University of Southern California Dissertations and Theses

Regulating functional amyloid formation: the promiscuous behavior of the Orb2A N-terminal amphipathic region

(USC Thesis Other)

Regulating functional amyloid formation: the promiscuous behavior of the Orb2A N-terminal amphipathic region

PDF

Download

Open document

Flip pages

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

Regulating functional amyloid formation:
The promiscuous behavior of the Orb2A N-terminal
amphipathic region

Maria A. Soria

A Dissertation Presented to the faculty of the
USC Graduate School

In partial fulfillment of the requirements for the
Doctorate of Philosophy in Medical Biophysics

Keck School of Medicine

University of Southern California

10 May 2019

2
Dedication

This dissertation is dedicated to:
my husband, for his constant love and support,
my parents, for their guidance and encouragement,
and God for sustaining all things.

3
Acknowledgments

Funding support from from the USC Graduate School Provost Fellowship,
the Whitehall Foundation and the National Institutes of Health NIGMS
Award R01GM110521.

My advisor, Ansgar Siemer, for his dedicated mentorship, insightful
scientific assistance, perseverance and unceasing optimism.

My lab members over the years: Silvia Cervantes, Alexander Falk, Thalia
Bajakian, Rachel Service, Ninad Agashe, Jina Kim, Majima Sarkar,
Samridhi Garg, Connor Hurd, Bethany Caulkins, Shruti Bendre, and
Rajashree Venkatraman for both their scientific and emotional support.

My committee members, Dr. Ralf Langen and Dr. Robert Farley, for their
fruitful scientific advice and mentorship.

All fellow first floor lab members, including J. Mario Isas, Mark Ambroso,
Alan Okada, Kazuki Teranishi, Natalie Kegulian, Meixin Tao, Alan Situ,
Thomas Shmidt, Benjamin Frey, Sean Chung, Fleur Lobo, Nitin Pandey,
Jobin Varkey, Anoop Rawat, Jose Bravo, Gincy George, and Naomi Sta.
Maria for all of their help and support.

Finally, my collaborators, Dr. Tobias Ulmer, Dr. Seth Ruffins, Dr. Andre
Ouellette, Dr. Hans Vogel, Dr. Janet Oldak and those at the Scripps
Institute Mass Spectroscopy Core.

4
Table of Contents

Page
Chapter 1: Introduction to Orb2 aggregation

6
Amyloid formation of the CPEB homolog Orb2 is important for LTM
6
Orb2A is important for initiating and regulating Orb2 aggregation
7
Comparison of Orb2A aggregation to pathological amyloid aggregation
9
Regulation of Orb2A

11
Chapter 2: The functional amyloid Orb2A binds to lipid
membranes

13
Introduction
13
Materials and Methods
15
Results
20
Discussion
29
Conclusions
32
Supporting Material

33
Chapter 3: Calmodulin binds the CPEB homolog Orb2A at the
aggregation-prone N-terminal region

40
Introduction
40
Materials and Methods
42
Results
45
Discussion

49
Chapter 4: Conclusions and outlook

51
Appendix A: Effects of calmodulin on huntingtin and α-
synculein

55
Rationale
55
Materials and Methods
55
Results and Discussion

56

5

Table of Contents (Cont)

Page
Appendix B: Aggregation of the N-terminal amino acids of
Orb2A

60
Rationale
60
Materials and Methods
60
Results and Discussion

64
Appendix C: Modulating intracellular Ca
2+
levels changes
Orb2A-eGFP puncta

70
Rationale
70
Materials and Methods
70
Results and Discussion

73
References 74

6
Chapter 1: Introduction to Orb2 aggregation

By Maria Soria

Amyloid formation of the CPEB homolog Orb2 is important for LTM

Amyloids are protein aggregates that form long β-sheet rich fibrils. They have a
cross-β sheet structure where the direction of the β-sheets runs perpendicular to the
axis of the fibril (1). These amyloids are self-perpetuating, with the fibril serving as a
template for the folding of soluble, non aggregated protein, resulting in its elongation.
Originally, amyloids were identified as being associated with diseases such as
Alzheimer’s, Huntington’s and Parkinson’s diseases and many more (reviewed
extensively by (2)). However, a growing number of amyloids are being discovered in
nature that serve a function in healthy organisms. For example, curli fibers are amyloids
that form outside bacteria as a component of the protective biofilm (3). In humans, the
Pmel17 protein forms amyloids that allow for melanin deposits to pigment hair, skin and
eyes (4). Another protein, cytoplasmic polyadenylation element binding protein (CPEB),
forms amyloid fibrils that are necessary for long term memory (LTM) (5).

Aplysia CPEB was the first CPEB protein discovered to have importance in LTM
by forming amyloid-like aggregates, and this was soon followed by a similar hypothesis
for a Drosophila CPEB homolog, Orb2 (6, 7). Kausik Si along with Susan Lindquist and
Eric Kandel showed Aplysia CPEB had prion-like properties, and that the monomeric
and prion-like form had different activities (5). Both Orb2 and Aplysia CPEB have a
glutamine-rich (Q-rich) region, which is reminiscent of other prion-like or amyloid
proteins such as yeast prions and the disease-related Q-repeat proteins such as
huntingtin (5). Because of this, the Q-rich region was of immediate relevance for
functional studies in fruit fly memory. When the Q-rich region was deleted, the short-
term memory of the fruit fly was not affected, however the long-term memory was
dramatically reduced when one allele was missing the Q-rich region and completely
abolished when both alleles were missing the Q-rich region (6). It was then concluded
that the Q-rich region of Orb2 was necessary for LTM and hypothesized that Orb2 may
be able to form prion-like aggregates similar to Aplysia CPEB (6).

7
Majumdar et al. showed that Orb2 did indeed form amyloid-like aggregates (8). A
high molecular weight Orb2 oligomer was present in extracts from fly heads but not fly
bodies, and this oligomer was resistant to many common denaturing methods, including
heat, guanidine-HCl, urea and detergents (8). Along with this, the amount of oligomeric
Orb2 increased upon neuronal stimulation by neurotransmitters, which further supported
the idea that aggregation was an important aspect of Orb2’s function in LTM (8). While
there are up to six splice variants at the orb2 gene locus, only two of them, Orb2A and
Orb2B, carry the RNA binding domains that make them CPEB homologs. While both of
these isoforms were able to form amyloid in vitro and were found together in aggregates
in immunoprecipitation assays, Orb2A has proven to be very important for regulating
aggregation in vivo.

Orb2A is important for initiating and regulating Orb2 aggregation

Orb2A and Orb2B were observed to be functionally different as well as being
structurally different. Orb2B knockout was lethal in flies, but Orb2A knockout flies
developed normally (9). Also, although Orb2B was found throughout the neuronal
cytoplasm, Orb2A levels were so low as to be nearly undetectable (8, 9). Flies with the
Orb2A knockout were normal in short term memory, but had a large detriment in LTM
(9). This again shows the very important role of Orb2A in LTM function. To try to
understand the role of the Q-rich domain and RRMs in each isoform, a series of deletion
mutants were generated. Flies with the Q-rich domain in only Orb2A and not in Orb2B
had no significant LTM differences from wild type flies (9). Furthermore, flies with intact
Orb2B but with Orb2A missing the Q-rich domain did not have significant LTM (9). This
shows that Orb2A is necessary for LTM, and that Orb2B is not sufficient for LTM. It was
also shown that when the RRMs of Orb2B are mutated, LTM is decreased, while when
the RRMs of Orb2A are mutated, there is no significant change in LTM (9). This shows
that the RRMs do not play a functional role in Orb2A in LTM, while in Orb2B they are
necessary for LTM function.

Other published experiments specifically detail the importance of the N-terminal
Q-rich domain of Orb2A. It was found that deletion of the unique eight amino acids at

8
the very N-terminal of Orb2A reduced Orb2 puncta in S2 cells, and a similar result was
seen for Orb2A that was missing the Q-rich domain (8). During a random mutagenesis
to screen for Orb2A point mutations that could reduce puncta formation, mutations in
the first 88 amino acids of Orb2A were greatly overrepresented, with many of these
mutations occurring within the first 8 amino acids (8). The mutation with the greatest
effect, F5Y, was tested in vivo to see the effects of reduced aggregation on LTM. Flies
with the F5Y mutation did indeed have a significant reduction in LTM after 48 hours as
compared to wild-type flies (8). This again supports the importance of Orb2A in the
aggregation and LTM function of Orb2 in fruit flies.

In order to test the role of aggregation in the LTM function of Orb2, an in-vitro-
reconstituted translation assay was designed using an mRNA called Tequila that is
known to be translated after activation by Orb2 (10). A luciferase gene was added to the
Tequila mRNA, creating a reporter gene and allowing researchers to track Orb2 activity
based on the luciferase activity. In the presence of fly head lysates, monomeric forms of
Orb2A and Orb2B repressed translation of the reporter gene, while oligomeric forms of
Orb2A and Orb2B activated translation of the reporter gene (10). This mainly occurred
through either stabilization or destabilization of the poly-A tail (10). In this way,
oligomerization of Orb2 acts as a molecular switch to control protein production. This
switch could potentially be the means by which Orb2 marks specific synapses in LTM
and maintains long term potentiation (LTP) (11).

An important part of understanding any amyloid structure is locating the specific
sequence that is involved in forming the β-sheet rich core of the amyloid fibril. Protease
accessibility studies, which can identify aggregated areas based on those areas which
are inaccessible by proteases, showed that the β-sheet core is potentially located within
the first 162 amino acids of Orb2A (12). The same group also tested the ability of
certain Orb2A constructs to aggregate with another construct missing the first 88 amino
acids of Orb2A. Wild-type Orb2A was able to induce aggregation of Orb2AΔ88, which
by itself was diffuse and unable to aggregate (12). Orb2AΔ88-162 was able to
aggregate by itself and co-aggregated with wild-type Orb2A. However, Orb2AΔ1-162
did not aggregate by itself or when mixed with wild-type Orb2 (12). This study revealed
that while the first 88 amino acids of Orb2A may be important for initiating amyloid

9
aggregation, the following residues through 162 may be responsible for recruitment into
already formed aggregates (12).

Structural studies reported something similar. Amyloid fibrils were visible on
electron microscopy for the first 88 amino acids of Orb2A, but also for the first 22 amino
acids of Orb2A (13). Solid-state NMR showed that for amyloid fibrils formed by the first
88 amino acids of Orb2A, only residues within the first 22 amino acids were static
enough to be a part of the amyloid core (13). Electron paramagnetic resonance showed
that these very N-terminal residues were in an in-register parallel β-sheet conformation
in these amyloid fibrils (13). Interestingly, the first eight amino acids of Orb2A are not
found in Orb2B, and solid-state NMR spectra for full length Orb2A shows a different
static core than for the first 88 amino acids (13). This is consistent with the cell fusion
experiments where amino acids 88-162 were able to recruit new monomers to
aggregates but not form de-novo aggregates themselves. This means there could be
another region of the protein that is able to aggregate into a β-sheet core after the initial
aggregation at the N-terminus.

Comparison of Orb2A aggregation to pathological amyloid aggregation

While we know that Orb2 forms aggregates that, far from being toxic, actually
perform necessary functions for the fruit fly, it still remains unclear why some amyloids
are associated with disease while others are functional and non-toxic. While it was
widely assumed that the amyloid fibril itself was the source of toxicity in disease, many
recent studies have shown that the fibril itself may be inert, and one of many possible
pre-fibrillar oligomers is responsible for the toxic effects. For example, the number of Aβ
amyloid plaques in a patient’s brain does not necessarily correlate with the extent of
disease symptoms (14). There have also been Aβ plaques found in brains of patients
whose cognitive abilities matched their healthy counterparts (15). Thus, the reason for
toxicity may lie in one or more of the various protein folding intermediates along the
pathway to amyloid. In non-toxic functional amyloids, this folding process is likely to be
controlled by speeding through toxic intermediates so as to prohibit the toxic effects
from occurring, or by bypassing these toxic intermediates altogether.

10
Using Orb2A as a model for understanding functional aggregation compared to
disease-related aggregation has shown some interesting results, and more work is
currently underway to fully understand this process. Single-molecule force spectroscopy
showed that a peptide containing the first 162 amino acids of Orb2A has a wide variety
of conformations possible in the monomeric state (12), and this is consistent with later
studies showing that the first 88 amino acids adopts a relatively random coil
conformation (16, 17). As time progresses, Orb2A forms aggregates recognizable by
the A11 antibody, which was originally developed to detect Aβ (12). However, while Aβ
oligomers may be detected for days by A11, the amount of Orb2A detected by A11
decreases over the same time scale. Another Aβ sensitive antibody, OC, was also able
to detect more mature Orb2A conformers (12). Amphotericin B, an amyloid-inhibiting
molecule, was able to arrest Orb2A in the A11 detecting state. This complex added to
cells causes cell toxicity and death within 24 hours (12). These data suggest that it is
possible for Orb2A to form toxic oligomeric species, but that perhaps these species are
too transient to cause any harm to cells.

Orb2 has routinely been compared to huntingtin, due to the similarity of Orb2’s
glutamine-rich region to huntingtin’s poly-Q region. Orb2A co-expressed with huntingtin
(Htt) Q128 were found in the same puncta together. The puncta also appeared larger
than when Orb2A was expressed on its own (12). A small, anti-amyloidogenic peptide,
polyglutamine-binding peptide 1 (QBP1), which inhibits Htt Q128 aggregation, also
inhibited Orb2A aggregation. When QBP1 was expressed in fruit fly neurons, long-term
memory was inhibited (12). These studies would suggest that Orb2 and Htt have related
amyloid structures and aggregation pathways. However, solid-state NMR of the first 88
amino acids of Orb2A, which contains the full glutamine-rich region, shows that the
glutamines are not actually a part of the β-sheet core of that Orb2A fragment (13). This
is quite different from Htt, where the poly-Q region forms the β-sheet core of the amyloid
fibril (18, 19). More work is needed to understand the precise amyloid-folding pathway
of Orb2 in order to be able to elucidate differences and similarities with the Htt pathway
and other disease-related amyloids, which may prove insightful into which phases of
aggregation are most toxic.

11
Regulation of Orb2A

A few studies have begun the process of looking into how Orb2 aggregation
might be regulated, focusing on the aggregation initiating isoform A. In vivo, Orb2A was
shown to associate with transducer of Erb-B2 (Tob), and this interaction stabilized
Orb2A which prevented degradation (20). Tob also promoted phosphorylation by
recruiting Lim Kinase (LimK) to phosphorylate Orb2A, which destabilized the Tob-Orb2A
interaction, but continued to promote Orb2A stability (20). Protein phosphatase 2A
(PP2A) was able to dephosphorylate Orb2A (20). Interestingly, both PP2A and LimK
activity are regulated by synaptic stimulation and have been shown to be important for
LTP, with LimK being upregulated and PP2A being downregulated (21, 22). While LimK
was the only kinase specifically tested in this study, it was apparent that other kinases
could phosphorylate Orb2A as well (20). It seems likely that phosphorylation is an
important regulator of Orb2A availability in a spacial and temporal manner. However,
the specific sites of these phosphorylations, the order in which they occur, and their
specific structural and functional effects have yet to be determined.

Few other regulatory mechanisms have been thoroughly investigated. One
potential mechanism could involve lipid binding. Many other amyloid forming proteins
are affected in their ability to aggregate by lipid membranes, such as
α-synuclein,
huntingtin, and human islet amyloid polypeptide (23–26). This is also seen in functional
amyloids such as Pmel17 (27). We originally noticed that Orb2A was enriched in the
synaptic membrane fraction of the fruit fly brain lysate (8). Along with that, Orb2A has a
slightly higher α-helical propensity than Orb2B, and the unique amino acids found at the
Orb2A N-terminus add an amphipathic nature to that region (12, 16). In the following
chapter of this dissertation, which was originally published in Biophysical Journal in
2017 (16), we explore the ability of Orb2A to bind to anionic lipid membranes by an
amphipathic helix at its N-terminus, and the effects of this binding on amyloid
aggregation.

Knowing that Orb2A has the ability to form an amphipathic helix and that there
were a few lysine residues located near the same area, we were also interested in its
ability to bind to the protein calmodulin (CaM). CaM preferentially binds to positively

12
charged amphipathic helices and is known to be an important Ca
2+
signal integrator in
LTP (28, 29). For example, Ca
2+
bound CaM activates CaM kinase II (CamKII), which in
turn was shown to phosphorylate CPEB in mouse hippocampal neurons (30). Thus far,
there has been no investigation into the effects of Ca
2+
on Orb2 aggregation or activity,
nor on potential direct CaM interactions with Orb2A. In chapter 3 of this dissertation,
which is being prepared for peer-reviewed publication, we show that Ca
2+
/CaM does
indeed interact with Orb2A and that this interaction inhibits amyloid formation. We also
discuss in Appendix C how Ca
2+
levels may affect puncta formation in S2 cells.

Ultimately, we hope that the following studies will provide a means of forming
new testable hypotheses for understanding how Orb2, and specifically Orb2A, may be
regulated in vivo. Understanding how Orb2 aggregation is regulated will allow for
comparison to the unregulated aggregation of disease-related amyloids. Knowing which
steps of Orb2 aggregation are highly regulated, and how the intermediates compare
with those of disease amyloids, we should be able to gain valuable insights into which
species are toxic, and on top of that, how these species are avoided or neutralized in
functional settings. This information may provide new therapy strategies for amyloid-
related diseases.

13
Chapter 2: The functional amyloid Orb2A binds to lipid
membranes*

Maria A. Soria
1
, Silvia A. Cervantes
1
, Thalia H. Bajakian
1
, and Ansgar B. Siemer
1

1
Department of Biochemistry and Molecular Medicine, Zilkha Neurogenetic Institute,
Keck School of Medicine, University of Southern California, 1501 San Pablo Street, Los
Angeles, California 90033, United States;

*Originally published in Volume 113 Issue 1 of Biophysical Journal, June 2017, as “The
functional amyloid Orb2A binds to lipid membranes”.

Introduction
Orb2 is a CPEB homolog found in Drosophila melanogaster that regulates mRNA
translation at the synapse and is important for long-term memory (6, 8, 31). In the fruit
fly brain, Orb2 has two isoforms, Orb2A and Orb2B. These isoforms differ only at the N-
terminus: Orb2B has an extended serine-rich N-terminus, while Orb2A has a shorter N-
terminus with only 8 unique amino acids prior to the first amino acid common to both
isoforms (Figure 1). Both isoforms share a glutamine/histidine (Q/H)-rich region, which
is followed by a glycine (G)-rich region and multiple RNA binding domains at the C-
terminus (Figure 1). Both Orb2 isoforms can form amyloid fibrils in vitro, and are found
as co-aggregates in vivo, where Orb2A is only found in trace amounts and Orb2B forms
the bulk of the aggregate (8). This aggregation is responsible for synapse specific
activation of dormant mRNA and stabilization of long-term memory (8, 10). Mounting
evidence suggests that Orb2A aggregation is key to initiating and regulating the
amyloid-like aggregation and function of Orb2B. Identifying factors that regulate the
aggregation of Orb2A is therefore necessary to our understanding of the regulation of
Orb2 function in general (8, 10). Recently, Cervantes et al. showed that the first 20 N-
terminal amino acids of Orb2A are sufficient to form amyloid fibrils on their own (13). It
was posited that this unique N-terminal region of Orb2A is important for the regulation of
amyloid formation, and thus long-term memory formation.
One common mechanism that can influence amyloid formation is lipid membrane
interaction. Several disease-related amyloids, such as α-synuclein (αS), human islet

14
amyloid peptide (IAPP), and huntingtin (HTT), have all been shown to interact with lipid
membranes via amphipathic helices, a process which affects their aggregation kinetics.
For example, αS is intrinsically disordered in solution. However, in the presence of
anionic lipid membranes αS forms an amphipathic α-helix and this interaction affects
amyloid fibril formation depending on vesicle curvature and composition (23, 24, 32–
34). IAPP, which is also intrinsically disordered in solution, interacts with anionic lipid
membranes by forming an amphipathic helix which increases the rate of aggregation
kinetics (26, 35, 36). The N-terminal 17 amino acids of HTT have been shown to be
particularly important in facilitating lipid-induced changes in aggregation kinetics (25).
These residues form an amphipathic helix that is involved in initiating amyloid formation
in the absence of membranes, but also enables HTT exon 1 to bind and aggregate on
lipid membrane surfaces (25). Lipid membrane interaction does not just affect
aggregation of disease-related amyloids. Pmel17, a protein that forms functional
amyloids involved in animal pigmentation, interacts with lysophospholipid monomers,
micelles and vesicles containing lysophospholipids (27). Lysophospholipids are
considerably enriched in early melanosomal interlumenal vesicle membranes, which
need to form highly curved membrane structures (37, 38). The
interaction can either
accelerate or inhibit aggregation depending on the charge and concentration of the
lysophospholipids present (27).

Figure 1: Orb2A has an amphipathic N-
terminus. A) Comparison of the domain
organization of Orb2A and Orb2B. Blue
indicates the mRNA binding motifs, green
indicates the glycine rich region, orange
indicates the glutamine/histidine rich region,
purple indicates the N-terminal region of
Orb2A, and gray indicates the serine-rich N-
terminus of Orb2B. The first 16 residues of
Orb2A are plotted on the helical wheel.
Yellow signifies hydrophobic residues and
pink shows charged or polar residues. B)
Sequence of Orb2A1-88. Purple indicates the
amphipathic N-terminus, orange indicates the
glutamine/histidine (Q/H)-rich, and green
indicates part of the glycine (G)-rich region.
C) Orb2A fragments used in circular
dichroism experiments.

15
Since Orb2A is found enriched in the synaptic membrane fraction in vivo (8), we
hypothesized that Orb2A could interact with lipid membranes. The unique N-terminus of
Orb2A might allow for the formation of an amphipathic helix. Based on the membrane
interaction of other amyloid-forming proteins, it is possible that membrane interaction
could affect the amyloid formation of Orb2A. To our knowledge, the interaction of Orb2
with lipid membranes has not yet been studied. We therefore wanted to address the
following questions: Is the N-terminal domain of Orb2A able to interact with lipid
membranes with an amphipathic helix similar to other amyloid forming proteins, and if
so which role does the charge and curvature of the membranes play for this interaction?
Finally, does the membrane interaction of Orb2A influence its ability to form fibrils?
The unique N-terminus of Orb2A was our primary focus, as this region has been
shown to be important for Orb2 aggregation and function. In the following we show that
the N-terminus of Orb2A does interact with lipids and, similar to other amyloid-forming
proteins, lipid composition and degree of vesicle curvature affect this interaction. Finally,
we also show that the presence of lipid vesicles affects the aggregation capability of the
N-terminus of Orb2A.
Materials and Methods
Protein and peptide plasmid constructs
Orb2A1-22 was ordered from AnaSpec, Inc. (Fremont, CA) as a synthetic
peptide and aliquoted under N
2
gas. The aliquots were stored away from light in a
vacuum desiccator containing Drierite.
Wild type Orb2A1-88 was cloned into a pET28b vector from a full length Orb2A
vector provided by Dr. Kausik Si (Stower’s Institute, Kansas City, MO). Site-directed
mutagenesis was used to mutate the wild-type C10 in the Orb2A1-88 plasmid to
methionine for electron paramagnetic resonance (EPR) studies. A QuickChange II XL
Site Directed Mutagenesis Kit (Agilent Technologies Inc, Santa Clara, CA) was used to
introduce the V6C, L18C and G84C mutants into the C10M plasmid. Vectors for the
Q23C, Q34C, and G12C mutants were produced by Genscript USA Inc. (Pescataway,
NJ). Orb2A21-88 was cloned into a pET28b vector from a full length Orb2A vector and a

16
tryptophan residue was added after the 6xHis affinity tag to allow for accurate
concentration determination also using the QuickChange II XL Site Directed
Mutagenesis Kit.
Protein
and
peptide
expression
and
purification

Orb2A1-88, Orb2A21-88 and Orb2A1-88 V6C, L18C, and G84C EPR mutants
were expressed in E. Coli Rosetta 2 (DE3) (EMD Millipore, Billerica MA). CaCl
2

chemically competent cells were transformed with the appropriate plasmid, and one
colony was used to inoculate a 25 mL culture in LB Miller medium with the appropriate
antibiotics, at 30°C. After ~16 hours, this culture was diluted into 1L LB Miller medium
with the appropriate antibiotics and was grown until OD
600
= 0.6. The culture was then
induced with isopropyl β-D-1-thiogalactopyranoside (IPTG), and protein was expressed
at 37°C for 4 hours or 25°C for ~16 hours. Cell pellets were spun down in a Sorvall
SLC-6000 rotor (Thermo Fisher Scientific Inc.) at 4000 rpm for 20 min at 4°C and used
immediately or stored at -80°C. Orb2A1-88 G12C, Q23C and Q34C mutants were
expressed similarly using BL21 (DE3) cell cultures.
Cells were suspended in Denaturing Buffer (8M Urea, 10mM Tris, 100mM
NaH
2
PO
4
, 0.05% β-mercaptoethanol, pH 8.0) and lysed using a Q125 Ultrasonic
Homogenizer (QSonica, Newton, CT). The solution was then centrifuged at 20,000 rpm
for 20 minutes using a Sorvall SS-34 rotor. The supernatant was collected and vacuum
filtered through a glass microfiber filter if needed. The supernatant was then poured
onto a pre-equilibrated Ni-NTA column equilibrated with Denaturing Buffer and
incubated with gentle shaking for ~1 hr. For samples meant for EPR studies the
incubation was skipped. The flow through was collected and the column was washed
with Denaturing Buffer containing 0.5% Triton-X followed by Denaturing Buffer
containing 500 mM NaCl and Denaturing Buffer at pH 6.75. Then the column was
washed with Renaturing Buffer (200 mM NaCl, 50 mM NaH
2
PO
4
, 10% glycerol, 0.05%
(v/v) β-mercaptoethanol, pH 8.0) and then with Renaturing Buffer containing 20 mM
imidazole. Then the protein was eluted in Renaturing Buffer using a 150 mM, 200 mM
and 250 mM imidazole step gradient. For samples intended for EPR experiments, the
column was first washed with Renaturing Buffer containing 20 mM imidazole, then
washed with Renaturing Buffer pH 7.4 with no β-mercaptoethanol, and finally eluted

17
with 250 mM imidazole at pH 7.4 with no β-mercaptoethanol. Elution fractions were
frozen in liquid nitrogen, and stored at -80°C.
Protein Concentration Determination
Protein concentration was determined using the compound fluorescamine
according to the FluoProbes protocol adapted from Bohlen et al. (39). Samples and
lysozyme standards were boiled in 0.1 M borate, 1% sodium dodecyl sulfate (SDS), pH
9.0 and then cooled to room temperature. 125 µl of 0.5 mg/ml fluorescamine in acetone
was added dropwise to each with vortexing. Fluorescence was measured at 25ºC using
the Eppendorf (Hamburg, Germany) PlateReader AF2200, with an excitation
wavelength of 360 nm with a 35 nm slitwidth, and an emission wavelength of 465 nm
with a 35 nm slitwidth. The protein sample concentration was then calculated from a
standard curve of known lysozyme concentrations.
Lipid vesicles
The phospholipids 1-palmitoyl-2-oleoyl-sn-glycero-3-phosphocholine (POPC), 1-
palmitoyl-2-oleoyl-sn-glycero-3-phospho-(1'-rac-glycerol) (sodium salt) (POPG), and 1-
palmitoyl-2-oleoyl-sn-glycero-3-phospho-L-serine (sodium salt) (POPS) were obtained
in chloroform from Avanti Polar Lipid, Inc. (Alabaster, Alabama). The appropriate
phospholipids in chloroform were mixed, and the chloroform was evaporated with N
2

gas for at least 30 min. The dried lipids were then placed in a vacuum desiccator for at
least 2 hours. Then 1 mL of the appropriate buffer was added to match the buffer of the
protein samples in the following experiments, and this mixture was vortexed until no
visible lipid residue remained, forming multilamellar vesicles (MLVs).
To make small unilamellar vesicles (SUV)s, the MLVs were sonicated with a
QSonica Q-125 ultrasonic homogenizer (Laboratory Supply Network, Inc., Atkinson,
NH) three times for one minute at 70% amplitude, with two-minute breaks on ice, until
the solution became clear. The lipid solution was then centrifuged at 13,500 rpm (FA45-
30-11 Eppendorf, Hamburg, Germany) for 20 min. The supernatant, which contained
the SUVs, was collected.
To make large unilamellar vesicles (LUVs), the MLVs were put through five
freeze-thaw cycles using liquid nitrogen and a 42ºC water bath, and then extruded at

18
25ºC on a Mini-Extruder (Avanti Polar Lipids, Inc., Alabaster, Alabama) the first ten
times through a 400 nm nitrocellulose filter, then eleven times through a 100 nm filter.
Vesicle sizes for SUVs and LUVs were verified using electron microscopy, and both
were consistent with what has been observed for vesicles produced using similar
protocols (i.e. 25 nm and 100 nm for SUVs and LUVs, respectively).
Circular dichroism
Purified fractions of Orb2A1-88 and Orb2A21-88 were transferred into CD Buffer
(75mM NaHPO
4
, 100 mM NaF, pH 7.6) using a PD-10 column. Orb2A1-22 synthetic
peptide was dissolved and disaggregated in hexafluoroisopropanol with 0.1%
trifluoroacitic acid, then lyophilized. The lyophilized Orb2A1-22 was then re-dissolved in
CD Buffer. Concentrations of Orb2A1-88 and Orb2A1-22 were measured using the
fluorescamine assay described above. The concentration of Orb2A21-88 was
determined using UV absorption at 280 nm, as Orb2A21-88 has no lysine residues to
react with the fluorescamine. We then used the concentrations of each protein sample
(each between 10-13 µM, except for the sample used in the POPC only vesicle titration
which was 5 µM) as well as the concentration of the lipid stocks to calculate the
appropriate volume of lipids to reach the desired protein to lipid ratios.
CD spectroscopy was performed at 25ºC on a Jasco J-810 Spectropolarimeter
(Jasco Inc., Easton, MD). Data points were measured every 0.5 nm at a scan speed of
50 nm/min from 260 nm to 195 nm, and the data were averaged over 16 scans.
Background measurements were taken of buffer with each lipid concentration used in
the experiments and averaged over 12 scans (16 scans for the Orb2A1-22 experiments
due to low signal). The background measurements were then subtracted from the
protein-lipid samples with the same concentration lipid. Mean residue ellipticity (MRE)
was calculated as previously described (40) taking into account the change in
concentration with the addition of lipids.
Estimation of fraction of helicity of a peptide undergoing a coil to helix transition
at a constant temperature was performed as previously described (41–43) using the
following equation:
𝑓
!"
=
!
!"#
! !
!"
!
!
! !
!"
(Equation 1)

19

where 𝑓
!"
is the fraction of helicity, 𝜃
!"#
is the observed MRE at 222 nm, 𝜃
!"
= 2220−
53 𝑇 is the theoretical MRE at 222 nm for random coil conformation at T temperature,
and 𝜃
!
= −44000+250𝑇 ∗ (1−
!
!
) is the theoretical value of the MRE at 222 nm for a
fully helical peptide of N residues in length.

Electron paramagnetic resonance
Purified protein aliquots (pH 7.4 with no β-mercaptoethanol) were thawed and 3 µl of 40
mg/ml MTSL spin label (Toronto Research Chemicals, Inc., North York, ON, Canada)
was added for a final concentration of over 200 µM (in great excess of protein). The
protein and MTSL were incubated at room temperature for ~1 hr. This mixture was then
diluted 1:15 in dH
2
O and added to a pre-equilibrated cation exchange column (S
Ceramic HyperD F, Pall Life Sciences, Port Washington, NY). The flowthrough was
collected, and the column was washed with 7 mL dH
2
O. The protein was then eluted
with 2 mL 8 M guanidine hydrochloride and dialyzed at 4ºC three times against 1L of
Renaturing Buffer (200 mM NaCl, 50 mM NaH
2
PO
4
, 10% glycerol, pH 7.4). The dialyzed
protein was then centrifuged at 13,500 rpm on an Eppendorf (Hamburg, Germany)
FA45-30-11 rotor for 20 min. The concentration of protein in the supernatant was
determined using the fluorescamine assay described above (each concentration
between 3 and 13 µM). The protein was then mixed with the desired ratio of lipids
(calculated similarly to the CD experiments) in the same Renaturing Buffer and then
loaded into a boro capillary tube (0.6 mm inner diameter, 0.84 mm outer diameter, Vitro-
Com, Mt. Lakes, NJ).
Continuous wave EPR spectra were collected at 25ºC on a Bruker X-band EMX
spectrometer (Bruker Biospin Corporation, Billerica, MA) using 15 scans at a scan width
of 150 gauss in an HS cavity. The microwave power was 12.6 mW. For kinetics curves,
5 scans were taken every 15 min, using the same parameters.
Amplitude measurements were read from the parameters file given by the Bruker
WinEPR program. The program reads the value of the highest and lowest point in the
spectrum, which can then be subtracted to obtain the amplitude. Linewidths were
measured as the full width of the half height of each absorption spectrum obtained by

20
integrating and baseline correcting the original EPR spectra. These calculations were
done with a custom python scripts using SciPy (44). Linewidth uncertainties were
determined as described by Tseitlin et al. for Gaussian lineshapes (45). This analysis
relies on the signal to noise ratio and the number of points per linewidth to determine
the uncertainty. The uncertainties for all linewidths in this paper are given in Tables S1-
S3. To determine the dissociation constant, EPR center line amplitudes at different lipid
concentrations were measured in triplicate and fitted to a one-site binding hyperbolic
function using the method of least squares in SciPy (44) and plotted with matplotlib (46).
The average, standard deviation, and error for each set of amplitudes are reported in
Table S4.
Electron microscopy
Purified Orb2A1-88 was buffer exchanged using a PD-10 column into 1 M urea,
100 mM NaCl, 10 mM HEPES, 1 mM DTT, pH 7.6. Urea was present to prevent
bundling of fibers for clearer visualization using EM (13). Concentration of protein in the
elution was determined using the fluorescamine protocol described above to be 6 µM.
The elution was then split in two and one fraction was combined with SUVs (containing
2 POPS:1 POPC) for a final protein to lipid ratio of 1:100, calculated similarly to the
above CD and EPR experiments. As a control, we made a sample containing only
buffer and the same concentration of SUVs. Copper formvar grids (Electron Microscopy
Sciences, Hatfield, PA) were placed on sample droplets for 5 min, stained for 2 min in
1% uranyl acetate, and then washed twice with 1% uranyl acetate and once with dH
2
O.
Grids were made of each sample every two days for 14 days, and were imaged on a
JEOL (Tokyo, Japan) JEM-1400 electron microscope.
Results
The Orb2A N-terminus becomes more helical with the addition of lipids
To investigate the lipid membrane binding properties of Orb2A, we used the
hydropathy analysis program MPEx to analyze the primary sequence of Orb2A (47). We
found that the first 15 amino acids, when plotted on a helical wheel, revealed a possible
amphipathic helix (Figure 1a). Hydrophobic residues on one side of the helix allow for

21
interaction with the hydrophobic lipid chains, and the polar residues on the other side
are able to interact with the phospholipid head groups and the aqueous environment
(48). This hypothesis was further supported when we noticed inconsistencies in our
Orb2A1-88 samples based on purification protocol (data not shown). Lipid extraction of
these samples followed by phosphorus determination revealed the presence of
phosphate in some samples depending on the purification protocol (Figure S1).
Consequently, additional column washing steps were added to the Ni
2+
affinity
chromatography purification protocol as described in the Materials and Methods for all
experiments performed in this paper.
To test whether the N-terminus of Orb2A can indeed form a helical structure, we
measured the circular dichroism (CD) of N-terminal Orb2A fragments in the presence of
lipid membranes. We designed these fragments by splitting the first 88 amino acids of
Orb2A into two sections (Figure 1b,c). The first section is the amphipathic region, which
Figure 2: The N-terminus of Orb2A becomes more helical with the addition of anionic SUVs. A) CD
spectra of Orb2A1-88 with increasing amounts of lipid vesicles (2 POPS:1 POPC). The minimum around
201 nm, which is indicative of a random coil conformation, shifts towards 208 nm, and the minimum at 222
nm, indicative of the amount of helical structure, decreases with the addition of lipids. B) Difference
between spectra of 1:0 and 1:100 protein to lipid ratio from panel A showing a typical helical CD spectrum
with minima at 208 nm and 222 nm. C) CD spectra of Orb2A21-88, which is missing the N-terminal
amphipathic region, show no change with increasing amounts of lipid vesicles (2 POPS:1 POPC). D) CD
spectra of the N-terminal 22 amino acids of Orb2A, which contains the amphipathic sequence, show a
strong increase in α-helicity with increasing amounts of lipid vesicles (2 POPS:1 POPC). The minimum
shifts from 201 nm to 208 nm and a local minimum appears at 222 nm.

22
contains the first 8 amino acids unique to Orb2A and other polar and non-polar residues
that could be a part of an amphipathic helix. The second section starts at the first
glutamine residue, continues through the glutamine/histidine (Q/H)-rich region, and
includes the first few residues of the glycine (G)-rich region. These first and second
sections are divided by two prolines, P21 and P22, which, from a structural perspective,
create a natural boundary between these two sections. Consequently, we studied three
N-terminal Orb2A fragments: the first 88 amino acids of Orb2A (Orb2A1-88), the
amphipathic first section (Orb2A1-22), and the second section, amino acids 21-88
(Orb2A21-88) (Figure1c).
Small unilamellar vesicles (SUVs) composed of POPS and POPC in a 2:1 ratio
were titrated into the CD cuvette with known concentrations of peptide. Both Orb2A1-88
and Orb2A1-22 showed an increase in α-helicity with the addition of lipid vesicles
(Figure 2). Changes in structure with each addition of lipid lessened towards a protein
to lipid ratio of 1:100, where binding appeared to saturate. Not surprisingly, the increase
in helicity for Orb2A1-88 is smaller compared to Orb2A1-22 since the proposed helical
region forms a smaller fraction of the overall peptide. Though the change seen for
Orb2A1-88 is small, it is clear from the difference spectrum in Figure 2b that the SUV-
containing sample has gained α-helical structure. From the MRE at 222 nm of Orb2A1-
88 at a 1:100 protein to lipid ratio, we estimated the percent helicity to be about 22%
which corresponds to 21 amino acids (see Materials and Methods). This number is
consistent with the CD data of the other peptides in Figure 2c and d. To more precisely
identify the extent of the α-helical region, we measured EPR as described below. We
attribute the fact that the CD spectrum of Orb2A1-88 in the absence of lipids (Figure
2a) does not go through the isosbestic point to a systematic concentration error that
affects the first point of our lipid titration. Orb2A21-88 does not show any change in
structure upon the addition of lipids as seen in Figure 2c.
The increase in helicity is dependent on lipid charge
Considering the anionic nature of the vesicles in the above CD experiments, we
decided to test if the charge of the vesicles was important for binding. Therefore, we
measured CD spectra of Orb2A1-88 in the presence of vesicles made from two different
lipid compositions. One titration was done with SUVs composed of only POPC, which

23
as a zwitterionic lipid, forms neutral membranes. The other titration was done with a 2:1
ratio of POPG:POPC, in which POPG replaces the POPS from the previous CD
experiments as an alternative negatively charged lipid. In Figure 3a and 3b, a similar
increase in α-helicity of Orb2A1-88 with the addition of the anionic POPG/POPC
vesicles can be seen compared to the previous results with POPS/POPC. The
difference spectrum of the spectrum with no lipids (1:0 protein to lipid) subtracted from
the spectrum with the highest lipid concentration (1:100 protein to lipid) revealed that
the change in structure is again α-helical. However, with the POPC only vesicles, we
observed no significant structural changes up to the 1:100 lipid to protein ratio used in
our titration experiments (Figure 3c,d). The difference spectrum in Figure 3d confirms
this.
Figure 3: Orb2A1-88 interaction requires negatively charged membranes. A) CD spectra showing
an increase in helicity for Orb2A1-88 with increasing amounts of SUVs containing POPG and POPC in a
2:1 ratio. The observed spectral change is similar to that in Figure 2a, with the minima shifting towards
208 nm and a decrease in MRE at 222 nm indicative of an increase in α helicity. B) Subtraction of the
spectrum in the absence of SUVs in panel A from the 1:100 protein to lipid spectrum in the same panel.
A typical α-helix spectrum is the result. C) CD spectra of Orb2A1-88 with the addition of POPC only
SUVs. The only change in spectra is the loss of signal around 201 nm due to the systematic error for this
first spectrum. No wavelength shift in the minimum or increase in MRE at 222 is observed. D)
Subtraction of the CD spectrum in the absence of SUVs in panel C from the 1:100 protein to lipid
spectrum in the same panel. The 1:0 spectrum was normalized to the 1:100 spectrum at their lowest
points before subtraction to account for the concentration error. No significant α-helical structure is
forming with the addition of POPC SUVs.

24
EPR reveals increased rigidity in the N-terminus with the addition of lipid vesicles
To identify more specifically which region in the Orb2A N-terminus interacts with
lipids, we performed site-directed spin labeling (SDSL) of various cysteine mutants of
Orb2A1-88 followed by continuous wave EPR (CW EPR) (Figure 4). For these titration
experiments, MTSL (S-(1-oxyl-2,2,5,5-tetramethyl-2,5-dihydro-1H-pyrrol-3-yl)methyl
methanesulfonothioate) labeled, soluble Orb2A1-88 was mixed with increasing amounts
of SUVs (2 POPS:1 POPC) and continuous wave X-band EPR spectra were recorded
(Figure 4a, Figure S2). The EPR spectrum, which is very sensitive to the dynamics of
the MTSL spin label, broadens and consequently its amplitude decreases when the spin
label becomes more immobilized. The decrease in amplitude can reflect both the
transition from a disordered state into a state with stable secondary structure as well as
the reduced tumbling of the protein upon membrane interaction. Generally, the EPR
spectra showed changes up to a protein to lipid ratio of 1:100. The change in inverse
linewidth was plotted to give a quantitative idea of how much the dynamics changed at
each labeled site (Figure 4b). We saw the strongest changes in EPR linewidth between
the unbound and fully bound states in the very N-terminal region at residues 6, 10 and
12 (Figure 4b). The sites following these, 18 and 23, showed intermediate EPR line
broadening in the presence of SUVs. Residue 34 and 84 both showed little to no
change in their EPR spectra. To make sure that the EPR linewidth changes we
observed were not due to lipid vesicle independent amyloid formation, we measured the
EPR spectra of residues 6 and 12, that were previously shown to be part of the amyloid
core of Orb2A1-88 (13), before and after the series of titration experiments in the
absence of lipids. Both of these residues showed no SUV independent spectral
changes indicating that the observed changes are caused by the presence of the lipid
vesicles (Figure S3). To test whether the MTSL label influenced the structure or ability
of Orb2A1-88 to interact with lipid vesicles, CD spectra we recorded of MTSL labeled
proteins and did not show any difference in structure or lipid binding properties (Figure
S4).
The EPR linewidth is dominated by the sharpest component, corresponding, in
our case, to the unbound protein. This dependence makes linewidth measurements
unsuitable for determining dissociation constants. The amplitude of the EPR spectra, in

25
contrast, is directly proportional to the ratio of bound vs unbound protein and, therefore,
more suitable for determining dissociation constant (see Supporting Text 1). To
determine a protein-lipid dissociation constant, we measured the amplitude of Orb2A1-
88 labeled at residue 6, which showed some of the strongest EPR line broadening with
the addition of anionic SUVs (Figure 4c). Since the receiver gain influences EPR
Figure 4: EPR reveals extent of Orb2A’s N-terminal lipid binding site. A) Location of each cysteine
mutant for MTSL spin labeling in Orb2A1-88. Purple represents the N-terminal amphipathic region, orange
the Q/H rich region, and green the glycine rich region. B) EPR spectra of MTSL labeled Orb2A1-88
cysteine mutants and wild type C10. Spectra acquired in the absence of lipids and spectra with 1:100
protein to lipid ratio are overlaid. Spectra are normalized by the central line intensity for better visualization
of the increase in linewidth. C) Change in inverse linewidth with the addition of lipids for each residue
measured by EPR. D) Binding curve shows the typical rectangular hyperbolic curve for a single ligand
species binding to another uniform population. The amplitude of Orb2A1-88 labeled at residue 6 was
plotted versus lipid concentration (black circles) as increasing amounts of lipid are added, and the best-fit
curve was calculated (black line) using the method of least squares.

26
intensity, it was held constant during these experiments. The protein concentration was
also held constant (3µM) with increasing amounts of lipids. Consequently, the number
of spins in each EPR spectrum is the same, and we were able to use direct
measurements of amplitude to plot the binding curve shown in Figure 4d. The data
follow a typical rectangular hyperbolic curve for the binding of a specific ligand species
to one particular uniform receptor population (49), and the fit resulted in a dissociation
constant of K
d
=58 ± 18 µM.
Orb2A1-88 membrane binding depends on membrane curvature
The degree of membrane curvature is an important factor determining the lipid
binding capacity of other amyloid-forming proteins (32, 50). To test whether the Orb2A1-
88 membrane affinity depends on membrane curvature, we repeated our EPR based
membrane-binding assay with large unilamellar vesicles (LUVs). LUVs, which have an
average diameter of 100 nm, are significantly less curved compared to the SUVs used
in the experiments described above, which have an average diameter of 25 nm. Using
CW EPR on an Orb2A1-88 sample that was MTSL spin labeled at residue 6, we
compared the lipid to protein ratio required for saturation when adding SUVs versus
LUVs (Figure 5). We plotted inverse linewidth to quantify the visual changes in the
Figure 5: Orb2A1-88 binds more strongly to
SUVs compared to LUVs. EPR spectra of
Orb2A1-88 spin labeled at residue 6 with
increasing amounts of LUVs and SUVs. The
protein to lipid ratios are given for each spectrum.
The binding was saturated by a protein to lipid
ratio of 1:100 with SUVs, but had yet to show full
saturation at a ratio of 1:500 with LUVs. Inverse
linewidth is plotted to quantify visual
observations.

27
spectra. With the addition of SUVs, MTSL-labeled Orb2A1-88 V6C was saturated at a
protein to lipid ratio of 1:100. With the addition of LUVs, Orb2A1-88 V6C showed a
much more gradual increase in EPR linewidth and did not saturate at a protein to lipid
ratio of 1:500.
Anionic SUVs inhibit the formation of Orb2A1-88 amyloid fibrils
Does the lipid membrane interaction of Orb2A1-88 have an effect on its ability to form
amyloid fibrils? To answer this question, we used electron microscopy to evaluate the
aggregation state of Orb2A1-88 in the presence and absence of anionic SUVs
(containing 2 POPS:1 POPC). Both samples were imaged at regular intervals to identify
the approximate time required for fibril formation and detect any morphological
differences (Figure 6a, Figure S5). Without SUVs, small Orb2A1-88 fibrils could be
detected after 2 days. These fibrils persisted throughout the experiment and slowly
elongated over time. We observed 30-50 nm long fibrils after two days, while after 21
days, we observed fibrils that ranged from 100-300 nm in length (Figure S5). The
Orb2A1-88 sample that contained SUVs never showed fibrillar structures similar to
those found in the sample without lipids over the course of the experiment. However,
besides the lipid vesicles, we detected small rod-like structures around 30 nm long that
resemble grains of rice visible on day 2. These structures were not seen in control
samples that contained only buffer and SUVs. Therefore, these particles must be a
result of the Orb2A1-88 SUV mixture. Whether these samples are only protein or a
mixture of protein and lipid cannot be determined from our electron micrographs. These
rod-like structures appeared to slowly elongate over time, up to 50 or 60 nm after 21
days (Figure S5), but never reached the length of the fibrils that were seen for Orb2A1-
88 in the absence of SUVs.
We used EPR with SDSL at the wild type C10 residue to confirm that the rod-like
structures were not structurally similar to amyloid fibrils (Figure 6b and c). Orb2A1-88
was mixed with anionic SUVs (2 POPS:1 POPC) for a final protein to lipid ratio of 1:100.

28

Figure 6: Aggregation of Orb2A1-88 into amyloid fibrils is inhibited by the presence of anionic
SUVs. A) Transmission electron microscope images taken of Orb2A1-88 with and without anionic
SUVs (2 POPS:1 POPC) over a period of 12 days. The black bar represents 200 nm. Fibrils are visible
on day 2 for Orb2A1-88 and elongate over a 21-day time period (full time course in Supporting Figure
5). Small, rod-like particles are visible on day 2 in the Orb2A1-88 sample with SUVs. These rod-like
particles are seen throughout the rest of the experiment and slightly elongate over time. They do not
develop into structures comparable to the fibrils seen for Orb2A1-88 without lipids. B) Time
dependence of the EPR amplitude of Orb2A1-88 MTSL labeled at residue 10 in the absence and
presence of SUVs (protein to lipid ratio of 1:100 using 2 POPS:1 POPC). C) EPR spectrum of Orb2A1-
88 labeled at the wild type C10 with the same lipid vesicles after 48 hours of incubation (top) and the
published EPR spectrum of Orb2A1-88 amyloid fibrils formed without lipids (13) (bottom).

29
We then monitored the EPR amplitude over time, which decays as the protein
aggregates (Figure 6b). Without lipids, Orb2A1-88 shows a typical amyloid aggregation
curve with a lag phase prior to aggregation and resultant decrease in amplitude.
Orb2A1-88 with lipid vesicles shows a rapid decrease in amplitude with no noticeable
lag phase, and by 20 hours begins to slowly increase in amplitude again. This increase
is not consistent with the long-term stability normally seen for amyloid fibrils. The
intensity EPR of Orb2A1-88 without lipids did not increase again even when checked
72h after the end of the kinetics measurements shown in Figure 6b. At 12 hours, the
EPR spectra of the Orb2A1-88 SUV mixture show the typical line shape of the lipid
bound state and we observed no spin exchange that is a hallmark of the amyloid fibril
spectrum at this residue (13) (Figure 6c). Combined, our EPR data suggests that the
structures seen in the lipid-containing EMs are not amyloid fibrils.
Discussion
Based on our CD and EPR experiments, Orb2A1-88 interacts with anionic lipid
vesicles. This interaction increases the amount of α-helicity inside the amphipathic N-
terminal region, but not in the Q/H-rich region. Furthermore, this interaction affects the
formation of amyloid fibrils as visualized by electron microscopy and confirmed with
EPR. The N-terminal region of Orb2A has previously been shown to be important for
aggregation and long-term memory (8, 12, 20). Orb2A has 8 amino acids at its N-
terminus that distinguish it from Orb2B (6). Full length Orb2A has been shown to
contain slightly more α-helical structure than Orb2B (12). This suggests that the N-
terminus of full-length Orb2A has a helix-forming propensity comparable to Orb2A1-88
and could similarly interact with lipid membranes.
Another important observation is that Orb2A1-88 interacts with vesicles
containing the anionic lipids POPS and POPG, but not with vesicles containing only the
neutral zwitterionic lipid POPC. There are three positive charges at the N-terminus,
namely Lys4, Lys20, and the N-terminal amine, which could facilitate binding through an
initial or sustained electrostatic interaction with the lipid head group. Similarly, αS binds
preferentially to negatively charged lipid membranes (32, 51). This binding has been
thought to be due to the presence of positively charged amino acids in the helix-forming

30
repeats found in αS. These repeats form Class A amphipathic helices, which are
characterized by basic amino acid residues that are positioned close to the
hydrophobic/hydrophilic interface of the helix which allow the helix to interact with the
negative charges in the lipid backbone and headgroup (52). The Lys4 and Lys20 in
Orb2A could act in a comparable way, allowing the positively charged amines to interact
with the negatively charged lipid headgroups, while the long acyl chain is inserted into
the hydrophobic fatty acid environment of the lipid membrane.
Orb2A1-88 was able to bind much more readily to the highly curved SUVs than
to LUVs. Both αS and IAPP have also been shown to interact preferentially with highly
curved vesicles (32, 51). This preference is thought to be due to packing defects that
are common between lipid headgroups in highly curved vesicles. Gaps that occur
between headgroups allow peptides to interact more easily with the hydrophobic acyl
chains of the lipids, which then promote helix formation. Therefore amphipathic helix
formation becomes thermodynamically more favorable in vesicles with higher curvature
(53).
In the presence of anionic membranes, IAPP becomes α-helical, and the
conversion from this α-helical structure to β-sheet only takes a few minutes (35, 54).
This increase in rate of fibril formation, compared to the absence of lipids, is seen for
both SUVs and LUVs, but is faster for SUVs (54). However, this only holds true up to a
certain charge density above which vesicles inhibit IAPP fibrilization (26). It is also
interesting to note that even though IAPP adopts a similar α-helical structure in micelles,
the α-helical structure is stable for several days, indicating lipid interaction can both
induce and inhibit fibril formation (54).
Orb2A1-88 formed amyloid fibrils over a three-week period, with small fibrils
visible after two days. However, when Orb2A1-88 was mixed with anionic lipid vesicles,
we did not observe the formation of mature amyloid fibrils. The smaller rod-like particles
we observed in Orb2A1-88 SUV mixtures are clearly distinct in size and shape from the
fibrils seen for Orb2A1-88 in the absence lipids. The EPR spectra of Orb2A1-88 did not
show any spin exchange, as a measure of amyloid formation, in the presence of lipids
but were very similar to the initial spectra even after several days of incubation. These
results indicate that the interaction of Orb2A1-88 with lipids, particularly anionic vesicles

31
of high curvature, can inhibit the formation of amyloid fibrils. Though we conclude that
the rod-like structure is not a canonical amyloid fibril, we currently don’t know exactly
what these particles are. Since they are not observed in EM images of SUVs alone,
they are likely a result of protein-lipid interaction. Possible compositions of these rod-like
structures could be lipid enriched protein particles, protein-lipid micellar or tubular
structures, or even just Orb2A1-88 oligomeric structures.
The aggregation of the human functional amyloid Pmel17 behaves similarly to
Orb2A1-88 in the presence of anionic lipid micelles (27). In the presence of negatively
charged lysolipid micelles, Pmel17 is prevented from forming amyloid fibrils. Pmel17
shows an increase in α-helicity when added to these micelles, and the region predicted
to form the α-helix is the same region that forms the β-sheet core of Pmel17 fibrils (27).
High concentrations of lysolipids are present in the melanosome, along with highly
curved membrane structures (37, 38). Thus it has been hypothesized that this inhibition
of fibril formation by particular membrane compositions is an important regulatory
mechanism in functional Pmel17 aggregation, and that the stabilized α-helix in the
presence of highly curved negative membranes inhibits the formation of a β-sheet in the
same area (27). The β-sheet amyloid core of Orb2A1-88 fibrils was recently shown to
be located within the first 20 amino acids, with no indication of β-sheet formation in the
Q/H rich region (13). If this region of the protein is involved in both α-helical lipid
interaction and amyloid formation, lipid interaction could prevent this region from
forming the β-sheets of the amyloid core (Figure 7). The focus of this study was to
show that Orb2A1-88 does bind to lipids, and to begin to understand the effects of this
Figure 7: Model of the possible regulatory
role of membranes for Orb2A amyloid
formation. Monomers that interact with lipid
membranes form an amphipathic helix at the N-
terminal region, which prevents β-sheet formation
and amyloidogenic aggregation. The Q-rich and
G-rich regions are represented in orange and
green respectively.

32
binding on amyloid formation. Vesicles of larger size, different amounts of charge, or
different hydrocarbon chain lengths may prove to have different effects on Orb2A fibril
formation than the conditions tested here. If the binding of Orb2A1-88 to the vesicle is
weaker, it could promote fibril formation as discussed above for other amyloid forming
proteins. More in-depth systematic studies of lipid compositions are necessary and are
currently underway to fully understand this mechanism.
Conclusions
We have shown that the N-terminus specific to Orb2A can bind to lipid
membranes. This binding occurs preferentially to curved membranes and requires the
membranes to be negatively charged. When binding to lipid membranes, the N-terminus
of Orb2A undergoes a transition from a dynamic, intrinsically disordered state into a less
dynamic α-helix. Finally, we have shown that this lipid membrane interaction can
prevent the N-terminus from forming amyloid fibrils.
We previously showed that the very region binding to lipid membranes is also
able to form in-register parallel β-sheet amyloid fibrils similar to disease-associated
amyloids (8, 12, 13). Considering that Orb2A is essential for long-term memory in
Drosophila and was hypothesized to act as a seed for fibril formation of the dominant
isoform Orb2B, we think that its membrane interaction might play a regulatory role. This
lipid-mediated regulation of Orb2 fibril formation might be important to suppress toxic
aggregation intermediates inside the cellular environment (12). Such lipid interaction
might be a common mechanism to suppress toxicity in functional amyloids. For
example, different lysolipid compositions in the intralumenal membrane vesicles were
hypothesized to be a major spaciotemporal regulatory mechanism of amyloid fibrils
composed of the Pmel17 protein that prevents the production of toxic oligomers (27).
Similarly, aggregated Orb2 is found enriched within the synapses of mushroom body
neurons (8), and it is possible that the highly curved synaptic vesicles and anionic
membrane lipids found at the synapse (55) could act in a regulatory capacity towards
the fibrilization of Orb2 in vivo.

33
Supporting Material

Supporting Figure 1: Orb2A1-88 co-purifies with lipids. Quantitative phosphate
determination of differently purified Orb2A1-88 samples with phospholipids extracted
using an acidified Bligh and Dyer protocol (1). Purified protein aliquots were mixed with
2:1 methanol to chloroform by volume, and acidified with 0.5% HCl, mixed and
incubated until separated. Then 1:1 chloroform to 100 mM HCl was added, vortexed,
and centrifuged for 5 min at 1000 rpm (A-4-44, Eppendorf, Hamburg, Germany) to
separate phases. The organic layer was removed and evaporated with N
2
. To
determine the total phosphorus (2), the residue was then redissolved in H
2
SO
4
and
heated above 200°C for 25 min. 30% H
2
O
2
was then added and samples were heated
for another 30 min. Samples were cooled and 3.9 ml water was added to each as well
as 0.5 ml 2.5% ammonium molybdate (VI) tetrahydrate and vortexed. Then 0.5 ml 10%
ascorbic acid was added and vortexed. Finally, the samples were heated to 100°C,
cooled, and their absorbance was measured at 820 nm on an Agilent (Santa Clara, CA)
Cary UV-Vis spectrometer. Phosphate standards (Sigma Aldrich, St. Louis, MO) that
were treated similarly to the samples were used to construct a standard curve by which
the concentration of phosphate in each sample was calculated. The Ni-NTA affinity
column with lipid wash was used to purify recombinant peptides for the experiments in
this paper, and showed negligible amounts of phosphate in the sample.

34

Supporting Figure 2: SUV (2 POPS:1 POPC) titration EPR data for all spin labeled
sites. EPR spectra of each MTSL labeled Orb2A1-88 cysteine mutant and wild type
C10 in the presence of increasing amounts of SUVs (2 POPS: 1 POPC) are shown. The
protein to lipid ratio of each spectrum is indicated.

35

Supporting Figure 3: Structural change in SUV titrations is not due to aggregation
of protein. For Orb2A1-88 MTSL labeled at residues 12 (G12C) and 6 (V6C) an EPR
spectrum was taken in the absence of SUVs before the lipid titration series was started
(Before). This sample was then saved and re-measured after the titration was
completed (After). Both residue 6 and 12 show no change in structure over the time of
the titration experiment.

Supporting Figure 4: The MTSL spin label does not perturb Orb2A1-88 structure
or lipid binding abilities. Orb2A1-88 was spin labeled with MTSL (R1) at residues 6,
18, and 34 with excess label removed via cation exchange followed by dialysis as
described in the Materials and Methods section. CD shows similar changes with the
addition of lipid vesicles containing 2 POPS:1 POPC comparted to those seen for non-
MTSL labeled Orb2A1-88 in Figures 2A and 3A

36

Supporting Figure 5: Full EM time course for Orb2A1-88 with and without anionic
SUVs. Transmission electron microscope images taken of Orb2A1-88 with and without
anionic SUVs (2 POPS:1 POPC) over a period of 21 days. The black bar represents
200 nm. Amyloid fibrils can be seen for Orb2A1-88 alone after two days. Similar fibrils
are not seen for Orb2A1-88 with anionic SUVs, though small, rod-like structures are
visible. Days 0, 2, 6 and 12 are the same as Figure 6 of the main text. Days 4, 14, and
21 are shown in addition here.

37
Supporting Text 1: Determining the fraction of bound protein from EPR spectra
The central line intensity of an EPR spectrum composed of an unbound and a bound
component can be described as a sum of two Laurentian absorption lines as follows:
𝑆= 𝑐 ∗
!
!
(!
!
!
!!
!
)
+ (1−𝑐)
!
!
(!
!
!
!!
!
)
(1)
where c is the the fraction of free protein, λ
a
and λ
b
are the half-widths of the bound and
the free protein, respectively, and H is the resonance offset assuming the line is
centered at 𝐻= 0. From this equation follows that the line intensity
𝑆
!"#
=
!
!
!
+
!!!
!
!
(2)
is a linear combination of the line intensities of both individual components (𝑆
!
=
𝜆
!
!!
; 𝑆
!
= 𝜆
!
!!
). The fraction c can, therefore, be determined from line intensity as
𝑐= 𝑆
!"#
−𝑆
!
/ (𝑆
!
−𝑆
!
) (3).
Similarly, we can determine c from the linewidth (i.e. the full width at half heights 𝛥𝐻) by
setting 𝑆=
!
!
𝑆
!"#
(4)
and solving for c
𝑐=
!
!
!!
!
! ! !
!
!
!!
!
! ! !
!
!
!
!
!!
!
!!
!
!!"!
!
!
!
!
!
!!!!
!
!
!
!!
!
!
(5).
The problem with the latter method is that c is not a linear function𝛥𝐻. This becomes
especially problematic when there is a large difference between 𝜆
!
and 𝜆
!
resulting in a
𝛥𝐻 that does not change substantially over large parts of the binding curve. To illustrate
this behavior, we assumed the half-widths of the individual components to be 𝜆
!
= 5G
and 𝜆
!
= 1G. Where the line intensity changes linearly with c as discussed above, the
linewidth 𝛥𝐻is dominated by the sharper component 𝜆
!
and only changes significantly
when the bound state becomes dominant (Supporting Figure 6). Therefore, we chose to
use the EPR line intensity to determine the protein-membrane dissociation constant.

38

Supporting Figure 6: Determining the fraction of bound protein c from EPR line
intensity and linewidth. The half-widths of the individual components were set to
𝜆
!
= 5G and 𝜆
!
= 1G and the line intensity S
max
was plotted using equation 2. The
linewidth 𝛥𝐻was determined by solving equation 4 for H and calculating the difference
of the two real solutions versus c using the program Mathematica (Wolfram Research,
Inc.,Champaign, IL).

Supporting Table 1: Values and uncertainties for 𝞓 inverse linewidth plotted in
Figure 4c. The values and uncertainties for the difference in inverse linewidth between
Orb2A1-88 in the absence of lipids and at a protein to lipid ratio of 1:100 are reported
for each residue number.
Residue Number
𝞓 Inverse LW
(1/G)
Uncertainty (1/G)
6 0.2012 0.00196
10 0.2273 0.00227
12 0.1914 0.00148
18 0.1618 0.00122
23 0.1113 0.00350
34 0.0466 0.00238
84 0.0312 0.00226

39
Supporting Table 2: Values and uncertainties for inverse linewidth of Orb2A1-88
spin labeled at residue 6 with SUVs. The values and uncertainties are listed for each
inverse linewidth measured at different protein to lipid ratios for SUVs.
Protein:Lipid
Inverse
LW
(1/G)
Uncertainty
(1/G)

1:0 0.4132 0.00022
1:25 0.2785 0.00106
1:50 0.2352 0.00177
1:75 0.2173 0.00045
1:100 0.1949 0.00087
1:200 0.2066 0.00027
1:300 0.2197 0.00046

Author Contributions

M.A.S. prepared samples, designed and performed experiments, wrote paper. S.A.C.
cloned mutants and prepared samples. T.H.B. prepared samples. A.B.S. designed
experiments, wrote paper.
Acknowledgements
The authors would like to thank Dr. Kausik Si for providing the Orb2A plasmid, as well
as Dr. Ralf Langen and his lab members, in particular Dr. Alan Okada and Dr. J. Mario
Isas, for valuable discussions and technical assistance.
Support from National Institutes of Health NIGMS Award R01GM110521 (A.B.S);
NINDS Award R01NS084345 (A.B.S. and S.A.C.), the Whitehall Foundation (A.B.S),
and the University of Southern California (A.B.S. and M.A.S.) is gratefully
acknowledged.

40
Chapter 3: Calmodulin binds the CPEB homolog Orb2A at the
aggregation-prone N-terminal region.*

Maria A. Soria
1
, Silvia A. Cervantes
1
, and Ansgar B. Siemer
1

1
Department of Physiology and Biophysics, Zilkha Neurogenetic Institute, Keck School
of Medicine, University of Southern California, 1501 San Pablo Street, Los Angeles,
California 90033, United States

*In preparation for publication.

Introduction

Cytoplasmic polyadenylation element-binding (CPEB) proteins are important
mRNA translational regulators. In Drosophila melanogaster, the CPEB homolog Orb2
has been shown to be important for long-term memory and regulating mRNA translation
at the synapse (6, 8, 31). There are two isoforms of Orb2, known as Orb2A and Orb2B.
Both are present in the synapse, though Orb2B is found abundantly throughout the
cytoplasm whereas Orb2A concentrations are low (8). Both Orb2A and Orb2B can form
amyloid-like aggregates in vitro, and are found together in aggregates in vivo. This
aggregation is necessary for long-term memory (8). Much of the work on Orb2
aggregation in memory suggests that Orb2A may be the key to regulating overall Orb2
aggregation for long-term memory (8, 10, 31). Regulation of amyloid aggregation would
be crucial to preventing toxic intermediates and end products that are commonly
associated with neurodegenerative diseases like Alzheimer’s disease and Parkinson’s
disease (12). Currently, the regulatory mechanisms for Orb2A aggregation are not
known.

Orb2A and Orb2B differ only in their N-terminal residues. Orb2A has a truncated
N-terminus, with only the first 8 amino acids being unique to Orb2A. Orb2B has an
extended serine-rich N-terminus with over 150 amino acids preceding the first common
residue with Orb2A (Figure 1A). Both Orb2A and Orb2B then have a
glutamine/histidine-rich region followed by a glycine-rich region, and finally two RRM
RNA binding domains and a zinc finger at the C-terminal end. The N-terminus of Orb2A
has so far proven to be very interesting. In a random mutagenesis experiment by
Majumdar et al., point mutations in the first eight amino acids were over-represented as

41
mutations that inhibited aggregation (8). Amyloids formed by the first 88 amino acids of
Orb2A have their β-sheet core within the first 20 amino acids of Orb2A. In fact, these
first 20 amino acids are able to form amyloid on their own (13). The amphipathic nature
of these first 20 amino acids (Figure 1A) led originally to our discovery of the lipid
binding capabilities of Orb2A. We showed that Orb2A forms an amphipathic helix in the
presence of anionic lipid vesicles and binds to these vesicles (16). This binding of
Orb2A inhibited amyloid fiber formation.

One common protein that interacts with amphipathic helices is the calcium-
sensing protein calmodulin (CaM) (28). Among its many functions in a large number of
different cell types, CaM is an important signal integration protein in neurons for long-
term memory (LTM) (29). CaM is present in very high concentrations in the CNS, up to
10-100 uM (56), and has been shown to directly regulate several proteins involved in
long-term potentiation (LTP). These include adenylyl cyclases AC1 and AC8 (57, 58),
calcineurin, and CaM kinases (CaMK) II and IV (59–63). Depending on the level of Ca
2+

present, CaM can activate proteins specifically for LTP or long-term depression
(LTD)(64), both of which are important in LTM. CaM interacts with sequences that are

Figure 1: Structural components of Orb2A with
typical CaM binding mechanism. A) Domain
layout of Orb2A and Orb2B, including the
amphipathic helix at the N-terminus of Orb2A. Gray
is the extended N-terminus of Orb2B, purple is the
amphipathic region, the green represents the
glutamine/histidine rich region, yellow is the glycine
rich region, and orange is the RNA binding
domains and zinc finger. B) Sequence of the N-
terminal 88 amino acids of Orb2A used throughout
this study. The colors correspond to the domains in
A. C) Mechanism by which Ca2+/CaM binds to
amphipathic helices. Blue circles represent Ca2+,
the orange helix is the linker domain of CaM, the
purple helix is a hypothetical CaM binding partner.

42
both amphipathic and positively charged. While there are many modes of binding with
CaM depending on the target (reviewed in (65)) the most observed mode relies on the
binding of Ca
2+
to each of the four EF hand domains present in CaM. These four EF
hands are separated, two to each side, by an unstructured linker, which becomes α-
helical when Ca
2+
binds to the EF hands (Figure 1C). This creates hydrophobic binding
pockets with which many of CaM’s target proteins interact.

In hippocampal dendrites, the CaM/Ca
2+
-regulated kinase CaMKII is responsible
for activating mouse CPEB via phosphorylation, which in turn activates mRNA
translation (30). Considering the amphipathic and positively charged nature of Orb2A’s
N-terminus, its involvement in LTM and the high concentration of CaM in the neuron, we
decided to investigate whether CaM might directly bind to Orb2A’s N-terminus. Here we
show that the N-terminal amphipathic sequence of Orb2A does bind to CaM in a
calcium dependent manner and that this binding affects aggregation of Orb2A into
amyloid fibrils.

Materials and Methods

Protein expression and purification

All protein vectors were cloned into pET28b vectors as described previously (13).
Wild-type Orb2A88 (Figure 1B) and cysteine mutants V6C, L18C, and G84C were
expressed in E. coli Rosetta™ 2 (DE3) cells (EMD Millipore, expressed using BL21
(DE3) cell cultures. For all protein constructs, the appropriate plasmid was transformed
into CaCl
2
chemically competent cells, and one colony was used to inoculate a 25 mL
culture in lysogeny broth (LB) Miller medium with the appropriate antibiotics at 37°C.
After approximately 4 h, this culture was diluted into 1 L LB Miller medium with the
appropriate antibiotics and was grown until OD
600
=0.6. The culture was then induced
with isopropyl b-D-1-thiogalactopyranoside (IPTG), and protein was expressed at 25°C
for approximately 16 h. Cell cultures were spun down at 4000 rpm in a Sorvall SLC-
6000 rotor (Thermo Fisher Scientific, Waltham, MA) for 20 min at 4°C, and the pellets
were stored at -80°C.

43
In order to purify protein, cell pellets were thawed and suspended in denaturing
buffer (8 M urea, 10 mM Tris, 100 mM NaH
2
PO
4
, and 0.05% β-mercaptoethanol (pH
8.0)). This was then lysed on ice using a Q125 ultrasonic homogenizer (QSonica,
Newton, CT). The cell lysate was centrifuged at 20,000 rpm for 20 min using a Sorvall
SS-34 rotor. The supernatant was collected and poured onto a Ni-NTA column
equilibrated with denaturing buffer. The cell lysate supernatant was then incubated with
gentle shaking for ~1 h. For samples meant for EPR studies, the incubation was
skipped. The flowthrough was collected and the column was washed with the following
series of solutions: 1) denaturing buffer containing 0.5% Triton-X 100, 2) denaturing
buffer containing 500 mM NaCl, 3) denaturing buffer at pH 6.75, 4) renaturing buffer
(200 mM NaCl, 50 mM NaH
2
PO
4
, 10% glycerol, and 0.05% (v/v) β-mercaptoethanol (pH
8.0)) and 5) renaturing buffer containing 20 mM imidazole. Elution of the protein was
achieved using renaturing buffer with 250 mM imidazole. Samples intended for EPR
experiments included an extra step: 6) renaturing buffer (pH 7.4) with no β-
mercaptoethanol. These proteins were also eluted with 250 mM imidazole but at pH 7.4
with no β-mercaptoethanol.

Electron paramagnetic resonance studies

Protein aliquots were thawed on ice, and 2.5 µl of 40 mg/ml S-(1-oxyl-2,2,5,5-
tetramethyl-2,5-dihydro-1H-pyr-rol-3-yl)methyl methanesulfonothioate (MTSL) spin label
(Toronto Research Chemicals, North York, Ontario, Canada) was added (in great
excess of the protein). This was allowed to incubate at room temperature for 1 h. Then
the protein was diluted 15x with water and added to a cation exchange column (S
Ceramic HyperD F, Pall Life Sciences, Port Washington, NY), which had been
equilibrated with dH
2
O. The flow-through was collected and the column was washed
with 7 ml dH
2
O. The protein was then eluted with 2 ml of 8 M guanidine-HCl. This was
then dialyzed 3x against 1L buffer with 20mM HEPES, 100 mM NaCl, with or without 10
mM CaCl
2
, depending on if the experiment required Ca
2+
free conditions or not. Protein
aggregates were spun out of solution and the supernatant concentration was taken
using UV at 280 nm. Absorbance signal due to scattering was mitigated by subtracting a
fitted exponential function out of the spectrum. Finally, the desired amount of CaM,

44
Ca
2+
, EDTA or buffer was added to the protein depending on the experiment, and the
sample was loaded into a borosilicate capillary tube (0.6 mm inner diameter, 0.84 mm
outer diameter; Vitro-Com, Mt. Lakes, NJ).

Continuous-wave EPR spectra were collected using a Bruker X-band EMX
spectrometer (Bruker Biospin, Billerica, MA) at room temperature. Each spectrum is the
accumulation of 15 scans in a high sensitivity cavity at a scan width of 150 gauss. The
microwave power was 12.6 mW. Amplitude was calculated as the distance between the
highest and lowest points of the spectrum. This was then normalized based on the
spectrum amplitude for the protein without any CaM added. For kinetics curves,
readings were taking of the specified sample over the indicated period of time.
Amplitude was normalized to the largest amplitude of each data set, and then plotted
together for comparison.

Thioflavin-T fluorescence assays

Orb2A88 protein aliquots were thawed on ice and exchanged into either 20 mM
HEPES, 100 mM NaCl with or without 10mM CaCl
2
depending on the necessity of the
presence or absence of Ca
2+
in the solution. In a 96-well plate, Orb2A88 was mixed with
either buffer or a 1:1 molar ratio of CaM to maintain the same final concentration of 10
µM for Orb2A88. A buffer blank and CaM alone blank were also run. ThT was added to
each well for a final concentration of 50uM ThT. This was then read at room
temperature with gentle periodic mixing by an Eppendorf (Hamburg, Germany) AF2200
plate reader over the indicated period of time using an excitation wavelength of 440 nm
and emission wavelength of 480 nm. Error bars represent the standard deviation of
three biological replicates.

Electron microscopy

Orb2A88 protein aliquots were buffer exchanged as for the Thioflavin-T assays and
mixed with the appropriate amounts of Ca
2+
, EDTA, and CaM to match the ratios of the
corresponding Thioflavin-T assays. At least 0.02% sodium azide was added to prevent
microbial growth. Samples were incubated at room temperature with slight agitation. To
image, copper formvar grids
(Electron Microscopy Sciences, Hat- field, PA) were

45
incubated with drops of protein sample for 5 min, and then were incubated with uranyl
acetate (1%) for another 5 min. Grids were than washed with two drops of uranyl
acetate and one drop of water and allowed to dry. Finally, grids were imaged using a
JEOL JEM-1400 EM (Tokyo, Japan).

Results

Orb2A88 binds to Calmodulin in a calcium-dependent manner

We first wanted to investigate whether Orb2A could bind to CaM. To do this, we
used electron paramagnetic resonance (EPR) with site-directed spin labeling. The first
88 amino acids of Orb2A (Orb2A88) contain several positive charges and an
amphipathic sequence at the very N-terminus (Figure 1A and B). We decided to use
this construct to test whether it could bind to CaM, as it contains the area we
hypothesized would bind to CaM, and also contains only one natural cysteine. When
bound to MTSL, we can use this wild-type cysteine as a reporter since it is in the region
of interest and leads to minimal perturbation of the protein sequence.
After labeling with MTSL, EPR spectra of Orb2A88 were taken in varying
conditions. EPR line broadening is an indicator of rigidity at the site of the spin label,
and so we used this as an indicator of CaM binding at position 10. Only in the presence
of both CaM and Ca
2+
does the line broaden for labeled position 10 (Figure 2A).
Neither CaM in the presence of EDTA nor Ca
2+
alone caused line broadening in the
spectra.

Along with line broadening, a decrease in amplitude indicates rigidity of the spin
label in an EPR spectrum. We used change in amplitude to indicate the fraction of
Orb2A88 binding to CaM and constructed a binding curve. This is possible because,
unlike line broadening, the amplitude changes linearly with binding (16). We again
labeled the wild type cysteine 10 with MTSL, and measured EPR spectra at different
Orb2A88 to CaM ratios (Figure 2B). From these data, we were also able to calculate a
binding constant of 2.1±0.3 μM.

46

Orb2A88 binds to CaM within its N-terminal amphipathic region

While the natural cysteine used for labeling in the above experiments was within
the N-terminal amphipathic region, we wanted to get a more specific idea of where on
the peptide CaM was binding. We used cysteine mutants along the length of Orb2A88
for MTSL labeling as previously described (13) (Figure 3A). We then performed EPR in
the presence of a 10mM Ca
2+
both without CaM present and with a 1:1 molar ratio of
CaM for each construct and the relative change in amplitude at each site with the
addition of CaM was plotted (Figure 3B). We saw the largest decrease in amplitude at
the N-terminal positions, which indicates that CaM is binding in the N-terminal
amphipathic region. The decrease in amplitude quickly becomes smaller towards the C-
terminus of the protein indicating that neither the glutamine/histidine-rich nor the
glycine-rich region are involved in binding CaM.

Figure 2: Orb2A88 binds CaM in a calcium
dependent manner. A) Orb2A88 labeled at
position 10 shows no change in linewidth
with the addition of CaM with EDTA or Ca
2+
,
but linewidth increases with the addition of
CaM with Ca
2+
. CaM is present at a 2:1 molar
ratio to Orb2A88, Ca
2+
or EDTA is present at
1 mM. B) Amplitude of spectra from 6 µM
Orb2A88 labeled at position 10 with
increasing concentrations of CaM. Ca
2+
was
present in excess at 10 mM.

47

CaM binding inhibits aggregation of Orb2A88

Previously, we showed that when Orb2A88 aggregates, the β-sheet rich fibril
core is located within the N-terminal amphipathic region (13). Considering that CaM
binds to the same location on Orb2A88, we wanted to know whether CaM binding
influenced Orb2A88 aggregation. We first used EPR to track changes in amplitude over
time in the N-terminal region using a label at the wild type cysteine 10. As the protein
aggregates, we expect to see a decrease in amplitude, as we have observed before
(16). Even though Orb2A88 labeled at position 10 shows an initially lower amplitude
when Ca
2+
/CaM is added, we expect the amplitude to continue to decrease over time as
free monomer is bound into amyloid fibrils. However, the amplitude of the spectra of
Orb2A88 with CaM did not decrease over time, indicating that Orb2A88 in the presence
of CaM was not aggregating (Figure 4A).

We then wondered if perhaps Orb2A88 was aggregating but using a different
region as its β-sheet core. Many proteins form amyloid using a glutamine repeat or
glutamine-rich region as the β-sheet core of the fibril. In order to know whether or not
Orb2A88 was aggregating in the glutamine/histidine-rich region, we again used EPR to
track change in amplitude over time, but this time the label was at position 34 (Figure
Figure 3: Orb2A binds CaM using the N-
terminal amphipathic region.
A) Location of each cysteine point mutation
along the length of Orb2A88. Purple
indicates the amphipathic region, green the
glutamine/histidine rich region, and yellow
the glycine rich region. B) The amplitude of
the EPR spectrum at each labeled position
after the addition of CaM as compared to
the amplitude of each labeled position
without CaM present. CaM was added in a
1:1 molar ratio to Orb2A88. All samples
were measured in the presence of 10mM
Ca
2+
.

48
4B). We saw a slight decrease in amplitude over time for Orb2A88 without CaM, but we
saw no decrease in amplitude over time in the presence of CaM. This confirms our
initial conclusion that without CaM, Orb2A88 aggregates in the N-terminal region, but in
the presence of CaM, Orb2A88 does not aggregate.

The aggregation inhibited by CaM is amyloid in nature

We next wanted to confirm that CaM inhibited amyloid aggregation, so we decided to
use Thioflavin-T (ThT) fluorescence, which has been commonly used as an indicator of
amyloid formation (66). It is thought that ThT binds to the β-sheet core of an amyloid
fibril, which stabilizes the rotating bond and causes fluorescence (67, 68). Because of
this, the change in ThT fluorescence can be used to measure amyloid fibril forming
kinetics. We tracked ThT signal over time for both Orb2A88 alone, and Orb2A88 with
CaM and subtracted baseline fluorescence values for both (Figure 5A). Orb2A88 alone
gave a characteristic amyloid aggregation curve, with a lag phase followed by
aggregation, which finally reaches equilibrium. However, for samples where CaM was
present, we do not see any net change in ThT signal. In fact, the apparent ThT
fluorescence values are negative when a baseline of the same concentration of CaM
alone was subtracted. We believe this is because of the ability of ThT to bind
hydrophobic pockets, which form in Ca
2+
bound CaM. When Orb2A88 is present, it
binds to these hydrophobic pockets and removes the ThT, resulting in a negative ThT
reading when the CaM baseline is subtracted.

Figure 4: The N-terminus of
Orb2A does not aggregate in
the presence of CaM.
A) Amplitude of Orb2A88 labeled
at position 10 over time. B)
Amplitude of Orb2A88 labeled at
position 34 over time. For both A
and B, CaM was present at 2:1
ratio to Orb2A88, with 10mM
Ca
2+
present in all samples. Data
was normalized to the first point
of each experiment.

49
To confirm these ThT fluorescence results, we used electron microscopy (EM).
We saw that while the ThT positive Orb2A88 sample forms amyloid fibrils visible by EM,
the sample containing Orb2A88 with CaM did not form any structures resembling
amyloid fibrils (Figure 5B). This again supports the conclusion that CaM inhibits
amyloid aggregation of Orb2A88.

Discussion

Here, we show that the first 88 amino acids of Orb2, which forms amyloid-like
aggregates that are involved in long-term memory, binds to CaM using its amphipathic
N-terminal region. We previously showed that this region is able to form an amphipathic
helix that is capable of binding anionic lipid membranes, and proposed that Orb2A88
preferred anionic lipid vesicles because of the lysines present at position 4 and 20, as
Figure 5: CaM inhibits amyloid
aggregation of Orb2A 88. A)
ThT fluorescence over time of
samples containing just Orb2A88
and samples of Orb2A88 with
CaM at a 1:1 molar ratio. B) EMs
of representative samples from
part A, taken after 9 days of
aggregation. Black bars
represent 1 um, white bars
represent 200 nm.

50
well as the positively charged N-terminus (16). Positively charged amphipathic helices
have also been shown to be preferential binding partners for CaM (69). Despite the fact
that CaM binding domain prediction programs did not identify Orb2A as a potential CaM
binding partner (data not shown), our EPR studies and the effects of CaM on Orb2A88
aggregation show that CaM does indeed interact with Orb2A88, and this interaction
occurs at the N-terminus.

Our EPR, ThT and EM studies show that the presence of CaM inhibits Orb2A88
amyloid aggregation. This was not surprising because the N-terminal amphipathic
region that interacts with CaM is also involved in forming the β-sheet rich amyloid core
of the Orb2A88 amyloid fibril (13). Previously we showed that the binding of the N-
terminal amphipathic region to lipid membranes also inhibited amyloid aggregation (16).
This binding to lipid membranes was hypothesized to stabilize the α-‐helical
conformation while not allowing the β-sheet to form. Orb2A88 binding to CaM could
create a similar situation where the N-terminus is stabilized in a helical conformation
due to CaM binding and is then unable to interact with other Orb2A88 monomers or
fibers to form a β-sheet (Figure 6).

Calcium signaling is very important for LTM, and CaM is an integral mediator of
this calcium signaling inside the neuron (29, 57–63). Orb2, a CPEB homolog in fruit
flies, forms amyloid-like aggregates necessary for LTM (8). In mouse hippocampal
neurons, CPEB can be activated by CaM through the activation of CaMKII and
subsequent phosphorylation (30). This phosphorylation occurs on a regulatory site,
threonine 171 (30). A homologous site exists for CPEB in Xenopus oocytes at serine
174 (70). There are a few potential homologous sites in Orb2A, including Thr 169 or Ser
Figure 6: Model for the inhibition of
Orb2A88 aggregation by CaM. Addition of
CaM with Ca
2+
sequesters the N-terminal
amphipathic region of Orb2A88, inhibiting it
from forming the β-sheet conformation
necessary to form amyloid fibrils.

51
165. Previously, Orb2A was shown to have multiple phosphorylation sites, with more
than one kinase contributing to these phosphorylations (20). It is possible that CaMKII
could regulate Orb2A similar to the regulation of CPEB in mouse hippocampal neurons.
It is still unknown how the direct binding of CaM to Orb2A shown in this study
affects learning and memory in vivo. Orb2A has been shown to be important for
regulating and initiating amyloid fibrils necessary for LTM (8, 10, 31). CaM binding could
potentially play a regulatory role at the very N-terminus along with phosphorylation
occurring further down the chain. It is interesting that CaM inhibits Orb2A aggregation,
while phosphorylation has been shown to be activating (20, 30). This could have
something to do with the fact that Orb2 aggregation is only associated with LTP
maintenance and not initiation. Perhaps as the balance between these two shifts as
CaMKII switches from CaM regulated to autoregulated, a switch occurs that signals for
the shifting of the cell from initiation of LTP to maintenance of LTP.
Recently, it was hypothesized that the β-sheet core found within the N-terminal
amphipathic region is potentially transient or regulatory for a more permanent core that
can be found when Orb2A and Orb2B fibrilize together (13). In this case, Ca
2+
/CaM
could be important for inhibiting the formation of the Orb2A N-terminal β-sheet
formation, and allowing other regions to interact and fold into β-sheet cores. More
studies are needed in order to fully understand the molecular processes involved in
LTM in general, and particularly more studies are needed on the regulation of Orb2A
fibrilization in cells and in vivo.

52
Chapter 4: Conclusions and Outlook

By Maria Soria

Throughout the course of this dissertation, we have discussed the importance of
the N-terminal region of Orb2A in regulating the aggregation of Orb2 necessary for long-
term memory (LTM). Here I showed that this amphipathic N-terminus is capable of
binding to both anionic lipid membranes as well as the calcium-sensing protein
calmodulin (CaM). In both cases, the unique 8 N-terminal amino acids that set Orb2A
apart from Orb2B are involved in these interactions.
Previous studies showed the unique function of Orb2A as initiator and regulator
of Orb2 aggregation and LTM (8, 12, 20, 31). The N-terminus of Orb2A is both
truncated and amphipathic, compared to Orb2B’s extended, serine-rich N-terminal
region. Orb2A does not need its RRMs to function in LTM, but Orb2B does need the
RRMs (31). What about Orb2A’s N-terminus makes it able to perform its function in
LTM? Why do the few amino acids present in Orb2A’s N-terminus add an amphipathic
nature to that region of the protein? These are questions that need to be answered in
order to fully understand Orb2A’s role in regulating aggregation as well as in LTM. We
have put forth two examples here of ways in which Orb2A may use its N-terminal
amphipathic region, though we do not yet have solid answers for the questions posed.
We have, however, learned a few things about the possibilities available for the N-
terminal region of Orb2A. From here, I believe it is possible to continue the journey of
discovering the purpose of Orb2A’s N-terminus in order be able to fully understand the
role of Orb2A as well as the process of Orb2’s amyloid-like aggregation.

In both cases presented, the amphipathic N-terminal amino acids were shown to
interact with another macromolecule, which resulted in the N-terminus unable to form
the β-sheets necessary for amyloid-like aggregation. This makes sense because we
showed previously that those same N-terminal amino acids were found in the β-sheet
rich amyloid core of Orb2A88 fibrils (13). Circular dichroism also showed that full-length
Orb2A was slightly more helical in structure than Orb2B (12), and we showed that the
N-terminus of Orb2A adopts a helical structure when bound to lipids. It is possible that
stabilizing a helical structure in the N-terminal amino acids is one form of aggregation

53
control in vivo. We have shown this to be a possibility for lipid binding or CaM binding,
but there are probably a many ways this could happen in vivo. Perhaps there are other
chaperone proteins or specific lipids that help stabilize Orb2A in a monomeric form. One
protein that was shown to stabilize monomeric Orb2A was the protein Tob (20).
However, not much is known about where Tob interacts with Orb2A or what the
structural implications of this interaction are. More work is needed to investigate Tob as
well as other interaction partners of Orb2A in order to understand from a structural
perspective how they can control Orb2A aggregation.

Other than the previous work studying Tob and the subsequent phosphorylation
of Orb2A, there is not much in vivo work to investigate the interactions of Orb2A in cells
or fruit flies. In Appendix C of this dissertation, we show some preliminary studies of the
effects of Ca
2+
on Orb2A puncta formation in Drosophila S2 cells. However, more work
is needed to really understand the relationship of Orb2A to calcium signaling in the cell.
While we know that Orb2A is capable of interacting with Ca
2+
bound CaM, we do not
know yet if this interaction occurs in vivo, and if it does, what circumstances lead to
binding or unbinding. Where in the Ca
2+
signaling pathway would Orb2A act? Studies of
protein-protein interaction such as yeast two-hybrid assays and FRET may be able to
help start answering this question, as well as over expression and knockout assays of
calcium signaling proteins with subsequent analysis of Orb2A quantities and puncta
formation. Especially since it is known that Orb2A is phosphorylated by more than one
kinase (20), these assays might help identify those kinases.

Is it possible for both lipid interaction and CaM interactions to both occur naturally
in vivo and have a purpose in Orb2A’s LTM function? Without Ca
2+
stimulation, CaM is
sequestered to the cell membrane by a complex of proteins including neuromodulin and
neurogranin (71, 72). Upon Ca
2+
binding, CaM is released. Orb2A may also be
sequestered in the cell membrane under normal conditions, as it was found mainly in
the membrane fraction of the neuron (8). According to the binding constants we
determined, Orb2A88 has a much higher affinity for Ca
2+
bound CaM than for lipid
membranes. Upon CaM’s release into the cytosol, CaM could bind Orb2A and also
bring it into the cytosol as well for the next steps in the signaling cascade that eventually
lead to aggregation and stabilization of long-term potentiation.

54
The importance of the N-terminal region of Orb2A has been apparent from very
early on in the studies of Orb2 and LTM, and yet we still do not have a clear idea of the
purpose of this unique N-terminus nor what pathways lead to its regulation. We also do
not have an atomic resolution idea of how this regulation actually occurs. More in vivo
and in vitro studies are needed to identify how this process works. As discussed in the
introduction, our understanding of these processes will continue to contribute to our
understanding how other amyloids are related to disease. We are keenly interested in
comparing regulation, kinetics and structural conformations of Orb2 with other disease
related amyloids. How could we maintain an environment or change an environment to
better control amyloid formation? Another interesting question is why amyloid formation
is particularly associated with neurological dysfunction. Is it possible, if the engram of
memory is an amyloid fibril, that formation of other amyloids can disrupt this process?
How often do Orb2 aggregates and other protein aggregates mix? Are these aggregate
mixtures the same in healthy brains and diseased brains?

Finally, studying the Orb2A N-terminus will fundamentally expand what we know
about how memory is formed and maintained in healthy cells for up to a lifetime. Before
CPEB was shown to form amyloid fibrils associated with LTM, it was difficult to conceive
of a synapse specific structure that could last as long as a human lifetime inside of a
cell. However, this novel discovery has pushed the study of memory to a new level and
has enabled continuing innovative work on signaling, structure and molecular biology
involved with LTM. This body of knowledge will only continue to grow as we explore the
last frontier of human biology: the brain.

55
Appendix A: Effects of Calmodulin on Huntingtin and α-
Synculein

By Maria A. Soria
1
, J. Mario Isas
2
, Jobin Varkey
2
, and Ansgar B. Siemer
1

1
Department of Physiology and Biophysics, Zilkha Neurogenetic Institute, Keck School
of Medicine, University of Southern California, 1501 San Pablo Street, Los Angeles,
California 90033, United States;
2
Department of Biochemistry and Molecular Medicine,
Zilkha Neurogenetic Institute, Keck School of Medicine, University of Southern
California, 1501 San Pablo Street, Los Angeles, California 90033, United States

Rationale

In chapter 3 of this dissertation, we show that the calcium-binding protein
calmodulin (CaM) inhibits the ThT positive aggregation of Orb2A88. We were interested
if the same effect could be seen on other amyloid-forming proteins. We first tested
huntingtin exon 1 with a Q-repeat length of 46 (Htt Q46). This has been shown to
quickly form ThT positive amyloid fibers by various studies in the Langen lab (73, 74).
We also tried α-synuclein which has also been shown to form ThT positive aggregates
(24)

Methods

Htt Q46 and its seeds were a gift from J. Mario Isas in the Langen lab. α-
synuclein was a gift from Jobin Varkey of the Langen lab.

For Htt Q46 aggregation, Htt Q46 was diluted to a final concentration of 20 uM
and 1% seeds and EKMax were added to initiate fibril formation as previously described
(73). This was combined with the indicated molar ratio of CaM. Fluorescence was
monitored for the indicated amount of time on an Eppendorf (Hamburg, Germany) AF-
2200 plate reader on a 96 well plate with intermittent shaking and a final concentration
of 50 uM ThT and 10 mM CaCl
2
. Excitation was 440 nm and emission was 480 nm.
Because we needed to add seeds to Htt Q46 and these contain β-sheets which should
bind to ThT, blanks including seeds alone were run to subtract out any effect of seeds

56
on overall ThT level. Orb2A88 alone and with CaM was also run with the initial Htt Q46
and CaM experiment on the same plate with the same CaM stock to as a positive
control. Htt Q46 was present in 20 mM Tris, 300 mM NaCl, pH 7.4 buffer and Htt Q46
seeds were present in 0.1% trifluoroacetic acid, while Orb2A88 and CaM were present
in 20 mM HEPES, 100 mM NaCl, pH 7.4. HEPES buffer and Htt Q46 were mixed in the
same volume ratio that it would be as when mixed with CaM to identify any effects the
addition of HEPES would have on the aggregation of Htt Q46. We saw no noticeable
effect on the aggregation of Htt Q46 when HEPES buffer was mixed with it. For the Htt
Q46 disaggregation experiment, 20 uM Htt Q46 was aggregated in the presence of 1%
seeds as previously described. After 20 hours, a 1:1 molar ratio of CaM was added to
three replicate wells, the same volume of buffer was added to three other replicate
wells, and three other wells were left alone with nothing added.

For the experiment with α-synuclein, a final concentration of 75 uM α-synuclein
was combined with the 75 uM CaM or the equivalent volume of buffer. This was then
allowed to aggregate for the indicated amount of time and with a final concentration of
150 uM ThT. This was done at room temperature, and not at 37°C as was the Langen
lab protocol.

Results and Discussion

We were successfully able to use ThT to track aggregation of Htt Q46 using
seeds as previously described (73). In the presence of a 1:1.75 ratio of Htt Q46: CaM
we did not see any net increase in ThT over the same time period (Figure 1A). At the
same time, using the same stock of CaM, Orb2A88 and Orb2A88 with CaM were
observed with ThT, and the expected results seen in Chapter 3 of this dissertation, were
observed (Figure 1B). This was used as a control to be certain that the CaM we were
using in the Htt Q46 experiment was functioning properly and thus we were able to
determine that it was indeed functioning in the way we had observed it before. From
these experiments we concluded that Htt Q46 fibril formation was inhibited by the
presence of CaM. However, as we noted in our previous studies (see Chapter 3), these
results need to be confirmed using another method, as ThT with CaM can be hard to
interpret due to the binding of ThT to CaM without the presence of amyloid.

57
Recently, Htt Q46 aggregation was shown to begin with a tetrameric species
(74). If this is true, then CaM should inhibit Htt Q46 aggregation down to a 0.25 molar
ratio. We tested smaller molar ratios of CaM to Htt Q46 and saw that both a 1:1 and
1:0.5 ratios of Htt Q46 to CaM were successful in inhibiting Htt Q46 aggregation, and
this was seen in a concentration dependent manner (Figure 1C). However, the 1:0.1
ratio of Htt Q46 to CaM did not inhibit aggregation in any way. This suggests that CaM
inhibition of Htt Q46 could be occurring before the tetrameric intermediate forms. It is
possible that Htt Q46 could be interacting with CaM through its N-terminal helix, similar
to Orb2A88. It is possible that this interaction sterically inhibits Htt Q46 monomers from
interacting with each other, or that something about the interaction inhibits the
conversion of the polyQ domain into a β-sheet structure.

Figure 1: CaM inhibits Htt Q46 aggregation but does not disaggregate pre-formed
fibrils. A) 20 uM Htt in red and with 1.75 molar ratio CaM in blue. B) Orb2A88 aggregated with
(blue) and without (red) the same CaM from part A, as a CaM functionality control. C) Htt Q46
aggregated with the indicated molar ratios of CaM (1 Htt Q46 per 0, 0.1, 0.5 and 1 CaM). D)
Htt aggregated without the presence of CaM, but at hour 20, indicated by the black arrow,
either 1:1 molar ratio of CaM, the same volume of buffer or nothing was added.

58

Finally, we also attempted to disaggregate Htt Q46 with CaM using a 1:1 ratio of
the two proteins. Htt Q46 was allowed to aggregate with seeds similarly to above. After
20 hours of aggregation, there was no more increase in ThT fluorescence and the
aggregation was assumed to be complete (Figure 1D). CaM in a 1:1 ratio was added,
and compared to samples where either the same volume of buffer was added or nothing
was added in order to be able to identify any effects of dilution. Unfortunately, there was
no change in ThT after CaM addition. Interestingly, as the protein aggregated, the ThT
fluorescence readings, while higher than the baseline, became very erratic. It was
thought that this could be due to large bundled fibers transiently passing over the read
area of the well. When these samples were diluted, the readings became more stable,
perhaps indicating unbundling or more uniform dispersion of bundles throughout the
sample. Further work is needed to get more structural information or to visualize these
effects using electron microscopy.

We also wanted to investigate the effect of CaM on another disease-associated
amyloid, α-synculein, which forms amyloid associated with Parkinson’s disease. In
aggregation assays, α-synculein is usually incubated at both a high concentration of 100
uM and at a higher temperature of 37°C in the Langen lab. Unfortunately, since CaM
must be added to the protein of interest, it must be diluted to some extent. While we
were able to concentrate α-synculein about 150 uM, it still needed to be diluted down to
75 uM. We were also unaware of the temperature increase needed for these
Figure 2: ThT of α-synuclein with
CaM shows overall increase. ThT
of α-synuclein alone (red) and α-
synuclein with CaM (blue) in a 1:1
molar ratio.

59
experiments. However, there was a net increase of ThT fluorescence over time, and this
was during a much quicker timeframe than seen normally in the Langen lab (usually it
needs about 24 hours) (Figure 2). However, as ThT can be difficult to interpret without
other supporting techniques, we are unsure of whether these are aggregates that reflect
pathological amyloid formation. The sample with a 1:1 ratio of CaM added also showed
an increase in ThT, albeit slightly more slowly. It is possible, since α-synculein also uses
an amphipathic helix to help initiate aggregation, that CaM could inhibit aggregation in a
similar way to Orb2A88, but more work is needed to be able to form a stronger
conclusion.

60
Appendix B: Aggregation of the N-terminal amino acids of
Orb2A

By Maria Soria
1
, J. Mario Isas
2
and Ansgar Siemer
1

1
Department of Physiology and Biophysics, Zilkha Neurogenetic Institute, Keck School
of Medicine, University of Southern California, 1501 San Pablo Street, Los Angeles,
California 90033, United States;
2
Department of Biochemistry and Molecular Medicine,
Zilkha Neurogenetic Institute, Keck School of Medicine, University of Southern
California, 1501 San Pablo Street, Los Angeles, California 90033, United States

Rationale

The Q-rich domain was originally hypothesized to form the β-sheet core of the
Orb2 amyloid fibrils for several reasons. The Q-rich domain is necessary for Orb2A
function in LTM, and the first 88 amino acids of Orb2A (Orb2A88) can form puncta in
cells on their own (8). There are also several disease-related proteins that have a poly-
Q domain that, in the case of abnormal expansion, are associated with disease onset
and progression and form the β-sheet core of these fibrils. We originally ordered a
peptide containing only the first N-terminal 22 amino acids of Orb2A (Orb2A22) in order
to study whether it could form pre-fibrillar oligomers. However, when we initially
dissolved the peptide in buffer and analyzed its structure, we noticed that these amino
acids were capable of forming amyloid fibrils on their own (13). When the structure of
Orb2A88 amyloid fibrils was analyzed using solid-state NMR, the most static region of
the protein was actually the amphipathic region in the first 20 amino acids of Orb2A88
(13). The following studies show the identification and characterization of the fibrils
formed from the first 22 amino acids of Orb2A.

Methods

Orb2A22 aggregation
Orb2A22 lyophilized powder was ordered as a synthetic peptide from Anaspec
(Fremont, CA). A small amount (~0.2 g) of Orb2A22 lyophilized powder was dissolved in
Tris buffer (75 mM Tris HCl, 10 mM NaCl, at pH 7.6). Concentration was determined
using UV at 280 nm and the solution was diluted to a protein concentration of 45 µM.

61
The protein solution was frozen at -80°C and later thawed and the resulting protein
aggregates visualized using a Jeol (Tokyo, Japan) JEM 1400 electron microscope,
staining with 1% uranyl acetate for 2 min. Thioflavin-T (ThT) fluorescence
measurements of the same sample were performed 6 days later using 50 µM ThT on a
Jasco (Easton, MD) FP-6500 fluorimeter. Orb2A22 was also dissolved in 1 M urea, 100
mM KCl, 10 mM HEPES, and 1 mM DTT and was incubated for 2 days. Then it was
imaged using the same EM procedure described above.

Orb2A22 disaggregation

For the first disaggregation attempt, which was originally published by Lu and
Murphy (75), we dissolved about 0.05 g of synthetic lyophilized Orb2A22 in 20 µL of
pure formic acid and vortexed vigorously. This was then incubated for 30 min and then
either flash frozen and lyophilized or evaporated with N
2
gas. This was then dissolved in
CD buffer (75 mM phosphate, 100 mM NaF, pH 7.6) and measured using a Jasco J-810
circular dichroism (CD) spectropolarimeter.

For another attempt at disaggregation, a small portion (~0.2 g) of lyophilized
Orb2A22 powder was dissolved in 0.5% TFA in water. To 25 µL of the sample was
added 5 µL DMSO and 45 µL HEPES buffer. The original sample in TFA water was
measured using on the same CD spectropolarimeter as above. It was then centrifuged
in an Eppendorf 5804 R centrifuge in an F-45-30-11 rotor at 13,500 RPM for 15 min. CD
spectra were measured as described above.

Disaggregation was also tried by dissolving the same small portion of Orb2A22
(~0.2 g) in 50 µL trifluoroacetic acid (TFA) in 950 µL hexoisopropanol (HFIP), with the
TFA added to the peptide first, followed by the HFIP. This was then vortexed and
lyophilized. The resulting residue was dissolved in 70% formic acid and was lyophilized
again. The resulting residue was finally dissolved in 5 µL DMSO and then 95 µL TFA
water and visualized using EM as described above.
For another disaggregation attempt, 8M GuHCl was added directly to ~0.2 g of
Orb2A22 powder. This was visualized on the electron microscope as described above.
Another ~0.2 g portion of Orb2A22 powder was dissolved in 50 µL TFA and then 950 µL

62
HFIP and this was then evaporated off with N
2
gas. The resulting residue was
redissolved in CD buffer and analyzed using CD as described above.

Orb2A22 expression and purification
The ketosteroid isomerase (KSI)-Orb2A22 plasmid (ksi-(MYNKFVNFIC
GGLPNLNLNK PP)
3
MLLEHHHHHH) was designed in a pET32 vector with the KSI
sequence added to the N-terminal side of Orb2A22, and was ordered from GenScript
(Nanjing, China). This was expressed in BL-21 E. coli cells in LB broth with Ampicillin.
25 mL LB starter cultures were grown for approximately 3 hrs until cloudy, and then
transferred to 500 mL LB flasks. These were grown to an OD
600
=0.6 and then induced
using IPTG. The cells were allowed to express for 4 hours at 37°C and then spun down
into pellets, which were stored at -80°C. The inclusion bodies from these pellets were
purified as follows. Pellets were thawed and lysis buffer (50 mM Tris, 100 mM NaCl,
0.5% Triton-X 100) was added along with 0.04 g lysozyme, 0.05% β-mercaptoethanol,
and 0.8 mL 5x protease inhibitor. This was then vortexed to suspend the pellets,
sonicated 3 times for 1.5 min each and then centrifuged at 10,000 rpm in a Sorvall 6000
rotor for 15 min at 4°C. The supernatant was collected for analysis and the pellets were
resuspended by vortexing in buffer containing 50 mM Tris, 100 mM NaCl and 0.5%
Triton-X. This was then sonicated twice for 2 min and centrifuged in the same rotor
again for 15 min at 10,000 rpm. The same procedure was repeated with the pellets
twice more until final pellets were dissolved in buffer containing 8M urea, 250 mM NaCl,
200 mM Na
2
HPO
4
, 10 mM citrate and 10% glycerol. This was then purified on His-
Select ® Nickel Affinity Gel (Sigma Aldrich, St. Louis, MO), washed with 8M urea buffer
containing 0.5% Triton-X 100, then containing 500 mM NaCl, then slowly stepwise
decreasing pH for elution at pH 6.75, 5.25, 4.60, 4.20 and 3.75. SDS-PAGE gels were
run to identify which fractions contained protein. Fractions that contained protein were
dialyzed against deionized water and lyophilized.

Cleavage of the KSI tag was achieved as previously published (76). Lyophilized
KSI-Orb2A22 was dissolved in 700 µL formic acid, and this was purged with N
2
gas.
Then a single crystal of CNBr was added and the tub was capped, parafilmed and
wrapped in foil to incubate for 16 hrs in darkness. Then 7 mL dH
2
O was added to

63
neutralize the reaction and this was flash frozen and lyophilized. The powder was
dissolved in 8 M GuHCl, which was then diluted 8x and lyophilized again to remove any
traces of CNBr. The final protein powder was dissolved in 5% TFA in HFIP to attempt
disaggregation and then it was diluted to 150 mL with dH
2
O. This was then purified
using a GE Healthcare (Chicago, IL) HiTrap SP HP cation exchange column on a Bio-
Rad (Hercules, CA) NGC Chromatography System FPLC using a step increase in pH at
6, 7.5 and 10.5 or a salt gradient from 0 to 1 M. While the KSI tag showed in the
chromatograph and on the gel, it was impossible to confirm the presence of Orb2A22 in
any fraction due to its small size not appearing on an SDS PAGE gel. We tried to
directly fibrilize fractions that appeared on the chromatogram but not on the gel, and this
did not yield any fibrils either by visual inspection or ThT fluorescence.

The untagged Orb2A22 plasmid was developed using site-directed mutagenesis,
which added a stop codon after amino acid 22 in the Orb2A88 plasmid pET28b. This
was then expressed in BL-21 E. coli cells in LB broth with kanamycin. After induction at
OD
600
=0.6 with IPTG, the cultures were allowed to express at either 37°C for four hours
30°C or 18°C overnight. We were not able to use SDS-PAGE to verify expression due
to the small peptide size. Pellets were purified using the inclusion body purification
described for KSI-Orb2A22, and the final pellets were purified using the same Bio-Rad
NGC Chromatography system, but using a Phenomenex (Torrence, CA) Jupiter 15 µm,
300 Å, 250 x 4.6 mm C4 column. We used a methanol gradient to elute the protein and
compared chromatograms from both induced and non-induced cell culture lysate
(2/27/18). Multiple peaks were seen for both induced and non-induced lysate, with no
one peak that stood out significantly more in the chromatograph of the induced
compared with the non-induced. However, we were not able to positively identify any
Orb2A22 from these peaks. ESI-TOF mass spectroscopy was performed at the Scripps
Research Institute (La Jolla, CA) on fractions 41-43 and 50-51, but no peak matching
the mass of Orb2A22 was identified.

Orb2A21-88 expression, purification and aggregation
The Orb2A21-88 expression vector was made by deleting the first 20 amino
acids from the Orb2A88 expression plasmid. A tryptophan was inserted after the

64
histidine tag and before the stop codon to allow for concentration determination. This
construct was grown in Rosetta ® E. coli (EMD Millipore, Burlington, MA) and purified
using a His-Select ® Nickel Affinity Gel (Sigma Aldrich, St. Louis, MO) resin with Triton-
X 100, high salt, and low pH washes as previously described for Orb2A88 (16).
Orb2A21-88 was buffer exchanged from the high-imidazole elution buffer into either CD
buffer for ThT experiments without seeds, or HEPES buffer (20 mM HEPES, 100 mM
NaCl) for experiments with seeds. ThT was performed using an Eppendorf AF-2200
plate reader on a 96 well plate with intermittent shaking and a final concentration of 50
µM ThT. Excitation was 440 nm and emission was 480 nm. Htt exon 1 Q46 seeds were
prepared in the Langen lab by Mario Isas as previously described (73). Samples with
Htt seeds contained 5% seeds. Baseline fluorescence of buffer or buffer with 5% seeds
was subtracted from samples containing no seeds or 5% seeds respectively.

Results and Discussion

The first 22 amino acids are necessary and sufficient to form amyloid fibrils of
Orb2A88
The first 88 amino acids of Orb2A contain an amphipathic region followed by a
Q-rich region (Figure 1A). The entire amphipathic region is contained in the first 22
amino acids, which were of interest to us in oligomerization studies. Orb2A22 peptide
was initially dissolved in 75 mM Tris buffer and, after freezing for storage and thawing,
was shown to have fully-formed amyloid fibrils already present in solution by EM
(Figure 1B). These fibrils were ThT positive, showing a small but definite fluorescence
peak at 480 nm (Figure 1C). Orb2A22 also showed amyloid fibrils on EM after being
dissolved in 1 M urea buffer (Figure 1D). While this peptide was not expected to be
able to form amyloid fibrils on its own, we show here that it is able to form these fibrils
and that these fibrils are ThT positive, similar to many other amyloid fibrils.

We deleted the first 20 amino acids of Orb2A88, creating a construct which we
called Orb2A21-88, and used this to study the necessity of the N-terminal amphipathic
region in aggregation. We saw that when the first 20 amino acids of OrbA88 are
deleted, it is no longer able to form ThT-positive aggregates in the same time frame as

65
Orb2A88 (Figure 1E). Following these observations we wondered whether the Q-rich
region of Orb2A88 could be induced into forming β-sheets by adding Htt Q46 seeds into
the samples. In Htt samples, this promotes aggregation of monomeric Htt (73). Since
the β-sheets are located in the poly-Q region of Htt, we thought perhaps these seeds
would be able to seed Orb2A21-88 into ThT positive aggregates. However, we saw no
aggregation of Orb2A21-88, even with the addition of Htt Q46 seeds (Figure 1F).

Orb2A22 does not readily disaggregate

We wanted to be able to study and compare the kinetics of amyloid formation of
Orb2A22 with other fragments of Orb2A as well as other amyloid forming proteins.
Unfortunately, since the protein was already in amyloid fibrils upon dissolving in buffer,
Figure 1: Orb2A22 forms amyloid fibrils, but not Orb2A21-88. A) Sequence of the first 88
amino acids of Orb2A. B) Fibrils formed by Orb2A22 in Tris buffer. C) ThT reading of same
Orb2A22 sample in A. D) Fibrils formed by Orb2A22 in 1 M Urea after 2 days of incubation. E)
ThT over time of Orb2A88 and Orb2A21-88. F) ThT over time of Orb2A88 and Orb2A21-88
both with and without seeds.

66

Figure 2: Disaggregation of Orb2A22 amyloid fibrils is difficult. A) CD of Orb2A22 treated with formic
acid. Pink shows the fraction that was lyophilized, black shows the fraction that was evaporated with
N2 prior to redissolving. B) CD of Orb2A22 treated with TFA water and then re-dissolved in buffer
(pink) and the supernatant of that sample after centrifugation (black). C) CD of Orb2A22 after
treatment with 5% TFA in HFIP dried with N2 gas instead of lyophilization. D) Fibrils of Orb2A22 that
was treated with 5% TFA in HFIP, 70% formic acid and finally dissolved in DMSO before diluting
with buffer. Black bar represents 200 nm. E) Orb2A22 sample which was dissolved directly with 8M
GuHCl. Black bar represents 200 nm.

67
we needed a method for disaggregating the stock lyophilized peptide before performing
any further studies. The first method we tried was previously published for asparagine
repeat synthetic peptides using formic acid (75). To remove the formic acid from the
sample, we tried both lyophilization and evaporating under N
2
gas, then the residue was
dissolved in buffer. CD spectra show that the lyophilized sample contains a large
amount of β-sheet still in solution, while the N
2
gas evaporated sample shows massive
loss of signal (Figure 2A). This could be due the actual peptide being blown out of the
tube along with the formic acid, or it could have stuck to the sides of the plastic
microfuge tube, which was not siliconized. However, the CD of the lyophilized sample
showed that the formic acid alone did not disaggregate the β-sheet fibrils.
We were able to somewhat disaggregate Orb2A22 by dissolving it in 5%

trifluoroacetic acid (TFA) and 95% hexafluorisopropanol (HFIP), adding the TFA first
followed by the HFIP, and then lyophilizing. A 100 uL aliquot was removed, evaporated
with N
2
gas, and redissolved in 8M GuHCl, which was then used to calculate protein
concentration in the whole sample using absorbance at 280 nm. After the rest of the
sample was lyophilized, the powder could be mostly dissolved in buffer and the CD
showed a mostly random-coil conformation, as seen in Chapter 2, Figure 2D. This
procedure allowed for enough monomer to see an increase in α-helical content when
lipid vesicles were added, but visible solid particles were frequently present when
redissolving in buffer. These were centrifuged to the bottom of the tube, so only the
supernatant was taken and used for experiments. In this way, we were unable to
calculate an accurate final concentration for Orb2A22 using this method.

Several other attempts were made to completely disaggregate Orb2A22
lyophilized peptides. We wanted to measure binding affinity with calmodulin and needed
an exact concentration of monomeric Orb2A22. We tried first treating the peptide with
0.5% TFA water, as previously described (74) to keep Htt from fibrilizing and also help
to avoid clumping of fibrils. We then lyophilized the TFA water off, and tried dissolving
the resulting powder in DMSO and diluting with our desired buffer. However, this
resulted in a visibly cloudy solution indicating that the disaggregation did not work. We
also tried dissolving the same powder from lyophilization after TFA water treatment
straight into buffer, but that resulted in β-sheets as measured by CD (Figure 2B).

68
Despite centrifuging, the supernatant still contained β-sheets. Consequently, we
hypothesized that the lyophilization could cause aggregates to re-form after
disaggregation with TFA/HFIP. Therefore, we tried disaggregation with TFA/HFIP
followed by evaporation with N2 gas to remove the TFA/HFIP solution. However, we still
saw the β-sheets present in CD when the residue was dissolved in buffer (Figure 2C).

Another disaggregation attempt was made with both the 5% TFA in HFIP mixture
that we used for the lipid binding CD experiments (Chapter 2, Figure 2D), followed by
treatment with 70% formic acid. After lyophilizing off the formic acid, we dissolved the
resulting powder in 5% DMSO and in TFA water. The resulting protein solution was
visualized using EM, and we saw amyloid fibrils present (Figure 2D). We finally tried
dissolving in 8 M guanidine HCl. We saw no fibrils on the EM images of this sample
(Figure 2E), however it is quite difficult to exchange this solution into a buffer that would
be usable in aggregation and binding experiments, as Orb2A22 is quite small and also
quite sticky. We did see some interesting structures that looked to be phase separated
droplets, which makes sense for such a hydrophobic peptide in a very high salt buffer.

Detection of purified recombinant Orb2A22 is difficult due to small size
We also tried to express and purify Orb2A22 from bacteria, and thus get a
monomeric solution from the start. We first used a fusion construct that attached
Orb2A22 to a ketosteroid isomerase (KSI) tag following the protocol previously used for
Aβ by Sharpe et al. (76). This protocol used cyanogen bromide to cleave off the KSI tag
before the first Met of the peptide sequence. While we were able to express and purify
the recombinant protein (Figure 3A and 3B), there was great difficulty in identifying the
Orb2A22 cleavage product. We tried using a cation exchange column with a pH step
gradient to separate the KI tag (PI 4.78) and the Orb2A22 cleavage product (PI 9.19). It
was easy to identify the KSI tag eluting in a peak using SDS PAGE (Figure 3C), but we
were not able to identify Orb2A22 in SDS PAGE as the molecular weight is too small.
We then tried using a salt gradient, to see if it would cause a more obvious elution peak,
which could be Orb2A22 (Figure 3D). The putative Orb2A22 peaks were followed using
ThT, but no signal was detected over time. It is possible that Orb2A22 was present in
any of the peaks coming off of the cation exchange column, but since it is not detectable

69

Figure 3: Expression and purification of Orb2A22. A) SDS-PAGE gel stained with Coomassie Blue
showing pre- and post- induction aliquots of KSI-Orb2A22. The arrow indicates the location of the KSI-
Orb2A22 band based on molecular weight. B) SDS-PAGE gel stained with Coomassie Blue showing
purification aliquots from the KSI-Orb2A22 purification. S1, S2 and S3 are the supernatants from their
respective centrifugations. KSI-Orb2A22 was found in the pH 4.25 elution, indicated by the arrow. C)
FPLC chromatograph of the cation exchange purification of Orb2A22 from the KSI tag. The circled
peak was identified to contain KSI using SDS PAGE. D) FPLC chromatograph of the cation exchange
purification of the Orb2A22 peptide from the KSI tag using a salt gradient. The peak indicated was
followed with ThT, but no increase in ThT was seen. E) FPLC chromatograph of non-induced cell
lysate run on a C4 column. F) The same FPLC protocol for the C4 column was run using induced cell
lysate. The indicted peaks were analyzed by mass spectroscopy and showed no presence of
Orb2A22.

70
using SDS PAGE, and may not be detectable using ThT (Orb2A88 does not give very

high ThT) it was difficult to identify. UV was not helpful, as it cannot distinguish between
KSI and Orb2A22.

We decided to try another expression and purification method. This time, we
expressed the Orb2A22 peptide with no fusion tag, and used a C4 reverse-phase
chromatography column in an attempt to purify the peptide. This was complicated by the
fact that since Orb2A22 does not appear using SDS-PAGE, we were not able to verify
that expression occurred, or how robust the expression was. We used lysate from cell
cultures that had been both induced and not induced in order to compare
chromatographs and identify peaks that were unique to the induced lysate. We saw
very little difference between induced and non-induced lysate chromatographs (Figures
3E and 3F). The two most prominent peaks were collected and sent for mass
spectroscopy analysis, but this showed nothing in those fractions that was near the
mass of Orb2A22. This could mean the peptide never expressed in the first place, or
that the purification protocol is not adequate. However, due to the difficulty of detecting
Orb2A22 on an SDS PAGE gel, it will be complicated to understand which part of the
entire protocol leads to the absence of Orb2A22.
Designing a new plasmid construct with a solubility tag attached to the Orb2A22
peptide would possibly allow for easier expression, purification, concentration
determination and control over aggregation, which were all major problems with the
production methods we tried. Development of an antibody against Orb2A22 would also
help by making it possible to do Western blots for identification and even quantification
of Orb2A22 at various points of the protocol. Ultimately, more work is needed to
understand the nature of the amphipathic region of Orb2A in the context of the behavior
and function of the whole protein.

71
Appendix C: Modulating intracellular Ca2+ levels changes
Orb2A-eGFP puncta

By Maria A. Soria
1
and Ansgar B. Siemer
1

1
Department of Physiology and Biophysics, Zilkha Neurogenetic Institute, Keck School
of Medicine, University of Southern California, 1501 San Pablo Street, Los Angeles,
California 90033, United States

Rationale

While we have seen the effects of the presence of Ca
2+
bound CaM on the
aggregation of Orb2A88 (Chapter 3), we do not know if this interaction also occurs in
vivo. It is well known that Ca
2+
and CaM are necessary signaling mechanisms in
memory (29), but it has not yet been investigated how Ca
2+
can affect the aggregation
and long-term memory (LTM) function of Orb2A. Previously, Orb2A tagged with eGFP
was visualized in Drosophila Schneider 2 (S2) cells as puncta or diffusely distributed
indicating aggregation or no aggregation respectively (8). We decided to test the effects
of intracellular Ca
2+
using the same system. We used ionomycin as a promoter of Ca
2+

cellular influx and BAPTA-AM as an intracellular Ca
2+
chelator and visualized the effects
of these two compounds on Orb2A-eGFP puncta in S2 cells.

Materials and Methods

The Orb2A-eGFP pAc.5 plasmid and eGFP pAc.5 control plasmid were gifts from
the lab of Kausik Si at the Stowers Institute in Kansas City. S2 cells were purchased
from ThermoFisher Scientific (Waltham, MA).
S2 cells were thawed and cultured according to the protocol from Thermo
Scientific. Briefly, cells that had been stored in complete Schneider’s media with 10%
DMSO were quickly thawed and the DMSO solution washed off. Cells were plated in
fresh complete Schneider’s media and allowed to grow until between 2x10
6
and 2x10
7

cells/ml. Cell concentration and viability was measured on an Orflo (Ketchum, ID) Moxi
automated cell counter. Viabilities over 98% were typical. 3 ml of 1x10
6
cells/ml were
plated in a 35 mm plate with fresh complete media and were allowed to grow overnight
at room temperature with no shaking. These cells were then CaCl
2
transfected with the

72

appropriate plasmid and incubated again overnight. The cells were then pelleted by
centrifugation at 100 RPM for 5 min using an Eppendorf (Hamburg, Germany) 5804R
centrifuge in an A-44-4 rotor, and CaCl
2
media washed off. For experiments involving a
Figure 1: Influx of Ca2+ decreases Orb2A-eGFP puncta fluorescence in S2 cells.
A) Cells transfected with eGFP only. B) Cells transfected with Orb2A tagged with eGFP.
DMSO added 4 h prior to imaging. C) Orb2A-eGFP cells with 1
µM BAPTA-AM added 4
hrs prior to imaging. D) Orb2A-eGFP cells with 1uM ionomycin added 4 h prior to
imaging. E) Orb2A-eGFP cells with 1 µM ionomycin added 16 hrs prior to imaging. All
images shown are at 63x magnification.

73
16 hr incubation, ionomycin or BAPTA-AM in DMSO was added at this point to a final
concentration of 1 µM. Otherwise the cells were left overnight again. Finally, in the
morning, a final concentration of either ionomycin, BAPTA-AM or DMSO was added if it
had not been added previously and left for approximately 4 hours until visualized in the
afternoon using a Zeiss (Oberkochen, Germany) AxioImager2.

Results and Discussion

S2 cells were successfully transfected with eGFP only, which showed diffuse
green color present throughout the cell (Figure 1A). However, when cells were
transfected with Orb2A tagged with eGFP (Orb2A-eGFP), the fluorescence is located in
puncta throughout the cytoplasm (Figure 1B). This is consistent with previous results
and indicates the aggregation of Orb2A within the cell (8). When we added BAPTA-AM
4 hours prior to imaging, we saw similar puncta throughout the cell (Figure 1C). This
indicates that the sequestering of Ca
2+
out of the cytoplasm and the subsequent
decrease in Ca
2+
bound CaM does not have an effect on the aggregation abilities of
Orb2A. However, when we added ionomycin, both 4 hours prior and 16 hours prior to
imaging, the puncta seemed much smaller and less bright, and the green color more
diffuse throughout the cell (Figures 1D and 1E). This means that the increase in
intracellular Ca
2+
levels and subsequent increase in active Ca
2+
bound CaM may
decrease the amount of Orb2 aggregates in the cell. This is consistent with the finding
in Chapter 3 that Ca
2+
/CaM inhibits Orb2A88 aggregation. More work is needed to
quantify the puncta found under each condition in order to understand the statistical
significance of these differences.

These results are not necessarily expected because Ca
2+
/CaM is known to
activate memory related processes (29). However, this has mainly been studied in the
context of CamKII activity which, as a LTP initiator, could precede the aggregation of
Orb2 (77). How these two processes are exactly connected is unclear, and more work is
needed to link the activities of Ca
2+
/CaM and the aggregation of Orb2A in LTM.

74
References

1. Astbury, W.T., S. Dickinson, and K. Bailey. 1935. The X-ray interpretation of
denaturation and the structure of the seed globulins. Biochem. J. 29: 2351–
2360.1.
2. Chiti, F., and C.M. Dobson. 2017. Protein Misfolding, Amyloid Formation, and
Human Disease: A Summary of Progress Over the Last Decade. Annu. Rev.
Biochem. 86: 27–68.
3. Chapman, M.R., L.S. Robinson, J.S. Pinkner, R. Roth, J. Heuser, M. Hammar, S.
Normark, and S.J. Hultgren. 2002. Amyloid Fiber Formation Role of Escherichia
coli Curli Operons in Directing Amyloid Fiber Formation. Science (80-. ). 295:
851–856.
4. Fowler, D.M., A. V. Koulov, C. Alory-Jost, M.S. Marks, W.E. Balch, and J.W.
Kelly. 2006. Functional amyloid formation within mammalian tissue. PLoS Biol. 4:
0100–0107.
5. Si, K., S. Lindquist, and E.R. Kandel. 2003. A Neuronal Isoform of the Aplysia
CPEB Has Prion-Like Properties. 115: 879–891.
6. Keleman, K., S. Krüttner, M. Alenius, and B. Dickson. 2007. Function of the
Drosophila CPEB protein Orb2 in long-term courtship memory. Nat. Neurosci. 10:
1587–1593.
7. Si, K., M. Giustetto, A. Etkin, R. Hsu, A.M. Janisiewicz, M.C. Miniaci, J. Kim, H.
Zhu, E.R. Kandel, N. York, and N. York. 2003. A Neuronal Isoform of CPEB
Regulates Local Protein Synthesis and Stabilizes Synapse-Specific Long-Term
Facilitation in Aplysia. 115: 893–904.
8. Majumdar, A., W.C. Cesario, E. White-Grindley, H. Jiang, F. Ren, M.R. Khan, L.
Li, E.M.-L. Choi, K. Kannan, F. Guo, J. Unruh, B. Slaughter, and K. Si. 2012.
Critical role of amyloid-like oligomers of Drosophila Orb2 in the persistence of
memory. Cell. 148: 515–29.
9. Krüttner, S., B. Stepien, J.N. Noordermeer, M. a Mommaas, K. Mechtler, B.J.
Dickson, and K. Keleman. 2012. Drosophila CPEB Orb2A mediates memory
independent of Its RNA-binding domain. Neuron. 76: 383–95.
10. Khan, M.R., L. Li, C. Pérez-Sánchez, A. Saraf, L. Florens, B.D. Slaughter, J.R.
Unruh, and K. Si. 2015. Amyloidogenic Oligomerization Transforms Drosophila
Orb2 from a Translation Repressor to an Activator. Cell. 163: 1468–1483.
11. Si, K., and E.R. Kandel. 2016. The Role of Functional Prion-Like Proteins in the
Persistence of Memory. .
12. Hervás, R., L. Li, A. Majumdar, M. del C. Fernández-Ramírez, J.R. Unruh, B.D.
Slaughter, A. Galera-Prat, E. Santana, M. Suzuki, Y. Nagai, M. Bruix, S. Casas-
Tintó, M. Menéndez, D. V. Laurents, K. Si, and M. Carrión-Vázquez. 2016.
Molecular Basis of Orb2 Amyloidogenesis and Blockade of Memory
Consolidation. PLOS Biol. 14: e1002361.
13. Cervantes, S.A., T.H. Bajakian, M.A. Soria, A.S. Falk, R.J. Service, R. Langen,
and A.B. Siemer. 2016. Identification and Structural Characterization of the N-
terminal Amyloid Core of Orb2 isoform A. Sci. Rep. 6: 38265.
14. Perrin, R.J., A.M. Fagan, and D.M. Holtzman. 2009. Multimodal techniques for

75
diagnosis and prognosis of Alzheimer’s disease. Nature. 461: 916–922.
15. Katzman, R., R. Terry, R. DeTeresa, T. Brown, P. Davies, P. Fuld, X. Renbing,
and A. Peck. 1988. Clinical, Pathological and Neurochemical Changes in
Dementia: A Subgrou with Preserved Mental Status and Numerous Neocortical
Plaques. Ann Neurol. 23: 138–144.
16. Soria, M.A., S.A. Cervantes, T.H. Bajakian, and A.B. Siemer. 2017. The
Functional Amyloid Orb2A Binds to Lipid Membranes. Biophys. J. 113: 37–47.
17. Bajakian, T.H., S.A. Cervantes, M.A. Soria, M. Beaugrand, J.Y. Kim, R.J. Service,
and A.B. Siemer. 2017. Metal binding properties of the N-Terminus of the
functional amyloid orb2. Biomolecules. 7: 1–13.
18. Hoop, C.L., H.-K. Lin, K. Kar, Z. Hou, M. a Poirier, R. Wetzel, and P.C. a van der
Wel. 2014. Polyglutamine amyloid core boundaries and flanking domain dynamics
in huntingtin fragment fibrils determined by solid-state NMR. Biochemistry. .
19. Isas, J.M., R. Langen, and A.B. Siemer. 2015. Solid-State Nuclear Magnetic
Resonance on the Static and Dynamic Domains of Huntingtin Exon-1 Fibrils.
Biochemistry. 54: 3942–3949.
20. White-Grindley, E., L. Li, R. Mohammad Khan, F. Ren, A. Saraf, L. Florens, and
K. Si. 2014. Contribution of Orb2A stability in regulated amyloid-like
oligomerization of Drosophila Orb2. PLoS Biol. 12: e1001786.
21. Pi, H.J., and J.E. Lisman. 2008. Coupled phosphatase and kinase switches
produce the tristability required for LTP and LTD. J. Neurosci. 28: 13132–13138.
22. Schratt, G.M., F. Tuebing, E.A. Nigh, C.G. Kane, M.E. Sabatini, M. Kiebler, and
M.E. Greenberg. 2006. A brain-specific microRNA regulates dendritic spine
development. Nature. 439: 283–289.
23. Lee, H.J., C. Choi, and S.J. Lee. 2002. Membrane-bound alpha-synuclein has a
high aggregation propensity and the ability to seed the aggregation of the
cytosolic form. J. Biol. Chem. 277: 671–678.
24. Zhu, M., and A.L. Fink. 2003. Lipid binding inhibits alpha-synuclein fibril formation.
J. Biol. Chem. 278: 16873–7.
25. Burke, K.A., K.J. Kauffman, C.S. Umbaugh, S.L. Frey, and J. Legleiter. 2013. The
interaction of polyglutamine peptides with lipid membranes is regulated by
flanking sequences associated with huntingtin. J. Biol. Chem. 288: 14993–15005.
26. Knight, J.D., and A.D. Miranker. 2004. Phospholipid catalysis of diabetic amyloid
assembly. J. Mol. Biol. 341: 1175–87.
27. Jiang, Z., and J.C. Lee. 2014. Lysophospholipid-Containing Membranes Modulate
the Fibril Formation of the Repeat Domain of a Human Functional Amyloid,
Pmel17. J. Mol. Biol. 426: 4074–4086.
28. O’Neil, K.T., and W.F. DeGrado. 1990. How calmodulin binds its targets:
sequence independent recognition of amphiphilic alpha-helices. Trends Biochem.
Sci. 15: 59–64.
29. Xia, Z., and D.R. Storm. 2005. The role of calmodulin as a signal integrator for
synaptic plasticity. Nat Rev Neurosci. 6: 267–276.
30. Atkins, C.M., N. Nozaki, Y. Shigeri, and T.R. Soderling. 2004. Cytoplasmic
Polyadenylation Element Binding Protein-Dependent Protein Synthesis Is
Regulated by Calcium/Calmodulin-Dependent Protein Kinase II. J. Neurosci. 24:
5193–5201.

76
31. Krüttner, S., B. Stepien, J.N. Noordermeer, M.A. Mommaas, K. Mechtler, B.J.
Dickson, and K. Keleman. 2012. Drosophila CPEB Orb2A mediates memory
independent of Its RNA-binding domain. Neuron. 76: 383–95.
32. Davidson, W.S., A. Jonas, D.F. Clayton, and J.M. George. 1998. Stabilization of
alpha -Synuclein Secondary Structure upon Binding to Synthetic Membranes. J.
Biol. Chem. 273: 9443–9449.
33. Jao, C.C., B.G. Hegde, J. Chen, I.S. Haworth, and R. Langen. 2008. Structure of
membrane-bound alpha-synuclein from site-directed spin labeling and
computational refinement. Proc. Natl. Acad. Sci. U. S. A. 105: 19666–19671.
34. Cole, N.B., D.D. Murphy, T. Grider, S. Rueter, D. Brasaemle, and R.L. Nussbaum.
2002. Lipid droplet binding and oligomerization properties of the Parkinson’s
disease protein alpha-synuclein. J. Biol. Chem. 277: 6344–6352.
35. Jayasinghe, S. a, and R. Langen. 2005. Lipid membranes modulate the structure
of islet amyloid polypeptide. Biochemistry. 44: 12113–9.
36. Apostolidou, M., S. a Jayasinghe, and R. Langen. 2008. Structure of alpha-helical
membrane-bound human islet amyloid polypeptide and its implications for
membrane-mediated misfolding. J. Biol. Chem. 283: 17205–10.
37. Jimbow, M., H. Kanoh, and K. Jimbow. 1982. Characterization of biochemical
properties of melanosomes for structural and functional differentiation: analysis of
the compositions of lipids and proteins in melanosomes and their subfractions. J.
Invest. Dermatol. 79: 97–102.
38. Ward, W.C., and J.D. Simon. 2007. The differing embryonic origins of retinal and
uveal (iris/ciliary body and choroid) melanosomes are mirrored by their
phospholipid composition. Pigment Cell Res. 20: 61–69.
39. Bohlen, P., S. Stein, W. Dairman, and S. Udenfriend. 1973. Fluorometric assay of
proteins in the nanogram range. Arch. Biochem. Biophys. 155: 213–220.
40. Greenfield, N.J. 2006. Using circular dichroism spectra to estimate protein
secondary structure. Nat. Protoc. 1: 2876–90.
41. Okada, A.K., K. Teranishi, J.M. Isas, S. Bedrood, R.H. Chow, and R. Langen.
2016. Diabetic Risk Factors Promote Islet Amyloid Polypeptide Misfolding by a
Common, Membrane-mediated Mechanism. Sci. Rep. 6: 31094.
42. Luo, P., and R.L. Baldwin. 1997. Mechanism of helix induction by trifluoroethanol:
A framework for extrapolating the helix-forming properties of peptides from
trifluoroethanol/water mixtures back to water. Biochemistry. 36: 8413–8421.
43. Rohl, C.A., and R.L. Baldwin. 1997. Comparison of nh exchange and circular
dichroism as techniques for measuring the parameters of the helix-coil transition
in peptides. Biochemistry. 36: 8435–8442.
44. Jones, E., E. Oliphant, P. Peterson, and E. Al. SciPy: Open Source Scientific
Tools for Python. http://www.scipy.org/. : [Online; accessed 2017-5-18].
45. Tseitlin, M., S.S. Eaton, and G.R. Eaton. 2012. Uncertainty analysis for absorption
and first-derivative EPR spectra. Concepts Magn. Reson. 40(A): 295–305.
46. Hunter, J.D. 2007. Matplotlib: A 2D graphics environment. Comput. Sci. Eng. 9:
90–95.
47. Snider, C., S. Jayasinghe, K. Hristova, and S.H. White. 2009. MPEx: A tool for
exploring membrane proteins. Protein Sci. 18: 2624–2628.
48. Hristova, K., W.C. Wimley, V.K. Mishra, G.M. Anantharamiah, J.P. Segrest, and

77
S.H. White. 1999. An amphipathic α-helix at a membrane interface: a structural
study using a novel X-ray diffraction method. J. Mol. Biol. 290: 99–117.
49. Hulme, E.C., and M.A. Trevethick. 2010. Ligand binding assays at equilibrium:
Validation and interpretation. Br. J. Pharmacol. 161: 1219–1237.
50. Caillon, L., A.R.F. Hoffmann, A. Botz, and L. Khemtemourian. 2016. Molecular
structure, membrane interactions, and toxicity of the islet amyloid polypeptide in
type 2 diabetes mellitus. J Diabetes Res. 2016.
51. Nuscher, B., F. Kamp, T. Mehnert, S. Odoy, C. Haass, P.J. Kahle, and K. Beyer.
2004. α-synuclein has a high affinity for packing defects in a bilayer membrane: A
thermodynamics study. J. Biol. Chem. 279: 21966–21975.
52. Segrest, J.P., M.K. Jones, H. De Loof, C.G. Brouillette, Y. V Venkatachalapathi,
and G.M. Anantharamaiah. 1992. The amphipathic helix in the exchangeable
apolipoproteins: a review of secondary structure and function. J. Lipid Res. 33:
141–166.
53. Wieprecht, T., M. Beyermann, and J. Seelig. 2002. Thermodynamics of the coil-
alpha-helix transition of amphipathic peptides in a membrane environment: The
role of vesicle curvature. Biophys. Chem. 96: 191–201.
54. Caillon, L., O. Lequin, and L. Khemtémourian. 2013. Evaluation of membrane
models and their composition for islet amyloid polypeptide-membrane
aggregation. Biochim. Biophys. Acta - Biomembr. 1828: 2091–2098.
55. Lim, L., and M.R. Wenk. 2010. Neuronal Membrane Lipids - Their Role in the
Synaptic Vesicle Cycle. In: Lajtha A, G Tettamanti, G Goracci, editors. Handbook
of Neurochemistry. Boston, MA: Springer US. pp. 224–234.
56. Cimler, B.M., T.J. Andreasen, K.I. Andreasen, and D.R. Storm. 1985. P-57 is a
neural specific calmodulin-binding protein. J. Biol. Chem. 260: 10784–10788.
57. Xia, Z., E. ‐J Choi, F. Wang, C. Blazynski, and D.R. Storm. 1993. Type I
Calmodulin‐Sensitive Adenylyl Cyclase Is Neural Specific. J. Neurochem. 60:
305–311.
58. Wang, H., and D.R. Storm. 2003. Calmodulin-Regulated Adenylyl Cyclases:
Cross-Talk and Plasticity in the Central Nervous System. Mol. Pharmacol. 63:
463–468.
59. Sanhueza, M., and J. Lisman. 2013. The CaMKII/NMDAR complex as a
molecular memory. Mol. Brain. 6: 1–8.
60. Lisman, J. 1994. The CaM kinase II hypothesis for the storage of synaptic
memory. Trends Neurosci. 17: 406–412.
61. Impey, S., A.L. Fong, Y. Wang, J.R. Cardinaux, D.M. Fass, K. Obrietan, G.A.
Wayman, D.R. Storm, T.R. Soderling, and R.H. Goodman. 2002. Phosphorylation
of CBP mediates transcriptional activation by neural activity and CaM kinase IV.
Neuron. 34: 235–244.
62. Bito, H., K. Deisseroth, and R.W. Tsien. 1996. CREB phosphorylation and
dephosphorylation: A Ca2+- and stimulus duration-dependent switch for
hippocampal gene expression. Cell. 87: 1203–1214.
63. Wu, G.-Y., K. Deisseroth, and R.W. Tsien. 2001. Activity-dependent CREB
phosphorylation: Convergence of a fast, sensitive calmodulin kinase pathway and
a slow, less sensitive mitogen- activated protein kinase pathway. Proc. Natl.
Acad. Sci. 98: 2808–2813.

78
64. Shi, J., M. Townsend, and M. Constantine-Paton. 2000. Activity-dependent
induction of tonic calcineurin activity mediates a rapid developmental
downregulation of NMDA receptor currents. Neuron. 28: 103–114.
65. Tidow, H., and P. Nissen. 2013. Structural diversity of calmodulin binding to its
target sites. FEBS J. 280: 5551–5565.
66. Biancalana, M., and S. Koide. 2010. Molecular mechanism of Thio flavin-T
binding to amyloid fibrils. Biochim. Biophys. Acta. 1804: 1405–1412.
67. Stsiapura, V.I., A.A. Maskevich, V.A. Kuzmitsky, K.K. Turoverov, and I.M.
Kuznetsova. 2007. Computational Study of Thioflavin T Torsional Relaxation in
the Excited State. J. Phys. Chem. A. 111: 4829–4835.
68. Voropai, E.S., M.P. Samtsov, K.N. Kaplevskii, A.A. Maskevich, V.I. Stepuro, O.I.
Povarova, I.M. Kuznetsova, K.K. Turoverov, A.L. Fink, and V.N. Uverskii. 2003.
Spectral Properties of Thioflavin T and its Complexes with Amyloid Fibrils. J. Appl.
Spectrosc. 70: 767–773.
69. Erickson-Viitanen, S., and W.F. DeGrado. 1987. Recognition and
Characterization of Calmodulin-Binding Sequences in Peptides and Proteins.
Methods Enzymol. 139: 455–478.
70. Mendez, R., L.E. Hake, T. Andresson, L.E. Littlepage, J. V Ruderman, and J.D.
Richter. 2000. Phosphorylation of CPE binding factor by Eg2 regulates translation
of c-mos mRNA. Lett. to Nat. 404: 302–307.
71. Alexander, K.A., B.T. Wakim, G.S. Doyle, K.A. Walsh, and D.R. Storm. 1988.
Identification and characterization of the calmodulin-binding domain of
neuromodulin, a neurospecific calmodulin-binding protein. J. Biol. Chem. 263:
7544–7549.
72. Andreasen, T.J., C.W. Luetje, W. Heideman, and D.R. Storm. 1983. Purification of
a Novel Calmodulin Binding Protein from Bovine Cerebral Cortex Membranes.
Biochemistry. 22: 4615–4618.
73. Isas, J.M., A. Langen, M.C. Isas, N.K. Pandey, and A.B. Siemer. 2017. Formation
and Structure of Wild Type Huntingtin Exon ‑ 1 Fibrils. Biochemistry. : 1–8.
74. Pandey, N.K., J.M. Isas, A. Rawat, R. V Lee, J. Langen, P. Pandey, and R.
Langen. 2017. The 17-residue-long N terminus in huntingtin controls step-wise
aggregation in solution and on membranes via different mechanisms. J. Biol.
Chem. .
75. Lu, X., and R.M. Murphy. 2014. Synthesis and disaggregation of asparagine
repeat-containing peptides. J. Pept. Sci. 20: 860–7.
76. Sharpe, S., W. Yau, and R. Tycko. 2005. Expression and purification of a
recombinant peptide from the Alzheimer’s beta-amyloid protein for solid-state
NMR. Protein Expr. Purif. 42: 200–210.
77. Buard, I., S.J. Coultrap, R.K. Freund, Y. Lee, M.L.D. Acqua, A.J. Silva, and K.U.
Bayer. 2010. CaMKII “ Autonomy ” Is Required for Initiating But Not for
Maintaining Neuronal Long-Term Information Storage. J. Neurosci. 30: 8214–
8220.

Abstract (if available)

Abstract Orb2 is a protein found in D. melanogaster that has been shown to form functional amyloid-like aggregates in fruit fly brains that are necessary for long-term memory. There are two isoforms of Orb2, A and B. Mounting evidence points to Orb2A as the initiator and regulator of Orb2 amyloid formation. This process must be highly controlled to avoid toxic folding byproducts and intermediates. Many other amyloid-forming proteins are regulated by lipid membranes. In this dissertation, we use electron paramagnetic resonance (EPR) and circular dichroism to show that Orb2A also is able to bind lipid membranes using an amphipathic helix at its unique N-terminus. EPR and electron microscopy show that this binding inhibits amyloid aggregation within the N-terminus. Amphipathic helices are also favored binding partners of the calcium-sensing protein, calmodulin (CaM). Calcium signaling is known to be a major component of long-term memory formation. We use EPR, electron microscopy and thioflavin-t staining to show that CaM binds to the N-terminus of Orb2A in a calcium dependent manner, and this binding also inhibits amyloid aggregation in that region. This leads to the conclusion that the N-terminal amphipathic region of Orb2A is important for regulating N-terminal amyloid formation, and could occur via a number of potential pathways in vivo.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Structure and kinetics of the Orb2 functional amyloid

PDF

Overexpression and interaction of Orb2 in S2 insect cells

PDF

Oligomer formation of functional amyloid protein - Orb2A

PDF

Structural characterization of the functional amyloid Orb2A using EPR spectroscopy

PDF

Structural studies on functional amyloids and the mechanism of aggregation and disaggregation of Huntingtin Exon 1

PDF

Facilitating unambiguous NMR assignment by solid–state NMR using segmental isotope labeling through split-inteins

PDF

The structure and function of membrane curving proteins on different membrane shapes and their regulation by post-translational modifications

PDF

Molecular basis of CD33 receptor function, and effects of a destabilizing transmembrane motif on αXβ2 integrin

PDF

Uncovering the influence of N-terminal phosphorylation on conformational dynamics of huntingtin exon 1 monomer

PDF

Enhancing and inhibiting diabetic amyloid misfolding

PDF

Dual effects of transmembrane proline residues on integrin function

PDF

Role of CD33 structure and function in Alzheimer’s disease

PDF

Interaction of monoclonal antibodies MW1 and PHP1 with huntingtin exon1 protein

PDF

Structural studies of the IAPP membrane-mediated aggregation pathway

PDF

Role of mitochondrially derived peptides in the inhibition of IAPP misfolding

PDF

Before they were amyloid: understanding the toxicity of disease-associated monomers and oligomers prior to their aggregation

PDF

Structural-functional study of the paracellular ion pore

PDF

Developing a method for measuring kinetic rates of RNA/protein interaction using switchSENSE® technology on the DRX2

PDF

The effect of familial mutants of Parkinson's disease on membrane remodeling

PDF

The role of endoplasmic reticulum chaperone protein GRP78 in breast cancer

Asset Metadata

Creator Soria, Maria Ann (author)

Core Title Regulating functional amyloid formation: the promiscuous behavior of the Orb2A N-terminal amphipathic region

School Keck School of Medicine

Degree Doctor of Philosophy

Degree Program Medical Biophysics

Publication Date 10/24/2019

Defense Date 03/07/2019

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

Tag Amyloid,Biophysics,calmodulin,long term memory,NMR,nuclear magnetic resonance,OAI-PMH Harvest,Orb2,structural biology

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Langen, Ralf (committee chair), Farley, Robert (committee member), Siemer, Ansgar (committee member)

Creator Email mariaaco@usc.edu,soriama@laccd.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-142185

Unique identifier UC11676757

Identifier etd-SoriaMaria-7229.pdf (filename),usctheses-c89-142185 (legacy record id)

Legacy Identifier etd-SoriaMaria-7229.pdf

Dmrecord 142185

Document Type Dissertation

Format application/pdf (imt)

Rights Soria, Maria Ann

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