Close
About
FAQ
Home
Collections
Login
USC Login
Register
0
Selected
Invert selection
Deselect all
Deselect all
Click here to refresh results
Click here to refresh results
USC
/
Digital Library
/
University of Southern California Dissertations and Theses
/
The impact of genetic pleiotropy on heterogeneity in the developing forebrain and hindbrain
(USC Thesis Other)
The impact of genetic pleiotropy on heterogeneity in the developing forebrain and hindbrain
PDF
Download
Share
Open document
Flip pages
Contact Us
Contact Us
Copy asset link
Request this asset
Transcript (if available)
Content
THE IMPACT OF GENETIC PLEIOTROPY ON HETEROGENEITY IN THE
DEVELOPING FOREBRAIN AND HINDBRAIN
by
RAMIN ALI MARANDI GHODDOUSI
A Dissertation Presented to the
FACULTY OF THE USC GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPY
(NEUROSCIENCE GRADUATE PROGRAM)
August 2023
Copyright 2023 Ramin Ali Marandi Ghoddousi
ii
Dedication
To my parents Faride Ali Marandi Ghoddousi and Ahad Ali Marandi Ghoddousi and all
the sacrifices that they have made.
To my sister Nazanin Ali Marandi Ghoddousi, who will forever be cooler than me.
And lastly, to my partner Natalie Danielle Beck, the actual best person in the world.
iii
Acknowledgements
The number of friends, loved ones, mentors, and colleagues that have supported
me throughout my decades living on this planet greatly outnumber the number of cell
types in the brain. I consider myself extremely fortunate for these relationships and the
important roles that they have played in my development as a human being and as a
scientist. It is impossible to think how I could have gotten here without all of you.
I have had some wonderful mentors throughout my career. I would like to thank
Drs. Sacha Nelson and David Levitan for giving me my first real taste of the exciting world
of neuroscience. Dr. Anna Kamitakahara was my guide into the phenomenal world of the
brainstem and the gut-brain connection. Anna, thank you for guiding this wide-eyed kid
through the tumultuous first few years of graduate school, and for always being willing to
lend an ear to new ideas or your eyes to a draft. Dr. Kathie Eagleson has been a constant
source of no-nonsense wisdom and knowledge in the lab. Kathie, thank you for your
friendship and for teaching me how to write science! The hit that my ego took after my
first attempt at writing a fellowship application made me question my career choice. Your
help and feedback have been instrumental for every new application and manuscript that
I have written.
Me and ChatGPT combined could not imagine a better mentor than Dr. Pat Levitt.
There is a very large overlap between the qualities that make someone and top-notch
human being and those that make one an excellent mentor, and Pat possesses all those
iv
qualities. Pat, there was never one moment in the six years as your mentee where I didn’t
feel supported and encouraged. I want to thank you for all the guidance, words of wisdom,
compassion, and friendship throughout these years. Thank you for teaching me how to
step back and look at the big picture, and how to home in on the most important questions.
Most of all, thank you for leaving me alone and letting me figure things out! The level of
trust that you showed me as a young scientist was extremely formative in my
development. You have set up a phenomenal work and training environment and have
made it very difficult to want to finish up and leave.
I have worked closely with people in this wonderful environment on numerous
projects. I would like to thank Valerie Magalong, Zia Rady, Anna Kamitakahara, and
Kathie Eagleson for all the hard work that they put into our collaborative projects.
Amanda, my PhD would have taken 12 years if it wasn’t for all the important mouse colony
work that you do. Thank you for everything. A big thank you to all the other postdocs
and research technicians in the lab for your friendship and support over the years. Lastly,
I want to thank the newly minted Dr. Alexandra Lanjewar for being a wonderful lab mate
and friend. Alex, thanks for joining the lab 6 months ahead of me and letting me ride your
coattails through the wasteland of administrative documents and deadlines. I’m going to
miss discussing our lab-related stress dreams.
I would also like to thank my dissertation committee; Drs. Alexandre Bonnin, Ruth
Wood, Alan Watts, and Scott Kanoski. The well though-out questions that you all asked
v
me for my competency exam and the feedback that you gave me during my yearly
meetings was invaluable for my progression as a scientist.
A big thank you to all the wonderful friends that I have made over the years all
across the world. You guys have helped me stay young and have exposed me to so many
new things! There are way too many of you to name, but you know who you are. I look
forward to all the new adventures. I want to send a special thanks to Sara’s Fan Club for
being the most well-rounded group of friends that I have ever had. I also want to send a
special thanks to Chris, my college roommate and one of my oldest friends. Chris, I have
cherished all our wonderful debates over the years. I’m looking forward to the next Death
Valley trip.
Then there is my large and wonderful family. There is no more appropriate place
to start than with my two grandmothers, Molook and Fatemeh. Molook Mamman I miss
you and your sense of humor very much. Thank you for always loving me and making
me laugh. Rest in peace. Fati Mamman, your constant reminders to keep my head up
and your long-distance prayers have helped me though many rough patches. Your words
are always in my mind. I can’t wait until I get a chance to see you again. I want to thank
my uncle Saeed for introducing me for philosophy and psychoanalysis as a teenager.
Daei, you planted the initial seed that sprouted into my current fascination with the mind
and the natural world.
vi
I couldn’t have done any of this without Ahad and Faride Ali Marandi Ghoddousi,
my wonderful mother and father. They picked up their two young kids and moved across
the world from Iran to the US, far from their homelands and most of their family, so that
me and my sister could have the best opportunities as we grew up. I can’t imagine the
courage that that took. I want to thank you both for all that you have done for me and for
never letting me forget my culture, language, and where I am from. I have learned many
valuable skills and lesson from you two over the years. Dad, you are the epitome of hard
work and not letting things get to you. Thank you for teaching me the benefits of hard
work and reminding me to take it easy. Mom, I get my persistence from you. Thank you
for teaching me the value of dusting yourself off and trying again, and again, and again
until you succeed. I could not have made it past all the failed experiments if it wasn’t for
you.
A big thank you goes to my little sister Nazanin, the bravest person that I know.
Naz, your ability to jump head-first (sometimes literally, especially when you were a kid)
into things is inspiring and has always motivated me to do the same. I don’t know if I
would have done half the things that I have done if it wasn’t for you, including starting a
PhD. Very glad to have you in my life and look forward to the next activity, friend, or
music genre that I’m going to steal from you.
Lastly, I would like to thank my wonderful partner, Natalie Beck. I can’t believe
that it has been a decade since we met! Seeing the care and effort that you put into your
work and the difference that you make in your students’ lives has been nothing short of
vii
inspirational. The world needs people that care and I feel honored to be partnered with
someone who cares so much. Your steady and kind support over the years, even when I
was circling cells for hours on end, has been an absolute rock for me. Thank you for being
willing to hop all over the US with me as I try to achieve this goal. And most importantly,
thank you for surrounding me with plants. You wouldn’t believe how many compliments I
get during zoom meetings.
viii
TABLE OF CONTENTS
DEDICATION ............................................................................................................................................... ii
ACKNOWLEDGEMENTS........................................................................................................................... iii
LIST OF FIGURES ..................................................................................................................................... x
CHAPTER 1 : GENERAL INTRODUCTION .............................................................................................. 1
1.1 CONTRIBUTIONS OF CELL TYPE DIVERSITY AND GENETIC PLEITROPY TO
STRUCTURE/FUNCTION RELATIONS IN THE NERVOUS SYSTEM ......................................... 1
1.2 DEVELOPMENT OF MOTOR NEURON DIVERSITY IN THE VAGAL BRAINSTEM ............. 3
1.3 DEVELOPMENT OF TRANSCRIPTOMIC DIVERSITY IN CORTICAL PROJECTION
NEURONS ................................................................................................................................... 16
1.4 SINGLE CELL SPATIAL TRANSCRIPTOMICS METHODS FOR MULTIMODAL
INVESTIGATION OF CELL TYPE DIVERSITY IN THE NERVOUS SYSTEM ............................ 23
CHAPTER 2 : MET RECEPTOR TYROSINE KINASE REGULATES LIFESPAN ULTRASONIC
VOCALIZATION AND VAGAL MOTOR NEURON DEVELOPMENT ...................................................... 25
2.1 ABSTRACT ........................................................................................................................... 25
2.2 INTRODUCTION .................................................................................................................... 26
2.3 MATERIALS AND METHODS ............................................................................................... 28
2.4 RESULTS .............................................................................................................................. 38
2.5 DISCUSSION ......................................................................................................................... 63
2.6 SUPPLEMENTARY MATERIALS ......................................................................................... 69
ix
CHAPTER 3 : SINGLE CELL GENE EXPRESSION ANALYSIS OF MET+ EXCITATORY
PROJECTION NEURONS IN THE DEVELOPING CORTEX .................................................................. 79
3.1 ABSTRACT ........................................................................................................................... 79
3.2 INTRODUCTION .................................................................................................................... 79
3.3 MATERIALS AND METHODS ............................................................................................... 82
3.4 RESULTS .............................................................................................................................. 87
3.5 DISCUSSION ....................................................................................................................... 109
CHAPTER 4 : SCAMPR, A SINGLE-CELL AUTOMATED MULTIPLEX PIPELINE FOR RNA
QUANTIFICATION AND SPATIAL MAPPING ....................................................................................... 117
4.1 ABSTRACT ......................................................................................................................... 117
4.2 INTRODUCTION .................................................................................................................. 117
4.3 MATERIALS AND METHODS ............................................................................................. 119
4.4 RESULTS ............................................................................................................................ 132
4.5 DISCUSSION ....................................................................................................................... 158
4.6 STUDY LIMITATIONS ......................................................................................................... 162
4.7 SUPPLEMENTARY MATERIALS ....................................................................................... 163
CHAPTER 5 : CONCLUDING REMARKS ............................................................................................. 173
REFERENCES ....................................................................................................................................... 177
APPENDIX.............................................................................................................................................. 208
x
List of Figures
Figure 2.1: Deletion of MET results in severely impaired ultrasonic vocal production and
syllable repertoire early postnatally…………………………………………………
43
Figure 2.2: Sustained vocalization deficits in Met cKO adult mice…………………………… 48
Figure 2.3: Axonal quantification and innervation status of αBT labeled AChR clusters in
laryngeal muscles following MET deletion…………………………………………
52
Figure 2.4: Topology of MET-EGFP expression and motor neuron loss in the nAmb
following conditional deletion of Met………………………………………………..
55
Figure 2.5: Analyses of lifespan loss of vagal motor neurons in the nAmb following
conditional deletion of MET………………………………………………………….
58
Figure 2.6: No difference in the number of neurons in the NTS or DMV is observed
following conditional deletion of MET………………………………………………
62
Figure S2.1: Conditional deletion of Met results in early embryonic loss of MET protein in
vagal motor neuron axons…………………………………………………………...
69
Figure S2.2: Early Cre-mediated recombination driven by the Phox2b promoter in the
developing brainstem………………………………………………………………...
71
Figure S2.3: Mechanical stimulation assay to directly elicit calls………………………………. 72
Figure S2.4: Respiratory pattern during vocalization……………………………………………. 73
Figure S2.5: Body weight does not correlate with vocalization disturbances observed
following MET deletion……………………………………………………………….
74
Figure S2.6: Reduction of nAmb neurons in cKO mice is not different between sexes…… 75
Figure S2.7: Limited colocalization of MET and Phox2b outside of the vagal nuclei………… 76
Figure 3.1: P8 visual and frontal cortex cells cluster into classically defined neural cell
types……………………………………………………………………………….......
89
Figure 3.2: Met is enriched in neurons and Hgf is enriched in astrocytes in both regions… 92
xi
Figure 3.3: Met is enriched in subsets of superficial and deep layer projection neurons in
each region…………………………………………………………………………....
95
Figure 3.4: Met+ and Met- projection neurons in the superficial layers of the VC are
transcriptomically different………………………………………………………......
99
Figure 3.5: Met+ and Met- projection neurons in the superficial layers of the FC are
transcriptomically different…………………………………………………………...
102
Figure 3.6: DEGs in superficial Met+ PNs are involved in neurodevelopmental and
neuronal activity processes………………………………………………………….
106
Figure 3.7: DEGs in superficial Met+ PNs are involved in neurodevelopmental and lipid
metabolism processes……………………………………………………………….
108
Figure 4.1: Schematic of SCAMPR pipeline……………………………………………………. 136
Figure 4.2: Combined IHC and HiPlex RNAscope allows for high-resolution, single-cell
mRNA expression in the central and peripheral nervous system……………….
140
Figure 4.3: Automated segmentation of cytoplasm of single cells using Cellpose…………. 144
Figure 4.4: Analysis and spatial visualization of gene expression and co-expression of
HiPlex RNAscope…………………………………………………………………….
150
Figure 4.5: High-dimensional clustering analysis of HiPlex RNAscope data……………….. 152
Figure 4.6: Utilization of HiPlex RNAscope for comparison of gene expression at single-
cell resolution in nodose ganglion neurons from CAU and ELS experimental
groups…………………………………………………………………………………..
154
Figure S4.1: Semiautomated thresholding methods using individual image intensities……... 165
Figure S4.2: Quantitative comparison of gene expression between nuclear and cytoplasmic
segmentation…………………………………………………………………………..
166
Figure S4.3: Pairwise gene co-expression comparisons between CAU and ELS in nodose
ganglion………………………………………………………………………………...
167
Figure S4.4: Comparison of gene expression between CAU and ELS nodose ganglia
neurons…………………………………………………………………………………
169
Figure S4.5: Accurate segmentation on maximum intensity projection images………………. 171
xii
Figure 5.1: MN loss in E16.5 nAmb after cKO of Met………………………………………….. 179
Figure 5.2: Changes in Met+ nAmb MNs NTFR expression pattern with cKO of Met……... 181
Figure 5.3: Changes in Gfra1 expression patterns in Met+ nAmb MNs after cKO of Met…. 183
1
CHAPTER 1: General Introduction
CONTRIBUTIONS OF CELL TYPE DIVERSITY AND GENETIC PLEIOTROPY TO
STRUCTURE/FUNCTION RELATIONS IN THE NERVOUS SYSTEM
The brain is a complex organ made up of distinct anatomical regions (structure)
that connect via local and distant circuits to process sensory information and elicit
behaviors (function) (L. W. Swanson, 2000; Tau & Peterson, 2009). The establishment of
structure-function relations in the vertebrate nervous system results from early
polarization of the embryo at the blastula/gastrula stages, followed by segmentation into
5 subdomains (forebrain, interbrain, midbrain, hindbrain, and spinal cord), and completed
by further segmentation of each subdomain into progenitor regions (Kudoh et al., 2002;
L. Swanson, 2012). These transient progenitor domains give rise to heterogenous cell
types that migrate and organize into anatomically and transcriptomically distinguishable
structures. Connections between different structures are established as neurons within
each structure undergo axon pathfinding, synaptogenesis, and synaptic pruning,
ultimately forming the functional circuits and subsystems that drive voluntary and intrinsic
behaviors (L. W. Swanson et al., 2022; Tau & Peterson, 2009).
Due to their accessibility and highly developed arsenal of molecular and genetic
tools, mice have been an excellent model for studying the role of these developmental
processes on structure-function organization. In mice, the formation of functional circuits
involves the development of ~111 million heterogenous neural cell types (72 million
neurons and 39 million glia), with different morphological, electrophysiological, molecular,
connectivity, and functional characteristics (Bota & Swanson, 2007; Cossart & Garel,
2
2022; Erö et al., 2018a; Zeng & Sanes, 2017). The developmental processes that form
these different characteristics are driven by tightly-controlled temporal expression
patterns of the 22,018 protein-coding genes that make up the mouse genome (Breschi et
al., 2017). This genetic blueprint is similar in quantity to the 19,099 protein-coding genes
found in the much more primitive and functionally restricted C. Elegans, which only has
352 neural cells (302 neurons and 50 glial cells) (Shaham, 2006; The C. elegans
Sequencing Consortium*, 1998). Genetic pleiotropy, the ability of one gene to effect
multiple phenotypes, expands the genetic repertoire in mice and allows for the
development of highly complex cellular and behavioral diversity.
My dissertation work has focused on the role of the Met gene on cell type diversity
in the nervous system (Eagleson et al., 2017a). Met codes for a highly pleiotropic
transmembrane receptor tyrosine kinase (Met), and interacts with its only known ligand,
hepatocyte growth factor (Hgf), to drive multiple developmental processes through the
Erk, MAPK, PI3k/Akt, and Ras intercellular signaling pathways (Eagleson et al., 2016a;
Ebens et al., 1996). Met is expressed in heterogenous structures and neuronal subsets
within these structures, including in cortical projection neurons (PNs) across the
neocortex, spinal and cranial motor neurons (MNs), small subsets of brainstem
interneurons, pyramidal cells in the hippocampus, myenteric neurons, retinal ganglion
cells (RGCs), sensory neurons of the dorsal root ganglion (DRG), and serotonergic cells
of the caudal dorsal raphe. It has numerous known developmental roles across these
structures. For example, Met is known to promote dendrite growth and synaptic
arborization in cortical projection neurons, modulate synapse formation and maturation in
3
hippocampal pyramidal neurons, and be required for cell survival and neurite outgrowth
in spinal MNs, RGCs, and DRG sensory neurons (Avetisyan et al., 2015; Eagleson et al.,
2011, 2016a; Heun-Johnson & Levitt, 2016; Judson et al., 2010b; Kast, Wu, Williams, et
al., 2017; F. Lamballe et al., 2011; Lanjewar et al., 2023; F Maina et al., 1997; Tönges et
al., 2011).
The work outlined in the chapters below investigate the role of Met in the
specialization and functional specification of vagal motor neurons, a highly heterogenous
and functionally important group of neurons that act as one of the primary communication
pathways between the gut/periphery and the brain. I discuss results from experiments
that trancriptomically profiled diverse Met-expressing PN populations from distinct
structures of the developing cortex. And finally, I present a new analytical pipeline for
simultaneously investigating transcriptomic heterogeneity and structural topography in
the developing mouse nervous system. These results add to our knowledge about how
genetic pleiotropy and cellular heterogeneity play overlapping roles to drive the function
outputs of multiple nervous system structures.
DEVELOPMENT OF MOTOR NEURON DIVERSITY IN THE VAGAL BRAINSTEM
Vagal circuits mediate critical peripheral functions, yet developmental studies of
genes that influence vagal circuit formation and function are very limited. This is
particularly true for the two brainstem vagal motor nuclei, the nucleus ambiguus (nAmb)
and dorsal motor nucleus of the vagus (DMV), which control numerous important
4
functions through their innervation of different peripheral organs. The nAmb innervate
supradiaphramatic peripheral organs such as the larynx, pharynx, esophagus, heart, and
controls vocal communication, respiration, cardiac function, and food ingestion, while the
DMV innervates the stomach and gastrointestinal tract and is involved in ingestion,
digestion, and metabolism (Bieger & Hopkins, 1987). Like spinal motor neurons (MNs),
MNs in the nAmb and DMV are topologically organized into distinct functional pools during
early development, when genes involved in neurogenesis, cell migration, nucleogenesis,
cell survival, and axon guidance promote MNs to specialize into differentially projecting
subtypes (Barsh et al., 2017a; Caubit et al., 2010; Guthrie, 2007b). However, compared
to their spinal MN counterparts, there is a significant knowledge gap in understanding the
developmental mechanisms that drive the diversity and functional specialization of vagal
MNs in these two nuclei. We believe that Met-signaling plays an important role in the
developmental diversification of vagal MNs.
The early segmentation of vertebrate embryos along the anterior-posterior (AP)
and dorsal ventral (DV) axes is imperative for the generation of diverse cell types and
their organization into distinct structures with distinct functions. The generation of distinct
functional structures in the hindbrain, including the two vagal motor nuclei, requires
neurons to be born from the correct progenitor pool, migration of these neurons into the
correct location, fusion into distinct groups, and projections into the correct targets. These
processes are driven by transcription factors, cell surface proteins, extracellular matrix
proteins, and early spontaneous activity.
5
The neurogenic niches from with vagal and other brainstem MNs arise result from
an AP segmentation in the hindbrain that forms several discrete, transient, compartments
called rhombomeres (Lumsden & Keynes, 1989; Lumsden & Krumlaf, 1996; Tomás-Roca
et al., 2016). Hindbrain segmentation plays an integral role in neuronal differentiation. The
rhombomeres provide specialized molecular environments for different progenitor pools
and newborn neurons by mechanically separating them from one-another via their
rhombomere boundaries, helping ensure a diversity of cell types (Calzolari et al., 2014).
It is generally accepted that the mouse hindbrain is comprised of 8 serial rhombomeres
(r1-r8), with r1 at the midbrain-hindbrain boundary and r8 bordering the spinal cord in the
caudal portions of the brainstem. Some groups have argued that the caudal
rhombomeres can be further divided into 5 “crypto-rhombomeres” based on transcription
factor expression patterns, bringing the total to 11 (Tomás-Roca et al., 2016).
In mice, the branchiomotor (BM) of the nAmb and visceromotor (VM) motor
neurons of the DMV are born in the basal plate of rhombomeres 6, 7, and 8 on embryonic
(E) days 9.5 and 10.5 and exhibit a rostral to caudal spatiotemporal pattern of birth
(Gilland & Baker, 2005; Jarrar et al., 2015; Pierce, 1973). That is to say, prospective
nAmb BM MNs are born first in the rostral portions r7/r8 and prospective DMV VM MNs
are born soon after in the more caudal portions of this proliferative zone (Caton et al.,
2000; Lumsden & Keynes, 1989; Mark et al., 1993). The non-vagal, glossopharyngeal
BM MNs of the nAmb are born more rostrally in r6 (Caton et al., 2000).
6
AP segmentation of the vertebrate embryonic brainstem into distinct rhombomeres
is required for motor neuron diversity. This segmentation is driven by the expression of
diffusible morphogens such as fibroblast growth factors (FGFs, Fgf8/Fgf3 in the
brainstem), and retinoic acid (RA), which interact with one another to form the initial
polarization of the hindbrain. There is a gradient of RA expression in the brainstem, which
high levels caudally and low levels rostrally. This gradient is maintained by caudal FGF
expression, which inhibits one of the key RA metabolizing enzymes. This RA gradient
drives the expression of unique transcription factor combinations in each rhombomere,
which in turn drive the proliferation and differentiation of distinct populations of vagal
motor neurons (Guthrie, 2007b). Disruption in morphogen signaling can disrupt the
rhombomere formation and neuronal differentiation. For example, reductions in RA
signaling in the posterior brainstem leads to a loss of the caudal rhombomeres (r4-r7) and
expansion of anterior rhombomeres (r1-3) towards the hindbrain-spinal cord boundary.
Overexpression of RA in the posterior hindbrain has the opposite effect, with posterior
rhombomeres expanding towards the anterior end of the hindbrain (Gavalas & Krumlauf,
2000; Marshall et al., 1992). In zebrafish, disruptions in RA expression elicit changes in
vagal MN number during early development. In mature vagal MNs, loss of RA expression
leads to deficits in axon guidance, possibly through RA’s regulation of Met expression in
vagal MNs and Hgf expression in the vagal MN targets (pharyngeal arches) (Isabella et
al., 2020; Linville et al., 2004).
These phenotypes are partially mediated by the transcriptional regulation of the
HOX and Three-Amino acid Loop Extension (TALE) families of homeodomain
7
transcription factors by the above morphogens (Isabella et al., 2020; Linville et al., 2004).
First discovered in invertebrates (Drosophila melanogaster), the Hox family of
transcription factors were shown to play important roles in whole body segmentation in
fruit flies. In vertebrates, there are approximately 39 Hox genes which are split into 4
groups of 13 paralogs (Hox1-Hox13, a-d), and which are initially expressed in a nested
fashion along the AP axis of the hindbrain. As the embryo develops the expression of
different Hox genes become more limited and start localizing to specific rhombomeres,
helping create constrained neurogenic niches that will give rise to the diversity of cells
(Frank & Sela-Donenfeld, 2019; Krumlauf, 2016). The role of Hox genes in rhombomere
formation have been demonstrated in KO studies of paralog groups 1-7, where loss of
Hox expression results in hindbrain patterning and rhombomere formation/maintenance
abnormalities (Barsh et al., 2017b; Frank & Sela-Donenfeld, 2019; McNulty et al., 2005;
Tomás-Roca et al., 2016).
The role of Hox expression in the caudal vagal motor neuron proliferative region
(r7/8) have been minimally investigated. Histological methods in mice demonstrate that
r7/r8 express the Hox1-Hox5 paralogs, with the highest expression levels found for Hox4
(Dasen & Jessell, 2009). A recent paper in zebrafish showed that differential expression
of Hox5 within r8 was important for the targeting and temporal patterning of vagal motor
axons to pharyngeal arches 4 and 5. Using in vivo time-lapse imaging, they demonstrate
that the more rostral vagal motor neurons produce axons first and project to pharyngeal
arch 4, while the more caudal neurons produce axons after their rostral counterparts and
project to the more caudal pharyngeal arch 5. They show that Hox5 regulates the
8
temporal patterning of axonogenesis by regulating the AP location of the motor neuron
cell bodies (Barsh et al., 2017b). The developmental functions of Hox1 and Hox2
expression in vagal MNs has also been demonstrated in mice. KO of Hoxa1 results in a
shift of r7/8 towards the rostral end of the embryo (Carpenter et al., 1993). As Hoxa1 is
expressed in the caudal rhombencephalon prior to and during motor neuron proliferation,
it likely plays a role in the specification and differentiation of vagal motor neurons and
other neuronal populations that originate in this region. When Hoxa1/b1 are both
disrupted, there is a loss of nerve fibers in the vagus nerve and a loss MNs and aberrant
patterning of other cranial nerves (Barsh et al., 2017a; D’Elia & Dasen, 2018; Philippidou
& Dasen, 2013). These studies demonstrate the importance of Hox genes in rhombomere
formation and their downstream effects of vagal motor neuron projection patterning.
Further refinement of rhombomere boundary formation and maintenance depend
on the transcriptional regulation of cell surface molecules by location-specific non-
homeodomain transcription factors. These transcription factors, which are all activated by
Fgf3, include Krox20 and Kreisler. They regulate the expression of Hox paralogs in a
location-specific manner and interact with each other to increase the expression of cell
surface molecules in the Eph/ephrin family, which are integral for the formation of
rhombomere boundaries. Transcriptional activation of Eph4a by Korx20 and Kreisler and
repression of its ligand ephrin2a by Kriesler lead to the expression of the receptor and its
ligand in alternating rhombomeres. Studies using antisense morpholino knock down
strategies and ectopic transplantation studies to disrupt proper expression of
Eph4a/ephrin2a show deficits in boundary formation (“leaky” boundaries) and cell sorting
9
(ectopic cells) at the rhombomere boundaries. These studies show that transcription
factor mediated Eph/Ephrin interactions are required for proper hindbrain segmentation
and boundary formation (J. Cooke et al., 2001; J. E. Cooke et al., 2005; Terriente &
Pujades, 2015). The roles of ephrin signaling and rhombomere boundary formation on
vagal MNs has not been directly investigated.
Dorsalventral polarization of the brainstem is also driven by morphogen gradients
and is important for MN differentiation. Sonic hedgehog (Shh) gradients repress Class I
transcription factors (Pax6) and induce expression of Class II transcription factors (Nkx6.1
and Nkx6.2) ventrally near the basal plate of the r7/r8 neural tube, giving the progenitors
in this proliferative region BM and VM neural fates (Ericson et al., 1997). This contrasts
from the non-vagal somatic motor neuron population which are born more dorsally (Jarrar
et al., 2015; Puelles et al., 2019). Disruption in this signaling can lead to patterning
deficits. For example, dorsal expansion of the Nkx domain in Pax6 mutants leads to the
dorsalateral expansion of vagal motor neuron pools and the loss of the hypoglossal
nucleus (Ericson et al., 1997). Further DV spatial organization prior to neurogenesis takes
places through differential expression of Hox transcription factors, which divide the MN
progenitor regions (alar and basal plates) into microzones with unique properties. These
microzones have roles in the proliferation, migration, and axon guidance patterns for cells
that are born in or tangentially migrate to the alar and basal plates (Puelles et al., 2019;
Tomás-Roca et al., 2016). Once the neurons are born, they undergo further specification
through the activity of LIM genes (Isl1/2), Tbx20, and Phox2b. These secondary
transcription factors are regulated by Hox genes rostrocaudally and Class I/II TFs
10
dorsoventrally. The combinations of these secondary TF’s are needed to differentiate the
BM and VM vagal motor neurons from their dorsal, non-vagal somatic motor neurons
(Guthrie, 2007b).
Once the MNs are born, they start projecting axons towards their prospective
targets and undergo nuclear translocation to migrate into the appropriate locations. Both
developmental processes are important for MN specialization; distinct cell populations
that are born in the same rhombomere region anatomically segregate during migration
and gain functional specificity through innervation of the appropriate targets. There is
recent evidence showing that once nAmb BM MNs are born they undergo and initial
dorsal migration to the alar plate (Caubit et al., 2010; Puelles et al., 2019). During this
time the neurons polarize by extending apical dendrites towards the fourth ventricle and
axons radially through the mantle and marginal layers towards their exit points. Over the
next few days, the cell bodies will follow their axons and either migrate a short distance
away from the ventricle to form the DMV, or a longer distance radially to form the nAmb
(Guthrie, 2007b; Rinaman & Levitt, 1993). This secondary migration by the vagal BM
neurons results from a change in cadherin expression once the neurons reach the alar
plate (Astick et al., 2014; Ju et al., 2004).
Initial BM and VM motor neuron axon guidance is dependent on Slit-Robo
signaling. Transient expression of Slit1 and Slit2 in r7/r8 repel the axon growth cones of
newly born vagal BM and VM neurons away from the midline and floor plate, towards
their dorsal exit points (Guthrie, 2007b; Hammond et al., 2005). Once they are repelled
11
from the ventral midline, other molecular cues likely take over to guide the axons to the
appropriate exit point in the ventrolateral portion or r7. Here the neurons exit as rootlets
and fasciculate with each other and with the dorsolateral sensory rootlets to form the
vagus nerve (Caton et al., 2000; Chase & Ranson, 1914; Gilland & Baker, 2005). How
these neurons get to this exit point is still unclear.
One possibility is that Slit-mediated repulsion from the midline and from the alar
plate drive the axons laterally to their correct exit point location (Hammond et al., 2005).
Another possibility is axons are guided to the exit point through attractive guidance cues
emitted by structures near the exit point. Early embryonic studies in chicks and mice have
shown that brainstem-adjacent sensory ganglia provide both axon guidance and nuclear
migration cues to cranial motor neurons, but this finding has yet to be replicated in other
vertebrate species (Caton et al., 2000; Moody & Heaton, 1983). This is an attractive
theory as the jugular ganglion, one of the sensory ganglia of the vagus nerve that both
the VM and BM vagal motor axons pass through, is right next to the r7/r8 exit point of the
mouse brainstem during early development. It has been shown that Hgf, which is
expressed in the trigeminal ganglion, the jugular ganglion, and the brachial arch muscle
plate in early developing mouse/chick embryos, plays an attractive role for Met-
expressing BM/VM motor neurons. Knockout of Hgf leads to aberrant development of the
hypoglossal nerve, which is partially made up of Met-positive BM motor neurons. The
vagus nerve looked normal in whole-mount neurofilament stains of these Hgf-KO
embryos, but this could have been due to a limitation in their methods. More intricate
12
deficits in development of vagal axons after a loss of Met-Hgf signaling would have been
difficult to observe in their whole-mount staining methods (Caton et al., 2000).
Once neurons reach their correct brainstem locations, they coalesce to form
functionally diverse nuclei. The topographic stabilization of the neurons at their terminal
locations requires the expression of type II cadherin family cell adhesion molecules.
Experiments in chicks have shown that the cranial nuclei originating from r8 (dorsal and
ventral hypoglossal, nAmb, and DMV) all express unique combinations of type II
cadherins. Early cadherins (Cdh20) are expressed ubiquitously in all motor neurons at
the time of birth and as they migrate to their locations, where the emergence of late,
location-specific cadherins stabilize nuclear topography. Disruptions in early cadherin
function through the unilateral electroporation of a dominant negative isoform led to
abnormalities in nucleogenesis, with all r8 nuclei exhibiting more dispersed patterning
when compared to the non-electroporated control side counterparts. General
overexpression of Cdh20 leads to comingling of these normally segregated nuclei (Astick
et al., 2014).
A more recent study shows that spontaneous activity in early cranial motor
neurons, mediated by gap junction-forming connexin proteins, is required for proper
nucleogenesis and nuclear stabilization. Using live in vivo imaging, the authors
demonstrated that both the blocking of T-type Ca+ channels and disruptions in cadherin
expression led to a decrease in spontaneous activity and deficits in nucleogenesis (diffuse
nuclei) (Montague et al., 2017). These studies together show a role for cell adhesion
13
molecules (cadherins), gap junction molecules (connexins), and spontaneous activity in
the formation and consolidation of cranial nuclei.
In summary, vagal MN diversity depends on early AP and DV segmentation and
boundary formation in the brainstem by morphogens, transcription factor cascades, and
cell surface proteins. This early patterning sets up transcriptional compartments that
express specific combinations of cell adhesion molecules, mitogens, and transmembrane
receptors to drive cell migration, axon pathfinding, and nucleogenesis within
compartments. These processes are required for the differentiation, survival, migration,
and projection patterns vagal MNs. Further work is needed to understand how the diverse
functional cell types develop within each vagal motor nucleus.
To this end, we have studied the expression patterns and functional roles of Met
in these nuclei to try gain insight into the development of functional diversity in these MNs.
Our lab has shown that Met is differentially expressed across the two vagal motor nuclei,
with only a very small subsets of DMV being Met+ but over 1/3 of nAmb expressing the
gene (A. Kamitakahara et al., 2017). These findings suggest two things; 1) Met signaling
plays a more extensive developmental role in nAmb MNs compared to those in the DMV,
and 2) Met is involved in the development of MN diversity within the nAmb, as it is only
expressed in a subset of nAmb MNs.
MN diversity within the nAmb is of great functional importance. The nAmb provides
motor input to the larynx, pharynx, and esophagus, and coordinates muscle contractions
14
within these three peripheral organs during different functional outputs. nAmb input into
the pharyngeal muscles control the epiglottis to direct food towards the esophagus during
feeding or air through the larynx during respiration and/or vocalization. Innervation of the
striated muscles of upper esophageal sphincter and the more caudal esophageal
peristaltic smooth muscles move food into the stomach. nAmb innervation of the laryngeal
muscles control the position of the vocal cords to protect the airways during swallowing
or to manipulate air during expiration to produce vocalization (Haney et al., 2020;
Heckman et al., 2016a; Pitts & Iceman, 2022; Portfors, 2007). Any deficits in coordination
between or activity in one of these organs could lead food/water aspiration during
swallowing, esophageal air swallow, or deficits in vocalizations and/or respiration, and
can dramatically affect the health outcomes of the organism.
This functional synchrony is coordinated by inhibition or activation of
viscerotopically segregated pools of nAmb MNs that innervate the musculature in these
three organs (Bieger & Hopkins, 1987). Tracing studies in rats show that cell bodies of
MNs that innervate the esophagus are grouped together in the rostral nAmb (compact
formation), caudal to these are the MNs innervating the pharynx (semi-compact
formation), and in the most caudal portion of the nAmb (loose formation) are MNs
innervating different laryngeal muscle groups, which are segregated into mostly non-
overlapping populations depending on the laryngeal muscle that they innervate
(Hernández-Morato et al., 2014; Hydman & Mattsson, 2008). As discussed above, the
specialization and topological organization of diverse pools of MNs that innervate and
control the distinct muscle groups takes place during early development, and is controlled
15
by morphogens, transcription factors, cell adhesion molecules, and environment-sensing
transmembrane receptors like Met.
Met is expressed right after birth in nAmb MNs, when they have started undergoing
migration and have already started projecting axons towards their terminal targets. The
early targets of nAmb MNs are the transient Hgf-expressing caudal pharyngeal arches
and the surrounding mesodermal areas, which will ultimately develop into the cartilage,
epithelium and musculature of the larynx, pharynx, and upper esophagus (Ebens et al.,
1996; A. Kamitakahara et al., 2017; A. K. Kamitakahara et al., 2021; Wu & Levitt, 2013b).
Met-Hgf signaling has been shown to induce axon grown in in vitro tissue culture
experiments of the developing brainstem. Inhibition of Met-Hgf interaction using
neutralizing antibodies reverses this axonal growth, demonstrating the importance of Met
signaling for brainstem MN axonogenesis (Caton et al., 2000). A second developmental
role for Met-Hgf signaling has also been demonstrated in spinal MNs, where it has been
shown to act as a neurotrophic factor receptor (NTFR) to promote MN survival in culture
(Ebens et al., 1996; Schaller et al., 2017a). Interestingly, deletion of Lifr or Il6st (gp130),
two other NTFRs, lead to a 34-50-% loss of MNS in the nAmb (M. Li et al., 1995;
Nakashima et al., 1999a). Further evidence of a pleiotropic role for Met in MNs can be
found in in vivo spinal MN studies, where Met signaling promotes cell survival in some
MN pools and axon growth and guidance in others (Ebens et al., 1996; Fabienne
Lamballe et al., 2011).
16
In chapter 2, we investigate whether Met had a similar pleiotropic role in the
development of nAmb MNs. We use a conditional KO model in mice to investigate the
biological consequences of the loss of Met signaling during MN development. We also
assess functional changes in these mice through analysis of weight gain
(esophageal/pharyngeal function) or ultrasonic vocalizations (laryngeal function).
DEVELOPMENT OF THE TRANSCRIPTOMIC DIVERSITY IN CORTICAL PROJECTION NEURONS.
A major goal in the field of neuroscience has been to understand how functional
diversity is generated in the cortex. The cortex, which develops from the posterior
portions of the neural tube, can be broken up into distinct structures based on their
cytoarchitectures, connectivity patterns, and functional outputs (Cadwell et al., 2019).
These cortical structures are broadly categorized as primary sensory areas which
integrate and process internal and external sensory information, motor areas which
coordinate motor decision, movements, and motor learning, and association areas which
integrate information from other cortical and subcortical regions to generate and regulate
higher level, goal-directed behaviors (Cadwell et al., 2019; Heindorf et al., 2018; Le Merre
et al., 2021; Lee et al., 2022; Marshel et al., 2011). The areal and laminar architecture of
mammalian cortex, the diversity of projection neuron types within each region and layer,
and the highly complex projection patterns that connect each cortical area with the rest
of the brain help drive the myriad functional outputs of this region. Cellular diversity and
its contribution to structure-function relations in the cortex is set up throughout
development. This starts with the early generation of neuronal progenitor pools and
17
neurogenesis, continues as neurons undergo migration and axon growth/guidance, and
develops further at postnatal timepoints when already formed circuits undergo
synaptogenesis, axon and dendrite elaboration, and circuit maturation via synaptic
pruning (Greig et al., 2013; Jain & Zipursky, 2023; Levitt et al., 2003; Molyneaux et al.,
2007).
As with other brain regions, there is a strong link between tissue structure and cell
type composition in the cortex. Investigation of cytoarchitecture has been combined with
molecular and connectivity profiling to split the mouse cortex into four superficial (1, 2, 3,
and 4) and 2 deep layers (5a/5b, 6a/6b), with a notable variation of laminar topography in
agranular anterior cortical areas which lack the tightly packed, small cell bodies (granular)
that constitute layer 4 (Angevine & Sidman, 1961; Rakic, 1972, 1974). The PNs that serve
as the that connect each cortical region with other cortical and subcortical brain regions
are classified based on their molecular signatures and connectivity patterns and have
distinguishable topographic organizations across the different cortical layers.
Intratelencephalic (IT) PNs, which make up the vast majority of PNs in layers 2/3,
innervate distal cortical areas as well as subcortical areas like the striatum, nucleus
accumbens, and amygdala and. IT PNs can also be found in layers 5 and 6 but at lower
percentages, making the most diverse class of PNs based on projection patterns and
laminar location. Subcerebral (SC) PNs are in layers 5 and 6, and project to areas outside
of the telencephalon, such as the hindbrain, and spinal cord. Corticothalamic (CT) PNs,
which innervate the different thalamic nuclei, are primarily located in layer 6. Layer 4,
which is the main recipient of sensory information from the thalamus, contain mostly
18
locally-projecting IT PNs that innervates cells in layers 2/3 in the same cortical column.
Recently, single-cell RNA sequencing and high-throughput sequencing studies have
demonstrated that some of these layers can be further split into molecularly and
anatomically distinct sublayers (Cheng et al., 2022a; Muñoz-Castañeda et al., 2021).
Additionally, it has been demonstrated that some of the anterior agranular cortical areas
exhibit both the molecular and cellular markers consistent with the presence of a layer 4,
and that neurons within layer 4 of the auditory cortex send distal projections to the striatum
(Bertero et al., 2022; Muñoz-Castañeda et al., 2021; Shepherd, 2009). The discoveries
of new cell types and sublayers demonstrate that our understanding of cortical
organization and cortical neuron heterogeneity are continually being refined.
Investigation of the transient developmental processes that lead to the production and
maturation of diverse cell types and their migration to and innervation of appropriate
regions will contribute to this refinement and will help us understand the functional
architecture of the cortex.
Genes coding for transcription factors and diffusible proteins such as Pax6,
Couptf1, Sp8, and Fgf8 are expressed in gradients across the anterior-posterior and
medial-lateral extent of the germinal zones of the cortex. These early compartmentalized
gene expression patterns serve as a protomap for generating diverse PNs across cortical
structures that integrate into distinct circuits (Kast & Levitt, 2019; Molnár et al., 2019;
Rakic, 1988; Rakic et al., 2009). Similarly, a mixture of encoded genetic program and
temporal dynamics drive the laminar organization of neurons within each cortical region.
It has long been established that the heterogenous PNs comprising the different cortical
19
layers are born in waves in the ventricular and subventricular zones and generate the
cortical column in an inside-out manner. The prospective deep layer PNs are the first to
be born and migrate towards the cortical plate and are displaced towards the ventricle by
later waves of prospective superficial layer PNs, helping establish the laminar architecture
of the cortex (Cooper, 2008; Greig et al., 2013; Rakic, 1972). The genetic programs
involved in laminar identity include transcription factors that can be expressed pre- and
post-mitotically and which have known cross-regulatory interactions. They include the
layer 2/3/4 transcription factors Cux1 and Cux2, layer 5 transcription factors like Bclllb
and Fezf2, and layer 6 transcription factors like Tbr1 and Sox5 (B. Chen et al., 2008; Han
et al., 2011; Iulianella et al., 2008; Kwan et al., 2008). The laminarly segregated
expression of these and other transcription factors regulate the expression of
environment-sensing receptors and cell-cell interaction molecules, ultimately effecting
early processes like migration which are important for laminar organization, and later
processes like axon guidance and targeting which establish neuronal projection subclass
identity (Arlotta et al., 2005; Gil-Sanz et al., 2015; Hevner et al., 2001; Leone et al., 2015;
McEvilly et al., 2002; Sugitani et al., 2002).
At early developmental timepoints, the diversity of PNs and their projection
patterns across each layer could arise in several ways. One possibility is that molecularly
diverse pools of fate restricted progenitor cells give rise to the diverse subclasses of PNs.
A second possibility is that PN diversity arises from multipotent progenitor populations
that get progressively fate restricted over developmental time. A third possibility is that
the diversification takes place post-mitotically as neurons migrate and receive time-
20
dependent signals from their environments, and a fourth possibility is some kind of
combination the three aforementioned theories (Mukhtar et al., 2022; Park et al., 2022).
Transplantation studies provide some insight into this process. Prospective superficial
layer progenitor cells specify into deep layer projection neurons when heterochronically
transplanted into the developing cortex at early timepoints. Conversely, prospective deep
layer progenitors are inflexible and retain their deep layer identities when transplanted at
a time corresponding to superficial layer neurogenesis (Desai & McConnell, 2000). These
findings demonstrate a timing-dependent potency for different progenitor cell populations
in the developing cortex. Other histological and fate mapping studies have identified
several PN subtype markers (Fezf2, Cux1, and Cux2) in progenitor cell populations, and
have shown that their expression levels correspond in neurogenesis of the PN subtypes
in which they are expressed (Franco et al., 2012; Molyneaux et al., 2005; Nieto et al.,
2004; Zimmer et al., 2004). These studies provide evidence for the existence of fate-
restricted progenitor pools. Recent transcriptomic studies have allowed researchers to
molecularly or computationally follow and molecularly profile precursor cells or newborn
neurons over developmental time, leading to some insights into these processes. Two
studies show molecular homogeneity in precursor cell populations in early periods prior
to and during the first few days of neurogenesis, and do not find any molecular signals in
these progenitor populations that would match future PN identity. Both studies also
demonstrate that transcriptional differences in cell populations start after neurogenesis,
with different subsets of cells expressing known gene markers and modules for PN
identity (Di Bella et al., 2021; Telley et al., 2019). One of these studies demonstrates that
subsets of neuronal precursors start to differ molecularly by upregulating their expression
21
of membrane receptors and cell-cell signaling proteins, going from an “introverted” to an
“extroverted” state from E12.5 to E15.5. These studies combined suggest that PN
diversification takes place post-mitotically. Expanding on prior histological findings, a
more recent study found cortical layer and PN marker gene expression in progenitor cell
populations, suggesting possible fate restriction in progenitor pools prior to neurogenesis
(Mukhtar et al., 2022). More studies that combine fate mapping and transcriptomics can
help determine the early mechanisms for diversification in the cortex.
There is no evidence of Met playing a role in cortical neuron diversification at
timepoints corresponding to neurogenesis. In situ hybridization and western blot
experiments first detect Met mRNA and protein at the very late stages of cortical
neurogenesis (E15.5 and 16.5), and a recent sequencing experiment was able to localize
this early Met transcript expression to both cortical progenitor cells and newborn neurons
at E15.5. Importantly, no abnormalities in lamination or cortical histology were noted after
conditional deletion (Emx1-Cre) of Met in cortical progenitor cells and neurons starting in
early cortical development (E10.5) (Chan et al., 2001; Judson et al., 2009; Mukhtar et al.,
2022). More detailed studies can help identify whether there are more subtle differences
in superficial PN diversity or development due to a loss of Met signaling.
Met signaling has a more established role in cortical neuron diversification at early
postnatal developmental periods, when PN undertake axonal guidance to project to their
targets, and synaptogenesis and synapse maturation to integrate into a functional circuit.
At the protein and transcript level Met expression is present at low levels during the late
22
embryonic periods that correspond with upper layer PN neurogenesis. Its expression
levels ramp up after birth and peaks around the first postnatal week, when PNs are
undergoing dendritic arborization and synaptogenesis, and starts going back down during
periods corresponding to synaptic maturation and pruning (Eagleson et al., 2016b;
Judson et al., 2009). Interestingly, at the cellular level Met protein gets shuttled to the
processes of PNs, and its active phosphorylated state (pMet) is enriched in the neuropil
and largely absent from axon tracts, suggesting roles for Met in axon and dendrite
development (Eagleson et al., 2016b). Evidence of this can be found in neuronal culture
studies, which have demonstrated that Met activation by Hgf leads to increases in
dendrite growth an arborization and synapse density (Eagleson et al., 2016b; Finsterwald
& Martin, 2011). In vivo studies where Met is conditionally deleted results in abnormal
dendritic architecture and spine volume in cortical PNs (Judson et al., 2010a). In addition,
Met is only expressed in subsets of PNs within different PN subclasses, depending on
cortical regions (Kast, Wu, & Levitt, 2017; Lanjewar et al., 2023). These results taken
together suggest that Met signaling has diverse subclass- and subset-specific
developmental roles in cortical PNs across different cortical regions, and thus contributes
to the developmental and perhaps long-term diversity of cortical PNs. In chapter 3, I
discuss results from our experiments that investigated the transcriptional profiles of Met-
expressing cortical PNs during development to try and determine how these neurons
differ from their neighbors. These data provide the basis for further functional studies on
the role of a pleiotropic gene on structural and functional heterogeneity in the cortex.
23
SINGLE CELL SPATIAL TRANSCRIPTOMIC METHODS FOR MULTIMODAL INVESTIGATION OF
CELL TYPE DIVERSITY IN THE NERVOUS SYSTEM
Over the last century, much progress has been made in understanding cellular
diversity in the nervous system. Even so, due to the limitations of classical histological
and molecular methods and the high level of cellular and functional complexity in the
nervous system, we have barely scratched the surface. This holds true even in model
vertebrate organisms such as mice, where ~70 million neurons connect into functional
circuits through trillions of synapses (Erö et al., 2018b). Understanding a system of such
complexity requires high throughput methods. Fortunately, recent technological
advancements in tissue clearing, imaging, electrophysiological, molecular, tracing, and
sequencing have allowed the field to generate large amounts of data on the different
classification criteria involved in the generation and maintenance of neuronal diversity.
Furthermore, combination of these new tools now allows us to apply these high-
throughput methods to multiple cellular modalities, allowing us to understand cellular
diversity and circuits formation with very high resolution and specificity.
For example, combining classical viral tracing methods with novel tissue clearing,
immunohistological, light sheet imaging, and image analysis methods have allowed
researches to identify the anatomical topography of neurons that are active during a
behavior, and correlate this to their projection patterns (Renier et al., 2016). In another
example, researchers combine classical and novel (FlashTag) cell tagging and birthdating
methods with new single-cell transcriptomic methods to the track the developmental
trajectories of progenitor cell populations as the undergo neurogenesis and cell
24
differentiations, and link this with their gene expression profiles at different developmental
timepoints (Klingler et al., 2021). In a final example, investigators combined classical or
barcoding-based (MAPseq) tracing methods with single-cell sequencing to investigation
whether gene expression patterns correlate with neuronal projection diversity both
developmentally and in adult mice (Kebschull et al., 2016; Klingler et al., 2021).
Investigation of multiple cellular characteristics in a high-throughput fashion does
not always require the combination of multiple methods. This is the case for the expanding
toolkit of newly developed high-dimensional spatial transcriptomics methods. These
methods, some of which have been commercialized, allow for high-resolution anatomical
profiling of gene expression pattern in intact tissue. A number of these methods go down
to single-cell resolution, empowering investigators to link the molecular diversity of single-
cells with their tissue topography, and to concurrently investigate gene expression and
anatomical changes of specific cell types between experimental groups (Eng et al., 2019;
Shah et al., 2017; Ståhl et al., 2016; F. Wang et al., 2012).
In chapter 4, I describe a bench and computer analytical pipeline that we
developed to quantify and analyze data generated from a commercialized spatial
transcriptomics method (HiPlex RNAscope). This analytical tool will allow us to gain more
insight into the roles that genes play in the development of diverse cell types and their
organization within different structures and substructures of the nervous system.
25
CHAPTER 2: MET RECEPTOR TYROSINE KINASE REGULATES LIFESPAN ULTRASONIC
VOCALIZATION AND VAGAL MOTOR NEURON DEVELOPMENT
Kamitakahara, A. K.*, Ali Marandi Ghoddousi, R.*, Lanjewar, A. L., Magalong, V. M., Wu,
H. H., & Levitt, P. (2021). MET Receptor Tyrosine Kinase Regulates Lifespan Ultrasonic
Vocalization and Vagal Motor Neuron Development. Frontiers in Neuroscience, 15,
768577.
* Co-first authors
2.1 ABSTRACT
The intrinsic muscles of the larynx are innervated by the vagal motor nucleus
ambiguus (nAmb), which provides direct motor control over vocal production in humans
and rodents. Here, we demonstrate in mice using the Phox2b
Cre
line, that conditional
embryonic deletion of the gene encoding the MET receptor tyrosine kinase (MET) in the
developing brainstem (cKO) results in highly penetrant, severe deficits in ultrasonic
vocalization in early postnatal life. Major deficits and abnormal vocalization patterns
persist into adulthood in more than 70% of mice, with the remaining recovering the ability
to vocalize, reflecting heterogeneity in circuit restitution. We show that underlying the
functional deficits, conditional deletion of Met results in a loss of approximately one-third
of MET
+
nAmb motor neurons, which begins as early as embryonic day 14.5. The loss of
motor neurons is specific to the nAmb, as other brainstem motor and sensory nuclei are
unaffected. In the recurrent laryngeal nerve, through which nAmb motor neurons project
to innervate the larynx, there is a one-third loss of axons in cKO mice. Together, the data
26
reveal a novel, heterogenous MET-dependence, for which MET differentially affects
survival of a subset of nAmb motor neurons necessary for lifespan ultrasonic vocal
capacity.
2.2 INTRODUCTION
The ability to vocalize is an essential form of communication in nearly all
vertebrates (Barkan & Zornik, 2020). Analysis of laboratory rodent ultrasonic vocalizations
(USVs) has served as a powerful model system for understanding the neural circuits
underlying vocal communication, social and affiliative behaviors, and
neurodevelopmental disorders (Fischer & Hammerschmidt, 2011; Hofer, 1996; Holy &
Guo, 2005; Portfors & Perkel, 2014; Scattoni et al., 2009; Van Segbroeck et al., 2017;
Wöhr & Schwarting, 2013). Much of the circuitry responsible for innate vocalization (e.g.
rodent USVs, cries, and other nonverbal emotional utterances) is evolutionarily conserved
in mammals, and develops prenatally, allowing infants and rodent pups to vocalize readily
after birth (Arriaga et al., 2012; Barkan & Zornik, 2020; Hernandez-Miranda et al., 2017;
Hofer, 1996; Roubertoux et al., 1996). Identifying the precise molecular instructive signals
that are required for the development of neural-mediated vocal function will provide a
foundational understanding of the circuitry underlying vocal production and has
implications for understanding how communication deficits arise.
A few studies have begun to dissect early genetic and molecular factors that are
necessary for the development of vocalization circuits. Constitutive loss or functional
27
mutations of Foxp2 result in the complete absence of USVs in early postnatal mice and
are associated with multilevel alterations in the cortex, striatum, and cerebellum (Chabout
et al., 2016; French et al., 2007; Fujita et al., 2008; Groszer et al., 2008; Shu et al., 2005;
Spiteri et al., 2007; Urbanus et al., 2020) . Additionally, deletion of Olig3 or Tlx3 result in
the loss of large portions of the nucleus tractus solitarius (NTS) and dramatic reductions
in early postnatal USVs . The causality is unclear, however, because these phenotypes
occur prior to the onset of respiratory impairments in these mice, which leads to lethality
within the first 12h of birth (Hernandez-Miranda et al., 2017). Deletion of Tshz3 reveals
that its expression is critical for both development of respiratory neurons and survival of
a large portion of the nucleus Ambiguus (nAmb), where phonatory laryngeal motor
neurons reside (Caubit et al., 2010). We reported previously in prenatal mice that highly
restricted subsets of developing medullary neurons in the NTS and larynx-projecting
nAmb motor neurons express MET (A. Kamitakahara et al., 2017; Wu & Levitt, 2013a),
positioning this receptor tyrosine kinase as a candidate for influencing the development
of vocalization circuits.
MET is a pleiotropic receptor that is important for synapse maturation and critical
period regulation in the cerebral cortex (K. Chen et al., 2021; Ma & Qiu, 2020; Tsyporin
et al., 2021; Xie, Eagleson, et al., 2016) and for the differentiation of several different
motor neuron pools in the brainstem and spinal cord (Caton et al., 2000; Ebens et al.,
1996; Tallafuss & Eisen, 2008; Wong et al., 1997). Upon binding to its only known ligand,
hepatocyte growth factor (HGF), MET confers neuronal survival in developing spinal
motor neurons that innervate the pectoralis minor muscle (F. Lamballe et al., 2011). By
28
contrast, in adjacent limb-innervating motor neurons, MET is dispensable for neuronal
survival but instead stimulates axonal elaboration (F. Lamballe et al., 2011). A recent
study in zebrafish shows that similar to its role in spinal motor neurons, MET signaling is
involved in axon growth and guidance of vagal motor neurons that innervate the
pharyngeal arches (Isabella et al., 2020). Mammalian species exhibit considerable
differences from fish in vagal anatomy and pharyngeal arch-derived structures, as
mammals form a larynx instead of gill arches. However, homologous to patterns observed
in zebrafish, HGF is expressed embryonically in the developing airways of mice (A.
Kamitakahara et al., 2017). This suggests that in mammals, MET may serve as a signal
for instructing the development of medullary vagal laryngeal neurons.
Here, we used a conditional knockout strategy to delete MET from vagal motor
neurons in mice expressing Cre recombinase under the control of the Phox2b promoter
(cKO
mice). Using this transgenic mouse model, in combination with several transgenic
reporter lines, the functional and neuroanatomical effects of conditional deletion of MET
on USV production and nAmb development were examined. These studies reveal a novel
role for MET signaling in vagal motor neuron development and functional vocalization
across the lifespan.
2.3 MATERIALS AND METHODS
Animals
29
Animal care and experimental procedures were performed in accordance with the
Institutional Animal Care and Use Committee of The Saban Research Institute, Children's
Hospital Los Angeles. Mice were housed in the vivarium on a 13:11 hour light:dark cycle
(lights on at 06:00 hours, lights off at 19:00 hours) at 22°C with ad libitum access to a
standard chow diet (PicoLab Rodent Diet 20, #5053, St. Louis, MO).
A number of genotypically unique strains were generated through the mating of
available transgenic mice in order to perform the current studies. The Phox2b
cre
and Cre-
dependent TdTomato reporter lines (TdTom) were obtained from The Jackson Laboratory
(B6(Cg)-Tg(Phox2b-cre)3Jke/J, stock 016223, RRID:IMSR_JAX:016223 and B6.Cg-
Gt(ROSA)26Sortm14(CAG-tdTomato)Hze/J, stock 007914; Ai14,
RRID:IMSR_JAX:007914, respectively). Initial validation studies were performed to
assess recombination efficiency in the vagal motor nuclei using Phox2b
cre
;TdTom mice.
The TdTom reporter was used as a marker of recombination and choline
acetyltransferase (ChAT) was used as a marker to identify all vagal motor neurons in the
dorsal motor nucleus of the vagus (DMV) and nAmb. Nearly 100% of ChAT+ neurons
were also TdTom+, demonstrating high recombination efficiency in the vagal motor nuclei
using the Phox2b
cre
line (data not shown). We employed a strategy of utilizing the
Phox2b
cre
driver line with the Met
fx
line to leverage the highly selective expression overlap
in vagal motor neurons, very limited overlap in NTS neurons, and no overlap in nodose
(sensory) ganglion neurons.
30
The Met
fx
mouse line, in which exon 16 of the Met allele is flanked by loxP sites,
was shared by the laboratory of Dr. Snorri S. Thorgeirsson (National Cancer Institute,
NIH, Bethesda, MD) and is available at the Jackson Laboratory (Stock 016974). Met
EGFP
mice were generated as previously described (A. Kamitakahara et al., 2017; Kast, Wu,
Williams, et al., 2017) using GENSAT project clone BX139 (Rockefeller University,
RRID:SCR_002721)(Geschwind, 2004). In this mouse model, the EGFP transcript is
downstream of the MET promoter, and does not result in additional MET receptor
expression because of the insertion. Expression concordance between Egfp and Met
transcript or protein in neurons residing in the neocortex, raphe and vagal motor complex
is nearly 100% in Met
EGFP
mice. Both the Met
fx
and Met
EGFP
transgenic lines have been
backcrossed for more than 10 generations and maintained isogenically on a C57BL/6J
background in our laboratory for all experiments described.
To simplify the strain names of mice used, Cre negative (Cre-) will be used to describe
any mouse lacking the Phox2b
cre
allele, conditional wild type (cWT) will be used to
describe any mouse carrying the Phox2b
cre
allele but lacking the Met
fx
allele, and
conditional knockout (cKO) will be used to describe any mouse carrying the Phox2b
cre
allele and two copies of the Met
fx
allele. A complete listing of genotypes that fall under the
Cre-, cWT, and cKO designations is included in Extended Table 1.
Ultrasonic Vocalization
Isolation-evoked ultrasonic vocalizations (USVs) were recorded on P7 using a
CM16/CMPA ultrasound microphone positioned 5 cm above the recording chamber floor,
and an UltraSoundGate 116H recorder (Avisoft Bioacoustics). Mice were maintained in
31
the home nest environment in a cage warmed by a heating pad until the moment of
recording. For each 5-minute recording, a single pup was removed from the nest and
recorded in isolation without heat support in the recording chamber. Following each
recording, an additional 15-second recording was made to determine whether USVs
could be evoked by acute tail pinch. Computed values for USV number and duration were
determined using Avisoft-SAS Lab Pro software (Avisoft Bioacoustics,
RRID:SCR_014438). Genotype was unknown to the operator.
Adult male USVs were recorded during a direct social interaction task using a
CM16/CMPA ultrasound microphone positioned 16 cm above the chamber floor, and an
UltraSoundGate 116H recorder (Avisoft Bioacoustics). Prior to the day of testing, males
from each genotype group were exposed to a 3- to 4-month old C57BL/6J female partner
for three days in their home cage to gain social experience. Females were then removed
from the cage, and males were housed in isolation for two days to increase motivation to
call during the direct social interaction task. On the day of testing, males were habituated
to the recording chamber for 10 minutes, then recorded for 6 minutes with a novel 3- to
4-month old C57BL/6J female. Computed values for USV number and duration were
determined using Avisoft-SAS Lab Pro software (Avisoft Bioacoustics,
RRID:SCR_014438). Genotype was unknown to the operator.
Respiratory Analysis
USV recordings from P7 mouse pups were analyzed based on parameters
determined by Sirotin, et al. and used by Urbanus et al. to examine respiratory patterns
32
of mouse pups during vocalization (Sirotin et al., 2014; Urbanus et al., 2020). To
specifically examine respiratory pattern during vocalization, analysis was confined to
vocalizations within a bout, defined as a string of 2 or more calls or clicks (e.g. events)
made within 300ms of each other. Sirotin et al demonstrated that USVs that occur within
60ms of each other are produced within the same breath (Sirotin et al., 2014). To estimate
the number of multi-event breaths, the number of events within 60ms of one another was
quantified. Within a bout, the average time between events, or inter-syllable-interval was
quantified as a measure of the inspiratory pattern during respiration.
MUPET Analysis
To assess possible alterations in specific patterns of vocalization, Mice Ultrasonic
Profile ExTractor (MUPET) v2.0 (Van Segbroeck et al., 2017), an open-access MATLAB
(MATLAB_R2021a) NeuroResource, was used to analyze experience-evoked USVs
recorded on P7 and P60. MUPET uses a complex clustering algorithm to categorize
individual USV syllables into a ‘repertoire unit’, based on syllable shape (Van Segbroeck
et al., 2017). The collection of all the repertoire units made by each genotype or group is
referred to as a ‘repertoire’. Audio files were processed in MUPET, and a dataset was
created for each genotype or genotype subset. At each age and for each genotype,
repertoires of 20 to 140 repertoire units were built to determine the appropriate repertoire
size best suited for the analysis. For P7, a repertoire size of 60 units was determined to
be optimal based on an overall repertoire modeling score, average log-likelihood,
Bayesian Information Criterion, and the number of repertoire units containing only one
syllable. The same criteria were used at P60 to determine 100 units as the optimal
33
repertoire size at this age. The ‘Best Match Sorting’ feature was used to generate matrices
of Pearson correlations, comparing the similarity of each repertoire unit from the Cre-
dataset to repertoire units in all other genotype or genotype subset datasets. The ‘Unit
Activity Sorting’ feature was used to generate a Pearson correlation coefficient for overall
repertoire similarity. To estimate the Pearson correlation coefficient for the Cre-
repertoire, six jackknife resampled datasets were analyzed using MUPET (Efron, 1982;
Quenouille, 1956). These jackknife resampled datasets were generated by systematically
leaving out one recording in each dataset from the three highest and three lowest
vocalizers. Together, the Best Match Sorting and Unit Activity Sorting features measure
the similarity of individual repertoire units, and the overall similarity of repertoires between
groups, respectively.
Immunohistochemistry
Tissue for immunofluorescence staining was collected on embryonic days (E) 9.5
and 14.5, postnatal day (P) 7, and in P60-P80 (denoted elsewhere as P60 or adult) mice.
For embryonic tissue collection, timed pregnant breeding pairs were set in the evening,
and observance of a vaginal plug the following morning was designated as E0.5. Sex was
not determined for embryonic samples. Embryonic tissue was collected and immersed
overnight in fixative (4% paraformaldehyde in 0.1M phosphate-buffered saline (PBS, pH
7.4)). Postnatal mice were deeply anesthetized by intraperitoneal injection of
ketamine:xylazine (100mg/kg:10mg/kg, Henry Schein, Melville, NY) and perfused
transcardially with 0.9% saline, followed by fixative. Collected tissues were postfixed,
cryoprotected in 20% sucrose in PBS, embedded in Tissue-Tek® Optimal Cutting
34
Temperature Compound, and frozen over liquid nitrogen vapors or powdered dry ice. For
brain tissue, 20µm-thick cryostat sections were collected in five coronal or two sagittal
series representing the entire rostral-caudal or medial-lateral extent of the nAmb. For
laryngeal tissue, 30µm-thick cryostat sections were collected in five coronal series. For
recurrent laryngeal nerve (RLN) axon cross sections, ten or more 20µm coronal sections
were collected between the fifth tracheal ring and the larynx. Slides were stored at -20
o
C
until processed.
For immunofluorescence labeling, slides were blocked and then incubated in a
solution containing one or more of the following primary antibodies: chicken anti-Green
Fluorescent Protein (GFP) (1:500, Abcam Cat# ab13970, RRID:AB_300798), goat anti-
MET receptor tyrosine kinase (MET) (1:500, R and D Systems Cat# AF527,
RRID:AB_355414), mouse monoclonal anti-Neurofilament H 1G9 (NF) (1:500,
(Pennypacker et al., 1991)), or rabbit anti-Red Fluorescent Protein (RFP) (1:750,
Rockland Immunochemicals Cat# 600-401-379). Sections were washed in PBS, then
incubated in solution with one or more of the following secondary or tertiary antibodies:
Alexa Fluor® 488 AffiniPure F(ab')2 Fragment Donkey Anti‐Chicken IgG (Jackson
ImmunoResearch Cat# 703–546‐155, RRID:AB_2340376), Biotin‐SP‐AffiniPure
F(ab')2Fragment Donkey Anti‐Goat IgG (Jackson ImmunoResearch Cat# 705–066‐147,
RRID:AB_2340398), Alexa Fluor® 488 Streptavidin (Jackson ImmunoResearch Cat#
016-540-084, RRID: AB_2337249), Alexa Fluor® 647 AffiniPure Donkey Anti-Mouse IgM,
µ chain specific (Jackson ImmunoResearch Cat# 715-605-020, RRID: AB2340860), or
Alexa Fluor® 594 AffiniPure F(ab')2 Fragment Donkey Anti‐Rabbit IgG (Jackson
35
ImmunoResearch Cat# 711-586-152, RRID:AB_2340622). To label acetylcholine
receptor clusters in muscular tissue, ⍺-Bungarotoxin (⍺BT) Alexa Fluor® 488 conjugate
(Thermo Fisher Scientific Cat# B13422) was added to the secondary antibody solution.
Sections were counterstained with DAPI (Thermo Fisher Scientific Cat# D1306) and
embedded with ProLong Gold Antifade Mountant (Thermo Fisher Scientific Cat# P36930)
prior to applying a coverslip.
Image Acquisition
A Zeiss LSM 710 laser scanning confocal microscope equipped with 10x, 20x, 40x
water-corrected, and 63x oil-corrected objectives was used to acquire
immunofluorescence images. Confocal image stacks were collected through the z-axis at
a frequency optimally determined by the Zeiss Zen software based on the optics of the
microscope and the wavelength of the fluorophores used for analysis. Slides were coded
so that the operators were blind to experimental group
For cell counts of EGFP+ and tdTomato+ neurons in the nAmb, DMV, and NTS, every
fifth consecutive coronal section and second sagittal section was imaged through the
entire rostral-caudal or medial-lateral length of the nucleus using a 20x objective. For
imaging of the RLN and laryngeal muscle innervation, anatomical regions of interest for
each muscle or nerve were identified, and confocal image stacks were captured using a
20x, 40x water-corrected, or 63x oil-corrected objective.
Image Analysis
36
For cell counts of tdTomato+ and EGFP+ neurons in the nAmb, DMV, or NTS, all
image stacks for each animal were manually analyzed using the ‘Cell Counter’ plugin
within the Fiji/ImageJ software. The boundaries of the nAmb and cell inclusion/exclusion
criteria were agreed upon by two different experimenters prior to cell counts. For
tdTomato+ counts, all neurons within each medullary nucleus that had DAPI colocalized
with tdTomato were counted. For EGFP+ counts, all neurons within the medullary nucleus
that had DAPI colocalized with both tdTomato and EGFP were counted. Diameters of at
least 5% of nuclei were measured in Fiji/ImageJ and averaged per animal. The total
number of cells in the nAmb, DMV, or NTS of each animal was estimated in accordance
with Abercrombie’s formula (Abercrombie, 1946). Count validation was established by
two operators independently counting samples.
For the generation of sagittal topology maps, each cell position was labeled with a
region of interest (ROI) in Fiji/ImageJ and the ROIs were uploaded into R Studio using
the RImageJROI package. The X and Y coordinates of each cell ROI were then
aggregated and plotted onto a reference coordinate denominated by the most dorsal-
caudal point of the facial nucleus. Mapped cellular coordinates from each section and
animal were aggregated by group and plotted to generate reconstructed topological maps
of the nucleus.
For analyses of laryngeal muscle innervation, postsynaptic ⍺BT labeled
acetylcholine receptor clusters were counted using the ‘Cell Counter’ plugin in Fiji/ImageJ.
37
The number of ⍺BT labeled clusters closely opposed to tdTomato labeled fibers were
then counted to generate a measurement of percent innervation.
For analysis of RLN cross sections, each image was collapsed into a maximum
intensity z-stack in Fiji/ImageJ, a rolling-ball radius (rbr) background removal step was
performed (‘Subtract background’ with rbr of 15; ‘Disable Smoothing’), the axon bundle
was manually circled, and the 'Clear Outside' option was used to remove non-specific
signal inside and outside of the nerve bundle. Single axon fibers were quantified in an
automated manner using the ‘StarDist’ plugin (DSB 2018 model; Probability Threshold of
0.4; Overlap Threshold of 0.0) (Schmidt et al., 2018).
Quantification and Statistical Analysis
Data were analyzed statistically and graphed using GraphPad Prism software
(RRID:SCR_002798) and expressed as mean values ± standard error of the mean. The
number of animals required for quantitative analysis was calculated based on power
analysis, with the aim of detecting differences between groups that were 1-2 standard
deviations from the mean with at least 80% power and p<0.05 for significance. Each
mouse is considered a sample, with sample sizes of each genotype included in the figure
legends for each analysis. For each genotype, a D’Agostino-Pearson normality test was
used to determine whether parametric or nonparametric statistical analyses should be
performed. For data following a normal distribution, an ordinary one-way ANOVA or a
parametric two-tailed unpaired t-test was used to compare means. For data that failed to
pass the D’Agostino-Pearson normality test, a nonparametric Kruskal-Wallis test
38
(correcting for multiple comparisons using Dunn’s test) or a nonparametric two-tailed
Mann-Whitney test was used to compare mean rank difference. Due to small sample
sizes, normality tests were not performed for the sagittal counts, and a t-test was used to
compare means. To compare correlation coefficients, Fisher’s r-to-z transformation was
applied followed by z-tests with Bonferroni correction. A p<0.05 was used for significance.
The individual statistical tests used and sample sizes are further indicated in each figure
legend.
2.4 RESULTS
Ultrasonic vocalization is impaired in early postnatal life following loss of MET
The activity of vagal laryngeal motor neurons in the nAmb is required for rodent
ultrasonic vocalization (USV) (Nunez et al., 1985; Wetzel et al., 1980; Yajima et al., 1982).
To examine whether expression of MET by vagal neurons is necessary for vocal function,
we used a Cre-Lox strategy to conditionally delete Met using a histologically validated
Phox2b
cre
;Met
fx/fx
(cKO) transgenic mouse model (Supplementary Figure 1-1 and 1-2
(Huh et al., 2004)), in which there is extensive overlap in expression of both genes in
specific vagal motor neuron subtypes (A. Kamitakahara et al., 2017; Scott et al., 2011).
To test whether vagal motor output is functionally altered following loss of MET, cKO mice
were used to examine USVs evoked by a standard method of brief isolation of mouse
pups from the nest recorded on postnatal day (P) 7. Following removal from the nest,
control (Cre-) pups made robust USV calls over the course of a 5-minute recording period.
39
Both the number and duration of calls were similar between Cre- and conditional wild-
type (cWT) control pups, demonstrating that Cre expression alone in this transgenic
model has no effect on vocalization (Figure 1A and B). By contrast, cKO mutant pups
exhibited severely reduced numbers of calls and decreased call duration (Figure 1A and
B), indicating that the loss of MET in vagal neurons impairs USV production. Remarkably,
at this age, the USV disrupted phenotype was fully penetrant for all cKO mice tested.
To determine whether the loss of isolation-evoked USVs is due to a motor deficit
to call, or perhaps a reduced motivation to call, short 15-second recordings were made
following a brief tail pinch stimulus to directly elicit vocalization. The mechanical tail pinch
stimulus provokes an involuntary painful vocalization in mouse pups. Therefore, a lack of
USVs evoked by tail pinch would further demonstrate motor deficits in the ability to call.
While both Cre- and cWT pups evoked many complex USVs following the tail pinch, cKO
pups made very few USVs, if any, and those emitted were very short in duration
(Supplementary Figure 1-3A and D). cKO mice also did not make any calls in the audible
range, but did produce significantly more clicks (a straight vertical shape on the
spectrogram) than Cre- pups (Supplementary Figure 1-3B, C, and D), suggesting that
emitted clicks may be generated by cKO pups in place of audible calls and complex USVs.
Together, these results demonstrate that pup vocalization is severely impacted by loss of
MET in vagal neurons.
To examine whether the loss of MET results in major changes in respiratory
patterns during vocalization, a secondary analysis of USV recordings was performed.
40
Mice typically take a single breath between each call, but occasionally will make more
than one call in the same breath (Sirotin et al., 2014). The ability to produce these multi-
event breaths are indicative of respiratory control. To examine irregularities in breathing
pattern during vocalization, the number of breaths that have more than one call or click
(e.g. event) were analyzed in cWT and cKO pups. Although cKO pups vocalize much
less, for those bouts that did occur, there were no differences in the percentage of multi-
event breaths. This analysis indicates that cKO pups possess sufficient respiratory control
necessary to make multiple calls within a single breath (Supplementary Figure 1-4A).
Furthermore, within a call bout, there was no difference in the inter-event-interval
(Supplementary Figure 1-4B), suggesting that the timing of the call-breath-call respiratory
pattern is similar between genotypes.
There was a small, but significant decrease (~20%) in the body weight of mutant
cKO pups on P7 (Supplementary Figure 1-5A). To determine if such changes impacted
vocalization capacity, the number of calls were examined as a function of pup body
weight. The analysis revealed that both Cre- and cWT pups of comparable body weights
to the knockout mice readily emitted isolation-evoked USVs, suggesting that the inability
of cKO pups to make calls is not a result of being smaller (Supplementary Figure 1-5B).
Furthermore, there were no differences in body weight in adult males and females
between genotypes (Supplementary Figure 1-5C and D).
Rodent USV repertoires are composed of a diverse compliment of syllable shapes
used to generate strain, sex, and context-specific vocal communication across
41
development and in adulthood (Arriaga et al., 2012; Grimsley et al., 2011; Holy & Guo,
2005; Van Segbroeck et al., 2017). Given that loss of MET results in a major reduction in
the number and duration of calls made by cKO mice in early postnatal life, we assessed
whether the shape of the few syllables emitted also might be altered, reflecting motor
impairments in the ability to produce complex vocalizations. To analyze differences in
USV repertoires between genotypes, Mouse Ultrasonic Profile ExTraction (MUPET)
software was used to perform unsupervised signal processing of isolation-evoked USVs
recorded on P7 (Van Segbroeck et al., 2017). Within each genotype, all recorded
syllables were clustered into repertoires composed of 60 repertoire units (RUs). Each RU
represents the syllable centroid, or average of syllable shapes in that RU cluster. The
Cre- repertoire was composed of several distinct RUs (short simple, chevrons, frequency
jumps, complex harmonics, etc.) (Figure 1D). The shape of each of these RUs in the Cre-
repertoire was then statistically compared to RUs clustered in cWT and cKO repertoires
(Figure 1F and H). For this analysis, MUPET generates a matrix of individual Pearson
correlations for pairwise comparisons of RUs in each repertoire. The matrix is color coded
such that warmer colors correspond to highly similar RUs. Fifty-four out of 60 RUs had
Pearson correlations greater than 0.7 when comparing the cWT and Cre- repertoires
(colored red to pink across the matrix diagonal in Figure 1E), demonstrating that the
shape of the syllables in these repertoires are highly similar to one another. In contrast,
the cKO repertoire was primarily composed of very short syllables, and a straight vertical
shape consistent with emitted clicking sounds. Only 28 out of 60 RUs in the cKO
repertoire had Pearson correlations greater than 0.7 when compared to the Cre-
repertoire (Figure 1G) and were primarily made up of very short simple shapes,
42
demonstrating the production of both a very limited and distinct syllable repertoire in the
cKO mice. In addition to comparing individual RUs, the entire Cre- repertoire was
compared to the entire cWT and cKO repertoires. Pearson correlation coefficients were
calculated for 100% of the syllables produced in each repertoire. While the cWT repertoire
was highly similar to the Cre- repertoire, with an overall Pearson’s r of 0.86, the cKO
repertoire was significantly different from the Cre- repertoire, with an overall Pearson’s r
of 0.60 (Figure 1C). These data indicate that expression of MET in vagal neurons is critical
for generating the diversity and complexity of syllables in the vocal repertoire produced
by P7 pups.
43
44
FIGURE 1. Deletion of MET results in severely impaired ultrasonic vocal production and syllable
repertoire early postnatally. (A) Quantification of the number of isolation-evoked USVs over the 5-min
recording period on P7. n = 37 Cre-, n = 12 cWT, n = 11 cKO. “∗∗” indicates p < 0.01, “****” indicates p <
0.0001 as analyzed by non-parametric Kruskal-Wallis test and Dunn’s correction for multiple comparisons.
(B) Quantification of the duration of isolation-evoked USVs over the 5-min recording period on P7. n = 37
Cre-, n = 12 cWT, n = 11 cKO. “∗∗” indicates p < 0.01, “****” indicates p < 0.0001 as analyzed by non-
parametric Kruskal-Wallis test and Dunn’s correction for multiple comparisons. (C) MUPET boxplot
comparing the Cre- repertoire to the cWT repertoire (blue) or cKO repertoire (red). Similarity of the top 5%
most frequently used RUs in the Cre negative repertoire indicated by ∗. Similarity of the top 25% most
frequently used RUs in the Cre negative repertoire indicated by top of the box. Similarity of the top 50%
most frequently used RUs in the Cre- repertoire indicated by the horizontal line. Similarity of the top 75%
most frequently used RUs in the Cre- repertoire indicated by the bottom of the box. Similarity of the top
95% most frequently used RUs in the Cre- repertoire indicated by +. r-values below boxes indicate the
overall Pearson correlation coefficient for the entire repertoire. # indicates overall Pearson’s r-values that
are significantly different from the Cre- repertoire analyzed using a Fisher r-to-z transformation to make
pairwise p-value calculations followed by Bonferroni correction for multiple comparisons. (D) Each
repertoire unit (RU) in the Cre- repertoire, displayed in the order of frequency of use. (E,G) Pearson
correlation matrices comparing each of the Cre- RUs (y-axis) to RUs in the cWT repertoire (x-axis, E) or
cKO repertoire (x-axis, G), ordered from most to least similar in shape. Warmer colors indicate higher
Pearson correlation, cooler colors indicate lower Pearson correlation. Boxed area shows the number of
RUs with Pearson correlations above 0.7, with the corresponding number of RUs indicated in the upper
right corner. (F) Each repertoire unit (RU) in the cWT repertoire, displayed in the order of frequency of use.
(H) Each repertoire unit (RU) in the cKO repertoire, displayed in the order of frequency of use. Total syllable
number in each MUPET repertoire: Cre- = 12,070; cWT = 3,416; cKO = 1,196.
USVs are impaired in the majority of male cKO mice in adulthood
45
To determine whether the inability to emit USVs is sustained into adulthood,
separate cohorts of Cre-, cWT, and cKO male mice were examined at P60 using a
standard direct social interaction task with a wild-type female C57BL/6J conspecific. Only
males were analyzed in this paradigm, as females did not produce USVs in sufficient
numbers in this behavioral paradigm to distinguish from baseline (data not
shown)(Heckman et al., 2016b). Both Cre- and cWT males averaged several hundred
calls over the course of the 6-minute recording period (Figure 2A). By contrast, cKO male
mice produced significantly fewer USVs over that same time period (Figure 2A).
Furthermore, average USV call duration was significantly reduced in mutant mice (Figure
2B). Intriguingly, the responses of cKO male mice appeared to follow two distinct patterns:
more than 70% of mice tested exhibited very limited vocalizations during interaction with
the females, using only short simple syllables (cKO low vocalizers); a smaller group of
mice vocalized fairly robustly and were able to produce more complex syllables (cKO high
vocalizers). These data, together with the nearly fully penetrant vocalization phenotype
at P7, indicate that while the majority of adult cKO mice continued to have a very limited
ability to vocalize, a small subset (5/18; 27.8%) appeared to regain USV-related laryngeal
function.
To determine whether USV repertoire impairments observed in early postnatal life
are sustained into adulthood, MUPET was used to examine syllable repertoires in
recordings from P60 males paired with females in the direct social interaction task. Within
each genotype, all recorded syllables were clustered into 100 RUs, based on modeling
parameters. The Cre- repertoire was composed of a diverse array of simple and complex
46
syllable shapes (Figure 2C). cWT and cKO repertoires were composed of syllables with
high similarity (Pearson’s correlations above 0.7) to the Cre- repertoire, 86/100 and
83/100 respectively (Figure 2F, G, L, M). Given observations that the USV recordings
from cKO mice at P60 were either low vocalizers or high vocalizers, repertoires of these
sub-sampled groups were compared. USV recordings from cKO mice were grouped into
either low (n=13) or high vocalizing (n=5) groups split by the mean number of USVs in
that genotype. Leveraging the vocalization variation in control mice, Cre- mice were also
grouped into either low or high vocalizing groups split by the mean number of USVs in
that genotype. When each of these four subgroups was compared to the entire Cre-
repertoire, it was found that both Cre- high and low vocalizers had high Pearson
correlations for syllable shape when compared to the entire Cre- group (Figure 2D, E, J,
K). These data indicate that the syllables produced by control, low vocalizing mice are
very similar to those produced by control, high vocalizing mice. Similarly, 78/100 syllables
from the five high vocalizing cKO mice exhibited high Pearson correlations for syllable
shape (Figure 2H and N), indicating that both the frequency and quality of vocalizations
in this small group of mice were normal, even in the absence of MET. In contrast, for the
cKO low vocalizer group, only 43/100 syllables had Pearson correlations above 0.7
(Figure 2I and O), revealing both that the vocalizations were less frequent and that the
repertoires of these mice were distinct from the Cre- controls. Similar to recordings at
P7, the cKO low vocalizer recordings were composed primarily of very short syllables with
a straight vertical shape that is often associated with either a clicking sound or noise
artifact. The entire Cre- repertoire also was compared to all repertoires from other
genotypes by calculating Pearson correlation coefficients for 100% of used syllables. Both
47
Cre- high and low vocalizers had large Pearson correlations for syllable shape when
compared to the entire Cre- group, 0.87 and 0.80, respectively (Figure 2P). Similarly,
cWT, cKO, and cKO high vocalizers had high overall Pearson correlations coefficients
when compared to the full Cre- repertoire (Figure 2P). However, cKO low vocalizers had
a significantly smaller Pearson’s r value (Figure 2P). This demonstrates that the
apparently "normal" repertoire of the entire cKO group is being driven by the few high
vocalizers, while major deficits in ultrasonic vocalizations are sustained into adulthood in
most cKO mice.
48
49
FIGURE 2. Sustained vocalization deficits in Met cKO adult mice. (A) Quantification
of the number of USVs produced by P60 males paired with females over a 6-min
recording period during a direct social interaction task. n = 26 Cre-, 13 cWT, 18 cKO. “∗”
indicates p < 0.05 as analyzed by non-parametric Kruskal-Wallis test and Dunn’s test to
correct for multiple comparisons. (B) Quantification of the duration of USVs made by P60
males over the 6-min recording period. n = 26 Cre-, 13 cWT, 18 cKO. “∗” indicates p <
0.05 as analyzed by non-parametric Kruskal-Wallis test and Dunn’s test to correct for
multiple comparisons. (C) Each repertoire unit (RU) in the Cre- repertoire, displayed in
the order of frequency of use. (D) Each repertoire unit (RU) in the Cre- high vocalizer
repertoire, displayed in the order of frequency of use. Repertoire generated from 13 Cre-
recordings with the highest number of calls. (E) Each repertoire unit (RU) in the Cre- low
vocalizer repertoire, displayed in the order of frequency of use. Repertoire generated from
13 Cre Negative recordings with the lowest number of calls. (F) Each repertoire unit (RU)
in the cWT repertoire, displayed in the order of frequency of use. (G) Each repertoire unit
(RU) in the cKO repertoire, displayed in the order of frequency of use. (H) Each repertoire
unit (RU) in the cKO high vocalizer repertoire, displayed in the order of frequency of use.
Repertoire generated from 5 cKO recordings with the highest number of calls. (I) Each
repertoire unit (RU) in the cKO low vocalizer repertoire, displayed in the order of frequency
of use. Repertoire generated from 13 cKO recordings with the lowest number of calls. (J–
O) Pearson correlation matrices comparing each of the Cre- RUs (y-axis) to RUs in the
Cre- high vocalizer repertoire (x-axis, J), Cre- low vocalizer repertoire (x-axis, K), cWT
repertoire (x-axis, L), cKO repertoire (x-axis, M), cKO high vocalizer repertoire (x-axis, N),
or cKO low vocalizer repertoire (x-axis, O), ordered from most to least similar in shape.
50
Warmer colors indicate higher Pearson correlation, cooler colors indicate lower Pearson
correlation. Boxed area shows the number of RUs with Pearson correlations above 0.7,
with the corresponding number of RUs indicated in the upper right corner. (P) MUPET
boxplot comparing the Cre- repertoire to each group. Similarity of the top 5% most
frequently used RUs in the Cre- repertoire indicated by ∗. Similarity of the top 25% most
frequently used RUs in the Cre- repertoire indicated by top of the box. Similarity of the top
50% most frequently used RUs in the Cre- repertoire indicated by the horizontal line.
Similarity of the top 75% most frequently used RUs in the Cre- repertoire indicated by the
bottom of the box. Similarity of the top 95% most frequently used RUs in the Cre-
repertoire indicated by +. r-values below boxes indicate the overall Pearson correlation
coefficient for the entire repertoire. # Indicates overall Pearson’s r-values that are
significantly different from the Cre- repertoire analyzed using a Fisher r-to-z
transformation to make pairwise p-value calculations followed by Bonferroni correction
for multiple comparisons. Total syllable number in each MUPET repertoire: All Cre- =
21,002; Cre- (High Vocalizer) = 14,651; Cre- (Low Vocalizer) = 6,351; cWT = 11,494; All
cKO = 6,370; cKO (High Vocalizer) = 4,803; cKO (Low Vocalizer) = 1,567.
Reduced recurrent laryngeal branch axons following deletion of Met in nAmb
Production of rodent USVs requires fine motor control of the muscles within the
larynx (Riede, 2011, 2013). Laryngeal muscle contraction is driven by motor neurons
located in the nAmb that project predominantly through the recurrent laryngeal branch of
the vagus nerve(Bieger & Hopkins, 1987; Kelm-Nelson et al., 2018; Nunez et al., 1985;
51
Van Daele & Cassell, 2009; Wetzel et al., 1980). To examine whether the loss of MET
expression alters axon projections within the recurrent laryngeal nerve, neurofilament
immunofluorescence staining was used to quantify the number of axons in recurrent
laryngeal nerve cross sections collected on P7. Compared to control cWT mice, cKO mice
exhibited an approximately 30% reduction in the number of axons traveling to the
laryngeal musculature (Figure 3A-C), suggesting that vocal impairments resulting from
Met deletion are the result of reduced motor input to the vocal organ.
To determine whether the loss of axons in the recurrent laryngeal branch results
in the loss of motor endplate innervation to the intrinsic laryngeal muscles, cKO mice were
bred to a conditional reporter line allowing visualization of tdTomato+ (tdTom+) axons.
Laryngeal tissue sections were co-stained with fluorescently tagged alpha-bungarotoxin
(𝛼 BT) to label acetylcholine receptor (AChR) clusters at the neuromuscular junction. The
percentage of stained 𝛼 BT profiles that were closely apposed to tdTom+ axons was then
quantified in cWT and cKO mice on P7. In all of the intrinsic muscles in the larynx
responsible for vocal production that were assayed (thyroarytenoid, TA; lateral
cricoarytenoid, LCA; cricothyroid, CT; posterior cricoarytenoid, PCA; and alar portion of
the thyroarytenoid, AlarTA), nearly all of the 𝛼 BT labelled AChR clusters were juxtaposed
by tdTom+ axons in cWT and cKO genotypes (Figure 3D), indicating that the remaining
two-thirds of motor axons form laryngeal neuromuscular junctions following conditional
deletion of Met (Figure 3E-H; LCA, CT, and AlarTA not shown).
52
Neuromuscular junction maintenance was also assessed in laryngeal samples
from cKO mice at P60. Similar to the P7 time point, at P60 nearly all 𝛼 BT labelled AChR
clusters were closely apposed by tdTom+ axons in the intrinsic laryngeal muscles of both
genotypes (data not shown), suggesting that the remaining axons continue to maintain
neuromuscular junctions through early adulthood.
FIGURE 3. Axonal quantification and innervation status of αBT labeled AChR
clusters in laryngeal muscles following MET deletion. (A) Quantification of the
number of axons in the recurrent laryngeal nerve in cWT and cKO mice on P7. n = 5 cWT,
53
5 cKO. t-test, “∗∗” indicates p ≤ 0.05. (B,C) Representative images of recurrent laryngeal
nerve cross sections labeled with neurofilament (red) in cWT and cKO mice on P7.
Scalebars = 4 μm. (D) Quantification of laryngeal motor end plate innervation in cWT and
cKO mice on P7. n = 3–5 cWT, 5–6 cKO. Analyzed by two-way ANOVA followed by Sidak
correction for multiple comparisons. (E,F) Representative images of tdTom + axons (red)
innervating αBT labeled AChR clusters (green) in the thyroarytenoid (TA) muscle of cWT
and cKO mice on P7. Insets show enlarged examples of neuromuscular junction
morphology. (G,H) Representative images of tdTom + axons (red) innervating αBT
labeled AChR clusters (green) in the posterior cricoarytenoid (PCA) muscle of cWT and
cKO mice on P7. Insets show enlarged examples of neuromuscular junction morphology.
(I,J) Representative images of tdTom + axons (red) innervating αBT labeled AChR
clusters (green) in the lateral cricoarytenoid (LCA) and cricothyroid (CT) muscle of cWT
and cKO mice on P7. All laryngeal muscle images shown are from tissue sectioned in the
coronal plane (prepared along the length of the larynx in sections from dorsal to ventral).
Scalebars = 50 μm in laryngeal muscle sections. The brightness and contrast of each
channel was adjusted separately for visualization purposes.
MET is required for the development of a subset of nAmb motor neurons
To determine whether the loss of axons in the recurrent laryngeal branch of cKO
mice is due to a reduction in the neurons that supply this branch located in the nAmb, the
distribution of motor neurons along the rostral-caudal axis of the nAmb was examined.
Previously published work from our laboratory demonstrated that MET expression is
54
primarily confined to neurons located in the rostral compact and the caudal loose
formations of the nAmb (A. Kamitakahara et al., 2017). To investigate structural changes
in nAmb topology following conditional deletion of Met, cWT and cKO mice were crossed
with mice expressing a Cre-dependent tdTomato reporter and enhanced green
fluorescent protein (EGFP) under the control of the MET promoter (Met
EGFP
), for which
there is near absolute fidelity between EGFP and intrinsic Met expression (A.
Kamitakahara et al., 2017; Kast, Wu, Williams, et al., 2017). Because the Met promoter
driven Egfp transgene (Met
EGFP
) is inserted independently from the Met
fx
allele, this
mouse model permits visualization and comparison of neurons expressing MET in cWT
mice and neurons expressing non-functional MET protein in cKO mice (see Methods).
Using these mice, sagittal sections of the P7 brainstem were obtained and used to
generate topologic maps of nAmb motor neurons. These maps confirmed previous
findings from our laboratory showing that MET-expressing motor neurons are primarily
confined to the compact and loose formations of the nAmb (Figure 4A and B). Very few
MET-expressing cells were present in the middle portion of the nucleus, which
corresponds to the semi-compact formation. This anatomical segregation of MET
expressing motor neurons to the rostral and caudal ends of the nAmb also is present in
cKO mice, demonstrating that the distribution of nAmb motor neurons is not altered in the
absence of MET signaling (Figure 4A and B). In addition, qualitative comparison of the
cWT and cKO maps highlight a prominent reduction in the size of the compact formation
of the nAmb in cKO mice (Figure 4B).
55
To investigate the aforementioned qualitative changes in the nAmb after
conditional deletion of MET, sagittal cell counts of all tdTom+ nAmb neurons from cKO
were compared to those from the cWT mice. Quantification of tdTom+ neurons in cWT
and cKO mice revealed a statistically significant 31.7% loss of motor neurons in the rostral
compact portion of the nAmb (250-500µm caudal to the facial nucleus) (Figure 4C). In
agreement with tdTom+ cell counts, quantification of the EGFP+ subpopulation revealed
a 49.2% loss of this motor neuron subgroup in the rostral nAmb (Figure 4D). The more
caudal loose formation of the nucleus (500-1500µm caudal to the facial nucleus), where
MET is also expressed, exhibited a trend toward a reduced number of neurons in cKO
mice that did not reach statistical significance.
56
FIGURE 4. Topology of MET-EGFP expression and motor neuron loss in the nAmb
following conditional deletion of Met. (A) Single-section representation of Phox2b +
and MET + motor neuron topology in the nAmb. (B) Comparative topology of tdTom
(Phox2b) and EGFP (MET) positive nAmb motor neurons. n = 3 per genotype. (C) Binned
Abercrombie-corrected counts of TdTom + cells denoting Phox2b + motor neurons in
nAmb of cWT and cKO mice. n = 3 per genotype. t-test, “*” indicates p ≤ 0.05. (D) Binned
Abercrombie-corrected counts of GFP + cells denoting MET + motor neurons in nAmb of
WT and cKO mice. n = 3 per genotype. t-test, p ≤ 0.05. Scale bars = 250 μm. The
brightness and contrast of each channel was adjusted separately for visualization
purposes.
To verify the loss of MET-expressing neurons viewed in the sagittal plane of the
medulla, we next processed additional sets of cWT and cKO mice to perform cell counts
in nAmb imaged in the coronal plane. Analysis in this plane of section reduces potential
variability in section orientations that can occur in the sagittal plane, particularly across
multiple developmental timepoints. Brainstem coronal sections were collected at several
developmental time points, beginning in embryonic life through adulthood in cWT and
cKO mice. On E14.5, approximately four days after nAmb neurons are generated (Pierce,
1973), a decrease in the size of the developing nAmb in cKO samples already was evident
(Figure 5A). Quantification of the number of neurons in the nAmb at this time point
revealed a statistically significant 23.3% decrease in the number of tdTom+ neurons in
cKO mice compared to cWT controls (Figure 5B). While the number of nAmb neurons
57
was reduced, there were no ectopically located tdTom neurons evident in the embryonic
and postnatal medulla, consistent with a lack of obvious deficits in cell migration. To
determine whether this loss in neuron number is maintained later in life, tdTom+ cell
counts were quantified in the brainstem of early postnatal and adult mice. Consistent with
our earlier sagittal counts, there was a 36.4% loss of tdTom+ neurons on P7 (Figure 5C
and D). The magnitude of neuronal loss in the nAmb remained consistent into adulthood,
with a 36.9% decrease in tdTom+ neurons on P60 (Figure 5E and F). Together these
data demonstrate that MET is required during embryonic and postnatal development for
sustaining a normal number of nAmb neurons, particularly in the compact formation of
the rostral nAmb.
cWT and cKO mice also were analyzed to quantify the neuron loss in the MET-
expressing subpopulation of neurons in the rostral nAmb. To accomplish this, multi-
transgenic mice were generated that express a Cre-dependent tdTomato reporter and
the Met
EGFP
allele to facilitate comparison of neurons expressing MET in cWT mice and
neurons expressing non-functional MET protein in cKO mice (Phox2b
Cre
, tdTomato, Met
fx
,
and Met
EGFP
alleles; Figure 5G). Cell counts in the nAmb at P60 revealed a 36.4%
decrease in the number of EGFP+ neurons (Figure 5H), consistent with the absence of a
subpopulation neurons in the nAmb following deletion of Met. Furthermore, within each
genotype, no sex differences were observed in the number of EGFP+ or tdTom+ neurons
at any of the postnatal ages examined (Supplementary Figure 5-1).
58
FIGURE 5. Analyses of lifespan loss of vagal motor neurons in the nAmb following
conditional deletion of MET. (A) Representative images of the nAmb from cWT and
cKO mice on E14.5. (B) tdTom + cell counts from cWT and cKO mice on E14.5. n = 11
59
cWT, 12 cKO. “**” indicates p < 0.01 as analyzed by parametric unpaired t-test. (C)
Representative images of the nAmb of cWT and cKO mice on P7. (D) tdTom + cell counts
from cWT and cKO mice on P7. n = 10 cWT, 8 cKO. “***” indicates p < 0.0001 as analyzed
by parametric unpaired t-test. (E) Representative images of the nAmb of cWT and cKO
mice on P60. (F) tdTom + cell counts from cWT and cKO mice on P60. n = 10 cWT, 17
cKO. “****” indicates p < 0.0001 as analyzed by parametric unpaired t-test. (G)
Representative images of EGFP + (green) motor neurons (tdTom +; red) in nAmb of cWT
and cKO mice on P60. Images in E and G demonstrate EGFP and tdTom overlap from
the same samples. (H) EGFP + cell counts from cWT and cKO mice on P60. n = 10 cWT,
17 cKO. “**” indicates p < 0.01 as analyzed by parametric unpaired t-test. Scalebars =
100 μm. The brightness and contrast of each channel was adjusted separately for
visualization purposes. Some images were cropped to center the nAmb in the image
plane.
Overlap between Phox2b
cre
and MET-expressing cells is restricted to the NTS, DMV, and
nAmb
To determine whether other brain regions were affected in our cKO model, whole
brain samples from P2 pups and adult control mice expressing tdTomato in all Phox2b+
cells and MET
EGFP
in all MET-expressing cells were systematically examined to
investigate areas of Phox2b
cre
and MET co-expression. Consistent with previous reports,
vagal Cre recombinase expression in this Phox2b
cre
line (designated line 3, JAX stock
016223) was observed in the nAmb, the dorsal motor nucleus of the vagus (DMV), the
60
nucleus of the solitary tract (NTS) and the nodose-jugular ganglia (Scott et al., 2011).
Phox2b-driven recombination was much broader than MET expression. MET was not
expressed in the vagal sensory nodose-jugular complex. Within the nAmb, more than
34% of the neurons co-expressed Phox2b and MET. A much more limited subset of
neurons in the DMV (~6.7%) and NTS (~3.0%) co-expressed Phox2b and MET.
Examination of other brainstem structures revealed that there was minimal to no overlap
between MET and Phox2b expression in the hypoglossal (XII), facial (VII), prepositus
(PRP), and medial vestibular (MV) nuclei, or any other brainstem region (Supplementary
Figure 6-1). In addition, no Phox2b+ neurons were located in the nucleus retroambiguus,
consistent with other studies (Hernandez-Miranda et al., 2017). No overlap of MET and
Phox2b was found in the cortex, hippocampus, hypothalamus, thalamus, nor
periaqueductal gray. All forebrain regions examined exhibited minimal to no overlap
between MET and Phox2b. The analyses demonstrate that in this conditional deletion
model, Phox2b
cre
and MET-expressing cells primarily overlap in nAmb, with minimal co-
expression in the NTS, DMV. Thus, loss of Met in one or a combination of these nuclei
are likely to be responsible for the severe vocal deficits.
Our earlier analysis demonstrated that MET is necessary for proper nAmb
development. To determine whether MET may be required for the development of
neurons in the NTS or DMV, the number of EGFP+ and tdTom+ neurons was quantified.
Colocalization of EGFP+ and tdTom+ neurons in the NTS are of particular interest, as a
major cell loss in the NTS causes early postnatal lethality with impaired respiration and
vocal production (Hernandez-Miranda et al., 2017). While the vast majority of EGFP+
61
neurons in the DMV also were tdTom+, only one third of EGFP+ neurons in the NTS were
observed to colocalize with the tdTom reporter (Figure 6A, B, and E). Cell counts further
revealed that there are approximately 32 neurons, out of more than 3300 tdTom+ neurons
in the entire bilateral NTS, that express both MET and Phox2b (compared to ~190 in the
unilateral nAmb; Figure 6F). Additionally, the ~3300 tdTom+ neurons represent a small
fraction of NTS neurons, as the vast majority of neurons in the NTS do not express tdTom
in this model (Figure 6A and B). While it is very unlikely that this very limited number of
Phox2b/MET expressing neurons in the NTS are responsible for the dramatic vocalization
phenotype observed in cKO mice, we nonetheless quantitated cell numbers in cWT and
cKO mice. There were no differences in the number of EGFP+ or tdTom+ cells in either
the DMV or NTS between genotypes (Figure 6C and D). Additionally, there are no known
connections between the DMV and the branchial organs involved in USV production in
rodents. Together, these data analyses provide strong evidence that a selective and
specific subset of vagal motor neurons located in the nAmb are lost due to Met deletion
in cKO mice, indicating that developmental MET signaling is required by the neurons in
nAmb that contribute to the circuitry underlying the production of USVs.
62
FIGURE 6. No difference in the number of neurons in the NTS or DMV is observed
following conditional deletion of MET. (A,B) Representative images of the DMV, NTS,
and reticular nucleus (RN) from cWT and cKO mice on P60. (C) EGFP + cell counts in
the DMV and NTS from cWT and cKO mice on P60. n = 9 cWT, 6 cKO. No significant
difference was found between genotypes using a non-parametric Mann-Whitney test. (D)
tdTom + cell counts in the DMV and NTS from cWT and cKO mice on P60. n = 9 cWT, 6
cKO. No significant difference was found between genotypes using a non-parametric
Mann-Whitney test. (E) Percentage of EGFP + cells co-expressing tdTom in the DMV and
NTS from cWT and cKO mice on P60. No significant difference was found between
genotypes using a non-parametric Mann-Whitney test. (F) Number of EGFP + cells co-
expressing tdTom in the DMV and NTS from cWT and cKO mice on P60. No significant
difference was found between genotypes using a non-parametric Mann-Whitney test. All
63
scale bars = 100 μm. The brightness and contrast of each channel was adjusted
separately for visualization purposes.
2.5 DISCUSSION
Like other motor circuits in the central nervous system, the present study
demonstrates that the development of functional vocalization is dependent upon the
timely exposure of neurons, and their responsiveness, to specific molecular cues
(Guthrie, 2007a; Jessell, 2000; Puelles et al., 2019). In the current study, we have newly
identified MET as a critical neurodevelopmental signal that is necessary for establishing
normal function and anatomy of the vagal motor nAmb. Specifically, the present data
demonstrate that the loss of MET signaling results in enduring, major deficits in the ability
to communicate using USVs, a significant decrease in the number of axonal projections
within the recurrent laryngeal nerve supplying motor innervation of the larynx, and a
significant decrease in a subset of vagal efferent motor neurons in the nAmb. The
functional changes in the number of vocalizations are profound, with mutant mice
exhibiting severe vocalization abnormalities. The functional analyses thus suggest that
expression of MET in a subset of vagal motor neurons is necessary for proper USV
production and the full repertoire of syllable shapes. Together with the anatomical studies
across different timepoints, the experimental results reveal a relationship between USV
production and developmental expression of MET by nAmb motor neurons.
64
Deficits in ultrasonic vocalization in early development and adulthood
The most dramatic phenotype observed in cKO mice is the severe disruption of
ultrasonic vocal communication. This is a fully penetrant phenotype developmentally, as
all cKO pups analyzed made few, if any, isolation-evoked USVs. Furthermore, analyses
of repertoires demonstrated that when there is vocalization, the cKO pups produce only
short, simple calls compared to the more complex long duration calls that are made by
Cre- and cWT pups. Studies in which the recurrent laryngeal nerves are unilaterally or
bilaterally transected demonstrate the importance of the connection between the nAmb
and the larynx for vocal fold movement and USV production (Bieger & Hopkins, 1987;
Kelm-Nelson et al., 2018; Nunez et al., 1985; Van Daele & Cassell, 2009; Wetzel et al.,
1980). Additionally, the current study demonstrates that MET expression during
development in a subset of vagal motor neurons of the nAmb is required for proper vocal
function.
Peripheral innervation abnormalities and functional deficits
Despite the loss of one third of the neurons in the nAmb and axonal projections
within the recurrent laryngeal nerve in cKO mice, the remaining laryngeal motor end
plates appeared to be innervated. Based on the functional analyses of vocalizations
during development and in the adult, the remaining subset of nAmb-larynx connections
are not able to mediate normal functions. This may be due to the primary loss of a subset
of neurons in the nAmb, secondary maladaptive functional changes of presynaptic input
65
onto the remaining nAmb neurons, and/or secondary inappropriate segregation of the
functionally correct motor pools for vocalization. Innervation of the laryngeal musculature
was examined in a small number of cWT and cKO mice expressing MET
EGFP
on P7 (data
not shown). This revealed that all of the intrinsic laryngeal muscles except the cricothyroid
received innervation from MET
EGFP
expressing axons in both genotypes, suggesting that
if compensatory sprouting does occur it is supplied by the remaining EGFP+ neurons that
survive following MET deletion. The approximate one fourth of adult cKO mice that exhibit
normal vocalization indicate an adaptive process that can occur, but only sporadically.
Thus, aberrant motor neuron targeting or non-functional sprouting could underlie what
appears to be relatively normal structural innervation in the laryngeal muscle groups, and
in some instances result in recovery of function. These are very interesting possibilities
that will require extensive further anatomical, physiological and functional experiments.
There are several relevant examples in humans that reflect aberrant laryngeal
innervation. A neurological condition, synkinesis, involves co-contraction of different
muscle groups that typically do not contract together. For example, during thyroid surgery,
some patients experience recurrent laryngeal nerve injury. Subsequent nonselective
reinnervation of laryngeal muscle groups can lead to synkinesis (Crumley, 1979; Flint et
al., 1991). Synkinesis also can arise developmentally as demonstrated in mouse models
of strabismus where chemokine receptor deletion has been shown to result in improper
ocular muscle innervation by trigeminal motor neurons (Whitman et al., 2018). To our
knowledge, no studies have specifically demonstrated laryngeal synkinesis of a
developmental origin. However, HGF exhibits survival, growth promoting, and
66
chemoattractive properties on the neuronal projections innervating the developing
brachial arches (Caton et al., 2000; Ebens et al., 1996; Isabella et al., 2020). Therefore,
the loss of MET expression in laryngeal motor neurons in the nAmb during development
could allow for non-selective innervation by other motor neuron types in cKO mice and
account for the observed vocal impairments. In addition, girls with Rett Syndrome may
exhibit disrupted vagal motor circuit function, reflected in atypical feeding, swallowing,
and vocalization(Einspieler & Marschik, 2019; Morton et al., 2007). This is of interest here,
because MET transcription is attenuated by mutations in MECP2, and MET expression is
nearly undetectable in postmortem brain samples of girls diagnosed with Rett Syndrome
compared to matched controls (Aldinger et al., 2020; Plummer et al., 2013).
Developmental loss of a subset of nAmb neurons
The MET receptor mediates a pleiotropic neurodevelopmental signal with well-
established roles in neuronal proliferation and survival, chemoattraction, dendritic
elaboration, synapse formation and maturation, and critical period plasticity (Avetisyan et
al., 2015; Caton et al., 2000; K. Chen et al., 2021; Eagleson et al., 2017b; Ebens et al.,
1996; Gutierrez et al., 2004; Isabella et al., 2020; Judson et al., 2010b; F. Lamballe et al.,
2011; Lim & Walikonis, 2008; Ma & Qiu, 2020; Flavio Maina et al., 1998; Nakamura et al.,
1989; Y. Peng et al., 2016; Xie, Eagleson, et al., 2016). Consistent with a developmental
role, conditional deletion of Met in vagal motor neurons resulted in a statistically significant
reduction in the number of neurons in the embryonic nAmb, detected just a few days after
the onset of neurogenesis occurs in this population of cells (Pierce, 1973). Though we
67
cannot completely exclude that some cells in the ventricular zone express MET, cell
labeling is most evident in the mantle zone of the medulla, suggesting that MET
expression is not involved in regulating the proliferation of these neurons (Supplementary
Figure 1-2). This is consistent with our studies of the developing telencephalon (Judson
et al., 2010b) and those in the spinal cord that found the majority of MET expression to
be restricted to non-proliferating cells (Ebens et al., 1996). We suggest that the prenatal
phenotype is consistent with MET signaling being required for survival of a subset of
nAmb motor neurons, the same type of survival heterogeneity exhibited by developing
spinal motor neurons.
Virtually all motor neuronal populations are produced in excess during
development, and then undergo a period of naturally occurring cell death (NOCD),
eliminating up to half of these neurons between E11.5 and P1 (Hamburger & Levi‐
Montalcini, 1949; M. Hollyday et al., 1977; Margaret Hollyday & Hamburger, 1976; R. W.
Oppenheim, 1991; White et al., 1998). The classic model of NOCD posits that competition
for a limited supply of trophic factors produced by the target muscle serves as one of the
central mechanisms through which apoptosis is regulated. Various trophic factors have
been identified, including HGF, which protect specific neuronal subpopulations during
NOCD. For example, there is a dose-dependent survival of lumbar motor neurons with
increasing concentrations of supplemented HGF or chick muscle-derived extract (Ebens
et al., 1996). While HGF supports the survival of lumbar neurons, it has little effect on
brachial or thoracic motor neuron survival (Novak et al., 2000), demonstrating the
specificity of trophic factor matching. We suggest that the differential MET-dependence
68
for survival exhibited by different nAmb neuron subpopulations is consistent with the
findings from spinal motor neurons, lending additional support for the data reported here.
Previous studies from our laboratory demonstrate expression of HGF at nAmb target
innervation sites in the muscles of the esophagus and larynx at E13.5 and E15.5 (A.
Kamitakahara et al., 2017), positioning it as a likely candidate mediating survival or other
developmental functions in this subpopulation of nAmb neurons.
While loss of MET signaling results in a reduction of neurons in the compact
formation of the nAmb, we demonstrate that the survival of other subsets of EGFP+
neurons are not impacted. The mechanism for subset-specific neuronal survival in the
nAmb of cKO mice is not known, but the current results are consistent with subset-specific
loss of neuronal populations observed in other developmental trophic factor interactions.
For example, GDNF-deficient mice exhibit a loss of approximately one quarter of spinal
lumbar neurons despite expression of the GDNF receptors, Ret and Gfra1, in all spinal
motor neurons (Moore et al., 1996; Ronald W. Oppenheim et al., 2000). Similarly, in
addition to expression of MET, the nAmb has been reported to express receptors for a
number of other trophic factors, including p75NTR, TrkA (Gibbs & Pfaff, 1994; Koh et al.,
1989) TrkB (Q. Liu & Wong-Riley, 2013), TrkC (Helke et al., 1998), Lifr (M. Li et al., 1995),
Cntfr (MacLennan et al., 1996), gp130 (Nakashima et al., 1999a), and Gfra1 (Mikaels et
al., 2000). Here it is possible that the remaining EGFP+ neurons that express non-
functional MET in cKO mice are able to maintain their survival through expression or
upregulation of receptors for other trophic factors. Yet the surviving nAmb neurons have
limited ability to replace functionally the loss of the subset of MET-expression neurons.
69
Further studies focusing on combinatorial trophic factor expression and their impact on
laryngeal connectivity and neuronal survival will inform a greater understanding of vocal
motor circuit formation and mechanisms through which developmentally or surgically-
induced deficits in vocal function arise.
2.6 SUPPLEMENTARY MATERIALS
70
Supplementary Figure 1-1. Conditional deletion of Met results in early embryonic
loss of MET protein in vagal motor neuron axons.
In the cKO mouse model, Cre-mediated recombination facilitates excision of exon 16 of
the Met allele (the functional ATP binding site), such that only truncated, non-functional
protein is transcribed (Huh et al., 2004). To confirm loss of MET from vagal motor
neurons, immunohistochemistry was used to examine protein expression in vagal axonal
projections. While no antibodies are available to exclusively detect the portion of the
receptor transcribed by exon 16, studies from our laboratory and others have
demonstrated the signaling incompetence and rapid degradation of MET protein following
Cre-mediated recombination in Metfx/fx mice (Huh et al., 2004; Judson et al., 2010; Peng
et al., 2016). In agreement with these results, in cWT mice, abundant MET
immunolabeling (green) could be visualized in both the nAmb and axonal projections of
developing Phox2b+ vagal motor neurons (tdTomato, red) on E14.5 (A, B, E, and E’). By
contrast, in cKO mice, MET immunolabeling could only be detected in the nAmb cell
soma, but not in axonal projections, consistent with truncated MET protein being
degraded in the cell soma and not transported to innervation targets (C, D, F, and F’).
All scalebars = 100μm. n = 4 mice per group.
The brightness and contrast of each channel was adjusted separately for visualization
purposes.
71
Supplementary Figure 1-2. Early Cre-mediated recombination driven by the Phox2b
promoter in the developing brainstem.
The Cre-dependent reporter tdTomato (tdTom) was used to identify all vagal populations
following functional deletion of MET. On embryonic day (E) 9.5, tdTom+ neurons were
present in rhombomeres 7/8, the region from which vagal neurons originate (Lumsden
and Keynes, 1989), indicating that recombination and deletion of MET occurs early, just
following cell birth (Pierce, 1973). Representative images show MET protein
immunoreactivity (green) and tdTom endogenous fluorescence (red) on E9.5. r5,
rhombomere 5; r6, rhombomere 6; r7/8, rhombomeres 7 and 8. Scale bars = 100μm. The
brightness and contrast of each channel was adjusted separately for visualization
purposes.
72
Supplementary Figure 1-3. Mechanical stimulation assay to directly elicit calls.
Audio recordings were made on P7 following a tail pinch stimulus. The number of USVs
(A), clicks (B), and audible calls (C) were quantified for each genotype. Despite high
motivation to vocalize (pain), cKO pups made significantly fewer calls in the ultrasonic
73
range compared to Cre- and cWT littermates, and no calls in the audible range. However,
cKO mice did produce significantly more clicks than Cre- pups, suggesting that emitted
clicks may be generated by cKO pups in place of USVs. (D) Illustration of representative,
one second spectrograms from P7 Cre-, cWT, and cKO mice following the tail pinch. Red
arrowheads indicate clicks emitted by cKO pups.
Supplementary Figure 1-4. Respiratory pattern during vocalization.
Audio recordings made on P7 to examine isolation evoked USVs were subsequently
analyzed to examine respiratory pattern during bouts of vocalization. The percent of
breaths containing more than one call or click (e.g. event) was not significantly different
between genotypes (A). Furthermore, the time between each call was not affected by
conditional MET deletion (B).
74
Supplementary Figure 1-5. Body weight does not correlate with vocalization
disturbances observed following MET deletion.
(A) Quantification of body weight in grams (g) of mouse pups from all genotypes on P7.
n = 34 Cre-, 12 cWT, 11 cKO. Analyzed by one-way ANOVA with Tukey correction for
multiple comparisons.
(B) Scatterplot of body weights vs the number of calls recorded during the 5-minute test
period. n = 34 Cre-, 12 cWT, 11 cKO
(C) Quantification of body weight in grams (g) of male mice from all genotypes on P60. n
= 17 Cre-, 9 cWT, 7 cKO. Analyzed by one-way ANOVA with Tukey correction for multiple
comparisons.
75
(D) Quantification of body weight in grams (g) of female mice from all genotypes on P60.
n = 9 Cre-, 11 cWT, 11 cKO. Analyzed by one-way ANOVA with Tukey correction for
multiple comparisons.
Supplementary Figure 5-1. Reduction of nAmb neurons in cKO mice is not different
between sexes.
(A) tdTom+ cell counts from cWT and cKO mice on P7, by sex. n = 4 cWT males, 6 cWT
females; n = 4 cKO males, 4 cKO females. ‘ns’ indicates no significant sex effect as
analyzed by 2way ANOVA.
(B) tdTom+ cell counts from cWT and cKO mice on P60, by sex. n = 5 cWT males, 5 cWT
females; n = 11 cKO males, 6 cKO females. ‘ns’ indicates no significant sex effect as
analyzed by 2way ANOVA.
(C) EGFP+ cell counts from cWT and cKO mice on P60, by sex. n = 5 cWT males, 5 cWT;
n = 11 cKO males, 6 cKO females. ‘ns’ indicates no significant sex effect as analyzed by
2way ANOVA.
76
Supplementary Figure 6-1. Limited colocalization of MET and Phox2b outside of
the vagal nuclei.
(A, B) Representative images of the nucleus prepositus (PRP) and medial vestibular
nucleus (MV) from cWT and cKO mice on P60.
(C, D) Representative images of the facial (VII) and retrotrapezoid (RTN) nuclei from cWT
and cKO mice on P60.
(E, F) Representative images of the hypoglossal nucleus (XII) from cWT and cKO mice
on P60.
All scalebars = 100μm. The brightness and contrast of each channel was adjusted
separately for visualization purposes.
77
78
Supplementary Table 1. Simplified Genotype Categories. To simplify the genotype
names of mice used, Cre- is used to describe mice lacking the Phox2bcre allele,
conditional wild type (cWT) is used to describe any mouse with the Phox2bcre allele but
lacking the Metfx allele, and conditional knockout (cKO) is used to describe mice with the
Phox2bcre allele and two copies of the Metfx allele. ‘+’ indicates the presence of the wild-
type allele. ‘cre’ indicates the presence of the Cre recombinase allele. ‘fx’ indicates the
presence of the floxed Met allele. ‘0’ indicates the absence of a reporter allele. ‘Tg’
indicates the presence of a reporter allele.
79
CHAPTER 3: Single Cell Gene Expression Analysis of Met+ Excitatory Projection
Neurons in the Developing Cortex
R. Ali Marandi Ghoddousi, P. Levitt, Zia Rady, K. Eagleson (In preparation)
3.2 ABSTRACT
The primary visual (VC) and frontal (FC) cortices, which is composed of heterogenous
neuron populations with different molecular phenotypes and projection patterns, are
involved in diverse social and sensory behaviors. We have demonstrated that subgroups
of cortical projection neurons express Met, a pleiotropic gene that encodes the
multifunctional c-Met receptor tyrosine kinase (MET) and whose peak expression
corresponds with the period of rapid synaptogenesis. It is unknown whether Met
expressing (Met+) cortical neurons possess unique transcript signatures at these
developmental timepoints. We performed single cell RNA sequencing on mouse VC and
FC at P8, and identified developmentally important differential gene expression (DGE)
between Met+ and Met- neurons in subpopulations in superficial (layers 2-4) and deep
(layers 5-6) cortical layers in both regions. In addition, we sole ligand of MET, Hgf, is
largely confined to subsets of astrocytes in both regions.
3.2 INTRODUCTION
The mammalian cerebral cortex is composed of anatomically distinct regions with
diverse functional outputs. The functional outputs of each subregion are conveyed by
80
heterogenous populations of excitatory projection neurons (PNs), which acquire their
distinguishing morphological, topological, and projection characteristics during
development (Cadwell et al., 2019; Cheng et al., 2022b; Harb et al., 2016; Kast & Levitt,
2019; Levitt et al., 2003; Molyneaux et al., 2005; Sohur et al., 2014). Understanding how
different PN subpopulations integrate into functional circuits during development and
contribute to region-specific cortical functions such as cognitive and sensory processing,
working memory, and social behaviors requires investigation of the developmental gene
expression programs that drive cortical PN heterogeny.
Studies in developing and adult mice have discovered sets of genes that define
PN identity. These genes, which are expressed in all cells that comprise a PN subclass,
drive PN specialization during early development through control of their laminar
organization and appropriate innervation of their prospective target regions (Di Bella et
al., 2021; Du et al., 2022; Kast, Lanjewar, et al., 2019; Kwan et al., 2008; Leone et al.,
2008; Park et al., 2022; Samata et al., 2020; Tsyporin et al., 2021; Woodworth et al.,
2016). Yet PN neurons within a subclass are not homogenous. There is further
heterogeneity, in terms of genes expressed, within subclasses of PNs that reside within
the same layer and cortical subregion and project to the same targets (Ortiz et al., 2020;
Park et al., 2022; Tasic et al., 2016, 2018a; Zeng & Sanes, 2017). The developmental
roles of these genes, which further divide PNs within a subclass into distinct cell types,
has not been thoroughly investigated. To this end, we focused on the pleiotropic gene
Met, which codes for a neuropil-enriched transmembrane receptor tyrosine kinase (Met)
that has been shown to promote synapse formation and maturation, and dendritic
81
arborization and spine formation in cortical PNs (Judson et al., 2009, 2010b; Xie, Li, et
al., 2016). Despite Met’s involvement in these broad and essential developmental roles,
it is only expressed in subsets of cells within different PN subclasses and has variable
expression patterns across different cortical regions. For example, in the primary visual
and sensory cortices Met is enriched in subsets of intratelencephalic (IT) PNs across all
granular layers and in very small subsets of layer 5 subcerebral (SC) and corticothalamic
(CT) PNs (Kast, Wu, et al., 2019; Kast, Wu, Williams, et al., 2017; Lanjewar et al., 2023).
Conversely, in the medial prefrontal cortex Met is excluded from IT neurons, and primarily
expressed in subsets of layer 5 SC and layer 6 CT PNs (Lanjewar et al., 2023).
To better understand how PN heterogeneity develops at the cell-type level,
experiments were designed to transcriptomically compare Met-expressing (Met+) PNs to
similar PN subclasses that don’t express Met (Met-). Comparisons were performed in two
different cortical areas during the first postnatal week, corresponding to high Met
expression, rapid synaptogenesis, and high transcriptomic heterogeneity in PNs (Klingler,
2023; H. Li et al., 2017; Özel et al., 2020). Complimenting prior results that used mouse
driver lines and immunohistochemistry, our results demonstrate that the Met transcript is
enriched in specific excitatory neuron subpopulations in the primary visual (VC) and
frontal (FC) cortices (Eagleson et al., 2011). Our analyses also identified non-neuronal
cell types that also express Met, albeit at lower levels. We show, for the first time, the
cortical cell type responsible for the expression of Met’s only known ligand, hepatocyte
growth factor (Hgf). And finally, we identify genes that are differentially expressed in
82
different Met+ PN subclasses in a layer- and region-specific manner and relate them with
several putative developmental functions.
3.3 MATERIALS AND METHODS
Animals
All procedures were approved by the Institutional Animal Care and Use Committee
(IACUC). Mice of both sexes were used in the study. All mice were collected at postnatal
day 8.
Tissue Collection and Library Preparation
5 independent biological replicates were collected for the VC, and 7 independent
biological replicates were collected for the FC. Each independent replicate consisted of 2
animals for the VC and 5-6 animals for the FC. Similar numbers of males and females
were used in each replicate.
HABG cell culture media was prepared by combining 49ml Hibernate A, 1ml B-27
and 125ul GlutaMAX. In another tube a ~35U/ml concentration papain enzymatic
digestion solution was prepared by mixing lyophilized papain with 75ul GlutaMAX and
30mL HA-CA. This digestion solution was incubated in a 37C water bath for 30min, filter
sterilized, and set on ice.
83
P7-9 mice were anesthetized with isoflurane prior to decapitation and dissection
of the brain. The FC and VC were dissected in HABG media and cut into 0.5mm pieces.
A siliconized non-polished Pasteur pipette was used to transfer the tissues into a 15-mL
polystyrene tube containing 2ml HABG and then placed on ice. The tissue samples,
papain solution, and 10 ml of HABG media placed in 37C water bath for 10 min. To digest
the tissue in preparation for cellular dissociation, the HABG media in each tissue sample
was replaced with and equal amount of papain solution and the tissue samples re-
incubated in a 30C water bath for 30 min. The tubes containing the tissue samples were
gently inverted several times every 5 minutes during the incubation period. The papain
was removed from each tube and replaced with 2 ml of warmed HABG media and the
tubes were placed at room temperature for 2 minutes in preparation for mechanical
isolation of single cells.
The digested tissues were gently triturated ten times with a flame-polished pasture
pipette and allowed to settle for 2 minutes after which the supernatant (debris) was
removed and 2ml of room temperature HABG was used to resuspend the cells. The
trituration and resuspension steps were repeated 2 more times and the final solution was
filtered using 40-micron cell filter and then centrifuged at 200g for 10 min. The supernatant
was then removed, and the cells were resuspended in 1ml HABG.
Library preparation, sequencing, and data preprocessing
84
To prepare libraries, single cell suspensions were sent to the Spatial Biology Core
at CHLA and run through Chromium Next GEM Automated Single Cell 3′ Library and Gel
Bead Kit v3.1(10X Genomics), using manufacturer’s protocol, with the output target of
10,000 cells per sample. Libraries were sent offsite for sequencing on either the Illumina
NextSeq 500 or the Illumina Hiseq X. The resulting FASTQ files were used for sequence
alignment and expression quantification by running them through the complete Cell
Ranger pipeline (10X Genomics) with default settings. H5AD and gene expression matrix
CSV files were used for downstream analysis in R and Python.
Data Analysis
Quality control
For each brain region, gene expression CSV files were uploaded into R (version
4.2.2) and combined into a single S4 object using the Seurat package (version 4.3.0).
Scatterplots, showing the number of genes by the number of transcripts in each cell, were
inspected to identify ambient RNA in each sample. Cells with less than 325 - 450 genes
for VC and 300-700 genes for FC were identified as ambient RNA and were discarded.
The Seurat MitoQC package wrapper was used on each sample separately to filter
dead/dying cells (posterior.cutoff = 0.85, model.type = "linear", model.slot =
"flexmix_model", backup.option = "percent", backup.percent = 10). The DoubletFinder
package (version 2.0.3) was employed on each sample separately to identify heterotypic
multiplets. 6,143 heterotypic multiplets in the VC and 6,156 in the FC were identified and
85
removed. Reinspection of the # genes by # transcripts scatterplot allowed for
identification and removal of homotypic doublets and further removal of putative ambient
RNA (nCount_RNA < 30000 & nFeature_RNA < 6500 & nFeature_RNA > 650). 16,874
cells in the VC and 15,318 cells in the FC were identified as putative homotypic doublets
or ambient RNA and removed. After filtering, our VC and FC datasets contained 59,206
and 61,739 high quality cells, respectively. Ribosomal genes as well as sex-linked genes
were removed to mitigate sex and transcriptional state effects on the downstream data.
Clustering and cell type identification
Dimensionality reduction was performed on both datasets using principal
component analysis and the top 50 principal components were carried into scTransform
for data normalization. Each sample for each dataset was normalized separately using
SCTransform, then integrated using Seurat’s integration pipeline (3000 genes used,
information on anchors, etc.). Neighbors were determined for each normalized and
integrated dataset and cells were clustered using 0.32 as a resolution for both FC and
VC. Clusters were plotted using UMAP projections in Seurat.
Cells were annotated into glial (astrocytes, oligodendrocytes, and microglia),
neuronal (excitatory and inhibitory), and other neural cell types (ependymal,
macrophages, vascular, and erythrocytes) using classical cell type markers. Cells from
neuronal clusters were combined and their cell IDs were saved. Iterative re-clustering
was performed by pulling neurons from the original unprocessed dataset and running
86
them through the normalization, integration, and clustering pipeline described above. The
neuronal clusters were annotated and merged into excitatory, inhibitory, and non-
neuronal subsets based on canonical marker genes, and an additional iteration of
clustering was performed on the excitatory neurons. Excitatory neuron clusters were
annotated into deep, superficial, unknown, and non-neuronal populations using a curated
list of markers genes. High-confidence marker genes for layer and projection types were
determine for each brain region separately via a thorough literature search and secondary
validation for correct laminar and region expression patterns through inspection of the P4
and P14 Allen Brain Atlas ISH database. High-confidence markers were used to annotate
each excitatory neuron cluster.
Differential gene expression and GO analysis
Met+ and Met- cells were identified in each subgroup in each region (Superficial,
Deep, etc). Counts for each gene were then aggregated and a pseudobulk DGE analysis
was performed using a multi-factor design in DESeq2 (version 1.38.3). A Benjamini-
Hochsberg test was used to correct for multiple comparisons, and genes showing LFC >
0.5 or < -0.5, and an adjusted p-value of < 0.1 were retained for downstream analysis.
DEGs in each cluster pair were used for downstream gene ontology (GO) analysis. GO
analysis for biological processes, molecular functions, and cellular components was
performed using the ClusterProfiler (version 4.6.2) R package (pvalueCutoff and
qvalueCutoff of 0.1). The background genes for the GO analysis were those that were
expressed in the cluster (universe). Duplicate terms were initially removed by using the
87
simplify() function in ClusterProfilerGO (cutoff=0.5, by="p.adjust", select_fun=min), then
a secondary manual inspection was performed to remove any remaining duplicate terms
(similar terms called by the exact same genes) terms. The remaining GO terms were
ordered by adjusted p-value prior to plotting.
3.4 RESULTS
Met and Hgf expression is enriched in specific cellular subtypes in the developing
VC and FC
The transcriptomic heterogeneity of Met expressing (Met+) cell populations in the
developing frontal (FC) and primary visual (VC) cortices was assayed using single cell
sequencing. Microdissections of the postnatal day 8 (P8) FC and VC were obtained to
generate single-cell suspensions. A total of 15 mice in the VC and 35 mice in the FC were
loaded into 5 lanes on the 10X chromium microfluidics-based platform to generate cDNA
libraries for sequencing (Figure 1A, left). Clustering, analysis of cell type composition,
and differential gene expression analyses were performed using Seurat and DESeq2.
After quality control, normalization, and integration of batches in each region, clustering
was performed to first identify broad cell types. Further iterative clustering was performed
in cell types enriched for Met until at the level of excitatory neuron subclasses. Differential
gene expression analysis between Met+ and Met- cells from excitatory neuron
subclasses of interest was performed using a pseudo-bulk strategy in DESeq2, and
88
differentially expressed genes (DEGs) were used to perform gene ontology (GO) analysis
(Figure 1A, right).
Filtering each region’s gene expression dataset for dead/dying cells,
doublets/multiplets, and ambient RNA yielded 69,206 high quality cells in the VC and
61,739 in the FC. After normalization and integration of runs in each dataset, a shared
nearest neighbor graph was constructed and the cells were clustered using a Louvain
clustering, resulting in 26 cluster in the VC and 24 in the FC. Cluster markers were
identified using a Wilcox test and each cluster was manually annotated using a list of
classical neural cell type markers. Clusters with the same annotation were combined. In
each region we successfully identified neurons, microglia, macrophages,
oligodendrocytes, astrocytes, ependymal cells, vascular cells, and erythrocytes, which
were spatially segregated in UMAP projections (Figures 1B and 1D). Apart from small
clusters that were enriched for both neuronal and oligodendrocyte marker genes
(annotated as unknown), the remaining annotated clusters were highly enriched for
specific subsets of neural cell type markers, demonstrating the accuracy of the
annotations (Figure 1C and E).
89
Figure 1. P8 visual and frontal cortex cells cluster into classically defined neural cell types
90
A. Outline of the bench (top) and analytical (bottom) methods used to perform single
cell RNAseq experiments in P8 VC and FC.
B. UMAP plot showing the different cell types identified in the VC.
C. Dot plots showing the gene expression levels and percentage of cells in each
annotated cluster expressing defined cell-type marker genes in the VC.
D. UMAP plot showing the different cell types identified in the FC.
E. Dot plots showing the gene expression levels and percentage of cells in each
annotated cluster expressing defined cell-type marker genes in the FC.
A total of 2,491 and 3,433 Met+ cells (> 0 Met transcripts) were identified in the VC
and the FC respectively. In both regions Met expression was most highly enriched in
neurons, which showed both the highest average Met expression and the highest
proportion of Met+ cells (Figures 2A and B). In the VC, 9.0% of neurons expressed Met
at an average scaled expression level of 2.4. The second and third highest Met
expression levels by percentage were observed in ependymal cells (4.7%, -0.4 average
scaled expression) and erythrocytes (3.3%, 0.8 average scaled expression). All other
cell types in the VC had minimal levels of Met expression. Similar expression patterns
were present in the FC, with 8.6% of neurons expressing Met at an average scaled
expression level of 2.4. The second and third highest levels were observed in erythrocytes
(6.6%, 0.3 average scaled expression) and ependymal cells (5.3%, -0.7 average scaled
expression), with minimal expression in other cell types. These data demonstrate and
91
confirm prior results showing that Met expression in the mouse cortex is primarily
neuronal.
The gene coding for Met’s only known ligand, Hgf, was expressed in 1,436 cells in the
VC and 2,158 cells in the FC. Interestingly, Hgf was very highly enriched in astrocytes in
both regions, with 8.8% of astrocytes demonstrating a 2.5 average scaled expression
level in the VC and 10.5% of astrocytes having a 2.5 average scaled expression level in
FC. Hgf expression was low in other cell types in both regions (Figures 2C and D). Prior
studies showed that Hgf is expressed in deeper cortical layers at early postnatal ages
(P14), but the cell types were unknown. We show for the first time that cortical Hgf is
primarily constrained to astrocytes across these two cortical regions during early
postnatal development.
92
Figure 2. Met is enriched in neurons and Hgf is enriched in astrocytes in both regions
A. Dot plots showing the percentage of cells and the level of Met transcript in each
VC cell type.
93
B. Dot plots showing the percentage of cells and the level of Met transcript in each
FC cell type.
C. Dot plots showing the percentage of cells and the level of Hgf transcript in each
VC cell type.
D. Dot plots showing the percentage of cells and the level of Hgf transcript in each
FC cell type.
Met is expressed in excitatory neuron subclasses in the developing VC and FC
Subclustering of neuronal clusters from each region demonstrated the enrichment
of Met+ cells in excitatory neuron clusters (10.4% VC and 18.2% FC) and low expression
in inhibitory neuron clusters (2.5% VC and 2.7% FC), confirming previous histological
studies at different developmental timepoints (data not shown). Further subsetting and
subclustering of the 14,233 VC and 8,295 FC excitatory neurons yielded 11 and 9 clusters
respectively. As with the initial broad cell type clustering, the subclustering of excitatory
neurons was performed separately in each brain region.
Classical cortical layer marker genes were assessed for both layer and region
specificity at early postnatal ages (P4 and P14) using the Allen Brain Atlas, and validated
marker genes were used to annotate the excitatory neuron clusters in each region as
either superficial (layers 2-4) or deep (layers 5-6) layer neurons. Clusters that were
enriched for non-excitatory neuron markers were annotated as Neurons, Inhibitory, or
94
Glial depending on the markers. Clusters enriched for multiple regional, or cell type
markers were considered ambiguous and annotated as Unknown. Using these criteria,
we identified 2 Superficial clusters, 6 Deep clusters, and 3 Unknown clusters in the VC
(Figure 3A). In the FC, we identified 1 broad Neuronal cluster, 1 Superficial cluster, 5
Deep clusters, 1 Glial cluster, and 1 Inhibitory cluster (Figure 3B). Within each region,
clusters with the same annotations were numbered based on decreasing cell numbers
(i.e VC Deep 1 with 4,018 cells and VC Deep 2 with 3,652 cells).
The three clusters in the VC that were most highly enriched for Met were the
Superficial 1 (24.5%, 1.92 average scaled expression), Deep 3 (34%, 1.14 average
scaled expression) and Deep 5 (27.2%, 1.02 average scaled expression) clusters (Figure
3C). Of the clusters that were not annotated as either deep or superficial in the VC, only
Unknown 3 had relatively high Met expression levels (16.4%, 0.11 average scaled
expression). This cluster was enriched for multiple deep layer markers (Foxp2, Tle4,
Cdh9, Fezf2, Slc17a7, Bcl11b, and Etv1), but was also enriched for the excitatory neuron
marker Slc17a7 and the microglial marker Hexb. It was thus not included in the
downstream analysis. In the FC, the clusters that were most highly enriched for Met were
Superficial (31.2%, 2.07 average scaled expression), Deep 1 (19.63%, -0.51 average
scaled expression), and Deep 4 (22.4%, 0.04 average scaled expression) (Figure 3D). Of
the clusters that were not annotated as either deep or superficial in the FC, only the
relatively small Glial cluster (223 cells) had high Met expression (26.9%, 0.71 average
scaled expression). This cluster was not included in the downstream analysis due to its
enrichment of the astrocyte marker Aqp4.
95
Figure 3. Met is enriched in subsets of superficial and deep layer projection neurons in
each region
A. UMAP plot showing the different projection neuron subclasses identified in the VC.
B. UMAP plot showing the different projection neuron subclasses identified in the FC.
C. Dot plots showing the percentage of cells and the level of Met expression in each
VC projection neuron subclass.
96
D. Dot plots showing the percentage of cells and the level of Met expression in each
FC projection neuron subclass.
Cluster annotation using projection-type markers in adult and developmentally older
mice can inform us about projection-type enrichment for Met+ PNs at P8. Using these
markers to annotate deep clusters in the VC show that Met is enriched in SC (Deep 3,
Bcl11b) and CT (Deep 5, Cdh9) PNs. Lower levels of Met were seen in the sole deep
cluster that exclusively expressed IT PN markers (Deep 1, Etv1, 7.4%). Overall, 30% of
deep layer Met+ VC PNs could be annotated as IT, 32% as SC, and 30% as CT.
Annotation of deep clusters in the FC demonstrate that Met is enriched in SC (Deep 1,
Fezf2) and CT (Deep 4, Cdh9) PNs in the FC. None of the deep layer clusters in this
region had sole expression of markers for IT PNs. The overall breakdown of FC Met+
PNs was 54% SC and 20% CT.
We demonstrate that Met is enriched in one of two superficial subclusters in the
developing VC and in the lone superficial subcluster in the developing FC. Additionally,
Met is enriched in some but not all deep excitatory neuron subclusters (and thus
projection types) in the developing VC and FC, confirming prior finding revealing that Met
is expressed in specific subsets of layer 5 and 6 projection neurons.
97
Met+ excitatory neurons have distinct gene expression patterns in the developing
VC and FC
To transcriptionally profile different Met+ excitatory neuron subclasses in each
region, we performed differential gene expression analysis between Met+ and Met-
neurons in each Met-enriched cluster. These clusters were Superficial 1, Deep 3, and
Deep 5 in the VC and Superficial, Deep 1, and Deep 4 in the FC. Within each cluster, we
aggregated the gene counts across cells in each sample (batch/run) and used a
pseudobulk method to identify DEGs (Murphy et al., 2022; Squair et al., 2021). Genes
with a log2 fold change (L2FC) between -0.5 and 0.5 and adjusted p-values < 0.1 were
considered differentially expressed. Gm42418 and Ay036118, shown previously to be
enriched in single cell DGE analyses due to very high expression levels, were removed
from DEG list and were not considered in the final counts or the downstream analysis. In
both the VC and FC, the two deep clusters that were enriched for Met were combined
prior to the analysis, as they could not be differentiated further based on the validated
projection neuron markers (Y. Liu et al., 2020; Nguyen et al., 2021).
In the VC, differential gene expression analysis identified 167 DEGs in the
Superficial 1 cluster, 119 of which were expressed at lower levels and 48 which expressed
at higher levels in Met+ cells. In contrast, for the combined Deep 3 and Deep 5 clusters,
only 11 DEGs were identified, 4 of which were expressed at higher levels in Met+ cells
(Figure 4A). In the Superficial 1 Met+ neurons of the VC, genes that were differentially
expressed include those coding for matricellular/matrix proteins (Sparcl1 and Chgb),
98
transcription factors (Id2), cell adhesion molecules (Fat3, Pcdh8, Pcdh17, and Cdh13),
calcium-binding proteins (Rcn1), nuclear hormone receptors (Rorb), lipoprotein
processing enzymes (Lpl), long non-coding RNAs (Snhg11 and Meg3), and growth factor
binding proteins (Igfbp5) (Figure 4B). Comparison of the top 5 DEGs with higher
expression in Met+ cells showed similar average expression levels between Met+ and
Met- cells but a higher percentage of cells expressing the genes in Met+ subset.
Conversely, the top 5 DEGs with lower expression in Met+ cells showed higher average
expression levels in the Met- cells but had similar percentages of cells expressing the
gene in both groups (Figure 4C).
In the Deep 3 and Deep 5 clusters of the VC, DEGs included those coding for
transcription factors (Mef2c and Zbtb20), guanylate cyclase enzymes (Gucy1b3), calcium
binding proteins (Calb1), transmembrane receptors (Nptxr and Islr2), secreted
glycoprotein ligands (Slit2 and Edil3), E3 ubiquitin ligases (Rnf182) and glutamate
transporters (Slc1a2) (Figure 4D). Comparison of the top DEGs between Met+ and Met-
neurons showed that the differential expression was mostly due to differences in average
expression, with the percentages of cells expression each cell being similar between the
two groups (Figure 4E).
99
Figure 4. Met+ and Met- projection neurons in the superficial layers of the VC are
transcriptomically different
100
A. Diagram showing the number of DEGs identified between Met+ and Met- PNs in
different layers of the VC. The DEG counts are further split into those expressed
at lower (L2FC <0) and higher (L2FC > 0) levels in Met+ PNs.
B. Volcano plot showing the DEGs that are expressed at lower (blue) and higher (red)
levels in superficial layer Met+ PNs in the VC.
C. Dot plot showing the expression level differences of the top 10 most DEGs (sorted
by p.adj.value) from B. Dashed red line denotes the boundary between genes
expressed at higher or lower levels in Met+ cells.
D. Volcano plot showing the DEGs that are expressed at lower (blue) and higher (red)
levels in deep layer Met+ PNs in the VC.
E. Dot plot showing the expression level differences of the top 10 most DEGs (sorted
by p.adj.value) from B. Dashed red line denotes the boundary between genes
expressed at higher or lower levels in Met+ cells.
Thirty-five DEGs were identified in the Superficial cluster of the FC, 11 of which were
expressed at lower levels and 24 which were expressed at higher levels in Met+ cells. In
the combined Deep 1 and Deep 4 clusters, there were a total of 3 DEGs, all of which were
expressed at lower levels in Met+ cells (Figure 5A). In the Superficial FC Met+ neurons,
genes that were differentially expressed included those coding for cell adhesion
molecules (Pcdh7, Cndn5, and Cadm2), aminoacyl-tRNA synthetases (Lars2), actin
binding proteins (Enc1), transmembrane proteins (Tsapn13 and Laptm4a), receptor
tyrosine kinases (Epha4), GTPase activators (Elmod1), structural and cytoskeletal
remodeling proteins (Rims2 and Mical2), RNA processors (Gpatch8), transcription factors
101
(Brd9), lipid and fatty acid binding proteins and transporters (Apoe and Fabp7), long non-
coding RNAs (Gm26917), and ion transporters (Slc45a5) (Figure 5B). Comparison of the
expression levels of the top 5 DEGs with higher expression in Met+ cells showed both
higher average expression and a higher percentage of cells expressing the genes in the
Met+ cells compared to the Met- cells. In contrast, the top 5 DEGs with lower expression
in Met+ cells were expressed in similar percentages of Met+ and Met- cells but had higher
average expression in the Met- cells (Figure 5C).
The 3 genes that were differentially expressed in the Deep 1 and Deep 4 clusters
coded for aminoacyl-tRNA synthetases (Lars2), hemoglobin subunits (Hbb-bs), and
proteolipid ion channel regulators (Nnat) (Figure 5D). All three genes were expressed in
similar percentages of cells in the Met+ and Met- subsets of cells but had higher average
expression levels in the Met- subset (Figure 5E).
102
Figure 5. Met+ and Met- projection neurons in the superficial layers of the FC are
transcriptomically different
103
A. Diagram showing the number of DEGs identified between Met+ and Met- PNs in
different layers of the FC. The DEG counts are further split into those expressed
at lower (L2FC <0) and higher (L2FC > 0) levels in Met+ PNs.
B. Volcano plot showing the DEGs that are expressed at lower (blue) and higher (red)
levels in superficial layer Met+ PNs in the FC.
C. Dot plot showing the expression level differences of the top 10 most DEGs (sorted
by p.adj.value) from B. Dashed red line denotes the boundary between genes
expressed at higher or lower levels in Met+ cells.
D. Volcano plot showing the DEGs that are expressed at lower (blue) and higher (red)
levels in deep layer Met+ PNs in the FC.
E. Dot plot showing the expression level differences of the top 10 most DEGs (sorted
by p.adj.value) from B. Dashed red line denotes the boundary between genes
expressed at higher or lower levels in Met+ cells.
In both regions, superficial Met+ neurons contained more DEGs than deep Met+
neurons, suggesting more transcriptomic similarity between Met+ neurons and Met-
neurons of the same subtype in deep layers compared to the superficial layers during
development. In both the superficial and deep layers of both regions, genes and gene
classes with multiple known developmental functions were differentially expressed in
Met+ neurons, suggesting functional differences in Met+ neurons during the early
postnatal periods.
104
Developing VC and FC Met+ neurons differentially express genes involved in
numerous functional processes
To address the potential biological significance of the DEGs in Met+ neurons, and
to identify possible functional differences between Met+ and Met- cells of the same
subtype, we performed gene ontology (GO) analysis in each region. Due to the low
numbers of significant DEGs in the deep subsets in the VC and FC, only the DEGs
identified in the superficial clusters were used for GO analysis. In the VC, GO analysis
using the 167 DEGs in the Superficial 1 cluster identified 39 enrichment terms; 27 were
classified as Biological Processes (BP), 5 as Cellular Components (CC), and 7 as
Molecular Functions (MF). When sorted by increasing adjusted p-value, the top 10 GO
terms included 7 BPs outlining broad biological (cell-cell adhesion via plasma-membrane
adhesion molecules), nervous system specific (fear response and sensory perception of
sound), neurodevelopmental (axon guidance and retinal rod cell development), and non-
nervous system developmental processes (myoblast proliferation and gland
development). In addition, there were 2 CC terms defining broad (secretory vesicle) and
brain related components (presynaptic membrane) and 1 MF term involved in broad,
multi-organ functions (calcium ion bonding). Between 3-17 genes contributed to these
top 10 GO terms (Figure 6A).
105
To further identify functional class associations for the DEGs, we generated a
chord plot visualization of the identities and directionality of the DEGs that contributed to
each of the top 10 GO terms. For ease of visualization, we only plotted genes that
contributed to at least 2 GO terms. The retinal rod cell development GO term was not
plotted as none of its contributing genes overlapped with any of the other top 10 terms.
Interestingly, 6 of the 7 DEGs that contributed to the presynaptic membrane and all 6
DEGs that contributed to the secretory vesicle GO terms were expressed at lower levels
in Met+ cells (Figure 6B). When the full plot was visualized without any limits, the tubulin
subunits (Tubb2b and Tubb3), which contributed to the axon guidance GO term, were
both more highly expressed in Met+ neurons compared to Met- neurons (data not shown).
106
107
Figure 6. DEGs in superficial Met+ PNs are involved in neurodevelopmental and neuronal
activity processes
A. Bar plot showing top 10 biological processes, molecular functions, and cellular
components GO terms (sorted by p.adjusted value) from DEGs detected in Met+
superficial PNs in VC.
B. Chord plot outlining the top genes that contribute to each GO term from A, and the
direction of their differential expression (red – expressed at higher levels in Met+
cells, blue – expressed at lower levels in Met+ cells)
GO analysis on the 35 DEGs in the Superficial cluster of the FC identified 7 enrichment
terms (5 BPs and 2 CCs). The BP terms involved broad (negative regulation of response
to external stimulus, regulation of guanyl-nucleotide exchange factor activity, and fatty
acid transport) as well as brain specific processes (negative regulation of long-term
synaptic potentiation and regulation of fear response). The two CC terms were related to
broad (intrinsic components of organelle membrane) and neuron specific (synaptic
membrane) components. Between 2-6 genes contributed to these GO terms (Figure 7A).
Chord plot visualization of the GO analysis demonstrate that the intrinsic component of
organelle membrane and regulation of guanyl-nucleotide exchange factor activity GO
terms are driven by genes that are more highly expressed in Met+ neurons compared to
Met- neurons. Interestingly, two genes involved in lipid metabolism (Apoe and Fabp7)
which contribute to the fatty acid transport and negative regulation of response to external
stimulus GO terms among others, are both expressed at lower levels in Met+ neurons
(Figure 7B).
108
109
Figure 7. DEGs in superficial Met+ PNs are involved in neurodevelopmental and lipid
metabolism processes
A. Bar plot showing top biological processes, molecular functions, and cellular
components GO terms (sorted by p.adjusted value) from DEGs detected in Met+
superficial PNs in FC.
B. Chord plot outlining the top genes that contribute to each GO term from A, and the
direction of their differential expression (red – expressed at higher levels in Met+
cells, blue – expressed at lower levels in Met+ cells)
Our results identify DEGs in VC and FC Met+ neurons that are involved in several
important neurodevelopmental and neuronal functions. In the VC, superficial Met+ cells
differentially express genes with known function in axon guidance as well as synaptic
structure and signaling. In the FC, superficial Met+ cells differentially express genes
involved in lipid metabolism and neuronal signal transmission (long-term potentiation).
3.5 DISCUSSION
Met is enriched in excitatory neurons and Hgf is expressed in astrocytes
In this study we used scRNAseq to determine the cell-type specific expression
patterns of Met and identified transcriptomic differences between Met+ and Met- PN
110
populations in the two distinct areas of the developing cortex. The study focused on a
time of high Met expression that correlates with active synapse formation and dendritic
growth. Our results demonstrate that Met is most highly expressed in excitatory neuron
populations in both the VC and FC and that it is also expressed at lower levels in smaller
subsets of cortical inhibitory neurons. While prior slice culture experiments have shown
Met expression and physiological roles for Met signaling in forebrain inhibitory neurons
and interneuron precursors, in vivo studies have failed to localize Met expression outside
of excitatory neurons in the forebrain (Eagleson et al., 2011; Frias et al., 2019; Powell et
al., 2001). We have previously hypothesized that ectopic Met expression in inhibitory
neurons may be due to injury and microenvironment differences that occur during the
generation and maintenance of tissue cultures (Eagleson et al., 2017a). Thus, in the
current study the low levels of Met expression in subsets of interneurons could have
resulted from injury-related ectopic expression of Met during the single cell dissociation.
Alternatively, it is possible that there is a transient expression of Met in small subsets of
interneurons during this early postnatal period which was not captured in previous
studies. In addition to expression in neurons, we show for the first time that Met is
expressed in small subsets of erythrocytes and ependymal cells. Interestingly, Hgf has
been shown to be expressed in the deep cortical layers developmentally, which are
closely opposed to the ependymal cells that line the ventricles (Judson et al., 2009). We
show in both regions that Hgf expression is almost entirely in astrocytes at P7, opening
the possibility of an unexplored astrocyte-ependyma interaction through Met-Hgf
signaling during development. Met’s role in brain-derived erythrocytes is unexplored.
111
Previous work in our laboratory demonstrated that Met expressing neurons are
located in multiple lamina, and largely confined to IT PNs in primary sensory cortices
(somatosensory and primary visual cortices). Met expression in mPFC, an association
cortex, is primarily confined to SC PNs in layer 5 and CT PNs in layer 6 (Kast, Wu, et al.,
2019; Lanjewar et al., 2023). The results from this study validate the laminar Met
expression patterns of the VC, demonstrating Met enrichment in both superficial and deep
clusters. Surprisingly, similar patterns are seen in the FC, where Met expression was
expected to be confined to the deep layers. The presence of Met+ cells in the superficial
layers of the FC in this current study is likely due to inclusion of dorsal prefrontal cortex
and secondary motor areas in our microdissections. These areas express Met (data not
shown) superficially, and preliminary data show regional differences when compared to
the more constrained mPFC. Spatial ISH experiments at P8 will validate Met expression
in non-excitatory neuron cell types and Met distribution patterns in the in the FC.
In both cortical regions, our results show that Met expression levels are not
ubiquitous across all deep layer PN clusters; some deep layer clusters have high
proportions of Met+ cells while in others, Met expression is nearly absent. These different
deep layer clusters may represent different projection neuron subtypes or neurons in
different developmental states of maturation. Our annotation of deep clusters using
projection-type markers from adult mice showed Met is enrichment in SC and CT PNs in
the FC, matching the results from our immunocytochemical experiments in the mPFC
(Lanjewar et al., 2023). However, the projection type enrichment patterns in the
scRNAseq data in the VC do not match exactly with our previous immunocytochemical
112
studies reporting that > 80% of Met-expressing cells in the primary VC are likely IT PNs
(Satb2+/Bcl11b-) at P7 (Lanjewar et al., 2023). In this study our results indicate that Met
is expressed at lower levels in one deep IT PN cluster, and that it is highly enriched in
clusters corresponding to SC and CT PNs. We show that across all Met+ deep layer VC
PNs, 30% could be annotated as IT, 32% as SC, and 30% as CT. This discrepancy could
arise from the differential loss of deep IT neurons during the cell dissociation process, the
unsuitability of adult derived PN markers genes for application of developmental
timepoints, or the incomplete capture of all cells within a PN subtype by a single marker.
In fact, there is ~30% overlap between the IT and SC PN markers (Satb2 and Bcl11b
respectively) in the deep layers of the P7 VC. In addition, in the current experiments we
used additional adult-derived projection-subtype markers such as Cdh9 and Zfpm2 to
annotate CT PNs. These were not used in previous experiments and could theoretically
capture CT neuron subpopulations that were missed by use of a single PN marker.
Further multi-projection-type-marker spatial transcriptomics and developmental tracing
studies are needed to help shed light on projection neuron subtype composition in the
developing VC.
Met+ and Met- PNs in VC and FC superficial layers are more transcriptomically
disctinct than those in deep layers
The transcriptomic heterogeneity of PNs within deep and superficial layers
changes throughout development. Within 24 hours of neurogenesis, early-born putative
deep layer PNs are more transcriptomically diverse than late-born putative superficial
113
PNs in the primary somatosensory (S1) cortex. This difference in PN diversity between
the layers disappears by the 4
th
day after neurogenesis and reappears once again in
adulthood (Magrinelli et al., 2022). Other scRNAseq studies spanning multiple cortical
regions have identified more deep layer than superficial layer PN clusters in adult mice,
demonstrating that heterogeneity within deep layer adult PNs is greater than those in
superficial layers across all cortical areas (Tasic et al., 2018b; Yao et al., 2021). In the
early postnatal period, one study that looked at the experience-driven PN differentiation
in V1 PN during the second postnatal week showed that superficial PNs are more plastic
than deep PNs, implying more transcriptomic diversity in superficial layer neurons than
deep layer neurons around P14 (Cheng et al., 2022b). The presence of laminar difference
in heterogeneity prior to eye opening, when PNs refine their connections within their target
regions and undergo experience-independent synaptic and dendritic development and
maturation, have not been thoroughly investigated.
Although the analysis resulted in more deep than superficial clusters, we identified
more transcriptomic differences between Met+ and Met- PNs in superficial layers than in
deep layers. This pattern was consistent for both the VC and FC, suggesting a broad,
developmental phenomenon spanning multiple cortical areas. The data revealed more
deep layer PN clusters, but the higher transcriptomic heterogeneity between superior
Met+ and Met- PN subclasses could be explained by more staggered developmental
states within the superficial population than in the deep population. That is, the deep layer
PNs may be more broadly diverse in terms of projection targets and overall molecular
profiles, leading them to segregate into more distinct clusters. At the same time, PNs
114
within each of these deep clusters may be closer to their maturity ceilings and are thus in
much more similar developmental states than the superficial PNs. Conversely, the
superior PNs are more homogenous overall but are in different developmental states and
express developmentally relevant genes in a more graded manner. Deep PNs are born
1-2 days earlier than superficial PNs, and thus, consistent with reaching maturity at an
earlier timepoint. In fact, staggered developmental pattern between different PN subtypes
has been shown in the primary somatosensory cortex (S1), in which S1 neurons
projecting to the secondary somatosensory cortex are in a more mature state than those
projecting to the motor cortex at multiple early postnatal ages (Klingler et al., 2021). In
future experiments, the inclusion of more than one developmental timepoint can help
identify whether there are differing developmental trajectories in different PN subtypes
across layers in the first developmental week.
DEGs in Met+ cells are involved in development and suggest different
developmental states
In each region, genes that were differentially expressed between Met+ and Met-
subtypes in superficial layers had developmentally important roles, suggesting either
differences in developmental states or pathways between genetically distinct neuron
subtypes. For example, the VC genes involved in neuronal activity (presynaptic
membrane and secretory vesicle GO terms) were expressed at lower levels in superficial
Met+ PNs. Interestingly, synaptic refinement, the development of retinotopic maps, and
visual circuit priming in the V1 takes place prior to eye opening and are driven by intrinsic
115
spontaneous neuronal activity (Suárez et al., 2023). Given that slow-wave spontaneous
activity is first detected in subsets of layer 2/3 V1 cells at P8, the same timepoint used in
our current experiments, the lower expression of genes involved in neuronal firing in Met+
VC PNs suggest that may be less mature than their Met- counterparts (Rochefort et al.,
2009). In addition, two genes coding for the β-tubulin isotypes, Tubb2b and Tubb3, were
more highly expressed in superficial Met+ VC PNs. Both genes have been shown to be
required for cortico-cortical projections to the contralateral side of the cortex as well as
axonal branching into the appropriate layers within the target site (Cederquist et al.,
2012). These match with prior known dendrite and synaptic maturation and arborization
roles for Met. Genes involved in neuronal activity and the combinatorial effects of tubulin
isotypes will provide new avenues to explore in out attempts to understand the
mechanisms behind the development of PN heterogeneity.
In the FC, we identified three genes involved in lipid metabolism that were
differentially expressed in superficial Met+ PNs and that may be interesting targets for
future study. Acls4, the only one of the three genes that was more highly expressed in
Met+ neurons, has been shown to be involved in synapse formation through its increase
of spines in cultured hippocampal neurons. Fabp7, which was decreased in Met+
neurons, has been shown to be expressed in both astrocytes and neurons. While the
developmental and morphological roles of neuron-derived Fabp7 expression on PNs
have not been studied, KO of Fabp7 astrocytes results in decreased synapse number
and altered dendritic morphology of mPFC PNs in neurons-astrocyte co-cultures (Asaro
et al., 2021; Ebrahimi et al., 2016). Apoe expression was also reduced in Met+ PNs. To
116
our knowledge there is only one report of the developmental role of neuronal Apoe in
neurons. This study shows that Apoe mRNA is trafficked to the dendrites of cultured
hippocampal neurons and that it is involved in structural synaptic plasticity (Oh et al.,
2010). These results identify lipid metabolism as a future biological process of interest for
studying the development of neuronal diversity in the forebrain.
Finally, it was notable that in both regions, the differences in gene expression
between Met+ and Met- cells were not absolute. There were not any statistically
significant DEGs that were expressed in one population and not the other. Rather, the
differential expression resulted from graded gene expression differences, the type of gene
expression changes seen during development as neurons undergo specification and
maturation. Taken together, we demonstrate in both the VC and FC that Met+ PNs,
particularly those in the superficial layers, are in different transcriptomic states. These
differences may represent distinct developmental trajectories, or the utilization of different
gene pathways to undergo the same developmental processes or functional outputs.
Further studies can test these hypotheses across the cortex.
117
CHAPTER 4: SCAMPR, A Single-Cell Automated Multiplex Pipeline for RNA
Quantification and Spatial Mapping
Ali Marandi Ghoddousi, R., Magalong, V. M., Kamitakahara, A. K., & Levitt, P. (2022).
Cell Reports Methods, 2(10).
4.1 ABSTRACT
Spatial gene expression, achieved classically through in situ hybridization, is a
fundamental tool for topographic phenotyping of cell types in the nervous system. Newly
developed techniques allow for visualization of multiple mRNAs at single-cell resolution
and greatly expand the ability to link gene expression to tissue topography, yet there are
challenges in efficient quantification and analysis of these high-dimensional datasets. We
have therefore developed the Single-Cell Automated Multiplex Pipeline for RNA
(SCAMPR), facilitating rapid and accurate segmentation of neuronal cell bodies using a
dual immunohistochemistry-RNAscope protocol and quantification of low and high
abundance mRNA signals using open-source image processing and automated
segmentation tools. Proof of principle using SCAMPR focused on spatial mapping of gene
expression by peripheral (vagal nodose) and central (visual cortex) neurons. The
analytical effectiveness of SCAMPR is demonstrated by identifying the impact of early life
stress on gene expression in vagal neuron subtypes.
4.2 INTRODUCTION
118
The murine brain contains approximately 70 million individual neurons (Herculano-
Houzel et al., 2006). Understanding a tissue of such complexity requires knowledge about
its component parts, both at the level of single neurons and at the level of cell groups that
constitute structural and functional nodes in circuits. Analysis of the individual
characteristics of single neurons, such as their topographical organization, morphology,
and molecular signature (among other properties), can be used to categorize them into
distinct functional cell groups (Erö et al., 2018a; Zeng & Sanes, 2017). The
commercialization and ongoing development of single molecule fluorescent in situ
hybridization (smFISH) techniques (e.g. RNAscope, HCR RNA-FISH, MERFISH, etc.),
supplemented by a variety of image analysis tools and software that quantitate molecular
signal with cellular specificity, allow for multimodal comparison of transcriptomics and
histology at a single-cell level (K. H. Chen et al., 2015; Choi et al., 2018; Eng et al., 2019;
Goh et al., 2020; Ståhl et al., 2016; F. Wang et al., 2012). Existing methods for quantitative
analysis of smFISH face several challenges, including incomplete single-cell
segmentation and a loss of cell size information due to the use of the nuclear marker
DAPI to delineate cell boundaries, limited compatibility with multi-round versions of
smFISH that are required for analysis of more than 4 genes, and resource/time intensive
pipelines (Carine Stapel et al., 2016; Codeluppi et al., 2018; Dries et al., 2019; Maynard
et al., 2020; Y. Wang et al., 2021). Here, we introduce SCAMPR, a quantitative analysis
pipeline that combines commercialized smFISH (HiPlex RNAscope) with
immunohistochemistry (IHC), and employs existing and newly developed semi-
automated image processing tools for accurate segmentation and quantification of mRNA
119
expression in neurons (F. Wang et al., 2012). SCAMPR includes open-source code for
quantitative analysis and spatial mapping of mRNA expression in neurons, as well as
precise comparison of gene expression between distinct experimental groups. In addition,
neuron size is accurately preserved due to whole-soma segmentation, and SCAMPR
generates datasets that require relatively small amounts of digital storage and processing
power for downstream analysis. Lastly, the entirety of the SCAMPR pipeline, from the
benchwork to the generation of quantitative plots, takes 5 days to complete when
multiplexing 12 genes, making it possible to perform comparative, multimodal gene
expression experiments in a short amount of time.
We validate the compatibility of SCAMPR on two mouse nervous system
structures that have distinct neuronal packing densities and topographical organizations:
the jugular-nodose complex of the peripheral nervous system, hereto referred to as the
nodose ganglion (NG) for simplicity, and the primary visual cortex (V1) of the central
nervous system. In V1, we demonstrate that SCAMPR can be used to correlate single
and multidimensional gene expression to cortical layer topography and to distinguish cell
types based on soma size, gene expression, and location. We use SCAMPR in the NG
to demonstrate cell-type specific gene expression changes resulting from early life stress
(ELS). Our findings demonstrate the accuracy and utility of SCAMPR for descriptive and
comparative analysis of neuronal gene expression and topography.
4.3 MATERIALS AND METHODS
120
Lead contact
Further information and help with the pipeline should be directed to the lead contact, Pat
Levitt (plevitt@med.usc.edu).
Materials availability
This study did not generate new unique reagents or mouse lines.
Data and code availability
Microscopy data, HiPlex images, and any additional data reported in this paper will
be shared by the lead contact upon request.
HiPlex Gene-Count matrices and all code used to analyze each Gene-Count
matrix are publicly available at GitHub as of the date of publication
(https://github.com/ramin-ali-marandi-ghoddousi/SCAMPR). DOI numbers are
listed in the key resources table.
Any additional information required to reanalyze the data reported in this paper is
available from the lead contact upon request.
Experimental models and subject details
Animals
One female C57BL/6J mouse, sacrificed at postnatal day 100 (P100), was used
for the V1 cortex HiPlex RNAscope experiments. Four ELS and four CAU litters were
121
used for the ELS experiments. From each litter, 1 male and 1 female postnatal day 9
(P9) C57BL/6J mouse was chosen at random and sacrificed for the NG HiPlex RNAscope
experiments.
All experimental procedures were performed in accordance with the Institutional Animal
Care and Use Committee of The Saban Research Institute, Children’s Hospital Los
Angeles.
Any additional information required to reanalyze the data reported in this paper is
available from the lead contact upon request.
METHOD DETAILS
Sequential Steps for SCAMPR Pipeline
Tissue collection and preparation
Prior to collection of adult visual cortex and early postnatal nodose ganglion, mice
were anesthetized with a Ketamine:Xylazine mixture (100 mg/kg: 10 mg/kg) and
transcardially perfused with 4% paraformaldehyde (pH 7.4). Due to the small size of the
tissue, a dissecting microscope was used to collect the nodose ganglion from each
mouse. All collected tissue was postfixed in 4% paraformaldehyde for 2-3 hours at 4°C,
then cryoprotected overnight in 20% sucrose. Tissue samples were embedded in Tissue-
122
Tek Optimal Cutting Temperature Compound (OCT) (Sakura Finetek, Cat. #4583), frozen
over dry ice, and stored at -80°C.
20 µM cryosections were collected in six series through the entirety of the nodose
ganglion and mounted directly onto slides. For the visual cortex, 20 µM cryosections were
collected in the sagittal orientation in 4 series. All slides were stored at -20°C until
processing.
Dual Immunohistochemistry-HiPlex RNAscope and Imaging
The RNAscope HiPlex Assay was performed according to the manufacturer’s
standard protocol using the RNAscope HiPlex12 Kit (Advanced Cell Diagnostics Cat.
#324194). Tissue sections were baked for 1 hour at 60°C and dehydrated in an ethanol
series, followed by antigen retrieval (5 minutes at 100°C) and protease treatment
(Protease IV for 30 minutes at room temperature for NG, Protease III for 30 minutes at
40° C for V1). Probes for twelve target genes were hybridized for 2 hours at 40°C,
washed, and hybridized with target-binding amplifiers allowing for signal amplification of
single RNA transcripts. Hybridization with negative control probes targeting bacterial
genes was performed in parallel. The final step of the first round of hybridization attached
three fluorophores to the first three of twelve of the target genes (T1-T3). Once the
fluorophores were hybridized to the three genes, the sections were counterstained with
DAPI for 30 seconds, then mounted/coverslipped with ProLong Gold Antifade Mountant
(Jackson ImmunoResearch, Cat. #017-000-121). For the NG samples, signal detection
123
was performed in four rounds, where three target genes were labeled with three cleavable
fluorophores and imaged with a 40X water objective each round on a Zeiss LSM 710
confocal microscope. Experimenters were blind to the genotype of all animals during
imaging and signal quantification of the NG sections. For the V1 sample, signal detection
was performed in three rounds (4 probes each round) due to the availability of an infrared
fluorophore (ACD Biosciences Catalog No. 322830) and a Leica Stellaris 5 confocal
microscope with infrared detector. The V1 sections were imaged using a 40x water
objective on a Leica Stellaris 5 and automated, inter-tile registration and blending was
performed for the visual cortex images using the Stellaris imaging software (LAS X 4.1)
to resolve any coordinate discrepancies between adjoining tiles.
For each section, the gain and laser power were qualitatively optimized by the
experimenter for each channel (DAPI, Alexa Fluor 488, Atto 550, Atto 647N, AF750 – V1
only) and a 20 µM z-stack image was obtained (1.33 µM step-size). No signal was
observed in the negative control tissue sections, or after cleaving all fluorophores,
demonstrating the specificity of the assay. For the NG, only the middle 6-7 µM of each
image was used for downstream analysis to minimize packing density-related
segmentation issues.
After the sections were imaged, the coverslips were removed in 4X SSC buffer (G.
Biosciences, Cat. #786-023) and the fluorophores were cleaved using the cleaving
solution provided in the kit. A new set of fluorophores targeting the next three genes (T4-
T6 for NG, T5-T9 for V1) were hybridized onto the tissue sections, another round of DAPI
124
counterstaining was performed, and the sections were reimaged as described above.
This was repeated until all 12 target genes were imaged. In order to identify the neurons
in the NG and V1, immunofluorescence labeling was performed in the same tissue
sections following the cleaving of the fluorophores from the last round of the HiPlex Assay.
The coverslips were washed off once again, tissue sections were briefly washed in
0.005% Tween-20 (Sigma-Aldrich Cat. #P2287) in PBS before incubation for 30 minutes
in blocking solution containing 10% normal donkey serum (Jackson ImmunoResearch
Cat. #017-000-121) and 1% bovine serum albumin (Gemini Bio Cat. #700-106P) in PBS.
Slides were incubated overnight at room temperature with antibodies against the neuronal
marker HuC/HuD (1:500, ThermoFisher Scientific Cat. #A-21271, RRID:AB_221448) with
1% BSA in PBS. Slides were washed in 0.005% PBST, then incubated for 1 hour at room
temperature in Alexa Fluor 594 AffiniPure F(ab’)2 Fragment Donkey Anti-Mouse IgG
(H+L) (1:500, Jackson ImmunoResearch Cat. #715-586-151, RRID:AB_2340858) with
1% BSA in PBS. Following three washes in 0.005% PBST and one wash in PBS, slides
were counterstained with DAPI (Advanced Cell Diagnostics Cat. #324108) and
coverslipped using ProLong Gold Antifade Mountant (ThermoFisher Scientific Cat.
#36930). One final round of imaging was performed as described above to capture the
HuC/D and DAPI signals.
Image preprocessing
Each confocal LSM (NG) or LIF (V1) image file was loaded into FIJI ImageJ. A
maximum intensity z-projection was created, each channel (corresponding to a specific
125
gene / HuC/D / DAPI) was split into a separate image file, and each new image was saved
as a TIFF file. Generating flattened z-projections at this step provides major advantages
to efficiency in the downstream processing; image registration and cell segmentation
only needs to be performed once for each z-projected image, in contrast to separate
registration and segmentation of each of the multiple z-planes in each image. To increase
the speed of this step, a macro was generated in FIJI ImageJ to create and save
maximum intensity z-projections for all images in a folder or set of subfolders.
Next, images from multiple rounds of imaging the same tissue were registered
together using the ACD Biosciences Image Registration Software, following the
manufacturers protocol (ACD Biosciences Document No. 30065UM). In brief, a DAPI
image from one of the rounds of imaging was used as the reference image. DAPI images
from all other rounds of imaging were registered to this reference image, generating a
transformation matrix of coordinate conversions that was applied to the remaining gene
and HuC/D images from each imaging round to create one, unified coordinate system for
all images from all rounds. In the same software, non-overlapping regions around the
edges of the images are cropped, and each registered image is saved separately as a
TIFF file for downstream processing. An alternative registration method using ImageJ is
detailed on the SCAMPR GitHub.
Cell Segmentation and Signal Quantification
126
Manual cell segmentation was performed by opening all registered HuC/D
maximum intensity projection images in FIJI ImageJ, then using the freehand selection
tool to circle individual cells, which were each outlined as separate regions of interest
(ROIs). ROIs for each image were then saved into a ZIP file for use in later processing.
Automated cell segmentation was performed using Cellpose (Stringer et al., 2021). A
Google Colab notebook that was provided by the authors of Cellpose was modified and
used to generate the ROIs for each image based on the HuC/D images. The ROI outlines
that are generated by Cellpose were converted into FIJI ImageJ compatible ROIs using
the Cellpose_Outline_to_ROI_Converter.py script and saved as a ZIP file for use in later
processing. We demonstrate that even in densely packed tissues such as the NG, the
maximum intensity projections are suitable for cellular ROI generation as changes in cell
morphology or the appearance/disappearance of cells across the z-stack have minimal
impact on segmentation accuracy (Supplemental Figure 5A-B).
Semi-automated signal quantification using SCAMPR requires limited user inputs.
In ImageJ, background subtraction was used to remove imaging artifacts such as
autofluorescence and other noise. Next, global image thresholding was performed to
designate each pixel in the image as either signal or background. Both steps require the
selection of specific parameters: a rolling-ball radius pixel size for the background
subtraction, and a minimum threshold value for the global image thresholding. Through
trial and error, a rolling-ball radius of radius 1 works well across all images and is therefore
set as the default. This can be changed by the user when analyses are performed. Due
to differential tissue quality, imaging, and mRNA expression levels between multiple
127
genes, using only one set of threshold parameters or one automated thresholding method
does a poor job of separating signal from background for all mRNAs across multiple
tissues (Supplemental Figure 1A). For this reason, experimenter-blinded manual
selection of the minimum threshold value is required for each image prior to signal
quantification. The rolling-ball radii and threshold values for image files of interest are
saved in one CSV file, which is utilized by the FIJI ImageJ macros that perform automated
signal quantification in downstream steps.
For users that require a more time friendly alternative for determining the minimum
threshold value, we have included a semi-automated method to SCAMPR that calculates
image-specific threshold values for the entire image set by using image intensities and
manual threshold specification of one of the images. To do this: 1) calculate the mean
pixel intensity of all background corrected images in an experiment, 2) calculate the
maximum of these mean intensity values (Max Mean Intensity), 3) identify the image for
which the mean is the closest to the Max Mean (Representative Image), 4) perform
manual thresholding of the Representative Image and save the lower threshold value
(Representative Threshold), 5) divide the mean intensity of each individual image
(Individual Mean Intensity) by the Max Mean Intensity to obtain an image-specific
thresholding weight, and 3) multiply the Representative Threshold by the image-specific
thresholding weight to obtain the intensity-adjusted lower threshold (Supplemental Figure
1B-E).
𝑰𝒏𝒅𝒊𝒗𝒊𝒅𝒖𝒂𝒍 𝑻𝒉𝒓𝒆𝒔𝒉𝒐𝒍𝒅 = 𝑹𝒆𝒑𝒓𝒆𝒔𝒆𝒏𝒕𝒂𝒕𝒊𝒗𝒆 𝑻𝒉𝒓𝒆𝒔𝒉𝒐𝒍𝒅 ×
𝑰𝒏𝒅𝒊𝒗𝒊𝒅𝒖𝒂𝒍 𝑴 𝒆 𝒂𝒏 𝑰𝒏𝒕𝒆𝒏𝒔𝒊𝒕𝒚 𝑴𝒂𝒙 𝑴𝒆𝒂𝒏 𝑰𝒏𝒕𝒆𝒏𝒔𝒊𝒕𝒚
128
A FIJI ImageJ macro was developed as part of SCAMPR to quantify whole cell
mRNA signal from RNAscope experiments. The inputs for this macro, which were
generated and processed in previous steps, are as follows: 1) the maximum intensity z-
projected images each gene, 2) the cellular ROIs generated manually or through
utilization of Cellpose, and 3) the CSV file containing the optimal, image-specific rolling-
ball radius and minimum threshold as determined by the experimenter. Using this
information, these macros determine the expression of all genes in all cells for the input
images, thereby generating a Gene by Cell matrix file with the results. Each row of this
matrix corresponds to a single cell and contains the animal ID, the image ID, the ROI
(cell) area, and the expression level of each gene of interest. The gene expression level
is displayed as an area-fraction calculation (the number of pixels expressing the gene
within the cell divided by the area of the cell in pixels). This area-fraction expression value
can be used as-is for the downstream analysis or can be converted to a value
representing the number of pixels expressing the gene by multiplying the area-fraction by
the cell area (i.e. pixel coverage in a cell).
Quantitative Analysis and Spatial Mapping
The first step prior to analysis is to import the Gene by Cell matrix created in the
previous step into RStudio. Cells not expressing any genes can be removed from the
dataset if desired. Depending on the genes chosen for the experiment, this can help
129
remove noise due to misclassification of tissue imperfections or other structures as cells.
Next, the columns containing the gene-expression data can be left as area-fraction
calculations or converted into pixel coverage. These values are then natural log
normalized to reduce high-expressing gene bias during analysis and data visualization
and used to generate violin plots, co-expression charts, spatial maps, and cluster
heatmaps/dendrograms.
Violin plots: Violin plots can be generated in two ways: 1) by plotting the gene expression
levels of each gene in each cell, or 2) by only plotting the expression levels of genes after
excluding cells that do not express the gene of interest. The latter provides a comparison
of gene expression levels, while the former will give a comparison of both expression
levels as well as a relative comparison of the number of cells that express a particular
gene.
Expression, co-expression, and spatial mapping: Cellular ROIs are uploaded into R and
matched to their appropriate cells in the Gene by Cell matrix. If desired, manual
annotations that mark structural boundaries (e.g the layers in the V1) or other image
features are also generated in ImageJ and uploaded into R. The gene expression levels
for a gene of interest are binned into different expression ranges and added as a column
to the matrix. Each bin is assigned a color, and these assigned colors are also appended
to the matrix. Each cell is then plotted based on its ROI coordinates and colored according
to its gene expression bin.
130
Graphical representation of gene co-expression is generated in R by using the log
normalized gene expression values. Correlation coefficient matrices can be plotted alone
or combined with ordinary least squares regression lines for each gene pair. For these
matrices, the data can be stratified based on group, making it useful for exploratory
analysis of differences in gene co-expression between groups.
For co-expression spatial mapping, a local polynomial regression model (LOESS)
is fitted to the log normalized gene expression values of two genes of interest. Residual
distances are calculated for each cell then binned based on size, and a color code from
a red-blue scale is assigned to each bin. This color code is added as a column to the
Gene by Cell matrix. Cells are plotted according to their ROI coordinates, and the new
matrix is then used to color each cell according to its residual size bin.
Clustering and heatmaps: Log normalized data are hierarchically clustered. Depending
on the dataset, different distance measures (Euclidean, Pearson, and Spearman) can be
utilized as inputs into the clustering algorithm, and a choice can be made between multiple
agglomeration methods (Ward.d2, complete, etc.). Here, we employed Ward.d2
clustering on Spearman correlation coefficient distance measures. Cluster identities are
assigned to each cell, and heatmaps are generated using normalized gene expression
data. The rows (cells) in the heatmap are organized by hierarchical clusters.
Cellular cluster identities are appended to the Gene by Cell matrix and used in
conjunction with ROIs to spatially map cells based on cluster. Cluster identities also can
131
be utilized in conjunction with gene expression levels and cell area to generate violin and
boxplots, allowing one to look for gene expression differences between clusters.
Early Life Stress Paradigm
A limited bedding and nesting (LBN) model was used to expose pups in our
experimental group to chronic early life stress (ELS) from P2 to P9 (Peña et al., 2017). In
brief, each ELS cage is lined with a mesh platform and just enough bedding is placed
under the platform to sparsely cover the cage floor. The dam is provided with half of
standard nesting material. Combined limited bedding and nesting materials have been
shown to act as a stressor on the dam, and consequently leads to fragmented and erratic
maternal care of the pups. Care-as-usual (CAU), control litters are raised in cages with
normal bedding and nesting resources. At P9, one male and one female mouse were
chosen at random from each of 4 different litters and sacrificed for HiPlex experiments.
12 genes that are involved in classifying NG neurons into 18 putative functionally
projecting cell-types were compared between groups in these experiments (Kupari et al.,
2019).
Manual and Cellpose Segmentation Comparisons
Eight 384 x 344 pixel ROIs, containing a combined total of 307 cells with a mean
of 38 cells per ROI, were randomly cropped from different locations in the visual cortex
132
image and were used to assess the time demand and accuracy associated with manual
segmentation, automated segmentation using Cellpose, and semi-automated
segmentation by combining Cellpose and manual correction.
Time requirement experiments: To determine the time required for manual segmentation,
each image was opened in ImageJ, and the amount of time needed to segment each cell
(ROI) by hand was determined. For these experiments, each cell was segmented using
a stylus on a touchscreen laptop or tablet in combination with the free hand selection tool
on ImageJ. For Cellpose segmentation, the amount of time required to load all eight files
into the Cellpose Google Colab notebook, to run the program, and to output a ZIP file of
all ROI masks was calculated. For the Corrected Cellpose time calculation, Cellpose ROIs
were overlayed onto each image in ImageJ, and the amount of time needed to delete
inaccurate Cellpose ROIs and redraw boundaries by hand using the free hand selection
tool was calculated. To determine the total time required for Corrected Cellpose
segmentation, the time to perform ROI corrections was added to the average time for
segmentation using Cellpose alone.
Accuracy Experiments: Segmentation accuracy was calculated by comparing the ROIs
from Cellpose and Corrected Cellpose to the ROIs from the manual segmentation. Using
the centroid of each ROI, the nearest Cellpose or Cellpose Corrected neighbor for each
Manual ROI was determined in ImageJ. The percentage of overlap was calculated for
each neighboring pair. In R, the number of true positive (TPs), false positives (FPs), and
133
false negatives (FNs) was determined at a 70% overlap threshold. The F1 score was
calculated by using the following equation:
𝐹 1
=
2𝑇𝑃
2𝑇𝑃 +𝐹𝑃 +𝐹𝑁
QUANTIFICATION AND STATISTICAL ANALYSIS
All statistics were performed in R and are described in the results section as well
as the figure legends. In the segmentation method comparisons, n represents the number
of images that were used for the time and accuracy calculations in each group. In the ELS
NG comparison experiments, n represents the number of animals per group. For all
comparisons, means were compared using a Student’s T test and significance was
reached at p < 0.05.
4.4 RESULTS
Overview of SCAMPR
A comprehensive snapshot of the SCAMPR pipeline is provided, outlining the
steps involved in processing and analyzing images from a single tissue section (Figure
1). SCAMPR has two workflows: 1) an assay workflow (Figure 1A), which comprises
benchtop in situ hybridization, immunostaining, microscopy, and image pre-processing,
and 2) an image processing and data analysis workflow (Figure 1B), which includes cell
segmentation, additional batch image processing in FIJI/ImageJ, mRNA quantification,
134
and data analysis (Schindelin et al., 2012; Schneider et al., 2012). The SCAMPR assay
workflow (Figure 1A) starts with the published protocol for the Advanced Cell Diagnostics
(ACD) HiPlex RNAscope Assay (F. Wang et al., 2012). Briefly, probes targeting three
distinct mRNAs are hybridized to the tissue, each with different fluorescent signal, a DAPI
counterstain is performed, the tissue is imaged, the fluorophores are cleaved, and this
process is repeated with the hybridization of three new mRNA probes. This is repeated
in the same tissue until all genes of interest have been hybridized and imaged. If infrared
confocal imaging is available, four probes can be hybridized simultaneously. After all
probe signals have been imaged, DAPI counterstaining and HuC/D immunochemistry are
performed to mark the boundaries of cellular nuclei and the neuronal cytoplasm,
respectively, and the tissue is imaged one final time. Flattened, maximum intensity
projection images from the different hybridization/staining rounds are spatially registered
(overlayed) using the DAPI signal with the ACD RNAscope HiPlex Image Registration
Software or in ImageJ. The resulting registered images then proceed through the
SCAMPR Image Processing & Data Analysis Workflow (Figure 1B).
In this second workflow (Figure 1B), cell segmentation and batch image processing
are performed. The HuC/D image is used to segment the cells into individual regions of
interest (ROIs). These ROIs may be achieved by manual drawing in FIJI/ImageJ, or to
save time, by using an open-source automated cell segmentation program such as
Cellpose (Stringer et al., 2021). In parallel, the images containing the mRNA signal are
visually inspected by the user in ImageJ to determine the optimal parameters for applying
a threshold to convert the mRNA signal into a binary signal/no-signal image. To add
135
flexibility to this process, an alternative, semiautomated method was designed to
determine the thresholding parameters (Supplemental Figure 1). The ROIs are then used
in combination with these parameters to quantify signal for each mRNA probe on a single
cell level in each tissue section. New macros were written to automate this process in
ImageJ. These macros produce a “Gene by Count” matrix in the form of a CSV file, which
is used for subsequent data analyses. The data analysis component of the SCAMPR
pipeline contains numerous methods to quantify gene expression, co-expression, and
global expression patterns, as well as methods for spatial mapping and comparative gene
expression analysis between groups. Each analysis method has its own R script that can
be accessed on GitHub.
136
137
Figure 1. Schematic of SCAMPR pipeline
(A) Assay workflow. (1) All rounds of HiPlex RNAscope are performed and sections are
imaged. (2) HuC/D IHC is performed. (3) Images from all rounds are pre-processed and
registered using the DAPI signal as the anchor.
(B) Image processing and data analysis workflow. (1) Batch cell boundary segmentation
using Cellpose. (2) Images are inspected by the user, then processed using macros to
automate gene expression quantification for each cell in ImageJ. (3) Gene expression
(top) and co-expression, clustering (middle), and spatial mapping (bottom) are performed
in R. Scale bars: 100 μm.
Dual IHC-RNAscope accurately demarcates cell boundaries in peripheral and
central nervous system tissues
HuC and HuD are mRNA-binding proteins that are involved in transcript regulation,
and are selectively expressed in neurons starting in embryonic development, initially
appearing in progenitor cells, and continuing to be expressed in postmitotic neurons into
adulthood (Okano & Darnell, 1997). HuC and HuD are present at the protein level in both
the nucleus and cytosol (Diaz-Garcia et al., 2021). The whole-cell expression, neuron
specificity, and broad expression time course of HuC/D make them excellent cell
boundary markers for quantitative studies that require neuronal segmentation. Two
experiments employing SCAMPR demonstrate that HuC/D IHC is compatible in multiple
nervous system tissues at different developmental timepoints. First, multi-round HiPlex
138
RNAscope is successfully applied in postnatal day 9 (P9) nodose ganglion (NG) of the
peripheral nervous system as well as in adult primary visual cortex (V1) in the central
nervous system (Figures 2 A, B, E, and F). Subsequent HuC/D immunostaining is
compatible with HiPlex RNAscope in these same tissues and accurately labels neuronal
cell bodies following multi-round HiPlex RNAscope (Figures 2 C and G). Comparison of
HuC/D labeling to nuclear DAPI labeling demonstrates that HuC/D more precisely
captures the complete individual cellular profile than the nuclear DAPI signal, resulting in
accurate localization of single-cell RNAscope signals in both tissues (Figures 2 C & D,
and G & H).
Additional validation of RNAscope-HuC/D co-staining was performed in the
embryonic NG and early postnatal brainstem (data not shown). In all instances, intense
HuC/D signal allowed for distinct segmentation between individual neurons. There is
minimal immunolabeling of neuronal processes, further enhancing the ability to
distinguish between closely apposed neuronal somata.
139
140
Figure 2. Combined IHC and HiPlex RNAscope allows for high-resolution, single-
cell mRNA expression in the central and peripheral nervous system
(A) HiPlex RNAscope visualized in postnatal day 9 nodose ganglion. Scale bar: 100 μm.
(B) High-magnification inset from (A). Scale bar: 15 μm.
(C and D) High-magnification insets from (A) co-visualized with HuC/D (C) and DAPI (D).
Scale bar: 10 μm.
(E) HiPlex RNAscope visualized in adult mouse visual cortex. Scale bar: 100 μm.
(F) High-magnification inset from (E). Scale bar: 5 μm.
(G and H) High-magnification inset from (E) co-visualized with HuC/D (G) and DAPI (H).
Scale bar: 5 μm.
Automated methods for accurate segmentation of single neurons
Quantitative analysis of spatial transcriptomics data requires segmentation of cells
into single, discrete units. The most accurate segmentation of cells in a tissue section will
be obtained with more precise cell boundary markers that demarcate the cytoplasm,
rather than using a single organelle label such as the nucleus. DAPI marks cell nuclei,
and because mRNA is localized to both nuclear and cytosolic cell compartments,
segmentation based on the nucleus can result in mRNA fluorescent signal that is
unassigned to a cell profile and therefore not measured. Using expanded nuclear labeling
(cytoDAPI) allows for more accurate segmentation of whole neurons but makes it difficult
to separate neurons from non-neurons in densely packed tissues with large amounts of
141
non-neuronal cell types (Y. Wang et al., 2021). Using the localization of two genes,
manual segmentation was performed to demonstrate potential differences in mRNA
signal detection between nuclear segmentation using DAPI signal and whole-cell
segmentation using HuC/D staining after HiPlex RNAscope in a sample section of the V1
cortex (Figure 3A). The HuC/D signal allows for segmentation of the cell body, and more
accurately encircles the entirety of the mRNA signals (here Pvalb and Chd13) in the
segmented cells compared to the DAPI signal (Figure 3B). Quantification of gene
expression after DAPI segmentation and HuC/D segmentation on a subset of images
further demonstrates that the HuC/D segmentation captures more of the RNA signal and
therefore higher average gene expression across most of the genes assayed
(Supplemental Figure 2A). The size of this increase is not consistent across all genes.
Thus, some genes such as Cckar are affected more than others by the incomplete capture
of RNA signal after DAPI segmentation (Supplemental Figure 2A). These gene-specific
differences are likely due to the predominant cytosolic localization of certain transcripts
such as Cckar (Supplemental Figure 2B-C).
Furthermore, using HuC/D staining for segmentation facilitates selective capture
of signal in neurons, as both the V1 and NG contain large numbers of non-neuronal cells.
Segmentation of a portion of the V1 cortex using the nuclear DAPI signal results in a 28%
increase in segmented cells (1,213 HuC/D vs 1,547 DAPI). This difference is much larger
in the NG, where there was a 209% increase in segmented cells (398 HuC/D vs 1230
DAPI), likely due to the large number of non-neuronal satellite cells.
142
The time requirement for manual segmentation increases dramatically as the
surface area of the tissue of interest and the cell packing density increases. Popular
automated segmentation methods such as watershed segmentation greatly reduce this
time requirement but perform poorly on images of densely packed cell bodies, a particular
challenge in most developing and certain adult tissues. More recent methods such as
StarDist are trained on specific image sets (DAPI signal) and require further user-guided
training to accurately define whole-cell boundaries that are more closely apposed
(Weigert et al., 2020). Cellpose, a generalized, deep learning, automated segmentation
algorithm that has been trained on multiple different image sets, obviates these
challenges (Stringer et al., 2021). Application of Cellpose on HuC/D immunolabeling
signal produces highly comparable results to manual segmentation in the same tissue
sections. Small deviations (missed cells, incorrect boundaries, etc.) can be corrected
manually after Cellpose segmentation (Corrected Cellpose) (Figure 3C). On a per cell
basis, data from two different operators demonstrates that there is an approximately 20-
fold difference in segmentation time between using Cellpose and manual segmentation,
with Cellpose segmentation requiring 0.25 seconds per cell and manual segmentation
requiring 4.77 seconds per cell. Corrected Cellpose takes ~6 times longer than Cellpose
alone but is also ~3 times faster than manual segmentation (Figure 3D). The time savings
estimates from limited ROIs are likely an undervaluation, as the rate of Cellpose
segmentation scales up in a non-linear fashion with an increase in number of analyzed
cells. Thus, Cellpose segmentation on ~6,500 cells using the entire V1 cortex tissue
section results in time reduction from 0.25 to 0.085 seconds per cell. This translates into
Cellpose segmentation being approximately 56 times faster than manual segmentation.
143
The calculation of an F1 score, representing the accuracy of Cellpose and Corrected
Cellpose segmentation as compared to manual segmentation, demonstrates the
precision of this segmentation method (See Methods for details). To generate the F1
score, ROIs that showed a ≥70% overlap with the manually generated ROIs were
considered true positives (Caicedo et al., 2019). This comparison resulted in an F1 score
of 0.76 for Cellpose and 0.79 for Corrected Cellpose (Figures 3D and E). To better
understand whether this accuracy affected the overall dynamics of segmentation in each
image, the number of segmented cells and the distribution of ROI areas were compared
between each method. Cellpose and Corrected Cellpose resulted in only 1.6-3.0% fewer
ROIs than manual segmentation, and the distribution of the ROI areas of the neurons
between the three methods were nearly identical, providing further evidence for the
overall accuracy of the fully automated segmentation method compared to the manual
method (Figure 3F). These data demonstrate that Cellpose segmentation alone can be
employed for both high segmentation accuracy and substantial time-savings.
144
Figure 3. Automated segmentation of cytoplasm of single cells using Cellpose
(A) Differences in nuclear segmentation boundaries (red outline) and whole-soma
segmentation boundaries (green outline) in V1 cortex.
145
(B) Nuclear signal that is present in the cytoplasm for two genes is not included when
using nuclear segmentation.
(C) Visualization of whole-cell segmentation. Automated Cellpose segmentation is
compared with manual segmentation and a mix of Cellpose and manual segmentation
(Corrected Cellpose). Red arrowhead denotes a cell identified during manual
segmentation but not during automated Cellpose segmentation.
(D) Mean cell counts, time burden, and F1 scores for manual, Cellpose, and Corrected
Cellpose segmentation methods. n = 8 images per group.
(E) Mean (red text) and individual F1 scores for ROIs generated by Cellpose and
Corrected Cellpose segmentation compared with the ground-truth ROIs generated by
manual segmentation. n = 8 images per group.
(F) The total number of ROIs generated after Cellpose, Corrected Cellpose, or manual
segmentation (see total number of cells in red). Note that cell area distributions are nearly
indistinguishable between the three methods. n = 8 images per group.
Quantification and spatial projection of pairwise mRNA expression patterns
One of the main advantages of SCAMPR is the ability to accurately quantify
multiplexed fluorescent in situ signal on a single cell level while preserving the
topographic location of each cell in the tissue. The pipeline provides an opportunity to
analyze both average gene expression and expression topography with single cell spatial
resolution across all neurons in a single tissue section. Single neuron expression of nine
146
different genes was quantified in V1 (Figure 4A). The analysis of V1 not only reveals
differences in mean gene expression between transcripts (e.g. Slc17a7 vs Chat), but also
demonstrates that for some genes with similar mean expression (Chd13 vs Piezo1), the
neurons have distinct expression distribution ranges. Specifically, Piezo1 expression is
normally distributed and exhibits a smaller distribution range while Chd13 expression is
skewed leftward towards zero and exhibits a larger overall range of expression (Figure
4A).
SCAMPR also facilitates the mapping of gene expression and gene co-expression
patterns at single-cell spatial resolution onto the tissue of origin. This is demonstrated by
using SCAMPR to map 3 genes, Met, Slc17a7, and Pvalb, each having distinct levels of
expression and neuronal distributions in V1 cortical layers. Consistent with prior results
from our laboratory, neurons expressing the highest levels of Met are distributed
predominantly in layer 2 and those expressing the lowest levels of Met localize to layer 4
(Figure 4B) (Eagleson et al., 2016a; Judson et al., 2009). Slc17a7 is expressed in neurons
across all layers in V1, with slightly lower expression in layer 4 (Figure 4C). Neurons
expressing Pvalb at high levels also are scattered across all layers of V1, with noticeable
enrichment in layer 5 (Figure 4D). In addition to single gene spatial mapping, SCAMPR
also can be used to quantify and spatially map gene co-expression patterns. Gene co-
expression was quantitated by generating pairwise linear correlation coefficients between
all nine genes in V1. A variety of relations are revealed, with some genes highly co-
expressed (Slc17a7 and Met), and others discordant (Slc17a7 and Pvalb) (Figure 4E). In
V1, Slc17a7 and Pvalb are known to be expressed in two different neuronal populations
147
(excitatory vs inhibitory neurons), whereas Met is highly enriched in excitatory neurons.
Comparison of the expression levels of these genes across the layers of V1 at single cell
resolution as well as co-expression analysis confirms these prior findings. As an added
dimension, the accurate segmentation of neuron cell bodies, rather than nuclei, provides
cell size (ROI_Area) as a variable in this co-expression analysis. Matching prior reports
in AMPA and NMDA gene expression in motor neurons, the quantitative expression levels
of most genes in V1 exhibited positive correlations with cell area (Rana et al., 2020). One
obvious exception was noted, with Vip (expressed in inhibitory neurons) exhibiting
expression levels that were independent of cell size (Figure 4E).
A unique feature of data analysis using SCAMPR is the single-cell, topographic
characterization of gene co-expression patterns between two genes. This is
accomplished by fitting a trend line to the co-expression scatterplot of the two genes, and
subsequently measuring residual distances between each cell and the trend line. Here,
the residual distances were mapped back onto the cells in the tissue. Fitting a locally
estimated scatterplot smoothing (LOESS) line to the Met and Slc17a7 co-expression
scatterplot shows a positive relationship between Met expression and Slc17a7 expression
for the large majority of Met/Slc17a7 co-expressing cells. Cells that fall on or near the
trend line represent the majority or “expected” co-expression pattern for Met and Slc17a7,
whereas those that fall far from the trend line express relatively more or relatively less
Met and Slc17a7 than expected. On the tissue section rendition, these subtypes were
marked in increasingly darker shades of red if they expressed more Met and/or less
Slc17a7 than expected, and increasingly darker shades of blue if they expressed less
148
Met and/or less Slc17a7 than expected (Figure 4F). The positive (red), negative (blue),
or neutral (grey) distances of these cells from the trend line were then mapped back onto
their location in the V1 tissue, showing that the Met/Slc17a7 ratio was larger than the
trend in layer 2 and smaller than the trend in multiple layers, particularly in layer 6a (Figure
4G). To validate the accuracy of this method for characterizing both trends and thus, the
heterogeneity in the co-expression pattern of two genes, high-magnification insets from
layers 2 and 5 of the spatial map were compared to the original HiPlex RNAscope images
from the same regions (Figure 4H).
149
150
Figure 4. Analysis and spatial visualization of gene expression and co-expression
of HiPlex RNAscope
(A) Violin plot of log-normalized gene expression for nine genes expressed in V1 cortex.
(B) Spatial map of log-normalized Met expression in the V1 cortex. Note the greater
expression in layers 2–3 and low expression in layer 4.
(C) Spatial map of log-normalized Slc17a7 expression in V1 cortex. Note the greater
expression in layers 5 and 6 and medium-low expression in layer 4.
(D) Spatial map of log-normalized Pvalb expression in V1 cortex. Note high-expressing
red profiles scattered across all layers, with an enrichment in layer 5.
(E) Pairwise correlation plot for log-normalized gene expression in V1 cortex. Asterisk
denotes correlation between Met and Slc17a7, which is analyzed further in (F) and (G).
(F) LOESS regression plot between log-normalized Met and Slc17a7 expression in V1
cortex. Points denote single cells and are colored based on their distance from the
regression line (residual size).
(G) Spatial map of residuals from (F). Red cells have large positive residuals and express
more Met and/or less Slc17a7 than is predicted by the model. Blue cells have large
negative residuals and express less Met and/or more Slc17a7 than is predicted by the
model. Note layer 2 enrichment of red cells.
(H) Left: high-magnification micrographs, denoted by green (layer 2) and blue (layer 4–5
border) boxes in (G). Top image illustrates the relatively high expression of Met (green)
compared with Slc17a7 (red) in layer 2. The resulting large positive residuals (pink/red)
are shown in the top right graphic. Bottom image illustrates cells with a high Slc17a7 (red)
151
in layer 4, lower co-expression in one profile, and higher expression of Met (green) in a
layer 5 neuron. These expression patterns match with their residual-denoted colors in the
spatial map (bottom right).
Clustering and spatial mapping of high dimensional mRNA expression patterns
In addition to single and dual gene expression patterns, SCAMPR can utilize
hierarchical clustering methods to categorize neurons into groups based on global gene
expression patterns. Using nine genes assayed in the V1 cortex HiPlex RNAscope
experiments, SCAMPR was used to cluster the cells into eight distinct groups. The
clusters are represented as a joint dendrogram/heatmap (Figure 5A). Because ROI
identity and spatial coordinates are preserved during the clustering, each cluster can be
spatially mapped back onto the tissue section. Notably, some of the clusters segregate
spatially. Cells that comprise cluster 1 are predominantly located in layer 4, whereas cells
in cluster 8 are predominantly located in layers 5 and 6a (Figure 5B). Additionally, the
cells that comprise cluster 8 are larger than those in cluster 1 (Figure 5C). Cells in cluster
8 also exhibit either higher or equivalent log normalized gene expression when compared
to cells in cluster 1, with prominent differences in genes such as Cdh13 and Foxo1 (Figure
5D). Examination of the original images for mRNA fluorescence validates significantly
higher Cdh13 and Foxo1 expression in deeper compared to superficial layers, with the
highest co-expression in layer 5 and a near absence of these two genes in layers 2-4
(Figure 5E).
152
Figure 5. High-dimensional clustering analysis of HiPlex RNAscope data
(A) Heatmap of scaled gene expression in V1 cortex after hierarchical clustering.
(B) Spatial maps of cells in cluster 1, which are mostly localized in layers 2/4/6, and cluster
8, which are predominantly localized in layers 5/6a.
153
(C) Boxplots demonstrating differences in cell size (area) between the clusters. Largest
difference in cell size is between cluster 1 (brown) and cluster 8 (green). Data are
represented as medians.
(D) Violin plots demonstrating higher expression of several genes in cluster 1 compared
with cluster 8.
(E) Hiplex RNAscope image of Chd13 (white) and Foxo1 (red), which are enriched in
cluster 8 and absent in cluster 1. Note that both genes are more highly expressed in layers
5 and 6a, the primary location of cluster 8 cells.
154
Figure 6. Utilization of HiPlex RNAscope for comparison of gene expression at
single-cell resolution in nodose ganglion neurons from CAU and ELS experimental
groups
(A and B) Pvalb/Scn1a and Ntsr1/Scn10a co-expression scatterplots for all cells in the
dataset. Points denote single cells in CAU (blue) and ELS (red) mice. Solid blue and red
lines denote the LOESS lines of CAU and ELS mice, respectively. Pearson’s r and
corresponding p values for the correlations were determined for each group separately.
155
(C) Mean, per-animal Pvalb expression in ELS and CAU Scn1a-expressing (left) and non-
Scn1a-expressing (right) cells (n = 4 per group). No statistically significant difference
noted between groups. Data are represented as medians.
(D) Mean, per-animal Ntsr1 expression in ELS and CAU Scn10a-expressing (left) and
non-Scn10a-expressing (right) cells (n = 4 per group). No statistically significant
difference noted between groups. Data are represented as medians.
(E) Percentage of cells from CAU and ELS animals that express Pvalb in Scn1a+ (left)
and Scn1a− cells (right) (n = 4 per group). Note the statistically significant decrease in the
percentage of neurons expressing Pvalb in Scn1a+ neurons in the ELS mice. Data are
represented as medians.
(F) Percentage of cells from CAU and ELS animals that express Ntsr1 in Scn10a+ (left)
andScn10a− cells (right) (n = 4 per group). No statistically significant difference noted
between groups. Data are represented as medians.
Using SCAMPR to identify gene expression pattern differences between two
experimental groups
Differences in expression levels of specific genes due to a manipulation may either
vary across all neurons that express a specific transcript, or in specific subtypes, and can
be resolved using quantitative analyses at the single cell level. Thus, high dimensional in
situ hybridization can be used to compare the expression of a carefully selected set of
genes, such as cell subtype markers, to determine potential phenotypic differences that
occur due to an experimental manipulation between groups. Here, we used a well-
156
validated model of limited bedding and nesting (LBN), applied from postnatal day (P)2-9
in mice. LBN disrupts maternal care and produces an early-life-stress (ELS) response in
pups (Eagleson et al., 2020; Heun-Johnson & Levitt, 2016; Rice et al., 2008). Care-as-
usual (CAU) pups were raised under normal conditions. Development of vagal circuitry is
sensitive to ELS, though information is limited at a molecular level (Banihashemi &
Rinaman, 2010; Card et al., 2005). To investigate prospective ELS-induced vagal gene
expression changes in early postnatal mice, we used SCAMPR to analyze the expression
of 12 genes that demarcate adult subtypes of vagal sensory neurons located in the
nodose ganglion (NG) of ELS and CAU mice (Kupari et al, 2019). Tissue was harvested
for processing on postnatal day (P) 9 at the end of the period of ELS exposure. Pairwise
correlation coefficients and least squares regression analyses were performed for both
groups separately to identify possible gene co-expression differences between groups
(Supplemental Figure 3). Visual inspection of the plot matrix was sufficient to narrow
possible differences in gene co-expression between CAU and ELS. Analyses began with
the identification of well-correlated gene pairs for all cells from all groups (R > 0.40). From
these, gene pairs that had both prominent between-group differences in correlation
coefficients and linear regression line slopes were selected. Two gene pairs fit these
criteria: Scn10a/Ntsr1 and Scn1a/Pvalb (Supplemental Figure 3). To identify differences
between the co-expression of these genes in CAU and ELS groups, scatterplots of
Scn10a/Ntsr1 co-expression and Scn1a/Pvalb co-expression were generated, and
LOESS models were fitted for each group. These plots demonstrated lower Pvalb and
Ntsr1 expression with rising values of Scn10a and Scn1a in the ELS group when
compared to the CAU group (Figure 6A and 6B). Since Scn10a and Scn1a have been
157
shown to be broad cell-type markers in the NG, and because Pvalb and Ntsr1 have been
shown to be differentially expressed in NG populations with varying putative functions
(control of pulmonary volume vs gastrointestinal chemo/mechano sensation), these
analyses at single-cell resolution suggest that there are subtype-specific decreases in
Pvalb and Ntsr1 expression in NG cells after early life stress exposure (Kupari et al.,
2019).
The subtype-specific reductions observed could be driven either by an average
decrease in expression of Pvalb or Ntsr1 across many cells, or a decrease in the
proportion of cells that express each gene. To investigate these possibilities, Scn1a-
expressing (Scn1a+), Scn10a-expressing (Scn10a+), non-Scn1a-expressing (Scn1a-),
and non-Scn10a-expressing (Scn10a-) data subsets were generated and mean Pvalb
and Ntsr1 expression was calculated for NG neurons from each ELS and CAU pup. The
averages of these mean expression values were compared between groups for each
subset. There was no difference in mean Pvalb and Ntsr1 expression in the Scn1a- and
Scn10a- subsets of neurons, and while mean Pvalb and Ntsr1 expression in ELS mice
trended towards a decrease in in the Scn1a+ and Scn10a+ subsets of neurons, these
differences did not reach statistical significance (Figure 6C and D).
It has been shown that the proportion of cells expressing a particular gene can be
altered in an experience-dependent manner during development (Cheng et al., 2022b).
To determine whether the reduction of Pvalb and Ntsr1 expression in ELS animals reflects
a reduction in the proportion of cells expressing the genes rather than overall mean
158
expression levels in the specific subpopulations, the percent of cells expressing Pvalb
and Ntsr1 in the different cell-type specific subsets was quantitated (Figure 6E and 6F).
In Scn1a+ cells, there was a significant 24.1% decrease in the percentage of cells
expressing Pvalb in ELS animals when compared to CAU animals (51.8% CAU vs 39.3%
ELS, p ≤ 0.05), whereas in Scn1a- cells, there was no significant difference between
groups (23.5% CAU vs 19.7% ELS, p = 0.56) (Figure 6E). In both Scn10a+ and Scn10a-
cells, there was a trending decrease in Ntsr1-expressing cells in ELS animals when
compared to CAU animals (64.7% CAU vs 52.2% ELS, p = 0.12 and 25.8% CAU vs
21.1% ELS, p = 0.38) (Figure 6F). These analyses suggest that in ELS animals, Pvalb
expression is reduced in a subset of NG cells that express Scn1a and that this reduction
is partially driven by a reduction in the fraction of cells expressing Pvalb transcript. In
addition, broad gene expression comparisons of the remaining ten genes demonstrated
a decrease in mean Gpr65 expression in ELS animals, yet no difference in the percentage
of cells expressing this gene (Supplemental Figure 4). Together, these experiments
demonstrate that SCAMPR can be used as a tool to quantitatively compare global and
cell-type specific differences at single cell spatial resolution, as well as gene co-
expression patterns across experimental groups.
4.5 DISCUSSION
Here we present SCAMPR, an open-source, user-friendly, time-saving, bench-to-
desk pipeline for accurate quantification and analysis of spatial neuronal gene expression
with single-cell resolution. The quantification and analytical tools that comprise SCAMPR
159
allow for comparison of the expression patterns of multiple genes, spatial mapping of
relative gene expression and co-expression patterns across a tissue, and cell-type
specific comparisons of gene expression between experimental groups, all at single
neuron resolution. The comparative analyses of cell labeling methods demonstrated
superior savings of time, fully completing bench and analytical work in approximately 5
days, without sacrificing accuracy, while using fully automated or semi-automated
segmentation of the entire neuronal soma.
Achieving accurate spatial analyses at single neuron resolution
Utilization of HuC/D staining in the SCAMPR pipeline allows for demarcation of
neurons across multiple developmental timepoints and multiple subregions of the nervous
system that exhibit different cell packing densities and spatial organization patterns,
allowing broad application of SCAMPR for analyzing topographic gene expression in
neurons. The stability of the anti-HuC/D antibody under the harsh protease digestions
and antigen retrieval processing steps of HiPlex RNAscope allows for HuC/D staining at
the very end of hybridization protocol, circumventing the early occupation of an imaging
channel or the need for protein stain degradation/quenching or fluorophore cleaving.
Other sufficiently stable primary antibodies marking glial cells in the nervous system, or
other antibodies marking non-nervous system cell classes present in other tissue types
can be validated for tissue processing stability and used with SCAMPR to quantify gene
expression differences in myriad cell types.
160
SCAMPR utilizes a combination of Cellpose and ImageJ scripts to quantify gene
expression for each neuron. In contrast to methods that perform puncta counting, which
rely on intense particles standing out from their surroundings, SCAMPR employs a
detection-by-thresholding method to quantify the number of pixels in a binary image that
are positive for a fluorescent signal representing a particular gene. Because adjoining
fluorescent molecules do not need to be separated from one another based on intensity
in detection-by-thresholding, this method allows for accurate quantification of genes
expressed at low or high levels within the same tissue section or even in adjacent
neurons. Furthermore, SCAMPR quantifies gene expression using flattened, z-projected
images, and outputs gene expression as the percentage of the cell area or the number of
pixels within a cell area that are positive for a gene. This contrasts with some puncta
count methods which match puncta with cellular ROIs by assigning x, y, and z coordinates
to each individual puncta, thus generating large data matrices that increase the
computational burden in the requisite downstream analyses. This is an important feature
when considering data management and computational time savings. Lastly,
segmentation based on the use of both DAPI and HuC/D, in conjunction with modified
versions of the ImageJ quantification macros, can be used to obtain cell compartment
information by separately quantifying nuclear and cytosolic mRNA signals.
Spatial mapping tools enable visualization of complex expression, co-expression,
and cell-type clustering patterns
161
The capability of SCAMPR for fast and accurate segmentation of neurons and
quantification of in situ hybridization signal were combined with its tools for spatial
analysis of gene expression, co-expression, and clustering driven cell-type localization in
the V1, a highly topographic structure with a laminar organization. We posited that
challenging the ability of SCAMPR to accurately segment, measure, and map gene
expression in a highly organized brain structure that is characterized by region-specific
differences in packing density would reveal strengths and weaknesses of the pipeline.
The data demonstrate the capabilities of SCAMPR for detecting layer-enriched
differences in the expression of single and multiple genes in this brain region. Analyses
of the co-expression of two genes (Met and Slc17a7), in which each cell was binned
based on residual proximity to a fitted line, facilitated spatial mapping of the general co-
expression patterns of these genes in the majority of neurons. Using this same analysis
method, SCAMPR was able to extract minority populations of neurons with more unique
co-expression patterns based on their deviation from the general co-expression patterns
in the majority population of neurons. This holistic analysis method will compliment more
direct strategies of separating neuron populations based on gene co-expression patterns,
such as the semi-quantitative categorization of neurons based on the expression levels
of both genes (low-low, low-high, high-low, and high-high), or by considering their
expression distributions and summary statistics (min, max, mean, quartiles, etc). Lastly,
it is possible to integrate HiPlex RNAscope and other smFISH data with published single
cell RNA sequencing datasets (Hashikawa et al., 2020; Y. Wang et al., 2021). These
integrated datasets could theoretically be used with SCAMPR to spatially map and
validate the expression levels of genes from single cell RNA sequencing experiments.
162
SCAMPR enables analysis of differentially expressed genes in subsets of cell types
SCAMPRs utility for comparing gene expression between neurons developing
under different environmental conditions was demonstrated. In the NG, a tissue without
a known cell-type specific topography, employing SCAMPR successfully identified unique
cell-type specific changes in gene expression due to ELS. Furthermore, we demonstrated
varied responses to ELS across cell subtypes, where some genes had lower mean
expression across a cell-type (Gpr65), while others were expressed in a lower number of
neurons in specific cell-types (Pvalb) (Figure S2, 6D, 6F). It should be noted that many of
the genes assayed in these experiments are expressed in both the jugular and nodose
ganglion of the jugular-nodose complex. Future experiments can employ ganglion specific
markers such as Phox2b to interrogate both cell type and tissue specific changes in gene
expression after ELS. In summary, SCAMPR also provides a cost-effective approach to
high resolution detection of potential, single-cell differences in gene expression in
experimental models, with spatial details retained.
4.6 STUDY LIMITATIONS
SCAMPR, which utilizes HuC/D signal for cellular segmentation and ROI
generation, is currently limited to the study of neurons in the nervous system. While the
stability of reagents under multiplex in situ hybridization conditions would need to be
assessed, it is likely that antibodies detecting proteins to mark other non-neural cells can
163
be utilized with SCAMPR to assay gene expression in unique cell populations. This would
facilitate analysis in pathogenic tissues, organoids and other sources of tissue to assay
gene expression spatially.
The SCAMPR pipeline utilizes three cellular attributes — gene expression, soma
size, and cellular topography — to distinguish neuron populations. Yet, neuronal
innervation patterns, often in combination with the three attributes utilized by SCAMPR,
are also important for categorizing neurons into distinct subtypes. It has been
demonstrated that one-round of RNAscope is compatible with retrograde tracers, making
it feasible to incorporate a fourth modality, circuit topology, into the SCAMPR pipeline
(Rana et al., 2020). For this to be successfully incorporated with multi-round HiPlex
RNAscope and SCAMPR, the signal from the traced cells would require bleaching or
degradation after an initial round of imaging, making the microscopy channel available for
use in the proceeding rounds to assay mRNA expression.
To accurately quantify mRNA signal in image sets with large variations in gene
expression and signal intensity, SCAMPR utilizes either manual or semi-automated
threshold selection to distinguish positive signal from noise in an image-specific manner.
To mitigate the effects of between-user variation during threshold selection, it is
imperative that the user utilize proper experimental blinding procedures to the identity of
the image and the experimental group during imaging and throughout the SCAMPR
pipeline.
4.7 SUPPLEMENTARY MATERIALS
164
Supplemental Figure 1: Semiautomated thresholding methods using individual
image intensities. Related to Figure 1B and the Cell Segmentation and Signal
Quantification section in the Methods.
A) A histogram-based automated thresholding method was applied to a high-intensity
image (Glp1r) and low intensity image (Piezo1) in the same tissue section. Comparison
165
of images that have undergone this type of thresholding to ground truth images and
images that have undergone manual thresholding demonstrates the suitability of default
ImageJ automated thresholding method for the high intensity image but poor performance
for the low intensity image. In the low intensity image, default ImageJ automated
thresholding greatly overestimates true signal. All other automated thresholding methods
in ImageJ demonstrated similar inaccuracies in thresholding both low and high intensity
images. B) The distribution of log-mean image intensities for 96 nodose ganglion images,
showing a wide variation in image intensity. C) Demonstration of the manual thresholding
of the “Representative Image” (image with the highest intensity). The image that is
undergoing manual thresholding (right) is compared to the ground truth image (left). D) A
table of corrected threshold values is generated from the individual image intensities
based on manual thresholding of a “Representative Image”. E) The results of the
semiautomated thresholding on the high and low intensity images from 1A compared to
their ground truth counterparts, demonstrating the accuracy of our semiautomated
method on images with differing image intensities.
166
Supplemental Figure 2: Quantitative comparison of gene expression between
nuclear and cytoplasmic segmentation. Related to Figure 3.
A) Box and whisker plots of gene expression using nuclear (DAPI) or cytoplasmic (HuC/D)
segmentation demonstrating that more signal is captured using HuC/D for segmentation.
Note that segmentation using DAPI leads to varying amounts of a loss in quantified signal
depending on the gene (Green Box, Cckar). B) NG tissue section with HuC/D cell
boundary ROIs and DAPI cell boundary ROIs. C) Cckar expression in the same NG tissue
section. High magnification insets with and without DAPI and HuC/D ROIs demonstrate
that much of the Cckar signal is cytoplasmic and not captured by nuclear segmentation.
167
Supplemental Figure 3: Pairwise gene co-expression comparisons between CAU
and ELS in nodose ganglion. Related to Figure 6.
In the bottom-left of the diagonal, Pearson’s correlation coefficients were calculated for
the expression of each gene pair for the mice in the CAU group (blue), for the mice in the
ELS group (red), and for all mice from both groups (grey). In the top-right of the diagonal,
ordinary least squares regression models were fitted to the expression of each gene pair
168
in the CAU (blue) and ELS (red) mice. The equation for each regression line is also
plotted. Red boxes indicate gene pairs that exhibited sizable differences in correlation
coefficients and regression slopes between groups and were selected for further analysis
using modern statistical methods.
169
170
Supplemental Figure 4: Comparison of gene expression between CAU and ELS
nodose ganglia neurons. Related to Figure 6.
The percent of cells expressing each gene in each animal in the CAU group are
averaged and compared to the averages calculated for the ELS animals. Mean gene
expression in each animal in the CAU group is calculated and compared to the mean
gene expression in the ELS group. Student’s t-tests are performed to compare means
between groups and p-values are calculated to assess significance. Note the significant
decrease in the percentage of all cells expressing Pvalb, but no change in mean
expression level. Conversely, the percentage of Gpr65 neurons is unchanged between
CAU and ELS groups, but the mean expression level is decreased in the ELS group.
171
Supplemental Figure 5: Accurate segmentation on maximum intensity projection
images. Related to the Cell Segmentation and Signal Quantification section in the
Methods.
A) Maximum intensity projection (MIP) of HuC/D in the Nodose Ganglion and Cellpose
cell boundary ROIs. Even in tissues with high cellular density, most cell boundaries can
be clearly distinguished. B) The first, middle, and last z-plane images from the MIP in
figure A. Cells appear/disappear and change morphologically as you progress through
172
the stack, however, the cell profiles are still properly segmented in the MIP (green arrow).
A small number of cells are not properly segmented, likely due to overlapping cell
boundaries (red arrows).
173
CHAPTER 5: Concluding Remarks
The highly complex functional and behavioral outputs of the mammalian nervous
system require temporally precise expression of gene networks throughout development.
The functional roles of these gene networks are greatly expanded by genetic pleiotropy,
which allows differential functional capabilities for single genes depending on
developmental time and anatomical space. This expanded repertoire of genetic function
drives the diversity of cell types that are present throughout different structures in the
nervous system and allows for their interaction and integration into functional structures
and circuits. Understanding these fundamental developmental mechanisms are
imperative for understanding how the brain drives so many different functional processes,
and the etiology of neurodevelopmental disorders and diseases of the nervous system.
In the chapters above, I outline what is known about the development of neuronal
diversity in the cortex and the brainstem and discuss the ever-expanding arsenal of
technologies that allow us to answer more nuanced questions about complex structure-
function relations in the brain. I present novel molecular, anatomical, and functional
findings on the role of the highly pleiotropic Met gene on the development of neuronal
diversity in multiple structures within the nervous system. I also present a novel analytical
tool for investigating the topographic gene expression patterns of neurons at multiple
developmental timepoints and in a high-dimensional manner.
174
In chapter 1 I describe the behavioral and functional complexity of the vertebrate
brain and describe how cellular heterogeneity and pleiotropy interact throughout
development to establish structure-function relations. I summarize the known
developmental roles for Met in multiple regions and cell types within the nervous system.
Next, I provide background on the innervation patterns and functional significance of
vagal motor nuclei and describe in detail how gene programs drive early brainstem
patterning, cell migration, nucleogenesis, and axon guidance and link these
developmental processes to the generation of vagal motor neuron diversity. Further
background is provided on the known developmental processes involved in neuronal and
functional diversity within different regions of the mouse cortex. Lastly, I describe how
novel methods are being used and combined to achieve a multidimensional
understanding of cellular heterogeneity within the nervous system.
Chapter 2 provide novel evidence on the developmental role of Met in vagal MNs.
The work outlined in this chapter reveal a trophic role for Met in MNs of the nAmb and
demonstrate that the cell loss is anatomically restricted to a subregion containing
esophagus-projecting MNs. highlighting the importance of Met-signaling in the
development of structure-function relations in the vagal brainstem, I show that this MN
loss leads to transient weight and lifelong ultrasonic vocalization deficits in mice. One of
the most interesting findings from this work was that only a subset of Met+ MNs were lost
after conditional deletion of Met. This suggests even further heterogeneity within the Met+
subpopulations of nAmb MNs and sets up the preliminary experiments discussed in
chapter 5.
175
Chapter 3 concentrates on the molecular heterogeneity of Met-expressing cells in
the developing cortex. I describe prior studies from our lab showing that Met is enriched
in the neuropil of subsets of PN subclasses, and that its peak expression levels
correspond with periods of synapse formation and neurite outgrowth. This sets up the
rationale of using scRNAseq to compare Met+ cells to Met- cells in the developing cortex.
The main takeaway from these studies is that there are subtle molecular differences
between Met+ and Met- PNs of the same subclass across two cortical regions that may
be related to the maturity state of different PN populations. In addition, I show that the
DEGs that were discovered are involved in numerous developmental processes and
contain genes that have been minimally explored in the context of neuronal development,
providing new avenues of investigation. Further comparisons between Met+ cells in the
VC and FC and HiPlex RNAscope experiments will help expand our knowledge of the
molecular diversity of different PN subclasses during development.
In chapter 4, I describe a new bench-to-computer pipeline that we developed for
robust and accurate quantification of neuronal gene expression. I demonstrate that our
ISH-IHC dual protocol and utilization of computation tools allows for fast and accurate
segmentation of cell bodies and better capture of mRNA signal in each cell. I also
demonstrate the analysis capabilities of SCAMPR for analysis of gene expression
topography and for comparing experimental groups. Tools like SCAMPR, which allow for
multimodal analysis of neuronal characteristic (gene expression and anatomical location),
176
will be fundamentally important for understanding the link between cellular heterogeneity
and the functional architecture of the brain.
The appendix that follows provides some preliminary answers for the unanswered
questions that arose from the work in chapter 2. Using the technique outlines in chapter
4 (SCAMPR), I asses the combinatorial NTFR expression pattering in the developing
nAmb in the context of conditional Met deletion. I show that nAmb vagal MNs co-express
different combinations of NTFRs, like what had previously been shown their spinal MNs
using different methods. This work identified two possible candidates that may serve
trophic roles in subsets of Met+ neurons that are lost in cKO mice. A larger sample size
and further analysis will determine whether these combinatorial NTFR expression
patterns help define MN diversity in the extremely functionally diverse nAmb.
Most of the work and background included above focuses basic biological
mechanisms involved in the development of structure-function relations. While we
focused on the basic biology, the studies outline above were performed on cortical and
vagal circuitry, both of which have been implicated in ASD (Campbell et al., 2010; Kast,
Wu, Williams, et al., 2017; Morton et al., 2007; Sheinkopf et al., 2019; van Hoorn et al.,
2019). Furthermore, decreased expression of Met has been correlated with ASD and Rett
Syndrome (Aldinger et al., 2020; Campbell et al., 2017; Eagleson et al., 2017a; Yun Peng
et al., 2013; Plummer et al., 2013). Ultimately, further advancements in determining how
genetic pleiotropy effects the development of structure-function relations may help us
177
understand the roles of genetic pleiotropy and cellular heterogeneity on neuropsychiatric
disease (Gandal et al., 2016).
REFERENCES
1. ABERCROMBIE, M. (1946). ESTIMATION OF NUCLEAR POPULATION FROM
MICROTOME SECTIONS. THE ANATOMICAL RECORD, 94(2), 239–247.
HTTPS://DOI.ORG/10.1002/AR.1090940210
2. ALDINGER, K. A., TIMMS, A. E., MACDONALD, J. W., MCNAMARA, H. K., HERSTEIN,
J. S., BAMMLER, T. K., EVGRAFOV, O. V., KNOWLES, J. A., & LEVITT, P. (2020).
TRANSCRIPTOME DATA OF TEMPORAL AND CINGULATE CORTEX IN THE RETT SYNDROME
BRAIN. SCIENTIFIC DATA, 7(1), 192. HTTPS://DOI.ORG/10.1038/S41597-020-0527-2
3. ALI MARANDI GHODDOUSI, R., MAGALONG, V. M., KAMITAKAHARA, A. K., & LEVITT,
P. (2022). SCAMPR, A SINGLE-CELL AUTOMATED MULTIPLEX PIPELINE FOR RNA
QUANTIFICATION AND SPATIAL MAPPING. CELL REPORTS METHODS, 2(10), 100316.
HTTPS://DOI.ORG/10.1016/J.CRMETH.2022.100316
4. ANGEVINE, J. B., & SIDMAN, R. L. (1961). AUTORADIOGRAPHIC STUDY OF CELL
MIGRATION DURING HISTOGENESIS OF CEREBRAL CORTEX IN MOUSE. NATURE,
192(4804), 766–768. HTTPS://DOI.ORG/10.1038/192766B0
5. ARLOTTA, P., MOLYNEAUX, B. J., CHEN, J., INOUE, J., KOMINAMI, R., & MACKLIS,
J. D. (2005). NEURONAL SUBTYPE-SPECIFIC GENES THAT CONTROL CORTICOSPINAL
MOTOR NEURON DEVELOPMENT IN VIVO. NEURON, 45(2), 207–221.
HTTPS://DOI.ORG/10.1016/J.NEURON.2004.12.036
6. ARRIAGA, G., ZHOU, E. P., & JARVIS, E. D. (2012). OF MICE, BIRDS, AND MEN:
THE MOUSE ULTRASONIC SONG SYSTEM HAS SOME FEATURES SIMILAR TO HUMANS
AND SONG-LEARNING BIRDS. PLOS ONE, 7(10), E46610.
HTTPS://DOI.ORG/10.1371/JOURNAL.PONE.0046610
7. ASARO, A., SINHA, R., BAKUN, M., KALNYTSKA, O., CARLO-SPIEWOK, A. S.,
RUBEL, T., ROZEBOOM, A., DADLEZ, M., KAMINSKA, B., ARONICA, E., MALIK, A. R., &
WILLNOW, T. E. (2021). APOE4 DISRUPTS INTERACTION OF SORTILIN WITH FATTY ACID-
BINDING PROTEIN 7 ESSENTIAL TO PROMOTE LIPID SIGNALING. JOURNAL OF CELL
SCIENCE, 134(20). HTTPS://DOI.ORG/10.1242/JCS.258894
178
8. ASTICK, M., TUBBY, K., MUBARAK, W. M., GUTHRIE, S., & PRICE, S. R. (2014).
CENTRAL TOPOGRAPHY OF CRANIAL MOTOR NUCLEI CONTROLLED BY DIFFERENTIAL
CADHERIN EXPRESSION. CURRENT BIOLOGY : CB, 24(21), 2541–2547.
HTTPS://DOI.ORG/10.1016/J.CUB.2014.08.067
9. AVETISYAN, X. M., WANG, H., SCHILL, X. E. M., BERY, X. S., GRIDER, J. R.,
HASSELL, J. A., STAPPENBECK, X. T., & HEUCKEROTH, X. R. O. (2015). HEPATOCYTE
GROWTH FACTOR AND MET SUPPORT MOUSE ENTERIC NERVOUS SYSTEM
DEVELOPMENT , THE PERISTALTIC RESPONSE , AND INTESTINAL EPITHELIAL
PROLIFERATION IN RESPONSE TO INJURY. 35(33), 11543–11558.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.5267-14.2015
10. BANIHASHEMI, L., & RINAMAN, L. (2010). REPEATED BRIEF POSTNATAL MATERNAL
SEPARATION ENHANCES HYPOTHALAMIC GASTRIC AUTONOMIC CIRCUITS IN JUVENILE
RATS. NEUROSCIENCE, 165(1), 265.
HTTPS://DOI.ORG/10.1016/J.NEUROSCIENCE.2009.09.081
11. BARKAN, C. L., & ZORNIK, E. (2020). INSPIRING SONG: THE ROLE OF RESPIRATORY
CIRCUITRY IN THE EVOLUTION OF VERTEBRATE VOCAL BEHAVIOR. DEVELOPMENTAL
NEUROBIOLOGY, 80(1–2), 31–41. HTTPS://DOI.ORG/10.1002/DNEU.22752
12. BARSH, G. R., ISABELLA, A. J., & MOENS, C. B. (2017A). VAGUS MOTOR NEURON
TOPOGRAPHIC MAP DETERMINED BY PARALLEL MECHANISMS OF HOX5 EXPRESSION
AND TIME OF AXON INITIATION. CURRENT BIOLOGY, 27(24), 3812-3825.E3.
HTTPS://DOI.ORG/10.1016/J.CUB.2017.11.022
13. BARSH, G. R., ISABELLA, A. J., & MOENS, C. B. (2017B). VAGUS MOTOR NEURON
TOPOGRAPHIC MAP DETERMINED BY PARALLEL MECHANISMS OF HOX5 EXPRESSION
AND TIME OF AXON INITIATION. CURRENT BIOLOGY, 27(24), 3812-3825.E3.
HTTPS://DOI.ORG/10.1016/J.CUB.2017.11.022
14. BERTERO, A., VERRILLO, L., & APICELLA, A. J. (2022). A NOVEL LAYER 4
CORTICOFUGAL CELL TYPE/PROJECTION INVOLVED IN THALAMO-CORTICO-STRIATAL
SENSORY PROCESSING. JOURNAL OF NEUROSCIENCE, 42(8), 1383–1405.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.1738-21.2021
15. BIEGER, D., & HOPKINS, D. A. (1987). VISCEROTOPIC REPRESENTATION OF THE
UPPER ALIMENTARY TRACT IN THE MEDULLA OBLONGATA IN THE RAT: THE NUCLEUS
AMBIGUUS. THE JOURNAL OF COMPARATIVE NEUROLOGY, 262(4), 546–562.
HTTPS://DOI.ORG/10.1002/CNE.902620408
16. BLOUNT, Q., HERNANDEZ-MORATO, I., MOAYEDI, Y., & PITMAN, M. J. (2022).
EXPRESSION OF GLIAL CELL-DERIVED NEUROTROPHIC FACTOR RECEPTORS WITHIN
NUCLEUS AMBIGUUS DURING RAT DEVELOPMENT. THE LARYNGOSCOPE.
HTTPS://DOI.ORG/10.1002/LARY.30440
179
17. BOTA, M., & SWANSON, L. W. (2007). THE NEURON CLASSIFICATION PROBLEM.
BRAIN RESEARCH REVIEWS, 56(1), 79–88.
HTTPS://DOI.ORG/10.1016/J.BRAINRESREV.2007.05.005
18. BRESCHI, A., GINGERAS, T. R., & GUIGÓ, R. (2017). COMPARATIVE
TRANSCRIPTOMICS IN HUMAN AND MOUSE. NATURE REVIEWS. GENETICS, 18(7), 425.
HTTPS://DOI.ORG/10.1038/NRG.2017.19
19. CADWELL, C. R., BHADURI, A., MOSTAJO-RADJI, M. A., KEEFE, M. G., &
NOWAKOWSKI, T. J. (2019). DEVELOPMENT AND AREALIZATION OF THE CEREBRAL
CORTEX. NEURON, 103(6), 980–1004.
HTTPS://DOI.ORG/10.1016/J.NEURON.2019.07.009
20. CAICEDO, J. C., ROTH, J., GOODMAN, A., BECKER, T., KARHOHS, K. W., BROISIN,
M., MOLNAR, C., MCQUIN, C., SINGH, S., THEIS, F. J., & CARPENTER, A. E. (2019).
EVALUATION OF DEEP LEARNING STRATEGIES FOR NUCLEUS SEGMENTATION IN
FLUORESCENCE IMAGES. CYTOMETRY PART A, 95(9), 952–965.
HTTPS://DOI.ORG/10.1002/CYTO.A.23863
21. CALZOLARI, S., TERRIENTE, J., & PUJADES, C. (2014). CELL SEGREGATION IN THE
VERTEBRATE HINDBRAIN RELIES ON ACTOMYOSIN CABLES LOCATED AT THE
INTERHOMBOMERIC BOUNDARIES. THE EMBO JOURNAL, 33(7), 686.
HTTPS://DOI.ORG/10.1002/EMBJ.201386003
22. CAMPBELL, D. B., BUIE, T. M., WINTER, H., BAUMAN, M., SUTCLIFFE, J. S.,
PERRIN, J. M., & LEVITT, P. (2017). DISTINCT GENETIC RISK BASED ON ASSOCIATION
OF MET IN FAMILIES WITH CO-OCCURRING AUTISM AND.
HTTPS://DOI.ORG/10.1542/PEDS.2008-0819
23. CAMPBELL, D. B., WARREN, D., SUTCLIFFE, J. S., LEE, E. B., & LEVITT, P. (2010).
ASSOCIATION OF MΕT WITH SOCIAL AND COMMUNICATION PHENOTYPES IN INDIVIDUALS
WITH AUTISM SPECTRUM DISORDER. AMERICAN JOURNAL OF MEDICAL GENETICS, PART
B: NEUROPSYCHIATRIC GENETICS, 153(2), 438–446.
HTTPS://DOI.ORG/10.1002/AJMG.B.30998
24. CARD, J. P., LEVITT, P., GLUHOVSKY, M., & RINAMAN, L. (2005). EARLY
EXPERIENCE MODIFIES THE POSTNATAL ASSEMBLY OF AUTONOMIC EMOTIONAL MOTOR
CIRCUITS IN RATS. JOURNAL OF NEUROSCIENCE, 25(40), 9102–9111.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.2345-05.2005
25. CARINE STAPEL, L., LOMBARDOT, B., BROADDUS, C., KAINMUELLER, D., JUG, F.,
MYERS, E. W., & VASTENHOUW, N. L. (2016). AUTOMATED DETECTION AND
QUANTIFICATION OF SINGLE RNAS AT CELLULAR RESOLUTION IN ZEBRAFISH EMBRYOS.
DEVELOPMENT (CAMBRIDGE), 143(3), 530–539.
HTTPS://DOI.ORG/10.1242/DEV.128918
180
26. CARPENTER, E. M., GODDARD, J. M., CHISAKA, O., MANLEY, N. R., CAPECCHI, M.
R., GODDARD, J. M., CHISAKA, O., & CAPECCHI, M. R. (1993). LOSS OF HOX-A1
(HOX-1.6) FUNCTION RESULTS IN THE REORGANIZATION OF THE MURINE HINDBRAIN.
DEVELOPMENT (CAMBRIDGE, ENGLAND), 118(4), 1063–1075.
HTTP://WWW.NCBI.NLM.NIH.GOV/PUBMED/7903632
27. CATON, A., HACKER, A., NAEEM, A., LIVET, J., MAINA, F., BLADT, F., KLEIN, R.,
BIRCHMEIER, C., & GUTHRIE, S. (2000). THE BRANCHIAL ARCHES AND HGF ARE
GROWTH-PROMOTING AND CHEMOATTRACTANT FOR CRANIAL MOTOR AXONS.
DEVELOPMENT (CAMBRIDGE, ENGLAND), 127(8), 1751–1766.
HTTP://WWW.NCBI.NLM.NIH.GOV/PUBMED/10725250
28. CAUBIT, X., THOBY-BRISSON, M., VOITURON, N., FILIPPI, P., BÉVENGUT, M.,
FARALLI, H., ZANELLA, S., FORTIN, G., HILAIRE, G., & FASANO, L. (2010). TEASHIRT 3
REGULATES DEVELOPMENT OF NEURONS INVOLVED IN BOTH RESPIRATORY RHYTHM
AND AIRFLOW CONTROL. JOURNAL OF NEUROSCIENCE, 30(28), 9465–9476.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.1765-10.2010
29. CEDERQUIST, G. Y., LUCHNIAK, A., TISCHFIELD, M. A., PEEVA, M., SONG, Y.,
MENEZES, M. P., CHAN, W. M., ANDREWS, C., CHEW, S., JAMIESON, R. V., GOMES,
L., FLAHERTY, M., GRANT, P. E., GUPTA, M. L., & ENGLE, E. C. (2012). AN INHERITED
TUBB2B MUTATION ALTERS A KINESIN-BINDING SITE AND CAUSES POLYMICROGYRIA,
CFEOM AND AXON DYSINNERVATION. HUMAN MOLECULAR GENETICS, 21(26), 5484.
HTTPS://DOI.ORG/10.1093/HMG/DDS393
30. CHABOUT, J., SARKAR, A., PATEL, S. R., RADDEN, T., DUNSON, D. B., FISHER, S.
E., & JARVIS, E. D. (2016). A FOXP2 MUTATION IMPLICATED IN HUMAN SPEECH
DEFICITS ALTERS SEQUENCING OF ULTRASONIC VOCALIZATIONS IN ADULT MALE MICE.
FRONTIERS IN BEHAVIORAL NEUROSCIENCE, 10(OCT).
HTTPS://DOI.ORG/10.3389/FNBEH.2016.00197
31. CHAN, C. H., GODINHO, L. N., THOMAIDOU, D., TAN, S. S., GULISANO, M., &
PARNAVELAS, J. G. (2001). EMX1 IS A MARKER FOR PYRAMIDAL NEURONS OF THE
CEREBRAL CORTEX. CEREBRAL CORTEX, 11(12), 1191–1198.
HTTPS://DOI.ORG/10.1093/CERCOR/11.12.1191
32. CHASE, M. R., & RANSON, S. W. (1914). THE STRUCTURE OF THE ROOTS, TRUNK
AND BRANCHES OF THE VAGUS NERVE. THE JOURNAL OF COMPARATIVE NEUROLOGY,
24(1), 31–60. HTTPS://DOI.ORG/10.1002/CNE.900240103
33. CHEN, B., WANG, S. S., HATTOX, A. M., RAYBURN, H., NELSON, S. B., &
MCCONNELL, S. K. (2008). THE FEZF2-CTIP2 GENETIC PATHWAY REGULATES THE FATE
CHOICE OF SUBCORTICAL PROJECTION NEURONS IN THE DEVELOPING CEREBRAL
CORTEX. PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED
STATES OF AMERICA, 105(32), 11382–11387.
HTTPS://DOI.ORG/10.1073/PNAS.0804918105
181
34. CHEN, K. H., BOETTIGER, A. N., MOFFITT, J. R., WANG, S., & ZHUANG, X. (2015).
SPATIALLY RESOLVED, HIGHLY MULTIPLEXED RNA PROFILING IN SINGLE CELLS.
SCIENCE,348(6233).
HTTPS://DOI.ORG/10.1126/SCIENCE.AAA6090/SUPPL_FILE/CHEN-SM.PDF
35. CHEN, K., MA, X., NEHME, A., WEI, J., CUI, Y., CUI, Y., YAO, D., WU, J.,
ANDERSON, T., FERGUSON, D., LEVITT, P., & QIU, S. (2021). TIME-DELIMITED
SIGNALING OF MET RECEPTOR TYROSINE KINASE REGULATES CORTICAL CIRCUIT
DEVELOPMENT AND CRITICAL PERIOD PLASTICITY. MOLECULAR PSYCHIATRY, 26(8),
3723–3736. HTTPS://DOI.ORG/10.1038/S41380-019-0635-6
36. CHENG, S., BUTRUS, S., TAN, L., XU, R., SAGIREDDY, S., TRACHTENBERG, J. T.,
SHEKHAR, K., & ZIPURSKY, S. L. (2022A). VISION-DEPENDENT SPECIFICATION OF CELL
TYPES AND FUNCTION IN THE DEVELOPING CORTEX. CELL, 185(2), 311-327.E24.
HTTPS://DOI.ORG/10.1016/J.CELL.2021.12.022
37. CHENG, S., BUTRUS, S., TAN, L., XU, R., SAGIREDDY, S., TRACHTENBERG, J. T.,
SHEKHAR, K., & ZIPURSKY, S. L. (2022B). VISION-DEPENDENT SPECIFICATION OF CELL
TYPES AND FUNCTION IN THE DEVELOPING CORTEX. CELL, 185(2), 311-327.E24.
HTTPS://DOI.ORG/10.1016/J.CELL.2021.12.022
38. CHOI, H. M. T., SCHWARZKOPF, M., FORNACE, M. E., ACHARYA, A., ARTAVANIS,
G., STEGMAIER, J., CUNHA, A., & PIERCE, N. A. (2018). THIRD-GENERATION IN SITU
HYBRIDIZATION CHAIN REACTION: MULTIPLEXED, QUANTITATIVE, SENSITIVE, VERSATILE,
ROBUST. DEVELOPMENT (CAMBRIDGE), 145(12).
HTTPS://DOI.ORG/10.1242/DEV.165753/48466
39. CODELUPPI, S., BORM, L. E., ZEISEL, A., LA MANNO, G., VAN LUNTEREN, J. A.,
SVENSSON, C. I., & LINNARSSON, S. (2018). SPATIAL ORGANIZATION OF THE
SOMATOSENSORY CORTEX REVEALED BY OSMFISH. NATURE METHODS, 15(11), 932–
935. HTTPS://DOI.ORG/10.1038/S41592-018-0175-Z
40. COOKE, J. E., KEMP, H. A., & MOENS, C. B. (2005). EPHA4 IS REQUIRED FOR CELL
ADHESION AND RHOMBOMERE-BOUNDARY FORMATION IN THE ZEBRAFISH. CURRENT
BIOLOGY : CB, 15(6), 536–542. HTTPS://DOI.ORG/10.1016/J.CUB.2005.02.019
41. COOKE, J., MOENS, C., ROTH, L., DURBIN, L., SHIOMI, K., BRENNAN, C., KIMMEL,
C., WILSON, S., & HOLDER, N. (2001). EPH SIGNALLING FUNCTIONS DOWNSTREAM OF
VAL TO REGULATE CELL SORTING AND BOUNDARY FORMATION IN THE CAUDAL
HINDBRAIN. DEVELOPMENT (CAMBRIDGE, ENGLAND), 128(4), 571–580.
HTTP://WWW.NCBI.NLM.NIH.GOV/PUBMED/8575301
42. COOPER, J. A. (2008). A MECHANISM FOR INSIDE-OUT LAMINATION IN THE
NEOCORTEX. TRENDS IN NEUROSCIENCES, 31(3), 113–119.
HTTPS://DOI.ORG/10.1016/J.TINS.2007.12.003
182
43. COSSART, R., & GAREL, S. (2022). STEP BY STEP: CELLS WITH MULTIPLE
FUNCTIONS IN CORTICAL CIRCUIT ASSEMBLY. NATURE REVIEWS NEUROSCIENCE 2022
23:7, 23(7), 395–410. HTTPS://DOI.ORG/10.1038/S41583-022-00585-6
44. CRUMLEY, R. L. (1979). MECHANISMS OF SYNKINESIS. LARYNGOSCOPE, 89(11),
1847–1854. HTTPS://DOI.ORG/10.1288/00005537-197911000-00020
45. D’ELIA, K. P., & DASEN, J. S. (2018). TOPOGRAPHIC MAPS: MOTOR AXONS WAIT
THEIR TURN. CURRENT BIOLOGY, 28(2), R86–R88.
HTTPS://DOI.ORG/10.1016/J.CUB.2017.11.047
46. DASEN, J. S., & JESSELL, T. M. (2009). CHAPTER SIX HOX NETWORKS AND THE
ORIGINS OF MOTOR NEURON DIVERSITY. CURRENT TOPICS IN DEVELOPMENTAL
BIOLOGY, 88, 169–200. HTTPS://DOI.ORG/10.1016/S0070-2153(09)88006-X
47. DESAI, A. R., & MCCONNELL, S. K. (2000). PROGRESSIVE RESTRICTION IN FATE
POTENTIAL BY NEURAL PROGENITORS DURING CEREBRAL CORTICAL DEVELOPMENT.
DEVELOPMENT, 127(13), 2863–2872. HTTPS://DOI.ORG/10.1242/DEV.127.13.2863
48. DI BELLA, D. J., HABIBI, E., STICKELS, R. R., SCALIA, G., BROWN, J.,
YADOLLAHPOUR, P., YANG, S. M., ABBATE, C., BIANCALANI, T., MACOSKO, E. Z.,
CHEN, F., REGEV, A., & ARLOTTA, P. (2021). MOLECULAR LOGIC OF CELLULAR
DIVERSIFICATION IN THE MOUSE CEREBRAL CORTEX. NATURE 2021 595:7868,
595(7868), 554–559. HTTPS://DOI.ORG/10.1038/S41586-021-03670-5
49. DIAZ-GARCIA, S., KO, V. I., VAZQUEZ-SANCHEZ, S., CHIA, R., GOV OLUBANKOLE,
R. C., AROGUNDADE, A., RODRIGUEZ, M. J., CLEVELAND, D., TRAYNOR, B. J., &
RAVITS, J. (2021). NUCLEAR DEPLETION OF RNA BINDING PROTEIN ELAVL3 (HUC) IN
SPORADIC AND FAMILIAL AMYOTROPHIC LATERAL SCLEROSIS. BIORXIV,
2021.06.03.446017. HTTPS://DOI.ORG/10.1101/2021.06.03.446017
50. DRIES, R., ZHU, Q., DONG, R., LINUS ENG, C.-H., LI, H., LIU, K., FU, Y., ZHAO, T.,
SARKAR, A., BAO, F., GEORGE, R., PIERSON, N., CAI, L., & YUAN, G.-C. (2019).
GIOTTO, A TOOLBOX FOR INTEGRATIVE ANALYSIS AND VISUALIZATION OF SPATIAL
EXPRESSION DATA. HTTPS://DOI.ORG/10.1101/701680
51. DU, H., WANG, Z., GUO, R., YANG, L., LIU, G., ZHANG, Z., XU, Z., TIAN, Y., YANG,
Z., LI, X., & CHEN, B. (2022). TRANSCRIPTION FACTORS BCL11A AND BCL11B ARE
REQUIRED FOR THE PRODUCTION AND DIFFERENTIATION OF CORTICAL PROJECTION
NEURONS. CEREBRAL CORTEX, 32(17), 3611–3632.
HTTPS://DOI.ORG/10.1093/CERCOR/BHAB437
52. EAGLESON, K. L., CAMPBELL, D. B., THOMPSON, B. L., BERGMAN, M. Y., & LEVITT,
P. (2011). THE AUTISM RISK GENES MET AND PLAUR DIFFERENTIALLY IMPACT
CORTICAL DEVELOPMENT. AUTISM RESEARCH : OFFICIAL JOURNAL OF THE
INTERNATIONAL SOCIETY FOR AUTISM RESEARCH, 4(1), 68.
HTTPS://DOI.ORG/10.1002/AUR.172
183
53. EAGLESON, K. L., LANE, C. J., MCFADYEN-KETCHUM, L., SOLAK, S., WU, H. H., &
LEVITT, P. (2016A). DISTINCT INTRACELLULAR SIGNALING MEDIATES C-MET
REGULATION OF DENDRITIC GROWTH AND SYNAPTOGENESIS. DEVELOPMENTAL
NEUROBIOLOGY, 76(10), 1160–1181. HTTPS://DOI.ORG/10.1002/DNEU.22382
54. EAGLESON, K. L., LANE, C. J., MCFADYEN-KETCHUM, L., SOLAK, S., WU, H. H., &
LEVITT, P. (2016B). DISTINCT INTRACELLULAR SIGNALING MEDIATES C-MET
REGULATION OF DENDRITIC GROWTH AND SYNAPTOGENESIS. DEVELOPMENTAL
NEUROBIOLOGY, 76(10), 1160–1181. HTTPS://DOI.ORG/10.1002/DNEU.22382
55. EAGLESON, K. L., VILLANEUVA, M., SOUTHERN, R. M., & LEVITT, P. (2020).
PROTEOMIC AND MITOCHONDRIAL ADAPTATIONS TO EARLY-LIFE STRESS ARE DISTINCT IN
JUVENILES AND ADULTS. NEUROBIOLOGY OF STRESS, 13, 100251.
HTTPS://DOI.ORG/10.1016/J.YNSTR.2020.100251
56. EAGLESON, K. L., XIE, Z., & LEVITT, P. (2017A). THE PLEIOTROPIC MET
RECEPTOR NETWORK: CIRCUIT DEVELOPMENT AND THE NEURAL-MEDICAL INTERFACE
OF AUTISM. BIOLOGICAL PSYCHIATRY, 81(5), 424–433.
HTTPS://DOI.ORG/10.1016/J.BIOPSYCH.2016.08.035
57. EAGLESON, K. L., XIE, Z., & LEVITT, P. (2017B). THE PLEIOTROPIC MET
RECEPTOR NETWORK: CIRCUIT DEVELOPMENT AND THE NEURAL-MEDICAL INTERFACE
OF AUTISM. BIOLOGICAL PSYCHIATRY, 81(5), 424–433.
HTTPS://DOI.ORG/10.1016/J.BIOPSYCH.2016.08.035
58. EBENS, A., BROSE, K., LEONARDO, E. D., JR, M. G. H., BLADT, F., BIRCHMEIER,
C., BARRES, B. A., & TESSIER-LAVIGNE, M. (1996). HEPATOCYTE GROWTH
FACTOR/SCATTER FACTOR IS AN AXONAL CHEMOATTRACTANT AND A NEUROTROPHIC
FACTOR FOR SPINAL MOTOR NEURONS. NEURON, 17(6), 1157–1172.
HTTPS://DOI.ORG/10.1016/S0896-6273(00)80247-0
59. EBRAHIMI, M., YAMAMOTO, Y., SHARIFI, K., KIDA, H., KAGAWA, Y., YASUMOTO, Y.,
ISLAM, A., MIYAZAKI, H., SHIMAMOTO, C., MAEKAWA, M., MITSUSHIMA, D.,
YOSHIKAWA, T., & OWADA, Y. (2016). ASTROCYTE-EXPRESSED FABP7 REGULATES
DENDRITIC MORPHOLOGY AND EXCITATORY SYNAPTIC FUNCTION OF CORTICAL NEURONS.
GLIA, 64(1), 48–62. HTTPS://DOI.ORG/10.1002/GLIA.22902
60. EFRON, B. (1982). 6. THE INFINITESIMAL JACKKNIFE, THE DELTA METHOD AND THE
INFLUENCE FUNCTION. THE JACKKNIFE, THE BOOTSTRAP AND OTHER RESAMPLING
PLANS, 37–47. HTTPS://DOI.ORG/10.1137/1.9781611970319.CH6
61. EINSPIELER, C., & MARSCHIK, P. B. (2019). REGRESSION IN RETT SYNDROME:
DEVELOPMENTAL PATHWAYS TO ITS ONSET. NEUROSCIENCE AND BIOBEHAVIORAL
REVIEWS, 98, 320–332. HTTPS://DOI.ORG/10.1016/J.NEUBIOREV.2019.01.028
184
62. ENG, C. H. L., LAWSON, M., ZHU, Q., DRIES, R., KOULENA, N., TAKEI, Y., YUN, J.,
CRONIN, C., KARP, C., YUAN, G. C., & CAI, L. (2019). TRANSCRIPTOME-SCALE SUPER-
RESOLVED IMAGING IN TISSUES BY RNA SEQFISH+. NATURE 2019 568:7751,
568(7751), 235–239. HTTPS://DOI.ORG/10.1038/S41586-019-1049-Y
63. ERICSON, J., RASHBASS, P., SCHEDL, A., BRENNER-MORTON, S., KAWAKAMI, A.,
VAN HEYNINGEN, V., JESSELL, T. M., & BRISCOE, J. (1997). PAX6 CONTROLS
PROGENITOR CELL IDENTITY AND NEURONAL FATE IN RESPONSE TO GRADED SHH
SIGNALING. CELL, 90(1), 169–180. HTTPS://DOI.ORG/10.1016/S0092-
8674(00)80323-2
64. ERÖ, C., GEWALTIG, M. O., KELLER, D., & MARKRAM, H. (2018A). A CELL ATLAS
FOR THE MOUSE BRAIN. FRONTIERS IN NEUROINFORMATICS, 12, 84.
HTTPS://DOI.ORG/10.3389/FNINF.2018.00084/BIBTEX
65. ERÖ, C., GEWALTIG, M. O., KELLER, D., & MARKRAM, H. (2018B). A CELL ATLAS
FOR THE MOUSE BRAIN. FRONTIERS IN NEUROINFORMATICS, 12, 421921.
HTTPS://DOI.ORG/10.3389/FNINF.2018.00084/BIBTEX
66. FINSTERWALD, C., & MARTIN, J. L. (2011). CELLULAR MECHANISMS UNDERLYING
THE REGULATION OF DENDRITIC DEVELOPMENT BY HEPATOCYTE GROWTH FACTOR.
EUROPEAN JOURNAL OF NEUROSCIENCE, 34(7), 1053–1061.
HTTPS://DOI.ORG/10.1111/J.1460-9568.2011.07839.X
67. FISCHER, J., & HAMMERSCHMIDT, K. (2011). ULTRASONIC VOCALIZATIONS IN
MOUSE MODELS FOR SPEECH AND SOCIO-COGNITIVE DISORDERS: INSIGHTS INTO THE
EVOLUTION OF VOCAL COMMUNICATION. GENES, BRAIN AND BEHAVIOR, 10(1), 17–27.
HTTPS://DOI.ORG/10.1111/J.1601-183X.2010.00610.X
68. FLINT, P. W., DOWNS, D. H., & COLTRERA, M. D. (1991). LARYNGEAL SYNKINESIS
FOLLOWING REINNERVATION IN THE RAT: NEUROANATOMIC AND PHYSIOLOGIC STUDY
USING RETROGRADE FLUORESCENT TRACERS AND ELECTROMYOGRAPHY. ANNALS OF
OTOLOGY, RHINOLOGY & LARYNGOLOGY, 100(10), 797–806.
HTTPS://DOI.ORG/10.1177/000348949110001003
69. FRANCO, S. J., GIL-SANZ, C., MARTINEZ-GARAY, I., ESPINOSA, A., HARKINS-
PERRY, S. R., RAMOS, C., & MÜLLER, U. (2012). FATE-RESTRICTED NEURAL
PROGENITORS IN THE MAMMALIAN CEREBRAL CORTEX. SCIENCE, 337(6095), 746–749.
HTTPS://DOI.ORG/10.1126/SCIENCE.1223616
70. FRANK, D., & SELA-DONENFELD, D. (2019). HINDBRAIN INDUCTION AND
PATTERNING DURING EARLY VERTEBRATE DEVELOPMENT. CELLULAR AND MOLECULAR
LIFE SCIENCES, 76(5), 941–960. HTTPS://DOI.ORG/10.1007/S00018-018-2974-X
185
71. FRENCH, C. A., GROSZER, M., PREECE, C., COUPE, A. M., RAJEWSKY, K., &
FISHER, S. E. (2007). GENERATION OF MICE WITH A CONDITIONAL FOXP2 NULL ALLELE.
GENESIS (UNITED STATES), 45(7), 440–446. HTTPS://DOI.ORG/10.1002/DVG.20305
72. FRIAS, C. P., LIANG, J., BRESSER, T., SCHEEFHALS, L., VAN KESTEREN, M., VAN
DORLAND, R., YIN HU, H., BODZETA, A., VAN BERGEN EN HENEGOUWEN, P. M. P.,
HOOGENRAAD, C. C., & WIERENGA, C. J. (2019). SEMAPHORIN4D INDUCES
INHIBITORY SYNAPSE FORMATION BY RAPID STABILIZATION OF PRESYNAPTIC BOUTONS
VIA MET COACTIVATION. JOURNAL OF NEUROSCIENCE, 39(22), 4221–4237.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.0215-19.2019
73. FRIEDLAND, D. R., EDEN, A. R., & LAITMAN, J. T. (1995). NATURALLY OCCURRING
MOTONEURON CELL DEATH IN RAT UPPER RESPIRATORY TRACT MOTOR NUCLEI: A
HISTOLOGICAL, FAST DIL AND IMMUNOCYTOCHEMICAL STUDY IN THE NUCLEUS
AMBIGUUS. JOURNAL OF NEUROBIOLOGY, 26(4), 563–578.
HTTPS://DOI.ORG/10.1002/NEU.480260409
74. FUJITA, E., TANABE, Y., SHIOTA, A., UEDA, M., SUWA, K., MOMOI, M. Y., & MOMOI,
T. (2008). ULTRASONIC VOCALIZATION IMPAIRMENT OF FOXP2 (R552H) KNOCKIN MICE
RELATED TO SPEECH-LANGUAGE DISORDER AND ABNORMALITY OF PURKINJE CELLS.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF
AMERICA, 105(8), 3117–3122. HTTPS://DOI.ORG/10.1073/PNAS.0712298105
75. GANDAL, M. J., LEPPA, V., WON, H., PARIKSHAK, N. N., & GESCHWIND, D. H.
(2016). THE ROAD TO PRECISION PSYCHIATRY: TRANSLATING GENETICS INTO DISEASE
MECHANISMS. NATURE NEUROSCIENCE 2016 19:11, 19(11), 1397–1407.
HTTPS://DOI.ORG/10.1038/NN.4409
76. GAVALAS, A., & KRUMLAUF, R. (2000). RETINOID SIGNALLING AND HINDBRAIN
PATTERNING. CURRENT OPINION IN GENETICS & DEVELOPMENT, 10(4), 380–386.
HTTPS://DOI.ORG/10.1016/S0959-437X(00)00100-3
77. GESCHWIND, D. (2004). GENSAT: A GENOMIC RESOURCE FOR NEUROSCIENCE
RESEARCH. LANCET NEUROLOGY, 3(2), 82. HTTPS://DOI.ORG/10.1016/S1474-
4422(03)00658-6
78. GIBBS, R. B., & PFAFF, D. W. (1994). IN SITU HYBRIDIZATION DETECTION OF TRKA
MRNA IN BRAIN: DISTRIBUTION, COLOCALIZATION WITH P75NGFR AND UP‐REGULATION
BY NERVE GROWTH FACTOR. JOURNAL OF COMPARATIVE NEUROLOGY, 341(3), 324–
339. HTTPS://DOI.ORG/10.1002/CNE.903410304
79. GIL-SANZ, C., ESPINOSA, A., FREGOSO, S. P., BLUSKE, K. K., CUNNINGHAM, C. L.,
MARTINEZ-GARAY, I., ZENG, H., FRANCO, S. J., & MÜLLER, U. (2015). LINEAGE
TRACING USING CUX2-CRE AND CUX2-CREERT2 MICE. NEURON, 86(4), 1091–1099.
HTTPS://DOI.ORG/10.1016/J.NEURON.2015.04.019
186
80. GILLAND, E., & BAKER, R. (2005). EVOLUTIONARY PATTERNS OF CRANIAL NERVE
EFFERENT NUCLEI IN VERTEBRATES. BRAIN, BEHAVIOR AND EVOLUTION, 66(4), 234–
254. HTTPS://DOI.ORG/10.1159/000088128
81. GOH, J. J. L., CHOU, N., SEOW, W. Y., HA, N., CHENG, C. P. P., CHANG, Y. C.,
ZHAO, Z. W., & CHEN, K. H. (2020). HIGHLY SPECIFIC MULTIPLEXED RNA IMAGING IN
TISSUES WITH SPLIT-FISH. NATURE METHODS 2020 17:7, 17(7), 689–693.
HTTPS://DOI.ORG/10.1038/S41592-020-0858-0
82. GREIG, L. C., WOODWORTH, M. B., GALAZO, M. J., PADMANABHAN, H., & MACKLIS,
J. D. (2013). MOLECULAR LOGIC OF NEOCORTICAL PROJECTION NEURON
SPECIFICATION, DEVELOPMENT AND DIVERSITY. NATURE REVIEWS NEUROSCIENCE
2013 14:11, 14(11), 755–769. HTTPS://DOI.ORG/10.1038/NRN3586
83. GRIMSLEY, J. M. S., MONAGHAN, J. J. M., & WENSTRUP, J. J. (2011).
DEVELOPMENT OF SOCIAL VOCALIZATIONS IN MICE. PLOS ONE, 6(3).
HTTPS://DOI.ORG/10.1371/JOURNAL.PONE.0017460
84. GROSZER, M., KEAYS, D. A., DEACON, R. M. J., DE BONO, J. P., PRASAD-
MULCARE, S., GAUB, S., BAUM, M. G., FRENCH, C. A., NICOD, J., COVENTRY, J. A.,
ENARD, W., FRAY, M., BROWN, S. D. M., NOLAN, P. M., PÄÄBO, S., CHANNON, K. M.,
COSTA, R. M., EILERS, J., EHRET, G., … FISHER, S. E. (2008). IMPAIRED SYNAPTIC
PLASTICITY AND MOTOR LEARNING IN MICE WITH A POINT MUTATION IMPLICATED IN
HUMAN SPEECH DEFICITS. CURRENT BIOLOGY, 18(5), 354–362.
HTTPS://DOI.ORG/10.1016/J.CUB.2008.01.060
85. GUTHRIE, S. (2007A). PATTERNING AND AXON GUIDANCE OF CRANIAL MOTOR
NEURONS. NATURE REVIEWS NEUROSCIENCE, 8(11), 859–871.
HTTPS://DOI.ORG/10.1038/NRN2254
86. GUTHRIE, S. (2007B). PATTERNING AND AXON GUIDANCE OF CRANIAL MOTOR
NEURONS. NATURE REVIEWS NEUROSCIENCE, 8(11), 859–871.
HTTPS://DOI.ORG/10.1038/NRN2254
87. GUTIERREZ, H., DOLCET, X., TOLCOS, M., & DAVIES, A. (2004). HGF REGULATES
THE DEVELOPMENT OF CORTICAL PYRAMIDAL DENDRITES. DEVELOPMENT, 131(15),
3717–3726. HTTPS://DOI.ORG/10.1242/DEV.01209
88. HAMBURGER, V., & LEVI‐MONTALCINI, R. (1949). PROLIFERATION,
DIFFERENTIATION AND DEGENERATION IN THE SPINAL GANGLIA OF THE CHICK EMBRYO
UNDER NORMAL AND EXPERIMENTAL CONDITIONS. JOURNAL OF EXPERIMENTAL
ZOOLOGY, 111(3), 457–501. HTTPS://DOI.ORG/10.1002/JEZ.1401110308
89. HAMMOND, R., VIVANCOS, V., NAEEM, A., CHILTON, J., MAMBETISAEVA, E.,
MAMBITISAEVA, E., ANDREWS, W., SUNDARESAN, V., & GUTHRIE, S. (2005). SLIT-
MEDIATED REPULSION IS A KEY REGULATOR OF MOTOR AXON PATHFINDING IN THE
187
HINDBRAIN. DEVELOPMENT (CAMBRIDGE, ENGLAND), 132(20), 4483–4495.
HTTPS://DOI.ORG/10.1242/DEV.02038
90. HAN, W., KWAN, K. Y., SHIM, S., LAM, M. M. S., SHIN, Y., XU, M., ZHU, Y., LI, M.,
& ŠESTAN, N. (2011). TBR1 DIRECTLY REPRESSES FEZF2 TO CONTROL THE LAMINAR
ORIGIN AND DEVELOPMENT OF THE CORTICOSPINAL TRACT. PROCEEDINGS OF THE
NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 108(7),
3041–3046. HTTPS://DOI.ORG/10.1073/PNAS.1016723108
91. HANEY, M. M., HAMAD, A., WOLDU, H. G., CIUCCI, M., NICHOLS, N., BUNYAK, F., &
LEVER, T. E. (2020). RECURRENT LARYNGEAL NERVE TRANSECTION IN MICE RESULTS
IN TRANSLATIONAL UPPER AIRWAY DYSFUNCTION. JOURNAL OF COMPARATIVE
NEUROLOGY, 528(4), 574–596. HTTPS://DOI.ORG/10.1002/CNE.24774
92. HARB, K., MAGRINELLI, E., NICOLAS, C. S., LUKIANETS, N., FRANGEUL, L., PIETRI,
M., SUN, T., SANDOZ, G., GRAMMONT, F., JABAUDON, D., STUDER, M., & ALFANO, C.
(2016). AREA-SPECIFIC DEVELOPMENT OF DISTINCT PROJECTION NEURON SUBCLASSES
IS REGULATED BY POSTNATAL EPIGENETIC MODIFICATIONS. ELIFE, 5(JANUARY2016).
HTTPS://DOI.ORG/10.7554/ELIFE.09531
93. HASHIKAWA, Y., HASHIKAWA, K., ROSSI, M. A., BASIRI, M. L., LIU, Y., JOHNSTON,
N. L., AHMAD, O. R., & STUBER, G. D. (2020). TRANSCRIPTIONAL AND SPATIAL
RESOLUTION OF CELL TYPES IN THE MAMMALIAN HABENULA. NEURON, 106(5), 743-
758.E5. HTTPS://DOI.ORG/10.1016/J.NEURON.2020.03.011
94. HECKMAN, J., MCGUINNESS, B., CELIKEL, T., & ENGLITZ, B. (2016A).
DETERMINANTS OF THE MOUSE ULTRASONIC VOCAL STRUCTURE AND REPERTOIRE.
NEUROSCIENCE & BIOBEHAVIORAL REVIEWS, 65, 313–325.
HTTPS://DOI.ORG/10.1016/J.NEUBIOREV.2016.03.029
95. HECKMAN, J., MCGUINNESS, B., CELIKEL, T., & ENGLITZ, B. (2016B).
DETERMINANTS OF THE MOUSE ULTRASONIC VOCAL STRUCTURE AND REPERTOIRE.
NEUROSCIENCE & BIOBEHAVIORAL REVIEWS, 65, 313–325.
HTTPS://DOI.ORG/10.1016/J.NEUBIOREV.2016.03.029
96. HEINDORF, M., ARBER, S., & KELLER, G. B. (2018). MOUSE MOTOR CORTEX
COORDINATES THE BEHAVIORAL RESPONSE TO UNPREDICTED SENSORY FEEDBACK.
NEURON, 99(5), 1040. HTTPS://DOI.ORG/10.1016/J.NEURON.2018.07.046
97. HELKE, C. J., ADRYAN, K. M., FEDOROWICZ, J., ZHUO, H., PARK, J. S., CURTIS, R.,
RADLEY, H. E., & DISTEFANO, P. S. (1998). AXONAL TRANSPORT OF NEUROTROPHINS
BY VISCERAL AFFERENT AND EFFERENT NEURONS OF THE VAGUS NERVE OF THE RAT.
J. COMP. NEUROL, 393, 102–117. HTTPS://DOI.ORG/10.1002/(SICI)1096-
9861(19980330)393:1
188
98. HELMBACHER, F., DESSAUD, E., ARBER, S., DELAPEYRIÈRE, O., HENDERSON, C.
E., KLEIN, R., & MAINA, F. (2003). MET SIGNALING IS REQUIRED FOR RECRUITMENT OF
MOTOR NEURONS TO PEA3-POSITIVE MOTOR POOLS. NEURON, 39(5), 767–777.
HTTPS://DOI.ORG/10.1016/S0896-6273(03)00493-8
99. HERCULANO-HOUZEL, S., MOTA, B., & LENT, R. (2006). CELLULAR SCALING RULES
FOR RODENT BRAINS. PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES,
103(32), 12138–12143. HTTPS://DOI.ORG/10.1073/PNAS.0604911103
100. HERNANDEZ-MIRANDA, L. R., RUFFAULT, P.-L., BOUVIER, J. C., MURRAY, A.
J., MORIN-SURUN, M.-P., ZAMPIERI, N., CHOLEWA-WACLAW, J. B., EY, E., BRUNET,
J.-F., CHAMPAGNAT, J., FORTIN, G., & BIRCHMEIER, C. (2017). GENETIC
IDENTIFICATION OF A HINDBRAIN NUCLEUS ESSENTIAL FOR INNATE VOCALIZATION.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES, 114(30), 8095–8100.
HTTPS://DOI.ORG/10.1073/PNAS.1702893114
101. HERNÁNDEZ-MORATO, I., BERDUGO-VEGA, G., SAÑUDO, J. R., MCHANWELL,
S., VÁZQUEZ, T., VALDERRAMA-CANALES, F. J., & PASCUAL-FONT, A. (2014).
SOMATOTOPIC CHANGES IN THE NUCLEUS AMBIGUUS AFTER SECTION AND
REGENERATION OF THE RECURRENT LARYNGEAL NERVE OF THE RAT. ANATOMICAL
RECORD, 297(5), 955–963. HTTPS://DOI.ORG/10.1002/AR.22877
102. HEUN-JOHNSON, H., & LEVITT, P. (2016). EARLY-LIFE STRESS PARADIGM
TRANSIENTLY ALTERS MATERNAL BEHAVIOR, DAM-PUP INTERACTIONS, AND
OFFSPRING VOCALIZATIONS IN MICE. FRONTIERS IN BEHAVIORAL NEUROSCIENCE, 10,
142. HTTPS://DOI.ORG/10.3389/FNBEH.2016.00142
103. HEVNER, R. F., SHI, L., JUSTICE, N., HSUEH, Y. P., SHENG, M., SMIGA, S.,
BULFONE, A., GOFFINET, A. M., CAMPAGNONI, A. T., & RUBENSTEIN, J. L. R. (2001).
TBR1 REGULATES DIFFERENTIATION OF THE PREPLATE AND LAYER 6. NEURON, 29(2),
353–366. HTTPS://DOI.ORG/10.1016/S0896-6273(01)00211-2
104. HOFER, M. A. (1996). MULTIPLE REGULATORS OF ULTRASONIC
VOCALIZATION IN THE INFANT RAT. PSYCHONEUROENDOCRINOLOGY, 21(2), 203–217.
HTTPS://DOI.ORG/10.1016/0306-4530(95)00042-9
105. HOLLYDAY, M., HAMBURGER, V., & FARRIS, J. M. G. (1977). LOCALIZATION
OF MOTOR NEURON POOLS SUPPLYING IDENTIFIED MUSCLES IN NORMAL AND
SUPERNUMERARY LEGS OF CHICK EMBRYO. PROCEEDINGS OF THE NATIONAL ACADEMY
OF SCIENCES OF THE UNITED STATES OF AMERICA, 74(8), 3582–3586.
HTTPS://DOI.ORG/10.1073/PNAS.74.8.3582
106. HOLLYDAY, MARGARET, & HAMBURGER, V. (1976). REDUCTION OF THE
NATURALLY OCCURRING MOTOR NEURON LOSS BY ENLARGEMENT OF THE PERIPHERY.
JOURNAL OF COMPARATIVE NEUROLOGY, 170(3), 311–320.
HTTPS://DOI.ORG/10.1002/CNE.901700304
189
107. HOLY, T. E., & GUO, Z. (2005). ULTRASONIC SONGS OF MALE MICE. PLOS
BIOLOGY, 3(12), E386. HTTPS://DOI.ORG/10.1371/JOURNAL.PBIO.0030386
108. HUH, C. G., FACTOR, V. M., SÁNCHEZ, A., UCHIDA, K., CONNER, E. A., &
THORGEIRSSON, S. S. (2004). HEPATOCYTE GROWTH FACTOR/C-MET SIGNALING
PATHWAY IS REQUIRED FOR EFFICIENT LIVER REGENERATION AND REPAIR.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF
AMERICA, 101(13), 4477–4482. HTTPS://DOI.ORG/10.1073/PNAS.0306068101
109. HYDMAN, J., & MATTSSON, P. (2008). COLLATERAL REINNERVATION BY THE
SUPERIOR LARYNGEAL NERVE AFTER RECURRENT LARYNGEAL NERVE INJURY. MUSCLE
AND NERVE, 38(4), 1280–1289. HTTPS://DOI.ORG/10.1002/MUS.21124
110. ISABELLA, A. J., BARSH, G. R., STONICK, J. A., DUBRULLE, J., & MOENS, C.
B. (2020). RETINOIC ACID ORGANIZES THE ZEBRAFISH VAGUS MOTOR TOPOGRAPHIC
MAP VIA SPATIOTEMPORAL COORDINATION OF HGF/MET SIGNALING. DEVELOPMENTAL
CELL, 53(3), 344. HTTPS://DOI.ORG/10.1016/J.DEVCEL.2020.03.017
111. IULIANELLA, A., SHARMA, M., DURNIN, M., VANDEN HEUVEL, G. B., &
TRAINOR, P. A. (2008). CUX2 (CUTL2) INTEGRATES NEURAL PROGENITOR
DEVELOPMENT WITH CELL-CYCLE PROGRESSION DURING SPINAL CORD NEUROGENESIS.
DEVELOPMENT, 135(4), 729–741. HTTPS://DOI.ORG/10.1242/DEV.013276
112. JAIN, S., & ZIPURSKY, S. L. (2023). TEMPORAL CONTROL OF NEURONAL
WIRING. SEMINARS IN CELL & DEVELOPMENTAL BIOLOGY, 142, 81–90.
HTTPS://DOI.ORG/10.1016/J.SEMCDB.2022.05.012
113. JARRAR, W., DIAS, J. M., ERICSON, J., ARNOLD, H.-H., & HOLZ, A. (2015).
NKX2.2 AND NKX2.9 ARE THE KEY REGULATORS TO DETERMINE CELL FATE OF
BRANCHIAL AND VISCERAL MOTOR NEURONS IN CAUDAL HINDBRAIN. PLOS ONE,
10(4), E0124408. HTTPS://DOI.ORG/10.1371/JOURNAL.PONE.0124408
114. JESSELL, T. M. (2000). NEURONAL SPECIFICATION IN THE SPINAL CORD:
INDUCTIVE SIGNALS AND TRANSCRIPTIONAL CODES. NATURE REVIEWS GENETICS, 1(1),
20–29. HTTPS://DOI.ORG/10.1038/35049541
115. JU, M. J., AROCA, P., LUO, J., PUELLES, L., & REDIES, C. (2004).
MOLECULAR PROFILING INDICATES AVIAN BRANCHIOMOTOR NUCLEI INVADE THE
HINDBRAIN ALAR PLATE. NEUROSCIENCE, 128(4), 785–796.
HTTPS://DOI.ORG/10.1016/J.NEUROSCIENCE.2004.06.063
116. JUDSON, M. C., BERGMAN, M. Y., CAMPBELL, D. B., EAGLESON, K. L., &
LEVITT, P. (2009). DYNAMIC GENE AND PROTEIN EXPRESSION PATTERNS OF THE
AUTISM-ASSOCIATED MET RECEPTOR TYROSINE KINASE IN THE DEVELOPING MOUSE
190
FOREBRAIN. JOURNAL OF COMPARATIVE NEUROLOGY, 513(5), 511–531.
HTTPS://DOI.ORG/10.1002/CNE.21969
117. JUDSON, M. C., EAGLESON, K. L., WANG, L., & LEVITT, P. (2010A).
EVIDENCE OF CELL-NONAUTONOMOUS CHANGES IN DENDRITE AND DENDRITIC SPINE
MORPHOLOGY IN THE MET-SIGNALING-DEFICIENT MOUSE FOREBRAIN. JOURNAL OF
COMPARATIVE NEUROLOGY, 518(21), 4463–4478.
HTTPS://DOI.ORG/10.1002/CNE.22467
118. JUDSON, M. C., EAGLESON, K. L., WANG, L., & LEVITT, P. (2010B).
EVIDENCE OF CELL-NONAUTONOMOUS CHANGES IN DENDRITE AND DENDRITIC SPINE
MORPHOLOGY IN THE MET-SIGNALING-DEFICIENT MOUSE FOREBRAIN. THE JOURNAL OF
COMPARATIVE NEUROLOGY, 518(21), 4463–4478.
HTTPS://DOI.ORG/10.1002/CNE.22467
119. KAMITAKAHARA, A. K., ALI MARANDI GHODDOUSI, R., LANJEWAR, A. L.,
MAGALONG, V. M., WU, H. H., & LEVITT, P. (2021). MET RECEPTOR TYROSINE
KINASE REGULATES LIFESPAN ULTRASONIC VOCALIZATION AND VAGAL MOTOR
NEURON DEVELOPMENT. FRONTIERS IN NEUROSCIENCE, 15, 768577.
HTTPS://DOI.ORG/10.3389/FNINS.2021.768577/BIBTEX
120. KAMITAKAHARA, A., WU, H. H., & LEVITT, P. (2017). DISTINCT PROJECTION
TARGETS DEFINE SUBPOPULATIONS OF MOUSE BRAINSTEM VAGAL NEURONS THAT
EXPRESS THE AUTISM-ASSOCIATED MET RECEPTOR TYROSINE KINASE. JOURNAL OF
COMPARATIVE NEUROLOGY, 525(18), 3787–3808.
HTTPS://DOI.ORG/10.1002/CNE.24294
121. KAST, R. J., LANJEWAR, A. L., SMITH, C. D., & LEVITT, P. (2019). FOXP2
EXHIBITS PROJECTION NEURON CLASS SPECIFIC EXPRESSION, BUT IS NOT REQUIRED FOR
MULTIPLE ASPECTS OF CORTICAL HISTOGENESIS. ELIFE, 8.
HTTPS://DOI.ORG/10.7554/ELIFE.42012
122. KAST, R. J., & LEVITT, P. (2019). PRECISION IN THE DEVELOPMENT OF
NEOCORTICAL ARCHITECTURE: FROM PROGENITORS TO CORTICAL NETWORKS.
PROGRESS IN NEUROBIOLOGY, 175, 77–95.
HTTPS://DOI.ORG/10.1016/J.PNEUROBIO.2019.01.003
123. KAST, R. J., WU, H.-H., & LEVITT, P. (2017). DEVELOPMENTAL
CONNECTIVITY AND MOLECULAR PHENOTYPES OF UNIQUE CORTICAL PROJECTION
NEURONS THAT EXPRESS A SYNAPSE-ASSOCIATED RECEPTOR TYROSINE KINASE.
CEREBRAL CORTEX, JANUARY 2018, 1–13.
HTTPS://DOI.ORG/10.1093/CERCOR/BHX318
124. KAST, R. J., WU, H.-H., WILLIAMS, P., GASPAR, P., LEVITT, P.,
NEUROSCIENCE, †, & PROGRAM, G. (2017). SPECIFIC CONNECTIVITY AND UNIQUE
MOLECULAR IDENTITY OF MET RECEPTOR TYROSINE KINASE EXPRESSING
191
SEROTONERGIC NEURONS IN THE CAUDAL DORSAL RAPHE NUCLEI.
HTTPS://DOI.ORG/10.1021/ACSCHEMNEURO.7B00020
125. KAST, R. J., WU, H. H., & LEVITT, P. (2019). DEVELOPMENTAL
CONNECTIVITY AND MOLECULAR PHENOTYPES OF UNIQUE CORTICAL PROJECTION
NEURONS THAT EXPRESS A SYNAPSE-ASSOCIATED RECEPTOR TYROSINE KINASE.
CEREBRAL CORTEX (NEW YORK, NY), 29(1), 189.
HTTPS://DOI.ORG/10.1093/CERCOR/BHX318
126. KEBSCHULL, J. M., GARCIA DA SILVA, P., REID, A. P., PEIKON, I. D.,
ALBEANU, D. F., & ZADOR, A. M. (2016). HIGH-THROUGHPUT MAPPING OF SINGLE-
NEURON PROJECTIONS BY SEQUENCING OF BARCODED RNA. NEURON, 91(5), 975–
987. HTTPS://DOI.ORG/10.1016/J.NEURON.2016.07.036
127. KELM-NELSON, C. A., LENELL, C., JOHNSON, A. M., & CIUCCI, M. R. (2018).
LARYNGEAL ACTIVITY FOR PRODUCTION OF ULTRASONIC VOCALIZATIONS IN RATS.
HANDBOOK OF BEHAVIORAL NEUROSCIENCE, 25, 37–43.
HTTPS://DOI.ORG/10.1016/B978-0-12-809600-0.00004-4
128. KLINGLER, E. (2023). TEMPORAL CONTROLS OVER CORTICAL PROJECTION
NEURON FATE DIVERSITY. CURRENT OPINION IN NEUROBIOLOGY, 79, 102677.
HTTPS://DOI.ORG/10.1016/J.CONB.2023.102677
129. KLINGLER, E., TOMASELLO, U., PRADOS, J., KEBSCHULL, J. M.,
CONTESTABILE, A., GALIÑANES, G. L., FIÈVRE, S., SANTINHA, A., PLATT, R., HUBER,
D., DAYER, A., BELLONE, C., & JABAUDON, D. (2021). TEMPORAL CONTROLS OVER
INTER-AREAL CORTICAL PROJECTION NEURON FATE DIVERSITY. NATURE 2021
599:7885, 599(7885), 453–457. HTTPS://DOI.ORG/10.1038/S41586-021-04048-3
130. KOH, S., OYLER, G. A., & HIGGINS, G. A. (1989). LOCALIZATION OF NERVE
GROWTH FACTOR RECEPTOR MESSENGER RNA AND PROTEIN IN THE ADULT RAT BRAIN.
EXPERIMENTAL NEUROLOGY, 106(3), 209–221. HTTPS://DOI.ORG/10.1016/0014-
4886(89)90154-4
131. KRUMLAUF, R. (2016). HOX GENES AND THE HINDBRAIN: A STUDY IN
SEGMENTS. CURRENT TOPICS IN DEVELOPMENTAL BIOLOGY, 116, 581–596.
HTTPS://DOI.ORG/10.1016/BS.CTDB.2015.12.011
132. KUDOH, T., WILSON, S., & DAWID, I. (2002). DISTINCT ROLES FOR FGF, WNT
AND RETINOIC ACID IN POSTERIORIZING THE NEURAL ECTODERM. DEVELOPMENT
(CAMBRIDGE, ENGLAND), 129(18), 4335–4346.
HTTPS://WWW.NCBI.NLM.NIH.GOV/PUBMED/12183385
133. KUPARI, J., HÄRING, M., AGIRRE, E., CASTELO-BRANCO, G., & ERNFORS, P.
(2019). AN ATLAS OF VAGAL SENSORY NEURONS AND THEIR MOLECULAR
192
SPECIALIZATION. CELL REPORTS, 27(8), 2508-2523.E4.
HTTPS://DOI.ORG/10.1016/J.CELREP.2019.04.096
134. KWAN, K. Y., LAM, M. M. S., KRSNIK, Ž., KAWASAWA, Y. I., LEFEBVRE, V., &
ŠESTAN, N. (2008). SOX5 POSTMITOTICALLY REGULATES MIGRATION, POSTMIGRATORY
DIFFERENTIATION, AND PROJECTIONS OF SUBPLATE AND DEEP-LAYER NEOCORTICAL
NEURONS. PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED
STATES OF AMERICA, 105(41), 16021–16026.
HTTPS://DOI.ORG/10.1073/PNAS.0806791105
135. LAMBALLE, F., GENESTINE, M., CARUSO, N., ARCE, V., RICHELME, S.,
HELMBACHER, F., & MAINA, F. (2011). POOL-SPECIFIC REGULATION OF MOTOR
NEURON SURVIVAL BY NEUROTROPHIC SUPPORT. JOURNAL OF NEUROSCIENCE,
31(31), 11144–11158. HTTPS://DOI.ORG/10.1523/JNEUROSCI.2198-11.2011
136. LAMBALLE, FABIENNE, GENESTINE, M., CARUSO, N., ARCE, V., RICHELME,
S., HELMBACHER, F., & MAINA, F. (2011). POOL-SPECIFIC REGULATION OF MOTOR
NEURON SURVIVAL BY NEUROTROPHIC SUPPORT. JOURNAL OF NEUROSCIENCE,
31(31), 11144–11158. HTTPS://DOI.ORG/10.1523/JNEUROSCI.2198-11.2011
137. LANJEWAR, A. L., JAGETIA, S., KHAN, Z. M., EAGLESON, K. L., & LEVITT, P.
(2023). SUBCLASS-SPECIFIC EXPRESSION PATTERNS OF MET RECEPTOR TYROSINE
KINASE DURING DEVELOPMENT IN MEDIAL PREFRONTAL AND VISUAL CORTICES. JOURNAL
OF COMPARATIVE NEUROLOGY, 531(1), 132–148.
HTTPS://DOI.ORG/10.1002/CNE.25418
138. LE MERRE, P., ÄHRLUND-RICHTER, S., & CARLÉN, M. (2021). THE MOUSE
PREFRONTAL CORTEX: UNITY IN DIVERSITY. NEURON, 109(12), 1925–1944.
HTTPS://DOI.ORG/10.1016/J.NEURON.2021.03.035
139. LEE, C., KIM, Y., & KAANG, B. K. (2022). THE PRIMARY MOTOR CORTEX:
THE HUB OF MOTOR LEARNING IN RODENTS. NEUROSCIENCE, 485, 163–170.
HTTPS://DOI.ORG/10.1016/J.NEUROSCIENCE.2022.01.009
140. LEONE, D. P., HEAVNER, W. E., FERENCZI, E. A., DOBREVA, G.,
HUGUENARD, J. R., GROSSCHEDL, R., & MCCONNELL, S. K. (2015). SATB2
REGULATES THE DIFFERENTIATION OF BOTH CALLOSAL AND SUBCEREBRAL PROJECTION
NEURONS IN THE DEVELOPING CEREBRAL CORTEX. CEREB CORTEX, 25(10), 3406–
3419. HTTPS://DOI.ORG/10.1093/CERCOR/BHU156
141. LEONE, D. P., SRINIVASAN, K., CHEN, B., ALCAMO, E., & MCCONNELL, S. K.
(2008). THE DETERMINATION OF PROJECTION NEURON IDENTITY IN THE DEVELOPING
CEREBRAL CORTEX. CURRENT OPINION IN NEUROBIOLOGY, 18(1), 28.
HTTPS://DOI.ORG/10.1016/J.CONB.2008.05.006
193
142. LEVITT, P., BARBE, M. F., & EAGLESON, K. L. (2003). PATTERNING AND
SPECIFICATION OF THE CEREBRAL CORTEX. HTTPS://DOI-
ORG.LIBPROXY2.USC.EDU/10.1146/ANNUREV.NEURO.20.1.1, 20, 1–2.
HTTPS://DOI.ORG/10.1146/ANNUREV.NEURO.20.1.1
143. LI, H., HORNS, F., XIE, Q., XIE, Q., LI, T., LUGINBUHL, D. J., LUO, L., &
QUAKE, S. R. (2017). CLASSIFYING DROSOPHILA OLFACTORY PROJECTION NEURON
SUBTYPES BY SINGLE-CELL RNA SEQUENCING. CELL, 171(5), 1206.E22-1220.E22.
HTTPS://DOI.ORG/10.1016/J.CELL.2017.10.019
144. LI, M., SENDTNER, M., & SMITH, A. (1995). ESSENTIAL FUNCTION OF LIF
RECEPTOR IN MOTOR NEURONS. NATURE, 378(6558), 724–727.
HTTPS://DOI.ORG/10.1038/378724A0
145. LIM, C. S., & WALIKONIS, R. S. (2008). HEPATOCYTE GROWTH FACTOR AND
C-MET PROMOTE DENDRITIC MATURATION DURING HIPPOCAMPAL NEURON
DIFFERENTIATION VIA THE AKT PATHWAY. CELLULAR SIGNALLING, 20(5), 825–835.
HTTPS://DOI.ORG/10.1016/J.CELLSIG.2007.12.013
146. LINVILLE, A., GUMUSANELI, E., CHANDRARATNA, R. A. S., & SCHILLING, T.
F. (2004). INDEPENDENT ROLES FOR RETINOIC ACID IN SEGMENTATION AND NEURONAL
DIFFERENTIATION IN THE ZEBRAFISH HINDBRAIN. DEVELOPMENTAL BIOLOGY, 270(1),
186–199. HTTPS://DOI.ORG/10.1016/J.YDBIO.2004.02.022
147. LIU, Q., & WONG-RILEY, M. T. T. (2013). POSTNATAL DEVELOPMENT OF
BRAIN-DERIVED NEUROTROPHIC FACTOR (BDNF) AND TYROSINE PROTEIN KINASE B
(TRKB) RECEPTOR IMMUNOREACTIVITY IN MULTIPLE BRAIN STEM RESPIRATORY-RELATED
NUCLEI OF THE RAT. JOURNAL OF COMPARATIVE NEUROLOGY, 521(1), 109–129.
HTTPS://DOI.ORG/10.1002/CNE.23164
148. LIU, Y., COOK, C., SEDGEWICK, A. J., ZHANG, S., FASSETT, M. S.,
RICARDO-GONZALEZ, R. R., HARIRCHIAN, P., KASHEM, S. W., HANAKAWA, S.,
LEISTICO, J. R., NORTH, J. P., TAYLOR, M. A., ZHANG, W., MAN, M. Q., CHARRUYER,
A., BELIAKOVA-BETHELL, N., BENZ, S. C., GHADIALLY, R., MAURO, T. M., … CHENG,
J. B. (2020). SINGLE-CELL PROFILING REVEALS DIVERGENT, GLOBALLY PATTERNED
IMMUNE RESPONSES IN MURINE SKIN INFLAMMATION. ISCIENCE, 23(10), 101582.
HTTPS://DOI.ORG/10.1016/J.ISCI.2020.101582
149. LUMSDEN, A., & KEYNES, R. (1989). SEGMENTAL PATTERNS OF NEURONAL
DEVELOPMENT IN THE CHICK HINDBRAIN. NATURE, 337(6206), 424–428.
HTTPS://DOI.ORG/10.1038/337424A0
150. LUMSDEN, A., & KRUMLAF, R. (1996). PATTERNING THE VERTEBRATE
NEURAXIS. SCIENCE, 274(5290), 1109–1116.
HTTP://GO.GALEGROUP.COM.LIBPROXY2.USC.EDU/PS/I.DO?ID=GALE%7CA18928373
&V=2.1&U=USOCAL_MAIN&IT=R&P=AONE&SW=W
194
151. MA, X., & QIU, S. (2020). CONTROL OF CORTICAL SYNAPSE DEVELOPMENT
AND PLASTICITY BY MET RECEPTOR TYROSINE KINASE, A GENETIC RISK FACTOR FOR
AUTISM. JOURNAL OF NEUROSCIENCE RESEARCH, 98(11), 2115–2129.
HTTPS://DOI.ORG/10.1002/JNR.24542
152. MACLENNAN, A. J., VINSON, E. N., MARKS, L., MCLAURIN, D. L., PFEIFER,
M., & LEE, N. (1996). IMMUNOHISTOCHEMICAL LOCALIZATION OF CILIARY
NEUROTROPHIC FACTOR RECEPTOR Α EXPRESSION IN THE RAT NERVOUS SYSTEM.
JOURNAL OF NEUROSCIENCE, 16(2), 621–630.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.16-02-00621.1996
153. MAGRINELLI, E., BAUMANN, N., WAGENER, R. J., GLANGETAS, C., BELLONE,
C., JABAUDON, D., & KLINGLER, E. (2022). HETEROGENEOUS FATES OF
SIMULTANEOUSLY-BORN NEURONS IN THE CORTICAL VENTRICULAR ZONE. SCIENTIFIC
REPORTS 2022 12:1, 12(1), 1–11. HTTPS://DOI.ORG/10.1038/S41598-022-09740-6
154. MAINA, F, HILTON, M. C., PONZETTO, C., DAVIES, A. M., & KLEIN, R.
(1997). MET RECEPTOR SIGNALING IS REQUIRED FOR SENSORY NERVE DEVELOPMENT
AND HGF PROMOTES AXONAL GROWTH AND SURVIVAL OF SENSORY NEURONS. GENES &
DEVELOPMENT, 11(24), 3341–3350. HTTPS://DOI.ORG/10.1101/GAD.11.24.3341
155. MAINA, FLAVIO, HILTON, M. C., ANDRES, R., WYATT, S., KLEIN, R., &
DAVIES, A. M. (1998). MULTIPLE ROLES FOR HEPATOCYTE GROWTH FACTOR IN
SYMPATHETIC NEURON DEVELOPMENT. NEURON, 20(5), 835–846.
HTTPS://DOI.ORG/10.1016/S0896-6273(00)80466-3
156. MARK, M., LUFKIN, T., VONESCH, J. L., RUBERTE, E., OLIVO, J. C., DOLLE,
P., GORRY, P., LUMSDEN, A., & CHAMBON, P. (1993). TWO RHOMBOMERES ARE
ALTERED IN HOXA-1 MUTANT MICE. DEVELOPMENT, 119(2).
HTTPS://DEV.BIOLOGISTS.ORG/CONTENT/119/2/319.LONG
157. MARSHALL, H., NONCHEV, S., SHAM, M. H., MUCHAMORE, I., LUMSDEN, A.,
& KRUMLAUF, R. (1992). RETINOIC ACID ALTERS HINDBRAIN HOX CODE AND INDUCES
TRANSFORMATION OF RHOMBOMERES 2/3 INTO A 4/5 IDENTITY. NATURE, 360(6406),
737–741. HTTPS://DOI.ORG/10.1038/360737A0
158. MARSHEL, J. H., GARRETT, M. E., NAUHAUS, I., & CALLAWAY, E. M. (2011).
FUNCTIONAL SPECIALIZATION OF SEVEN MOUSE VISUAL CORTICAL AREAS. NEURON,
72(6), 1040. HTTPS://DOI.ORG/10.1016/J.NEURON.2011.12.004
159. MAYNARD, K. R., TIPPANI, M., TAKAHASHI, Y., PHAN, B. D. N., HYDE, T. M.,
JAFFE, A. E., & MARTINOWICH, K. (2020). DOTDOTDOT: AN AUTOMATED APPROACH TO
QUANTIFY MULTIPLEX SINGLE MOLECULE FLUORESCENT IN SITU HYBRIDIZATION
(SMFISH) IMAGES IN COMPLEX TISSUES. NUCLEIC ACIDS RESEARCH, 48(11), E66.
HTTPS://DOI.ORG/10.1093/NAR/GKAA312
195
160. MCEVILLY, R. J., ORTIZ DE DIAZ, M., SCHONEMANN, M. D., HOOSHMAND,
F., & ROSENFELD, M. G. (2002). TRANSCRIPTIONAL REGULATION OF CORTICAL NEURON
MIGRATION BY POU DOMAIN FACTORS. SCIENCE, 295(5559), 1528–1532.
HTTPS://DOI.ORG/10.1126/SCIENCE.1067132
161. MCNULTY, C. L., PERES, J. N., BARDINE, N., VAN DEN AKKER, W. M. R., &
DURSTON, A. J. (2005). KNOCKDOWN OF THE COMPLETE HOX PARALOGOUS GROUP 1
LEADS TO DRAMATIC HINDBRAIN AND NEURAL CREST DEFECTS. DEVELOPMENT
(CAMBRIDGE, ENGLAND), 132(12), 2861–2871. HTTPS://DOI.ORG/10.1242/DEV.01872
162. MIKAELS, Å., LIVET, J., WESTPHAL, H., DE LAPEYRIÈRE, O., & ERNFORS, P.
(2000). A DYNAMIC REGULATION OF GDNF-FAMILY RECEPTORS CORRELATES WITH A
SPECIFIC TROPHIC DEPENDENCY OF CRANIAL MOTOR NEURON SUBPOPULATIONS DURING
DEVELOPMENT. EUROPEAN JOURNAL OF NEUROSCIENCE, 12(2), 446–456.
HTTPS://DOI.ORG/10.1046/J.1460-9568.2000.00924.X
163. MOLNÁR, Z., CLOWRY, G. J., ŠESTAN, N., ALZU’BI, A., BAKKEN, T., HEVNER,
R. F., HÜPPI, P. S., KOSTOVIĆ, I., RAKIC, P., ANTON, E. S., EDWARDS, D., GARCEZ,
P., HOERDER-SUABEDISSEN, A., & KRIEGSTEIN, A. (2019). NEW INSIGHTS INTO THE
DEVELOPMENT OF THE HUMAN CEREBRAL CORTEX. JOURNAL OF ANATOMY, 235(3),
432–451. HTTPS://DOI.ORG/10.1111/JOA.13055
164. MOLYNEAUX, B. J., ARLOTTA, P., HIRATA, T., HIBI, M., & MACKLIS, J. D.
(2005). FEZL IS REQUIRED FOR THE BIRTH AND SPECIFICATION OF CORTICOSPINAL
MOTOR NEURONS. NEURON, 47(6), 817–831.
HTTPS://DOI.ORG/10.1016/J.NEURON.2005.08.030
165. MOLYNEAUX, B. J., ARLOTTA, P., MENEZES, J. R. L., & MACKLIS, J. D.
(2007). NEURONAL SUBTYPE SPECIFICATION IN THE CEREBRAL CORTEX. NATURE
REVIEWS NEUROSCIENCE 2007 8:6, 8(6), 427–437.
HTTPS://DOI.ORG/10.1038/NRN2151
166. MONTAGUE, K., LOWE, A. S., UZQUIANO, A., KNÜFER, A., ASTICK, M.,
PRICE, S. R., & GUTHRIE, S. (2017). THE ASSEMBLY OF DEVELOPING MOTOR NEURONS
DEPENDS ON AN INTERPLAY BETWEEN SPONTANEOUS ACTIVITY, TYPE II CADHERINS AND
GAP JUNCTIONS. DEVELOPMENT (CAMBRIDGE, ENGLAND), 144(5), 830–836.
HTTPS://DOI.ORG/10.1242/DEV.144063
167. MOODY, S. A., & HEATON, M. B. (1983). DEVELOPMENTAL RELATIONSHIPS
BETWEEN TRIGEMINAL GANGLIA AND TRIGEMINAL MOTONEURONS IN CHICK EMBRYOS. II.
GANGLION AXON INGROWTH GUIDES MOTONEURON MIGRATION. THE JOURNAL OF
COMPARATIVE NEUROLOGY, 213(3), 344–349.
HTTPS://DOI.ORG/10.1002/CNE.902130309
196
168. MOORE, M. W., KLEIN, R. D., FARINAS, I., SAUER, H., ARMANINI, M.,
PHILLIPS, H., REICHARDT, L. F., RYAN, A. M., CARVER-MOORE, K., & ROSENTHAL, A.
(1996). RENAL AND NEURONAL ABNORMALITIES IN MICE LACKING GDNF. NATURE,
382(6586), 76–79. HTTPS://DOI.ORG/10.1038/382076A0
169. MORTON, R. E., PINNINGTON, L., & ELLIS, R. E. (2007). AIR SWALLOWING IN
RETT SYNDROME. DEVELOPMENTAL MEDICINE & CHILD NEUROLOGY, 42(4), 271–275.
HTTPS://DOI.ORG/10.1111/J.1469-8749.2000.TB00084.X
170. MUKHTAR, T., BREDA, J., ADAM, M. A., BOARETO, M., GROBECKER, P.,
KARIMADDINI, Z., GRISON, A., ESCHBACH, K., CHANDRASEKHAR, R., VERMEUL, S.,
OKONIEWSKI, M., PACHKOV, M., HARWELL, C. C., ATANASOSKI, S., BEISEL, C., IBER,
D., VAN NIMWEGEN, E., & TAYLOR, V. (2022). TEMPORAL AND SEQUENTIAL
TRANSCRIPTIONAL DYNAMICS DEFINE LINEAGE SHIFTS IN CORTICOGENESIS. THE EMBO
JOURNAL, 41(24), 1–29. HTTPS://DOI.ORG/10.15252/EMBJ.2022111132
171. MUÑOZ-CASTAÑEDA, R., ZINGG, B., MATHO, K. S., CHEN, X., WANG, Q.,
FOSTER, N. N., LI, A., NARASIMHAN, A., HIROKAWA, K. E., HUO, B., BANNERJEE, S.,
KOROBKOVA, L., PARK, C. S., PARK, Y. G., BIENKOWSKI, M. S., CHON, U., WHEELER,
D. W., LI, X., WANG, Y., … DONG, H. W. (2021). CELLULAR ANATOMY OF THE MOUSE
PRIMARY MOTOR CORTEX. NATURE 2021 598:7879, 598(7879), 159–166.
HTTPS://DOI.ORG/10.1038/S41586-021-03970-W
172. MURPHY, A. E., SKENE, N. G., ARISING, F., ZIMMERMAN, K. D., LANGEFELD,
M. A. &, & NATURE, C. D. (2022). A BALANCED MEASURE SHOWS SUPERIOR
PERFORMANCE OF PSEUDOBULK METHODS IN SINGLE-CELL RNA-SEQUENCING
ANALYSIS. NATURE COMMUNICATIONS 2022 13:1, 13(1), 1–4.
HTTPS://DOI.ORG/10.1038/S41467-022-35519-4
173. NAKAMURA, T., NISHIZAWA, T., HAGIYA, M., SEKI, T., SHIMONISHI, M.,
SUGIMURA, A., TASHIRO, K., & SHIMIZU, S. (1989). MOLECULAR CLONING AND
EXPRESSION OF HUMAN HEPATOCYTE GROWTH FACTOR. NATURE, 342(6248), 440–
443. HTTPS://DOI.ORG/10.1038/342440A0
174. NAKASHIMA, K., WIESE, S., YANAGISAWA, M., ARAKAWA, H., KIMURA, N.,
HISATSUNE, T., YOSHIDA, K., KISHIMOTO, T., SENDTNER, M., & TAGA, T. (1999A).
DEVELOPMENTAL REQUIREMENT OF GP130 SIGNALING IN NEURONAL SURVIVAL AND
ASTROCYTE DIFFERENTIATION. JOURNAL OF NEUROSCIENCE, 19(13), 5429–5434.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.19-13-05429.1999
175. NAKASHIMA, K., WIESE, S., YANAGISAWA, M., ARAKAWA, H., KIMURA, N.,
HISATSUNE, T., YOSHIDA, K., KISHIMOTO, T., SENDTNER, M., & TAGA, T. (1999B).
DEVELOPMENTAL REQUIREMENT OF GP130 SIGNALING IN NEURONAL SURVIVAL AND
ASTROCYTE DIFFERENTIATION. JOURNAL OF NEUROSCIENCE, 19(13), 5429–5434.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.19-13-05429.1999
197
176. NGUYEN, H. T. T., GUEVARRA, R. B., MAGEZ, S., & RADWANSKA, M.
(2021). SINGLE-CELL TRANSCRIPTOME PROFILING AND THE USE OF AID DEFICIENT MICE
REVEAL THAT B CELL ACTIVATION COMBINED WITH ANTIBODY CLASS SWITCH
RECOMBINATION AND SOMATIC HYPERMUTATION DO NOT BENEFIT THE CONTROL OF
EXPERIMENTAL TRYPANOSOMOSIS. PLOS PATHOGENS, 17(11), E1010026.
HTTPS://DOI.ORG/10.1371/JOURNAL.PPAT.1010026
177. NIETO, M., MONUKI, E. S., TANG, H., IMITOLA, J., HAUBST, N., KHOURY, S.
J., CUNNINGHAM, J., GOTZ, M., & WALSH, C. A. (2004). EXPRESSION OF CUX-1 AND
CUX-2 IN THE SUBVENTRICULAR ZONE AND UPPER LAYERS II–IV OF THE CEREBRAL
CORTEX. JOURNAL OF COMPARATIVE NEUROLOGY, 479(2), 168–180.
HTTPS://DOI.ORG/10.1002/CNE.20322
178. NOVAK, K. D., PREVETTE, D., WANG, S., GOULD, T. W., & OPPENHEIM, R.
W. (2000). HEPATOCYTE GROWTH FACTOR/SCATTER FACTOR IS A NEUROTROPHIC
SURVIVAL FACTOR FOR LUMBAR BUT NOT FOR OTHER SOMATIC MOTONEURONS IN THE
CHICK EMBRYO. JOURNAL OF NEUROSCIENCE, 20(1), 326–337.
HTTPS://DOI.ORG/10.1523/JNEUROSCI.20-01-00326.2000
179. NUNEZ, A. A., POMERANTZ, S. M., BEAN, N. J., & YOUNGSTROM, T. G.
(1985). EFFECTS OF LARYNGEAL DENERVATION ON ULTRASOUND PRODUCTION AND
MALE SEXUAL BEHAVIOR IN RODENTS. PHYSIOLOGY AND BEHAVIOR, 34(6), 901–905.
HTTPS://DOI.ORG/10.1016/0031-9384(85)90011-3
180. OH, J. Y., NAM, Y. J., JO, A., CHEON, H. S., RHEE, S. M., PARK, J. K., LEE,
J. A., & KIM, H. K. (2010). APOLIPOPROTEIN E MRNA IS TRANSPORTED TO DENDRITES
AND MAY HAVE A ROLE IN SYNAPTIC STRUCTURAL PLASTICITY. JOURNAL OF
NEUROCHEMISTRY, 114(3), 685–696. HTTPS://DOI.ORG/10.1111/J.1471-
4159.2010.06773.X
181. OKANO, H. J., & DARNELL, R. B. (1997). A HIERARCHY OF HU RNA BINDING
PROTEINS IN DEVELOPING AND ADULT NEURONS. JOURNAL OF NEUROSCIENCE, 17(9),
3024–3037. HTTPS://DOI.ORG/10.1523/JNEUROSCI.17-09-03024.1997
182. OPPENHEIM, R. W. (1991). CELL DEATH DURING DEVELOPMENT OF THE
NERVOUS SYSTEM. ANNUAL REVIEW OF NEUROSCIENCE, 14, 453–501.
HTTPS://DOI.ORG/10.1146/ANNUREV.NE.14.030191.002321
183. OPPENHEIM, RONALD W., HOUENOU, L. J., PARSADANIAN, A. S., PREVETTE,
D., SNIDER, W. D., & SHEN, L. (2000). GLIAL CELL LINE-DERIVED NEUROTROPHIC
FACTOR AND DEVELOPING MAMMALIAN MOTONEURONS: REGULATION OF PROGRAMMED
CELL DEATH AMONG MOTONEURON SUBTYPES. JOURNAL OF NEUROSCIENCE, 20(13),
5001–5011. HTTPS://DOI.ORG/10.1523/JNEUROSCI.20-13-05001.2000
198
184. ORTIZ, C., NAVARRO, J. F., JUREK, A., MÄRTIN, A., LUNDEBERG, J., &
MELETIS, K. (2020). MOLECULAR ATLAS OF THE ADULT MOUSE BRAIN. SCIENCE
ADVANCES, 6(26).
185. HTTPS://DOI.ORG/10.1126/SCIADV.ABB3446/SUPPL_FILE/ABB3446_S
M.PDF
186. ÖZEL, M. N., SIMON, F., JAFARI, S., HOLGUERA, I., CHEN, Y. C., BENHRA,
N., EL-DANAF, R. N., KAPURALIN, K., MALIN, J. A., KONSTANTINIDES, N., & DESPLAN,
C. (2020). NEURONAL DIVERSITY AND CONVERGENCE IN A VISUAL SYSTEM
DEVELOPMENTAL ATLAS. NATURE 2020 589:7840, 589(7840), 88–95.
HTTPS://DOI.ORG/10.1038/S41586-020-2879-3
187. PARK, S. H. E., ORTIZ, A. K., & KONOPKA, G. (2022). CORTICOGENESIS
ACROSS SPECIES AT SINGLE-CELL RESOLUTION. IN DEVELOPMENTAL NEUROBIOLOGY
(VOL. 82, ISSUE 6, PP. 517–532). DEV NEUROBIOL.
HTTPS://DOI.ORG/10.1002/DNEU.22896
188. PEÑA, C. J., KRONMAN, H. G., WALKER, D. M., CATES, H. M., BAGOT, R.
C., PURUSHOTHAMAN, I., ISSLER, O., LOH, Y. E., LEONG, T., KIRALY, D. D.,
GOODMAN, E., NEVE, R. L., SHEN, L., & NESTLER, E. J. (2017). EARLY LIFE STRESS
CONFERS LIFELONG STRESS SUSCEPTIBILITY IN MICE VIA VENTRAL TEGMENTAL AREA
OTX2. 1188(JUNE), 1185–1188.
189. PENG, Y., LU, Z., LI, G., PIECHOWICZ, M., ANDERSON, M., UDDIN, Y., WU,
J., & QIU, S. (2016). THE AUTISM-ASSOCIATED MET RECEPTOR TYROSINE KINASE
ENGAGES EARLY NEURONAL GROWTH MECHANISM AND CONTROLS GLUTAMATERGIC
CIRCUITS DEVELOPMENT IN THE FOREBRAIN. MOLECULAR PSYCHIATRY, 21(7), 925–
935. HTTPS://DOI.ORG/10.1038/MP.2015.182
190. PENG, YUN, HUENTELMAN, M., SMITH, C., & QIU, S. (2013). MET
RECEPTOR TYROSINE KINASE AS AN AUTISM GENETIC RISK FACTOR. IN INTERNATIONAL
REVIEW OF NEUROBIOLOGY (VOL. 113, PP. 135–165). ACADEMIC PRESS INC.
HTTPS://DOI.ORG/10.1016/B978-0-12-418700-9.00005-8
191. PENNYPACKER, K., FISCHER, I., & LEVITT, P. (1991). EARLY IN VITRO
GENESIS AND DIFFERENTIATION OF AXONS AND DENDRITES BY HIPPOCAMPAL NEURONS
ANALYZED QUANTITATIVELY WITH NEUROFILAMENT-H AND MICROTUBULE-ASSOCIATED
PROTEIN 2 ANTIBODIES. EXPERIMENTAL NEUROLOGY, 111(1), 25–35.
HTTPS://DOI.ORG/10.1016/0014-4886(91)90047-G
192. PHILIPPIDOU, P., & DASEN, J. S. (2013). HOX GENES: CHOREOGRAPHERS IN
NEURAL DEVELOPMENT, ARCHITECTS OF CIRCUIT ORGANIZATION. NEURON, 80(1), 12–
34. HTTPS://DOI.ORG/10.1016/J.NEURON.2013.09.020
193. PIERCE, E. T. (1973). TIME OF ORIGIN OF NEURONS IN THE BRAIN STEM OF
THE MOUSE. PROGRESS IN BRAIN RESEARCH, 40(C), 53–65.
HTTPS://DOI.ORG/10.1016/S0079-6123(08)60679-2
199
194. PITTS, T., & ICEMAN, K. E. (2022). DEGLUTITION AND THE REGULATION OF
THE SWALLOW MOTOR PATTERN. HTTPS://DOI.ORG/10.1152/PHYSIOL.00005.2021,
38(1). HTTPS://DOI.ORG/10.1152/PHYSIOL.00005.2021
195. PLUMMER, J. T., EVGRAFOV, O. V., BERGMAN, M. Y., FRIEZ, M., HAIMAN, C.
A., LEVITT, P., & ALDINGER, K. A. (2013). TRANSCRIPTIONAL REGULATION OF THE MET
RECEPTOR TYROSINE KINASE GENE BY MECP2 AND SEX-SPECIFIC EXPRESSION IN
AUTISM AND RETT SYNDROME. TRANSLATIONAL PSYCHIATRY, 3(10), E316–E316.
HTTPS://DOI.ORG/10.1038/TP.2013.91
196. PORTFORS, C. V. (2007). TYPES AND FUNCTIONS OF ULTRASONIC
VOCALIZATIONS IN LABORATORY RATS AND MICE. IN JOURNAL OF THE AMERICAN
ASSOCIATION FOR LABORATORY ANIMAL SCIENCE.
HTTPS://WWW.INGENTACONNECT.COM/CONTENT/AALAS/JAALAS/2007/00000046/00000
001/ART00005?CRAWLER=TRUE&MIMETYPE=APPLICATION/PDF
197. PORTFORS, C. V, & PERKEL, D. J. (2014). THE ROLE OF ULTRASONIC
VOCALIZATIONS IN MOUSE COMMUNICATION. CURRENT OPINION IN NEUROBIOLOGY, 28,
115–120. HTTPS://DOI.ORG/10.1016/J.CONB.2014.07.002
198. POWELL, E. M., MARS, W. M., & LEVITT, P. (2001). HEPATOCYTE GROWTH
FACTOR/SCATTER FACTOR IS A MOTOGEN FOR INTERNEURONS MIGRATING FROM THE
VENTRAL TO DORSAL TELENCEPHALON. NEURON, 30(1), 79–89.
HTTPS://DOI.ORG/10.1016/S0896-6273(01)00264-1
199. PUELLES, L., TVRDIK, P., & MARTÍNEZ-DE-LA-TORRE, M. (2019). THE
POSTMIGRATORY ALAR TOPOGRAPHY OF VISCERAL CRANIAL NERVE EFFERENTS
CHALLENGES THE CLASSICAL MODEL OF HINDBRAIN COLUMNS. THE ANATOMICAL
RECORD, 302(3), 485–504. HTTPS://DOI.ORG/10.1002/AR.23830
200. QUENOUILLE, M. H. (1956). NOTES ON BIAS IN ESTIMATION. BIOMETRIKA,
43(3/4), 353. HTTPS://DOI.ORG/10.2307/2332914
201. RAKIC, P. (1972). MODE OF CELL MIGRATION TO THE SUPERFICIAL LAYERS
OF FETAL MONKEY NEOCORTEX. J. COMP. NEUROL., 145(1), 61–83.
HTTPS://DOI.ORG/10.1002/CNE.901450105
202. RAKIC, P. (1974). NEURONS IN RHESUS MONKEY VISUAL CORTEX:
SYSTEMATIC RELATION BETWEEN TIME OF ORIGIN AND EVENTUAL DISPOSITION. SCIENCE,
183(4123), 425–427. HTTPS://DOI.ORG/10.1126/SCIENCE.183.4123.425
203. RAKIC, P. (1988). SPECIFICATION OF CEREBRAL CORTICAL AREAS.
SCIENCE, 241(4862), 170–176. HTTPS://DOI.ORG/10.1126/SCIENCE.3291116
200
204. RAKIC, P., AYOUB, A. E., BREUNIG, J. J., & DOMINGUEZ, M. H. (2009).
DECISION BY DIVISION: MAKING CORTICAL MAPS. TRENDS IN NEUROSCIENCES, 32(5),
291. HTTPS://DOI.ORG/10.1016/J.TINS.2009.01.007
205. RANA, S., SIECK, G. C., & MANTILLA, C. B. (2020). HETEROGENEOUS
GLUTAMATERGIC RECEPTOR MRNA EXPRESSION ACROSS PHRENIC MOTOR NEURONS IN
RATS. JOURNAL OF NEUROCHEMISTRY, 153(5), 586–598.
HTTPS://DOI.ORG/10.1111/JNC.14881
206. RENIER, N., ADAMS, E. L., KIRST, C., WU, Z., AZEVEDO, R., KOHL, J.,
AUTRY, A. E., KADIRI, L., UMADEVI VENKATARAJU, K., ZHOU, Y., WANG, V. X., TANG,
C. Y., OLSEN, O., DULAC, C., OSTEN, P., & TESSIER-LAVIGNE, M. (2016). MAPPING
OF BRAIN ACTIVITY BY AUTOMATED VOLUME ANALYSIS OF IMMEDIATE EARLY GENES.
CELL, 165(7), 1789. HTTPS://DOI.ORG/10.1016/J.CELL.2016.05.007
207. RICE, C. J., SANDMAN, C. A., LENJAVI, M. R., & BARAM, T. Z. (2008). A
NOVEL MOUSE MODEL FOR ACUTE AND LONG-LASTING CONSEQUENCES OF EARLY LIFE
STRESS. ENDOCRINOLOGY, 149(10), 4892. HTTPS://DOI.ORG/10.1210/EN.2008-0633
208. RIEDE, T. (2011). SUBGLOTTAL PRESSURE, TRACHEAL AIRFLOW, AND
INTRINSIC LARYNGEAL MUSCLE ACTIVITY DURING RAT ULTRASOUND VOCALIZATION.
JOURNAL OF NEUROPHYSIOLOGY, 106(5), 2580–2592.
HTTPS://DOI.ORG/10.1152/JN.00478.2011
209. RIEDE, T. (2013). STEREOTYPIC LARYNGEAL AND RESPIRATORY MOTOR
PATTERNS GENERATE DIFFERENT CALL TYPES IN RAT ULTRASOUND VOCALIZATION.
JOURNAL OF EXPERIMENTAL ZOOLOGY PART A: ECOLOGICAL GENETICS AND
PHYSIOLOGY, 319(4), 213–224. HTTPS://DOI.ORG/10.1002/JEZ.1785
210. RINAMAN, L., & LEVITT, P. (1993). ESTABLISHMENT OF VAGAL
SENSORIMOTOR CIRCUITS DURING FETAL DEVELOPMENT IN RATS. JOURNAL OF
NEUROBIOLOGY, 24(5), 641–659. HTTPS://DOI.ORG/10.1002/NEU.480240509
211. ROCHEFORT, N. L., GARASCHUK, O., MILOS, R. I., NARUSHIMA, M.,
MARANDI, N., PICHLER, B., KOVALCHUK, Y., & KONNERTH, A. (2009). SPARSIFICATION
OF NEURONAL ACTIVITY IN THE VISUAL CORTEX AT EYE-OPENING. PROCEEDINGS OF THE
NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 106(35),
15049. HTTPS://DOI.ORG/10.1073/PNAS.0907660106
212. ROUBERTOUX, P. L., MARTIN, B., LE ROY, I. LE, BEAU, J., MARCHALAND, C.,
PEREZ-DIAZ, F., COHEN-SALMON, C., & CARLIER, M. (1996). VOCALIZATIONS IN
NEWBORN MICE: GENETIC ANALYSIS. BEHAVIOR GENETICS, 26(4), 427–437.
HTTPS://DOI.ORG/10.1007/BF02359487
213. SAMATA, B., TAKAICHI, R., ISHII, Y., FUKUSHIMA, K., NAKAGAWA, H., ONO,
Y., & TAKAHASHI, J. (2020). L1CAM IS A MARKER FOR ENRICHING CORTICOSPINAL
201
MOTOR NEURONS IN THE DEVELOPING BRAIN. FRONTIERS IN CELLULAR
NEUROSCIENCE, 14. HTTPS://DOI.ORG/10.3389/FNCEL.2020.00031/FULL
214. SCATTONI, M. L., CRAWLEY, J., & RICCERI, L. (2009). ULTRASONIC
VOCALIZATIONS: A TOOL FOR BEHAVIOURAL PHENOTYPING OF MOUSE MODELS OF
NEURODEVELOPMENTAL DISORDERS. NEUROSCIENCE AND BIOBEHAVIORAL REVIEWS,
33(4), 508. HTTPS://DOI.ORG/10.1016/J.NEUBIOREV.2008.08.003
215. SCHALLER, S., BUTTIGIEG, D., ALORY, A., JACQUIER, A., BARAD, M.,
MERCHANT, M., GENTIEN, D., DE LA GRANGE, P., & HAASE, G. (2017A). NOVEL
COMBINATORIAL SCREENING IDENTIFIES NEUROTROPHIC FACTORS FOR SELECTIVE
CLASSES OF MOTOR NEURONS. PROCEEDINGS OF THE NATIONAL ACADEMY OF
SCIENCES OF THE UNITED STATES OF AMERICA, 114(12), E2486–E2493.
HTTPS://DOI.ORG/10.1073/PNAS.1615372114
216. SCHALLER, S., BUTTIGIEG, D., ALORY, A., JACQUIER, A., BARAD, M.,
MERCHANT, M., GENTIEN, D., DE LA GRANGE, P., & HAASE, G. (2017B). NOVEL
COMBINATORIAL SCREENING IDENTIFIES NEUROTROPHIC FACTORS FOR SELECTIVE
CLASSES OF MOTOR NEURONS. PROCEEDINGS OF THE NATIONAL ACADEMY OF
SCIENCES, 114(12), E2486–E2493. HTTPS://DOI.ORG/10.1073/PNAS.1615372114
217. SCHINDELIN, J., ARGANDA-CARRERAS, I., FRISE, E., KAYNIG, V., LONGAIR,
M., PIETZSCH, T., PREIBISCH, S., RUEDEN, C., SAALFELD, S., SCHMID, B., TINEVEZ, J.
Y., WHITE, D. J., HARTENSTEIN, V., ELICEIRI, K., TOMANCAK, P., & CARDONA, A.
(2012). FIJI: AN OPEN-SOURCE PLATFORM FOR BIOLOGICAL-IMAGE ANALYSIS. NATURE
METHODS 2012 9:7, 9(7), 676–682. HTTPS://DOI.ORG/10.1038/NMETH.2019
218. SCHMIDT, U., WEIGERT, M., BROADDUS, C., & MYERS, G. (2018). CELL
DETECTION WITH STAR-CONVEX POLYGONS. LECTURE NOTES IN COMPUTER SCIENCE
(INCLUDING SUBSERIES LECTURE NOTES IN ARTIFICIAL INTELLIGENCE AND LECTURE
NOTES IN BIOINFORMATICS), 11071 LNCS, 265–273. HTTPS://DOI.ORG/10.1007/978-
3-030-00934-2_30
219. SCHNEIDER, C. A., RASBAND, W. S., & ELICEIRI, K. W. (2012). NIH IMAGE
TO IMAGEJ: 25 YEARS OF IMAGE ANALYSIS. NATURE METHODS 2012 9:7, 9(7), 671–
675. HTTPS://DOI.ORG/10.1038/NMETH.2089
220. SCOTT, M. M., WILLIAMS, K. W., ROSSI, J., LEE, C. E., & ELMQUIST, J. K.
(2011). LEPTIN RECEPTOR EXPRESSION IN HINDBRAIN GLP-1 NEURONS REGULATES
FOOD INTAKE AND ENERGY BALANCE IN MICE. JOURNAL OF CLINICAL INVESTIGATION,
121(6), 2413–2421. HTTPS://DOI.ORG/10.1172/JCI43703
221. SHAH, S., LUBECK, E., ZHOU, W., & CAI, L. (2017). SEQFISH ACCURATELY
DETECTS TRANSCRIPTS IN SINGLE CELLS AND REVEALS ROBUST SPATIAL
ORGANIZATION IN THE HIPPOCAMPUS. NEURON, 94(4), 752-758.E1.
HTTPS://DOI.ORG/10.1016/J.NEURON.2017.05.008
202
222. SHAHAM, S. (2006). GLIA–NEURON INTERACTIONS IN THE NERVOUS SYSTEM
OF CAENORHABDITIS ELEGANS. CURRENT OPINION IN NEUROBIOLOGY, 16(5), 522–
528. HTTPS://DOI.ORG/10.1016/J.CONB.2006.08.001
223. SHEINKOPF, S. J., LEVINE, T. P., MCCORMICK, C. E. B., PUGGIONI, G.,
CONRADT, E., LAGASSE, L. L., & LESTER, B. M. (2019). DEVELOPMENTAL
TRAJECTORIES OF AUTONOMIC FUNCTIONING IN AUTISM FROM BIRTH TO EARLY
CHILDHOOD. BIOLOGICAL PSYCHOLOGY, 142, 13–18.
HTTPS://DOI.ORG/10.1016/J.BIOPSYCHO.2019.01.003
224. SHEPHERD, G. M. G. (2009). INTRACORTICAL CARTOGRAPHY IN AN
AGRANULAR AREA. FRONTIERS IN NEUROSCIENCE, 3(DEC), 337–343.
HTTPS://DOI.ORG/10.3389/NEURO.01.030.2009/BIBTEX
225. SHU, W., CHO, J. Y., JIANG, Y., ZHANG, M., WEISZ, D., ELDER, G. A.,
SCHMEIDLER, J., DE GASPERI, R., GAMA SOSA, M. A., RABIDOU, D., SANTUCCI, A. C.,
PERL, D., MORRISEY, E., & BUXBAUM, J. D. (2005). ALTERED ULTRASONIC
VOCALIZATION IN MICE WITH A DISRUPTION IN THE FOXP2 GENE. PROCEEDINGS OF THE
NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 102(27),
9643–9648. HTTPS://DOI.ORG/10.1073/PNAS.0503739102
226. SIROTIN, Y. B., COSTA, M. E., & LAPLAGNE, D. A. (2014). RODENT
ULTRASONIC VOCALIZATIONS ARE BOUND TO ACTIVE SNIFFING BEHAVIOR. FRONTIERS IN
BEHAVIORAL NEUROSCIENCE, 8(NOV).
HTTPS://DOI.ORG/10.3389/FNBEH.2014.00399
227. SOHUR, U. S., PADMANABHAN, H. K., KOTCHETKOV, I. S., MENEZES, J. R.
L., & MACKLIS, J. D. (2014). ANATOMIC AND MOLECULAR DEVELOPMENT OF
CORTICOSTRIATAL PROJECTION NEURONS IN MICE. CEREBRAL CORTEX, 24(2), 293–
303. HTTPS://DOI.ORG/10.1093/CERCOR/BHS342
228. SPITERI, E., KONOPKA, G., COPPOLA, G., BOMAR, J., OLDHAM, M., OU, J.,
VERNES, S. C., FISHER, S. E., REN, B., & GESCHWIND, D. H. (2007). IDENTIFICATION
OF THE TRANSCRIPTIONAL TARGETS OF FOXP2, A GENE LINKED TO SPEECH AND
LANGUAGE, IN DEVELOPING HUMAN BRAIN. AMERICAN JOURNAL OF HUMAN GENETICS,
81(6), 1144–1157. HTTPS://DOI.ORG/10.1086/522237
229. SQUAIR, J. W., GAUTIER, M., KATHE, C., ANDERSON, M. A., JAMES, N. D.,
HUTSON, T. H., HUDELLE, R., QAISER, T., MATSON, K. J. E., BARRAUD, Q., LEVINE, A.
J., LA MANNO, G., SKINNIDER, M. A., & COURTINE, G. (2021). CONFRONTING FALSE
DISCOVERIES IN SINGLE-CELL DIFFERENTIAL EXPRESSION. NATURE COMMUNICATIONS
2021 12:1, 12(1), 1–15. HTTPS://DOI.ORG/10.1038/S41467-021-25960-2
230. STÅHL, P. L., SALMÉN, F., VICKOVIC, S., LUNDMARK, A., NAVARRO, J. F.,
MAGNUSSON, J., GIACOMELLO, S., ASP, M., WESTHOLM, J. O., HUSS, M., MOLLBRINK,
203
A., LINNARSSON, S., CODELUPPI, S., BORG, Å., PONTÉN, F., COSTEA, P. I., SAHLÉN,
P., MULDER, J., BERGMANN, O., … FRISÉN, J. (2016). VISUALIZATION AND ANALYSIS
OF GENE EXPRESSION IN TISSUE SECTIONS BY SPATIAL TRANSCRIPTOMICS. SCIENCE,
353(6294), 78–82.
HTTPS://DOI.ORG/10.1126/SCIENCE.AAF2403/SUPPL_FILE/AAF2403_STAHL_S
M.PDF
231. STRINGER, C., WANG, T., MICHAELOS, M., & PACHITARIU, M. (2021).
CELLPOSE: A GENERALIST ALGORITHM FOR CELLULAR SEGMENTATION. NATURE
METHODS, 18(1), 100–106. HTTPS://DOI.ORG/10.1038/S41592-020-01018-X
232. SUÁREZ, R., BLUETT, T., MCCULLOUGH, M. H., AVITAN, L., BLACK, D. A.,
PAOLINO, A., FENLON, L. R., GOODHILL, G. J., & RICHARDS, L. J. (2023). CORTICAL
ACTIVITY EMERGES IN REGION-SPECIFIC PATTERNS DURING EARLY BRAIN DEVELOPMENT.
PROCEEDINGS OF THE NATIONAL ACADEMY OF SCIENCES, 120(22), E2208654120.
HTTPS://DOI.ORG/10.1073/PNAS.2208654120
233. SUGITANI, Y., NAKAI, S., MINOWA, O., NISHI, M., JISHAGE, K. I., KAWANO,
H., MORI, K., OGAWA, M., & NODA, T. (2002). BRN-1 AND BRN-2 SHARE CRUCIAL
ROLES IN THE PRODUCTION AND POSITIONING OF MOUSE NEOCORTICAL NEURONS.
GENES DEV., 16(14), 1760–1765. HTTPS://DOI.ORG/10.1101/GAD.978002
234. SWANSON, L. (2012). BRAIN ARCHITECTURE: UNDERSTANDING THE BASIC
PLAN - LARRY W. SWANSON - GOOGLE BOOKS (2ND ED.). OXFORD UNIVERSITY PRESS.
HTTPS://BOOKS.GOOGLE.COM/BOOKS?HL=EN&LR=&ID=4RALWYWVEROC&OI=FND&PG
=PP2&DQ=BRAIN+ARCHITECTURE:+UNDERSTANDING+THE+BASIC+PLAN&OTS=ZQQ7
ZUZSYT&SIG=KCWE5Z3L2OBMI9IQGAZAZ31JVXY#V=ONEPAGE&Q=BRAIN
ARCHITECTURE%3A UNDERSTANDING THE BASIC PLAN&F=FALSE
235. SWANSON, L. W. (2000). WHAT IS THE BRAIN? TRENDS IN NEUROSCIENCES,
23(11), 519–527. HTTPS://DOI.ORG/10.1016/S0166-2236(00)01639-8
236. SWANSON, L. W., HAHN, J. D., & SPORNS, O. (2022). STRUCTURE–
FUNCTION SUBSYSTEM MODEL AND COMPUTATIONAL LESIONS OF THE CENTRAL
NERVOUS SYSTEM’S ROSTRAL SECTOR (FOREBRAIN AND MIDBRAIN). PROCEEDINGS OF
THE NATIONAL ACADEMY OF SCIENCES OF THE UNITED STATES OF AMERICA, 119(45).
HTTPS://DOI.ORG/10.1073/PNAS.2210931119/-/DCSUPPLEMENTAL
237. TALLAFUSS, A., & EISEN, J. S. (2008). THE MET RECEPTOR TYROSINE
KINASE PREVENTS ZEBRAFISH PRIMARY MOTONEURONS FROM EXPRESSING AN
INCORRECT NEUROTRANSMITTER. NEURAL DEVELOPMENT, 3(1).
HTTPS://DOI.ORG/10.1186/1749-8104-3-18
238. TASIC, B., MENON, V., NGUYEN, T. N., KIM, T. K., JARSKY, T., YAO, Z.,
LEVI, B., GRAY, L. T., SORENSEN, S. A., DOLBEARE, T., BERTAGNOLLI, D., GOLDY, J.,
SHAPOVALOVA, N., PARRY, S., LEE, C., SMITH, K., BERNARD, A., MADISEN, L.,
204
SUNKIN, S. M., … ZENG, H. (2016). ADULT MOUSE CORTICAL CELL TAXONOMY
REVEALED BY SINGLE CELL TRANSCRIPTOMICS. NATURE NEUROSCIENCE, 19(2), 335–
346. HTTPS://DOI.ORG/10.1038/NN.4216
239. TASIC, B., YAO, Z., GRAYBUCK, L. T., SMITH, K. A., NGUYEN, T. N.,
BERTAGNOLLI, D., GOLDY, J., GARREN, E., ECONOMO, M. N., VISWANATHAN, S.,
PENN, O., BAKKEN, T., MENON, V., MILLER, J., FONG, O., HIROKAWA, K. E., LATHIA,
K., RIMORIN, C., TIEU, M., … ZENG, H. (2018A). SHARED AND DISTINCT
TRANSCRIPTOMIC CELL TYPES ACROSS NEOCORTICAL AREAS. NATURE 2018 563:7729,
563(7729), 72–78. HTTPS://DOI.ORG/10.1038/S41586-018-0654-5
240. TASIC, B., YAO, Z., GRAYBUCK, L. T., SMITH, K. A., NGUYEN, T. N.,
BERTAGNOLLI, D., GOLDY, J., GARREN, E., ECONOMO, M. N., VISWANATHAN, S.,
PENN, O., BAKKEN, T., MENON, V., MILLER, J., FONG, O., HIROKAWA, K. E., LATHIA,
K., RIMORIN, C., TIEU, M., … ZENG, H. (2018B). SHARED AND DISTINCT
TRANSCRIPTOMIC CELL TYPES ACROSS NEOCORTICAL AREAS. NATURE 2018 563:7729,
563(7729), 72–78. HTTPS://DOI.ORG/10.1038/S41586-018-0654-5
241. TAU, G. Z., & PETERSON, B. S. (2009). NORMAL DEVELOPMENT OF BRAIN
CIRCUITS. NEUROPSYCHOPHARMACOLOGY 2010 35:1, 35(1), 147–168.
HTTPS://DOI.ORG/10.1038/NPP.2009.115
242. TELLEY, L., AGIRMAN, G., PRADOS, J., AMBERG, N., FIÈVRE, S., OBERST,
P., BARTOLINI, G., VITALI, I., CADILHAC, C., HIPPENMEYER, S., NGUYEN, L., DAYER,
A., & JABAUDON, D. (2019). TEMPORAL PATTERNING OF APICAL PROGENITORS AND
THEIR DAUGHTER NEURONS IN THE DEVELOPING NEOCORTEX. SCIENCE, 364(6440).
HTTPS://DOI.ORG/10.1126/SCIENCE.AAV2522/SUPPL_FILE/AAV2522_TELLEY_
SM.PDF
243. TERRIENTE, J., & PUJADES, C. (2015). CELL SEGREGATION IN THE
VERTEBRATE HINDBRAIN: A MATTER OF BOUNDARIES. IN CELLULAR AND MOLECULAR
LIFE SCIENCES (VOL. 72, ISSUE 19, PP. 3721–3730).
HTTPS://DOI.ORG/10.1007/S00018-015-1953-8
244. THE C. ELEGANS SEQUENCING CONSORTIUM*. (1998). GENOME SEQUENCE
OF THE NEMATODE C. ELEGANS: A PLATFORM FOR INVESTIGATING BIOLOGY. IN SCIENCE
(VOL. 282, ISSUE 5396, PP. 2012–2018). AMERICAN ASSOCIATION FOR THE
ADVANCEMENT OF SCIENCE. HTTPS://DOI.ORG/10.1126/SCIENCE.282.5396.2012
245. TOMÁS-ROCA, L., CORRAL-SAN-MIGUEL, R., AROCA, P., PUELLES, L., &
MARÍN, F. (2016). CRYPTO-RHOMBOMERES OF THE MOUSE MEDULLA OBLONGATA,
DEFINED BY MOLECULAR AND MORPHOLOGICAL FEATURES. BRAIN STRUCTURE AND
FUNCTION, 221(2), 815–838. HTTPS://DOI.ORG/10.1007/S00429-014-0938-Y
246. TÖNGES, L., OSTENDORF, T., LAMBALLE, F., GENESTINE, M., DONO, R.,
KOCH, J.-C., BÄHR, M., MAINA, F., & LINGOR, P. (2011). HEPATOCYTE GROWTH
205
FACTOR PROTECTS RETINAL GANGLION CELLS BY INCREASING NEURONAL SURVIVAL AND
AXONAL REGENERATION IN VITRO AND IN VIVO. JOURNAL OF NEUROCHEMISTRY, 117(5),
892–903. HTTPS://DOI.ORG/10.1111/J.1471-4159.2011.07257.X
247. TOVAR-Y-ROMO, L. B., RAMÍREZ-JARQUÍN, U. N., LAZO-GÓMEZ, R., &
TAPIA, R. (2014). TROPHIC FACTORS AS MODULATORS OF MOTOR NEURON PHYSIOLOGY
AND SURVIVAL: IMPLICATIONS FOR ALS THERAPY. IN FRONTIERS IN CELLULAR
NEUROSCIENCE (VOL. 8, ISSUE FEB). FRONTIERS RESEARCH FOUNDATION.
HTTPS://DOI.ORG/10.3389/FNCEL.2014.00061
248. TSYPORIN, J., TASTAD, D., MA, X., NEHME, A., FINN, T., HUEBNER, L., LIU,
G., GALLARDO, D., MAKHAMREH, A., ROBERTS, J. M., KATZMAN, S., SESTAN, N.,
MCCONNELL, S. K., YANG, Z., QIU, S., & CHEN, B. (2021). TRANSCRIPTIONAL
REPRESSION BY FEZF2 RESTRICTS ALTERNATIVE IDENTITIES OF CORTICAL PROJECTION
NEURONS. CELL REPORTS, 35(12), 109269.
HTTPS://DOI.ORG/10.1016/J.CELREP.2021.109269
249. URBANUS, B. H. A., PETER, S., FISHER, S. E., & DE ZEEUW, C. I. (2020).
REGION-SPECIFIC FOXP2 DELETIONS IN CORTEX, STRIATUM OR CEREBELLUM CANNOT
EXPLAIN VOCALIZATION DEFICITS OBSERVED IN SPONTANEOUS GLOBAL KNOCKOUTS.
SCIENTIFIC REPORTS, 10(1). HTTPS://DOI.ORG/10.1038/S41598-020-78531-8
250. VAN DAELE, D. J., & CASSELL, M. D. (2009). MULTIPLE FOREBRAIN
SYSTEMS CONVERGE ON MOTOR NEURONS INNERVATING THE THYROARYTENOID
MUSCLE. NEUROSCIENCE, 162(2), 501–524.
HTTPS://DOI.ORG/10.1016/J.NEUROSCIENCE.2009.05.005
251. VAN HOORN, A., CARPENTER, T., OAK, K., LAUGHARNE, R., RING, H., &
SHANKAR, R. (2019). NEUROMODULATION OF AUTISM SPECTRUM DISORDERS USING
VAGAL NERVE STIMULATION. IN JOURNAL OF CLINICAL NEUROSCIENCE (VOL. 63, PP. 8–
12). CHURCHILL LIVINGSTONE. HTTPS://DOI.ORG/10.1016/J.JOCN.2019.01.042
252. VAN SEGBROECK, M., KNOLL, A. T., LEVITT, P., & NARAYANAN, S. (2017).
MUPET—MOUSE ULTRASONIC PROFILE EXTRACTION: A SIGNAL PROCESSING TOOL
FOR RAPID AND UNSUPERVISED ANALYSIS OF ULTRASONIC VOCALIZATIONS. NEURON,
94(3), 465-485.E5. HTTPS://DOI.ORG/10.1016/J.NEURON.2017.04.005
253. WANG, F., FLANAGAN, J., SU, N., WANG, L. C., BUI, S., NIELSON, A., WU,
X., VO, H. T., MA, X. J., & LUO, Y. (2012). RNASCOPE: A NOVEL IN SITU RNA
ANALYSIS PLATFORM FOR FORMALIN-FIXED, PARAFFIN-EMBEDDED TISSUES. JOURNAL OF
MOLECULAR DIAGNOSTICS, 14(1), 22–29.
HTTPS://DOI.ORG/10.1016/J.JMOLDX.2011.08.002
254. WANG, Y., EDDISON, M., FLEISHMAN, G., WEIGERT, M., XU, S., WANG, T.,
ROKICKI, K., GOINA, C., HENRY, F. E., LEMIRE, A. L., SCHMIDT, U., YANG, H.,
206
SVOBODA, K., MYERS, E. W., SAALFELD, S., KORFF, W., STERNSON, S. M., &
TILLBERG, P. W. (2021). EASI-FISH FOR THICK TISSUE DEFINES LATERAL
HYPOTHALAMUS SPATIO-MOLECULAR ORGANIZATION. CELL, 184(26), 6361-6377.E24.
HTTPS://DOI.ORG/10.1016/J.CELL.2021.11.024/ATTACHMENT/D0CE8F78-B9C5-
4C98-8D79-3FCF32F06C0F/MMC4.PDF
255. WEIGERT, M., SCHMIDT, U., HAASE, R., SUGAWARA, K., & MYERS, G.
(2020). STAR-CONVEX POLYHEDRA FOR 3D OBJECT DETECTION AND SEGMENTATION IN
MICROSCOPY. PROCEEDINGS - 2020 IEEE WINTER CONFERENCE ON APPLICATIONS OF
COMPUTER VISION, WACV 2020, 3655–3662.
HTTPS://DOI.ORG/10.1109/WACV45572.2020.9093435
256. WETZEL, D. M., KELLEY, D. B., & CAMPBELL, B. A. (1980). CENTRAL
CONTROL OF ULTRASONIC VOCALIZATIONS IN NEONATAL RATS: I. BRAIN STEM MOTOR
NUCLEI. JOURNAL OF COMPARATIVE AND PHYSIOLOGICAL PSYCHOLOGY, 94(4), 596–
605. HTTPS://DOI.ORG/10.1037/H0077699
257. WHITE, F. A., KELLER-PECK, C. R., MICHAEL KNUDSON, C., KORSMEYER, S.
J., & SNIDER, W. D. (1998). WIDESPREAD ELIMINATION OF NATURALLY OCCURRING
NEURONAL DEATH IN BAX- DEFICIENT MICE. JOURNAL OF NEUROSCIENCE, 18(4), 1428–
1439. HTTPS://DOI.ORG/10.1523/JNEUROSCI.18-04-01428.1998
258. WHITMAN, M. C., NGUYEN, E. H., BELL, J. L., TENNEY, A. P., GELBER, A., &
ENGLE, E. C. (2018). LOSS OF CXCR4/CXCL12 SIGNALING CAUSES OCULOMOTOR
NERVE MISROUTING AND DEVELOPMENT OF MOTOR TRIGEMINAL TO OCULOMOTOR
SYNKINESIS. INVESTIGATIVE OPHTHALMOLOGY AND VISUAL SCIENCE, 59(12), 5201–
5209. HTTPS://DOI.ORG/10.1167/IOVS.18-25190
259. WÖHR, M., & SCHWARTING, R. K. W. (2013). AFFECTIVE COMMUNICATION IN
RODENTS: ULTRASONIC VOCALIZATIONS AS A TOOL FOR RESEARCH ON EMOTION AND
MOTIVATION. CELL AND TISSUE RESEARCH, 354(1), 81–97.
HTTPS://DOI.ORG/10.1007/S00441-013-1607-9
260. WONG, V., GLASS, D. J., ARRIAGA, R., YANCOPOULOS, G. D., LINDSAY, R.
M., & CONN, G. (1997). HEPATOCYTE GROWTH FACTOR PROMOTES MOTOR NEURON
SURVIVAL AND SYNERGIZES WITH CILIARY NEUROTROPHIC FACTOR. JOURNAL OF
BIOLOGICAL CHEMISTRY, 272(8), 5187–5191.
HTTPS://DOI.ORG/10.1074/JBC.272.8.5187
261. WOODWORTH, M. B., GREIG, L. C., LIU, K. X., IPPOLITO, G. C., TUCKER, H.
O., MACKLIS, J. D., WOODWORTH, M. B., GREIG, L. C., LIU, K. X., IPPOLITO, G. C., &
TUCKER, H. O. (2016). CTIP1 REGULATES THE BALANCE BETWEEN SPECIFICATION OF
DISTINCT PROJECTION NEURON SUBTYPES IN DEEP ARTICLE CTIP1 REGULATES THE
BALANCE BETWEEN SPECIFICATION OF DISTINCT PROJECTION NEURON SUBTYPES IN
DEEP CORTICAL LAYERS. 999–1012.
HTTPS://DOI.ORG/10.1016/J.CELREP.2016.03.064
207
262. WU, H. H., & LEVITT, P. (2013A). PRENATAL EXPRESSION OF MET
RECEPTOR TYROSINE KINASE IN THE FETAL MOUSE DORSAL RAPHE NUCLEI AND THE
VISCERAL MOTOR/SENSORY BRAINSTEM. DEVELOPMENTAL NEUROSCIENCE, 35(1), 1–
16. HTTPS://DOI.ORG/10.1159/000346367
263. WU, H. H., & LEVITT, P. (2013B). PRENATAL EXPRESSION OF MET
RECEPTOR TYROSINE KINASE IN THE FETAL MOUSE DORSAL RAPHE NUCLEI AND THE
VISCERAL MOTOR/SENSORY BRAINSTEM. DEVELOPMENTAL NEUROSCIENCE, 35(1), 1.
HTTPS://DOI.ORG/10.1159/000346367
264. XIE, Z., EAGLESON, K. L., WU, H. H., & LEVITT, P. (2016). HEPATOCYTE
GROWTH FACTOR MODULATES MET RECEPTOR TYROSINE KINASE AND Β-CATENIN
FUNCTIONAL INTERACTIONS TO ENHANCE SYNAPSE FORMATION. ENEURO, 3(4).
HTTPS://DOI.ORG/10.1523/ENEURO.0074-16.2016
265. XIE, Z., LI, J., BAKER, J., EAGLESON, K. L., COBA, M. P., & LEVITT, P.
(2016). RECEPTOR TYROSINE KINASE MET INTERACTOME AND
NEURODEVELOPMENTAL DISORDER PARTNERS AT THE DEVELOPING SYNAPSE.
BIOLOGICAL PSYCHIATRY, 80(12), 933–942.
HTTPS://DOI.ORG/10.1016/J.BIOPSYCH.2016.02.022
266. YAJIMA, Y., HAYASHI, Y., & YOSHII, N. (1982). AMBIGUUS MOTONEURONS
DISCHARGING CLOSELY ASSOCIATED WITH ULTRASONIC VOCALIZATION IN RATS. BRAIN
RESEARCH, 238(2), 445–450. HTTPS://DOI.ORG/10.1016/0006-8993(82)90121-4
267. YAO, Z., VAN VELTHOVEN, C. T. J., NGUYEN, T. N., GOLDY, J., SEDENO-
CORTES, A. E., BAFTIZADEH, F., BERTAGNOLLI, D., CASPER, T., CHIANG, M.,
CRICHTON, K., DING, S. L., FONG, O., GARREN, E., GLANDON, A., GOUWENS, N. W.,
GRAY, J., GRAYBUCK, L. T., HAWRYLYCZ, M. J., HIRSCHSTEIN, D., … ZENG, H.
(2021). A TAXONOMY OF TRANSCRIPTOMIC CELL TYPES ACROSS THE ISOCORTEX AND
HIPPOCAMPAL FORMATION. CELL, 184(12), 3222.
HTTPS://DOI.ORG/10.1016/J.CELL.2021.04.021
268. ZENG, H., & SANES, J. R. (2017). NEURONAL CELL-TYPE CLASSIFICATION:
CHALLENGES, OPPORTUNITIES AND THE PATH FORWARD. IN NATURE REVIEWS
NEUROSCIENCE (VOL. 18, ISSUE 9, PP. 530–546). NATURE PUBLISHING GROUP.
HTTPS://DOI.ORG/10.1038/NRN.2017.85
269. ZIMMER, C., TIVERON, M. C., BODMER, R., & CREMER, H. (2004).
DYNAMICS OF CUX2 EXPRESSION SUGGESTS THAT AN EARLY POOL OF SVZ
PRECURSORS IS FATED TO BECOME UPPER CORTICAL LAYER NEURONS. CEREBRAL
CORTEX, 14(12), 1408–1420. HTTPS://DOI.ORG/10.1093/CERCOR/BHH102
208
APPENDIX: Spatial Profiling of Neurotrophic Factor Receptors in the Embryonic Nucleus
Ambiguus of Wildtype and Met Conditional Knockout Mice
INTRODUCTION
Heterogenous pools of motor neurons (MNs) in the spinal cord and brainstem for
efferent control of myriad peripheral functions. The supradiaphragmatic motor component
of the vagus nerve originates from pools of brainstem MNs within the nucleus Ambiguus
(nAmb) and controls vocal communication, respiration, cardiac function, and food
ingestion via its innervation of multiple organs systems (Bieger & Hopkins, 1987). The
nAmb is viscerotopically organized, and the structure-function relations of substructures
within the nAmb and its myriad peripheral targets requires circuit assembly during early
development. To form the appropriate circuits, vagal MNs must differentiate into distinct
subsets, defined developmentally by discrete gene expression, morphological, and
innervation patterns. For spinal cord MNs, interactions between neurotrophins and their
cognate neurotrophic factor receptors (NTFRs) have subtype specific roles in survival,
differentiation, and axon pathfinding (Helmbacher et al., 2003; F. Lamballe et al., 2011;
Schaller et al., 2017b). One such NTFR is encoded by the Met gene, which results in
highly pleiotropic receptor tyrosine that has pool-specific developmental roles in spinal
MNs. Met signaling promotes the survival of discrete MN pools, whereas Met signaling in
adjacent pools promotes axon growth and guidance (Ebens et al., 1996; F. Lamballe et
al., 2011). Thus, functional spinal MN projection classes can be defined by their position,
209
innervation targets, and unique responses to trophic factor combinations. It is unknown
whether a similar mechanism exists for the diversification of vagal MNs of the brainstem.
We had previously demonstrated a role for Met on the development of nAmb MNs.
Embryonic deletion of Met from vagal MNs resulted in a loss of 36% of nAmb MNs,
transient reduced body weight, and profound disruption of lifespan vocalizations. The
loss of MNs was primarily confined to the rostral pool of nAmb MNs, which predominantly
innervate the esophagus. Interestingly, we found that only a subset of Met-expressing
(Met+) nAmb MNs are lost with conditional deletion of Met. The loss of only subsets of
Met+ nAmb MNs is similar to findings in spinal MN pools, and consistent with the
hypothesis that combinatorial NTFRs regulate the development of functionally diverse
vagal MN subtypes (A. Kamitakahara et al., 2017; A. K. Kamitakahara et al., 2021).
In this study, we test this hypothesis through experiments designed to identify the
combinatorial NTFR expression patterns of nAmb MNs at a single-cell level. These
preliminary results identify multiple NTFRs that are co-expressed with Met in nAmb MNs.
From these, we identify two NTFRs that may play neurotrophic roles subsets of Met+
nAmb MNs after the deletion of Met. We also identify two other NTFRs that are also
expressed in Met+ MNs but which are likely dispensable to MN survival in the absence of
Met. Our methods also allow for the preservation of tissue topography, providing
important information on which pools of nAmb MNs co-express these putative survival-
promoting NTFRs. Once validated with increased sample sizes, these data can provide
210
important insight into the gene modules involved in vagal MN diversity during
development.
MATERIALS AND METHODS
Animals
Animal care and experimental procedures were performed in accordance with the
Institutional Animal Care and Use Committee of The Saban Research Institute, Children’s
Hospital Los Angeles. Breeding pairs and pregnant damns were housed in the vivarium
on a 13:11 hour light:dark cycle (lights on at 06:00 hours, lights off at 19:00 hours) at 22°C
with ad libitum access to a standard chow diet (PicoLab Rodent Diet 20, #5053, St. Louis,
MO).
Breeding pairs were set up to generate our control (Phox2b
Cre
;Met
wt/wt
) and
conditional knockout (Phox2b
Cre
;Met
fx/fx
) embryos. To simplify the strain names of mice
used, conditional wild type (cWT) will be used to describe any mouse carrying the
Phox2b
cre
allele but lacking the Met
fx
allele, and conditional knockout (cKO) will be used
to describe any mouse carrying the Phox2b
cre
allele and two copies of the Met
fx
allele.
The Phox2b
cre
was obtained from The Jackson Laboratory (B6(Cg)-Tg(Phox2b-
cre)3Jke/J, stock 016223, RRID:IMSR_JAX:016223). We had previously performed
validation studies to assess recombination efficiency in the vagal motor nuclei using
211
Phox2b
cre
;TdTom mice, and demonstrated a nearly 100% recombination via
quantification of the reporter (A. K. Kamitakahara et al., 2021).
The Met
fx
mouse line, in which exon 16 of the Met allele is flanked by loxP sites,
was shared by the laboratory of Dr. Snorri S. Thorgeirsson (National Cancer Institute,
NIH, Bethesda, MD) and is available at the Jackson Laboratory (Stock 016974). The Met
fx
transgenic line has been backcrossed for more than 10 generations and maintained
isogenically on a C57BL/6J background in our laboratory for all experiments described.
In Met cKO mice, Exon 16 of floxed Met alleles are conditionally deleted in all cells that
express Phox2b, a gene expressed in all vagal MNs starting at E9, leading to the
production and rapid degradation of truncated Met mRNA and signaling-deficient MET
protein. The truncated mRNA is still detectible through RNAscope, allowing us to identify
Met+ cells and quantify Met expression levels in the cKO nAmb MNs. Prior histological
and Western blot experiments have validated the efficiency of recombination in the Met
fx/fx
mice (Huh et al., 2004; Judson et al., 2009, 2010b). In addition, the topography of nAmb
MNs loss was consistent across all cKO animals, and the severe USV deficits were 100%
penetrance in postnatal day 7 (P7) mice, precluding a stochastic penetrance of floxed
allele recombination and demonstrating the reliability of the model (A. K. Kamitakahara
et al., 2021).
Tissue Collection and Processing
212
Dams were anesthetized with isoflurane and decapitated. Embryos were collected
and anesthetized on ice-cold PBS. The brains of all embryos were collected and fixed in
4% paraformaldehyde on a shaker overnight at 4C. Brains were cryoprotected using
serial sucrose exchanges (10% -- 2hrs, 20% -- overnight, and 30% -- overnight) and
frozen over dry ice. 20um sagittal cryosections were collected and mounted on permafrost
slides.
Hiplex RNAScope and Imaging
The dual HiPlex-immunohistochemistry experiments and imaging were performed
as previously described in (Ali Marandi Ghoddousi et al., 2022). An anti-Chat probe was
used to identify cholinergic nAmb cells. Eleven other probes were used to capture the
gene expression levels of Met and other NTFR mRNAs. The SCAMPR quantification
pipeline was used to generate cellular ROIs using the HUC/D signal, and the
SCAMPR_AreaFraction_SemiAuto SCAMPR ImageJ script was used to quantify the
amount of signal for each probe was in each individual cell to get a quantitative estimate
of NTFR expression levels. A rolling-ball-radius of 5 was used for both background
removal steps in this script. The results were saved into a gene-expression matrix and
used for downstream analysis.
Quantitative Data Analysis
213
The gene-expression matrix was analyzed as follows. The counts were normalized
prior to calculating mean expression in the following way. The total number of transcripts
were calculated for each sample. This number was divided by the number of cells in each
sample to get a sample-specific correction factor. The expression value of each gene in
each cell was divided by the sample-specific correction factor, multiplied by a scaling
factor (1000), and log-normalized.
The number of Chat+ and Chat+/Met+ cell numbers were quantified and compared
across groups. The percentage of Chat+ and Chat+/Met+ MNs that co-express each of
the other NTFRs was determined for each animal and compared across groups. The
average NTFR were also calculated and compared across groups in Chat+ and
Chat+/Met+ MNS. Gfra4 was removed from the analysis as it exhibited very low
expression levels and high background, making it difficult to separate signal from noise.
Although we are currently undersampled, a student’s T test was used to preliminarily
compare mean cell numbers, % of cells expression each gene, and average gene
expression across groups. cell survival between the different conditions and anatomical
locations to determine the trophic requirements of different MET+ nAmb MN
subpopulations.
PRELIMINARY RESULTS
HiPlex experiments recapitulate cell loss phenotype in E16.5 nAmb MNs
214
To test the robustness of our ability to detect all and Met+ MNs using HiPlex
RNAscope, we compared differences in cell numbers across groups to those from a
previous study in our lab that used mouse driver lines and immunohistochemistry to
identify cell groups (A. K. Kamitakahara et al., 2021). Qualitative comparison of HuC/D
images identified a reduction in the size of the caudal portion of the nAmb at E16.6 with
the conditional deletion of Met (Figure 1A), validating our previous findings at postnatal
day 7 (A. K. Kamitakahara et al., 2021). 69% of all Chat+ MNs expressed at least one
transcript of Met (not shown). Quantification of Chat+ and Chat+/Met+ MNs in the nAmb
revealed a 31.6% loss of all MNs and 46.6% loss of Met+ MNs (Figures 1B and 1C). This
matched well with our previous findings where we demonstrated a 31.7% and 49.2% loss
of MNs and Met+ MNs in the rostral nAmb at P7 (A. K. Kamitakahara et al., 2021). These
results demonstrate that we are able to identify and accurately quantify nAmb MNs in
cWT and cKO mice.
215
Figure 1 – MN loss in E16.5 nAmb after cKO of Met. A) Sagittal HuC/D
immunistochemistry images in representative sample of E16.5 nAmb from cWT and cKO
mice. Larger with brighter HuC/D signal are MNs of the nAmb. B) Bar plot representing
the MN counts in cWT and cKO E16.5 nAmb. C) Bar plot representing Met+ MN counts
in cWT and cKO E16.5 nAmb.
216
Loss of Met signaling leads to changes in NTFR expression patterns and cell composition
in nAmb
Only a subset of the Met+ MNs were lost, suggesting a heterogenous population
with diverse trophic requirements. We hypothesized that trophic support from other
NTFRs allow subsets of Met+ MNs to survive after cKO of Met. To test this hypothesis,
we compared the NTFR expression patterns and cellular composition of cWT and cKO
nAmb MNs. We first investigated the percentage of cells that expressed each NTFR in
both groups and identified two NTFRs with reduced expression percentages in the cKO
mice compared to controls, and two NTFRs that trended in the opposite direction. There
was a 25.9% and 21.4 reduction in the percentage of Met+ MNs expressing Ntrk3 and
Lifr in cKO mice compared to cWT mice, suggesting that these NTFRs do not play
redundant trophic roles in subsets of Met+ MNs that are lost. Insr and Gfra1 showed
trends towards being expressed in higher proportions of Met+ MNs in the cKO mice
(Figure 2). This could be due to a selective loss of MNs that do not express these NTFRs,
setting them up as possible sole or redundant neurotrophins in the subset of nAmb MNs
that are lost in cKO mice.
217
Figure 2 – Changes in Met+ nAmb MNs NTFR expression pattern with cKO of Met. The
percentage of Met+ MNs expressing each NTFR in cWT and cKO mice. Note a reduction
in the percentage of MNs expressing Lifr and Ntrk3 and the trend towards higher
percentages of MNs expressing Gfra1 and Insr in cKO mice.
We next investigated the gene expression levels and expression topographies of
the NTFRs that were expressed in higher percentages of MNs in cKO mice. We focused
on Gfra1 and demonstrated that this NTFR has similar expression levels in Met+ MNs
across both groups. Surprisingly, Gfra1 expression levels trend towards being expressed
at lower levels in cKO Met- MNs (Figure 3A). This suggests a possible non-cell-
autonomous role for Met in the expression of other NTFRs in nAmb MNs. We examined
218
the topography of Gfra1 expression in the nAmb to see if it is expressed in the rostral
compact formation of the nAmb, where the majority of the MN loss takes place in cKO
mice (A. K. Kamitakahara et al., 2021). Gfra1 is expressed in a diffuse manner across the
rostrocaudal extent of the nAmb in cWT and cKO group, including in the rostral compact
and the caudal loose formations, where the majority of Met+ MNs reside (Figure 3B).
Next, we next investigated the co-expression patterns of Met and Gfra1 in nAmb MNs.
We set cutoffs for co-expression after qualitative inspection of a co-expression scatterplot
(red box, Figure 3C), and found that 8.6 % of Met+ MNs co-express Gfra1 in cWT mice,
compared to 18.6% in cKO mice. Inspection of Met and Gfra1 co-expression topography
demonstrated similar expression patterns in cWT and cKO mice, with similar numbers
and locations of MNs that express both genes. Together, these results provide promising
evidence that Gfra1 may play a protective role for small subsets of Met+ nAmb MNs after
the loss of Met signaling. As the percentages of MNs expressing Gfra1 were quite variable
in the cKO mice, further samples are required to determine whether these changes are
statistically significant. In addition, analysis of other promising protective NTFRs like Insr
and their co-expression patterns with Met and Gfra1 will determine whether there is even
further diversity in trophic requirements of Met+ MNs.
219
Figure 3 – Changes in Gfra1 expression patterns in Met+ nAmb MNs after cKO of Met.
A) The mean expression of Gfra1 in Met+ and Met- cells in cWT and cKO mice. B)
Representative images of Gfra1 expression levels across the rostral-caudal extend of the
nAmb in cWT (top) and cKO (bottom) mice. C) Scatter plot representing the co-
220
expression patterns of Met and Gfra1 in cWT and cKO mice. Each dot represents a cell.
Closed and open dots correspond to cWT and cKO cells, respectively. Area highlighted
in red represents cells that co-express both genes are relatively high levels. D)
Representative images demonstrating cells that co-express Gfra1 and Met across the
rostral-caudal extend of the nAmb in cWT (top) and cKO (bottom) mice. Cells in red are
those that fall withing the red box in C.
DISCUSSION
MNs require trophic support to survive, especially during periods of naturally
occurring cell death (NOCD) when MNs are competing for trophic factors at their target
regions. In this study, we employed HiPlex RNAscope and a the SCAMPR analytical
pipeline to determine combinatorial expression patterns of eleven NTFRs with single-cell
resolution to explore the early molecular signals that underlie survival and specialization
of in the nAmb MNs at E16.5 (Ali Marandi Ghoddousi et al., 2022). Prior studies had
shown that nAmb MNs express NTFRs at different developmental timepoints. In this
study, we demonstrate that nAmb MNs co-express multiple NTFRs at an embryonic age
that corresponds with NOCD (Caubit et al., 2010; Friedland et al., 1995). Previous work
from our lab showed that that deletion of Met leads to a loss of subsets of MNs in the
caudal nAmb, starting at E14.5. Our preliminary results confirm these findings and
demonstrate that the cells which are lost and the ones that remain may express distinct
combinations of NTFRs.
221
We show that the percentages of cells expressing Lifr and Ntrk3 decreased in the
cKO mice, suggesting that the MNs that were lost expressed these genes. While the
trophic role of Ntrk3 has not been explored in the nAmb, one previous study has shown
that knockout of Lifr leads to a 50% loss of MN in the nAmb (M. Li et al., 1995). This
suggest that Met and Lifr have trophic roles in distinct populations of NTFRs. Another
NTFR, Il6st, has been shown to lead to a 34% loss of MN in the nAmb (Nakashima et al.,
1999b). The percentage of cell expressing this gene did not change in the cKO mice,
suggesting a possible neuroprotective role for this gene in the absence of Met. Further
analysis and in vitro studies can help clarify the combinatorial roles of these NTFRs in the
nAmb.
We identify two other genes that may play prospective neurotrophic roles in
subsets of Met+ MNs. One of these codes for Insr, which is activated insulin, Igf-I, and
Igf-2. The other codes for Gfra1, which is activated by glial-derived neurotrophic factor
(Gdnf) and neuretrin (Ntn). Both of these receptor have been shown to play neurotrophic
and other roles in generating neuronal diversity, but neither have been studies in the vagal
brainstem and their combinatorial effects are unknown (Blount et al., 2022; Moore et al.,
1996; Ronald W. Oppenheim et al., 2000; Tovar-y-Romo et al., 2014). We show that
Gfra1 expression is not constrained to a specific pool within the nAmb, which contrasts
with the expression of Met in the most rostral and most caudal compartments of the
nucleus. In addition, we demonstrate that Gfra1 mean expression levels are lower in cKO
Met- MNs when compared to those from cWT animals. It is possible that the loss of Met
signaling is leading to a non-cell-autonomous decrease of Gfra1 in surrounding neurons.
222
There is evidence of this in spinal MNs, where Met expression in specific pools of MNs
induces Pea3 transcription factor expression in nearby, Met- MNs (Helmbacher et al.,
2003). It is possible that Met plays a similar regulatory role on NTFR expression in nAmb
MNs.
Taken together, these preliminary results suggest possible combinatorial roles for
Met and other NTFRs in the generation of vagal MN diversity. Once powered, data from
these experiments will inform in vitro cell/tissue culture assays and future combinatorial
in vivo gene knockout experiments to determine the specific roles of NTFR combinations
on nAmb neuronal subset development and function.
Abstract (if available)
Abstract
The brain is a complex organ which processes numerous sensory inputs and drives myriad physiological and behavioral outputs. These functions rely greatly on the expression of pleiotropic (multifunctional) genes during development. The developmental expression of these genes drive the production of heterogenous neuron types within different structures of the nervous system, and play roles in the formation of functional circuits. The work outlined below describes the role of the pleiotropic Met receptor tyrosine kinase on the development of structure-function relations in projection neurons of the cortex and motor neurons of the vagal brainstem. This work demonstrates that Met-expressing cortical projection neurons have subtly distinct gene expression patterns during development, suggesting that they are in slightly different developmental stages then neighboring cells which do not express Met. In the vagal brainstem, my colleagues and I show that Met signaling has a neurotrophic effect in subsets of the nucleus ambiguus motor neurons and effects on the production of vocalizations. We also demonstrate the accuracy and effectiveness of a novel analytical method for spatial transcriptomics.
Linked assets
University of Southern California Dissertations and Theses
Conceptually similar
PDF
Genetics and the environment: evaluating the role of noncoding RNA in autism spectrum disorder
PDF
The role of Ryk and Smek in neurogenesis; Mechanisms of CBP/β-catenin signaling inhibitor and IL-6 mediators in head and neck cancer
PDF
Transcriptomic maturation of developing human cone precursors in fetal and 3D hESC-derived tissues
PDF
Gene-environment interactions in neurodevelopment
PDF
Molecular diversity of neurons in developing and mature brain circuits
PDF
Axon guidance cues in development of the mammalian auditory circuit
PDF
A single cell time course of senescence uncovers discrete cell trajectories and transcriptional heterogeneity
PDF
Elucidating neurodevelopmental consequences of syngap1 mutations and inactivated functional regions in human iPSC-derived neurons
PDF
Functional consequences of network architecture in rat hippocampus: a computational study
PDF
The role of vascular dysfunction in cognitive impairment
PDF
Defining the circuits and mechanisms mediating a pacemaker-controlled behavior in C. elegans
PDF
Dissecting the heterogeneity of mouse hematopoietic stem cells in vivo
PDF
Neural circuits underlying the modulation and impact of defensive behaviors
PDF
Exploiting novel transport properties of adeno-associated virus for circuit mapping and manipulation
PDF
Understanding human nephrogenesis and scaling synthesis of organoids facilitate modeling of kidney development and disease
PDF
Transcriptional regulation of lipid metabolism and exploring the specific roles of AKT isoforms
PDF
Transcription errors induce proteotoxic stress in mammalian cells
PDF
Patch aRNA in vitro amplification (PAIA): single cell RNA-seq to expand the understanding of the developing and developed nervous system
PDF
Robustness and stochasticity in Drosophila development
PDF
The role of inflammation in mediating effects of obesity on Alzheimer's disease
Asset Metadata
Creator
Ali Marandi Ghoddousi, Ramin
(author)
Core Title
The impact of genetic pleiotropy on heterogeneity in the developing forebrain and hindbrain
School
College of Letters, Arts and Sciences
Degree
Doctor of Philosophy
Degree Program
Neuroscience
Degree Conferral Date
2023-08
Publication Date
08/07/2023
Defense Date
07/18/2023
Publisher
University of Southern California. Libraries
(digital)
Tag
cortex,developmental neuroscience,neurodevelopment,neuronal specification,neurotrophic factor receptors,neurotrophins,nucleus ambiguus,OAI-PMH Harvest,pleiotropy,single cell RNA sequencing,smFISH,spatial transcriptimics,trophic signaling,vagal,vagus
Language
English
Contributor
Electronically uploaded by the author
(provenance)
Advisor
Bonnin, Alexandre (
committee chair
), Kanoski, Scott (
committee member
), Watts, Alan (
committee member
), Wood, Ruth (
committee member
), Levitt, Pat (
committee member
)
Creator Email
Alimaran@usc.edu,Raminamg@gmail.com
Permanent Link (DOI)
https://doi.org/10.25549/usctheses-oUC113296396
Unique identifier
UC113296396
Identifier
etd-AliMarandi-12203.pdf (filename)
Legacy Identifier
etd-AliMarandi-12203
Document Type
Dissertation
Rights
Ali Marandi Ghoddousi, Ramin
Internet Media Type
application/pdf
Type
texts
Source
20230807-usctheses-batch-1079
(batch),
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 author, as the original true and official version of the work, but does not grant the reader permission to use the work if the desired use is covered by copyright. It is the author, as rights holder, who must provide use permission if such use is covered by copyright.
Repository Name
University of Southern California Digital Library
Repository Location
USC Digital Library, University of Southern California, University Park Campus, Los Angeles, California 90089, USA
Repository Email
cisadmin@lib.usc.edu
Tags
cortex
developmental neuroscience
neurodevelopment
neuronal specification
neurotrophic factor receptors
neurotrophins
nucleus ambiguus
pleiotropy
single cell RNA sequencing
smFISH
spatial transcriptimics
trophic signaling
vagal
vagus