University of Southern California Dissertations and Theses

Exploiting novel transport properties of adeno-associated virus for circuit mapping and manipulation

(USC Thesis Other)

Exploiting novel transport properties of adeno-associated virus for circuit mapping and manipulation

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content i

Exploiting novel transport properties of adeno-associated virus
for circuit mapping and manipulation

By

Brian Zingg
Thesis Advisor: Dr. Li I. Zhang

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 PHILOSOPHY
(Neuroscience)

August 2019

ii

ACKNOWLEDGMENTS

I would like to thank my mentors, Drs. Li Zhang and Huizhong Tao, for their professional
insight and encouragement throughout my graduate experience. I am very grateful to have had the
opportunity to work in such a passionate and productive lab. I would also like to thank my committee
members, Drs. Hong-Wei Dong and Alexandre Bonnin, for their support and helpful advice, as well as
my labmates for generously sharing their technical expertise and for the delightfully random
conversations when I needed to take my mind off science. I would also like to acknowledge my
funding from NIH and thank the NGP leadership and staff for fostering such a fun and supportive
community. Finally, I would like to thank my family for investing so much in me, and I would
especially like to thank my wife Julie for her love and patience throughout this process.

iii

TABLE OF CONTENTS

Acknowledgments …………………………………………………………………………………………..……ii
Table of Contents ………………………………………………………………………………………………..iii
List of Figures …………………………………………………………………………………………………….v
List of Tables …………………………………………………………………………………………………….vi
Abstract ………………………………………………………………………………………………………….vii

CHAPTER 1: Introduction ………… ……… ……… … ………… ……… ……… … ……. . . ……… ……… …. . 1
1.1 Viruses as essential tools in systems neuroscience ………………………...……………..2
1.2 General characteristics of AAV ……………………………………………...………….…….3
1.2.1 Mechanism of entry into host neurons…………………………….……...…………5
1.2.2 Intracellular transport and gene expression………..……………………………….6
1.2.3 Potential mechanisms for viral release from host neurons…..……………………9
1.3 Limitations of alternative methods for accessing input-defined cell populations ...…….11

CHAPTER 2: Retrograde transport of AAV: Accessing projection-defined claustrum neurons
for brain-wide input-output mapping ... …… ……… … …… … … ……… … ……… ……… … ……… . …. .14
2.1 Introduction …………………………………………..………………………………………..15
2.2 Materials and methods ……………………………………………………………………….16
2.2.1 Animal preparation and stereotaxic surgery …………………...……….…………16
2.2.2 Histology ………………………………………………………………………...…….17
2.2.3 CTB labeling of claustrum projection neurons ……………………………….……18
2.2.4 AAV injections for mapping CLA output …………………….……………………..18
2.2.5 Monosynaptic rabies virus tracing…………………………….……...…………….19
2.2.6 Imaging and quantification …………………………………………………………..19
2.2.7 Statistics …………………………………………………………………………...….20
2.3 Results …………………………………………………………………………………………21
2.3.1 Distribution and co-localization of RSP-projecting CLA neurons ……………….21
2.3.2 Output of projection-defined CLA neurons ………………………………….…….24
2.3.3 Monosynaptic input to RSP-projecting CLA neurons ……………………….……27
2.4 Discussion ……………………………………………………………………………………..35

CHAPTER 3: Anterograde transsynaptic transfer of AAV: Mapping cortico-collicular input-
defined neural pathways for defense behavior ………… ……… ……… … ………… ……… …… … . ...40
3.1 Introduction …………………………………………..………………………………………..41
3.2 Materials and methods ……………………………………………………………………….43
3.2.1 Animal preparation and stereotaxic surgery ………………………………………43
iv

3.2.2 Injection of viruses for anterograde transneuronal labeling ………….………….44
3.2.3 Histology…………………………………………………………………...………….46
3.2.4 Imaging and quantification…………………………………………..………………47
3.2.5 Slice preparation and recording …………………………………………..………..47
3.2.6 In vivo optogenetic preparation and stimulation ………………………………….48
3.2.7 Behavioral testing and quantification ……………………………….……………...49
3.2.8 Statistical Methods …………………………………………………...………………50
3.3 Results ……………………………………………………………………………………...….51
3.3.1 Anterograde transneuronal spread of AAV1-Cre ………………………………...51
3.3.2 Dependence on viral type, serotype, and other factors ………………………….55
3.3.3 Synaptic specificity of anterograde transneuronal spread ………………………57
3.3.4 Mapping outputs of input-defined SC neuron subpopulations ……………….….60
3.3.5 Distinct behavioral functions for input-defined SC neuron subpopulations …....63
3.3.6 Anterograde transsynaptic labeling in a cell-type-specific manner …………..…68
3.4 Discussion ……………………………………………………………………………………..70

CHAPTER 4: Anterograde transsynaptic AAV techniques for probing neural circuitry …… … . ..75
4.1 Introduction …………………………………………..………………………………………..76
4.2 Materials and methods ……………………………………………………………………….77
4.2.1 Animal preparation and stereotaxic surgery ………………………………….......77
4.2.2 Injection of viruses for anterograde transsynaptic labeling ……………..……….78
4.2.3 Tetanus toxin expression ……………………………………………………………81
4.2.4 Virus injections for sparse labeling of neurons ...........…………..……………….82
4.2.5 Histology …………………………………………………………………….………..83
4.2.6 Imaging and quantification ………………………………………………………….83
4.2.7 Slice preparation and recording ……………………………………..……………..84
4.2.8 Statistical methods …………………………………………………….…………….85
4.3 Results …………………………………………………………………………………………87
4.3.1 Testing additional viruses and constructs for anterograde transsynaptic
transport ………………………………………………..………………………..……87
4.3.2 Functional, anatomical, and molecular examination of synaptic spread …….…91
4.3.3 Efficiency of spread across different types of synapses ……………….………..98
4.3.4 Application in corticofugal, thalamic, and spinal pathways …………...………..102
4.3.5 Application in sparse labeling approaches for single neuron
reconstruction .……………………………………………………………………...108
4.4 Discussion and Future Directions …………………………………………...…………….111

References ………… ……… ……. … ………… ……… ……… … ………… ……… ……… … ……… ……..119

v

LIST OF FIGURES

Figure 1.1 Recombinant AAVs and their use in neural circuits ……………………………………………4
Figure 1.2 AAV entry into host neurons ……………………………………………………………………..6
Figure 1.3 AAV transduction steps in a host neuron …………………………………...…………………..7
Figure 1.4 Axonal trafficking of AAV in neurons ……………………………………….…………………...8
Figure 1.5 Possible mechanisms for AAV transneuronal spread …………………….………………….10
Figure 2.1 Co-localization of projection-defined CLA neurons ………………………..…………………23
Figure 2.2 Axonal output of RSP-projecting CLA neurons ……………………………………………….25
Figure 2.3 Comparison of output for different projection-defined CLA neurons ……………………….27
Figure 2.4 Monosynaptic input to RSP-projecting CLA neurons …………………………...……………29
Figure 2.5 Summary of brain-wide input to RSP-projecting CLA neurons ………………………..……30
Figure 2.6 Quantification of presynaptic input to RSP-projecting CLA neurons ……………….………32
Figure 2.7 CLA boundaries and summary of brain-wide connectivity …………………………….…….34
Figure 3.1 Advantages of anterograde transsynaptic mapping of functional circuits ………..………..41
Figure 3.2 Anterograde transneuronal transport of AAV1-Cre ……………………………………..……52
Figure 3.3 Retrograde transport of AAV1 …………………………………………………………….……53
Figure 3.4 Quantification of labeling density in targets downstream of V1 …….……………………...54
Figure 3.5 Serotype specificity of transneuronal transport ……………………………………………....56
Figure 3.6 Synaptic specificity of anterograde viral spread ……………………………………..……….58
Figure 3.7 Pre-synaptic specificity of LED evoked currents recorded in slice preparation ……..…….59
Figure 3.8 Mapping of axonal outputs of input-defined neuronal populations in SC ………….………61
Figure 3.9 A1-recipient SC neurons drive an innate escape behavior ………………………….………64
Figure 3.10 Lack of retrograde ChR2 expression in cortex in animals used for behavior studies ……65
Figure 3.11 V1-recipent SC neurons drive freezing behavior ……………………………………….……66
Figure 3.12 Efficiency of anterograde transneuronal labeling within the V1-SC-LP pathway …..…….67
Figure 3.13 Cell-type specific anterograde transneuronal labeling ………………………………...........69
Figure 4.1 Comparison of anterograde transsynaptic spread for different viruses ……………………88
Figure 4.2 WPRE enhancer may contribute to Cre-induced toxicity at the injection site ……………..90
Figure 4.3 Verification of functional synaptic connectivity ……………………………………..…………92
vi

Figure 4.4 Anatomical evidence for the synaptic specificity of viral spread ……………………………93
Figure 4.5 Tetanus toxin inhibition of viral spread …………………………………………………..…….97
Figure 4.6 Efficiency of viral spread across different types of synapses …………………………….…99
Figure 4.7 Application in ascending thalamic pathways ……………………………………………...…103
Figure 4.8 Application in descending pathways to the spinal cord ……………………………….……105
Figure 4.9 Accessing topographically precise, input-defined cell populations through corticofugal
pathways ………………………………………………………………………………...………107
Figure 4.10 Application with sparse labeling approaches for reconstructing single neuron
morphology ………………………………………………………………………………...……109

LIST OF TABLES

Table 2.1 List of anatomical abbreviations …………………..…………………………….………………21
Table 4.1 List of viruses used in this study ……………………………………………….………………..86

vii

ABSTRACT

AAV is one of the most versatile tools in systems neuroscience. Relative to other neurotropic viruses,
it is unique in its ability to drive long-term, robust transgene expression without becoming toxic to host
cells. Combined with its broad tropism, high safety profile, and ease of manufacture, this has led to its
widespread use in both anatomical and functional investigations of neural circuitry. It is most
commonly applied for the local transduction of targeted cell populations, however AAV also displays
distal transport properties, including retrograde and anterograde transsynaptic transport, that are less
well characterized but show great potential for application.
In the first part of this study, we take advantage of a recently developed retrograde serotype of
AAV and use it to gain selective access to projection-defined claustrum neurons in the mouse brain.
This structure likely plays an important role in the modulation of higher-order cortical regions.
However, due to the difficult nature of targeting the claustrum, its precise function and anatomical
organization has remained enigmatic. To overcome this, we used injections of AAVretro-Cre to tag
claustrum neurons based on their projection to the retrosplenial cortex. We then used secondary
injections of a Cre-dependent AAV along with monosynaptic rabies virus tracing to map the brain-wide
input and output of this structure. Our results suggest the claustrum may be driven by emotionally
salient stimuli and may modulate the attentional state of the animal during rewarding or aversive
contexts. Using the same approach for accessing these cells, future studies may seek to test this
hypothesis by directly recording from optogenetically identified claustrum neurons in behaving
animals.
In the remaining portion of this study, we characterize a novel anterograde transsynaptic
transport property of AAV1. We find that AAV1-Cre from transduced presynaptic neurons can
effectively and specifically drive Cre-dependent transgene expression in selected postsynaptic
neuronal targets, thus allowing axonal tracing and functional manipulation of these input-defined cell
populations. This is especially important given that currently available tools for this purpose, such as
viii

herpesvirus and vesicular stomatitis virus, suffer from toxicity issues that limit their use in functional
studies. We then apply this tool in corticofugal pathways that converge on distinct populations of
neurons in the superior colliculus and reveal their ability to drive either freezing or flight, two different
forms of innate defense behavior. Finally, given its promise for application in other brain regions, we
provide a broader demonstration of its use in corticofugal, thalamic, and spinal pathways, and offer
additional lines of evidence supporting a synaptic, rather than extrasynaptic, mechanism of spread. In
particular, we find a strong correspondence between pre-synaptic connectivity and post-synaptic
labeling in slice recording studies, and find that co-expression of tetanus toxin light chain, an inhibitor
of pre-synaptic vesicle fusion, nearly abolishes transsynaptic spread of AAV. Overall, our results
suggest that AAV-mediated anterograde transsynaptic tagging shows great potential as a tool for the
forward screening of neural circuits.

1

CHAPTER 1

INTRODUCTION

2

1.1 Viruses as essential tools in systems neuroscience
A major goal in neuroscience is to gain experimental access to specific cell-types within a circuit.
From there, individual circuit components may be targeted for recording and manipulation to establish
their causal role in a given behavior or cognitive task. While many features must be included in the
definition of a cell-type (e.g. position in a circuit, morphology, intrinsic functional properties), arriving at
a genetic definition for cells with shared properties is a critical step in rendering them amenable to
further investigation. With a genetic definition, useful transgenes such as fluorescent markers or
optogenetic tools may then be introduced downstream of cell-type specific promoters in genetically
engineered mice. A major limitation with this approach, however, is that transgene expression may
be simultaneously driven throughout the entire genetically-defined population, rather than in spatially
restricted brain regions of interest. To overcome this, transgenic mice that express Cre recombinase
in specific cell-types may be used in combination with targeted viral injections that can locally drive
Cre-dependent expression of fluorophores and functional tools. This approach forms the basis for
most modern studies of circuit function and critically depends on the use of viruses for flexible
targeting of cell-types within spatially restricted nuclei.
In addition to probing function, another major goal is to establish chains of synaptic
connectivity between defined cell-types, which provides a framework for understanding information
flow in the brain. To achieve this, several methods may be used, including electron microscopy and
channelrhodopsin-assisted circuit mapping (Petreanu et al., 2009), however these approaches are
time consuming and offer only a low-throughput analysis of specific connection pathways.
Alternatively, viruses that naturally infect neurons and spread transsynaptically offer a convenient
means for characterizing brain-wide pre- or post-synaptic relationships between identified cell-types.
Among these, rabies virus has proven to be particularly useful for revealing all neurons upstream of a
given starter population, however, analogous tools for anterograde transsynaptic tracing remain under
development.
3

Given their natural ability to drive local transgene expression and spread predictably through
the nervous system, a wide variety of neurotropic viruses have been adopted for use in the study of
neural circuitry (Nassi et al., 2015). These include herpesvirus (HSV129), vesicular stomatitis virus
(VSV), pseudorabies virus (PRV), and rabies virus (RV) for the investigation of transsynaptic
connectivity; canine adenovirus (CAV) for the retrograde infection of projection neurons; and lentivirus
(LV), adenovirus (Ad5), and adeno-associated virus (AAV) for the local transduction of cell bodies.
Among these, AAV is the most versatile and widely used, and is prized for its lack of toxicity, robust
gene expression, and potential for retrograde and anterograde transsynaptic applications.

1.2 General characteristics of AAV
AAV is classified as a Dependovirus within the broader category of Parvoviridae viruses. Viruses of
this type are naturally replication-incompetent and therefore rely on gene products supplied by an
additional co-infecting virus, such as adenovirus or herpes simplex virus, to complete their life-cycle.
This feature of replication deficiency, combined with the ability to produce high titer stocks with
relative ease, makes AAV an especially attractive tool for achieving robust, stable transgene
expression in neurons without resulting in toxicity and cell death. The genome of wild-type AAV is
comprised of single-stranded DNA and contains two open reading frames (ORFs) that enable the
expression of proteins for replication functions (Rep78, Rep68, Rep52, Rep40) and viral capsid coat
formation (VP1, VP2, VP3) (Figure 1.1A). These genes are flanked by 145 bp inverted terminal
repeat (ITR) regions that are essential for proper packaging of the genome into the viral capsid and
for stabilizing and expressing genetic material in the host cell (Berry & Asokan, 2016). The viral
capsid proteins assemble into a coat structure that is approximately 20 nm in diameter, with VP2 and
VP3 proteins primarily forming the core, and VP1 contributing to the variable outer loops that form the
basis of its interactions with host cell membrane receptors (Figure 1.1B). At least eleven different
naturally occurring serotypes of AAV have been identified and each are characterized by their distinct
4

Figure 1.1. Recombinant AAVs and their use in neural circuits.
(A) Wild-type AAV genome depicting Rep and Cap genes flanked by ITR regions. Rep codes for 4 different
proteins (purple) involved in viral replication, and Cap codes for three proteins that comprise the capsid coat
(blue). (B) Capsid proteins assembled into a complete viral particle. (C) Recombinant AAV genome with all wild-
type genes replaced with a promoter and gene of interest (e.g. GFP). (D) Different AAV serotypes display
different uptake and transport properties, enabling retrograde, anterograde, and anterograde transsynaptic
access to neurons. Serotypes displaying each transport property are listed. AAVretro is used in Chapter 2,
AAV1 is used in Chapters 3 and 4. Adapted from Bedbrook et al., 2018; Sun & Schaffer et al., 2018.

differences in capsid protein amino acid sequence. These differences in turn confer variability in
tissue tropism among the serotypes, with AAV1, 5, 6, 8, and 9 exhibiting the greatest capacity for
transducing neurons (Aschauer et al., 2013). In addition, several engineered serotypes have been
created for specific applications in neuroscience, including AAVretro, which is preferentially taken up
by axons and retrogradely transported to the cell body (Tervo et al., 2016), and AAV-PHP.B, which is
capable of efficiently crossing the blood-brain barrier and systemically transducing neurons following
peripheral injection into the vasculature (Deverman et al., 2016). Recombinant AAVs for use in neural
circuit investigation may be generated by swapping out all genetic material between the ITRs and
replacing it with a construct capable of driving robust expression of a desired transgene, such as Cre
recombinase, a fluorescent reporter, or any number of tools for recording or manipulating neural
function (Figure 1.1C). The only limitation here is that the full construct must be no more than ~4.7 kb
5

in length in order to successfully package it into the capsid coat. Depending on the desired transport
properties in the nervous system, the viral construct may be packaged into different capsid serotypes
(Figure 1.1D). While most serotypes show robust transduction of local cell bodies at an injection site,
AAVretro exhibits superior retrograde transduction from axon terminals (applied in Chapter 2) and
AAV1 is capable of anterograde transsynaptic spread to second order neurons (the subject of
Chapters 3 & 4).

1.2.1 Mechanism of entry into host neurons
Following a localized injection of high titer AAV into the brain (typically 50 nL at a concentration of 10
13

viral particles per mL), the virus diffuses through the extracellular space and interacts with primary
attachment factors expressed on the surface of cell membranes. These factors are typically
proteoglycans and different serotypes have been shown to have preferential affinity for certain classes
of these receptors. For example, AAV1, AAV5, and AAV6 bind to sialic acid, AAV2 binds to heparin
sulfate, AAV8 binds to laminin, and AAV9 binds to galactose (Pillay & Carette, 2017). Primary
attachment to these factors then facilitates viral interaction with secondary receptors that initiate
endocytosis. The exact identities of these receptors are not well understood, however various
integrins have been implicated, as well as a novel receptor encoded by the gene KIAA0319L, termed
AAVR (Pillay et al., 2016). Knockout of this receptor was reported to abolish host cell transduction for
all AAV serotypes tested in cell culture studies. However, examination of uptake efficiency in vivo
using an AAVR-/- mouse revealed diminished, but not complete elimination of transduction following
systemic injection (Dudek et al., 2018). It has yet to be demonstrated if direct injections of AAV into
the brain lead to transduction in these mice. Following interaction with various glycans and
transmembrane co-receptors, AAV gains entry to the host cell through several mechanisms, including
macropinocytosis, clathrin-mediated endocytosis, and the clathrin-independent CLIC/GEEC pathway
(Figure 1.2). Not all of these pathways lead to transduction, however, and it is estimated that only
6

about 30% of endocytosed AAV particles are successfully delivered to the nucleus (Berry & Asokan,
2016). In general, the differential expression of extracellular factors within the soma, dendrites, and
axons of different cell-types in the brain underlies the variable tropism, transport properties, and
transduction efficiencies exhibited by different serotypes of AAV. Future work is needed to clarify the
essential set of receptors necessary for AAV entry, which will facilitate the creation of novel capsids
and receptor knockout mice that confer additional control over vector spread in circuit mapping
studies.
Figure 1.2. AAV entry into host neurons.
AAV first binds to cell surface glycan receptors, which facilitates interaction with secondary transmembrane
proteins that initiate endocytosis. Entry may occur through several mechanisms, including macropinocytosis,
clathrin-dependent endocytosis, or clathrin-independent (CLIC/GEEC) endocytosis. Adapted from Berry &
Asokan, 2016.

1.2.2 Intracellular transport and gene expression
Following endocytosis, successful transduction requires delivery of the AAV particle to the cell
nucleus (summarized in Figure 1.3). This path involves trafficking endocytosed AAV through
endosomal compartments that become increasingly acidified. The drop in pH triggers an irreversible
conformational change in the AAV capsid proteins, which is essential for viral escape from the
endosome and accumulation in the trans-Golgi network and perinuclear area. The conformationally
changed capsid then enters the nucleus via interactions with β-importin and sheds its coat proteins to
7

release its genetic material. The single-stranded genome then undergoes second strand synthesis
and the ITR regions interact to form a stable episome that drives long-term gene expression in the
host cell.

Figure 1.3. AAV transduction steps in a host neuron.
Following (1) attachment and (2) endocytosis, AAV is trafficked through the (3) early endosome, which matures
into a (4) late endosome/recycling endosome with reduced internal pH that results in a conformational change in
the AAV capsid proteins, leading to endosomal escape and transport the Golgi network, followed by (5) nuclear
pore entry and shedding of capsid proteins leading to (6) genome release and (7) second strand synthesis,
followed by (8) genome stabilization as episomal DNA or ITR-dependent multi-viral genome concatemer and (9)
transgene expression using host cell machinery. Example virally expressed effector proteins listed for mapping,
manipulating, or recording activity of host neuron. Adapted from Isayeva et al., 2005.

As mentioned, given the variability in entry mechanisms at the cell membrane, most
endocytosed AAV is directed through unproductive paths that do not lead to transduction. Some of
these lead to viral degradation in lysosomes or local release back into the extracellular space through
fusion of recycling endosomes with the plasma membrane. In addition, an estimated 14% of all AAV-
8

containing endosomes are rapidly trafficked down the axon and delivered to the synaptic terminal
(Castle et al., 2014a). This process was shown to depend on kinesin-2 and was observed to occur at
an average rate of 2 μm/sec, fast enough to deliver AAV from the cortex to the spinal cord in about
two hours in mice. In addition, AAV1, AAV8, and AAV9 were all shown to exhibit a similar capacity for
anterograde axonal transport and they appeared to share the same endosomal compartments in this
trafficking step, despite potential differences in uptake mechanisms at the host membrane. Lastly, the
amount of AAV delivered down the axon appeared to scale with uptake efficiency at the soma.
Among the serotypes tested, AAV1 displayed the greatest frequency of axonal transport due to its
ability to enter and reach greater concentrations within the host cell, relative to AAV8 and AAV9.
Whether or not AAV was capable of being released from the synaptic terminal in sufficient quantities
to transduce downstream neurons remained an important question to resolve. This is addressed in
Figure 1.4. Axonal trafficking of AAV in neurons.
Following endocytosis at the cell body (top), AAV is trafficked broadly throughout the endosomal system
(colored circles). Transport may be directed to the Golgi network and nucleus, leading to successful
transduction and viral gene expression. Alternatively, some AAV-containing endosomes may be rapidly
trafficked down the axon in a kinesin-2 dependent manner, possibly leading to synaptic release of viral particles.
Anterograde axonal trafficking may pause or reverse direction at any point, returning AAV to the host cell body
in a dynein/dynactin dependent fashion. AAV taken up at axon terminals (bottom) is similarly trafficked to the
cell body using dynein motors, but does not reverse direction in the axon. Adapted from Castle et al., 2014a.
9

Chapters 3 and 4. Lastly, in the same study, it was observed that AAV taken up by the axon terminal
undergoes rapid dynein-dependent transport along the axon toward the cell body. This trafficking
property enables distal transduction at the soma following injections of an AAV capable of efficient
axon terminal uptake, such as AAVretro (applied in Chapter 2).

1.2.3 Potential mechanisms for viral release from host neurons
Given the potential for AAV to be trafficked to synaptic terminals, how might it be released to enable
transduction of second-order neurons? More importantly, to what extent does spread to downstream
neurons require synaptic connectivity as opposed to simple proximity to axon terminals? One
possibility is that AAV-containing vesicles merge with late endosome compartments in the axon
terminal. Late endosomes can bind and take up other molecules from inside the cell through
endocytosis to create small, molecule-filled luminal vesicles, giving them a multivesicular appearance
and the common name, multivesicular bodies (MVBs). These MVBs have been shown to contain
exogenous materials such as neural tracers and viruses, as well as misfolded proteins such as Tau
and α-synuclein in pathological conditions (Von Bartheld & Altick, 2011; Janas et al., 2016). Instead
of binding with lysosomes, some of these MVBs may bind with the plasma membrane to release their
contents as “exosomes”. This process is thought to contribute to the transneuronal propagation of
misfolded proteins in Alzheimer’s (Wang et al., 2017) and other such diseases and may contribute to
the release of accumulated AAV particles in the axon. Fusion with the plasma membrane requires
VAMP7 and may occur at the presynaptic terminal in regions adjacent to the synaptic vesicle active
zone, or fusion may occur extrasynaptically, contributing to the potential spread of AAV to adjacent,
non-connected neurons (Figure 1.5).
On the other hand, AAV-containing vesicles may affiliate with early endosomes that give rise
to synaptic vesicles. Synaptic vesicles are about 40 nm in diameter and may support the co-
packaging of neurotransmitter and AAV cargo (~20 nm diameter). These may then dock pre-
10

synaptically and undergo Ca2+-dependent fusion using VAMP2/synaptobrevin-2, thus releasing AAV
into the synaptic cleft (Figure 1.5). Both of these mechanisms, in addition to potentially unknown
pathways, may contribute to synaptic release of AAV. Given the requirement of VAMP2 in synaptic
vesicle fusion and the ability of tetanus toxin (TeNT) to selectively cleave this protein, it is possible to
test the relative contribution of each of these mechanisms following co-expression of TeNT in host
neurons transduced with AAV1. We perform this experiment in Chapter 4.3.2.

Figure 1.5. Possible mechanisms for AAV transneuronal spread.
Following active trafficking of AAV-containing vesicles to the synaptic terminal, AAV may merge with
multivesicular bodies (MVBs, pink arrows), which in turn can fuse with the plasma membrane to release their
contents. Fusion may occur extrasynaptically (top right) or with the presynaptic membrane (bottom left).
Alternatively, AAV-containing vesicles may affiliate with endosomal compartments (blue arrows) that give rise to
synaptic vesicles (SV). AAV-containing SVs may then dock and fuse with the presynaptic membrane, co-
releasing neurotransmitter and viral particles into the synaptic cleft. For a review of MVBs in neurons, see Von
Bartheld & Altick, 2011 and Janas et al., 2016.

11

1.3 Limitations of alternative methods for accessing input-defined cell populations
A common approach for mapping connections between brain regions involves locally injecting a
fluorescent protein-expressing AAV and then visualizing the axonal output from that region to all
downstream targets. The overlap of axon terminals and cell bodies allows one to infer connectivity
between structures, however exactly which cell-types in a target region are synaptically innervated is
difficult to determine using this method. To overcome this, previous efforts have sought to develop
viral or molecular tools that are capable of spreading one synaptic step to all downstream neurons
following local expression in an upstream starter cell population. These approaches and their
limitations are briefly described.
Wheat germ agglutinin (WGA): WGA is a non-toxic plant lectin that was originally used as a
retrograde and anterograde tracer following direct injections into the brain (Schwab et al., 1978).
Following observations of potential anterograde transsynaptic spread, several transgenic mouse lines
were created that allowed promoter-specific or Cre-dependent expression of WGA from a starter cell
population (Yoshihara et al., 1999; Braz et al., 2002). Experiments using these mice demonstrated
that WGA was capable of anterograde transneuronal spread to first-order neurons downstream of a
particular cell-type but spread minimally to second-order neurons due to dilution of the tracer along
the circuit path. In addition, WGA was also observed to spread in a retrograde direction to
presynaptic neurons, albeit with less efficiency. To expand the utility of WGA, later studies used an
AAV construct capable of expressing WGA tagged with Cre recombinase to facilitate conditional
access to downstream (Gradinaru et al., 2010) or upstream (Xu & Sudhof, 2013) neurons following
viral injection. Further application of WGA for transneuronal mapping has not been widely adopted,
however, perhaps due to concerns of bidirectional transport, efficiency, and synaptic specificity of
release, though it was observed to localize in presynaptic vesicles using EM (along with many other
endosomal compartments) (Broadwell & Balin, 1985).
12

Herpesvirus (HSV129): Herpesvirus was initially used for the forward mapping of multi-synaptic
connectivity from peripheral tissues to the brain (Ugolini et al., 1989). HSV129 has a complex life
cycle and a large genome (~150 kb) such that genetic manipulations may often lead to unpredictable
impairments in viral replication efficiency. Genetically targetable polysynaptic (Lo & Anderson, 2011)
and monosynaptic (Zeng et al., 2017) versions of the virus have been characterized. A major
limitation for both, however, is that they lead to rapid toxicity and cell death in infected neurons which
prevents their use in functional studies. In addition, they display some capacity for infection of axon
terminals, leading to retrograde transduction of neurons projecting to the injection site. Moreover, the
mechanism of spread to post-synaptic neurons has not been well characterized and may not be
purely transsynaptic (Curanovic & Linquist, 2009; Kratchmarov et al., 2012).
Vesicular stomatitis virus (VSV): In contrast to herpesvirus, VSV has a simple genome consisting
of only five genes (similar to rabies virus), making genetic manipulations more straightforward. In
addition, it may be pseudotyped in a similar fashion as rabies to enable transduction of a genetically-
specified starter cell population. When trans-complemented with the lymphocytic choriomeningitis
virus glycoprotein (LCMV-G), VSV was able to spread one synaptic step to neurons downstream of
the starter population (Beier et al., 2011). Unfortunately, however, this virus appears to be severely
toxic to host cells and results in cell lysis after just 72 hours. This again limits its use in functional
work and confounds the synaptic specificity of viral spread due to its potential extrasynaptic release
from lysed axons.
TRACT and trans-TANGO: Given the consistent concerns for toxicity and bidirectional transport
associated with viral approaches for transsynaptic mapping, two promising new methods have been
reported that rely on molecular signaling, rather than viral vectors. These approaches are termed
TRACT (Huang et al., 2017) and trans-TANGO (Talay et al., 2017) and both rely on an engineered
receptor-ligand interaction that spans the synaptic cleft. The design involves expression of a
genetically encoded ligand that is targeted to the presynaptic membrane in a specified starter cell
13

population. This ligand then interacts with a receptor expressed on post-synaptic neurons, triggering
the release of a transcriptional activator that can unlock an endogenous reporter gene in the
downstream cells. Currently, this system has only been tested in drosophila, though efforts are
underway to demonstrate its use in mice. One initial concern related to this approach, however, is the
possibility that receptor-ligand interactions between closely apposed non-synaptic cell membranes
may trigger reporter gene expression in unconnected neurons (Huang et al., 2017). This may lead to
noisy expression that confounds interpretation of labeling associated with synaptic connectivity.
Given the current limitations associated with each of these methods, the development of an
efficient, non-toxic tool for revealing all monosynaptically connected cells downstream of a starter
population remains in high demand. While future optimization is needed, our work in Chapters 3 & 4
reveals that AAV1 exhibits many of these desired features and may be used in its current form to
access spatially restricted, input-defined cell populations for anatomical and functional interrogation.
Combined with the application of retrogradely transported AAV in Chapter 2, the work in this thesis
takes advantage of two novel transport properties of AAV to study specific circuits that would
otherwise be impossible to target with alternative tools and methods.

14

CHAPTER 2

Retrograde transport of AAV: Accessing projection-defined claustrum neurons
for brain-wide input-output mapping

Adapted from:

Zingg, B., Dong, H.-W., Tao, H. W., & Zhang, L. I. (2018). Input-output organization of the mouse
claustrum. Journal of Comparative Neurology, (June), 1–16. http://doi.org/10.1002/cne.24502

15

2.1 Introduction
The claustrum shares widespread connections with the cortex in a variety of mammalian species
(Olson & Graybiel, 1980; LeVay & Sherk, 1981; Dinopoulos et al., 1992; Sadowski et al., 1997;
Tanne-Gariepy et al., 2002; Smith & Alloway, 2014; Torgerson et al., 2014; Zingg et al., 2014; Reser
et al., 2016; Wang et al., 2016; White et al., 2016; Atlan et al., 2016), and has been postulated to play
an important role in processes ranging from attention (Goll et al., 2015; White et al., 2018) and
salience detection (Remedios et al., 2010; Remedios et al., 2014; Kim et al., 2016), to the
coordination of rapid eye movement (REM) sleep (Renouard et al., 2015; Billwiller et al., 2017; for
review see Crick & Koch, 2005; Mathur, 2014; Brown et al., 2017). Progress in determining its
precise function and anatomical organization, however, has been hindered by the difficulty in targeting
claustrum neurons for experimental manipulation due to its small, irregular shape. In particular,
functional studies recording from identified claustrum neurons in vivo remain scarce, and anatomical
studies directly targeting the claustrum have led to conflicting results due to the challenge of confining
neural tracer injections within the boundaries of this structure.
Recent developments in transgenic and viral tracing approaches in rodents, however, provide
a path forward by facilitating experimental access to genetically-defined claustrum neurons. Toward
this end, previous studies have sought to characterize claustrum-specific genetic markers, such as
Gng2, Gnb4, and Ntng2, and have led to the generation of several transgenic Cre-driver mouse lines
that enable conditional access to claustrum cells (Mathur et al., 2009; Watakabe et al., 2014; Wang et
al., 2016). These lines have been shown to provide relatively selective access to claustrum neurons,
however sparse expression in neighboring cortical and endopiriform regions again requires careful
targeting of viral injections and may result in non-specific labeling issues (Wang et al., 2016).
Alternatively, claustrum neurons may be defined based on their projection to a particular
cortical region using injections of a retrogradely transported Cre-expressing virus. This approach has
been reported using AAV6 (Kitanishi & Matsuo, 2017), however care must be taken in choosing a
16

cortical region that receives input only from claustrum, but not overlying insular cortex, to avoid non-
specific labeling. In this study, we found that the posterior retrosplenial cortex (RSP) was ideally
suited for this purpose as it receives dense input specifically from the claustrum, but not from
surrounding structures, allowing for the precise targeting of claustrum cells following secondary
injections of a Cre-dependent AAV vector.
Using this approach, we addressed several unresolved questions regarding the anatomical
organization of the claustrum. Specifically, we aimed to (1) determine whether or not the claustrum
projects subcortically, (2) examine the extent to which claustrum neurons collateralize broadly or form
segregated, pathway-specific projections, and (3) directly reveal the presynaptic inputs to the
claustrum, including those from previously unreported subcortical regions, using monosynaptic rabies
virus mapping. Together, our results provide a more complete understanding of the claustrum
network in the mouse brain, and provide an alternative approach for precisely targeting claustrum
neurons that may be adapted for use in future functional studies.

2.2 Materials and Methods
2.2.1 Animal preparation and stereotaxic surgery
All experimental procedures used in this study were approved by the Animal Care and Use
Committee at the University of Southern California. Male and female C57BL/6J and Ai14 Cre-
dependent tdTomato reporter mice (Jackson Laboratories), and GAD67-GFP mice (from Dr. Yuchio
Yanagawa, Brain Science Institute, RIKEN, Japan) aged 2-6 months were used in this study. Mice
were group housed in a light controlled (12 hr light: 12 hr dark cycle) environment with ad libitum
access to food and water.
Stereotaxic injections of viruses and neural tracers were carried out as previously described
(Xiong et al., 2015). Mice were anesthetized initially in an induction chamber containing 5%
isoflurane mixed with oxygen and then transferred to a stereotaxic frame equipped with a heating
17

pad. Anesthesia was maintained throughout the procedure using continuous delivery of 2%
isoflurane through a nose cone at a rate of 1.5 liters/min. The scalp was shaved and a small incision
was made along the midline to expose the skull. After leveling the head relative to the stereotaxic
frame, injection coordinates based on the Allen Reference Atlas (Dong, 2007) were used to mark the
location on the skull directly above the target area and a small hole (0.5mm diameter) was drilled.
The following coordinates were used to target RSP (3.9 mm posterior and 1.0 mm lateral to bregma
and 0.4 mm ventral from the cortical surface), ACA (0.5 mm anterior and 0.4 mm lateral to bregma
and 0.9 mm ventral from the cortical surface), and V1 (4.0 mm posterior and 2.6 mm lateral to
bregma and 0.4 mm ventral from the cortical surface). To target the CLA, a hole was drilled 0.1 to 0.7
mm anterior and 3.7 mm lateral to bregma, and the head was then rotated laterally 10˚ before
lowering the pipette through the center of the hole to a depth of 2.4 mm below the cortical surface.
Mice were injected in either the right or the left hemisphere, which resulted in identical, yet mirrored,
patterns of labeling for both CLA input and output. Viruses or neural tracers were delivered through
pulled glass micropipettes (inner diameter of tip: ~20 µm) using pressure injection via a micropump
(World Precision Instruments). Total injection volume was 60 to 80 nl, at 15 nl/min. After withdrawing
the micropipette, the scalp was sutured closed and animals were administered ketofen (5mg/kg) to
minimize inflammation and discomfort. Animals were recovered from anesthesia on a heating pad
and then returned to their home cage.
2.2.2 Histology
Following sufficient post-injection survival time, animals were deeply anesthetized and transcardially
perfused with 4% paraformaldehyde. Brains were extracted and post-fixed for 24 hours at 4˚C in 4%
paraformaldehyde and then sliced into 150 µm thick coronal sections using a vibratome (Leica,
VT1000s). The sections were serially mounted onto glass slides and coverslipped. For some
experiments, a fluorescent Nissl stain was added (Neurotrace 640, ThermoFisher, Cat# N21483,
RRID:AB_2572212) to reveal cell body location and cytoarchitectural information. To label cholinergic
18

neurons in the basal forebrain, brain sections were blocked with 10% donkey serum in PBS
containing 0.1% Triton-X100 (PBST) for 1 hour, followed by overnight incubation at 4˚C in PBST with
goat anti-ChAT antibody (Millipore Cat# AB144P, RRID:AB_2079751; 1:1000 dilution). Sections were
then washed with PBS 3 times and incubated in PBST with donkey anti-goat Alexa Fluor 647
conjugated secondary antibody (ThermoFisher Cat# A-21447, RRID:AB_141844; 1:200 dilution) for 2
hours at room temperature.
2.2.3 CTB labeling of claustrum projection neurons
To retrogradely label RSP-, V1-, and ACA-projecting CLA neurons, each region was injected in the
same animal unilaterally with 80 nl of fluorescently conjugated Cholera toxin subunit B (CTB 488, 555,
or 647, respectively; 0.5%; ThermoFisher). To compare the distribution of primary motor cortex
(MOp; coordinates: 0.5 mm anterior and 1.5 mm lateral to bregma, and 0.5 mm ventral from the
cortical surface)- and RSP-projecting CLA neurons, CTB 488 or CTB 555 was injected unilaterally
(120 nl injection volume) into MOp and RSP, respectively. Mice were euthanized 7 days following
injection to allow time for tracer transport and the claustrum was examined for the presence of CTB+
cell bodies.
2.2.4 AAV injections for mapping CLA output
To label the axonal output of projection-defined CLA neurons, 80 nl injections of AAVretro-hSyn-Cre-
WPRE-hGH (see Tervo et al.,, 2017; 1.2 x 10
14
GC/mL, custom ordered from ViGene Biosciences)
were targeted to RSP, V1, or ACA in Ai14 or wild-type C57BL/6J mice to express Cre-recombinase in
CLA neurons that project to each region, respectively. In the same surgical procedure, a second
injection (60 nl) of AAV1-CAG-FLEX-eGFP-WPRE-bGH (UPenn vector core, 1.7 x 10
13
GC/mL) was
targeted to the ipsilateral CLA for V1- and RSP-projecting CLA neurons. For ACA-projecting CLA
neurons, iontophoretic injections of AAV1-CAG-FLEX-eGFP-WPRE-bGH (Stoelting, 5 μA current,
alternating 7 s on/off for 5 mins) were applied to restrict viral spread and minimize the potential
19

labeling of adjacent Cre+ insular cortex neurons that also project to ACA. Animals were euthanized 3
weeks following injection to allow time for viral transport and transgene expression.
2.2.5 Monosynaptic rabies virus tracing
To label the presynaptic inputs to RSP-projecting CLA neurons, wild-type C57BL/6J mice were
injected in RSP with 80 nl of AAVretro-hSyn-Cre-WPRE-hGH and in the same surgical procedure, the
ipsilateral CLA was injected with 60 nl of a 1:1 mixture of AAV1-EF1a-DIO-HTB (Salk vector core, 5.7
x 10
12
GC/mL, virus expresses histone-bound GFP, TVA receptor, and rabies glycoprotein) and
AAV1-CAG-FLEX-RG (UNC vector core, 3.5 x 10
13
GC/mL, virus expresses only rabies glycoprotein),
which was added to boost rabies glycoprotein expression in starter cells and enhance the presynaptic
transfer of rabies virus. Following 3 weeks, a second injection of pseudotyped, glycoprotein-deficient
rabies virus (EnvA-ΔG-mCherry rabies, Salk vector core, 8.6 x 10
7
GC/mL) was injected (80 nl) into
the same location of the CLA. Animals were euthanized 1 week later and brains were extracted and
examined for presynaptic, mCherry+ cell bodies. To examine the identity of presynaptic basal
forebrain neurons, the same injection procedure was performed on GAD67-GFP mice (Tamamaki et
al., 2003) and combined with immunostaining for cholinergic neurons (see Histology).
2.2.6 Imaging and quantification
All images were generated using a confocal microscope (Olympus FluoView FV1000). To quantify
the number of double- and triple-labeled CTB+ CLA neurons projecting to ACA, V1, and RSP, one out
of every two coronal sections was collected across the entire length of the CLA (from about +1.2 to -
1.0 mm relative to bregma, 7 sections total, 150 μm thick) and imaged in a single focal plane at 40X
magnification. Individual CLA neurons were quantified manually and examined in separate
fluorescent channels for the presence of two or more tracers for n = 3 mice total.
To quantify the density of axonal output for RSP-projecting CLA neurons, one out of every two
coronal sections (150 μm thick) were collected across the entire brain and mounted in serial order.
Brain regions containing fluorescently labeled axons and terminal boutons were identified and
20

scanned at 10X magnification at the same laser intensity. Using Photoshop, background was then
subtracted from each image (thresholding to 4X greater than the mean background intensity) and
images were rendered binary. For each brain region, at least five 300 x 300 μm sample areas were
then selected across multiple sections to generate an average pixel density corresponding to axonal
fluorescence. Average values for each region were then expressed as a fraction of the total pixel
density quantified for all regions in a given animal to generate a normalized projection density profile
for n = 3 animals. To compare projection density profiles for ACA-, V1-, and RSP projecting CLA
neurons, the same approach was applied, however only seven brain regions were selected for
analysis and images were collected at higher magnification (40X). Average pixel density values for
each region were normalized by dividing by the total pixel density quantified for the 7 sampled
regions. Normalized projection density profiles were then compared for n = 3 animals in each group.
To determine the number of cells that provide monosynaptic input to RSP-projecting CLA
neurons, one out of every two coronal sections (150 μm thick) was collected across the entire brain
and imaged at 4X magnification. Brain regions were delineated using the Allen Reference Atlas and
mCherry+ cell bodies within each region were manually quantified and expressed as a fraction of the
total cells quantified for a given animal. Normalized input profiles were then compared for n = 4
animals. To quantify the number of cholinergic and GAD67+ basal forebrain neurons that project to
the CLA (Figure 2.6D-F), rabies labeled mCherry+ neurons were examined for co-expression of ChAT
antibody or GAD67-GFP at 10X magnification for n = 3 animals. Co-labeled neurons were expressed
as a percentage of the total number of rabies mCherry+ neurons in the basal forebrain for each
animal.
2.2.7 Statistics
All statistical analyses were performed using GraphPad Prism 6. For comparison of ipsilateral and
contralateral inputs to claustrum, percentages were first quantified from individual brain sections (at
least 3 values for each region) for each of n = 4 animals and were determined to have a normal
21

distribution using the Shapiro-Wilk normality test. Comparison of means was then performed by
paired Student’s t-test. Differences between data sets were considered significant if p < 0.05. To test
for any significant differences in the overall projection pattern for each population of target-defined
claustrum neurons, we performed two-way ANOVA analysis. Results for all histograms are
expressed as mean ± 95% confidence interval. Lower case “n” always denotes the number of
animals used in a set of experiments.
Table 2.1. List of anatomical abbreviations used in this study

2.3 Results
2.3.1 Distribution and co-localization of RSP-projecting CLA neurons
The claustrum (CLA) has previously been shown to project to the retrosplenial cortex (RSP) in rodents
using both retrograde and anterograde tracing approaches (Wang et al., 2016; White et al., 2016).
However, the extent to which RSP-projecting claustrum neurons form a segregated population or co-
localize with other projection-defined claustrum neurons remains uncertain. To directly test this, we
injected equal volumes of cholera toxin subunit B (CTB) conjugated with Alexa 647, 488, or 555 into
the anterior cingulate area (ACA), primary visual cortex (V1), and RSP, respectively, in the same
animal to retrogradely label claustrum neurons that project to each of these regions (Figure 2.1A).
22

The ACA was selected since retrograde tracer injections here have been shown to strongly label
claustrum neurons throughout the entire length of the structure, and the distribution of these cells has
been confirmed to fall within the boundaries of the claustrum as defined by dense parvalbumin (PV)
immunoreactivity and Gng2 expression (Mathur et al., 2009; White et al., 2016; White et al., 2018).
Similarly, V1 was selected as it has been shown to receive claustrum input, however, some reports
suggest that sensory cortical regions may receive segregated inputs from discrete parts of the
claustrum (White et al., 2016; Kitanishi & Matsuo, 2017). Interestingly, all three injections resulted in
claustrum cell body labeling that was spatially overlapped and evenly distributed along the length of
the structure (Figure 2.1C). However, approximately twice as many cells were observed projecting to
the ACA compared to RSP or V1, suggesting a denser innervation profile for this structure (Figure
2.1D). Occasionally, retrogradely labeled cells were found outside the claustrum in the surrounding
insular cortex following injections in ACA, but not RSP or V1, making ACA a less attractive target for
accessing claustrum neurons, despite its stronger connectivity.
To determine whether ACA-, V1-, and RSP-projecting claustrum neurons represent separate
populations with unique output or rather collateralize to multiple targets, individual neurons were
examined for the presence of two or more tracers using confocal imaging at 40X magnification (Figure
2.1B, bottom panels). For each population, a surprisingly large fraction of cells was found to be
double-labeled, ranging from 23-57% for each combination, while the occurrence of triple-labeled cells
ranged from 15-32% (Figure 2.1E). This suggests that claustrum neurons, including those projecting
to RSP, may frequently collateralize to innervate multiple cortical regions. Taken together, these data
reveal that RSP receives input from a population of uniformly distributed, broadly targeting claustrum
neurons, and may therefore serve as a suitable target for capturing a representative population of
claustrum cells using a retrograde viral tracing approach.
23

Figure 2.1. Co-localization of projection-defined claustrum neurons.
(A) Injection site locations in anterior cingulate (ACA, CTB 647, blue), primary visual cortex (V1, CTB 555, red),
and retrosplenial cortex (RSP, CTB 488, green). Scale bar: 500 µm.
(B) Retrograde labeling in claustrum (CLA). Bottom panels, 40X magnification of V1, RSP, and ACA projecting
CLA neurons. Scale bars: 500 µm, top panel; 50 µm, bottom panels.
(C) Distribution of retrogradely labeled cells along the claustrum axis for each injection site. Values in mm
relative to bregma. Scale bar: 250 µm.
(D) Fraction of co-labeling for total population of CLA neurons quantified for all 3 animals. 12% of the entire
population of cells project to all three injection sites.
(E) Quantification of co-labeling for V1, RSP, and ACA-projecting CLA neuron populations (n = 3 mice, error bar
= 95% CI).
24

2.3.2 Output of projection-defined claustrum neurons
To directly examine the output of RSP-projecting claustrum neurons, a recently developed retrograde
variant of AAV (AAVretro-hSyn-Cre; Tervo et al., 2016) was injected into the RSP in Ai14 Cre-
dependent tdTomato reporter mice (Madisen et al., 2012). Robust labeling of Cre+/Tom+ neurons
was observed throughout the length of the claustrum, but not in surrounding cortical or endopiriform
regions (Figure 2.2B). Injection of a second, Cre-dependent GFP-expressing virus (AAV1-CAG-
FLEX-GFP) within the CLA enabled selective expression of GFP in CLA neurons and characterization
of their axonal output across the entire brain (Figure 2.2B-C). Interestingly, a dense pattern of
innervation was observed throughout multiple cortical regions, including orbital and medial prefrontal
cortex, ACA, and the retrohippocampal complex, in addition to the expected labeling in RSP (the
primary target of these cells) (Figure 2.2C-D). This demonstrates that many RSP-projecting
claustrum neurons collateralize extensively to reach a wide variety of cortical targets. In addition,
relatively more diffuse labeling was observed throughout posterior parietal and visual cortical areas
(Figure 2.2C), while little or no labeling was found in primary motor, somatosensory, or auditory areas
(Figure 2.2D). Output was restricted almost entirely to the ipsilateral hemisphere. This projection
pattern was similar to that previously characterized for two genetically-defined populations of
claustrum neurons (Gnb4+ and Ntng2+; Wang et al., 2016), except that we did not observe axonal
labeling in the insular or piriform cortices, nor in any subcortical structure outside of the basolateral
amygdala (BLA), including the mediodorsal (MD) and parvicellular ventral posteromedial (VPMpc)
nuclei of the thalamus, which receive input from the surrounding insular cortex (Shi & Cassell, 1998;
Figure 2.2C). These results suggest that RSP-projecting claustrum neurons may comprise a subset
of “generic” CLA principal cells that collateralize broadly and overlap substantially with other target-
defined claustrum populations.

25

Figure 2.2. Axonal output of RSP-projecting CLA neurons.
(A) Viral injection strategy for labeling CLA neurons.
(B) Example injection of AAVretro-Cre in RSP (top left panel) and AAV1-FLEX-GFP in CLA (bottom left panel) in
an Ai14 tdTomato Cre-reporter mouse. Retrogradely labeled Cre+ neurons (red) are found in CLA, but not
surrounding brain structures, and co-express GFP (green, right panels). Scale bars: 500 µm, left panels; 100
µm, right panels.
26

(C) Overview of axonal projection pattern for RSP-projecting CLA neurons (right panel). No subcortical labeling
was observed, except in the basolateral amygdala. Density of axonal termination was greatest in layers (L)2/3
and L5 of orbital cortex and L1, L5a, and L6 of cingulate, but was more diffuse in parietal and visual cortical
areas (left panels). Scale bars: 1 mm, right panel; 250 µm, left panels.
(D) Quantification of axonal fluorescence density expressed as a percentage of the total for all brain regions
containing labeled axons (n = 3 animals, error bar = 95% CI).

To further test this idea, we injected AAVretro-Cre in either ACA or V1 and AAV1-FLEX-GFP
into the claustrum to directly compare the output of each of these populations with that of RSP-
projecting neurons. In the case of ACA-projecting neurons, iontophoretic injections were applied in
the claustrum to minimize the potential for labeling Cre+ neurons sparsely found in the overlying
insular cortex. Qualitatively, the pattern of output across the entire cortex for ACA- and V1-projecting
neurons matched that observed for the RSP-projecting population (Figure 2.3A), suggesting that each
group branches substantially to innervate the same set of cortical structures. To more closely
examine any targeting bias among these populations, we quantified and compared the relative axonal
projection density in several different cortical regions for each case (Figure 2.3B). Similar densities
were observed for each population (no significant difference revealed by two-way ANOVA), however
ACA-projecting claustrum neurons showed a slight bias toward innervating frontal cortical regions
(ORBvl, mPFC, ACA) relative to more posterior regions (RSP, ENTl, V1) when compared to RSP- and
V1-projecting populations. These results demonstrate that, overall, these projection-defined
claustrum neurons share a common pattern of cortical innervation and may represent largely
overlapping populations, however subtle bias may exist in the relative density of innervation for
different groups.

27

Figure 2.3. Comparison of output for different projection-defined CLA neurons.
(A) Claustrum injection sites (top panels, green) and example axonal labeling in several cortical regions for
RSP-, ACA-, and V1-projecting claustrum populations.
(B) Quantification of fluorescent axon density expressed as a fraction of the total density found for 7 sampled
areas for RSP-, ACA-, and V1-projecting claustrum populations. Two-way ANOVA showed no significant
difference in overall targeting preference between each population (n = 3 animals for each group, error bar =
95% CI).

2.3.3 Monosynaptic input to RSP-projecting claustrum neurons
We next aimed to characterize the presynaptic input to the claustrum using monosynaptic rabies virus
tracing and RSP-projecting claustrum cells as a representative starter population (Wickersham et al.,
2007; Wall et al., 2010; Sun et al., 2014). RSP-projecting claustrum neurons were primed for uptake
and monosynaptic transport of a pseudotyped mCherry-expressing rabies virus (EnvA-ΔG-mCherry
rabies) following injections of AAV1-DIO-TVA-GFP and AAV1-FLEX-RG into the claustrum (Figure
2.4A, see methods). Starter cells were identified based on their co-expression of GFP and mCherry
(Figure 2.4B) while presynaptic neurons expressed only mCherry (Figure 2.4C-E). Serial sections
were collected across the entire brain and imaged to reveal the distribution of claustrum inputs
28

(summarized in Figure 2.5) and individual cell bodies were quantified within ipsi- and contralateral
cortical and subcortical regions (Figure 2.6).
Overall, most input to the claustrum arose from the cortex (Figure 2.6A), which accounted for
about 84% of the total presynaptic cell population (Figure 2.6B). Among these inputs, higher-order
cortical regions were most prominently labeled, including orbital, medial prefrontal, cingulate,
temporal, and entorhinal areas (Figure 2.5 and Figure 2.6A). Cell bodies were found mostly in cortical
layers 2/3 and 5 for each of these regions, however layer 6 was also labeled in temporal cortex
(Figure 2.4D). Sparse labeling was also observed in the parietal cortex (layers 5 and 6), auditory
cortex (layers 2/3 and 5), and in higher visual areas (layers 2/3, 5, and 6) (Figure 2.4D; Figure 2.5).
Very little labeling was found specifically within primary visual cortex (VISp), however the few cells we
did observe tended to be located within cortical layer 6 (Figure 2.5). This is consistent with previous
findings in rat (Smith & Alloway, 2014) and in cat (LeVay & Sherk, 1983; Katz, 1987), though a far
greater number of layer 6 CLA-projecting neurons are present in cat VISp, perhaps representing a
species-specific difference. In addition, substantial intraclaustral connectivity was found in regions
posterior to the starter population (Figure 2.4D, bottom right panel). Higher magnification revealed
many of these neurons exhibited dendritic spines, a morphological feature of CLA principal cells,
suggesting these neurons may form a network of excitatory connections along the length of the
structure. Finally, little or no labeling was observed in the somatosensory or motor cortex, or in the
retrosplenial cortex (Figure 2.5; Figure 2.6A). Overall, the major sources of input to the claustrum
appeared to be the same regions that were most heavily targeted by claustrum output, suggesting
that these areas define a reciprocally connected claustro-cortical network. A notable exception,
however, is the retrosplenial cortex which receives dense input from, but does not project back to, the
claustrum.

29

Figure 2.4. Monosynaptic input to RSP-projecting CLA neurons.
(A) Viral injection strategy for mapping pre-synaptic input to RSP-projecting CLA neurons.
(B) Example injection site showing CLA starter cell population (yellow) co-expressing rabies glycoprotein (RG),
TVA receptor, GFP and rabies-mCherry (red). Scale bar = 200 µm.
(C) Example images of presynaptic rabies-mCherry labeling (red) in orbital, medial prefrontal, cingulate, and
entorhinal regions.
(D) Example images of cortical input to claustrum. Labeled cells were most prominent in layers (L)2/3 and L5 of
orbital and cingulate cortex (top panels), and L5 of parietal and entorhinal cortex (middle panels). More diffuse
labeling was seen across deep layers of temporal cortex (bottom left panel). Presynaptic labeling was also
observed within the claustrum in sections more posterior to the starter population (bottom right panel). Higher
magnification confirmed many of these cells were principal neurons with dendritic spines (inset).
(E) Examples images of subcortical input to the claustrum.
30

Figure 2.5. Summary of brain-wide input to RSP-projecting CLA neurons.
Example labeling for one case plotted on corresponding coronal atlas sections. Red dots indicate location of
retrogradely labeled cell bodies and yellow dots indicate extent of starter cell population. See Table 2.1 for list of
abbreviations.
31

In addition, while nearly all claustro-cortical output was found to target the ipsilateral cortex
(Figure 2.2C), we found prominent inputs to the claustrum from contralateral cortical regions,
suggesting an asymmetrical interhemispheric organization of the claustro-cortical network. In
particular, while most cortical input areas showed an ipsilateral bias in their projection to the
claustrum, cell body labeling was approximately equal in both hemispheres of the medial prefrontal
cortex and was about 30% greater in the contralateral cingulate cortex (Figure 2.6B). This is in line
with previous studies that have reported bilateral asymmetry in cortical projections to the claustrum
using retrograde (Smith & Alloway, 2010) and anterograde tracing methods (Smith & Alloway, 2014;
Atlan et al., 2016; Wang et al., 2016). Together, these results suggest that cortical inputs to the
claustrum are highly convergent, as a wide variety of cortical regions provide synaptic input to a
specific set of projection-defined claustrum cells. Moreover, these inputs arise primarily from the
same higher-order cortical regions that receive claustral output, thus defining a claustro-cortical
network that is largely reciprocal, but contains notable examples of asymmetry in its interhemispheric
connections.
Finally, inputs to the claustrum from subcortical regions have not been systematically
characterized. We therefore examined brain sections for any presynaptic cell body labeling in regions
outside the cortex. Substantial labeling was found in the basolateral amygdala; supramammillary
nucleus; paraventricular, anteromedial, and reuniens nuclei of the thalamus; and the CA1 region of
the ventral hippocampus (Figure 2.4E, Figure 2.5, and Figure 2.6A). In addition, we found prominent
input from neuromodulatory cell populations in the dorsal raphe (presumed serotonergic neurons),
and the basal forebrain, which contains cortically projecting cholinergic neurons. To further confirm
the identity of presynaptically labeled basal forebrain neurons, we performed monosynaptic rabies
virus tracing in GAD67-GFP mice (Tamamaki et al., 2003) combined with immunostaining for choline
acetyltransferase (ChAT), a marker for cholinergic neurons (Figure 2.6D-F). We found that
approximately 65% of claustrum-projecting basal forebrain neurons are cholinergic, while ~20% of
them are GAD67+. The remaining 15% may be either glutamatergic or GAD2+ inhibitory neurons.
32

Together, these data suggest that, in addition to a prominent network of cortical inputs, claustrum
neurons may be driven by a set of subcortical regions implicated in setting motivational and arousal
states (summarized in Figure 2.7E).

Figure 2.6. Quantification of presynaptic input to RSP-projecting CLA neurons.
(A) Number of cells projecting to CLA from ipsilateral (left) and contralateral (right) brain regions expressed as a
percentage of the total number of labeled cells quantified for the entire brain for each case (n = 4 animals, error
bar = 95% CI). See Table 2.1 for list of abbreviations.
33

(B) Quantification of bilateral asymmetry in the number of cells projecting to CLA from ipsilateral and
contralateral cingulate, medial prefrontal cortex (mPFC), and all other cortical regions. Percentages reflect the
number of cells quantified for ipsilateral or contralateral regions divided by the total number of cells quantified for
that specific region for each case (n = 4 animals; error bar = 95% CI; n.s., not significant, **p < 0.01, ***p <
0.001; paired t-test).
(C) Fraction of total cells projecting to CLA from cortical versus subcortical brain regions (n = 4 animals, error
bar = 95% CI, ***p < 0.001, paired t-test).
(D) Schematic of monosynaptic rabies virus injection strategy for labeling basal forebrain neurons that project to
CLA in GAD67-GFP mice.
(E) Retrograde labeling in basal forebrain (red). Cells co-localized with both GAD67-GFP inhibitory neurons
(green) and cholinergic neurons (ChAT+ immunostaining, blue).
(F) Quantification of co-localization with ChAT+ and GAD67+ neurons expressed as a percentage of total RAV+
neurons in the basal forebrain for each case (n = 3 mice, error bar = 95% CI, total cell counts for all 3 mice: 424
ChAT+/663 RAV+ cells, 137 GAD67+/663 RAV+ cells).

Lastly, it is worth noting that very little interaction was observed between RSP-projecting CLA
neurons and primary motor and somatosensory cortices in our input-output studies. This is similar to
previous observations made by Wang et al. (2016), however other studies have suggested that a
dorsal component of the CLA may be reciprocally connected with motor and somatosensory areas
(Sadowski et al., 1997; Smith & Alloway, 2014; White et al., 2016). To test if these motor cortex-
projecting neurons overlap with RSP-projecting CLA neurons, we injected CTB 488 into the primary
motor cortex (MOp) and CTB 555 into the RSP in the same animal and examined the claustrum for
retrogradely labeled cells (Figure 2.7A-B). Interestingly, each injection gave rise to two largely
segregated populations. RSP-projecting CLA neurons formed an oval-shaped cluster that extended
the length of the structure, while MOp-projecting neurons formed a thinner cluster just dorsal to this
region and were predominately located near the anterior half of the claustrum. The distribution of
these MOp-projecting neurons roughly matched that of several known genetic markers for CLA and
endopiriform (EP) cells, including Gnb4, Ntng2, and Latexin (Figure 2.7C; Watakabe et al., 2014;
Wang et al., 2016), however they appeared to lie outside of the parvalbumin-enriched region used to
define CLA in Mathur et al., (2009) which circumscribes cingulate- and RSP-projecting CLA
populations (Mathur et al., 2009; White et al., 2016; Kim et al., 2016). Whether or not these MOp-
34

projecting neurons co-express such genetic markers and should be considered part of a larger
claustrum complex, or rather comprise deep layers of gustatory and visceral cortex as proposed by
Wang et al. (2016), requires further investigation (see Figure 2.7D).

Figure 2.7. Claustrum boundaries and summary of brain-wide connectivity.
(A) Schematic of cholera toxin subunit B (CTB) injection sites in primary motor (MOp) and retrosplenial cortex
(RSP). Values shown in mm relative to bregma.
35

(B) Retrograde labeling in CLA (dashed circle) following injection of CTB 555 in RSP (red) and CTB 488 in MOp
(green) in the same animal. MOp projecting cells cluster just dorsal to the CLA region defined by RSP-
projecting cells. Values shown in mm relative to bregma. Scale bar 250 μm.
(C) Gene expression pattern for latexin, a known genetic marker for CLA and endopiriform (EPd) cell types (see
Watakabe et al., 2014). Gene expression is enriched in a dorsal population of cells that may correspond to
motor cortex projecting CLA neurons. In situ hybridization (ISH) data from the Allen Brain Atlas database
(www.brain-map.org, Experiment #72340108).
(D) Schematic showing distinct network relationships for EPd and CLA as well as the potential inclusion of a
dorsal CLA region that interacts specifically with somatosensory and motor cortex (see Smith & Alloway, 2014;
Watson et al., 2017; but also see Mathur et al., 2009; Wang et al., 2016).
(E) Schematic representation of major inputs to CLA characterized in this study. Major inputs include medial
prefrontal, cingulate, and retrohippocampal cortical regions (dark gray); and neuromodulatory (BF, DR), midline
thalamic (PVT, RE), amygdalar (BLA), supramammillary (SUMl), and ventral hippocampal (CA1v) subcortical
regions (light gray).
(F) Summary of claustrum output. Axonal projections are predominately restricted to ipsilateral cortex, except for
a sparse subcortical projection to basolateral amygdala (BLA). Dense projections are found along midline
cortical regions, including orbital, medial prefrontal, cingulate, and retrosplenial areas, as well as more laterally
located temporal (TEa), entorhinal (ENT), and retrohippocampal (PRE, POST) regions. Substantial projections
were also found diffusely throughout posterior parietal (PTLp) and visual (VIS) cortical regions.

2.4 Discussion
In this study, we used a retrograde viral injection strategy to access CLA neurons that project to
retrosplenial cortex. We found that injections of AAVretro-hSyn-Cre into the posterior-most parts of
RSP in Ai14 mice robustly labeled neurons throughout the length of the CLA, but not in any
surrounding structures. This enabled highly selective targeting of these CLA neurons with local
injections of Cre-dependent AAV or pseudotyped rabies virus to establish their input-output profile.
Using this approach, we (1) confirmed a lack of subcortical projections from CLA principal neurons,
with the exception of sparse projections to BLA; (2) demonstrated extensive collateralization of CLA
output, with similar targeting trends for three different projection-defined populations; (3) and showed
for the first time the presynaptic sources of input to the CLA from across the entire brain using
monosynaptic rabies tracing.
Several studies have systematically examined the output of the claustrum in rodents using
anterograde or retrograde tracing techniques (Sadowski et al., 1997; Smith & Alloway, 2014; Wang et
36

al., 2016; White et al., 2016). These studies characterized widespread projections to the ipsilateral
cortex, and, in particular, emphasized strong innervation of higher cortical centers in prefrontal,
cingulate, and retrohippocampal regions. Our results confirm these findings and clarify additional
unresolved issues related to claustrum output. Specifically, some reports suggest the claustrum may
project to subcortical regions such as the striatum, amygdala, thalamus, and hypothalamus (Carey &
Neal, 1986; Dinopoulos et al., 1992; Wang et al., 2016). These results are inconsistent, however, and
may be due to the spread of injections into the surrounding insular cortex. Using our projection-based
approach to selectively target CLA neurons, we observed sparse termination in the BLA, however no
other subcortical projections were identified. This suggests that CLA output is almost entirely
restricted to the cortex, and previously observed thalamic and striatal projections may have arisen
from labeled neurons in layers 5 and 6 of the insular cortex. In addition, we did not see any
projections directly into, or from, the overlying insular cortex, as has been noted in some previous
studies (Wang et al., 2016; Kitanishi & Matsuo, 2017). This is in line with previous work in cat
reporting such a lack of connections (Markowitsch et al., 1984), and suggests that the CLA functions
separately from the insula, despite residing within its deepest layers. These observed differences in
connectivity may be due to the enhanced selectivity in labeling CLA neurons, but not surrounding cell
types, afforded by our projection-based targeting approach.
Despite extensive work characterizing claustrum output, the degree to which these neurons
segregate to form target-specific output populations or collateralize to uniformly innervate the cortex
remains unknown. This is important to resolve as some studies suggest the claustrum may be
organized into modules with distinct input and output (Smith & Alloway, 2014; Atlan et al., 2016; White
et al., 2016; Kitanishi & Matsuo, 2017), however recent findings based on the reconstruction of
several individual claustrum neurons suggest that they are capable of collateralizing to a large number
of cortical targets and may therefore broadly influence cortical function (Wang et al., 2017). Our multi-
fluorescent retrograde tracing revealed that 23-57% of CLA neurons projecting to either V1, RSP, or
ACA, also projected to another target, and a modest fraction (15-32%) projected to all three targets.
37

The lower ends of these ranges corresponded to the ACA-projecting population, which consistently
showed a lower fraction of co-labeling due to the comparatively larger (roughly two-fold greater) total
population of CLA cells back-labeled with each injection (Figure 2.1C-D). These co-labeling
percentages are much higher than previously reported (White et al., 2016; Kitanishi & Matsuo, 2017),
yet still likely represent an underestimate of the total number of CLA neurons that collateralize to
multiple targets, given that each injection site samples only a small region of cortical space.
These data shed some light on the frequency of multi-targeting for CLA neurons, but still do
not resolve whether projection-defined populations exhibit bias in their output profile and thus
represent unique output channels from the claustrum. To address this, we examined the pattern of
axonal output from three different target-defined CLA populations. Surprisingly, no significant
difference in the cortical innervation pattern (as determined by two-way ANOVA) was observed for
CLA neurons defined by their projection to ACA, RSP, or V1, though caution must be taken in
interpreting these data given the high false discovery rate associated with smaller sample sizes (n = 3
in this case, see Colquhoun, 2014). In addition, each population innervated the full set of cortical
targets previously described for two genetically-defined CLA populations (Gnb4+ and Ntng2+; Wang
et al., 2016), suggesting that the output from each of these target-defined populations may be
representative of the CLA as a whole, at least as it is defined here (i.e. within the region characterized
by dense PV- and Gng2-immunoreactivity, which has been shown to circumscribe cingulate- and
retrosplenial-projecting CLA populations; see Mathur et al., 2009; White et al., 2016; Kim et al., 2016;
and Figure 2.7). Together, these data reveal that CLA principal neurons may frequently and
indiscriminately collateralize to multiple targets, and their output may therefore serve to globally
modulate processing across a wide variety of cortical areas. Given the laminar specificity of CLA
output, especially in layers 1, 5a, and 6 of cingulate and medial prefrontal cortex, determining the
specific post-synaptic cell classes that are innervated by the CLA, and the net effect of activation on
local cortical processing, remains an important question to resolve (Jiang et al., 2013; Harris &
Shepherd, 2015).
38

In addition, future studies may seek to clarify the existence of a more dorsally situated
population of claustrum neurons that preferentially interacts with primary motor and somatosensory
cortex, as suggested by previous work using anterograde and retrograde tracer injections in these
regions (Smith & Alloway, 2010; Smith & Alloway, 2014; White et al., 2016). These neurons appear to
be largely segregated from the ACA-, V1-, and RSP-projecting CLA neurons defined in this study, as
axonal output from the latter populations was extremely sparse or absent in primary motor and
somatosensory cortex (Figure 2.2C and Figure 2.3A), and dual-retrograde injections in RSP and
primary motor cortex revealed mostly non-overlapping cell populations along the claustrum axis
(Figure 2.7B). This population may therefore function in parallel with the more ventrally-defined CLA
region in this study, perhaps analogous to the endopiriform nucleus which lies below the CLA and
interacts specifically with olfactory-related cortical networks (Watson et al., 2017; Figure 2.7D).
Whether or not these motor cortex-projecting neurons share genetic and morphological features of
claustrum neurons, however, or rather belong to deep layers of overlying gustatory and visceral cortex
requires further investigation (Wang et al., 2016).
Several studies have systematically characterized inputs to the claustrum in rodents using
anterograde tracing (Atlan et al., 2016; Wang et al., 2016). This approach comes with some
limitations, however, as the extent to which axons projecting to the claustrum form synaptic
connections or merely pass through the structure remains uncertain, and the laminar distribution and
morphological identity of the input neurons are not revealed. We therefore used monosynaptic rabies
virus tracing to directly reveal this information and disambiguate the synaptic nature of the input.
Using RSP-projecting CLA neurons as a starter population, we demonstrated widespread inputs from
both cortical and subcortical regions. Most presynaptically labeled cells were found in the cortex
(84% of total) and the greatest sources of input were often the same cortical regions that received the
densest innervation from the CLA, including orbital, mPFC, cingulate, and entorhinal cortices,
suggesting a reciprocal network among these higher order cortical regions and the CLA. An
exception however, was retrosplenial cortex, which was found to receive strong input from CLA, but
39

did not project back. Additionally, most input to the CLA arose from sources ipsilateral to the injection
site, however about half of the input from mPFC and over 60% from ACA arose from the contralateral
hemisphere. This is in agreement with previous observations reported with retrograde (Smith &
Alloway, 2010) and anterograde tracing (Smith & Alloway, 2014; Atlan et al., 2016; Wang et al., 2016),
however the functional significance of this bilateral asymmetry remains to be determined. It is
interesting to note, as well, that such widespread presynaptic input, from nearly all of the potential
sources implicated in previous studies (e.g. Wang et al., 2016), was observed for this projection-
defined population of starter cells. This suggests that inputs to CLA neurons, in general, may be
highly convergent and points to a more uniform, rather than modular, circuit organization for this
structure in the mouse brain. Finally, it should be noted that this study reveals inputs specifically to
excitatory principal neurons in CLA, so it remains to be determined if there are any differences in the
sources or bias of inputs to local inhibitory cell-types within the CLA circuit.
Inputs to the CLA from subcortical regions have not been systematically reported. We
therefore examined the entire brain for any sources of subcortical input to the CLA. Interestingly, we
found prominent labeling in regions implicated in regulating the arousal and the affective state of the
animal, including the basolateral amygdala, CA1 region of the ventral hippocampus, paraventricular
(PVT) and reuneins (RE) nuclei of the thalamus, lateral supramammillary nucleus (SUMl), and
serotonergic and cholinergic inputs from the dorsal raphe and basal forebrain, respectively (Swanson,
2000; Vertes et al., 2015; Reppucci & Petrovich 2015; Hu 2016). Many of these regions have been
shown to modulate their activity in response to rewarding or aversive stimuli or to cues that predict
such outcomes (Hangya et al., 2015; Zhu et al., 2016; Burgos-Robles et al., 2017; Matias et al., 2017;
Beyeler et al., 2018; Jimenez et al., 2018). The CLA may therefore integrate some of these response
features and broadcast its output to the cortex to modulate processing under emotionally salient
conditions. Taken together, our results provide a more complete understanding of the claustrum
network in the mouse brain and provide an alternative approach for precisely targeting claustrum
neurons that may be adapted for use in future functional studies.
40

CHAPTER 3

Anterograde transsynaptic transfer of AAV: Mapping cortico-collicular input-defined
neural pathways for defense behavior

Adapted from:

Zingg, B., Chou, X., Zhang, Z., Mesik, L., Liang, F., Tao, H. W., & Zhang, L. I. (2017). AAV-Mediated
Anterograde Transsynaptic Tagging: Mapping Corticocollicular Input-Defined Neural Pathways for
Defense Behaviors. Neuron, 93(1), 33–47. http://doi.org/10.1016/j.neuron.2016.11.045

41

3.1 Introduction
A key for deciphering complex brain circuits is the ability to identify neural pathways underlying
distinct behaviors or brain functions. Given that a brain region (“X”, Figure 3.1) is known to be
involved in a specific behavior/function, determining which neural pathways downstream of X that
mediate this behavior/function remains challenging. Traditional anterograde tracing methods allow
the identification of all regions targeted by neurons in X. However, for each of these regions (“Y 1” to
“Yn”, Figure 3.1), the possibility exists that there are neurons that do not receive direct input from X
and are unrelated to the X-dependent function under study. These neurons are not necessarily
molecularly or genetically different from those directly involved in the function of interest based on our
current knowledge. Therefore, to precisely trace the neural pathway underlying the X-dependent
function under study, it is necessary to identify in a relevant downstream nucleus (Y) the neuronal
subpopulation that receives direct input from X. This requires afferent-dependent tagging of
postsynaptic neurons only in a selected target region, which to our knowledge has not been achieved

Figure 3.1. Advantages of anterograde transsynaptic mapping of functional circuits.
(A) A brain region “X” is known to mediate a behavior/function of interest. Neurons in X project to multiple target
nuclei (“Y”). To map the relevant downstream circuit, conventional method relies on activation of ChR2-
expressing X axon terminals in a given target nucleus. This may result in unwanted activation of collateral
targets via antidromic stimulation (marked by dash lines).
(B) A virus capable of anterograde transsynaptic spread would allow direct activation of postsynaptic cells in a
target region that specifically receive input from region X, by enabling Cre-dependent transgene expression
(green) in a Y nucleus.
42

with current approaches (Nassi et al., 2015). A straightforward strategy for labeling those input-
defined postsynaptic neurons is to apply an anterograde transsynaptic tracer in the source region
coupled with tracer-dependent transgene expression in the selected target region. This would allow
functional manipulations of a specific second-order downstream pathway starting from the source
nucleus.
Transcellular/transsynaptic viral tracers can be potentially powerful tools for mapping
functional circuits in the above described manner (Figure 3.1B). However, while retrograde
transsynaptic viral tracers, e.g. rabies virus and pseudorabies virus (PRV), have been widely used for
mapping presynaptic inputs to transduced neurons (Wickersham et al., 2007; Wall et al., 2010;
Defalco et al., 2001; Ekstrand et al., 2008), analogous anterograde transsynaptic tools for tracing
neural pathways immediately downstream of targeted postsynaptic neurons remain under
development. Although several viruses, e.g. herpes simplex virus (HSV) and vesicular stomatitis virus
(VSV), have been found previously to exhibit transneuronal/transsynaptic spread, the neurotoxicity of
these viruses and their uncontrollable spread across multiple serial synapses largely limit their
applications in mapping circuits in a more precise manner (Lo and Anderson, 2011; Beier et al.,
2011). Recent studies suggest that AAV1 can be transported anterogradely down the axon (Castle et
al., 2014a; Castle et al., 2014b). However, evidence for its potential transneuronal/transsynaptic
transduction remains largely unclear and controversial (Oh et al., 2014; Salegio et al., 2013; Hutson et
al., 2015; Harris et al., 2012; Aschauer et al., 2013). For example, anterograde transneuronal spread
of AAV1 has not been observed within the Allen Institute database for mouse brain connectivity (Oh et
al., 2014). Here, we investigated whether different serotypes of AAVs can spread from presynaptic to
postsynaptic cell populations. Together with the development of an intersectional approach, we have
established AAV1 as an effective anterograde transsynaptic tracer for mapping input-defined
functional neural pathways.

43

3.2 Materials and Methods
3.2.1 Animal preparation and stereotaxic surgery
All experimental procedures used in this study were approved by the Animal Care and Use
Committee at the University of Southern California. Male and female C57BL/6J and Ai14 (Cre-
dependent tdTomato reporter, RRID: IMSR_JAX:007914) mice (Jackson Laboratories) aged 2-6
months were used in this study. Mice were group housed in a light controlled (12 hr light: 12 hr dark
cycle) environment with ad libitum access to food and water.
Stereotaxic injection of viruses was carried out as we previously described (Ibrahim et al.,
2016; Liang et al., 2015; Xiong et al. 2015). Mice were anesthetized initially in an induction chamber
containing 5% isoflurane mixed with oxygen and then transferred to a stereotaxic frame equipped with
a heating pad. Anesthesia was maintained throughout the procedure using continuous delivery of 2%
isoflurane through a nose cone at a rate of 1.5 liters/min. The scalp was shaved and a small incision
was made along the midline to expose the skull. After leveling the head relative to the stereotaxic
frame, injection coordinates based on the Allen Reference Atlas (Dong, 2007) were used to mark the
location on the skull directly above the target area and a small hole (0.5mm diameter) was drilled.
Viruses were delivered through pulled glass micropipettes with a beveled tip (inner diameter of tip:
~20 µm) using pressure injection via a micropump (World Precision Instruments). Total injection
volume was 50 to 100 nl, at 15 nl/min. Following injection, the micropipette was left in place for
approximately 5 mins to minimize diffusion of virus into the pipette path. After withdrawing the
micropipette, the scalp was sutured closed and animals were administered ketofen (5mg/kg) to
minimize inflammation and discomfort. Animals were recovered from anesthesia on a heating pad
and then returned to their home cage.
For virus injection into the retina, animals were anesthetized and positioned in a stereotaxic
frame as described above. Using a pulled glass micropipette with a beveled tip (40 µm inner tip
diameter), 800 nl of virus was injected into the left eye at a rate of 20 nl/min at a depth of 0.7 mm from
44

the surface of the lateral most curvature of the eye using a 40º angle of approach relative to the optic
axis.
3.2.2 Injection of viruses for anterograde transneuronal labeling
To demonstrate the anterograde transneuronal properties of AAV, AAV2/1-hSyn-Cre-WPRE-hGH
(UPenn Vector Core, 2.5 x 1013 GC/mL) was injected into either V1 (60 nl total volume; 3.9 mm
posterior and 2.6 mm lateral to bregma and 0.5 mm ventral from the cortical surface) or retina (800 nl
total volume) of Ai14 mice using the pressure injection method. Animals were euthanized 4 weeks
following injection to allow time for viral transport and transgene expression. To characterize the
temporal progress of the transport, mice were also euthanized at 2-day, 5-day, 2-week and 3-month
post-injection time points.
To examine the potential anterograde transneuronal properties of other viruses, AAV2/1-CMV-
PlCre-rBG (UPenn Vector Core, 2.7 x 1013 GC/mL), AAV2/5-CMV-Pl-Cre-rBG (UPenn Vector Core,
2.8 x 1013 GC/mL), AAV2/6-CMV-Pl-Cre-rBG (UPenn Vector Core, 3.5 x 1013 GC/mL), AAV2/8-
CMV-Pl-Cre-rBG (UPenn Vector Core, 4.4 x 1013 GC/mL), AAV2/9-CMV-Pl-Cre-rBG (UPenn Vector
Core, 1.6 x 1014 GC/mL), AAV2/1-CB7-Cl-eGFP-WPRE-rBG (UPenn Vector Core, 4.2 x 1013
GC/mL), or CAV2-CMV-Cre (Montpellier Vector Core, 1.3 x 1012 GC/mL) was injected into V1 (60 nl
total volume) of Ai14 mice (for Cre-expressing viruses) or wild-type C57BL/6J mice (for GFP-
expressing virus). Original titers were used, while for testing dependence on titer AA1-CMV-Cre was
also diluted (10X or 20X). Note that the original titer was lowest for AAV1 among tested AAV
serotypes all expressing CMV-driven Cre. Animals were euthanized 4 weeks following injection and
postsynaptic structures were examined for the presence of cell body labeling.
Retrograde spread of AAV1 To test for potential retrograde transport of AAV, injections of
either AAV2/1-hSyn-Cre-WPREhGH (60 nl) or AAV2/1-CB7-Cl-eGFP-WPRE-rBG (60 nl) were made
into SC (3.9 mm posterior and 0.8 mm lateral to bregma and 1.5 mm ventral from the cortical surface)
45

of either Ai14 or wild-type C57BL/6J mice, respectively. Animals were euthanized 4 weeks following
injection.
Two-step viral injection For mapping the axonal outputs of subpopulations of neurons in
SC, AAV2/1-hSyn-Cre-WPREhGH was injected into V1, contralateral retina, A1 (3.1 mm posterior and
4.5 mm lateral to bregma and 0.75 mm ventral from the cortical surface) or M1 (0.5 mm anterior and
1.5 mm lateral to bregma and 0.5 mm ventral from the cortical surface) of Ai14 mice (60 nl total
volume). Following 2 to 7 days, a second injection of AAV2/1-CAG-FLEX-eGFP-WPRE-bGH (UPenn
vector core, 1.7 x 1013 GC/mL, originally created by Allen Institute) was made into the ipsilateral SC
(3.9 mm posterior and 0.8 mm lateral to bregma and 1.5 mm ventral from the cortical surface; 60 nl
total volume). The spacing of the two injections over several days was selected to allow sufficient
time for the clearance of any residual AAV-Cre virus that may have spread across the pial surface in
an effort to eliminate any local contamination of the Cre-dependent virus injection site. Animals were
allowed to recover for at least 4 weeks following the second injection.
For behavioral testing, AAV2/1-hSyn-Cre-WPRE-hGH was injected into V1, A1, or SC, as
described above. Following 2 to 7 days, a second injection of AAV2/1-EF1a-DIO-hChR2-eYFP
(UPenn vector core, 1.6 x 1013 GC/mL) was made into SC (for A1 and V1 injections of AAV-Cre) or
LP (for SC injections of AAV-Cre; 2.4 mm posterior and 1.6 mm lateral to bregma and 2.6 mm ventral
to the cortical surface). Animals were then prepared for optogenetic testing 4 weeks after the second
injection. To estimate the fraction of LP-projecting SC cells that are labeled following V1 injections of
AAV-Cre, AAV2/1-hSyn-Cre-WPRE-hGH was injected into V1 of Ai14 tdTomato mice as described
above, and LP was injected with cholera toxin subunit B, Alexa 488 (CTB-488, 100 nl injection
volume, 0.5% solution in PBS, ThermoFisher) using coordinates described above. Animals were
euthanized 4 weeks after injection.
For labeling glutamatergic and GABAergic SC neuronal populations receiving V1 input, AAV1-
EF1a-DIO-Flp (1.5 x 1014 GC/mL, custom design, ViGene Biosciences) was injected into V1 (60 nl
46

total volume) of Vglut2-Cre (Slc17a6tm2(cre)Lowl, Jackson Laboratories, RRID: IMSR_JAX:016963)
or GAD2-Cre (Gad2tm2(cre)Zjh, Jackson Laboratories, RRID: MGI:4418723) mice crossed with Ai14.
Following 2-7 days, a second injection of AAVDJ-EF1a-fDIO-YFP (UNC Vector Core, 1.6 x 1013
GC/mL, originally created by Karl Deisseroth) was injected into SC (60 nl total volume). Animals were
euthanized 4 weeks following the second injection.
3.2.3 Histology
Following desired post-injection survival time, animals were deeply anesthetized and transcardially
perfused with 4% paraformaldehyde. Brains were extracted and post-fixed for 24 hours at 4˚C in 4%
paraformaldehyde and then sliced into 150 µm sections using a vibratome (Leica, VT1000s). The
sections were serially mounted onto glass slides and coverslipped. For some experiments, a
fluorescent Nissl stain was added (Neurotrace 640, ThermoFisher, N21483) to reveal cell body
location and cytoarchitectural information. To label astrocytes, brain sections were blocked with 10%
donkey serum in PBS containing 0.1% Triton-X100 (PBST) for 1 hour, followed by overnight
incubation at 4˚C in PBST with rabbit anti-GFAP antibody (Millipore, 1:1000 dilution, RRID:
AB_2109645). Sections were then washed with PBS 3 times and incubated in PBST with donkey
antirabbit Alexa Fluor 488 conjugated secondary antibody (ThermoFisher, 1:200, RRID: AB_2535792)
for 2 hours at room temperature. To enhance Ai14 tdTomato signal in Vglut2-Cre+ neurons (Figure
3.13B), brain sections were treated as described above with rabbit anti-RFP antibody (Rockland,
1:500, RRID: AB_2209751), followed by donkey anti-rabbit Alexa Fluor 555 conjugated secondary
antibody (ThermoFisher, 1:200, RRID: AB_162543).
3.2.4 Imaging and quantification
All images were generated using a confocal microscope (Olympus FluoView FV1000). To quantify
the total number of cell bodies labeled in structures downstream of V1 or retina, serial sections across
the whole brain were collected and examined. Regions with labeled cells were imaged at 10X
47

magnification across the depth of the tissue (150 μm thickness, 15 μm z-stack interval) over all
sections containing labeling. TdTomato+ cell bodies that co-localized with fluorescent Nissl stain were
manually identified and counted. The total number of tdTomato+ cells were counted across sections
for each structure of interest. To estimate the percentage of tdTomato+ cells in innervated regions,
40X magnification images were taken across multiple sections for SC, LGNv, Str, and PN. A region of
interest of fixed size was defined for each structure (Str = 150 x 200 μm, PN = 100 x 200 μm, and SC
and LGNv = 200 x 300 μm area) and the total number of Tomato+ cells and Nissl+ cells were
quantified within the region (see Figure 3.4). We chose to focus this quantification on restricted local
regions with Tomato labeling, rather than the entire structure, as only a small portion of each target
structure is innervated by a given V1 injection due to the topographic nature of its output. We defined
the central regions of labeling in each structure to capture the majority of labeled neurons for this
estimation. To quantify the percentage of YFP+/Tomato+ GABAergic or glutamatergic cells in SC
(Figure 3.13E), 40X magnification images were collected and a 200 x 300 μm defined region of
interest centered on the region with YFP labeling was assigned to each image. All YFP+/Tomato+
cells were manually counted and compared to the total number of Cre+/Tomato+ cells found within
each region of interest. To quantify the percentage of LP-projecting SC cells labeled anterograde
transneuronally with AAV-Cre injections in V1 (Figure 3.126), 40X magnification images were taken
across all sections of SC containing Tomato+ cells. All CTB+ and CTB+/Tomato+ cells were
quantified manually within the local region of Tomato+ labeling.
3.2.5 Slice preparation and recording
To confirm synaptic connectivity to anterogradely labeled cells in the striatum (Figure 3.6B) and SC
(Figure 3.7), a 1:1 mixture of AAV2/1-hSyn-Cre-WPRE-hGH and AAV2/1-EF1a-DIO-hChR2-eYFP
was injected into V1 of Ai14 mice (80 nl total volume). Following a 4 week post-injection survival time,
acute brain slices containing striatum or SC were prepared. Following urethane anesthesia, the
animal was decapitated and the brain was rapidly removed and immersed in an ice-cold dissection
48

buffer (composition: 60 mM NaCl, 3mM KCl, 1.25 mM NaH2PO4, 25 mM NaHCO3, 115 mM sucrose,
10 mM glucose, 7 mM MgCl2, 0.5 mM CaCl2; saturated with 95% O2 and 5% CO2; pH= 7.4). Brain
slices of 350 μm thickness containing the rostral striatum were cut in a coronal plane using a vibrating
microtome (Leica VT1000s). Slices were allowed to recover for 30 min in a submersion chamber
filled with the warmed (35 °C) ACSF and then to cool gradually to the room temperature until
recording. The spatial expression pattern of ChR2-EYFP in each slice was examined under a
fluorescence microscope before recording. Striatal and SC neurons were visualized with IR-DIC and
fluorescence microscopy (Olympus BX51 WI) for specific targeting of tdTomato+ neurons surrounded
by EYFP+ fluorescent fibers. Patch pipettes (Kimax) with ~4-5 MΩ impedance were used for whole-
cell recordings. Recording pipettes contained: 130 mM K-gluconate, 4 mM KCl, 2 mM NaCl, 10 mM
HEPES, 0.2 mM EGTA, 4 mM ATP, 0.3 mM GTP, and 14 mM phosphocreatine (pH, 7.25;
290mOsm). Signals were recorded with an Axopatch 200B amplifier (Molecular Devices) under
voltage clamp mode at a holding voltage of –70 mV for excitatory currents or 0 mV for inhibitory
currents, filtered at 2 kHz and sampled at 10 kHz. 1 µM tetrodotoxin (TTX) and 1 mM 4-
aminopyridine (4-AP) was added to the external solution for recording only monosynaptic responses
(Petreanu et al. 2009) to blue light stimulation (3-10 ms pulse, 3 mW power, 10-30 trials, delivered via
a mercury Arc lamp gated with an electronic shutter). To test if ChR2 was expressed in connected
SC neurons, glutamate receptor antagonist DNQX (20 µM, Sigma-Aldrich) was added to the bath
following demonstration of LED evoked synaptic responses in patched neurons. LED pulses were
then delivered as before to test for non-synaptic, light evoked currents.
3.2.6 In vivo optogenetic preparation and stimulation
To examine whether SC neuron subgroups mediate different behavioral responses, mice were
implanted with an optical fiber (200 µm diameter, Thorlabs) three weeks after secondary injection of
AAV2/1-EF1a-DIO-hChR2-eYFP in SC, following our previous study (Xiong et al. 2015). Briefly, mice
were anesthetized with isoflurane and mounted into a stereotaxic apparatus. A small hole (~500 µm
49

diameter) was drilled in the skull directly above the targeted region and the optical fiber was lowered
to the desired depth and fixed in place using dental cement. For activating ChR2-expressing neurons
in superficial SC, the fiber was positioned 3.9 mm posterior and 0.6 mm lateral to bregma, and 0.9
mm ventral from the cortical surface. For activating SC neurons in deep layers, the fiber was
positioned as above, but lowered to a depth of 1.6 mm below the cortical surface. For activating the
SC axonal projection to LP, or LP neurons directly, the optical fiber was positioned at 2.4 mm
posterior and 1.6 mm lateral to bregma and 2.1 mm ventral to the cortical surface. Animals were
allowed to recover for 5-7 days prior to behavioral testing. During test sessions, the implanted optical
fiber was connected to a patch cord fiber secured with a plastic sleeve (Thorlabs, 200 µm Core, 0.22
NA (numerical aperture)). The latter fiber was equipped with an integrated rotary joint (Thorlabs) and
was supported from above to allow the animal to move freely in the chamber without being hindered.
To activate ChR2-expressing neurons or axons, optical stimulation was delivered using a blue LED
source (470 nm, 5 mW, Thorlabs) at a rate of 20 Hz (20 ms pulse) for a duration of 5 seconds.
Following testing sessions, animals were euthanized and each brain was sectioned and imaged to
verify the specificity of ChR2 expression and location of the implanted fiber.
3.2.7 Behavioral testing and quantification
Escape behaviors were tested in a box containing two chambers connected by a small opening, as
we previously described (Xiong, et al., 2015). Mice (n = 7 mice for A1-SC, n = 5 mice for V1-SC)
were allowed to acclimate to one chamber for 10 minutes prior to behavioral testing. During this time,
the opening connecting the two chambers was blocked with a removable door. After 10 mins the door
was removed and the mouse was free to explore the adjacent, novel chamber. Following complete
entry into the novel chamber, 5-s 20-Hz blue LED stimulation or 5-s 70 dB SPL white noise was
applied. The animal behavior was recorded with a camera mounted above the box. If the mouse went
back to the home chamber within 5 s of noise or LED stimulation, it was considered as a successful
escape trial. If not, it was a failure trial. Test process was repeated for 4 to 6 times depending on the
50

willingness of the animal to enter the novel chamber after the door removal. Between trials, animals
were allowed to rest for at least 5 mins. Each animal was tested for 2 sessions separated by one day.
Escape rate was calculated as the fraction of total trials for each animal that evoked escape.
Freezing behavior was tested in a single or double chamber box with a camera on top
recording the whole process (n = 5 mice for V1-SC, n = 6 mice for SC-LP). After the animal explored
the single chamber or the novel chamber for 5 minutes, 5-s 20-Hz blue LED stimulation was applied.
Each session contained 4-6 trials, with an interval of at least 5 mins. Each animal was tested for 2
sessions separated by one day. The trial number depended on the activity level of the mouse, and a
session would end if the animal stayed unmoved in the corner for more than 1 min. If the animal
stayed motionless for more than 1.5 s after the onset of stimulation, it was considered as a successful
freezing response (Shang et al., 2015; Tovote et al., 2016; Wei et al., 2015; Wolff et al., 2014).
Freezing time was quantified for each animal as the fraction of time spent freezing during the
optogenetic stimulation (total freezing time/ total LED stimulation time). Freezing rate was calculated
as the fraction of total trials that evoked freezing.
Control subjects for each behavioral test (sham, n = 5 each) received a sham injection of
AAV2/1- CAG-FLEX-GFP, and were then implanted with an optical fiber and tested with the same
stimulation parameters. In this study, the optogenetic stimulation was unilateral, but was sufficient for
inducing defense behaviors such as flight and freezing, which is consistent with previous studies
(Comoli et al., 2012; Dean et al., 1988; Liang et al., 2015; Sahibzada et al, 1986). For escape
responses, we did not find a correlation between the direction in which the body turned and the side of
SC that was stimulated (data not shown).
3.2.8 Statistical Methods
Samples were first determined to have normal distribution using the Shapiro-Wilk test. One-way
ANOVA and post hoc Tukey’s multiple comparison was used to test significance between samples in
51

Figure 3.5C. For behavioral results (Figures 3.9, 3.11), significance was determined using t-test. Data
were presented as mean ± s.d.

3.3 Results
3.3.1 Anterograde transneuronal spread of AAV1-Cre
To test for any anterograde transneuronal transport of AAV, we injected AAV2/1-hSyn-Cre into the
primary visual cortex (V1) of Ai14 (Cre-dependent tdTomato reporter) mice (Madisen et al. 2010).
Following a 4 week post-injection survival time, we observed tdTomato-expressing cell bodies
intermingled with terminal fields of tdTomato-labeled axons in all regions known to be directly targeted
by V1, including the superior colliculus (SC), striatum (Str), ventral lateral geniculate nucleus (LGNv),
and the pontine nucleus (PN) (Figure 3.2A). As these four regions do not project back to V1
(Simmons et al., 1982; Oh et al., 2014; Zingg et al., 2014), the presence of tdTomato+ cell bodies in
these areas may only be explained by transneuronal spread of the Cre virus from V1 axons to
neurons in their targeted structures. AAV1-Cre was also able to retrogradely spread to presynaptic
neurons, although with a low efficiency, resulting in Cre-dependent transgene expression in
presynaptic neurons (Figure 3.3, also see Tervo, et al., 2016; Rothermel, et al., 2013; Aschauer et al.,
2013). Therefore, in brain regions that are reciprocally connected with V1 (e.g. dorsal lateral
geniculate nucleus, LGNd), tdTomato labeling of cell bodies can result from both anterograde and
retrograde transneuronal transport of the Cre virus. These regions are not considered in the current
study because of the ambiguity.
To further confirm the anterograde transneuronal spread of AAV1, we injected AAV1-hSyn-Cre
unilaterally into the retina of Ai14 mice, and examined tdTomato expression in brain regions known to
receive retinal ganglion cell (RGC) inputs. In all regions examined, including the SC, LGNd, LGNv,
olivary pretectal nucleus (OP) and the suprachiasmatic nucleus (SCN), we observed tdTomato+ cell
bodies (Figure 3.2B). Again, since none of the above central structures project back to the retina
52

(Morin and Studholme, 2014), the tdTomato labeling of cell bodies can only be a result of anterograde
transneuronal spread of the virus from RGC axons to target cells.

Figure 3.2. Anterograde transneuronal transport of AAV1-Cre.
(A) Left, tdTomato (red) expression in the injection site following injection of AAV1-hSyn-Cre into V1 of an Ai14
reporter mouse. Blue is Nissl staining. Right four panels, tdTomato fluorescence in regions downstream of V1
(SC, striatum, LGNv, pontine nucleus). High-magnification images (bottom) show tdTomato-labeled cell bodies
53

in the region indicated by the white box. Scale bars: 500 µm, left panel; 250 µm, right top panels; 25 µm, right
bottom panels. The same scales apply to (B) and (C) below correspondingly.
(B) Anterograde transneuronal labeling in the SC, LGNd, LGNv, olivary pretectal nucleus (OP), and
suprachiasmatic nucleus (SCN) following injection of AAV1-hSyn-Cre into the contralateral retina. High-
magnification images (bottom panels) show tdTomato-labeled cell bodies in boxed regions.
(C) Similar experiment as in (A) except that AAV1-CAG-GFP was injected. Note that no GFP-labeled cell
bodies were found in regions downstream of V1 (4 week post-injection survival time).
(D) Quantification of total number of tdTomato-labeled cells in selected structures downstream of V1 following
injection of AAV1-hSyn-Cre (4 week post-injection survival time, n = 4 mice, Error bar = SD).
(E) Estimated percentage of tdTomato-labeled cells as compared with the total number of cells within a defined
local region in each downstream structure (D) that includes the majority of transneuronally labeled neurons (see
Figure 3.4 and Experimental Procedures). N = 4. Error bar = SD.
(F) Quantification of total number of labeled cells found in structures downstream of the contralateral retina
following injection of AAV1-hSyn-Cre (4 week post-injection survival, n = 4 mice, Error bar = SD).
(G) Left, injection of AAV1-hSyn-Cre in V1 of Ai14 reporter mice labels neurons in SC (1
st
order connection),
which in turn project strongly to PBG (2
nd
order connection). Middle and right, dense tdTomato+ axons (red)
were observed in PBG, but no tdTomato+ cell bodies (right panels) after 3 month post-injection survival time.
Scale bar: 250 µm, middle panel; 25 µm, right panels.
(H) Quantification of number of labeled cells in the first and second order downstream regions in four different
mice. No second order spread was observed after 3 month post-injection survival time. Abbreviations: GP,
globus pallidus; LA, lateral amygdala; LP, lateral posterior nucleus of thalamus; PBG, parabigeminal nucleus;
SCs, superior colliculus, superficial layer; SCd, superior colliculus, deep layer; SNr, substantia nigra reticulata.

Figure 3.3. Retrograde transport of AAV1.
(A) Schematic diagram of AAV1-hSyn-Cre injection in SC of Ai14 tdTomato mouse.
(B) Raw images showing injection site in SC (left panel, red) and neurons in V1 (right panel). Numerous
retrogradely labeled cells were observed in layer 5 (L5) of visual cortex.
(C) Schematic diagram of injection. AAV1-CAG-GFP was injected into SC of wildtype mice and V1 was
examined for retrogradely labeled cell bodies.
(D) Raw images showing injection site in SC (left panel, green) and V1 (right panel). No GFP+ cell bodies were
observed in V1. Scale bars, 1 mm (left panel) and 250 µm (right panel).

54

It is interesting to note that anterograde transneuronal transport of AAV has not been widely
reported, despite the extensive use of AAVs expressing GFP or channelrhodopsin2 (ChR2) (Harris et
al., 2012; Aschauer et al., 2013; Yizhar et al., 2011; Oh et al., 2014). Consistent with previous results,
at 4 weeks following injections of AAV1-CAG-GFP into V1, we did not find any GFP+ cell bodies in
regions directly targeted by V1, despite the presence of strongly labeled axonal terminals (Figure
3.2C). This result indicates that the efficiency of the transneuronal spread is relatively low. That is,
only a small number of viral particles might be transported transneuronally from the initial host cell,
leading to extremely weak GFP expression in the downstream neurons, which may be below the
detection threshold. Thus, in any application of AAV-mediated anterograde transneuronal tagging, an
amplification step is required, such as utilizing Cre to unlock robust transgene expression.
We counted the total number of anterogradely labeled neurons in each of the target structures of
interest. For V1 injections, SC contained the largest number of labeled neurons, followed by LGNv,
striatum (Str) and the pontine nucleus (PN) (Figure 3.2D). The number of tdTomato-labeled neurons
was compared with that of Nissl-stained cell somata within the same local area (Figure 3.4). We
Figure 3.4. Quantification of labeling density in targets downstream of V1.
(A) Cre+/Tom+ cells (red) and Nissl stained cells (blue) were quantified within a fixed 200 x 300 µm sample
space (white grid) centered on regions with Tom+ labeling across multiple sections of SC to generate an
estimate of average density of cell labeling. One example image is shown.
(B) An example image for quantifying density of Tom+ cells within a 150 x 200 µm sample region in striatum
(Str).
(C) Example image of ventral lateral geniculate nucleus (LGNv) quantification using a 200 x 300 µm sample
region.
(D) Example image of pontine nucleus (PN) quantification using a 100 x 200 µm sample region.
55

found that in the SC, LGNv, and PN, the percentage of anterogradely labeled cells reached about
40% (Figure 3.2E), suggesting that the number of neurons that receive the transneuronally
transported Cre virus is relatively high. For retina injections, SC also contained the largest number of
anterogradely labeled neurons, followed by LGNd and LGNv (Figure 3.2F). The OP, intergeniculate
leaflet (IGL), and SCN also contained a small number of labeled neurons.
Outside of the regions known to be directly innervated by V1 or retinal axons, we did not
observe any tdTomato+ cell bodies, suggesting that the virus does not further spread to second-order
downstream structures. An example of this is shown in Figure 3.2G. While SC neurons were labeled
by tdTomato expression unlocked by anterogradely transported AAV1-Cre from V1, in the
parabigeminal nucleus (PBG) which is directly targeted by SC (Huerta and Harting, 1984; Shang, et
al., 2015), no tdTomato+ cell bodies were found, even at 3 months post injection (Figure 3.2G, right).
In addition, no tdTomato-labeled cell bodies were found in several other second-order nuclei
examined (Figure 3.2H).

3.3.2 Dependence on viral type, serotype and other factors
We next asked whether the observed anterograde transneuronal spread is a general property
common to other viruses. To test this, we first injected Cre-expressing canine adenovirus (CAV2-
CMV-Cre; Soudais et al., 2001) in a similar way as AAV1. At 4 weeks following V1 injections in Ai14
mice, no tdTomato+ cell bodies were observed in any regions exclusively downstream of V1 (Figure
3.5A). This suggests that anterograde transneuronal transport might be a unique property of AAV
(and possibly some other viruses), due to specific interactions between the virus and host cells. To
test if this property varies with serotype, we compared transneuronal labeling with injections of AAV1,
AAV6, AAV9, AAV5, and AAV8 encoding CMV-driven Cre into V1 (Figure 3.5B-C). AAV6, AAV5 and
AAV8 did not exhibit anterograde transneuronal transport, unlike AAV1 and AAV9 (Figure 3.5B-C),
despite similarly strong local transduction of neurons at the injection site (Figure 3.5B). These results
further demonstrate that the observed anterograde transneuronal transport of AAV may reflect
56

Figure 3.5. Serotype specificity of transneuronal transport.
(A) CAV2-CMV-Cre injection in V1 (upper panel). No anterograde transneuronal labeling was observed in
regions downstream of V1 such as the SC (middle and bottom panels). High-magnification images (bottom) of
the boxed region are shown (4 week post-injection survival).
(B) Comparison of AAV1, AAV6, and AAV9. Injection into V1 resulted in robust anterograde transneuronal
labeling in SC with AAV1 (left) and AAV9 (right), but not AAV6 (middle) (4 week post-injection survival, 60 nl
injection each). Scale bars: 500 µm, top panels; 250 µm, middle panels; 25 µm, bottom panels. The same
scales also apply to (A) correspondingly.
(C) Quantification of total number of cells labeled in SC for different Cre-expressing viruses injected into V1 of
Ai14 mice (4 week post-injection survival, 60 nl injection, n = 4 mice each). Error bar = SD. **, p < 0.01,
different from all other groups; *, p < 0.05, different from AAV5, AAV6, AAV8 and CAV2 groups (one-way
ANOVA and post hoc test).
(D) Dependence of transneuronal labeling on viral concentration. Number of labeled cells in SC following
injection of undiluted AAV1-hSyn-Cre in V1, 1:10 dilution, or 1:20 dilution (4 week post-injection survival, 60 nl
injection, n = 4 each). Error bar = SD.
(E) Quantification of transneuronal labeling in SC following different post-injection survival times: 2 days, 5 days,
2 weeks, 4 weeks, and 3 months (60 nl injections of AAV1-hSyn-Cre in V1, n = 4 each). Error bar = SD.
57

specific interactions between viral capsid proteins and host cells. It is worth noting that due to the
higher titer of the injected AAV9 compared with AAV1 (see Experimental Procedures), the actual
efficiency of AAV9 might be considerably lower than AAV1. Moreover, as both CAV2 and AAV6
drove high levels of Cre expression at the injection site (Figure 3.5A, 3.5B), yet no neurons were
labeled in downstream targets, this suggests that Cre protein itself is incapable of transneuronal
spread. The observed transneuronal labeling thus results from the spread of the Cre-encoding virus,
rather than the Cre protein per se.
Relatively high titers of the virus were needed for the anterograde transneuronal spread, as
the number of labeled neurons rapidly reduced with decreasing titers of AAV1 (Figure 3.5D). The
anterograde transneuronal labeling could be observed as early as 5 days post injection, was fully
expressed at 2 weeks post injection and then remained stable for as long as 3 months (Figure 3.5E).

3.3.3 Synaptic specificity of anterograde transneuronal spread
Does AAV spread specifically to synaptically connected target neurons or is it just released from
axons locally capable of transducing neighboring cells? To address this issue, we first examined
whether AAV is capable of transducing glial cells surrounding axonal fibers of passage, which might
be expected if the virus was released non-specifically along the axon. We injected AAV1-CMV-Cre,
which is capable of expressing Cre in both neurons and glia, in V1 of Ai14 mice. In the corpus
callosum, which is comprised almost entirely of glia and axons (Sturrock, 1976; Aboitiz & Montiel,
2003), no tdTomato+ cell bodies were identified (Figure 3.6A, top panels), whereas strong
transneuronal cell body labeling still occurred in the target structures of V1 (e.g. pontine nucleus,
Figure 3.6A, bottom panels). These results suggest that the virus does not “leak” from fibers of
passage. Additionally, in the sections containing transneuronal labeling, we co-stained the tissue with
a marker for astrocytes, GFAP (Xu et al., 1999). In all sections examined, we did not observe co-
labeling of tdTomato+ cell bodies with GFAP antibody (e.g. Figure 3.6A, bottom panels). Given that
58

astrocytes can participate in a tripartite synaptic structure (Perea et al., 2009), the failure of their
transduction by the virus suggests that the viral spread is highly restricted to neuronal structures in
closest proximity to presynaptic terminals.

Figure 3.6. Synaptic specificity of anterograde viral spread.
(A) TdTomato expression (red) and GFAP labeling (green) in the corpus callosum (upper) and pontine nucleus
(lower) following injection of AAV1-CMV-Cre into V1 of an Ai14 mouse. High-magnification images (right
panels) show that Nissl stained cell bodies (blue) within the corpus callosum were negative for tdTomato
(upper), and that tdTomato-positive cell bodies within PN were negative for GFAP staining (lower). Scale bars:
250 µm, left panels; 25 µm, right panels.
(B) Slice recording from transneuronally labeled neurons in the striatum (red) following co-injection of AAV1-
hSyn-Cre and AAV1-EF1a-DIO-ChR2-YFP into V1 (left panel). Top right image shows ChR2-expressing axons
(green) surrounding tdTomato-labeled striatal neurons. Scale: 25 µm. Middle panel, average LED-evoked
excitatory (-70 mV) and inhibitory (0 mV) currents in an example tdTomato+ striatal neuron before and after
perfusing in TTX and 4AP. LED stimulation is marked by a blue bar. Right bottom, a summary of amplitudes of
average monosynaptic excitatory currents evoked by LED in 9 recorded striatal cells.

We next asked whether transneuronally labeled cells received functional synaptic input from
neurons within the starter population. To test this, we injected a 1:1 mixture of AAV1-hSyn-Cre and a
Cre-dependent ChR2 virus, AAV1-EF1α-DIO-ChR2-EYFP, into V1 of Ai14 mice (Figure 3.6B, left
panel). Only neurons co-transduced with the two viruses could express ChR2. In slice preparations,
we examined monosynaptic connectivity between ChR2-labeled V1 axons and tdTomato+ striatal
neurons with whole-cell recording in the presence of TTX and 4-AP (Petreanu et al., 2009). In all
59

recorded cells, blue LED pulses evoked excitatory synaptic currents which remained in the presence
of TTX and 4-AP, while the evoked inhibitory currents were abolished by TTX and 4-AP (Figure 3.6B,
middle and right panels). Since the ChR2-labeled V1 axons contained the Cre virus, the functional
coupling between the former and the transneuronally labeled striatal neurons provides strong
evidence for transsynaptic spread of the Cre virus to synaptically connected target neurons. To
further support this notion, our control experiments (Castro-Alamancos and Favero, 2016)
demonstrated that the LED-evoked current was only synaptic in nature, as antagonists of glutamate
receptors completely blocked the current (Figure 3.7).

Figure 3.7. Pre-synaptic specificity of LED evoked currents recorded in slice preparation.
(A) Schematic diagram of injection and recording strategy. AAV1-DIO-ChR2-YFP and AAV1-hSyn-Cre were
co-injected into V1 of Ai14 mice to label downstream neurons in SC (red) and SC-projecting axons with ChR2
(green). Tom+ cells in SC were then targeted for recording in slice preparation. Raw images of the injection site
and labeling in SC are shown. No ChR2-YFP+ cell bodies were observed in SC (bottom panel). Blue, Nissl
stain. Scale bars, 500 µm, top left panel, 25 µm, bottom right panel.
(B) Averaged LED evoked excitatory synaptic responses recorded from 5 Tom+ SC neurons in the presence of
TTX and 4-AP. Responses were eliminated after adding glutamate receptor antagonist DNQX, suggesting no
ChR2 expression within the recorded cells. Blue bar indicates 5 ms LED pulse. Graph below summarizes
amplitudes of LED evoked responses before and after addition of DNQX. Data points for the same cell are
connected with a line.

60

3.3.4 Mapping outputs of input-defined SC neuron subpopulations
The anterograde transsynaptic property of AAV-Cre allows its application in mapping the axonal
projections of neuronal populations with a specific presynaptic input source (i.e. input-defined). To
test this idea, we used a two-step viral injection procedure and selected SC as the target structure, as
it receives inputs from the retina as well as a variety of cortical areas, which terminate within distinct
layers of SC (Huerta and Harting, 1984; Comoli et al., 2012). We first injected AAV1-hSyn-Cre in V1
of Ai14 mice, to label SC neurons that receive V1 input. Several days later, a second injection of
AAV1-CAG-FLEX-GFP was made in SC (Figure 3.8A, left panel). After 4 weeks post-injection
survival time, we observed robust GFP expression in tdTomato+ neurons specifically within the
superficial gray layer of SC (i.e. SC-sg; Figure 3.8A, middle panel). 100% of GFP+ neurons were also
tdTomato+, consistent with a result of co-transduction of Cre-dependent GFP and transneuronally
transported Cre viruses. We examined the long-range axonal projections of this specific set of SC-sg
neurons to various target structures. The most prominent axonal labeling was found in the lateral
posterior nucleus (LP) of thalamus, with additional labeling seen in pretectal regions and PBG (Figure
3.8A, right panel). Next, we examined SC neurons receiving RGC projections, which also target the
superficial layer of SC. As with V1, paired injections in the retina and SC yielded a group of GFP+
neurons specifically within SC-sg (Figure 3.8B). Their axonal projection targets were found to closely
match those of the V1 input-defined population (compare Figure 3.8A and Figure 3.8B), suggesting
that the retina and V1 may target a similar group of SC neurons.
We further examined SC neuron groups specifically receiving inputs from the primary auditory
cortex (A1) and primary motor cortex (M1), respectively. Following paired injections into A1 and SC,
we observed numerous GFP+ cells within the medial aspect of deep grey layer (SC-dg), with a few
scattered in overlying intermediate grey layer (SC-ig) (Figure 3.8C). This labeling pattern is consistent
with the known distribution of A1 projections to SC (Xiong et al., 2015). Axonal outputs of A1 input-
defined SC neurons were apparently different from those of superficial SC neurons, as prominent
61

projections were seen in the PBG, periaqueductal gray (PAG), cuneiform nucleus (CUN), rostral
pontine reticular nucleus (PRNr), and among other regions (Figure 3.8C, E). Only very sparse input
to LP was observed as compared with the strong projection to LP from superficial SC neurons. In
comparison, paired injections in M1 and SC labeled a population of neurons restricted to the lateral
aspect of SC-ig (Figure 3.8D). Axonal outputs from this group of SC neurons differed further from
those defined by A1 and V1/retina inputs (Figure 3.8D).
Figure 3.8. Mapping of axonal outputs of input-defined neuronal populations in SC.
(A) Left, AAV1-hSyn-Cre was injected into V1 of Ai14 mice, followed by a second injection of AAV1-CAG-FLEX-
GFP into SC. Middle, GFP-labeled neurons in SC-sg were also tdTomato+. Right four panels, GFP-labeled
axons in various regions (LP, pretectal area, PBG, cuneiform nucleus (CUN)) downstream of SC. High-
magnification images (bottom) reveal ramified axons and their terminal and bouton structures. Blue, Nissl
62

staining. Scale bars: 250 µm, middle top panel; 500 µm, right top panels; 25 µm, bottom panels. The scales
also apply to (B), (C), (D) correspondingly.
(B) Axonal outputs of SC-sg neurons that receive input from the contralateral retina. Data are displayed in a
similar way as in (A).
(C) Axonal outputs of SC neurons that receive input from A1, which are located mainly in SC-dg and sparsely in
SC-ig.
(D) Axonal outputs of SC neurons that receive input from M1, which are located mainly in the lateral aspect of
SC-ig.
(E) Summary of observed target regions for SC neuron subpopulations receiving input from V1/retina (blue),
from A1 (red), and from M1 (yellow) respectively. Abbreviations: CL, central lateral nucleus of thalamus; PCN,
paracentral nucleus; VM, ventral medial nucleus of thalamus; CM, central medial nucleus of thalamus; APN,
anterior pretectal nucleus; PF, parafascicular nucleus; SPFm and SPFp, subparafascicular nucleus,
magnocellular and parvicellular; ZI, zona incerta; MRN, midbrain reticular nucleus; PRN, pontine reticular
nucleus; TRN, tegmental reticular nucleus; ICd and ICe, inferior colliculus, dorsal and external; PARN,
parvicellular reticular nucleus; GRN, gigantocellular reticular nucleus; IO inferior olivary complex.

To summarize the axonal output profiles of input-defined SC neuron subpopulations, we
examined all brain sections containing GFP+ axons and plotted those regions containing observable
synaptic boutons on corresponding atlas sections (Figure 3.8E). SC neurons receiving V1 and retinal
inputs exhibited very similar output profiles (Figure 3.8A-B; Figure 3.8E, blue), while those defined by
A1 and M1 inputs showed profiles distinct from each other and from the V1/retina-defined population
(Figure 3.8C-D; Figure 3.8E, red and yellow respectively). For example, unlike V1- and A1-defined
populations, M1-defined SC neurons exhibited prominent projections to some contralateral targets,
especially in the posterior part of the brainstem, such as the tegmental reticular nucleus (TRN),
gigantocellular reticular nucleus (GRN), and inferior olive (IO) (Figure 3.8D; Figure 3.8E). Together
these results demonstrate that AAV-Cre mediated transsynaptic labeling can be utilized to map
specific classes of projection neurons within a given brain region, each with distinct input and output
patterns.

63

3.3.5 Distinct behavioral functions for input-defined SC neuron subpopulations
The superior colliculus integrates diverse sensory and motor information, and has been implicated in
controlling orienting behaviors as well as innate defense behaviors, such as freezing and escape
(Schenberg et al., 2005; Sahibzada, et al., 1986; Dean et al., 1988; Liang et al., 2015; Wei et al.,
2015; Shang et al., 2015). Since SC neuron subpopulations receiving V1 and A1 inputs have different
downstream target profiles, we speculated that each might participate in driving a distinct SC-
mediated behavior. To test this, we performed paired injections of AAV1-Cre in either A1 or V1 and
AAV1-EF1α-DIO-ChR2-EYFP in SC to enable Cre-dependent expression of ChR2 in either deep or
superficial layer SC neurons, respectively (Figure 3.9A, Figure 3.11A). We could then activate the SC
neurons in awake, freely moving mice using pulses of blue LED light (at 20 Hz for 5 sec) delivered
through an implanted optical fiber (see Experimental Procedures). We tested both freezing and
escape responses using a two-chamber set up, in which the mouse was acclimated to a “home”
chamber on one side for 10 minutes, upon which a door was then removed to allow its exploration of
the adjoining, “novel” chamber (Figure 3.9B). As the mouse was exploring the novel chamber, SC
was optically activated and subsequent escape to home chamber or freezing was monitored (Figure
3.9B). Mice expressing ChR2 in A1-recipient, deep layer SC neurons demonstrated a robust escape
response following LED light activation (Figure 3.9C-D). Such LED-induced escape behavior was not
observed in sham mice receiving injection of only GFP expressing virus (Figure 3.9C-D). The effect
of optically activating A1-recipient SC neurons was similar as applying a loud noise sound (5 s, 70 dB
sound pressure level), presentation of which alone robustly drove an escape response in most trials
(Figure 3.9C-D, noise), consistent with our previous study (Xiong et al., 2015). No freezing response
was observed for these mice on any trials (Figure 3.9E). In our control slice recording experiments,
we demonstrated that in these injected mice, A1 corticofugal neurons transduced with AAV1-Cre did
not express ChR2 themselves (Figure 3.10). Therefore, the observed escape behavior could not be
attributed to activation of some other A1 targets.
64

Figure 3.9. A1-recipient SC neurons drive an innate escape behavior.
(A) Schematic illustration of paired injections in A1 and SC, as well as LED illumination applied (A1-SC). Right
panels, images of injection sites (red for tdTomato; green for ChR2) in an example animal. Scale bar: 500 µm.
(B) Schematic illustration of two-chamber behavior setup for testing freezing or escape.
(C) Movement tracking for an example A1-SC mouse under LED stimulation (left), A1-SC mouse under noise
stimulation (middle) and sham mouse under LED stimulation (right) in the novel chamber during 5 s LED
activation or 5 s noise stimulation. Each curve represents one trial. Blue dot indicates the starting location at
the initiation of LED or noise stimulus, and red dot indicates the location at the end of the stimulus. Red dot
beyond the novel chamber boundary indicates that the animal has returned to the adjacent home chamber
within 5 sec (bottom left). For “sham”, AAV1-FLEX-GFP was injected in SC.
(D) Summary of percentage of trials that induced escape behavior (n = 7 mice for A1-SC group, n = 5 mice for
sham). Error bar = SD. ***, p < 0.001, t test.
(E) Percentage of trials that induced freezing behavior. Error bar = SD.

65

Figure 3.10. Lack of retrograde ChR2 expression in cortex in animals used for behavior studies.
(A) Schematic diagram showing injection and recording strategy. To test if ChR2 is retrogradely expressed in
Cre+ cortical neurons, AAV1-hSyn-Cre was injected into A1 of Ai14 mice. Anterograde transneuronally labeled
cells in SC (red) were then targeted for Cre-dependent expression of ChR2 with a second injection of AAV1-
EF1a-DIO-ChR2-YFP. Following 4 weeks expression time, Cre+/Tom+ neurons in layer 5 (L5) of A1 were
targeted for slice recording.
(B) Raw images of a representative injection. No ChR2-YFP+ cell bodies were observed in A1 (top right panel).
Strong expression of ChR2 was observed in numerous Cre+/Tom+ cell bodies in A1-recipient SC neurons
(bottom panels). Scale bars, 500 µm, top left panel and bottom left panel, 25 µm, bottom right panel.
(C) Superimposed average current traces for 15 Tom+ L5 neurons in A1. Blue bar indicates 10 ms LED pulse.
No light evoked currents were observed in any of the recorded cells.
(D) Summary of LED evoked response amplitude for all 15 cells.

Mice expressing ChR2 in V1-recipient, superficial layer SC neurons (Figure 3.11A), on the other
hand, all demonstrated freezing response following LED light activation (Figure 3.11B-C), which was
not observed for sham mice (Figure 3.11B-C). None of the mice exhibited escape behavior following
LED activation (Figure 3.11G). Our results thus reveal distinct functional roles of superficial versus
deep layer SC neurons in controlling two different defense behaviors.
Our tracing result has demonstrated that LP is a major axonal target of V1-recipient SC neurons
(Figure 3.8A). In addition, combined anterograde transneuronal labeling and retrograde labeling
confirmed that a large fraction of V1-recipient SC neurons project to LP (Figure 3.126). We thus
explored whether the freezing behavior evoked by SC-sg activation might be mediated through this
structure. To test this, we first optogenetically activated the axon terminals of SC-sg neurons in LP
66

(Figure 3.11D) in freely behaving mice, which resulted in similar, though somewhat weaker
expression of freezing (Figure 3.11E-F, V1-SC-LP), implicating LP’s role in driving SC-mediated
freezing behavior. In contrast, optically activating SC axon terminals in PBG did not produce any
freezing behavior (Figure 3.11E-F, V1-SC-PBG), suggesting that PBG does not play a role in this
behavior. None of these activations induced escape behavior (Figure 3.11G).

Figure 3.11. V1-recipent SC neurons drive freezing behavior.
(A) Paired injections labeling SC neurons receiving V1 input. LED illumination was applied to cell bodies in SC
(V1-SC). Scale bar: 500 μm.
(B) Percentage of time spent freezing within the time window of LED illumination (n = 5 mice for each group).
Error bar = SD. ***, p < 0.001, t test.
(C) Percentage of trials that induced freezing. ***, p < 0.001, t test.
(D) LED illumination was applied to ChR2+ SC axon terminals in either LP (V1-SC-LP) or PBG (V1-SC-PBG).
Right, images showing ChR2 labeled SC axons in LP or PBG. Scale bar: 250 μm.
67

(E) Percentage of time spent freezing within the time window of LED illumination (n = 5 mice for each group).
***, p < 0.001, t test.
(F) Percentage of trials that induced freezing. ***, p < 0.001, t test.
(G) Percentage of trials that induced escape behavior.
(H) Paired injections labeling LP neurons that receive input from SC. LED illumination was applied to LP (SC-
LP). Right panel, images showing injection sites in SC and LP. Scale bar: 500 μm.
(I) Percentage of time spent freezing within the time window of LED illumination (n = 6 mice for SC-LP, n = 5
mice for sham). Error bar = SD. **, p < 0.05, t test.
(J) Percentage of trials that induced freezing behavior. Error bar = SD. ***, p < 0.001, t test.

Figure 3.12. Efficiency of anterograde transneuronal labeling within the V1-SC-LP pathway.
(A) Schematic diagram of injections. AAV1-hSyn-Cre was injected into V1 in Ai14 tdTomato mice and
fluorescently conjugated cholera toxin subunit b (CTB-488, green) was injected into LP to retrogradely label cell
bodies in SC that project to LP. Numerous LP-projecting cell bodies were observed in the deepest part of
superficial SC (middle and right panel, green) co-mingled with Cre+/Tom+ V1-recipient SC neurons (red). At
40X magnification, extensive co-labeling was observed. Scale bars, 250 µm, left panel, 25 µm, bottom right
panel.
(B) Quantification of CTB co-labeled cells within the region of anterograde transneuronal labeling. An average
of 66% of CTB+ cells were co-labeled with tdTomato (n =3 mice).

With terminal activation, it is possible however that nonspecific antidromic stimulation of collateral
targets could occur, resulting in activation of undesired SC targets. Moreover, LED light from the optic
fiber might also activate ChR2-labeled axons passing through the LP to other targets. Both scenarios
would confound interpretation of results. To overcome these limitations, we used AAV-Cre
transsynaptic labeling to express ChR2 specifically in LP neurons that receive input from SC (Figure
68

3.11H). To achieve this, AAV1-Cre was injected into SC and AAV1-DIO-ChR2 was injected into LP,
which has unidirectional connectivity with SC (Huerta and Harting, 1984, Comoli et al., 2012). ChR2-
expressing LP neurons were then selectively activated in freely moving animals. All mice
demonstrated freezing behavior comparable to that elicited by activation of SC projections in LP
(Figure 3.11I-J). This result provides strong support for the structure of LP as a downstream mediator
of the freezing response generated by activation of superficial SC neurons.

3.3.6 Anterograde transsynaptic labeling in a cell-type specific manner
In the above experiments, the cell type of transsynaptically labeled postsynaptic neurons could not be
selected. To achieve the ability to select postsynaptic cell types, we developed an intersectional
approach which depends on both Cre and flippase (Flp) recombinases (Fenno et al., 2014). A new
AAV1 vector expressing Cre-dependent Flp (AAV1-EF1a-DIO-Flp) was made for this purpose. We
injected this new virus into V1 of Vglut2-Cre or GAD2-Cre mice, and Flp-dependent YFP virus
(AAVDJ-fDIO-YFP) into SC of the same animal. Only in Cre-expressing SC neurons can Flp and thus
Flp-dependent YFP be expressed (Figure 3.13A). YFP-labeled SC neurons were found in both
Vglut2-Cre and GAD2-Cre mice (Figure 3.13B-C), indicating that both glutamatergic and GABAergic
SC neurons can receive direct V1 input. YFP-labeled axons were found in LP (target of SC) in
Vglut2-Cre but not GAD2-Cre mice (Figure 3.13D). Moreover, in the injected GAD2-Cre mice, YFP-
labeled axons were not found in any structures outside SC (data not shown). These observations
demonstrate that unlike glutamatergic neurons, GABAergic SC neurons are local interneurons without
long-range projections. Quantification of the number of YFP-labeled SC neurons in comparison with
tdTomato+ SC neurons within the same local area revealed that about 25% of tdTomato+
glutamatergic neurons were YFP+, while about 12% of tdTomato+ GABAergic neurons were YFP+
(Figure 3.13E). As tdTomato+ SC neurons might not all receive V1 input, these numbers gave an
estimation of the lower bound of transsynaptic labeling efficiency for these two types of SC neurons
69

respectively. Together, our results demonstrate that these AAV-based transsynaptic approaches
have the capacity to specifically and robustly label diverse populations of neurons downstream of a
given brain structure.

Figure 3.13. Cell-type specific anterograde transneuronal labeling.
(A) Strategy for labeling glutamatergic (Vglut2-Cre+) or GABAergic (GAD2-Cre+) subpopulations of SC neurons
that receive input from V1.
(B) Selective labeling of V1-recipient glutamatergic neurons in superficial layers of SC. Anterograde
transneuronal transport of AAV1-EF1a-DIO-Flp enables Cre-dependent expression of Flp in glutamateric
neurons (Ai14 tdTomato+) receiving input from V1. A second injection of AAVDJ-fDIO-YFP into SC enables
Flp-dependent expression of YFP specifically in V1-recipient glutamatergic neurons (green), filling soma,
70

dendrites, and axons. YFP+ neurons co-localize with Ai14 tdTomato expression in Vglut2-Cre+ neurons
(bottom panels, enhanced with anti-RFP immunostaining).
(C) Selective labeling of V1-recipient GABAergic neurons using same strategy as in (B) but with injections in a
GAD2-Cre mouse. Scale: 250 µm, top panel; 25 µm bottom panel. The scales also apply to (B)
correspondingly.
(D) Long range axonal projection to LP from glutamatergic, but not GABAergic, V1-recipient SC neurons. Scale:
250 µm.
(E) Percentage of YFP+/Tomato+ cells out of total Tomato+ cells quantified for local regions expressing YFP in
Vglut2-Cre and GAD2-Cre mice (4 week post-injection survival, 60 nl injections, n = 4 each). Error bar = SD.

3.4 Discussion
Significant progress has been made in developing viral tools for revealing the inputs to a given brain
region or cell type (Callaway and Luo, 2015; Junyent et al., 2015; Oyibo et al., 2014; Defalco et al.,
2001; Gradinaru et al., 2010; Schwarz et al., 2015; Ugolini, 2010). However, creation of a
complementary set of tools for labeling transsynaptic output remains an active area of investigation.
Here we show that AAV can be transported anterogradely and can transsynaptically transduce
neuronal populations in a wide variety of brain regions postsynaptic to a given injection site. When
combined with a conditional expression strategy (i.e. two-step viral injection approach), this
transsynaptic transport property enabled us to reveal the outputs of distinct subpopulations of SC
neurons that receive different corticocollicular inputs and demonstrate their unique functional roles in
driving different types of defense behavior.
AAV vectors expressing various fluorescence reporter genes have been widely used for
anterograde circuit mapping. However, anterograde transneuronal spread of this virus has remained
controversial (Salegio et al., 2013; Hutson et al., 2015; Harris et al., 2012; Oh et al., 2014). Indeed, in
this study, AAV1-CAG-GFP failed to result in any GFP+ cell bodies in regions postsynaptic to the
injection site (Figure 3.2C). This result is consistent with the experimental data provided by the online
database of the Allen Institute, where all neuronal projections were anterogradely traced with AAV1,
but with no evidence of transneuronal transduction. Only when AAV1-Cre was injected into Ai14
mice, or was combined with a second virus injection to conditionally express a Cre-dependent
71

fluorescence reporter, did we see robust labeling of neurons downstream of the injection location. In
this case, the lack of transneuronal cell body labeling with AAV1-CAG-GFP injections may be
attributed to the possibility that only a small number of viral particles escape the initial host cell and
enter downstream neurons, resulting in only extremely weak fluorescence expression which is below
the detection threshold for confocal imaging. With AAV1-hSyn-Cre, however, only a small amount of
Cre supplied by a small number of viral particles may be sufficient to unlock robust tdTomato
expression in Ai14 mice or GFP expression supplied by a second Cre-dependent virus, thus revealing
an otherwise hidden capacity of AAV for transneuronal spread.
Previous studies have examined the anterograde transneuronal properties of other viruses
including vesicular stomatitis virus (VSV-G) and herpes simplex virus (HSV-1) (Breier et al, 2011; Lo
et al, 2011). Though promising, these viruses suffer from toxicity issues which limit their use in
physiological studies, and their uncontrollable spread to higher order neurons complicates
interpretation of results from circuit mapping studies. AAV, therefore, offers several advantages over
those viruses currently available for such a purpose. Namely, AAV has a well characterized lack of
toxicity, which allows for the long term, robust expression of viral transgenes necessary for axonal
mapping and functional manipulations. Additionally, as we demonstrate in this study, AAV appears
not to spread beyond the first-order downstream neurons, thus facilitating a more conclusive
interpretation of anatomical connectivity results. One limitation of using AAV1 for anterograde
transneuronal studies, however, is the fact that AAV1-Cre can be transported also in the retrograde
direction, as has been reported (Tervo, et al., 2016; Rothermel, et al., 2013; Aschauer et al., 2013;
Aronoff et al., 2010; Masamizu, et al., 2014) and as we demonstrate in this study (Figure 3.3).
Application should therefore be limited to pathways that do not contain reciprocal connections
between targeted pre- and postsynaptic regions. Overall, its lack of toxicity and restricted spread to
first-order downstream neurons suggest that AAV may be a valuable new alternative to currently
available vectors for anterograde transneuronal circuit studies.
72

The exact mechanism by which AAV spreads transneuronally is unclear, though it appears to
reflect a specific interaction between the virus and host cells, rather than a physical property such as
passive diffusion, as injection of high titer Cre-expressing CAV2 and several other AAV serotypes
failed to replicate the anterograde transneuronal labeling seen with AAV1 and AAV9. AAV is a small
(~20 nm), replication-deficient virus that is normally endocytosed and transported to the nucleus
through interactions between capsid proteins and cell surface receptors (Pillay et al., 2016; Kotterman
et al., 2014). Evidence suggests that at high titers, some viral particles may be trafficked down axons
where they may be released from host cell synaptic terminals, enabling local infection of adjacent
neurons (Castle et al., 2014a; Castle et al., 2014b). Whether this always happens specifically through
synaptically connected neurons is still uncertain. Here we provide several lines of evidence that are
suggestive of a transsynaptic mechanism of spread. In particular, we show that (1) labeled cells are
found only in regions that conform to known innervation patterns of a given upstream structure; (2) no
labeling was found in cells surrounding fibers of passage; (3) labeled cells were never GFAP+ glial
cells, suggesting specificity in the viral spread to closely apposed neurons; and (4) labeled cells were
always functionally connected to their presynaptic starter population. These results support a
transsynaptic, rather than a nonspecific or extrasynaptic mechanism of spread in downstream target
regions.
Although the exact efficiency of the anterograde transsynaptic spread is unknown, our
experiments provide a rough estimation. First, 25% and 12% of local SC-sg neurons can be
transneuronally labeled by V1 injections in Vglut2-Cre and GAD2-Cre mice respectively (Figure 3.13),
suggesting a lower bound of transneuronal transport efficiency for glutamatergic and GABAergic SC
neuron populations respectively. Second, for SC-sg neurons retrogradely labeled from CTB injection
in LP, about 65% of the cells were transneuronally labeled from V1 injection (Figure 3.126),
suggesting that for the group of V1-recipient, LP-projecting SC neurons the efficiency of transneuronal
labeling is relatively high. Together, these data suggest that a reasonably high efficiency can be
achieved under our experimental conditions, by controlling the titer and volume of viral injection.
73

When combined with secondary injections of a conditional virus, the anterograde transneuronal
properties of AAV offer several unique opportunities for the study of neural circuits. In particular, this
method provides experimental access to spatially restricted groups of neurons that would otherwise
be impossible to target with conventional viral injections. These groups may then be targeted for
mapping their axonal output or for manipulation or recording using genetically encoded opsins or Ca
2+

indicators. In addition, the viral approach may be combined with transgenic mouse lines to gain
access to specific subtypes of neurons receiving input from a given upstream region, as we
demonstrated with injections of a novel Cre-dependent Flp expressing virus and a Flp-dependent YFP
expressing virus in Vglut2-Cre and GAD2-Cre mice (Figure 3.13). Such an approach enabled us to
separate glutamateric and GABAergic neurons in SC that were innervated by V1.
The two-step injection strategy provides a new way to functionally categorize neurons according
to their distinct input sources. As a proof of concept, we accessed discrete populations of neurons
residing in superficial, deep, and lateral aspects of SC. This enabled us to reveal that each
subpopulation, defined by its specific corticocollicular input, has a unique divergent axonal targeting
profile. We further show that two of these subpopulations—those in superficial layer SC receiving
input from V1, and those in deep layers of SC receiving input from A1—mediate distinct freezing and
escape behaviors, respectively. Together these results reveal the potential for AAV-based
anterograde transsynaptic labeling to access, map, and functionally manipulate specific groups of
neurons that would otherwise be challenging to investigate using conventional approaches.
Additionally, application of this method provides a straightforward means for examining the
potential role of downstream structures in mediating a function or behavior observed from activation of
a given upstream structure. Using this approach, we implicate LP as an important downstream
structure of SC-sg in mediating evoked freezing responses, as also suggested by a previous report
(Wei et al., 2015, but see Shang et al., 2015). In addition to optogenetically activating ChR2-
expressing SC axon terminals in LP, we went one step further and directly activated SC-recipient cell
bodies in LP and found robust evoked freezing behavior. Employing an anterograde transsynaptic
74

labeling approach to this functional test allowed us to bypass potential problems associated with
optogenetic activation of axon terminals. These could include activation of axon collaterals of SC
neurons, as illustrated in Figure 3.1A, and activation of fibers of passage in LP. Instead, directly
activating the LP neurons that receive SC input provides more conclusive evidence implicating this
structure’s role in mediating the freezing response evoked by SC-sg activation. Altogether, our results
demonstrate that the AAV-based anterograde transsynaptic tagging approaches described in this
study may be broadly applicable for use in the forward screening of functional circuits across multiple
synaptic steps.

75

CHAPTER 4

Anterograde transsynaptic AAV techniques for probing neural circuitry

Manuscript in preparation

76

4.1 Introduction
Revealing the organization and function of neural circuits is greatly facilitated by viral tools that spread
transsynaptically. Adeno-associated virus (AAV) has recently been shown to be capable of
anterograde transneuronal transport, with serotype 1 in particular exhibiting the greatest efficiency of
spread (Zingg et al., 2017). Given its well established lack of toxicity and apparent transduction of
only first-order post-synaptic neurons, AAV1 shows great promise as a tool for mapping and
manipulating input-defined cell populations.
Previous work suggests that AAV1 is released at or near axon terminals, and transduced
neurons downstream of the injection site show a high probability of receiving functional synaptic input
in slice recording experiments. However, the extent to which AAV1 spreads exclusively to
synaptically connected neurons remains uncertain. In addition, despite clear evidence for the active
trafficking of AAV-containing vesicles down the axon (Castle et al, 2014a; Castle et al., 2014b),
exactly how AAV is eventually released (e.g. through synaptic or extrasynaptic vesicle fusion) remains
unknown. Addressing these questions will be essential for establishing the synaptic nature of AAV
transneuronal transduction.
AAV1 has been shown to efficiently transduce both excitatory and inhibitory neurons
downstream of a variety of glutamatergic corticofugal pathways. In addition, this efficiency appears to
be critically dependent on viral titer, as reducing the titer from 10
13
to 10
11
GC/mL completely
eliminates transneuronal spread. Given the molecular diversity among different cell-types in the brain,
it remains uncertain whether differences in cell surface receptor expression, intracellular trafficking, or
synapse type might limit the efficiency of AAV spread in certain pathways. In particular, transneuronal
spread through inhibitory projection neurons or neuromodulatory cell populations has yet to be directly
demonstrated. Moreover, whether or not axonal length might diminish spread (e.g. from cortex to
spinal cord) remains to be tested.
77

In this study, we systematically examine the synaptic specificity of AAV1 transneuronal
transport using a variety of functional, anatomical, and molecular approaches. We find a strong
correspondence between pre-synaptic connectivity and post-synaptic labeling for different pathways,
and find that co-expression of tetanus toxin light chain, an inhibitor of pre-synaptic vesicle fusion,
nearly abolishes transsynaptic spread of AAV. In addition, we establish that AAV1 spreads efficiently
through inhibitory projection pathways, as well as long range pathways to the spinal cord, but shows
little or no spread through all neuromodulatory cell-types examined. Lastly, we expand the application
potential for this technique by combining Flp- and Cre-dependent mapping in dual-reporter mice and
incorporate its use with approaches for the sparse labeling of input- and genetically-defined neurons.

4.2 Materials and Methods
4.2.1 Animal preparation and stereotaxic surgery
All experimental procedures used in this study were approved by the Animal Care and Use
Committee at the University of Southern California. Male and female Ai14 (Cre-dependent tdTomato
reporter, Jackson Laboratories, stock #007914) and Frt-GFP (Flp-dependent GFP reporter, MMRRC,
stock #32038) mice aged 2-6 months were used in this study. Mice were group housed in a light
controlled (12 hr light: 12 hr dark cycle) environment with ad libitum access to food and water.
Stereotaxic injection of viruses was carried out as we previously described (Zingg et al., 2016;
Ibrahim et al., 2016; Liang et al., 2015; Xiong et al. 2015). Mice were anesthetized initially in an
induction chamber containing 5% isoflurane mixed with oxygen and then transferred to a stereotaxic
frame equipped with a heating pad. Anesthesia was maintained throughout the procedure using
continuous delivery of 2% isoflurane through a nose cone at a rate of 1.5 liters/min. The scalp was
shaved and a small incision was made along the midline to expose the skull. After leveling the head
relative to the stereotaxic frame, injection coordinates based on the Allen Reference Atlas (Dong,
78

2007) were used to mark the location on the skull directly above the target area and a small hole
(0.5mm diameter) was drilled. Viruses were delivered through pulled glass micropipettes with a
beveled tip (inner diameter of tip: ~20 µm) using pressure injection via a micropump (World Precision
Instruments). Total injection volumes ranged from 40 to 100 nL, at 15 nL/min. Following injection, the
micropipette was left in place for approximately 5 mins to minimize diffusion of virus into the pipette
path. After withdrawing the micropipette, the scalp was sutured closed and animals were
administered ketofen (5mg/kg) to minimize inflammation and discomfort. Animals were recovered
from anesthesia on a heating pad and then returned to their home cage.
For spinal injections of AAVretro-hSyn-GFP, mice were anesthetized and positioned into a
stereotaxic frame as described above. Cervical and lumbar injections were performed sequentially in
the same procedure using aseptic technique. For each injection, a small patch of skin above the
cervical or lumbar spinal region was shaved and a small incision was made to expose underlying
muscle tissue. Muscle was blunt dissected and retracted to expose cervical (near C7-C8) or lumbar
(near L3-L4) vertebrae and the spinal cord was stabilized using notched bars (Kopf, Model #987).
Virus was injected unilaterally (200 nL total volume, 15 nL/min) between vertebral segments using a
pulled glass micropipette with a beveled tip (inner diameter of tip: ~20 µm) at a depth of 0.7 mm from
the dorsal surface of the spinal cord. After withdrawing the micropipette, the skin was sutured closed
and animals were recovered from anesthesia on a heating pad. To minimize inflammation and
discomfort, animals were administered ketofen (5mg/kg) at the beginning of the surgical procedure
and again every 24 hours for 4 days following surgery.
4.2.2 Injection of viruses for anterograde transneuronal labeling
To compare the anterograde transneuronal transport properties of different viruses, self-
complementary (sc) AAV1-hSyn-Cre (Vigene Biosciences, 2.8 x 10
13
GC/mL), AAV1-hSyn-Flp
(Vigene Biosciences, 5.5 x 10
13
GC/mL), AAVretro-hSyn-Cre (Vigene Biosciences, 1.5 x 10
14
GC/mL),
AAVPHP.B-CMV-Cre (SignaGen, 2.3 x 10
13
GC/mL), Adenovirus (Ad5-CMV-Cre, Kerafast, 3.0 x 10
12

79

GC/mL), VSVG-pseudotyped Lentivirus (LV-CMV-Cre, Cellomics Tech, 1.0 x 10
8
GC/mL), VSVG-
pseudotyped Baculovirus (BAC-CMV-Cre, Uni. of Iowa, 3.7 x 10
10
GC/mL), or G-deleted Rabies virus
(RAV-Cre-GFP, Salk Institute, 8.6 x 10
8
GC/mL) was injected into primary visual cortex (V1, 60 nL
total volume; coordinates from bregma: anteroposterior -3.9 mm, mediolateral 2.6 mm, depth 0.5 mm)
of Ai14 mice (for Cre-expressing viruses) or Frt-GFP mice (for AAV1-hSyn-Flp). Original titers were
used for each virus injection. Animals were euthanized 4 weeks following injection and postsynaptic
structures were examined for the presence of cell body labeling.
To test for the specificity of viral spread to different cell-types in the cerebellum, scAAV1-hSyn-
Cre was injected into the pontine nucleus (PN, 80 nL total volume; coordinates from bregma:
anteroposterior -3.9 mm, mediolateral 0.4 mm, depth 5.5 mm) or inferior olive (IO, 80 nL total volume;
coordinates from bregma: anteroposterior -6.6 mm, mediolateral 0.4 mm, depth 5.5 mm) in Ai14 mice
crossbred with GAD67-GFP mice (from Dr. Yuchio Yanagawa, Brain Science Institute, RIKEN,
Japan). Animals were euthanized 2 weeks following injection and the cerebellar cortex was examined
for the presence of tdTomato+ cell body labeling in the molecular, granule, and purkinje cell layers.
To examine the efficiency of anterograde transsynaptic spread across different types of
synapses, scAAV1-hSyn-Cre injections were targeted to inhibitory projection neurons or
neuromodulatory cell populations in Ai14 mice. To test for viral spread through inhibitory synapses
onto downstream inhibitory neurons, scAAV1-hSyn-Cre was injected into the striatum (Str, 100 nL
total volume; coordinates from bregma: anteroposterior +0.5 mm, mediolateral 2.3 mm, depth 3.0 mm)
and the substantia nigra, pars reticulata (SNr) was examined for cell body labeling (2 week post-
injection survival time). To confirm the Str→SNr projection pathway is unidirectional, G-deleted
Rabies-GFP (Salk Institute, 5.5 x 10
8
GC/mL) was injected into the Str using the same coordinates as
above and the substantia nigra, pars compacta (SNc) and SNr were examined for the presence of
retrogradely labeled GFP+ cell bodies (50 nL injection, 1 week post-injection survival time). To test
for viral spread through inhibitory synapses onto presumed excitatory neurons, SNr was injected with
scAAV1-hSyn-Cre (50 nL total volume; coordinates from bregma: anteroposterior -3.3 mm,
80

mediolateral 1.6 mm, depth 4.3 mm) and downstream neurons in the ventromedial nucleus of the
thalamus (VM) were examined for the presence of tdTomato+ labeling (2 week post-injection survival
time). To test for viral spread through neuromodulatory synapses, scAAV1-hSyn-Cre was injected (80
nL total volume, 2 week post-injection survival time) into the diagonal band nucleus (NDB, cholinergic
neurons, coordinates from bregma: anteroposterior +0.5 mm, mediolateral 2.0 mm, depth 5.0 mm, 11°
angle), dorsal raphe (DR, serotonergic neurons, coordinates from bregma: anteroposterior -4.5 mm,
mediolateral 2.0 mm, depth 3.2 mm, 30° angle), or locus coeruleus (LC, noradrenergic neurons,
coordinates from bregma: anteroposterior -5.4 mm, mediolateral 0.8 mm, depth 3.1 mm) of Ai14 mice
and unidirectionally connected downstream targets were examined for cell body labeling (e.g. V1 for
NDB and LC, or dorsolateral geniculate (LGNd) and lateral entorhinal cortex (ENTl) for DR). To
demonstrate the location and axonal targeting of V1-projecting cholinergic or LGNd-projecting
serotonergic cell populations, AAV1-CAG-FLEX-GFP-WPRE (Addgene, 1.7 x 10
13
GC/mL) was
injected (60 nL total volume, 3 week post-injection survival time) into NDB of Chat-IRES-Cre mice
(Jackson Laboratories, stock #006410) or DR of Pet1-Cre mice (Jackson Laboratories, stock
#012712) using the same coordinates listed above.
To map the axonal output for different input-defined neurons in the thalamus, scAAV1-hSyn-
Cre was injected into either the anterior pretectal nucleus (APN, 80 nL total volume, coordinates from
bregma: anteroposterior -3.1 mm, mediolateral 1.3 mm, depth 3.3 mm) or principal sensory trigeminal
nucleus (PSV, 80 nL total volume, coordinates from bregma: anteroposterior -4.9 mm, mediolateral
2.0 mm, depth 4.4 mm) in Ai14 mice. In the same surgical procedure, AAV1-CAG-FLEX-GFP was
iontophoretically injected (Stoelting, 3 μA current, alternating 7 s on/off for 3 mins) into the PO/VPM
region of the ipsilateral thalamus (for APN injections) or contralateral thalamus (for PSV injections)
using the following coordinates from bregma: anteroposterior -1.5 mm, mediolateral 1.5 mm, depth
3.2 mm. Animals were euthanized 2 weeks post-injection and thalamocortical projections were
examined.
81

To label different input-defined cell populations in the spinal cord, descending pathways in the
brain were first identified by injecting AAVretro-hSyn-GFP (200 nL total volume, Vigene Biosciences,
1.7 x 10
14
GC/mL) unilaterally into the left cervical (C7-C8) and lumbar (L3-L4) spinal cord. Animals
were euthanized 3 weeks following injection and the entire brain was examined for retrogradely
labeled GFP+ cell bodies. Several brain regions were then selected for injection of scAAV1-hSyn-Cre
(100 nL total volume) in Ai14 mice using the following coordinates from bregma: Primary motor cortex,
upper-limb (MOp-ul, anteroposterior +0.7 mm, mediolateral 1.7 mm, depth 0.6 mm); primary motor
cortex, lower limb (MOp-ll, anteroposterior -0.8 mm, mediolateral 1.3 mm, depth 0.6 mm); primary
somatosensory cortex, lower-limb (SSp-ll, anteroposterior -0.9 mm, mediolateral 1.7 mm, depth 0.6
mm); lateral hypothalamic area (LHA, anteroposterior -1.5 mm, mediolateral 1.3 mm, depth 5.1 mm);
red nucleus (RN, anteroposterior -3.5 mm, mediolateral 0.5 mm, depth 3.7 mm); and superior
vestibular nucleus (SUV, anteroposterior -5.8 mm, mediolateral 1.5 mm, depth 3.3 mm). Following a
2 week post-injection survival time, animals were euthanized and cervical, thoracic, and lumbar
segments of the spinal cord were examined for tdTomato+ cell body labeling.
To reveal the topographical distribution of cells downstream of two different corticofugal
pathways in the same brain, AAV1-hSyn-Flp was injected into MOp-ul (100 nL total volume, same
coordinates as above) and scAAV1-hSyn-Cre was injected into MOp-ll (100 nL total volume, same
coordinates as above) in Ai14-tdTomato x Frt-GFP mice. Following a 2 week post-injection survival
time, downstream targets across the entire brain and spinal cord were examined for Flp+/GFP+ and
Cre+/Tom+ cell body labeling.
To label the axonal output of PN neurons that specifically receive input from upper limb-related
MOp, scAAV1-hSyn-Cre was injected into MOp-ul (80 nL total volume, same coordinates as above)
and AAV1-CAG-FLEX-GFP was injected into PN (80 nL total volume, same coordinates as above) in
Ai14 mice. Following a 2 week post-injection survival time, animals were euthanized and the
cerebellum was examined for GFP+ axons.
82

4.2.3 Tetanus toxin expression
To explore underlying mechanisms of AAV transneuronal release, tetanus toxin light chain was
expressed in V1 (same coordinates as described above) of Ai14 mice using injections of AAVDJ-
CMV-TeNT-P2A-GFP (Stanford Viral Core, 5.7 x 10
12
GC/mL, 100 nL total volume, n = 8 mice). As a
control, a separate group of Ai14 mice were injected with AAV1-hSyn-GFP-WPRE (Addgene, titer
reduced to 3.2 x 10
12
GC/mL, 100 nL total volume, n = 8 mice). Following 2 weeks to allow for
sufficient viral gene expression, scAAV1-hSyn-Cre was injected into the same location in V1 for each
group (60 nL total volume). After an additional 2 weeks post-injection, mice were euthanized and SC
was examined for the presence of tdTomato+ cell body labeling.
4.2.4 Virus injections for sparse labeling of neurons
To achieve sparse labeling in a given Cre-expressing cell population, a co-injection strategy was used
to obtain robust levels of YFP expression in only a few cells. Co-injections consisted of a 1:1 mixture
of high titer AAVDJ-EF1a-fDIO-YFP-WPRE (UNC Vector core, final titer 1.2 x 10
13
GC/mL) and low
titer AAV1-EF1a-DIO-Flp-WPRE (Vigene Bioscience, diluted within a final titer range of 7.5 x 10
8
–
10
10
GC/mL). To establish a relationship between titer and the resulting number of labeled cells,
AAV1-EF1a-DIO-Flp-WPRE was diluted to either 7.5 x 10
8
, 10
9
, or 10
10
GC/mL and co-injected with
AAVDJ-EF1a-fDIO-YFP-WPRE (50 nL total volume) in V1 of Ai14 x PV-Cre mice (Jackson
Laboratories, stock # 017320, n = 4 mice for each titer). Following a 2 week post-injection survival
time, animals were euthanized and V1 was examined for YFP+ cell labeling.
To achieve sparse labeling of both input- and genetically-defined cell populations, AAV1-
EF1a-DIO-Flp-WPRE was injected at reduced titer (1.5 x 10
12
GC/mL, 80 nL injection volume) into the
anterior cingulate area (ACA, coordinates from bregma: anteroposterior +0.5 mm, mediolateral 0.3
mm, depth 0.9 mm) in Ai14 x Vglut2-Cre mice (Jackson Laboratories, stock #016963). This titer was
chosen as previous results (Zingg et al., 2016) revealed the number of anterograde transsynaptically
labeled cells in a given target region are reduced by 85-90% when injection titer is lowered from 10
13

83

to 10
12
GC/mL. A test injection of scAAV1-hSyn-Cre (80 nL volume, 1.5 x 10
13
GC/mL) in ACA of
Ai14 x GAD67-GFP mice revealed an average of about 8 GAD67-GFP+/Tom+ cells and 35 GAD67-
GFP-/Tom+ cells (presumed excitatory) within a given 300 µm
3
sample space of the dorsolateral
periaqueductal gray (PAGdl). A reduced titer injection of AAV1-EF1a-DIO-Flp-WPRE (1.5 x 10
12

GC/mL, 80 nL injection volume) in ACA of Vglut2-Cre mice would then be expected to tag ~4 Vglut2-
Cre+/Flp+ cells per 300 µm
3
region of PAGdl. To robustly label these cells for morphological analysis,
injections of AAVDJ-EF1a-fDIO-YFP-WPRE (30 nL total volume, 2.5 x 10
13
GC/mL) were targeted to
the PAGdl (coordinates relative to bregma: anteroposterior -4.0 mm, mediolateral 0.6 mm, depth 2.3
mm). Following a 2 week post-injection survival, animals were euthanized and PAGdl was examined
for the presence of YFP+ cell bodies.
4.2.5 Histology
Following desired post-injection survival time, animals were deeply anesthetized and transcardially
perfused with 4% paraformaldehyde. Brains were extracted and post-fixed for 24 hours at 4˚C in 4%
paraformaldehyde and then sliced into 150 µm sections using a vibratome (Leica, VT1000s). The
sections were serially mounted onto glass slides and coverslipped. For some experiments, a
fluorescent Nissl stain was added (Neurotrace 640, ThermoFisher, N21483) to reveal cell body
location and cytoarchitectural information.
4.2.6 Imaging and quantification
All images were generated using a confocal microscope (Olympus FluoView FV1000). To quantify
the total number of cell bodies labeled in SC following virus injection in V1 (Figures 4.1, 4.5), serial
sections across the entire structure were collected and examined. Regions with labeled cells were
imaged at 10X magnification across the depth of the tissue (150 μm thickness, 15 μm z-stack
interval). TdTomato+ cell bodies that co-localized with fluorescent Nissl stain were manually identified
and counted.
84

To quantify cells labeled in the cerebellum following AAV1-Cre injection in PN or IO (Figure
4.4), coronal sections across the entire cerebellum were collected and the contralateral simple lobule
(SIM) was selected for quantification as it exhibited strong and consistent labeling in all examined
cases. For each animal, 10X magnification images were collected for 4 sections of SIM (from about -
5.8 to -6.4 mm posterior to bregma) across the depth of the tissue (150 μm thickness, 15 μm z-stack
interval). TdTomato+ neurons were quantified in the molecular layer, purkinje cell layer, and large
GAD67-GFP+/Tom+ cells were identified in the granule layer. To distinguish tdTomato+ granule cells
from mossy fiber terminals, 40X z-stack images were collected throughout the granule layer and
TdTomato+ cell bodies that co-localized with fluorescent Nissl stain were manually identified and
counted. Cell counts for all 4 sections were totaled for each animal and plotted as mean ± SD (n = 4
mice for both PN and IO).
To provide an estimate of the efficiency of viral spread across different types of synapses
(Figure 4.6), the percentage of tdTomato+ cells relative to Nissl+ cells was quantified within the axon
terminal field in either SNr (for striatum injections) or VM (for SNr injections). For diffuse
neuromodulatory output targeting entire structures or cortical regions, the percentage of tdTomato+
cells was quantified in relation to the total number of Nissl+ cells in the LGNd (for DR injections) or the
total number of tdTomato+ cells relative to Nissl+ cells within a 500 x 500 µm sample space in V1 (for
NDB and LC injections). 40X magnification images were used for quantification, and an average
percentage was generated for each animal using at least 4 sample images in the target region (n = 3
mice for each pathway).
To quantify the number of sparsely labeled YFP+/Tomato+ PV cells in V1 following co-
injections of AAV-fDIO-YFP and low titer AAV-DIO-Flp (Figure 4.10), serial sections through V1 were
collected and imaged at 10X magnification across the depth of the tissue (150 μm thickness, 15 μm z-
stack interval). The total number of YFP+/Tom+ cells were manually identified and counted for each
of the titers tested (n = 4 animals each).
85

4.2.7 Slice preparation and recording
To compare the rate of synaptic connectivity between tdTomato+, and neighboring non-labeled
neurons in IC, a 1:1 mixture of scAAV1-hSyn-Cre and AAV1-EF1a-DIO-hChR2-YFP (Addgene, 1.6 x
10
13
GC/mL) was injected into A1 of Ai14 mice (100 nL total volume, coordinates from bregma:
anteroposterior -3.1 mm, mediolateral 4.5 mm, depth 0.7 mm). Following a 2 week post-injection
survival time, acute brain slices containing the inferior colliculus (IC) were prepared. Following
urethane anesthesia, the animal was decapitated and the brain was rapidly removed and immersed in
an ice-cold dissection buffer (composition: 60 mM NaCl, 3mM KCl, 1.25 mM NaH2PO4, 25 mM
NaHCO3, 115 mM sucrose, 10 mM glucose, 7 mM MgCl2, 0.5 mM CaCl2; saturated with 95% O2 and
5% CO2; pH= 7.4). Brain slices of 350 μm thickness containing IC were cut in a coronal plane using
a vibrating microtome (Leica VT1000s). Slices were allowed to recover for 30 min in a submersion
chamber filled with warmed (35 °C) ACSF and then to cool gradually to room temperature until
recording. The spatial expression pattern of ChR2-EYFP in each slice was examined under a
fluorescence microscope before recording. IC neurons were visualized with IR-DIC and fluorescence
microscopy (Olympus BX51 WI) for specific targeting of both tdTomato+ neurons and nearby (within
150 µm) tdTomato- neurons surrounded by EYFP+ fluorescent fibers. Patch pipettes (Kimax) with ~4-
5 MΩ impedance were used for whole-cell recordings. Recording pipettes contained: 130 mM K-
gluconate, 4 mM KCl, 2 mM NaCl, 10 mM HEPES, 0.2 mM EGTA, 4 mM ATP, 0.3 mM GTP, and 14
mM phosphocreatine (pH, 7.25; 290mOsm). Signals were recorded with an Axopatch 200B amplifier
(Molecular Devices) under voltage clamp mode at a holding voltage of –70 mV for excitatory currents
or 0 mV for inhibitory currents, filtered at 2 kHz and sampled at 10 kHz. 1 µM tetrodotoxin (TTX) and
1 mM 4- aminopyridine (4-AP) was added to the external solution for recording only monosynaptic
responses (Petreanu et al. 2009) to blue light stimulation (3-10 ms pulse, 3 mW power, 10 trials,
delivered via a mercury Arc lamp gated with an electronic shutter).

86

4.2.8 Statistical Methods
All statistical analyses were performed using GraphPad Prism 6. Samples were first determined to
have normal distribution using the Shapiro-Wilk test. To determine a significant association between
anterograde transsynaptic labeling and synaptic connectivity (Figure 4.3E), Chi-square analysis was
performed. To test for a significant difference in the number of cerebellar cell-types labeled (Figure
4.4I) or the number of cells found SC in tetanus toxin experiments (Figure 4.5D), unpaired Student’s t-
test was used. Differences between data sets were considered significant if p < 0.05. Results for all
histograms are expressed as mean ± SD.

Table 4.1. List of viruses used in this study.

87

4.3 Results
4.3.1 Testing additional viruses and constructs for anterograde transsynaptic
transport
Our previous results indicated that AAV1, and to a lesser extent AAV9, are capable of anterograde
transsynaptic transport, while other tested serotypes (AAV5, 6, 8) or virus types (CAV2) did not
display this property (Zingg et al., 2017). It remains uncertain, however, whether other commonly
used viruses might also demonstrate such transport properties. To test this, we injected equal
volumes (60 nL) of Cre-expressing adenovirus (Ad5-CMV-Cre), lentivirus (LV-CMV-Cre), baculovirus
(BAC-CMV-Cre), or G-deleted rabies virus (RV-Cre-GFP) into the primary visual cortex (V1) of Ai14-
tdTomato Cre-reporter mice (Madisen et al., 2010; Jackson laboratories), and then examined the
superior colliculus (SC) for tdTomato+ cell bodies following a 4 week post-injection survival time
(Figure 4.1A). None of the tested viruses yielded any anterograde transsynaptic labeling in SC
(quantified in Figure 4.1D) or in any other downstream targets examined (e.g. striatum or pontine
nucleus, data not shown). We also tested two newly developed AAV serotypes, AAVretro (Tervo et
al., 2016) and AAV-PHP.B (Deverman et al., 2016), and similarly found no tdTomato+ cell bodies in
SC (Figure 4.1B, right panel; Figure 4.1D). Among all viruses tested, AAV1 thus appears unique in its
capacity to efficiently spread to neurons downstream of an injection site, and therefore remains the
best option for such applications.
Given that previous examples of AAV1-mediated anterograde transsynaptic spread have
primarily relied on viral expression of Cre in Ai14 tdTomato reporter mice, we asked whether this
technique could also be applied with a Flp recombinase reporter system. To test this, we injected
AAV1-hSyn-Flp (Vigene Biosciences) into V1 of Flp-dependent GFP-reporter mice (Frt-GFP; Sousa et
al., 2009; Jackson laboratories) and examined SC for GFP+ cell bodies after a 4 week post-injection
survival time. Similar to AAV1-Cre, we observed robust cell body labeling in SC (Figure 4.1C-D) and

88

in all other downstream targets of V1 (data not shown). This suggests AAV1-Flp may be a useful
alternative to Cre-based approaches for anterograde transsynaptic labeling. In addition, the two may
be combined in dual-reporter mice (Ai14 x Frt-GFP) to simultaneously examine the topographical or
convergent targeting of downstream neurons following any two injections of AAV1-Cre and AAV1-Flp
in the same brain (see Figure 4.9).

Figure 4.1. Comparison of anterograde transsynaptic spread for different viruses.
(A) Strategy for testing efficiency of transneuronal spread for different viruses. Injections of different Cre-
expressing viruses (60 nL total volume) were targeted to primary visual cortex (V1) in Ai14-tdTomato Cre-
reporter mice. Following a 4 week post-injection survival time, the superior colliculus (SC) was examined for
Cre+/tdTomato+ cell bodies.
(B) Anterograde transsynaptic labeling of cell bodies (red) was observed in SC following V1 injections of AAV1-
hSyn-Cre-WPRE (left) and self-complementary (sc) AAV1-hSyn-Cre (middle), but not AAV-PHP.B-CMV-Cre
(right). Blue, fluorescent Nissl stain.
(C) Anterograde transsynaptic labeling using an alternate recombinase-reporter system. Flp+/GFP+ cells
(green) were observed in SC following a 60 nl injection of AAV1-hSyn-Flp in V1 of Flp-dependent GFP reporter
mice (Frt-GFP). Scale bars (B and C), 250 µm, top panels; 25 µm, bottom panels.
(D) Quantification of total number of cells labeled in SC for different Cre-, or Flp-, expressing viruses injected
into V1 (4 week post-injection survival, 60 nL injection, n = 4 mice each). Error bar = SD.
89

We next explored whether modifications to the genetic construct of AAV1 might enhance the
transduction efficiency in post-synaptic neurons. In particular, it has been suggested that following
nuclear transport and capsid uncoating, a major rate limiting step in AAV transgene expression
involves the conversion of single-stranded viral DNA to double-stranded DNA (McCarty, 2008). This
may be overcome, however, by directly packaging AAV with a “self-complementary” double-stranded
DNA construct (scAAV). To test for any potential enhancement in anterograde transsynaptic
transduction by scAAV, we injected self-complementary Cre-expressing AAV1 (scAAV1-hSyn-Cre,
Vigene Biosciences) into V1 of Ai14 mice and quantified the number of tdTomato+ cell bodies in SC
(Figure 4.1B, mddle panel; Figure 4.1D). We observed ~30% more tdTomato+ cells in SC compared
with injections of AAV1-CMV-Cre, and ~24% more cells when compared with AAV1-hSyn-Flp (Figure
4.1D). This increase in efficiency may be attributed to the self-complementary vector design and/or
differences in the relative strength of the hSyn and CMV promoters (compared with AAV1-CMV-Cre)
or efficacy of recombinase activity (compared with AAV1-hSyn-Flp). Despite these increases,
however, scAAV1-hSyn-Cre proved to be only about half as efficient as AAV1-hSyn-Cre-WPRE
(Figure 4.1B, left panel; Figure 4.1D). Given that viral titer, injection volume, and promoter sequence
were similar for these two viruses (see Materials and Methods), the difference in labeling efficiency
may be due to the presence of the woodchuck hepatitis virus posttranscriptional regulatory element
(WPRE), which has been reported to increase vector gene expression by 7-fold (Loeb et al., 2002).
Inclusion of this enhancer element may therefore surpass any potential boost in gene expression
conferred by a self-complementary vector design.
Given its increased efficiency relative to other AAV1 constructs, AAV1-hSyn-Cre-WPRE
appears to be an ideal tool for anterograde transsynaptic experiments. It is important to note,
however, that very high levels of Cre expression may become toxic for host cells. Indeed, under our
experimental conditions we observed some evidence for cell death at the injection site following
injections of AAV1-hSyn-Cre-WPRE, but not scAAV1-hSyn-Cre or AAV1-hSyn-Flp, suggesting the
WPRE element may facilitate excessive Cre expression and lead to toxicity in host neurons (Figure
90

4.2A-B). To avoid this complication, we selected the latter two viruses for use throughout the rest of
this study.

Figure 4.2. WPRE enhancer may contribute to Cre-induced toxicity at the injection site.
(A) Injection of AAV1-hSyn-Cre-WPRE in V1 of Ai14 mice (top panel) may result in cell death at the injection
site, as seen by a reduction in Nissl stain intensity and irregular cell morphology (middle panel). Bottom panel,
anterograde transsynaptically labeled neurons in SC.
(B) Injections of either scAAV1-hSyn-Cre in Ai14 mice or AAV1-hSyn-Flp in Frt-GFP mice (top panels) show no
apparent toxicity in neurons at the injection site (middle panels) and lead to robust anterograde transsynaptic
labeling of cells in SC (bottom panels). All injections 60 nL total volume, 4 week post-injection survival time.

91

4.3.2 Functional, anatomical, and molecular examination of synaptic spread
Previously, we observed that AAV1-Cre appears to spread only to first order neurons in expected
target regions, does not leak from axons of passage, and labels neurons that show a high probability
of functional synaptic input (Zingg et al., 2017). These observations are suggestive of a transsynaptic
mechanism of spread, however the extent to which AAV1 passes exclusively through synaptic
connections remains uncertain. To provide a deeper understanding of the synaptic nature of AAV1
transport, we systematically characterized (1) the probability of functional connectivity between
labeled and non-labeled cells in a target region, (2) the specificity of spread to only expected cell-
types in a well-defined circuit, and (3) the effect of blocking presynaptic vesicle fusion on AAV spread
via co-expression of tetanus toxin.
To explore the association between anterograde transsynaptically labeled neurons and
functional connectivity, we performed slice recording from tdTomato+ neurons and neighboring
tdTomato- neurons in the inferior colliculus (IC). Co-injections of a 1:1 mixture of scAAV1-hSyn-Cre
and AAV1-EF1a-DIO-ChR2-YFP into the primary auditory cortex (A1) of Ai14 mice (Figure 4.3A)
enabled ChR2 expression only in neurons that were also co-transduced by AAV1-Cre, thus allowing
us to activate the same presynaptic axons that presumably transported AAV1-Cre to its downstream
targets. Using whole-cell recording, functionally connected cells were identified by their LED-evoked
excitatory synaptic responses, which persisted in the presence of tetrodotoxin (TTX) and 4-AP
(Petreanu et al., 2009; Figure 4.3C). Altogether, 100% of the tdTomato+ cells recorded exhibited
LED-evoked responses (28/28 cells), while only 46% of neighboring tdTomato- cells responded to
LED pulses (13/28 cells, Figure 4.3D-E). These results revealed a statistically significant association
between tdTomato+ labeling and functional connectivity when compared with non-labeled neurons
randomly recorded within the same region. AAV1 may therefore preferentially spread to synaptically
connected neurons downstream of the injection site.
92

Figure 4.3. Verification of functional synaptic connectivity.
(A) Strategy for slice recording from transsynaptically labeled neurons in the IC (red) following co-injection of
scAAV1-hSyn-Cre and AAV1-EF1a-DIO-ChR2-YFP into A1 (2 week post-injection survival, 100 nL injection
volume).
(B) ChR2-expressing axons (green) surrounding a tdTomato-labeled neuron and neighboring non-labeled
neurons (blue Nissl stain). Scale bar, 25 µm.
(C) Average LED-evoked excitatory (-70 mV) and inhibitory (0 mV) currents in an example tdTomato+ IC neuron
before and after perfusing in TTX and 4-AP. LED stimulation is marked by a blue bar.
(D) Fraction of transsynaptically labeled (red) and neighboring non-labeled (gray) neurons showing excitatory
monosynaptic current response to LED stimulation. ***p < 0.001, Chi-square test, 28 cells in each group.
(E) Summary of amplitudes of average monosynaptic excitatory currents evoked by LED for all labeled (red) and
non-labeled (gray) recorded neurons (28 cells in each group). Error bar = SD.

Given the relatively small number of neurons sampled in the above study, it is possible that
some non-connected tdTomato+ neurons may have been overlooked. To provide a broader
estimation of the synaptic specificity of AAV1 transneuronal spread, we screened two different
unidirectional pathways that converge on distinct cell-types within the cerebellar cortex (Figure 4.4).
Neurons in the pontine nucleus (PN) project to the granule layer of the cerebellum and form well-
characterized connections with granule cells and large inhibitory Golgi cells (Palay & Chan-Palay,
1974; Kanichay & Silver, 2008). On the other hand, neurons in the inferior olive (IO) send axons
through the granule layer and terminate primarily onto the dendrites of purkinje cells (Palay & Chan-
Palay, 1974; Mathews et al., 2012). We therefore expected injections of AAV1-Cre in the PN to label
granule cells, but not purkinje cells, and injections in the IO to label purkinje cells, but not granule
cells.
93

Figure 4.4. Anatomical evidence for the synaptic specificity of viral spread.
(A) Strategy for labeling post-synaptic neurons in the cerebellar cortex (CB) following injection of scAAV1-hSyn-
Cre in the pontine nucleus (PN) of Ai14 x GAD67-GFP transgenic mice (2 week post-injection survival, 80 nL
injection volume).
(B) Example injection site (red) in the pontine nucleus. Fluorescent Nissl stain, blue.
(C) Pontine afferents and post-synaptic neurons (red) labeled in the granule cell layer of the cerebellum (coronal
section through simple lobule shown). GAD67-GFP+ purkinje cells and molecular layer interneurons, green.
Fluorescent Nissl stain, blue. Scale bar (C and G), 250 µm.
(D) 40X magnification image of dashed region shown in (C). Cre+/tom+ granule cells (red, co-stained with
Nissl, blue, arrowheads) and mossy fiber terminals (red) were observed in the granule cell layer, along with
94

large GAD67-GFP+ neurons (yellow, asterisk). Right panels, 40X magnification examples of cells found in
regions (i-iv) shown in (C). Fluorescent Nissl and tdTomato channels only (left), and with GFP (right). Scale
bars (D and H), 25 µm.
(E) Strategy for labeling post-synaptic neurons in CB following injection of scAAV1-hSyn-Cre in the inferior olive
of Ai14 x GAD67-GFP mice (2 week post-injection survival, 80 nL injection volume).
(F) Example injection site in IO (red).
(G) Climbing fiber afferents and post-synaptic neurons (red) labeled in the granule, purkinje, and molecular
layers of the simple lobule.
(H) 40X magnification of dashed region shown in (G). Cre+/Tom+/GFP+ neurons were observed in the granule
layer (yellow, arrowheads), purkinje cell layer (yellow, asterisk), and molecular layer (top right panels).
(I) Quantification of the total number of cell-types counted across 4 sections of the simple lobule for each animal
injected in PN or IO (mean plotted for n = 4 animals for each pathway). Error bar = SD. *p <0.05, ***p < 0.001,
t-test.

To test this, we first injected scAAV1-hSyn-Cre into the PN or IO of Ai14 x GAD67-GFP mice
(Figure 4.4A-B). GAD67-GFP expression was used to facilitate identification of purkinje cells and
molecular layer interneurons (MLIs), as well as large inhibitory neurons in the granule cell layer
(Figure 4.4C). Following a 2 week post-injection survival time, PN injections yielded hundreds of
small tdTomato+/GFP- cells in the granule layer (presumed granule cells) along with numerous large
tdTomato+/GFP+ neurons (presumed Golgi cells, Figure 4.4C-D), while IO injections labeled a large
number of purkinje cells, as well as MLIs (Figure 4.4G-H). To quantify this labeling, we focused on 4
consecutive sections of the contralateral simple lobule (SIM; see Materials and Methods) and counted
the total number of granule cells, purkinje cells, MLIs, and GAD67+ granule layer neurons across all 4
sections for each animal (n = 4 animals for both PN and IO). Total cell counts for each cell-type were
then plotted for PN and IO groups to directly compare labeling results for each pathway (Figure 4.4I).
As expected, PN injections labeled a large number of granule cells (average 735 cells per animal),
while IO injections labeled a negligible number of granule cells (average 13 cells per animal).
Similarly, IO injections robustly labeled purkinje cells (average 48 cells per animal), while PN
injections labeled little or no purkinje cells (average 1 cell per animal). Given that each pathway is
expected to innervate one cell-type and exclude the other, the non-zero values observed for
unexpected cell-types may represent potential leakage of the virus to non-synaptically connected
95

neurons. For each pathway, leaky cell labeling comprised about 2% of the total cell population
labeled by the other pathway. Thus, despite this potential for viral leakage, anterograde
transneuronal transport was overall quite selective in labeling the expected post-synaptic granule or
purkinje cell populations.
Interestingly, while large GAD67-GFP+ cells in the granule layer (presumed Golgi cells) were
expected to be labeled following PN injections (Kanichay & Silver, 2008), it was surprising to see even
more of these cell-types labeled with IO injections (average 15.7 versus 22 neurons per animal;
Figure 4.4G-I). If these neurons are indeed Golgi cells, it may be possible that they receive direct
synaptic input onto their apical dendrites, which extend into the molecular layer and overlap with IO
climbing fiber terminals (Castejon & Castejon, 2000; however, see Galliano et al., 2013). While these
cells have been shown to be functionally coupled in vivo (Schulman & Bloom, 1981; Xu & Edgley,
2008) and in vitro (Nietz et al., 2017), evidence from slice recording studies suggests that IO climbing
fibers may drive Golgi cells via glutamate spillover, rather than through direct point-to-point synaptic
contacts (Nietz et al., 2017). The same may also be true of IO climbing fiber innervation of MLIs,
which were found to be prominently labeled following AAV1-Cre injections in IO, but not PN (Figure
4.4G-I). Some anatomical work has suggested the potential for direct synaptic connections between
climbing fibers and MLIs (Brown et al., 2012; Galliano et al., 2013), however slice recording studies
again point to a spillover mechanism of functional connectivity (Szapiro et al., 2007; Coddington et al.,
2013). Given these anatomical uncertainties, and the potential specialization of climbing fiber
terminals for high volume transmitter release, it is unclear whether viral spread to these two cell types
is synaptic in nature, and whether conclusions about the transsynaptic specificity of spread to these
cells are generalizable to other pathways that signal through more restricted point-to-point synaptic
contacts. We thus focus our estimation of the synaptic specificity of viral spread to the more
established purkinje cell and granule cell connections for each pathway and find AAV1 to be quite
selective in labeling each of these respective cell populations.
96

How might AAV1 be released from synaptic terminals? Previous studies have shown that,
following uptake at the soma, about 14% of all AAV-containing endosomes are rapidly trafficked down
the axon in a kinesin-2 dependent manner (Castle et al., 2014a; Castle et al., 2014b). It is possible
that some of these may merge with endosomal compartments that give rise to synaptic vesicles,
enabling co-release of AAV particles and neurotransmitter into the synaptic cleft (Figure 4.5A).
Alternatively, AAV may accumulate in other endosomal compartments capable of extrasynaptic
release near the axon terminal. For example, multivesicular bodies (MVBs) in axons have been
shown to contain exogenous materials such as neural tracers and viruses (Von Bartheld & Altick,
2011) and may fuse with the plasma membrane to release their contents into the extracellular space
(Janas et al., 2016). To examine the contribution of synaptic vesicle release in viral spread, we
expressed tetanus toxin light chain (TeNT) in V1 neurons using injections of AAVDJ-CMV-TeNT-P2A-
GFP in Ai14 mice (Figure 4.5A, 4.5C). TeNT has been shown to completely block Ca2+-evoked
synaptic vesicle fusion and transmitter release by cleaving VAMP2 (also known as synaptobrevin-2)
(Schaivo et al., 1992; Schoch et al., 2001; Yamamoto et al., 2003). Cleavage results in improper
SNARE complex formation specifically for synaptic vesicles, while presumably leaving other forms of
vesicular fusion intact. For example, instead of using VAMP2, MVB fusion has been proposed to
require VAMP7 or YKT6, which are not subject to TeNT cleavage (Raposo & Stoorvogel, 2013;
Hessvik & Llorente, 2018). Following a 2 week post-injection survival time to allow for TeNT
expression, a second injection of scAAV1-hSyn-Cre was targeted to the same location in V1 (Figure
4.5C, top panel), and SC was then examined 2 weeks later for the presence of tdTomato+ cell bodies
(Figure 4.5C, bottom panels). Remarkably, we observed a ~94% decrease in the number of
transsynaptically labeled neurons in SC, as compared with control animals that received only GFP-
expressing injections of AAV1 followed by scAAV1-hSyn-Cre injection (Figure 4.5B, 4.5D). The
remaining fraction of transsynaptically labeled cells observed may be the result of incomplete co-
transduction of starter cells with both viruses, and/or may reflect alternate vesicular mechanisms of
release (such as MVB fusion) that could contribute to viral spread. Lastly, it is also worth noting that
97

the reduction in labeling was accompanied by noticeable enlargements in TeNT-containing axon
terminals (Figure 4.5C, bottom panel), which may reflect a compensatory change following complete
block of synaptic vesicle release (see Woods et al., 2018). Overall, these results provide additional
insight into the synaptic specificity of viral spread and suggest that successful transduction of
downstream neurons by AAV1 may favor a synaptic mechanism of vesicle release.

Figure 4.5. Tetanus toxin inhibition of viral spread.
(A) Experimental design and timeline of virus injections. AAV trafficked to the synapse may be released through
synaptic vesicles in a VAMP2-dependent manner. Tetanus toxin cleaves VAMP2, preventing synaptic vesicle
fusion and potential release of AAV.
(B) Control experiment. AAV1-hSyn-GFP injection in V1 (100 nL volume) followed by scAAV1-hSyn-Cre
injection (60 nL) two weeks later in Ai14 mice (top panel). Two weeks after the second injection, GFP+ axons
(green) and anterograde transsynaptically labeled neurons (red) were observed in SC (middle panel, 10X;
bottom panel, 40X). Fluorescent Nissl, blue.
(C) AAVDJ-CMV-TeNT-P2A-GFP injection in V1 (100 nL) followed by scAAV1-hSyn-Cre injection (60 nL) two
weeks later in Ai14 mice (top panel). Two weeks after the second injection, GFP+/TeNT+ axons (green) were
98

found in SC, however very few transsynaptically labeled neurons were observed (red). Scale bars (B and C),
500 µm, top panel; 250 µm, middle; 25 µm, bottom.
(D) Quantification of anterograde transsynaptically labeled cells in SC (n = 8 mice for each group). Error bar =
SD. ***p < 0.001, t-test.

4.3.3 Efficiency of spread across different types of synapses
We previously demonstrated that AAV1-Cre can spread from excitatory projection neurons to both
excitatory and inhibitory post-synaptic cell-types. However, whether or not this transport occurs
efficiently across other types of synaptic connections in the brain (e.g. through inhibitory or
neuromodulatory pathways) remains to be determined. To test this, we first examined the efficiency
of anterograde transsynaptic spread from inhibitory neurons to downstream excitatory or inhibitory cell
populations in Ai14 mice (Figure 4.6A-B). To avoid possible retrograde labeling from AAV1 injections
(Rothermel et al., 2013; Aschauer et al., 2013; Zingg et al., 2017), only unidirectionally connected
regions downstream of each injection site were examined. To characterize the efficiency of spread
from inhibitory-to-inhibitory cells, we injected scAAV1-hSyn-Cre into the striatum (Str), which contains
GABAergic medium spiny neurons (MSNs) that project to inhibitory cells within the substantia nigra,
pars reticulata (SNr) (Figure 4.6A, left panel). Following a 2 week post-injection survival time,
numerous tdTomato+ cells were found intermingled with dense axon terminals in SNr (Figure 4.6A,
middle panels), suggesting the potential for AAV1 to spread through inhibitory synapses to
downstream inhibitory neurons. We also observed tdTomato+ cells in the overlying substantia nigra,
pars compacta (SNc, presumed dopaminergic neurons) (Figure 4.6A, second panel from left),
however these were excluded from analysis as they provide strong projections back to the Str and
may therefore have been retrogradely labeled by AAV1-Cre. This is supported by the observation
that injections of GFP-expressing G-deleted rabies virus into the same region of the Str robustly back-
label neurons in SNc, but not SNr (Figure 4.6A, right panel). To provide an estimate of the
99

Figure 4.6. Efficiency of viral spread across different types of synapses.
(A) Anterograde transsynaptic spread of scAAV1-hSyn-Cre in Ai14 mice (100 nL injection volume, 2 week post-
injection survival) from inhibitory projection neurons in striatum (left panel, red, scale bar, 500 µm) to inhibitory
neurons in the SNr (middle panels, scale bars, 250 µm, left; 25 µm, right). Retrograde labeling was restricted
primarily to SNc, as revealed by Rabies-GFP injection in striatum (right panel, green, scale bar, 250 µm).
(B) Anterograde transsynaptic spread of scAAV1-hSyn-Cre (50 nL injection volume, 2 week post-injection
survival) from inhibitory projection neurons in SNr (left panel, red, scale bar, 500 µm) to presumed excitatory
neurons in VM (right panels, scale bars, 500 µm, left; 250 µm, middle; 25 µm, right).
(C) Anterograde transsynaptic spread through cholinergic neurons. Cholinergic cells in the NDB project
unidirectionally to the primary visual cortex (left panels, green, scale bars, 500 µm, left; 250 µm, right). Injection
of scAAV1-hSyn-Cre (80 nL volume, 2 week post-injection survival) into the NDB sparsely labeled neurons in V1
throughout all cortical layers (right panels, red, scale bars, 500 µm, left; 250 µm, middle; 25 µm, right).
100

(D) Anterograde transsynaptic spread through serotonergic neurons. Serotonergic neurons in DR project
unidirectionally to LGNd (left panels, green, scale bars, 500 µm, left; 250 µm, right) and ENTl (data not shown).
Injection of scAAV1-hSyn-Cre (80 nL volume, 2 week post-injection survival) into the DR labeled little or no cells
in LGNd or ENTl (right panels, red, scale bars, 500 µm, left; 250 µm, middle; 250 µm, right).
(E) Anterograde transsynaptic spread through noradrenergic neurons. Injection of scAAV1-hSyn-Cre (80 nL
volume, 2 week post-injection survival) into the LC (left panel, red, scale bar, 500 µm). No labeling of cell
bodies was observed in downstream regions unidirectionally connected to LC, such as V1 (right panel, scale
bar, 250 µm).
(F) Estimation of the efficiency of transsynaptic spread across different types of synapses (# of tom+ cells / # of
Nissl+ cells) in each target region shown in (A-E) (n = 3 mice each). Data are compared with previous results
for excitatory projections from V1 to downstream excitatory (PN) and inhibitory (LGNv) cell-types (blue bars).
Error bar = SD.

anterograde transsynaptic labeling efficiency from Str→SNr, we quantified the number of tdTomato+
neurons relative to the total number of Nissl+ neurons within the boundaries of the axon terminal field
in SNr (see Materals and Methods). We found on average ~41% of the cells within this region were
tdTomato+, suggesting comparable efficiency to previously reported excitatory projection neuron
pathways (Zingg et al., 2017; Figure 4.6F). In addition, using the same approach, we also tested
AAV1 spread from inhibitory neurons in the SNr to presumed excitatory neurons in the ventromedial
nucleus of the thalamus (VM), another known unidirectional pathway (Figure 4.6B). Again, we found
relatively efficient anterograde transsynaptic labeling in VM (~36% tom+/total Nissl+ cells; Figure
4.6B, 4.6F), providing further evidence that AAV1 may be used to transsynaptically tag diverse cell
populations downstream of different classes of inhibitory projection neurons.
To determine whether neuromodulatory cell-types support anterograde transsynaptic spread
of AAV1, we performed similar injections in brain regions that contain cholinergic (ACh), serotonergic
(5-HT), or noradrenergic (NE) cell populations and examined targets that were exclusively
downstream for tdTomato+ cell bodies (Figure 4.6C-E). In particular, to reveal transsynaptic spread
through cholinergic neurons, we injected the diagonal band nucleus (NDB) in the anterior basal
forebrain, which projects strongly to V1 (Figure 4.6C, left panels), but does not receive input back
from V1. Following injections of scAAV1-hSyn-Cre into the NDB, we observed sparse tdTomato+
101

labeling scattered throughout all layers of V1, including layer 1 (Figure 4.6C, right panels). This
suggests AAV1 may have the capacity to spread post-synaptically through cholinergic neurons, albeit
less efficiently than expected given the dense termination pattern in this target structure. Indeed,
quantification of the number of tdTomato+ cells relative to Nissl+ cells within a given 500 x 500 µm
sample of V1 revealed only ~0.65% labeling efficiency (Figure 4.6F). One possible explanation for
this lack of efficiency may be that cholinergic neurons can signal through both classical synaptic
connections as well as extrasynaptic axonal varicosities that enable “volume” release of
neurotransmitter (Arroyo et al., 2014). It is possible that viral particles released into the extracellular
space via these varicosities are statistically less likely to transduce downstream neurons than those
that are released directly into the synaptic cleft. In line with this, we also observed inefficient spread
of AAV1 through serotonergic and noradrenergic cell populations, both of which have also been
proposed to signal extensively through volume release of neurotransmitter (Agnati et al., 1995).
Serotonergic projections from the dorsal raphe (DR) to the dorsolateral geniculate nucleus (LGNd) or
lateral entorhinal cortex (ENTl) produced almost no transsynaptic labeling (Figure 4.6D, right panels),
despite strong axonal innervation of these structures (Figure 4.6D, left panels). Furthermore,
transsynaptic labeling was completely absent in V1 following transduction of noradrenergic neurons in
the locus coeruleus (LC) (Figure 4.6E), which diffusely project to most cortical areas, including V1
(Polack et al., 2013; Schwarz et al., 2015). Thus, while AAV1 appears to be capable of efficient
spread through classically defined glutamatergic and GABAergic synaptic pathways, its application in
various neuromodulatory systems may be limited by potential volume release of virus, or by other
factors, such as differences in viral uptake or internal trafficking, that reduce efficiency of spread to
downstream neurons.

102

4.3.4 Application in corticofugal, thalamic, and spinal pathways
In our previous work, we primarily focused on a limited number of cortico-collicular projections to
demonstrate the utility of AAV1-mediated transsynaptic transport. We therefore aimed to expand the
potential application of this technique by characterizing its use in a broader set of neural pathways. In
particular, we (1) demonstrated its potential for accessing distinct thalamic cell populations defined by
their ascending input from mid- and hindbrain structures (Figure 4.7), (2) explored its use in labeling
spinal neurons downstream of various descending pathways to the spinal cord (Figure 4.8), and (3)
demonstrated its utility in revealing topographically-defined corticofugal-recipient cell populations
following dual injections of AAV1-Cre and AAV1-Flp in Flp/Cre-reporter mice (Figure 4.9).
Given that many of the ascending projections to the thalamus from mid- and hindbrain
structures are unidirectional, we asked whether anterograde transsynaptic tagging with AAV1-Cre
could be applied throughout this set of pathways to gain access to specific, input-defined thalamic
populations. To test this, we injected scAAV1-hSyn-Cre into either the principle sensory trigeminal
nucleus (PSV) (Figure 4.7A-B) or the anterior pretectal nucleus (APN) (Figure 4.7G-H) in Ai14 mice.
PSV is known to project strongly to the contralateral ventral posteromedial nucleus of the thalamus
(VPM), where it relays primary sensory information from the head, neck, and whisker pads (Petersen,
2007). On the other hand, APN projects to cells in the posterior nucleus of the thalamus (PO) and
may contribute to high-order modulation of somatosensory processing (Rees & Roberts, 1993; Bokor
et al., 2005; Giber et al., 2007), however its exact function remains unknown. As expected, following
each injection we saw strong anterograde transsynaptic labeling of cell bodies in either the VPM
(Figure 4.7C-D) or the PO (Figure 4.7I-J). As neurons from each of these thalamic structures are
known to have unique, laminar-specific projections in somatosensory cortex (Lu & Lin, 1993; Wimmer
et al., 2010), we targeted each population with iontophoretic injections of AAV1-CAG-Flex-GFP to
further confirm their identity. Selective expression of GFP in PSV-recipient VPM neurons revealed a
strong projection to layer (L)4 “barrels” in the somatosensory barrel cortex (SSp-bfd) (Figure 4.7E-F),
103

while output from APN-recipient PO neurons targeted L1, L2/3, and L5 of SSp-bfd and avoided direct
innervation of L4 barrels (Figure 4.7K-L). These distinct output profiles matched the expected
termination patterns for PO and VPM described in previous studies (Lu & Lin, 1993; Wimmer et al.,
2010), and suggest that the anterograde transsynaptic spread of AAV1 is selective enough to
parcellate these two populations, despite their close proximity to each other in the thalamus. Overall,
combined with previous demonstrations in a colliculo-thalamic pathway (Zingg et al., 2017; Beltramo
& Scanziani, 2019), these results suggest that transsynaptic tagging through various ascending
connections may provide a promising means of experimentally accessing specific thalamic
populations that might otherwise be difficult to target using conventional approaches.

Figure 4.7. Application in ascending thalamic pathways.
(A) Strategy for labeling thalamic neurons in VPM that receive input from PSV.
(B) Injection of scAAV1-hSyn-Cre in PSV in an Ai14 mouse (red, 80 nL volume).
(C) Injection of AAV1-CAG-Flex-GFP in VPM (green, iontophoresis). Both injections (B and C) performed in the
same surgical procedure, post-injection survival time = 2 weeks. Scale bar (B and C), 500 µm.
(D) 40X magnification of Cre+/tom+ transsynaptically tagged neurons in VPM (red, left panel) that co-express
GFP (right panel). Scale bar, 25 µm.
104

(E) Axonal projection to barrel cortex (green) from PSV-recipient VPM neurons labeled in (C). Scale bar, 500
µm.
(F) Axons terminate densely within L4 barrels and sparsely across the boarder of L5 and L6. Scale bar, 250 µm.
(G) Strategy for labeling thalamic neurons in PO that receive input from APN.
(H, I) Injections of scAAV1-hSyn-Cre in APN (red, 80 nL volume) and AAV1-CAG-Flex-GFP in PO (green,
iontophoresis). Post-injection survival time = 2 weeks. Scale bar, 500 µm.
(J) Transsynaptically labeled Cre+/tom+ neurons in PO (red, left panel) co-expressing GFP (green, right panel).
Scale bar, 25 µm.
(K) Axonal projection to barrel cortex (green) from APN-recipient PO neurons. Scale bar, 500 µm.
(I) Axons avoid L4 barrels and terminate most densely in L1, L2/3, and L5. Scale bar, 250 µm.

We next asked whether AAV1 transsynaptic tagging could be applied in long-range projection
pathways from the brain to the spinal cord. To test this, we first identified several brain regions that
contained spinal-projecting cell populations using injections of a retrogradely transported GFP-
expressing virus (AAVretro-hSyn-GFP; Tervo et al., 2016) in the cervical and lumbar spinal cord
(Figure 4.8A). Based on this labeling, we then selected three cortical regions (upper- and lower-limb-
related primary motor cortices, MOp-ul, MOp-ll; and lower-limb-related somatosensory cortex, SSp-ll)
and three subcortical regions (lateral hypothalamic area, LHA; red nucleus, RN; and superior
vestibular nucleus, SUV) for injections of scAAV1-hSyn-Cre in Ai14 (Figure 4.8B). Following a 2 week
post-injection survival time, robust tdTomato+ cell body labeling was observed in the spinal cord for
each descending pathway (Figure 4.8C), suggesting a capacity for AAV1-Cre to efficiently transduce
neurons over long distances. Interestingly, the distribution of this post-synaptic labeling appeared to
be regional and layer-specific for each pathway, suggesting different descending projections to the
spinal cord recruit unique sub-populations of spinal neurons that may underlie their distinct functional
roles. For example, MOp-ul and MOp-ll labeled neurons in lamina IV and V of the cervical or lumbar
spinal cord, respectively (Figure 4.8C, left columns), while output from SSp-ll was restricted mostly to
the medial portion of lamina IV in the lumbar region (Figure 4.8C, third column from left). These
results further highlight the specificity of anterograde transsynaptic AAV spread and reveal its
potential use in dissecting the downstream circuit components of different brain-spinal projection
pathways.
105

Figure 4.8. Application in descending pathways to the spinal cord.
(A) Retrograde labeling of spinal-projecting cortical and subcortical neuronal populations (green) following
injections of AAVretro-hSyn-GFP into the left side of the cervical and lumbar spinal cord. Scale bars (A and B),
500 µm.
(B) Corresponding injections of scAAV1-hSyn-Cre (red, 100 nL injection volume) into different spinal-projecting
brain regions in Ai14 mice.
(C) Following a 2 week post-injection survival time, different patterns of transsynaptic labeling were observed
across cervical, thoracic, and lumbar segments of the spinal cord for each injection. Each column corresponds
to the injection site shown above in (B). Scale bars, 250 µm.

106

Lastly, as corticofugal output represents one of the largest and most straight-forward systems
for applying anterograde transsynaptic tagging, we aimed to explore in greater detail its potential for
accessing input-defined cell populations throughout the entire brain. In particular, given the highly
topographical nature of corticofugal output, we asked whether AAV1-Cre and AAV1-Flp could be used
together to simultaneously reveal the topographic distribution of cells innervated by two different
functionally distinct cortical regions. To test this, we injected AAV1-hSyn-Flp in MOp-ul and scAAV1-
hSyn-Cre into MOp-ll in a universal Flp- and Cre-reporter mouse (Frt-GFP x Ai14-tdTomato) (Figure
4.9A-B). After 2 weeks, we then examined all subcortical targets for the presence of GFP+/Flp+ or
tdTomato+/Cre+ cell bodies corresponding to output from either MOp-ul or MOp-ll, respectively. The
thalamus was excluded from analysis due to its reciprocal connectivity with each injection site,
however topography was still evident in this structure (Figure 4.9C, top row, second panel from left).
Remarkably, several brain regions, including the striatum (Str), zona incerta (ZI), medial accessory
oculomotor nucleus (MA3), APN, RN, basal pontine nucleus (PN), and SC, contained discrete, non-
overlapping populations of GFP+ or tdTomato+ cells that subdivided each structure based on its input
from upper- or lower-limb-related motor cortex (Figure 4.9C). In addition, we also observed structures
such as the bed nucleus of the anterior commissure (BAC), caudal periaqueductal gray (PAG),
pontine reticular nucleus (PRN), and cuneate (CU) and gracile (GR) nuclei, that were labeled by one
pathway, but not the other (Figure 4.9C).
Given this potential to subdivide certain structures based on their somatotopic input, we next
asked whether PN neurons defined by their input from MOp-ul might in turn project to upper-limb
related portions of the cerebellum, thus bridging two somatotopically organized systems and providing
some cross-validation for the specificity of these transsynaptically tagged subpopulations. To test
this, we injected scAAV1-hSyn-Cre into MOp-ul and AAV1-CAG-Flex-GFP into the PN to Cre-
dependently express GFP in this subpopulation of input-defined PN neurons (Figure 4.9D). We then
examined the cerebellum for GFP+ axon terminals. Interestingly, we found these neurons projected
primarily to the contralateral paramedian lobule (PRM), which has been shown to respond to forelimb
107

Figure 4.9. Accessing topographically precise, input-defined cell populations through corticofugal
pathways.
(A) Strategy for labeling cell populations that receive input from upper limb- (green) and lower limb- (red) related
primary motor cortex in Ai14 x Frt-GFP Cre-/Flp-reporter mice.
(B) Injection sites for AAV1-hSyn-Flp in MOp-ul (top panel, green) and scAAV1-hSyn-Cre in MOp-ll (bottom
panel, red) in an Ai14 x Frt-GFP mouse. 100 nL injection volume each, 2 week post-injection survival time.
108

(C) Anterograde transsynaptic labeling of cells that receive input from upper limb- (green) or lower limb- (red)
related primary motor cortex. Many closely apposed, non-overlapping cell populations were observed in mid-
and hindbrain structures, including the PN (second row, right panel). None of the structures shown project back
to motor cortex, with the exception of the thalamus (top row, second panel from left), which may contain both
retrograde and anterograde transsynaptic labeling of cell bodies. Scale bars, 250 µm.
(D) Axonal output of PN neurons that receive input specifically from MOp-ul. Injections of scAAV1-hSyn-Cre
and AAV1-hSyn-Flex-GFP were made into the MOp-ul (red) and PN (green), respectively (80 nL each, 2 week
post-injection survival time). Bottom panel, Cre-dependent GFP expression in Cre+/tom+ PN neurons. Scale
bar, 250 µm.
(E) Axonal targeting in cerebellar cortex. Output was primarily restricted to contralateral PflS and ventral PRM.
PRM targeting appeared as segregated bands (top row, second panel from left). Bottom panels, 10X
magnification of regions in dashed boxes. Scale bars, 500 µm, top row; 250 µm, bottom row.
(F) Schematic summary of axonal projections to cerebellum (blue, posterior view) from PN neurons that receive
input from upper limb-related motor cortex.

stimulation in micromapping studies (Shambes et al., 1978; Odeh et al., 2005) and has been shown to
receive di-synaptic input specifically from the forelimb motor cortex via the pontine nucleus using
multi-synaptic rabies virus tracing (Suzuki et al., 2012) (Figure 4.9E-F). Together, these results
provide evidence that AAV1-transsynaptic tagging may be broadly applied in various corticofugal
pathways to access specific subpopulations of neurons defined by their topographic cortical input.

4.3.5 Application in sparse labeling approaches for single neuron reconstruction
To better characterize populations of neurons that receive input from a given pathway, it may be
useful to recover their individual morphological and axonal targeting features. Given the normally
dense arrangement of these processes, this is greatly facilitated by using a method to sparsely label
only a small number of cells at a time. To achieve sparse labeling in a given Cre-expressing cell
population, we reasoned that co-injections of a low titer Cre-dependent Flp-expressing virus (AAV1-
DIO-Flp) and a high titer Flp-dependent YFP-expressing virus (AAVDJ-fDIO-YFP) could be used to
obtain robust levels of YFP expression in only a few cells (Figure 4.10A). To establish a relationship
between titer and the resulting number of labeled cells, AAV1-DIO-Flp was diluted to a final
concentration of either 7.5 x 10
8
, 10
9
, or 10
10
GC/mL and co-injected with AAVDJ-fDIO-YFP (1.2 x
109

Figure 4.10. Application with sparse labeling approaches for reconstructing single neuron morphology.
(A) For a given Cre+ cell population (red), sparse labeling (green) may be achieved by co-injecting AAV1-DIO-
Flp at increasingly lower titers along with high titer AAVDJ-fDIO-YFP. To establish a titering curve, PV neurons
in V1 were targeted with co-injections of AAVDJ-fDIO-YFP (final titer: 1.2 x 10
13
GC/mL) and AAV1-DIO-Flp
(final titers: 7.5 x 10
10
, 10
9
, or 10
8
GC/mL) in PV-Cre x Ai14 mice (50 nL injection volume, 2 week post-injection
survival time).
110

(B) Examples of cell labeling (green) achieved at each titer step. Red, PV-Cre+/tom+ cells. Blue, fluorescent
Nissl. Scale bar, 250 µm.
(C) 40X magnification of a YFP+ PV neuron labeled in (B, middle panel). Scale bar, 25 µm.
(D) Quantification of the number of YFP+/PV+ cells labeled at each titer step (n = 4 mice each). Error bar, SD.
(E) Injection of scAAV1-hSyn-Cre in the ACA of GAD67-GFP x Ai14 mice transsynaptically labels neurons in
PAG (80 nL injection, 2 week post-injection survival time). Cell density is greatest in the PAGdl (middle panel)
and both inhibitory (GFP+/Tom+) and presumed excitatory (GFP-/Tom+) cell-types are labeled (bottom panel,
40X magnification). Scale bars, 250 µm, middle panel; 50 µm, bottom panel.
(F) Strategy for sparse labeling of input- and genetically-defined cell populations. AAV1-DIO-Flp titer is reduced
to 1.5 x 10
12
GC/mL to achieve sparse anterograde transsynaptic labeling in PAGdl following 80 nL injection in
ACA. Individual Vglut2-Cre+/Flp+ neurons may then be targeted with an injection of AAVDJ-fDIO-YFP (30 nL
volume) to specifically label glutamatergic neurons in PAGdl that receive input from ACA.
(G) Example of a single Vglut2-Cre+/Flp+ neuron labeled in PAGdl (green). An additional neuron was found in
the superior colliculus (arrowhead). Right panels, 40X magnification of YFP+ (green, top panel) and Vglut2-
Cre+/Tom+ (red, bottom panel) neuron in PAGdl. Fluorescent Nissl, blue. Scale bars, 250 µm, left panel; 25
µm, right panels.
(H) 40X magnification of local dendrites and axonal projection of PAGdl neuron in (G). Scale bar, 50 µm.

10
13
GC/mL, 50 nL total volume) in V1 of Ai14 x PV-Cre mice, which express Cre in parvalbumin+
cortical interneurons (Figure 4.10B). At this injection volume, we found that titers around 7.5 x 10
9
GC/mL consistently labeled ~4 PV+ cells per animal, enabling recovery of cell morphology (Figure
4.10C), while titers around 7.5 x 10
8
GC/mL labeled only 1 cell in one out of four animals (Figure
4.10B-D). Therefore, working around a concentration of 10
9
GC/mL and adjusting for the relative
difference in starter population density, it may be possible to apply this approach to any given group
Cre-expressing neurons, including those labeled using anterograde transsynaptic spread of AAV1-
Cre.
In our previous study, we showed that transsynaptic tagging may be applied in different Cre-
expressing transgenic mice to access both input- and genetically-defined cell-types in a given
downstream circuit. We therefore explored an additional means to achieve sparse labeling in cell
populations that meet these criteria. As a test system, we examined the projection from anterior
cingulate cortex (ACA) to the dorsolateral periaqueductal gray (PAGdl) (Figure 4.10E). Following
injections of scAAV1-hSyn-Cre in the ACA of Ai14 x GAD67-GFP mice, a mixture of transsynaptically
labeled GABAergic (GFP+/tdTomato+) and presumed glutamatergic (GFP/tdTomato+) neurons were
111

observed primarily within the PAGdl (Figure 4.10E, bottom panels), suggesting descending
projections from ACA may innervate both cell-types. To select only glutamatergic neurons for
analysis, similar injections may be performed in ACA of Vglut2-Cre mice using AAV1-DIO-Flp and a
secondary injection of AAVDJ-fDIO-YFP within the PAGdl to label input-defined glutamatergic
Cre+/Flp+ cells with YFP. With high titer AAV1-DIO-Flp (e.g. >10
13
GC/mL) and AAVDJ-fDIO-YFP,
these injections are expected to label a large number of cells in PAGdl, however AAV1-DIO-Flp may
be diluted to achieve less transsynaptic labeling, as we previously demonstrated that efficiency of
spread is titer-dependent (Zingg et al., 2017). In particular, previous results revealed that reducing
viral titer from ~10
13
GC/mL to ~10
12
GC/mL reduces transsynaptic labeling efficiency by about 85-
90%. As our test injection of scAAV1-hSyn-Cre at 10
13
titer yielded an average of 35 GFP-
/tdTomato+ cells (presumed glutamatergic neurons) per 300 µm
3
sample space of PAGdl, we
expected that similar injections of AAV1-DIO-Flp at 10
12
titer would tag ~4 Vglut2-Cre+/Flp+ within the
same sample region (see Materials and Methods for details). This sparse population could then be
targeted for robust YFP expression following local injection of high titer AAVDJ-fDIO-YFP (Figure
4.10F). Following this protocol, we were able to label a single input-defined glutamatergic neuron in
PAGdl (Figure 4.10G-H). Combined with methods for tissue clearing and whole-brain imaging, it may
be possible to systematically reconstruct both local morphology and all long-range axonal projections
for any given sparsely labeled group of input- and genetically-defined cells.

4.4 Discussion and Future Directions
In this study, we provide a systematic characterization of the synaptic specificity of AAV1
transneuronal spread and examine its transport efficiency throughout a diverse set of brain pathways.
Overall, we find it to be highly selective in its transduction of post-synaptic neurons and broadly
applicable in all pathways tested, with the exception of several neuromodulatory cell-types (i.e.
serotonergic, cholinergic, and noradrenergic neurons). In addition, we reaffirm the observation that
anterograde transsynaptic spread is unique to AAV1 (and to some extent AAV9), following an
112

expanded comparison with additional AAV serotypes and other common neurotropic viruses (Figure
4.1). As expected, transsynaptic spread is not specific to Cre-expressing AAV1, per se, as we
previously show it works equally well with a Cre-dependent Flp expressing virus (AAV1-EF1a-DIO-
Flp), and in this study, an unconditional Flp-expressing virus (AAV1-hSyn-Flp) in a Flp-reporter mouse
(Frt-GFP; Sousa et al., 2009). Thus, AAV1-mediated transsynaptic tagging may be expanded to
include a variety of recombinase systems to facilitate intersectional approaches for accessing cell-
types based on multiple criteria.
Why might transsynaptic spread be unique to AAV1 and not other serotypes? Previous work
has shown that, following endocytosis at the cell body, AAV1, AAV8, and AAV9 exhibit similar
patterns of trafficking throughout the endosomal system and display a similar capacity for anterograde
axonal transport (Castle et al., 2014b). Nevertheless, AAV1 was observed to undergo trafficking
down the axon with greater frequency than AAV8 or AAV9. Rather than reflecting a mechanistic
difference in transport, however, this was attributed to a greater efficiency of uptake at the host cell
membrane, thus leading to increased concentration in the cell body and a proportional increase in the
fraction of axon-directed AAV1-containing vesicles. In support of this idea, pretreatment of host
neurons with neuraminidase, which increases AAV9 attachment and endocytosis, resulted in a
significant increase in the frequency of AAV9 axon trafficking, even surpassing rates observed for
AAV1. It is therefore likely that, among all of the AAV serotypes tested, AAV1 displays the highest
efficiency of uptake by host cells, and this enables a greater fraction of the virus to be delivered to the
axon terminal for release. Given the rapid decrease in post-synaptic labeling efficiency observed
following small reductions in viral titer (Zingg et al., 2017), AAV1 injected at high titer (e.g. 10
13

GC/mL) may therefore marginally surpass the concentration threshold required within the host cell for
transneuronal spread, while other serotypes fail to reach this at similar titers and injection volumes.
Following this reasoning, future studies seeking to increase the efficiency of transneuronal spread
may focus on enhancing mechanisms of AAV uptake through capsid engineering and/or over-
113

expression of extracellular attachment factors in host neurons (Sun & Schaffer, 2018; Bedbrook et al.,
2018).
In our previous study, we described several lines of evidence that suggested AAV1 may
spread selectively to synaptically connected neurons. In particular, we noted that (1) post-synaptic
labeling was found only in expected target regions, (2) the virus did not appear to leak into cells
adjacent to fibers of passage, and (3) post-synaptic neurons all showed functional pre-synaptic input
in slice recording studies. It is possible, however, that non-connected neurons were overlooked in our
small recording sample. Moreover, based on the observed labeling alone, it is difficult to rule out the
contribution of locally released, extrasynaptic viral spread. We therefore aimed to test the specificity
of viral transport under conditions that might better reveal the relative contribution of extrasynaptic
leakage. To do this, we first performed a more comprehensive slice recording study to examine the
statistical association between post-synaptically labeled neurons and pre-synaptic connectivity, and
between non-labeled neurons and a lack of pre-synaptic connectivity. Using this approach, we may
reveal that labeled neurons have a low probability of pre-synaptic connectivity, indicating leaky viral
spread, or we may reveal a high probability of connectivity for labeled neurons that only becomes
meaningful if neighboring non-labeled neurons show a correspondingly low level of connectivity, thus
avoiding a false-positive association. Following this logic, we found that all of the labeled neurons that
we recorded from (28/28 cells) were mono-synaptically connected to their presynaptic partners, while
only ~46% of neighboring non-labeled cells (13/28) exhibited pre-synaptic input (Figure 4.3D). This
revealed a highly significant association between post-synaptic AAV spread and pre-synaptic
connectivity, suggesting AAV may favor a synaptic mechanism of spread.
Despite recording from a larger population of neurons in the above study, it again remains
possible that we overlooked some fraction of labeled cells that lacked pre-synaptic input. We
therefore aimed to provide a broader screen for capturing viral leakage. To do this, we considered
unidirectional pathways that featured anatomically well-defined connectivity with some, but not all,
cell-types within a heterogenous target structure. In this way, the specificity of viral transport could be
114

screened through hundreds of expected synaptic connections, while also capturing potential non-
specific spread to unconnected cell-types within the same region. To our knowledge, the projections
from the PN and IO to the cerebellar cortex represented one of the few pathways that met these
criteria and offered straightforward identification of expected cell-type labeling. These pathways come
fairly close to the ideal test scenario, though there is some spatial segregation of both cell-types and
axonal projections into different layers. For example, PN mossy fibers terminate exclusively within the
granule layer, while IO climbing fibers terminate primarily within the molecular layer. However, in
each case axons for each pathway are in close apposition to descending purkinje cell axons, as well
as all classes of granule layer cells. In addition, another potential caveat to interpretation is the
specialized nature of the synaptic terminals for each pathway. Rather than forming classical point-to-
point connections, PN mossy fibers form elaborated “rosette” terminal endings that connect with
granule cell dendrites and Golgi cell processes in complex glomeruli. In addition, IO climbing fibers
may be specialized for the high-volume release of glutamate, which plays a putative role in the extra-
synaptic excitation of local interneurons (Szapiro et al., 2007; Coddington et al., 2013; Nietz et al.,
2017). Nevertheless, AAV1 injections through each of these pathways demonstrated a high degree of
synaptic specificity, innervating either the expected granule cell population, but not purkinje cells (PN
pathway, Figure 4.4A), or purkinje cells, but not granule cells (IO pathway, Figure 4.4E). The synaptic
specificity of spread was not entirely perfect, however, as in each case a small fraction of the
unexpected cell-type was labeled. These comprised ~2% of the total population labeled by the other
connected pathway and may provide an estimate of viral leakage in this particular system (Figure
4.4I). Interestingly, we observed labeling of both MLIs and presumed Golgi cells following injections
in IO, however the existence of true synaptic connectivity between climbing fibers and these cell-types
is uncertain (Galliano et al., 2013), as is the potential for AAV1 to be excessively co-released with
glutamate in this pathway, presuming extra-synaptic signaling plays a prominent role in this circuit
(Szapiro et al., 2007; Coddington et al., 2013; Nietz et al., 2017).
115

Finally, to provide evidence for a pre-synaptic mechanism of AAV release, we expressed
TeNT in host cells in V1 to effectively block synaptic vesicle fusion (Schaivo et al., 1992; Schoch et
al., 2001; Yamamoto et al., 2003). We then assessed its impact on the efficiency of AAV1 post-
synaptic transduction in SC (Figure 4.5). Remarkably, we observed a ~94% reduction in
transsynaptic spread. This suggests that, following vesicular trafficking to the axon terminal, AAV1
may affiliate with endosomal compartments that give rise to synaptic vesicles, or may directly
transition to a Rab3+ secretory vesicle (Castle et al., 2014a; Binotti et al., 2016), leading to eventual
co-release with other neurotransmitter-containing vesicles into the synaptic cleft. The marked
reduction in transsynaptic AAV1 spread may reflect a predominately presynaptic mechanism of viral
release, or, given potentially widespread co-localization with other exocytic compartments, may reflect
a statistically greater likelihood of transneuronal transduction following release into the synaptic cleft
versus into the extracellular space. In support of this, we have observed in our lab that a minimum of
~15,000 AAV1-Cre particles must be pressure injected into the extracellular space in order to locally
transduce a single neuron in Ai14 Cre-reporter mice (data not shown). It may therefore be the case
that accumulation of a small number of AAV particles in the synaptic cleft may have a greater chance
of being taken up by a downstream neuron than release of the same number of particles into the
extracellular space. It should be noted here that transsynaptic spread of AAV was not completely
blocked by TeNT expression, suggesting either incomplete co-transduction of the starter neurons with
both viruses, or the possibility that other vesicle release mechanisms that do not require
VAMP2/synaptobrevin-2, such as fusion of multivesicular bodies (MVBs), contribute modestly to
transsynaptic spread (Von Bartheld & Altick, 2011; Raposo & Stoorvogel, 2013; Janas et al., 2016;
Hessvik & Llorente, 2018). These may release their contents extrasynaptically, or into the synaptic
cleft following fusion with the presynaptic membrane in regions adjacent to active zones for synaptic
vesicle docking (Janas et al., 2016; Figure 1.5). Additionally, given this potential synaptic vesicle
mechanism of viral release, it would be interesting to know if viral spread is more efficient in cells that
exhibit higher firing rates or are tonically active. Similarly, might viral spread be enhanced following
116

optogenetic stimulation or activation with DREADDs? Finally, along the lines of establishing cell-type
specific anterograde transsynaptic spread with AAV1, it may be possible to use TeNT expression to
selectively block AAV transport in all non-desired neurons at a given injection site while permitting
transsynaptic spread from a specified Cre-expressing population. This could be achieved by injecting
a “Cre-OFF” TeNT construct that contains lox sites flanking the gene, thus leading to default TeNT
expression in all Cre- cells and a lack of TeNT expression in Cre+ cells, allowing for AAV spread.
Limitations of this approach, however, may include incomplete transduction of all Cre- neurons at the
injection site, as well as concerns for disruption of normal circuit activity in functional applications.
Overall, these results collectively provide deeper insight into the synaptic nature of AAV1
transneuronal spread and suggest that transport through transsynaptic mechanisms contributes to the
bulk (~94-98%) of the observed labeling in each of the three pathways examined here.
Previously, we characterized the transsynaptic spread of AAV1 through glutamatergic
projection pathways (e.g. corticofugal, retino-collicular, colliculo-thalamic), revealing its capacity for
efficient transduction of both excitatory and inhibitory cell-types in all downstream targets. To what
extent might this technique be applied to other pathways utilizing different types of synapses? Here
we show that AAV1 is equally capable of transducing either excitatory (VM) or inhibitory (SNr) cell-
types downstream of two different respective GABAergic projection pathways (Str → SNr, and SNr →
VM; Figure 4.6A-B). Quantification of the percentage of labeled neurons in each target structure
revealed transduction efficiencies that were comparable to previously reported glutamatergic
pathways (~41% in SNr; ~36% in VM; Figure 4.6F), suggesting that AAV1 is effectively transported
across inhibitory synaptic connections and may be applied in these systems. On the other hand,
injections in three different neuromodulatory cell populations failed to yield efficient transsynaptic
labeling in their respective downstream targets (Figure 4.6C-E). For example, cholinergic neurons in
the basal forebrain labeled only a sparse population of cells in V1 (~0.65% of total number of cells
were tdTomato+; Figure 4.6C, F), while serotonergic neurons in DR and noradrenergic neurons in LC
exhibited almost no transsynaptic spread (Figure 4.6D,E). Several factors may account for this lack of
117

efficiency. First, given the sensitivity of this technique to viral titer and efficiency of uptake at the host
cell body (Castle et al., 2014), it may be possible that these neuromodulatory cell-types express
different proportions of extracellular receptors that show lower affinity for AAV1 binding and import.
Second, it is possible that AAV1 enters these cells with high efficiency, but, due to internal trafficking
differences or vesicle release properties, AAV fails to escape synaptic structures at sufficient
concentrations. Lastly, many of these neuromodulatory classes release neurotransmitter into the
extracellular space through specialized axonal varicosities, rather than through point-to-point synaptic
connections (Agnati et al., 1995; Arroyo et al., 2014). This so-called volume transmission may result
in diffuse release of AAV into the extracellular space, rather than the synaptic cleft, resulting in lower
probabilities for successful post-synaptic transduction of downstream neurons. Future studies may
seek to test additional AAV serotypes or engineer novel capsids that show enhanced transduction and
more efficient transsynaptic spread from these classes of projection neurons.
Finally, we highlight the potential application of AAV1 transsynaptic tagging in a variety of
other brain systems and incorporate its use with new strategies for achieving dual-tracer reporting
(Figure 4.9) or sparse labeling of input- and genetically-defined neurons (Figure 4.10). We show that
tagging approaches may be used in (1) ascending pathways from the brainstem to the thalamus
(Figure 4.7), (2) in descending pathways from the brain to the spinal cord (Figure 4.8), and (3) in
corticofugal pathways targeting mid- and hindbrain structures (Figure 4.9). Most pathways in these
systems are unidirectional and therefore provide extensive opportunity for the application of
transsynaptic tagging without concerns for retrograde transport of AAV.
Future directions
Overall, this technique offers a valuable alternative to existing approaches for anterograde
transsynaptic mapping of circuits (e.g. VSVg or HSV129). In its current form, it can be used to
conveniently access and manipulate spatially restricted, input-defined neurons that would otherwise
be challenging to study. There are two major limitations to this technique, however. First, it cannot be
118

used in reciprocally connected circuits due to the fact that AAV1 is also capable of retrograde
transport. Second, while we are able to determine post-synaptic targets for injected brain regions, we
cannot specify particular cell-types for transsynaptic mapping. Thus, to broaden the application of this
technique, these two issues need to be resolved.
One potential path forward may be to manipulate capsid-receptor interactions to achieve
selective import of AAV into genetically specified neurons. If an essential set of receptors for AAV
uptake can be identified and genetically knocked-out without physiological consequence, then they
could be conditionally restored in specific cell-types to prime them for AAV transduction. If uptake
efficiency is sufficient, this could support anterograde transsynaptic transfer of AAV through this
specific class of neurons, and it could be achieved without concern for retrograde transport,
considering axon terminals within the injection site region would not express the necessary receptors
for uptake. Alternatively, if knockout of these receptors is too disruptive, then expression of novel
transmembrane proteins combined with novel capsids engineered to recognize them might provide a
way around this. Interestingly, a receptor (LY6A) found only in vascular endothelial cells, but not
neurons, was reported to enable robust uptake of AAV-PHP.B (Huang et al., 2019), permitting it to
cross the blood-brain barrier with high efficiency. In addition, ectopic expression of this receptor in
cultured cells not normally permissive to AAV-PHP.B transduction enabled robust infection. It may be
possible then to exploit this interaction by over-expressing LY6A in a specific Cre+ population of
neurons using viral injection. These cells may then be primed for selective uptake and transport of
AAV-PHP.B-Flp, which may be transsynaptically transported to downstream neurons that could
subsequently be studied with Flp-dependent vectors. For enhanced selectivity, these injections may
take place in an AAVR-/- knockout mouse (Pillay et al., 2016) to eliminate uptake of AAV-PHP.B by
neighboring neurons. Gaining a more complete understanding of the interactions between variable
regions of capsid proteins and their extracellular receptors will help guide the future development of
these tools.

119

REFERENCES

Aboitiz, F., and Montiel, J. (2003). One hundred million years of interhemispheric communication:
The history of the corpus callosum. Brazilian J. Med. Biol. Res. 36, 409–420.
Agnati, L. F., Zoli, M., Strömberg, I., & Fuxe, K. (1995). Intercellular communication in the brain:
Wiring versus volume transmission. Neuroscience, 69(3), 711–726.
https://doi.org/10.1016/0306-4522(95)00308-6
Arroyo, S., Bennett, C., & Hestrin, S. (2014). Nicotinic modulation of cortical circuits. Frontiers in
Neural Circuits, 8(March), 30. https://doi.org/10.3389/fncir.2014.00030
Aschauer, D.F., Kreuz, S., and Rumpel, S. (2013). Analysis of transduction efficiency, tropism and
axonal transport of AAV serotypes 1, 2, 5, 6, 8 and 9 in the mouse brain. PLoS One 8,
e76310.
Atlan, G., Terem, A., Peretz-Rivlin, N., Groysman, M., & Citri, A. (2016). Mapping synaptic cortico-
claustral connectivity in the mouse. The Journal of Comparative Neurology, 520, 633–655.
http://doi.org/10.1002/cne.23997
Barnea, G., Johnson, M., Szymanski, J. R., Hartmann, G. G., Chou-Freed, C., Talay, M., … Snell, N.
J. (2017). Transsynaptic Mapping of Second-Order Taste Neurons in Flies by trans-Tango.
Neuron, 96(4), 783–795.e4. https://doi.org/10.1016/j.neuron.2017.10.011
Bedbrook, C. N., Deverman, B. E., & Gradinaru, V. (2018). Viral Strategies for Targeting the Central
and Peripheral Nervous Systems. Annual Review of Neuroscience, 41(1), annurev-neuro-
080317-062048. https://doi.org/10.1146/annurev-neuro-080317-062048
Beier, K.T., and Saunders, A. (2011). Anterograde or retrograde transsynaptic labeling of CNS
neurons with vesicular stomatitis virus vectors. Proc. Natl. Acad. Sci. USA. 108, 15414–
15419.
Beltramo, R., & Scanziani, M. (2019). A collicular visual cortex: Neocortical space for an ancient
midbrain visual structure. Science, 363(6422), 64–69.
https://doi.org/10.1126/SCIENCE.AAU7052
Berry, G. E., & Asokan, A. (2016). Cellular transduction mechanisms of adeno-associated viral
vectors. Current Opinion in Virology, 21, 54–60. https://doi.org/10.1016/j.coviro.2016.08.001
Beyeler, A., Chang, C.-J., Silvestre, M., Lévêque, C., Namburi, P., Wildes, C. P., & Tye, K. M.
(2018). Organization of Valence-Encoding and Projection-Defined Neurons in the Basolateral
Amygdala. Cell Reports, 22(4), 905–918. http://doi.org/10.1016/j.celrep.2017.12.097
Billwiller, F., Renouard, L., Clement, O., Fort, P., & Luppi, P. H. (2017). Differential origin of the
activation of dorsal and ventral dentate gyrus granule cells during paradoxical (REM) sleep in
the rat. Brain Structure and Function, 222(3), 1495–1507. http://doi.org/10.1007/s00429-016-
1289-7
Binotti, B., Jahn, R., Jia, J., & Chua, E. (2016). Functions of Rab Proteins at Presynaptic Sites.
Cells, 1–10. https://doi.org/10.3390/cells5010007
120

Bokor, H., Frère, S. G. A., Eyre, M. D., Slézia, A., Ulbert, I., Lüthi, A., & Acsády, L. (2005). Selective
GABAergic control of higher-order thalamic relays. Neuron, 45(6), 929–940.
https://doi.org/10.1016/j.neuron.2005.01.048
Braz, J. M., Rico, B., & Basbaum, A. I. (2002). Transneuronal tracing of diverse CNS circuits by Cre-
mediated induction of wheat germ agglutinin in transgenic mice. Proceedings of the National
Academy of Sciences, 99(23), 15148–15153. https://doi.org/10.1073/pnas.222546999
Broadwell, R. D., & Balin, B. J. (1985). Endocytic and exocytic pathways of the neuronal secretory
process and trans synaptic transfer of wheat germ agglutinin ‐horseradish peroxidase in vivo.
Journal of Comparative Neurology, 242(4), 632–650. https://doi.org/10.1002/cne.902420410
Brown, K. M., Sugihara, I., Shinoda, Y., & Ascoli, G. A. (2012). Digital Morphometry of Rat
Cerebellar Climbing Fibers Reveals Distinct Branch and Bouton Types. Journal of
Neuroscience, 32(42), 14670–14684. https://doi.org/10.1523/jneurosci.2018-12.2012
Brown, SP; Mathur, B; Olsen, SR; Luppi, P-H´; Bickford, ME; Citri, A. (2017). New Breakthroughs in
Understanding the Role of Functional Interactions between the Neocortex and the
Claustrum. The Journal of Neuroscience, 37(45), 10877–10881.
http://doi.org/DOI:10.1523/JNEUROSCI.1837-17.2017
Burgos-Robles, A., Kimchi, E. Y., Izadmehr, E. M., Porzenheim, M. J., Ramos-Guasp, W. A., Nieh,
E. H., … Tye, K. M. (2017). Amygdala inputs to prefrontal cortex guide behavior amid
conflicting cues of reward and punishment. Nature Neuroscience, 20(6), 824–835.
http://doi.org/10.1038/nn.4553
Callaway, X.E.M., and Luo, L. (2015). Monosynaptic Circuit Tracing with Glycoprotein-Deleted
Rabies Viruses. J. Neurosci. 35, 8979–8985.
Carey, R. G., & Neal, T. L. (1986). Reciprocal connections between the claustrum and visual
thalamus in the tree shrew (Tupaia glis). Brain Research, 386(1–2), 155–168.
http://doi.org/10.1016/0006-8993(86)90152-6
Castejon, O.J. and Castejon, H.V. (2000). Correlative microscopy of cerebellar Golgi cells. Biocell.
24, 13-30.
Castle, M.J., Gershenson, Z.T., Giles, A.R., Holzbaur, E.L.F., and Wolfe, J.H. (2014). Adeno-
associated virus serotypes 1, 8, and 9 share conserved mechanisms for anterograde and
retrograde axonal transport. Hum. Gene Ther. 25, 705–720.
Castle, M.J., Perlson, E., Holzbaur, E.L., and Wolfe, J.H. (2014). Long-distance Axonal Transport of
AAV9 Is Driven by Dynein and Kinesin-2 and Is Trafficked in a Highly Motile Rab7-positive
Compartment. Mol. Ther. 22, 1–13.
Castro-Alamancos, M.A., and Favero, M. (2016). Whisker-related afferents in superior colliculus. J.
Neurophysiol. 115, 2265–2279.
Coddington, L. T., Rudolph, S., VandeLune, P., Overstreet-Wadiche, L., & Wadiche, J. I. (2013).
Spillover-Mediated Feedforward Inhibition Functionally Segregates Interneuron Activity.
Neuron, 78(6), 1050–1062. https://doi.org/10.1016/j.neuron.2013.04.019
121

Colquhoun, D. (2014). An investigation of the false discovery rate and the misinterpretation of P
values. Royal Society Open Science, 1–15. http://doi.org/10.1098/rsos.140216
Comoli, E., Das Neves Favaro, P., Vautrelle, N., Leriche, M., Overton, P.G., and Redgrave, P.
(2012). Segregated anatomical input to sub-regions of the rodent superior colliculus
associated with approach and defense. Front. Neuroanat. 6, 1-19.
Crick, F. C., & Koch, C. (2005). What is the function of the claustrum? Philosophical Transactions of
the Royal Society of London - Series B: Biological Sciences, 360(1458), 1271–1279.
http://doi.org/10.1098/rstb.2005.1661
Curanovic, D., & Enquist, L. W. (2009). Directional transneuronal spread of α-herpesvirus infection.
Future Virology, 4(6), 591–603. https://doi.org/10.2217/fvl.09.62
Dean, P., Mitchell, I.J., and Redgrave, P. (1988). Responses resembling defensive behaviour
produced by microinjection of glutamate into superior colliculus of rats. Neuroscience 24,
501–510.
Defalco, J., Tomishima, M., Liu, H., Zhao, C., Cai, X.L., Marth, J.D., Enquist, L., and Friedman, J.M.
(2010). Virus-Assisted Mapping of Neural Inputs to a Feeding Center in the Hypothalamus.
Science 291, 2608-2613.
Deverman, B. E., Pravdo, P. L., Simpson, B. P., Kumar, S. R., Chan, K. Y., Banerjee, A., …
Gradinaru, V. (2016). Cre-dependent selection yields AAV variants for widespread gene
transfer to the adult brain. Nature Biotechnology, advance on(2), 1–7.
https://doi.org/10.1038/nbt.3440
Dinopoulos, A., Papadopoulos, G. C., Michaloudi, H., Parnavelas, J. G., Uylings, H. B. M., &
Karamanlidis, A. N. (1992). Claustrum Connections in the Hedgehog (Erinaceus europaeus)
Brain: Cytoarchitecture and With Cortical and Subcortical Structures. The Journal of
Comparative Neurology, 316, 187–205.
Dong, H.-W. (2007). The Allen Reference Atlas: A Digital Color Brain Atlas Of The C57BL/6J Male
Mouse (Hoboken: John Wiley and Sons, Inc.).
Dudek, A. M., Pillay, S., Puschnik, A. S., Nagamine, C. M., Cheng, F., Qiu, J., … Vandenberghe, L.
H. (2018). An alternate route for adeno-associated virus entry independent of AAVR. Journal
of Virology, 92(7), JVI.02213-17. https://doi.org/10.1128/JVI.02213-17
Ekstrand, M.I., Enquist, L.W., and Pomeranz, L.E. (2008). The alpha-herpesviruses: molecular
pathfinders in nervous system circuits. Trends Mol. Med. 14, 134–140.
Fenno, L.E., Mattis, J., Ramakrishnan, C., Hyun, M., Lee, S.Y., He, M., Tucciarone, J.,
Selimbeyoglu, A., Berndt, A., Grosenick, L., et al. (2014). Targeting cells with single vectors
using multiple-feature Boolean logic. Nat. Methods 11, 763–772.
Galliano, E., Baratella, M., Sgritta, M., Ruigrok, T. J. H., Haasdijk, E. D., Hoebeek, F. E., … De
Zeeuw, C. I. (2013). Anatomical investigation of potential contacts between climbing fibers
and cerebellar Golgi cells in the mouse. Frontiers in Neural Circuits, 7(April), 1–9.
https://doi.org/10.3389/fncir.2013.00059
122

Giber, K., Slezia, A., Bokor, H., Bodor, A. l., Ludanyi, A., Katona, I., & Acsady, L. (2007).
Heterogeneous Output Pathways Link the Anterior Pretectal Nucleus With the Zona Incerta
and the Thalamus in Rat. The Jounal of Comparative Neurology, 505(July 2007), 337–351.
https://doi.org/10.1002/cne
Goll, Y., Atlan, G., & Citri, A. (2015). Attention: the claustrum. Trends in Neurosciences, 1–10.
http://doi.org/10.1016/j.tins.2015.05.006
Gradinaru, V., Zhang, F., Ramakrishnan, C., Mattis, J., Prakash, R., Diester, I., Goshen, I.,
Thompson, K.R., and Deisseroth, K. (2010). Molecular and cellular approaches for
diversifying and extending optogenetics. Cell 141, 154–165.
Gronostajski, R.M., and Sadowski, P.D. (1985). The FLP recombinase of the Saccharomyces
cerevisiae 2 microns plasmid attaches covalently to DNA via a phosphotyrosyl linkage. Mol.
Cell. Biol. 5, 3274–3279.
Hangya, B., Ranade, S. P., Lorenc, M., & Kepecs, A. (2015). Central Cholinergic Neurons Are
Rapidly Recruited by Reinforcement Feedback. Cell, 162(5), 1155–1168.
http://doi.org/10.1016/j.cell.2015.07.057
Harris, J. A., Wook Oh, S., and Zeng, H. (2012). Adeno-associated viral vectors for anterograde
axonal tracing with fluorescent proteins in nontransgenic and Cre driver mice. Curr. Protoc.
Neurosci. 1, 1–18.
Hessvik, N. P., & Llorente, A. (2018). Current knowledge on exosome biogenesis and release.
Cellular and Molecular Life Sciences, 75(2), 193–208. https://doi.org/10.1007/s00018-017-
2595-9
Hu, H. (2016). Reward and Aversion. Annual Review of Neuroscience, 39(1), 297–324.
http://doi.org/10.1146/annurev-neuro-070815-014106
Huang, Q., Chan, K. Y., Tobey, I. G., Chan, Y. A., Poterba, T., Boutros, C. L., … Deverman, B. E.
(2019). Delivering genes across the blood-brain barrier: LY6A, a novel cellular receptor for
AAV-PHP.B capsids. BioRxiv, 538421. https://doi.org/10.1101/538421
Huang, T. H., Niesman, P., Arasu, D., Lee, D., De La Cru, A. L., Callejas, A., … Lois, C. (2017).
Tracing neuronal circuits in transgenic animals by transneuronal control of transcription
(Tract). ELife, 6, 1–23. https://doi.org/10.7554/eLife.32027
Huerta, M.F., and Harting, J.K. (1984). Connectional organization of the superior colliculus. Trends
Neurosci. 7, 286–289.
Hutson, T.H., Kathe, C., and Moon, L.D.F. (2015). Trans-neuronal transduction of spinal neurons
following cortical injection and anterograde axonal transport of a bicistronic AAV1 vector.
Gene Ther. 23, 231–236.
Ibrahim, L.A., Mesik, L., Ji, X.Y., Fang, Q., Li, H.F., Li, Y.T., Zingg, B., Zhang, L.I., and Tao, H.W.
(2016). Cross-Modality Sharpening of Visual Cortical Processing through Layer-1-Mediated
Inhibition and Disinhibition. Neuron 89, 1031-1045.
123

Isayeva, T., & Ren, C. (2005). Recombinant Adeno-Associated Virus 2-Mediated Antiangiogenic
Prevention in a Mouse Model of Intraperitoneal Ovarian Cancer. Clinical Cancer Research,
11, 1342–1347.
Janas, A. M., Sapoń, K., Janas, T., Stowell, M. H. B., & Janas, T. (2016). Exosomes and other
extracellular vesicles in neural cells and neurodegenerative diseases. Biochimica et
Biophysica Acta - Biomembranes, 1858(6), 1139–1151.
https://doi.org/10.1016/j.bbamem.2016.02.011
Ji, X.Y., Zingg, B., Mesik, L., Xiao, Z., Zhang, L.I., and Tao, H.W. (2016). Thalamocortical
Innervation Pattern in Mouse Auditory and Visual Cortex: Laminar and Cell-Type Specificity.
Cereb Cortex. 26, 2612-2625.
Jiang, X., Wang, G., Lee, A. J., Stornetta, R. L., & Zhu, J. J. (2013). The organization of two new
cortical interneuronal circuits. Nature Neuroscience, 16(2), 210–218.
http://doi.org/10.1038/nn.3305
Jimenez, J. C., Su, K., Goldberg, A. R., Luna, V. M., Biane, J. S., Ordek, G., … Kheirbek, M. A.
(2018). Anxiety Cells in a Hippocampal-Hypothalamic Circuit. Neuron, 0(0), 670–683.
http://doi.org/10.1016/j.neuron.2018.01.016
Junyent, F., and Kremer, E.J. (2015). CAV-2 - Why a canine virus is a neurobiologist’s best friend.
Curr. Opin. Pharmacol. 24, 86–93.
Kanichay, R. T., & Silver, R. A. (2008). Synaptic and Cellular Properties of the Feedforward
Inhibitory Circuit within the Input Layer of the Cerebellar Cortex. Journal of Neuroscience,
28(36), 8955–8967. https://doi.org/10.1523/jneurosci.5469-07.2008
Katz, L. C. (1987). Local circuitry of identified projection neurons in cat visual cortex brain slices.
Journal of Neuroscience, 7(4), 1223–1249.
Kim, J., Matney, C. J., Roth, R. H., & Brown, S. P. (2016). Synaptic Organization of the Neuronal
Circuits of the Claustrum. Journal of Neuroscience, 36(3), 773–784.
http://doi.org/10.1523/JNEUROSCI.3643-15.2016
Kitanishi, T., & Matsuo, N. (2017). Organization of the Claustrum-to-Entorhinal Cortical Connection
in Mice. The Journal of Neuroscience, 37(2), 269–280.
http://doi.org/10.1523/JNEUROSCI.1360-16.2017
Kotterman, M.A., and Schaffer, D. V. (2014). Engineering adeno-associated viruses for clinical gene
therapy. Nat. Rev. Genet. 15, 445–451.
Kratchmarov, R., Taylor, M. P., & Enquist, L. W. (2012). Making the case: Married versus Separate
models of alphaherpes virus anterograde transport in axons. Reviews in Medical Virology,
22(6), 378–391. https://doi.org/10.1002/rmv.1724
LeVay, S., & Sherk, H. (1981). The visual claustrum of the cat. I. Structure and connections. The
Journal of Neuroscience, 1(9), 956–80. Retrieved from
http://www.ncbi.nlm.nih.gov/pubmed/6169810
124

Liang, F., Xiong, X.R., Zingg, B., Ji, X., Zhang, L.I., and Tao, H.W. (2015). Sensory Cortical Control
of a Visually Induced Arrest Behavior via Corticotectal Projections. Neuron 86, 1–13.
Lo, L., and Anderson, D.J. (2011). A cre-dependent, anterograde transsynaptic viral tracer for
mapping output pathways of genetically marked neurons. Neuron 72, 938–950.
Loeb, J. E., Cordier, W. S., Harris, M. E., Weitzman, M. D., & Hope, T. J. (2002). Enhanced
Expression of Transgenes from Adeno-Associated Virus Vectors with the Woodchuck
Hepatitis Virus Posttranscriptional Regulatory Element: Implications for Gene Therapy.
Human Gene Therapy, 10(14), 2295–2305. https://doi.org/10.1089/10430349950016942
Loscher, W., Gillard, M., Sands, Z. A., Kaminski, R. M., Lo, W., & Klitgaard, H. (2016). Synaptic
Vesicle Glycoprotein 2A Ligands in the Treatment of Epilepsy and Beyond. CNS Drugs, 30,
1055–1077. https://doi.org/10.1007/s40263-016-0384-x
Lu, S. M., & Lin, R. C. S. (1993). Thalamic afferents of the rat barrel cortex: A light-and electron-
microscopic study using phaseolus vulgaris leucoagglutinin as an anterograde tracer.
Somatosensory & Motor Research, 10(1), 1–16. https://doi.org/10.3109/08990229309028819
Luppi, P. H., Billwiller, F., & Fort, P. (2017). Selective activation of a few limbic structures during
paradoxical (REM) sleep by the claustrum and the supramammillary nucleus: evidence and
function. Current Opinion in Neurobiology, 44, 59–64.
http://doi.org/10.1016/j.conb.2017.03.002
Madisen, L., Zwingman, T.A., Sunkin, S.M., Oh, S.W., Zariwala, H.A., Gu, H., Ng, L.L., Palmiter,
R.D., Hawrylycz, M.J., Jones, A.R., Lein, E.S., and Zeng, H. (2009). A robust and high-
throughput Cre reporting and characterization system for the whole mouse brain. Nat.
Neurosci. 13, 133–140.
Markowitsch, H. J., Irle, E., Bang-Olsen, R., & Flindt-Egebak, P. (1984). Claustral efferents to the
cat’s limbic cortex studied with retrograde and anterograde tracing techniques.
Neuroscience, 12(2), 409–425.
Mathews, P. J., Lee, K. H., Peng, Z., Houser, C. R., & Otis, T. S. (2012). Effects of Climbing Fiber
Driven Inhibition on Purkinje Neuron Spiking. Journal of Neuroscience, 32(50), 17988–
17997. https://doi.org/10.1523/jneurosci.3916-12.2012
Mathur, B. N. (2014). The claustrum in review. Frontiers in Systems Neuroscience, 8(April), 48.
http://doi.org/10.3389/fnsys.2014.00048
Mathur, B. N., Caprioli, R. M., & Deutch, A. Y. (2009). Proteomic analysis illuminates a novel
structural definition of the claustrum and insula. Cerebral Cortex, 19(10), 2372–2379.
http://doi.org/10.1093/cercor/bhn253
Matias, S., Lottem, E., Dugué, G. P., & Mainen, Z. F. (2017). Activity patterns of serotonin neurons
underlying cognitive flexibility. ELife, 6, 1–24. http://doi.org/10.7554/eLife.20552
McCarty, D. M. (2008). Self-complementary AAV vectors; advances and applications. Molecular
Therapy, 16(10), 1648–1656. https://doi.org/10.1038/mt.2008.171
125

Morin, L.P., and Studholme, K.M. (2014). Retinofugal Projections in the Mouse. J. Comp. Neurol. 00,
1–21.Nassi, J.J., Cepko, C.L., Born, R.T., and Beier, K.T. (2015). Neuroanatomy goes viral!
Front. Neuroanat. 9, 80.
Nassi, J.J., Cepko, C.L., Born, R.T., and Beier, K.T. (2015). Neuroanatomy goes viral! Front.
Neuroanat. 9, 80.
Nietz, A. K., Vaden, J. H., Coddington, L. T., Overstreet-Wadiche, L., & Wadiche, J. I. (2017). Non-
synaptic signaling from cerebellar climbing fibers modulates Golgi cell activity. ELife, 6, 1–22.
https://doi.org/10.7554/eLife.29215.001
Odeh, F., Ackerley, R., Bjaalie, J., & Apps, R. (2005). Pontine Maps Linking Somatosensory and
Cerebellar Cortices Are in Register with Climbing Fiber Somatotopy. Journal of
Neuroscience, 25(24), 5680–5690. https://doi.org/10.1523/JNEUROSCI.0558-05.2005
Oh, S.W., Harris, J.A., Ng, L., Winslow, B., Cain, N., Mihalas, S., Wang, Q., Lau, C., Kuan, L.,
Henry, A.M., et al. (2014). A mesoscale connectome of the mouse brain. Nature 508, 207–
214.
Olson, C. R., & Graybiel, a M. (1980). Sensory maps in the claustrum of the cat. Nature. Retrieved
from http://www.ncbi.nlm.nih.gov/pubmed/7442793
Oyibo, H.K., Znamenskiy, P., Oviedo, H. V, Enquist, L.W., and Zador, A.M. (2014). Long-term Cre-
mediated retrograde tagging of neurons using a novel recombinant pseudorabies virus.
Front. Neuroanat. 8, 86.
Palay S.L. & Chan-Palay V. 1974. Cerebellar Cortex: Cytology and Organization. Springer-Verlag
Berlin Heidelberg.
Perea, G., Navarrete, M., and Araque, A. (2009). Tripartite synapses: astrocytes process and control
synaptic information. Trends Neurosci. 32, 421–431.
Petersen, C. C. H. (2007). The functional organization of the barrel cortex. Neuron, 56(2), 339–355.
https://doi.org/10.1016/j.neuron.2007.09.017
Petreanu, L., Mao, T., Sternson, S.M., and Svoboda, K. (2009). The subcellular organization of
neocortical excitatory connections. Nature 457, 1142–1145.
Pillay, S., & Carette, J. E. (2017). Host determinants of adeno-associated viral vector entry. Current
Opinion in Virology, 24, 124–131. https://doi.org/10.1016/j.coviro.2017.06.003
Pillay, S., Meyer, N.L., Puschnik, A.S., Davulcu, O., Diep, J., Ishikawa, Y., Jae, L.T., Wosen, J.E.,
Nagamine, C.M., Chapman, M.S., et al. (2016). An essential receptor for adeno-associated
virus infection. Nature 530, 108–112.
Polack, P.-O., Friedman, J., & Golshani, P. (2013). Cellular mechanisms of brain state-dependent
gain modulation in visual cortex. Nature Neuroscience, 16(9), 1331–1339.
https://doi.org/10.1038/nn.3464
Raposo, G., & Stoorvogel, W. (2013). Extracellular vesicles: Exosomes, microvesicles, and friends.
Journal of Cell Biology, 200(4), 373–383. https://doi.org/10.1083/jcb.201211138
126

Rees, H., & Roberts, M. H. T. (1993). The anterior pretectal nucleus: a proposed role in sensory
processing. Pain, 53(2), 121–135. https://doi.org/10.1016/0304-3959(93)90072-W
Remedios, R., Logothetis, N. K., & Kayser, C. (2010). Unimodal responses prevail within the
multisensory claustrum. Journal of Neuroscience, 30(39), 12902–12907.
http://doi.org/10.1523/JNEUROSCI.2937-10.2010
Remedios, R., Logothetis, N. K., & Kayser, C. (2014). A role of the claustrum in auditory scene
analysis by reflecting sensory change. Frontiers in Systems Neuroscience, 8(April), 44.
http://doi.org/10.3389/fnsys.2014.00044
Renouard, L., Billwiller, F., Ogawa, K., Clement, O., Camargo, N., Abdelkarim, M., … Luppi, P.-H.
(2015). The supramammillary nucleus and the claustrum activate the cortex during REM
sleep. Science Advances, 1(3), e1400177–e1400177. http://doi.org/10.1126/sciadv.1400177
Reppucci, C. J., & Petrovich, G. D. (2016). Organization of connections between the amygdala,
medial prefrontal cortex, and lateral hypothalamus: a single and double retrograde tracing
study in rats. Brain Structure and Function, 221(6), 2937–2962.
http://doi.org/10.1007/s00429-015-1081-0
Reser, D. H., Majka, P., Snell, S., Chan, J., Watkins, K., Worthy, K., … Rosa, M. G. P. (2016).
Topography of claustrum and insula projections to medial prefrontal and anterior cingulate
cortex of the common marmoset (Callithrix jacchus ). Journal of Comparative Neurology, 00,
n/a-n/a. http://doi.org/10.1002/cne.24009
Rothermel, M., Brunert, D., Zabawa, C., Diaz-Quesada, M., and Wachowiak, M. (2013). Transgene
Expression in Target-Defined Neuron Populations Mediated by Retrograde Infection with
Adeno-Associated Viral Vectors. J. Neurosci. 33, 15195–15206.
Sadowski, M., Morys, J., Jakubowska-sadowska, K., & Narkiewicz, O. (1997). Rat’s claustrum
shows two main cortico-related zones. Brain Research, 147–152.
Sahibzada, N., Dean, P., and Redgrave, P. (1986). Movements resembling orientation or avoidance
elicited by electrical stimulation of the superior colliculus in rats. J Neurosci. 6, 723-733.
Salegio, E.A., Samaranch, L., Kells, A.P., Mittermeyer, G., San Sebastian, W., Zhou, S., Beyer, J.,
Forsayeth, J., and Bankiewicz, K.S. (2013). Axonal transport of adeno-associated viral
vectors is serotype-dependent. Gene Ther. 20, 348–352.
Schenberg, L.C., Povoa, R.M.F., Costa, A.L.P., Caldellas, A. V., Tufik, S., and Bittencourt, A.S.
(2005). Functional specializations within the tectum defense systems of the rat. Neurosci.
Biobehav. Rev. 29, 1279–1298.
Schiavo, G., Benfenati, F., Poulain, B., Rossetto, O., Laureto, P. P. de, DasGupta, B. R., &
Montecucco, C. (1992). Tetanus and botulinum-B neurotoxins block neurotransmitter release
by proteolytic cleavage of synaptobrevin. Nature.
Schoch, S., Deák, F., Königstorfer, A., Mozhayeva, M., Sara, Y., Südhof, T. C., & Kavalali, E. T.
(2001). SNARE function analyzed in synaptobrevin/VAMP knockout mice. Science. Science.
https://doi.org/10.1126/science.1064335
127

Schulman, J. A., & Bloom, F. E. (1981). Golgi cells of the cerebellum are inhibited by inferior olive
activity. Brain Research, 210, 350–355.
Schwab, M. E., Javoy-Agid, F., & Agid, Y. (1978). Labeled wheat germ agglutinin (WGA) as a new,
highly sensitive retrograde tracer in the rat brain hippocampal system. Brain Research,
152(1), 145–150. https://doi.org/10.1016/0006-8993(78)90140-3
Schwarz, L. A., Miyamichi, K., Gao, X.J., Beier, K.T., Weissbourd, B., DeLoach, K.E., Ren, J.,
Ibanes, S., Malenka, R.C., Kremer, E.J., et al. (2015). Viral-genetic tracing of the input–
output organization of a central noradrenaline circuit. Nature 524, 88-92.
Shambes, G., Gibson, J., & Welker, W. (1978). Fractured Somatotopy in Granule Cell Tactile Areas
of Rat Cerebellar Hemispheres Revealed by Micromapping. Brain, Behavior and Evolution,
15, 116–140.
Shang, C., Liu, Z., Chen, Z., Shi, Y., Wang, Q., Liu, S., Li, D., and Cao, P. (2015). A parvalbumin-
positive excitatory visual pathway to trigger fear responses in mice. Science. 348, 1472–
1477.
Shi, C., & Cassell, M. D. (1998). Connections of the Anterior and Posterior Insula of rat. Journal of
Comparative Neurology, 468(May), 440–468.
Simmons, P.A., Lemmon, V., and Pearlman, A.L. (1982). Afferent and efferent connections of the
striate and extrastriate visual cortex of the normal and reeler mouse. J. Comp. Neurol. 211,
295–308.
Smith, J. B., & Alloway, K. D. (2010). Functional specificity of claustrum connections in the rat:
interhemispheric communication between specific parts of motor cortex. Journal of
Neuroscience, 30(50), 16832–16844. http://doi.org/10.1523/JNEUROSCI.4438-10.2010
Smith, J. B., & Alloway, K. D. (2014). Interhemispheric claustral circuits coordinate sensory and
motor cortical areas that regulate exploratory behaviors. Frontiers in Systems Neuroscience,
8(May), 1–14. http://doi.org/10.3389/fnsys.2014.00093
Soudais, C., Laplace-Builhe, C., Kissa, K., and Kremer, E.J. (2001). Preferential transduction of
neurons by canine adenovirus vectors and their efficient retrograde transport in vivo. FASEB
J. 15, 2283–2285.
Sousa, V. H., Miyoshi, G., Hjerling-Leffler, J., Karayannis, T., & Fishell, G. (2009). Characterization
of Nkx6-2-derived neocortical interneuron lineages. Cerebral Cortex, 19(SUPPL. 1), 1–10.
https://doi.org/10.1093/cercor/bhp038
Sturrock, R.R. (1976). Light microscopic identification of immature glial cells in semithin sections of
the developing mouse corpus callosum. J. Anat. 122, 521–537.
Sun, S. & Schaffer, D. V. (2018). Engineered viral vectors for functional interrogation, deconvolution,
and manipulation of neural circuits. Current Opinion in Neurobiology, 50, 163–170.
https://doi.org/10.1016/j.conb.2017.12.011
128

Sun, Y., Nguyen, A. Q., Nguyen, J. P., Le, L., Saur, D., Choi, J., … Xu, X. (2014). Cell-type-specific
circuit connectivity of hippocampal CA1 revealed through cre-dependent rabies tracing. Cell
Reports, 7(1), 269–280. http://doi.org/10.1016/j.celrep.2014.02.030
Suzuki, L., Coulon, P., Sabel-Goedknegt, E. H., & Ruigrok, T. J. H. (2012). Organization of Cerebral
Projections to Identified Cerebellar Zones in the Posterior Cerebellum of the Rat. Journal of
Neuroscience, 32(32), 10854–10869. https://doi.org/10.1523/jneurosci.0857-12.2012
Swanson, L. W. (2000). Cerebral hemisphere regulation of motivated behavior. Brain Research,
886, 113–164.
Szapiro, G., & Barbour, B. (2007). Multiple climbing fibers signal to molecular layer interneurons
exclusively via glutamate spillover. Nature Neuroscience, 10(6), 735–742.
https://doi.org/10.1038/nn1907
Tamamaki, N., Yanagawa, Y., Tomioka, R., Miyazaki, J. I., Obata, K., & Kaneko, T. (2003). Green
Fluorescent Protein Expression and Colocalization with Calretinin, Parvalbumin, and
Somatostatin in the GAD67-GFP Knock-In Mouse. Journal of Comparative Neurology,
467(1), 60–79. http://doi.org/10.1002/cne.10905
Tanné-Gariépy, J., Boussaoud, D., & Rouiller, E. M. (2002). Projections of the claustrum to the
primary motor, premotor, and prefrontal cortices in the macaque monkey. The Journal of
Comparative Neurology, 454(2), 140–57. http://doi.org/10.1002/cne.10425
Tervo, D. G. R., Hwang, B. Y., Viswanathan, S., Gaj, T., Lavzin, M., Ritola, K. D., … Karpova, A. Y.
(2016). A Designer AAV Variant Permits Efficient Retrograde Access to Projection Neurons.
Neuron, 92(2), 372–382. http://doi.org/10.1016/j.neuron.2016.09.021
Torgerson, C. M., Irimia, A., Goh, S. Y. M., & Van Horn, J. D. (2014). The DTI connectivity of the
human claustrum. Human Brain Mapping, 00, n/a-n/a. http://doi.org/10.1002/hbm.22667
Tovote, P., Esposito, M. S., Botta, P., Chaudun, F., Jonathan, P., et al. (2016). Midbrain circuits for
defensive behaviour. Nature 534, 206-212.
Ugolini, G. (2010). Advances in viral transneuronal tracing. J. Neurosci. Methods 194, 2–20.
Ugolini, G., Kuypers, H.G., & Strick, P.L. (1989) Transneuronal transfer of herpes virus from
peripheral nerves to cortex and brainstem. Science 243, 89-91.
Vertes, R. P., Linley, S. B., & Hoover, W. B. (2015). Limbic circuitry of the midline thalamus.
Neuroscience and Biobehavioral Reviews, 54, 89–107.
http://doi.org/10.1016/j.neubiorev.2015.01.014
Von Bartheld, C. S., & Altick, A. L. (2011). Multivesicular bodies in neurons: Distribution, protein
content, and trafficking functions. Progress in Neurobiology, 93(3), 313–340.
https://doi.org/10.1016/j.pneurobio.2011.01.003
Wall, N. R., Wickersham, I. R., Cetin, A., De La Parra, M., & Callaway, E. M. (2010). Monosynaptic
circuit tracing in vivo through Cre-dependent targeting and complementation of modified
129

rabies virus. Proceedings of the National Academy of Sciences of the United States of
America, 107(50), 21848–53. http://doi.org/10.1073/pnas.1011756107
Wall, N.R., Wickersham, I.R., Cetin, A., De La Parra, M., and Callaway, E.M. (2010). Monosynaptic
circuit tracing in vivo through Cre-dependent targeting and complementation of modified
rabies virus. Proc. Natl. Acad. Sci. U. S. A. 107, 21848–21853.
Wang, Q., Ng, L., Harris, J.A., Feng, D., Li, Y., Royall, J.J., Oh, S.W., Bernard, A., Sunkin, S.M.,
Koch, C., & Zeng, H. (2016). Organization of the connections between claustrum and cortex
in the mouse. J Comp Neurol. 525:1317-1346.
Wang, Y., Balaji, V., Kaniyappan, S., Krüger, L., Irsen, S., Tepper, K., … Mandelkow, E. M. (2017).
The release and trans-synaptic transmission of Tau via exosomes. Molecular
Neurodegeneration, 12(1), 1–25. https://doi.org/10.1186/s13024-016-0143-y
Wang, Y., Gong, H., Li, Y., Kuang, X., Daigle, T.L., Madisen, L., Gu, H., Mills, M., Gray, L., Tasic, B.,
Li, A., Harris, J.A., Luo, Q., Koch, C., & Zeng, H. Whole-brain reconstruction and
classification of spiny claustrum neurons of mice. Program No. 259.02. 2017 Neuroscience
Meeting Planner. Washington, DC: Society for Neuroscience, 2017. Online.
Watakabe, A., Ohsawa, S., Ichinohe, N., Rockland, K. S., & Yamamori, T. (2014). Characterization
of claustral neurons by comparative gene expression profiling and dye-injection analyses.
Frontiers in Systems Neuroscience, 8(May), 98. http://doi.org/10.3389/fnsys.2014.00098
Watson, G. D. R., Smith, J. B., & Alloway, K. D. (2017). Interhemispheric connections between the
infralimbic and entorhinal cortices: The endopiriform nucleus has limbic connections that
parallel the sensory and motor connections of the claustrum. Journal of Comparative
Neurology, 525(6), 1363–1380. http://doi.org/10.1002/cne.23981
Wei, P., Liu, N., Zhang, Z., Liu, X., Tang, Y., He, X., Wu, B., Zhou, Z., Liu, Y., Li, J., et al. (2015).
Processing of visually evoked innate fear by a non-canonical thalamic pathway. Nat.
Commun. 6, 6756.
White, M. G., Cody, P. A., Bubser, M., Wang, H.-D., Deutch, A. Y., & Mathur, B. N. (2016). Cortical
hierarchy governs rat claustrocortical circuit organization. Journal of Comparative Neurology,
520(Sfb 655), n/a-n/a. http://doi.org/10.1002/cne.23970
White, M. G., Panicker, M., Mu, C., Carter, A. M., Roberts, B. M., Dharmasri, P. A., … Mathur, B. N.
(2018). Anterior Cingulate Cortex Input to the Claustrum Is Required for Top-Down Action
Control. Cell Reports, 22(1), 84–95. http://doi.org/10.1016/j.celrep.2017.12.023
Wickersham, I.R., Lyon, D.C., Barnard, R.J.O., Mori, T., Finke, S., Conzelmann, K.K., Young, J.
A.T., and Callaway, E.M. (2007). Monosynaptic Restriction of Transsynaptic Tracing from
Single, Genetically Targeted Neurons. Neuron 53, 639–647.
Wimmer, V. C., Bruno, R. M., de Kock, C. P. J., Kuner, T., & Sakmann, B. (2010). Dimensions of a
Projection Column and Architecture of VPM and POm Axons in Rat Vibrissal Cortex.
Cerebral Cortex, 20(10), 2265–2276. https://doi.org/10.1093/cercor/bhq068
130

Wolff, S., Gründemann, J., Tovote, P., Krabbe, S., Jacobson, G.A., Müller, C., Herry, C., et al.
(2014) Amygdala interneuron subtypes control fear learning through
disinhibition. Nature 509, 453-458.
Woods, N. I., Vaaga, C. E., Chatzi, C., Adelson, J. D., Collie, M. F., Perederiy, J. V., … Westbrook,
G. L. (2018). Preferential Targeting of Lateral Entorhinal Inputs onto Newly Integrated
Granule Cells. The Journal of Neuroscience, 38(26), 5843–5853.
https://doi.org/10.1523/jneurosci.1737-17.2018
Xiong, X. R., Liang, F., Zingg, B., Ji, X., Ibrahim, L. A., Tao, H. W., & Zhang, L. I. (2015). Auditory
cortex controls sound-driven innate defense behaviour through corticofugal projections to
inferior colliculus. Nature Communications, 6, 7224. http://doi.org/10.1038/ncomms8224
Xu, K., Malouf, A.T., Messing, A., and Silver, J. (1999). Glial fibrillary acidic protein is necessary for
mature astrocytes to react to B-amyloid. Glia 25, 390–403.
Xu, W., & Edgley, S. A. (2008). Climbing fibre-dependent changes in Golgi cell responses to
peripheral stimulation. Journal of Physiology, 586(20), 4951–4959.
https://doi.org/10.1113/jphysiol.2008.160879
Xu, W., & Südhof, T. C. (2013). A neural circuit for memory specificity and generalization. Science
(New York, N.Y.), 339(6125), 1290–1295. https://doi.org/10.1126/science.1229534
Yamamoto, M., Wada, N., Kitabatake, Y., Watanabe, D., Anzai, M., Yokoyama, M., … Nakanishi, S.
(2003). Reversible suppression of glutamatergic neurotransmission of cerebellar granule
cells in vivo by genetically manipulated expression of tetanus neurotoxin light chain. The
Journal of Neuroscience : The Official Journal of the Society for Neuroscience, 23(17), 6759–
6767. https://doi.org/23/17/6759 [pii]
Yizhar, O., Fenno, L.E., Davidson, T.J., Mogri, M., and Deisseroth, K. (2011). Optogenetics in neural
systems. Neuron 71, 9–34.
Yoshihara, Y., Mizuno, T., Nakahira, M., Kawasaki, M., Watanabe, Y., Kagamiyama, H., … Mori, K.
(1999). A genetic approach to visualization neurotechnique of multisynaptic neural pathways
using plant lectin transgene. Neuron, 22(1), 33–41. https://doi.org/10.1016/S0896-
6273(00)80676-5
Zhu, Y., Wienecke, C. F. R., Nachtrab, G., & Chen, X. (2016). A thalamic input to the nucleus
accumbens mediates opiate dependence. Nature, 530(7589), 219–222.
http://doi.org/10.1038/nature16954
Zingg, B., Chou, X., Zhang, Z., Mesik, L., Liang, F., Tao, H. W., & Zhang, L. I. (2017). AAV-Mediated
Anterograde Transsynaptic Tagging: Mapping Corticocollicular Input-Defined Neural
Pathways for Defense Behaviors. Neuron, 93(1), 33–47.
https://doi.org/10.1016/j.neuron.2016.11.045
Zingg, B., Hintiryan, H., Gou, L., Song, M. Y., Bay, M., Bienkowski, M. S., … Dong, H.-W. (2014).
Neural Networks of the Mouse Neocortex. Cell, 156(5), 1096–1111.
http://doi.org/10.1016/j.cell.2014.02.023

Abstract (if available)

Abstract AAV is one of the most versatile tools in systems neuroscience. Relative to other neurotropic viruses, it is unique in its ability to drive long-term, robust transgene expression without becoming toxic to host cells. Combined with its broad tropism, high safety profile, and ease of manufacture, this has led to its widespread use in both anatomical and functional investigations of neural circuitry. It is most commonly applied for the local transduction of targeted cell populations, however AAV also displays distal transport properties, including retrograde and anterograde transsynaptic transport, that are less well characterized but show great potential for application. ❧ In the first part of this study, we take advantage of a recently developed retrograde serotype of AAV and use it to gain selective access to projection-defined claustrum neurons in the mouse brain. This structure likely plays an important role in the modulation of higher-order cortical regions. However, due to the difficult nature of targeting the claustrum, its precise function and anatomical organization has remained enigmatic. o overcome this, we used injections of AAVretro-Cre to tag claustrum neurons based on their projection to the retrosplenial cortex. We then used secondary injections of a Cre-dependent AAV along with monosynaptic rabies virus tracing to map the brain-wide input and output of this structure. Our results suggest the claustrum may be driven by emotionally salient stimuli and may modulate the attentional state of the animal during rewarding or aversive contexts. Using the same approach for accessing these cells, future studies may seek to test this hypothesis by directly recording from optogenetically identified claustrum neurons in behaving animals. ❧ In the remaining portion of this study, we characterize a novel anterograde transsynaptic transport property of AAV1. We find that AAV1-Cre from transduced presynaptic neurons can effectively and specifically drive Cre-dependent transgene expression in selected postsynaptic neuronal targets, thus allowing axonal tracing and functional manipulation of these input-defined cell populations. This is especially important given that currently available tools for this purpose, such as herpesvirus and vesicular stomatitis virus, suffer from toxicity issues that limit their use in functional studies. We then apply this tool in corticofugal pathways that converge on distinct populations of neurons in the superior colliculus and reveal their ability to drive either freezing or flight, two different forms of innate defense behavior. Finally, given its promise for application in other brain regions, we provide a broader demonstration of its use in corticofugal, thalamic, and spinal pathways, and offer additional lines of evidence supporting a synaptic, rather than extrasynaptic, mechanism of spread. In particular, we find a strong correspondence between pre-synaptic connectivity and post-synaptic labeling in slice recording studies, and find that co-expression of tetanus toxin light chain, an inhibitor of pre-synaptic vesicle fusion, nearly abolishes transsynaptic spread of AAV. Overall, our results suggest that AAV-mediated anterograde transsynaptic tagging shows great potential as a tool for the forward screening of neural circuits.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Neural circuits underlying the modulation and impact of defensive behaviors

PDF

Neural circuits control and modulate innate defensive behaviors

PDF

Synaptic mechanism underlying development and function of neural circuits in rat primary auditory cortex

PDF

From sensory processing to behavior control: functions of the inferior colliculus

PDF

Improving adeno-associated viral vector for hematopoietic stem cells gene therapy

PDF

Contextual modulation of sensory processing via the pulvinar nucleus

PDF

Functional properties of the superficial cortical interneurons

PDF

Axon guidance cues in development of the mammalian auditory circuit

PDF

Synaptic circuits for information processing along the central auditory pathway

PDF

The impact of genetic pleiotropy on heterogeneity in the developing forebrain and hindbrain

PDF

The structural and functional configurations of glutamate, GABA, and catecholamine pre-synaptic terminals in the parvicellular neuroendocrine part of the paraventricular nucleus of the hypothalamus

PDF

On the electrophysiology of multielectrode recordings of the basal ganglia and thalamus to improve DBS therapy for children with secondary dystonia

Asset Metadata

Creator Zingg, Brian (author)

Core Title Exploiting novel transport properties of adeno-associated virus for circuit mapping and manipulation

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Neuroscience

Publication Date 07/24/2020

Defense Date 04/30/2019

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

Tag AAV,adeno-associated virus,anterograde transsynaptic,claustrum,defense behavior,OAI-PMH Harvest,superior colliculus,viral tracing

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bonnin, Alexandre (committee chair), Dong, Hong-Wei (committee member), Tao, Huizhong (committee member), Zhang, Li (committee member)

Creator Email brian.zingg@gmail.com,brian.zingg@loni.usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-195384

Unique identifier UC11663180

Identifier etd-ZinggBrian-7645.pdf (filename),usctheses-c89-195384 (legacy record id)

Legacy Identifier etd-ZinggBrian-7645.pdf

Dmrecord 195384

Document Type Dissertation

Format application/pdf (imt)

Rights Zingg, Brian

Type texts

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

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

Repository Name University of Southern California Digital Library

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