (2012) TREM-2, triggering receptor expressed on myeloid cell-2,
negatively regulates TLR responses in dendritic cells. Eur J Immunol 42:176–185
Available at: http://doi.wiley.com/10.1002/eji.201141679 [Accessed April 16, 2018].
Ito K, Awano W, Suzuki K, Hiromi Y, Yamamoto D (1997) The Drosophila mushroom
body is a quadruple structure of clonal units each of which contains a virtually
identical set of neurones and glial cells. Development 124:761–771 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/9043058.
Ivashkiv LB (2011) How ITAMs Inhibit Signaling. Sci Signal 4:pe20--pe20.
Jalili A, Marquez-Curtis L, Shirvaikar N, Wysoczynski M, Ratajczak M, Janowska-
Wieczorek A (2010) Complement C1q enhances homing-related responses of
hematopoietic stem/progenitor cells. Transfusion 50:2002–2010 Available at:
http://doi.wiley.com/10.1111/j.1537-2995.2010.02664.x [Accessed April 9, 2018].
Janeway CA Jr1 MR (2002) Innate immune recognition. Annu Rev Immunol 20:197–
216.
Jankowsky JL, Fadale DJ, Anderson J, Xu GM, Gonzales V, Jenkins NA, Copeland NG,
Lee MK, Younkin LH, Wagner SL, Younkin SG, Borchelt DR (2004) Mutant
presenilins specifically elevate the levels of the 42 residue β-amyloid peptide in
vivo: Evidence for augmentation of a 42-specific γ secretase. Hum Mol Genet
13:159–170.
Jankowsky JL, Slunt HH, Ratovitski T, Jenkins NA, Copeland NG, Borchelt DR (2001)
365
Co-expression of multiple transgenes in mouse CNS: a comparison of strategies.
Biomol Eng 17:157–165.
Jay TR et al. (2015) TREM2 deficiency eliminates TREM2+ inflammatory macrophages
and ameliorates pathology in Alzheimer{\textquoteright}s disease mouse models. J
Exp Med 212:287–295.
Jenett A et al. (2012) A GAL4-Driver Line Resource for Drosophila Neurobiology. Cell
Rep 2:991–1001.
Jiang D, Rauda I, Han S, Chen S, Zhou F (2012) Aggregation Pathways of the Amyloid
$β$(1{\textendash}42) Peptide Depend on Its Colloidal Stability and Ordered $β$-
Sheet Stacking. Langmuir 28:12711–12721.
Jiang K, Zhong B, Gilvary DL, Corliss BC, Vivier E, Hong-Geller E, Wei S, Djeu JY
(2002) Syk Regulation of Phosphoinositide 3-Kinase-Dependent NK Cell Function.
J Immunol 168:3155–3164 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/9862693 [Accessed April 17, 2018].
Jiang T, Tan L, Zhu XC, Zhang QQ, Cao L, Tan MS, Gu LZ, Wang HF, Ding ZZ, Zhang
YD, Yu JT (2014) Upregulation of TREM2 ameliorates neuropathology and rescues
spatial cognitive impairment in a transgenic mouse model of Alzheimer’s disease.
Neuropsychopharmacology 39:2949–2962.
Jiang T, Wan Y, Zhang Y-D, Zhou J-S, Gao Q, Zhu X-C, Shi J-Q, Lu H, Tan L, Yu J-T
(2017a) TREM2 Overexpression has No Improvement on Neuropathology and
Cognitive Impairment in Aging APPswe/PS1dE9 Mice. Mol Neurobiol 54:855–865.
Jiang T, Wan Y, Zhang Y-D, Zhou J-S, Gao Q, Zhu X-C, Shi J-Q, Lu H, Tan L, Yu J-T
(2017b) TREM2 Overexpression has No Improvement on Neuropathology and
Cognitive Impairment in Aging APPswe/PS1dE9 Mice. Mol Neurobiol 54:855–865
Available at: http://link.springer.com/10.1007/s12035-016-9704-x [Accessed April
18, 2018].
Jiang T, Zhang YD, Chen Q, Gao Q, Zhu XC, Zhou JS, Shi JQ, Lu H, Tan L, Yu JT
(2016) TREM2 modifies microglial phenotype and provides neuroprotection in
P301S tau transgenic mice. Neuropharmacology 105:196–206.
Jin SC et al. (2015) TREM2 is associated with increased risk for Alzheimer’s disease in
African Americans. Mol Neurodegener 10.
Jinno S, Fleischer F, Eckel S, Schmidt V, Kosaka T (2007) Spatial arrangement of
microglia in the mouse hippocampus: A stereological study in comparison with
astrocytes. Glia 55:1334–1347.
Jinno S, Kosaka T (2008) Reduction of Iba1-expressing microglial process density in
the hippocampus following electroconvulsive shock. Exp Neurol 212:440–447.
Johansson KC, Metzendorf C, Söderhäll K (2005) Microarray analysis of immune
challenged Drosophila hemocytes. Exp Cell Res 305:145–155.
Jones JDG, Vance RE, Dangl JL (2016) Intracellular innate immune surveillance
devices in plants and animals. Science 354:aaf6395 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/27934708 [Accessed April 22, 2018].
Jonsson T et al. (2013) Variant of TREM2Associated with the Risk of Alzheimer’s
Disease. N Engl J Med 368:107–116.
Jung S, Aliberti J, Graemmel P, Sunshine MJ, Kreutzberg GW, Sher A, Littman DR
(2000) Analysis of Fractalkine Receptor CX3CR1 Function by Targeted Deletion
and Green Fluorescent Protein Reporter Gene Insertion. Mol Cell Biol 20:4106–
366
4114 Available at: http://mcb.asm.org/cgi/doi/10.1128/MCB.20.11.4106-4114.2000
[Accessed April 7, 2018].
Kaneko M, Stellwagen D, Malenka RC, Stryker MP (2008) Tumor Necrosis Factor-α
Mediates One Component of Competitive, Experience-Dependent Plasticity in
Developing Visual Cortex. Neuron 58:673–680.
Kapitonov V V, Koonin E V (2015) Evolution of the RAG1-RAG2 locus: both proteins
came from the same transposon. Biol Direct 10:20 Available at:
http://www.biologydirect.com/content/10/1/20 [Accessed April 23, 2018].
Keren-Shaul H, Spinrad A, Weiner A, Matcovitch-Natan O, Dvir-Szternfeld R, Ulland TK,
David E, Baruch K, Lara-Astaiso D, Toth B, Itzkovitz S, Colonna M, Schwartz M,
Amit I (2017) A Unique Microglia Type Associated with Restricting Development of
Alzheimer’s Disease. Cell 169:1276--1290.e17 Available at:
https://www.sciencedirect.com/science/article/pii/S0092867417305780?via%3Dihu
b [Accessed April 16, 2018].
Kettenmann H, Hanisch U-K, Noda M, Verkhratsky A (2011) Physiology of Microglia.
Physiol Rev 91:461–553 Available at:
http://www.physiology.org/doi/10.1152/physrev.00011.2010 [Accessed April 7,
2018].
Khalturin K, Becker M, Rinkevich B, Bosch TCG (2003) Urochordates and the origin of
natural killer cells: Identification of a CD94/NKR-P1-related receptor in blood cells
of Botryllus. Proc Natl Acad Sci 100:622–627 Available at:
http://www.pnas.org/cgi/doi/10.1073/pnas.0234104100.
Kierdorf K et al. (2013) Microglia emerge from erythromyeloid precursors via Pu.1- and
Irf8-dependent pathways. Nat Neurosci 16:273–280 Available at:
http://www.nature.com/articles/nn.3318 [Accessed April 7, 2018].
Kim MS, Kewalramani G, Puthanveetil P, Lee V, Kumar U, An D, Abrahani A, Rodrigues
B (2008) Acute diabetes moderates trafficking of cardiac lipoprotein lipase through
p38 mitogen-activated protein kinase-dependent actin cytoskeleton organization.
Diabetes 57:64–76 Available at: http://www.ncbi.nlm.nih.gov/pubmed/17942824
[Accessed April 16, 2018].
Kim Y, Mylonakis E (2012) Caenorhabditis elegans immune conditioning with the
probiotic bacterium Lactobacillus acidophilus strain NCFM enhances gram-positive
immune responses. Infect Immun 80:2500–2508 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/22585961 [Accessed April 19, 2018].
Kimbrell DA, Beutler B (2001) The evolution and genetics of innate immunity. Nat Rev
Genet 2:256–267 Available at: http://www.nature.com/articles/35066006 [Accessed
April 22, 2018].
Kimura H, Usui F, Karasawa T, Kawashima A, Shirasuna K, Inoue Y, Komada T,
Kobayashi M, Mizushina Y, Kasahara T, Suzuki K, Iwasaki Y, Yada T, Caturegli P,
Takahashi M (2015) Immunoproteasome subunit LMP7 Deficiency Improves
Obesity and Metabolic Disorders. Sci Rep 5:1–12 Available at:
http://dx.doi.org/10.1038/srep15883.
Kipnis J, Mizrahi T, Hauben E, Shaked I, Shevach E, Schwartz M (2002)
Neuroprotective autoimmunity: naturally occurring CD4+CD25+ regulatory T cells
suppress the ability to withstand injury to the central nervous system. Proc Natl
Acad Sci U S A 99:15620–15625 Available at:
367
http://www.ncbi.nlm.nih.gov/pubmed/12429857 [Accessed April 17, 2018].
Kishore U, Reid KB. (2000) C1q: Structure, function, and receptors.
Immunopharmacology 49:159–170 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S016231090080301X.
Kleinberger G et al. (2014) TREM2 mutations implicated in neurodegeneration impair
cell surface transport and phagocytosis. Sci Transl Med 6:243ra86--243ra86.
Klesney-Tait J, Turnbull IR, Colonna M (2006) The TREM receptor family and signal
integration. Nat Immunol 7:1266–1273.
Kober DL, Alexander-Brett JM, Karch CM, Cruchaga C, Colonna M, Holtzman MJ, Brett
TJ (2016) Neurodegenerative disease mutations in TREM2 reveal a functional
surface and distinct loss-of-function mechanisms. Elife 5:e20391.
Kohl J, Jefferis GSXE (2011) Neuroanatomy: Decoding the fly brain. Curr Biol 21.
Koistinaho M, Koistinaho J (2002) Role of p38 and p44/42 Mitogen- Activated Protein
Kinases in Microglia. Glia 40:175–183 Available at:
https://onlinelibrary.wiley.com/doi/pdf/10.1002/glia.10151 [Accessed April 16,
2018].
Koonin E V., Krupovic M (2015) Evolution of adaptive immunity from transposable
elements combined with innate immune systems. Nat Rev Genet 16:184–192
Available at: http://www.nature.com/articles/nrg3859 [Accessed April 23, 2018].
Kounatidis I, Ligoxygakis P (2012) Drosophila as a model system to unravel the layers
of innate immunity to infection. Open Biol 2:120075–120075 Available at:
http://rsob.royalsocietypublishing.org/cgi/doi/10.1098/rsob.120075.
Kraft-Terry SD, Buch SJ, Fox HS, Gendelman HE (2009) A Coat of Many Colors:
Neuroimmune Crosstalk in Human Immunodeficiency Virus Infection. Neuron
64:133–145.
Krasemann S et al. (2017) The TREM2-APOE Pathway Drives the Transcriptional
Phenotype of Dysfunctional Microglia in Neurodegenerative Diseases. Immunity
47:566--581.e9.
Kremer A (2011) GSK3 and Alzheimer’s disease: facts and fiction…. Front Mol Neurosci
4:1–10 Available at:
http://journal.frontiersin.org/article/10.3389/fnmol.2011.00017/abstract.
Kruse PH, Matta J, Ugolini S, Vivier E (2014) Natural cytotoxicity receptors and their
ligands. Immunol Cell Biol 92:221–229 Available at:
http://dx.doi.org/10.1038/icb.2013.98.
Lange C, Hemmrich G, Klostermeier UC, López-Quintero JA, Miller DJ, Rahn T, Weiss
Y, Bosch TCG, Rosenstiel P (2011) Defining the origins of the NOD-like receptor
system at the base of animal evolution. Mol Biol Evol 28:1687–1702.
Lanier LL (2008) Evolutionary struggles between NK cells and viruses. Nat Rev
Immunol 8:259–268.
Lanier LL, Bakker ABH (2000) The ITAM-bearing transmembrane adaptor DAP12 in
lymphoid and myeloid cell function. Immunol Today 21:611–614.
Lawson LJ, Perry VH, Dri P, Gordon S (1990) Heterogeneity in the distribution and
morphology of microglia in the normal adult mouse brain. Neuroscience 39:151–
170.
Leavy O (2013) Innate-like lymphocytes: Will the real ILC1 please stand up? Nat Rev
Immunol.
368
Lee CYD, Daggett A, Gu X, Jiang L-L, Langfelder P, Li X, Wang N, Zhao Y, Park CS,
Cooper Y, Ferando I, Mody I, Coppola G, Xu H, Yang XW (2018a) Elevated
TREM2 Gene Dosage Reprograms Microglia Responsivity and Ameliorates
Pathological Phenotypes in Alzheimer’s Disease Models. Neuron 97:1032–1048.e5
Available at:
https://www.sciencedirect.com/science/article/pii/S0896627318301016?via%3Dihu
b [Accessed March 20, 2018].
Lee J-H, Poudel B, Ki H-H, Nepali S, Lee Y-M, Shin J-S, Kim D-K (2018b) Complement
C1q stimulates the progression of hepatocellular tumor through the activation of
discoidin domain receptor 1. Sci Rep 8:4908 Available at:
http://www.nature.com/articles/s41598-018-23240-6 [Accessed April 8, 2018].
Leiss F, Groh C, Butcher NJ, Meinertzhagen IA, Tavosanis G (2009) Synaptic
organization in the adult Drosophila mushroom body calyx. J Comp Neurol
517:808–824.
Lemaitre B, Nicolas E, Michaut L, Reichhart JM, Hoffmann JA (1996) The dorsoventral
regulatory gene cassette spatzle/Toll/Cactus controls the potent antifungal
response in Drosophila adults. Cell 86:973–983.
Lesch C, Goto A, Lindgren M, Bidla G, Dushay MS, Theopold U (2007) A role for
Hemolectin in coagulation and immunity in Drosophila melanogaster. Dev Comp
Immunol 31:1255–1263.
Leung BP, Doty KR, Town T (2015) Cerebral innate immunity in Drosophila
melanogaster. AIMS Neurosci 2.
Levashina EA, Moita LF, Blandin S, Vriend G, Lagueux M, Kafatos FC (2001)
Conserved role of a complement-like protein in phagocytosis revealed by dsRNA
knockout in cultured cells of the mosquito, Anopheles gambiae. Cell 104:709–718.
Leyns CEG, Ulrich JD, Finn MB, Stewart FR, Koscal LJ, Remolina Serrano J, Robinson
GO, Anderson E, Colonna M, Holtzman DM (2017) TREM2 deficiency attenuates
neuroinflammation and protects against neurodegeneration in a mouse model of
tauopathy. Proc Natl Acad Sci:201710311 Available at:
http://www.pnas.org/lookup/doi/10.1073/pnas.1710311114.
Li X, Long J, He T, Belshaw R, Scott J (2015) Integrated genomic approaches identify
major pathways and upstream regulators in late onset Alzheimer{\textquoteright}s
disease. Sci Rep 5:12393.
Li Y, Du X, Liu C, Wen Z, Du J (2012) Reciprocal regulation between resting microglial
dynamics and neuronal activity in vivo. Dev Cell 23:1189–1202 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S153458071200490X%5Cnpapers3://publ
ication/doi/10.1016/j.devcel.2012.10.027.
Li Z, Zhong L, Gu L, Huang W, Shi X, Zhang X, An X, Lin Q, Tzeng CM (2016)
Association study of TREM2 polymorphism rs75932628 with leucoaraiosis or
Parkinson’s disease in the Han Chinese population. BMJ Open 6.
Lin KC, Huang DY, Huang DW, Tzeng SJ, Lin WW (2016) Inhibition of AMPK through
Lyn-Syk-Akt enhances FcεRI signal pathways for allergic response. J Mol Med
94:183–194.
Liu Y, Liu F, Grundke-Iqbal I, Iqbal K, Gong C-X (2011) Deficient brain insulin signalling
pathway in Alzheimer’s disease and diabetes. J Pathol 225:54–62.
Liu Z, Chen Y, Wang D, Wang S, Zhang YQ (2010) Distinct Presynaptic and
369
Postsynaptic Dismantling Processes of Drosophila Neuromuscular Junctions during
Metamorphosis. J Neurosci 30:11624–11634 Available at:
http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.0410-10.2010.
López-Otín C, Blasco MA, Partridge L, Serrano M, Kroemer G (2013) The hallmarks of
aging. Cell 153.
Louveau A, Harris TH, Kipnis J (2015) Revisiting the Mechanisms of CNS Immune
Privilege. Trends Immunol 36:569–577.
Luis EO, Ortega-Cubero S, Lamet I, Razquin C, Cruchaga C, Benitez BA, Lorenzo E,
Irigoyen J, Pastor MA, Pastor P (2014) Frontobasal gray matter loss is associated
with the TREM2 p.R47H variant. Neurobiol Aging 35:2681–2690.
Mair F, Hartmann FJ, Mrdjen D, Tosevski V, Krieg C, Becher B (2016) The end of
gating? An introduction to automated analysis of high dimensional cytometry data.
Eur J Immunol 46:34–43.
Marsh EK, May RC (2012) Caenorhabditis elegans, A model organism for investigating
immunity. Appl Environ Microbiol 78:2075–2081.
Matarin M, Salih DA, Yasvoina M, Cummings DM, Guelfi S, Liu W, NahabooSolim MA,
Moens TG, Paublete RM, Ali SS, Perona M, Desai R, Smith KJ, Latcham J,
Fulleylove M, Richardson JC, Hardy J, Edwards FA (2015a) A Genome-wide gene-
expression analysis and database in transgenic mice during development of
amyloid or tau pathology. Cell Rep 10:633–645 Available at:
http://dx.doi.org/10.1016/j.celrep.2014.12.041.
Matarin M, Salih DA, Yasvoina M, Cummings DM, Guelfi S, Liu W, Solim MAN, Moens
TG, Paublete RM, Ali SS, Perona M, Desai R, Smith KJ, Latcham J, Fulleylove M,
Richardson JC, Hardy J, Edwards FA (2015b) A Genome-wide Gene-Expression
Analysis and Database in Transgenic Mice during Development of Amyloid or Tau
Pathology. Cell Rep 10:633–644.
Matcovitch-Natan O et al. (2016) Microglia development follows a stepwise program to
regulate brain homeostasis. 353:aad8670--aad8670.
Matzinger P, Kamala T (2011) Tissue-based class control: The other side of tolerance.
Nat Rev Immunol 11:221–230.
Mayer F, Mayer N, Chinn L, Pinsonneault RL, Kroetz D, Bainton RJ (2009) Evolutionary
conservation of vertebrate blood-brain barrier chemoprotective mechanisms in
Drosophila. J Neurosci 29:3538–3550 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/19295159 [Accessed April 22, 2018].
McGeer PL, Itagaki S, Tago H, McGeer EG (1987) Reactive microglia in patients with
senile dementia of the Alzheimer type are positive for the histocompatibility
glycoprotein HLA-DR. Neurosci Lett 79:195–200.
McGuire VA, Gray A, Monk CE, Santos SG, Lee K, Aubareda A, Crowe J, Ronkina N,
Schwermann J, Batty IH, Leslie NR, Dean JLE, O’Keefe SJ, Boothby M, Gaestel M,
Arthur JSC (2013) Cross Talk between the Akt and p38 Pathways in Macrophages
Downstream of Toll-Like Receptor Signaling. Mol Cell Biol 33:4152–4165 Available
at: http://mcb.asm.org/cgi/doi/10.1128/MCB.01691-12.
McQuillan K, Lynch MA, Mills KHG (2010) Activation of mixed glia by Aβ-specific Th1
and Th17 cells and its regulation by Th2 cells. Brain Behav Immun 24:598–607.
Medinas DB, Hetz C (2013) Proteostasis impairment: At the intersection between
Alzheimer’s disease and diabetes. Cell Metab 18:771–772.
370
Medway C, Morgan K (2014) Review: The genetics of Alzheimer’s disease; putting flesh
on the bones. Neuropathol Appl Neurobiol 40:97–105.
Medzhitov R, Janeway CA (2002) Decoding the patterns of self and nonself by the
innate immune system. Science 296:298–300 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/11951031 [Accessed April 21, 2018].
Mei S, Gu H, Ward A, Yang X, Guo H, He K, Liu Z, Cao W (2012) p38 mitogen-
activated protein kinase (MAPK) promotes cholesterol ester accumulation in
macrophages through inhibition of macroautophagy. J Biol Chem 287:11761–
11768.
Meister M, Lemaitre B, Hoffmann JA (1997) Antimicrobial peptide defense in
Drosophila. BioEssays 19:1019–1026.
Melchiotti R, Gracio F, Kordasti S, Todd AK, de Rinaldis E (2016) Cluster stability in the
analysis of mass cytometry data. Cytometry A.
Merle NS, Noe R, Halbwachs-Mecarelli L, Fremeaux-Bacchi V, Roumenina LT (2015)
Complement System Part II: Role in Immunity. Front Immunol 6:1–26.
Mhatre SD, Paddock BE, Saunders AJ, Marenda DR (2013) Invertebrate models of
Alzheimer’s disease. J Alzheimers Dis 33:3–16 Available at:
http://content.iospress.com.ezp-prod1.hul.harvard.edu/articles/journal-of-
alzheimers-
disease/jad121204?resultNumber=0&totalResults=3218&start=0&q=invertebrates+
models+of+alzheimer%27s+disease&resultsPageSize=10&rows=10.
Mitchell K, Shah JP, Tsytsikova L V., Campbell AM, Affram K, Symes AJ (2014) LPS
antagonism of TGF-β signaling results in prolonged survival and activation of rat
primary microglia. J Neurochem 129:155–168.
Mócsai A, Ruland J, Tybulewicz VLJ (2010) The SYK tyrosine kinase: a crucial player in
diverse biological functions. 10:387–402.
Moser VA, Pike CJ (2017) Obesity Accelerates Alzheimer-Related Pathology in APOE4
but not APOE3 Mice. eneuro 4:ENEURO.0077-17.2017 Available at:
http://eneuro.sfn.org/lookup/doi/10.1523/ENEURO.0077-17.2017.
Mosher KI, Wyss-Coray T (2014) Microglial dysfunction in brain aging and Alzheimer’s
disease. Biochem Pharmacol 88:594–604.
Mrdjen D, Pavlovic A, Hartmann FJ, Schreiner B, Utz SG, Leung BP, Lelios I, Heppner
FL, Kipnis J, Merkler D, Greter M, Becher B (2018a) High-Dimensional Single-Cell
Mapping of Central Nervous System Immune Cells Reveals Distinct Myeloid
Subsets in Health, Aging, and Disease. Immunity 48:380–395.e6 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S1074761318300323 [Accessed March
20, 2018].
Mrdjen D, Pavlovic A, Hartmann FJ, Schreiner B, Utz SG, Leung BP, Lelios I, Heppner
FL, Kipnis J, Merkler D, Greter M, Becher B (2018b) High-Dimensional Single-Cell
Mapping of Central Nervous System Immune Cells Reveals Distinct Myeloid
Subsets in Health, Aging, and Disease. Immunity 48.
Munshi MN (2017) Cognitive dysfunction in older adults with diabetes: What a clinician
needs to know. Diabetes Care 40:461–467.
N’Diaye E-N, Branda CS, Branda SS, Nevarez L, Colonna M, Lowell C, Hamerman JA,
Seaman WE (2009) TREM-2 (triggering receptor expressed on myeloid cells 2) is a
phagocytic receptor for bacteria. J Cell Biol 184:215–223 Available at:
371
http://www.ncbi.nlm.nih.gov/pubmed/19171755%0Ahttp://www.pubmedcentral.nih.g
ov/articlerender.fcgi?artid=PMC2654305.
Nahmias AJ, Schollin J, Abramowsky C (2011) Evolutionary-developmental
perspectives on immune system interactions among the pregnant woman,
placenta, and fetus, and responses to sexually transmitted infectious agents. Ann N
Y Acad Sci 1230:25–47 Available at: http://doi.wiley.com/10.1111/j.1749-
6632.2011.06137.x [Accessed April 19, 2018].
Newell EW, Cheng Y (2016) Mass cytometry: blessed with the curse of dimensionality.
Nat Immunol 17:890–895.
Novakova M, Dolezal T (2011) Expression of drosophila adenosine deaminase in
immune cells during inflammatory response. PLoS One 6.
Nussbaum K, Burkhard SH, Ohs I, Mair F, Klose CSN, Arnold SJ, Diefenbach A,
Tugues S, Becher B (2017) Tissue microenvironment dictates the fate and tumor-
suppressive function of type 3 ILCs. J Exp Med 214:2331–2347 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/28698286 [Accessed March 21, 2018].
O’Neill LAJ, Golenbock D, Bowie AG (2013) The history of Toll-like receptors —
redefining innate immunity. Nat Rev Immunol 13:453–460 Available at:
http://www.nature.com/doifinder/10.1038/nri3446.
Oikonomou G, Shaham S (2011) The glia of caenorhabditis elegans. Glia 59:1253–
1263.
Okello A, Edison P, Archer HA, Turkheimer FE, Kennedy J, Bullock R, Walker Z,
Kennedy A, Fox N, Rossor M, Brooks DJ (2009) Microglial activation and amyloid
deposition in mild cognitive impairment: a PET study. Neurology 72:56–62.
Ou J, He Y, Xiao X, Yu TM, Chen C, Gao Z, Ho MS (2014) Glial cells in neuronal
development: Recent advances and insights from Drosophila melanogaster.
Neurosci Bull 30:584–594.
Palgi M, Lindstrom R, Peranen J, Piepponen TP, Saarma M, Heino TI (2009) Evidence
that DmMANF is an invertebrate neurotrophic factor supporting dopaminergic
neurons. Proc Natl Acad Sci 106:2429–2434 Available at:
http://www.pnas.org/cgi/doi/10.1073/pnas.0810996106.
Pandey UB, Nichols CD (2011) Human disease models in Drosophila melanogaster and
the role of the fly in therapeutic drug discovery. Pharmacol Rev 63:411–436
Available at:
http://www.ncbi.nlm.nih.gov/pubmed/21415126%5Cnhttp://www.pubmedcentral.nih.
gov/articlerender.fcgi?artid=PMC3082451.
Paresce DM, Ghosh RN, Maxfield FR (1996) Microglial cells internalize aggregates of
the Alzheimer’s disease amyloid $β$-protein via a scavenger receptor. Neuron
17:553–565.
Park-Min K-H, Ji J-D, Antoniv T, Reid AC, Silver RB, Humphrey MB, Nakamura M,
Ivashkiv LB (2009) IL-10 Suppresses Calcium-Mediated Costimulation of Receptor
Activator NF-$κ$B Signaling during Human Osteoclast Differentiation by Inhibiting
TREM-2 Expression. J Immunol 183:2444–2455.
Park J-S, Ji IJ, Kim D-H, An HJ, Yoon S-Y (2017) The Alzheimer’s Disease-Associated
R47H Variant of TREM2 Has an Altered Glycosylation Pattern and Protein Stability.
Front Neurosci 10:26043.
Park M, Yi J-W, Kim E-M, Yoon I-J, Lee E-H, Lee H-Y, Ji K-Y, Lee K-H, Jang J-H, Oh S-
372
S, Yun C-H, Kim S-H, Lee K-M, Song M-G, Kim D-H, Kang H-S (2015) Triggering
Receptor Expressed on Myeloid Cells 2 (TREM2) Promotes Adipogenesis and
Diet-Induced Obesity. Diabetes 64:117–127 Available at:
http://diabetes.diabetesjournals.org/lookup/doi/10.2337/db13-1869.
Pawlas N, Płachetka A, Kozłowska A, Broberg K, Kasperczyk S (2015) Telomere length
in children environmentally exposed to low-to-moderate levels of lead. Toxicol Appl
Pharmacol 287:111–118.
Pegram HJ, Andrews DM, Smyth MJ, Darcy PK, Kershaw MH (2011) Activating and
inhibitory receptors of natural killer cells. Immunol Cell Biol 89:216–224.
Perrin RJ, Fagan AM, Holtzman DM (2009) Multimodal techniques for diagnosis and
prognosis of Alzheimer’s disease. Nature 461:916–922.
Petersen AJ, Katzenberger RJ, Wassarman DA (2013) The innate immune response
transcription factor relish is necessary for neurodegeneration in a Drosophila model
of ataxia-telangiectasia. Genetics 194:133–142.
Petersen AJ, Rimkus SA, Wassarman DA (2012) ATM kinase inhibition in glial cells
activates the innate immune response and causes neurodegeneration in
Drosophila. Proc Natl Acad Sci 109:E656–E664 Available at:
http://www.pnas.org/cgi/doi/10.1073/pnas.1110470109.
Pham LN, Dionne MS, Shirasu-Hiza M, Schneider DS (2007) A Specific Primed
Immune Response in Drosophila Is Dependent on Phagocytes. PLoS Pathog 3:e26
Available at: http://dx.plos.org/10.1371/journal.ppat.0030026 [Accessed April 19,
2018].
Piirainen S, Youssef A, Song C, Kalueff A V., Landreth GE, Malm T, Tian L (2017)
Psychosocial stress on neuroinflammation and cognitive dysfunctions in
Alzheimer’s diseasethe emerging role for microglia? Neurosci Biobehav Rev
77:148–164 Available at: http://dx.doi.org/10.1016/j.neubiorev.2017.01.046.
Poltorak A, He X, Smirnova I, Liu MY, Van Huffel C, Du X, Birdwell D, Alejos E, Silva M,
Galanos C, Freudenberg M, Ricciardi-Castagnoli P, Layton B, Beutler B (1998)
Defective LPS signaling in C3H/HeJ and C57BL/10ScCr mice: Mutations in Tlr4
gene. Science (80- ) 282:2085–2088.
Pottier C, Wallon D, Rousseau S, Rovelet-Lecrux A, Richard AC, Rollin-Sillaire A,
Frebourg T, Campion D, Hannequin D (2013) TREM2 R47H variant as a risk factor
for early-onset alzheimer’s disease. J Alzheimer’s Dis 35:45–49.
Pradeu T, Cooper EL (2012) The danger theory: 20 years later. Front Immunol 3.
Ragheb RRT, Kim D, Bandyopadhyay A, Chahboune H, Bulutoglu B, Ezaldein H,
Criscione JM, Fahmy TM (2013) Induced clustered nanoconfinement of
superparamagnetic iron oxide in biodegradable nanoparticles enhances transverse
relaxivity for targeted theranostics. Magn Reson Med 70:1748–1760.
Ransohoff RM, Brown MA (2012) Innate immunity in the central nervous system. J Clin
Invest 122:1164–1171.
Rast JP, Smith LC, Loza-Coll M, Hibino T, Litman GW (2006) Genomic insights into the
immune system of the sea urchin. Science 314:952–956 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/17095692 [Accessed April 19, 2018].
Reiner AP, Beleza S, Franceschini N, Auer PL, Robinson JG, Kooperberg C, Peters U,
Tang H (2012) Genome-wide Association and Population Genetic Analysis of C-
Reactive Protein in African American and Hispanic American Women. Am J Hum
373
Genet 91:502–512.
Reitz C, Mayeux R, Alzheimer\textquoterights Disease Genetics Consortium (2013)
TREM2 and neurodegenerative disease. N Engl J Med 369:1564–1565.
Ressl S, Vu BK, Vivona S, Martinelli DC, Südhof TC, Brunger AT (2015) Structures of
C1q-like Proteins Reveal Unique Features among the C1q/TNF Superfamily.
Structure 23:688–699 Available at:
https://www.sciencedirect.com/science/article/pii/S096921261500043X [Accessed
March 21, 2018].
Richards DM, Endres RG (2014) The mechanism of phagocytosis: Two stages of
engulfment. Biophys J 107:1542–1553.
Rodrigues J, Brayner FA, Alves LC, Dixit R, Barillas-Mury C (2010) Hemocyte
differentiation mediates innate immune memory in Anopheles gambiae mosquitoes.
Science (80- ) 329:1353–1355.
Roe K, Gibot SÃ bastien, Verma S (2014) Triggering receptor expressed on myeloid
cells-1 (TREM-1): a new player in antiviral immunity? Front Microbiol 5:567.
Rogers J, Webster S, Lue LF, Brachova L, Civin WH, Emmerling M, Shivers B, Walker
D, McGeer P (1996) Inflammation and Alzheimer’s disease pathogenesis.
Neurobiol Aging 17:681–686.
Rowitch DH, Kriegstein AR (2010) Developmental genetics of vertebrate glial-cell
specification. Nature 468:214–222.
Royet J, Reichhart JM, Hoffmann JA (2005) Sensing and signaling during infection in
Drosophila. Curr Opin Immunol 17:11–17.
Saijo K, Glass CK (2011) Microglial cell origin and phenotypes in health and disease.
11:775–787.
Sakano H (2010) Review Neural Map Formation in the Mouse Olfactory System.
Neuron 67:530–542 Available at: http://dx.doi.org/10.1016/j.neuron.2010.07.003.
Sankowski R, Mader S, ValdÃ©s-Ferrer SI (2015) Systemic Inflammation and the Brain:
Novel Roles of Genetic, Molecular, and Environmental Cues as Drivers of
Neurodegeneration. Front Cell Neurosci 9:1–20 Available at:
http://journal.frontiersin.org/Article/10.3389/fncel.2015.00028/abstract.
Schlepckow K, Kleinberger G, Fukumori A, Feederle R, Lichtenthaler SF, Steiner H,
Haass C (2017) An Alzheimer-associated TREM2 variant occurs at the ADAM
cleavage site and affects shedding and phagocytic function. EMBO Mol
Med:e201707672--10.
Schmucker D, Chen B (2009) Dscam and DSCAM: Complex genes in simple animals,
complex animals yet simple genes. Genes Dev 23:147–156.
Schneider D, Tate AT (2016) Innate Immune Memory: Activation of Macrophage Killing
Ability by Developmental Duties. Curr Biol 26:R503–R505 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/27326712 [Accessed April 19, 2018].
Schneider DS (2007) How and Why Does a Fly Turn Its Immune System Off? PLoS Biol
5.
Schwartz M, Moalem G, Leibowitz-Amit R, Cohen IR (1999) Innate and adaptive
immune responses can be beneficial for CNS repair. Trends Neurosci 22:295–299.
Sekar A, Bialas AR, de Rivera H, Davis A, Hammond TR, Kamitaki N, Tooley K,
Presumey J, Baum M, Van Doren V, Genovese G, Rose SA, Handsaker RE,
Consortium SWG of the PG, Daly MJ, Carroll MC, Stevens B, McCarroll SA (2016)
374
Schizophrenia risk from complex variation of complement component 4. Nature.
Selkoe DJ (2001) Alzheimer’s disease: genes, proteins, and therapy. Physiol Rev
81:741–766.
Selkoe DJ (2012) Preventing alzheimer’s disease. Science (80- ) 337:1488–1492.
Selkoe DJ, Hardy J (2016) The amyloid hypothesis of Alzheimer’s disease at 25~years.
EMBO Mol Med 8:595–608.
Sen GL, Webster DE, Barragan DI, Chang HY, Khavari PA (2008) Control of
differentiation in a self-renewing mammalian tissue by the histone demethylase
JMJD3. Genes Dev 22:1865–1870 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/18628393 [Accessed April 17, 2018].
Severance EG, Gressitt KL, Buka SL, Cannon TD, Yolken RH (2014) Maternal
complement C1q and increased odds for psychosis in adult offspring. Schizophr
Res 159:14–19.
Shah D, Romero F, Zhu Y, Duong M, Sun J, Walsh K, Summer R (2015) C1q
Deficiency Promotes Pulmonary Vascular Inflammation and Enhances the
Susceptibility of the Lung Endothelium to Injury. J Biol Chem 290:jbc.M115.690784-
-29651.
Shah S et al. (2016) In Situ Transcription Profiling of Single Cells Reveals Spatial
Organization of Cells in the Mouse Hippocampus. Neuron 92:342–357.
Sharif O, Gawish R, Warszawska JM, Martins R, Lakovits K, Hladik A, Doninger B,
Brunner J, Korosec A, Schwarzenbacher RE, Berg T, Kralovics R, Colinge J,
Mesteri I, Gilfillan S, Salmaggi A, Verschoor A, Colonna M, Knapp S (2014a) The
Triggering Receptor Expressed on Myeloid Cells 2 Inhibits Complement
Component 1q Effector Mechanisms and Exerts Detrimental Effects during
Pneumococcal Pneumonia. PLOS Pathog 10:e1004167.
Sharif O, Gawish R, Warszawska JM, Martins R, Lakovits K, Hladik A, Doninger B,
Brunner J, Korosec A, Schwarzenbacher RE, Berg T, Kralovics R, Colinge J,
Mesteri I, Gilfillan S, Salmaggi A, Verschoor A, Colonna M, Knapp S (2014b) The
Triggering Receptor Expressed on Myeloid Cells 2 Inhibits Complement
Component 1q Effector Mechanisms and Exerts Detrimental Effects during
Pneumococcal Pneumonia Orihuela CJ, ed. PLoS Pathog 10:e1004167 Available
at: http://dx.plos.org/10.1371/journal.ppat.1004167 [Accessed April 14, 2018].
Sharif O, Knapp S (2008) From expression to signaling: Roles of TREM-1 and TREM-2
in innate immunity and bacterial infection. Immunobiology 213:701–713.
Shaw JL, Chang KT (2013) Nebula/DSCR1 Upregulation Delays Neurodegeneration
and Protects against APP-Induced Axonal Transport Defects by Restoring
Calcineurin and GSK-3β Signaling. PLoS Genet 9.
Shi Q, Chowdhury S, Ma R, Le KX, Hong S, Caldarone BJ, Stevens B, Lemere CA
(2017) Complement C3 deficiency protects against neurodegeneration in aged
plaque-rich APP/PS1 mice. Sci Transl Med 9:1–15.
Sierra A, de Castro F, del Río-Hortega J, Rafael Iglesias-Rozas J, Garrosa M,
Kettenmann H (2016) The “Big-Bang” for modern glial biology: Translation and
comments on Pío del Río-Hortega 1919 series of papers on microglia. Glia
64:1801–1840.
Singh RP, Dhawan P, Golden C, Kapoor GS, Mehta KD (1999a) One-way cross-talk
between p38(MAPK) and p42/44(MAPK). Inhibition of p38(MAPK) induces low
375
density lipoprotein receptor expression through activation of the p42/44(MAPK)
cascade. J Biol Chem 274:19593–19600.
Singh RP, Dhawan P, Golden C, Kapoor GS, Mehta KD (1999b) One-way cross-talk
between p38(MAPK) and p42/44(MAPK). Inhibition of p38(MAPK) induces low
density lipoprotein receptor expression through activation of the p42/44(MAPK)
cascade. J Biol Chem 274:19593–19600 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/10391894 [Accessed April 16, 2018].
Skoog (2007) Principles of Instrumental Analysis. Anal Chim Acta 152:314 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S0003267000849363.
Song W, Hooli B, Mullin K, Jin SC, Cella M, Ulland TK, Wang Y, Tanzi R, Colonna M
(2016) Alzheimer’s disease-associated TREM2 variants exhibit either decreased or
increased ligand-dependent activation. Alzheimer’s Dement.
Song X et al. (2017) Long non-coding RNA RPAIN regulates the invasion and apoptosis
of trophoblast cell lines via complement protein C1q. Oncotarget 8:7637–7646
Available at: http://www.oncotarget.com/fulltext/13826 [Accessed April 18, 2018].
Spits H, Bernink JH, Lanier L (2016) NK cells and type 1 innate lymphoid cells: Partners
in host defense. Nat Immunol.
Stein-Streilein J, Caspi RR (2014) Immune Privilege and the Philosophy of Immunology.
Front Immunol 5:110 Available at:
http://journal.frontiersin.org/article/10.3389/fimmu.2014.00110/abstract [Accessed
April 17, 2018].
Stein KF, Filosa SH (1964) A new mutation causing circling behavior in mice. J Hered
55:247–252.
Stephan AH, Barres BA, Stevens B (2012) The Complement System: An Unexpected
Role in Synaptic Pruning During Development and Disease. Annu Rev Neurosci
35:369–389 Available at: http://www.annualreviews.org/doi/10.1146/annurev-neuro-
061010-113810.
Stephan AH, Madison D V., Mateos JM, Fraser DA, Lovelett EA, Coutellier L, Kim L,
Tsai H-H, Huang EJ, Rowitch DH, Berns DS, Tenner AJ, Shamloo M, Barres BA
(2013) A Dramatic Increase of C1q Protein in the CNS during Normal Aging. J
Neurosci 33:13460–13474 Available at:
http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.1333-13.2013.
Stet RJM, Hermsen T, Westphal AH, Jukes J, Engelsma M, Verburg-Van Kemenade
BML, Dortmans J, Aveiro J, Savelkoul HFJ (2005) Novel immunoglobulin-like
transcripts in teleost fish encode polymorphic receptors with cytoplasmic ITAM or
ITIM and a new structural Ig domain similar to the natural cytotoxicity receptor
NKp44. Immunogenetics 57:77–89.
Stevens B, Allen NJ, Vazquez LE, Howell GR, Christopherson KS, Nouri N, Micheva
KD, Mehalow AK, Huberman AD, Stafford B, Sher A, Litke AM, Lambris JD, Smith
SJ, John SWM, Barres BA (2007) The Classical Complement Cascade Mediates
CNS Synapse Elimination. Cell 131:1164–1178 Available at:
https://www.sciencedirect.com/science/article/pii/S0092867407013554 [Accessed
April 19, 2018].
Stojanovic A, Correia MP, Cerwenka A (2013) Shaping of NK cell responses by the
tumor microenvironment. Cancer Microenviron 6:135–146 Available at:
http://www.pubmedcentral.nih.gov/articlerender.fcgi?artid=3717064&tool=pmcentre
376
z&rendertype=abstract.
Stork T, Bernardos R, Freeman MR (2012) Analysis of glial cell development and
function in Drosophila. Cold Spring Harb Protoc 7:1–17.
Streilein JW (1995) Immunological non-responsiveness and acquisition of tolerance in
relation to immune privilege in the eye. Eye 9:236–240 Available at:
http://www.nature.com/articles/eye199546 [Accessed April 19, 2018].
Streilein JW, Takeuchi M, Taylor AW (1997) Immune privilege, T-cell tolerance, and
tissue-restricted autoimmunity. Hum Immunol 52:138–143.
Streit WJ (2006) Microglial senescence: does the brain’s immune system have an
expiration date? Trends Neurosci 29:506–510.
Strohbehn G, Coman D, Han L, Ragheb RRT, Fahmy TM, Huttner AJ, Hyder F,
Piepmeier JM, Saltzman WM, Zhou J (2014) Imaging the delivery of brain-
penetrating PLGA nanoparticles in the brain using magnetic resonance. J
Neurooncol 121:441–449.
Stuart LM, Ezekowitz RAB (2005) Phagocytosis: Elegant complexity. Immunity 22:539–
550.
Sun A (2003) P38 MAP kinase is activated at early stages in Alzheimer’s disease brain.
Exp Neurol 183:394–405 Available at:
http://www.sciencedirect.com/science/article/pii/S0014488603001808.
Sunyer JO (2013) Fishing for mammalian paradigms in the teleost immune system. Nat
Immunol 14:320–326.
Swatton JE, Sellers LA, Faull RLM, Holland A, Iritani S, Bahn S (2004) Increased MAP
kinase activity in Alzheimer’s and Down syndrome but not in schizophrenia human
brain. Eur J Neurosci 19:2711–2719 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/15147305.
Szaruga M, Munteanu B, Lismont S, Veugelen S, Horré K, Mercken M, Saido TC, Ryan
NS, De Vos T, Savvides SN, Gallardo R, Schymkowitz J, Rousseau F, Fox NC,
Hopf C, De Strooper B, Chávez-Gutiérrez L (2017) Alzheimer’s-Causing Mutations
Shift Aβ Length by Destabilizing γ-Secretase-Aβn Interactions.
Cell VL - IS - SP - EP - PY - T2 - 170:443--448.e14.
Takahashi K, Rochford CDP, Neumann H (2005) Clearance of apoptotic neurons
without inflammation by microglial triggering receptor expressed on myeloid cells-2.
J Exp Med 201:647–657.
Tan T, Song Z, Yuan L, Xiong W, Deng X, Ni B, Chen Y, Deng H (2016) Genetic
analysis of TREM2 variants in Chinese Han patients with sporadic Parkinson’s
disease. Neurosci Lett 612:189–192.
Tasdemir-Yilmaz OE, Freeman MR (2014) Astrocytes engage unique molecular
programs to engulf pruned neuronal debris from distinct subsets of neurons. Genes
Dev 28:20–33.
Terai K, Walker DG, McGeer EG, McGeer PL (1997) Neurons express proteins of the
classical complement pathway in Alzheimer disease. Brain Res 769:385–390.
Thielens NM, Tedesco F, Bohlson SS, Gaboriaud C, Tenner AJ (2017) C1q: A fresh
look upon an old molecule. Mol Immunol 89:73–83 Available at:
https://www.sciencedirect.com/science/article/pii/S0161589017301591 [Accessed
March 21, 2018].
Thomas LM (2015) Current perspectives on natural killer cell education and tolerance:
377
emerging roles for inhibitory receptors. ImmunoTargets Ther:45 Available at:
http://www.dovepress.com/current-perspectives-on-natural-killer-cell-education-
and-tolerance-em-peer-reviewed-article-ITT.
Thompson BL, Levitt P, Stanwood GD (2009) Prenatal exposure to drugs: effects on
brain development and implications for policy and education. Nat Rev Neurosci
10:303–312.
Ting JP-Y, Davis BK (2005) CATERPILLER: A Novel Gene Family Important in
Immunity, Cell Death, and Diseases. Annu Rev Immunol 23:387–414 Available at:
http://www.annualreviews.org/doi/10.1146/annurev.immunol.23.021704.115616.
Tomiyama AJ, O’Donovan A, Lin J, Puterman E, Lazaro A, Chan J, Dhabhar FS,
Wolkowitz O, Kirschbaum C, Blackburn E, Epel E (2012) Does cellular aging relate
to patterns of allostasis?: An e`xamination of basal and stress reactive HPA axis
activity and telomere length. Physiol Behav 106:40–45 Available at:
https://www.sciencedirect.com/science/article/pii/S0031938411005361?via%3Dihu
b [Accessed April 17, 2018].
Town T, Laouar Y, Pittenger C, Mori T, Szekely CA, Tan J, Duman RS, Flavell RA
(2008) Blocking TGF-$β${\textendash}Smad2/3 innate immune signaling mitigates
Alzheimer-like pathology. Nat Med 81:741.
Town T, Nikolic V, Tan J (2005) The microglial “activation” continuum: from innate to
adaptive responses. J Neuroinflammation 2:24 Available at:
http://www.ncbi.nlm.nih.gov/pmc/articles/PMC1298325/pdf/1742-2094-2-24.pdf.
Tremblay M-È, Lecours C, Samson L, Sánchez-Zafra V, Sierra A (2015) From the Cajal
alumni Achúcarro and Río-Hortega to the rediscovery of never-resting microglia.
Front Neuroanat 9:45 Available at:
http://journal.frontiersin.org/article/10.3389/fnana.2015.00045/abstract [Accessed
April 17, 2018].
Trotta R, Fettucciari K, Azzoni L, Abebe B, Puorro KA, Eisenlohr LC, Perussia B (2000)
Differential Role of p38 and c-Jun N-Terminal Kinase 1 Mitogen-Activated Protein
Kinases in NK Cell Cytotoxicity. J Immunol 165:1782–1789 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/8871635 [Accessed April 17, 2018].
Turnbull IR, Gilfillan S, Cella M, Aoshi T, Miller M, Piccio L, Hernandez M, Colonna M
(2006) Cutting Edge: TREM-2 Attenuates Macrophage Activation. J Immunol
177:3520–3524.
Ulland TK, Song WM, Huang SC-C, Ulrich JD, Sergushichev A, Beatty WL, Loboda AA,
Zhou Y, Cairns NJ, Kambal A, Loginicheva E, Gilfillan S, Cella M, Virgin HW,
Unanue ER, Wang Y, Artyomov MN, Holtzman DM, Colonna M (2017) TREM2
Maintains Microglial Metabolic Fitness in Alzheimer’s Disease. Cell 170:649--
663.e13.
Ulland TK, Wang Y, Colonna M (2015) Regulation of microglial survival and proliferation
in health and diseases. Semin Immunol 27:410–415.
Uribe C, Folch H, Enriquez R, Moran G (2011) Innate and adaptive immunity in teleost
fish: a review. Vet Med (Praha) 56:486–503.
van Schaarenburg RA, Magro-Checa C, Bakker JA, Teng YKO, Bajema IM, Huizinga
TW, Steup-Beekman GM, Trouw LA (2016) C1q Deficiency and Neuropsychiatric
Systemic Lupus Erythematosus. Front Immunol 7:647 Available at:
http://journal.frontiersin.org/article/10.3389/fimmu.2016.00647/full [Accessed April
378
18, 2018].
Vasek MJ et al. (2016) A complement{\textendash}microglial axis drives synapse loss
during virus-induced memory impairment. Nature 534:538–543.
Vehmas AK, Kawas CH, Stewart WF, Troncoso JC (2003) Immune reactive cells in
senile plaques and cognitive decline in Alzheimer’s disease. Neurobiol Aging
24:321–331.
Vergadi E, Ieronymaki E, Lyroni K, Vaporidi K, Tsatsanis C (2017) Akt Signaling
Pathway in Macrophage Activation and M1/M2 Polarization. J Immunol 198:1006–
1014 Available at: http://www.ncbi.nlm.nih.gov/pubmed/28115590 [Accessed April
16, 2018].
Vestergaard G, Garrett RA, Shah SA (2014) CRISPR adaptive immune systems of
Archaea. RNA Biol 11:156–167 Available at:
http://www.tandfonline.com/doi/abs/10.4161/rna.27990 [Accessed April 22, 2018].
Vincent CM, Sharp NP (2014) Sexual antagonism for resistance and tolerance to
infection in Drosophila melanogaster. Proc R Soc B Biol Sci 281:20140987–
20140987 Available at:
http://rspb.royalsocietypublishing.org/cgi/doi/10.1098/rspb.2014.0987.
Vivier E, Pavert SA Van De, Cooper MD, Belz GT (2016) The evolution of innate
lymphoid cells. 17.
Wang J, Kubes P (2016) A Reservoir of Mature Cavity Macrophages that Can Rapidly
Invade Visceral Organs to Affect Tissue Repair. Cell 165:668–678 Available at:
http://dx.doi.org/10.1016/j.cell.2016.03.009.
Wang P, Guo Q, Zhou Y, Chen K, Xu Y, Ding D, Hong Z, Zhao Q (2017) Lack of
association between triggering receptor expressed on myeloid cells 2
polymorphism rs75932628 and late-onset Alzheimer’s disease in a Chinese Han
population. Psychiatr Genet:1 Available at:
http://insights.ovid.com/crossref?an=00041444-900000000-
99581%0Ahttp://www.ncbi.nlm.nih.gov/pubmed/29256968.
Wang Y, Cella M, Mallinson K, Ulrich JD, Young KL, Robinette ML, Gilfillan S, Krishnan
GM, Sudhakar S, Zinselmeyer BH, Holtzman DM, Cirrito JR, Colonna M (2015a)
TREM2 Lipid Sensing Sustains the Microglial Response in an
Alzheimer{\textquoteright}s Disease Model. Cell 160:1061–1071.
Wang Y, Cella M, Mallinson K, Ulrich JD, Young KL, Robinette ML, Gilfillan S, Krishnan
GM, Sudhakar S, Zinselmeyer BH, Holtzman DM, Cirrito JR, Colonna M (2015b)
TREM2 lipid sensing sustains the microglial response in an Alzheimer’s disease
model. Cell.
Wang Y, Ulland TK, Ulrich JD, Song W, Tzaferis JA, Hole JT, Yuan P, Mahan TE, Shi
Y, Gilfillan S, Cella M, Grutzendler J, DeMattos RB, Cirrito JR, Holtzman DM,
Colonna M (2016) TREM2-mediated early microglial response limits diffusion and
toxicity of amyloid plaques. J Exp Med 213:667–675 Available at:
http://www.jem.org/lookup/doi/10.1084/jem.20151948.
Webster S, Glabe C, Rogers J (1995) Multivalent Binding of Complement Protein C1q
to the Amyloid $β$-Peptide (A$β$) Promotes the Nucleation Phase of A$β$
Aggregation. Biochem Biophys Res Commun 217:869–875.
Webster SD, Yang AJ, Margol L, Garzon-Rodriguez W, Glabe CG, Tenner AJ (2000)
Complement Component C1q Modulates the Phagocytosis of A$β$ by Microglia.
379
Exp Neurol 161:127–138.
Wellen KE, Thompson CB (2010) Cellular Metabolic Stress: Considering How Cells
Respond to Nutrient Excess. Mol Cell 40:323–332.
Wendeln A, Degenhardt K, Kaurani L, Gertig M, Ulas T, Jain G, Wagner J, Häsler LM,
Wild K, Skodras A, Blank T, Staszewski O, Datta M, Centeno TP, Capece V, Islam
R, Kerimoglu C, Staufenbiel M (2018) Innate immune memory in the brain shapes
neurological disease hallmarks. Nature.
Westra ER, Buckling A, Fineran PC (2014) CRISPR-Cas systems: Beyond adaptive
immunity. Nat Rev Microbiol 12:317–326.
Wiens M, Korzhev M, Krasko A, Thakur NL, Perović-Ottstadt S, Breter HJ, Ushijima H,
Diehl-Seifert B, Müller IM, Müller WEG (2005) Innate immune defense of the
sponge Suberites domuncula against bacteria involves a MyD88-dependent
signaling pathway. Induction of a perforin-like molecule. J Biol Chem 280:27949–
27959 Available at: http://www.ncbi.nlm.nih.gov/pubmed/15923643 [Accessed April
22, 2018].
Williams DW (2005) Cellular mechanisms of dendrite pruning in Drosophila: insights
from in vivo time-lapse of remodeling dendritic arborizing sensory neurons.
Development 132:3631–3642 Available at:
http://dev.biologists.org/cgi/doi/10.1242/dev.01928.
Williams MJ (2007) Drosophila Hemopoiesis and Cellular Immunity. J Immunol
178:4711–4716 Available at:
http://www.jimmunol.org/cgi/doi/10.4049/jimmunol.178.8.4711.
Willmann E, Pandurovic V, Bohlson S (2017) Complement component C1q regulates
AMPK and ERK1/2 activation in mouse and human macrophages. J Immunol
198:222.6 Available at: http://www.jimmunol.org/content/198/1_Supplement/222.6.
Winkler EA, Bell RD, Zlokovic B V. (2011) Central nervous system pericytes in health
and disease. Nat Neurosci 14:1398–1405.
Wirths O, Bayer TA (2010) Neuron loss in transgenic mouse models of Alzheimer’s
disease. Int J Alzheimers Dis 2010 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/20871861 [Accessed June 14, 2018].
Wood W, Jacinto A (2007) Drosophila melanogaster embryonic haemocytes: Masters of
multitasking. Nat Rev Mol Cell Biol 8:542–551.
Wu K, Byers DE, Jin X, Agapov E, Alexander-Brett J, Patel AC, Cella M, Gilfilan S,
Colonna M, Kober DL, Brett TJ, Holtzman MJ (2015a) TREM-2 promotes
macrophage survival and lung disease after respiratory viral infection. J Exp Med
212:681–697 Available at: http://www.ncbi.nlm.nih.gov/pubmed/25897174
[Accessed April 16, 2018].
Wu K, Byers DE, Jin X, Agapov E, Alexander-Brett J, Patel AC, Cella M, Gilfilan S,
Colonna M, Kober DL, Brett TJ, Holtzman MJ (2015b) TREM-2 promotes
macrophage survival and lung disease after respiratory viral infection. J Exp Med
212:681–697.
Yang Y (2005) Loss of Neuronal Cell Cycle Control in Ataxia-Telangiectasia: A Unified
Disease Mechanism. J Neurosci 25:2522–2529 Available at:
http://www.jneurosci.org/cgi/doi/10.1523/JNEUROSCI.4946-04.2005.
Yeh FL, Wang Y, Tom I, Gonzalez LC, Sheng M (2016) TREM2 Binds to
Apolipoproteins, Including APOE and CLU/APOJ, and Thereby Facilitates Uptake
380
of Amyloid-Beta by Microglia. Neuron 91:328–340.
Yoshida K, Maekawa T, Zhu Y, Renard-Guillet C, Chatton B, Inoue K, Uchiyama T,
Ishibashi K, Yamada T, Ohno N, Shirahige K, Okada-Hatakeyama M, Ishii S (2015)
The transcription factor ATF7 mediates lipopolysaccharide-induced epigenetic
changes in macrophages involved in innate immunological memory. Nat Immunol
16:1034–1043 Available at: http://www.nature.com/doifinder/10.1038/ni.3257
[Accessed April 19, 2018].
Yu HH, Awasaki T, Schroeder MD, Long F, Yang JS, He Y, Ding P, Kao JC, Wu GYY,
Peng H, Myers G, Lee T (2013) Clonal development and organization of the adult
Drosophila central brain. Curr Biol 23:633–643.
Yu J-T, Jiang T, Wang Y-L, Wang H-F, Zhang W, Hu N, Tan L, Sun L, Tan M-S, Zhu X-
C, Tan L (2014) Triggering receptor expressed on myeloid cells 2 variant is rare in
late-onset Alzheimer’s disease in Han Chinese individuals. Neurobiol Aging
35:937.e1-3.
Yu X, Buttgereit A, Lelios I, Utz SG, Cansever D, Becher B, Greter M (2017) The
Cytokine TGF-β Promotes the Development and Homeostasis of Alveolar
Macrophages. Immunity 47:903–912.e4 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/29126797 [Accessed March 21, 2018].
Yuan P, Condello C, Keene CD, Wang Y, Bird TD, Paul SM, Luo W, Colonna M,
Baddeley D, Grutzendler J (2016) TREM2 Haplodeficiency in Mice and Humans
Impairs the Microglia Barrier Function Leading to Decreased Amyloid Compaction
and Severe Axonal Dystrophy. Neuron 90:724–739.
Zhang B et al. (2013) Integrated systems approach identifies genetic nodes and
networks in late-onset Alzheimer’s disease. Cell 153:707–720 Available at:
http://linkinghub.elsevier.com/retrieve/pii/S0092867413003875 [Accessed
December 4, 2017].
ZHANG W, LIU HT (2002) MAPK signal pathways in the regulation of cell proliferation in
mammalian cells. Cell Res 12:9–18.
Zhang X, Zhang Y (2012) DBL-1, a TGF- , is essential for Caenorhabditis elegans
aversive olfactory learning. Proc Natl Acad Sci 109:17081–17086 Available at:
http://www.pnas.org/cgi/doi/10.1073/pnas.1205982109.
Zhang Y, Shan B, Boyle M, Liu J, Liao L, Xu T, Yates JR (2014) Brain proteome
changes induced by olfactory learning in drosophila. J Proteome Res 13:3763–
3770.
Zhao Y, Wu X, Li X, Jiang L-L, Gui X, Liu Y, Sun Y, Zhu B, Piña-Crespo JC, Zhang M,
Zhang N, Chen X, Bu G, An Z, Huang TY, Xu H (2018) TREM2 Is a Receptor for β-
Amyloid that Mediates Microglial Function. Neuron 97:1023–1031.e7 Available at:
http://www.ncbi.nlm.nih.gov/pubmed/29518356 [Accessed March 20, 2018].
Zheng H, Liu CC, Atagi Y, Chen XF, Jia L, Yang L, He W, Zhang X, Kang SS,
Rosenberry TL, Fryer JD, Zhang YW, Xu H, Bu G (2016) Opposing roles of the
triggering receptor expressed on myeloid cells 2 and triggering receptor expressed
on myeloid cells-like transcript 2 in microglia activation. Neurobiol Aging 42:132–
141 Available at: http://dx.doi.org/10.1016/j.neurobiolaging.2016.03.004.
Zhong L, Chen X-F, Wang T, Wang Z, Liao C, Wang Z, Huang R, Wang D, Li X, Wu L,
Jia L, Zheng H, Painter M, Atagi Y, Liu C-C, Zhang Y-W, Fryer JD, Xu H, Bu G
(2017) Soluble TREM2 induces inflammatory responses and enhances microglial
381
survival. J Exp Med 35:jem.20160844--934.e10.
Zhou J, Fonseca MI, Pisalyaput K, Tenner AJ (2008) Complement C3 and C4
expression in C1q sufficient and deficient mouse models of
Alzheimer{\textquoteright}s disease. J Neurochem 106:2080–2092.
Zhu M, Li D, Wu Y, Huang X, Wu M (2014) TREM-2 Promotes Macrophage-Mediated
Eradication of Pseudomonas aeruginosa via a PI3K/Akt Pathway. Scand J Immunol
79:187–196 Available at: http://doi.wiley.com/10.1111/sji.12148 [Accessed April 18,
2018].
Zugasti O, Ewbank JJ (2009) Neuroimmune regulation of antimicrobial peptide
expression by a noncanonical TGF-beta signaling pathway in Caenorhabditis
elegans epidermis. Nat Immunol 10:249–256 Available at:
http://www.ncbi.nlm.nih.gov/entrez/query.fcgi?cmd=Retrieve&db=PubMed&dopt=Ci
tation&list_uids=19198592.

Abstract (if available)

Abstract Alzheimer’s disease (AD) now afflicts more than 5 million Americans, and an effective treatment or cure does not exist. Risk for late onset AD (LOAD) is likely determined by complex interplay between environmental and genetic factors. Multiple genome-wide association studies and gene network analyses have implicated two major innate immune pathways as key risk factors for late-onset Alzheimer’s disease: Triggering Receptor Expressed on Myeloid cells 2 (TREM2) and protein complement component C1q. Further, a more recent GWAS specifically identified peripheral macrophage genes as AD risk factors. We show that both TREM2 and C1q proteins are dysregulated in AD mouse brains and in LOAD brains vs. non-demented controls. While classically thought to regulate different types of immune responses, our data raise the tantalizing possibility that the TREM2 and C1q pathways intersect in AD. Specifically, we have made the novel observation that C1q, TREM2 and Aβ physically interact, forming a heteromeric complex on innate immune cells. We further show that small Aβ assemblies preferentially bind to TREM2, while C1q more avidly associates with Aβ aggregates. In the APP/PS1 mouse model, compound genetic loss of TREM2 and C1q abrogates extracellular regulated kinase (ERK)1/2 innate immune signaling. Peripheral mononuclear phagocyte experiments demonstrate that phagocytosis of C1q opsonized Aβ is both Trem2 and pERK1/2 dependent, whereas phagocytosis of Aβ alone requires Trem2 and p-p38 mitogen activated protein kinase (MAPK). For the first time, these results shed light on the biological significance of the TREM2-C1q-Aβ neuroimmune complex in transition from the healthy CNS to the AD brain, and have key implications for myeloid Aβ phagocytosis, clearance, and immunoproteostasis.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Macrophage-specific Stat3 deletion in a mouse model of Alzheimer’s disease

PDF

The role of inflammation in mediating effects of obesity on Alzheimer's disease

PDF

Adaptive immunity in the central nervous system

PDF

Estrogen and progesterone-based hormone therapy and the development of Alzheimer's disease

PDF

Role of CD33 structure and function in Alzheimer’s disease

PDF

Using neuroinformatics to identify genomic and proteomic markers of suboptimal aging and Alzheimer's disease

PDF

Nebula/DSCR1 upregulation preserves axonal transport and memory function in a Drosophila model for Alzheimer's disease

PDF

Innate immune response to amyloid-beta: relevance to Alzheirmer’s disease and neuroinflammation

PDF

Understanding the role of APP and DYRK1A in human brain pericytes

PDF

Affective neuropsychiatric symptoms and neural connectivity in the early stages of Alzheimer’s disease

PDF

Exploring the relationship between menopausal hot flushes and Alzheimer's disease biomarkers: a cross-sectional analysis in postmenopausal women

PDF

Insulin sensitivity in cognition, Alzheimer's disease and brain aging

PDF

The role of specific gut microbiota in regulating the development of Alzheimer抯 disease

PDF

Expression and purification of extracellular domain of human CD33 in Escherichia coli and Pichia pastoris

PDF

Long-term treadmill exercise delays the onset of motor dysfunction, cognitive impairments, and mood disturbances in the CAG 140 mouse model of Huntington's disease via restoration of dopamine neu...

Asset Metadata

Creator Leung, Brian Pak Yan (author)

Core Title TREM2 and C1q signaling regulates immunoproteostasis in Alzheimer's disease

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Neuroscience

Publication Date 08/07/2018

Defense Date 08/04/2018

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag aging,Alzheimer's disease,C1q,complement,Immune System,immunity,immunoproteostasis,macrophage,microglia,neurodegeneration,Neuroscience,OAI-PMH Harvest,TREM2

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Finch, Caleb (committee chair), Chang, Karen (committee member), Levitt, Pat (committee member), Tower, John (committee member), Town, Terrence Christopher (committee member)

Creator Email brian.p.leung@gmail.com,brianple@USC.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-56563

Unique identifier UC11670327

Identifier etd-LeungBrian-6676.pdf (filename),usctheses-c89-56563 (legacy record id)

Legacy Identifier etd-LeungBrian-6676.pdf

Dmrecord 56563

Document Type Dissertation

Format application/pdf (imt)

Rights Leung, Brian Pak Yan

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA