University of Southern California Dissertations and Theses

Neural circuits control and modulate innate defensive behaviors

(USC Thesis Other)

Neural circuits control and modulate innate defensive behaviors

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content Copyright 2021 Xiyue Wang

NEURAL CIRCUITS CONTROL AND MODULATE
INNATE DEFENSIVE BEHAVIORS

by

Xiyue Wang

A Dissertation Presented to the
FACULTY OF THE GRADUATE SCHOOL
UNIVERSITY OF SOUTHERN CALIFORNIA
In Partial Fulfillment of the
Requirements for the Degree
DOCTOR OF PHILOSOPHY
(Neuroscience)

May 2021

ii

Acknowledgements

This work would not have been possible without the support of my advisors, committee
members, colleagues, families and friends.
Foremost, I would like to express my deepest gratitude to my advisor Dr. Huizhong W. Tao and
Dr. Li I Zhang, who continuously guide me through my graduate study and shared the
excitement of six years of discovery. Their unwavering wisdom, enthusiasm, patience and
encouragement have kept me engage with my research. I could not have achieved anything
without their support and supervision.
I would like to thank my committee members, Dr. Alexandre Bonnin, Dr. Samuel Andrew Hires,
Dr. Hongwei Dong, Dr. Jeannie Chen and Dr. Chien-ping Ko. They have guided me through my
graduate study with prompt inspirations and insightful suggestions.
My sincere thanks also go to my colleagues. Dr. Xiaolin Chou’s mentoring and support have
been especially valuable, and his early insights launched a great part of this dissertation. Dr.
Lukas Mesik, Junxiang Huang enlivened the third floor with good scientific discussions and
good humor. I would also like to thank all my other lab members who have helped me and made
the lab a positive atmosphere in which to do science. Above ground, I am indebted to my family
and friends. I would like to thank my mom; whose love and guidance are with me in whatever I
pursue. My dear ‘Forum’ friends and Lei Peng, who provide unending inspirations and
encouragement. Finally, I wish to thank my loving husband, YC, who brings me a life of joy in
the hours when the lab lights went off.

iii
Table of Contents
Acknowledgements ....................................................................................................................... ii
Table of Figures ............................................................................................................................. v
Table of Tables .............................................................................................................................. vi
Abstract ....................................................................................................................................... vii
Chapter 1: Introduction ................................................................................................................ 1
1.1 Overview of defensive behaviors .................................................................................. 2
1.1.1 Different types of defensive behaviors ........................................................................... 2
1.1.2 Learning of defensive behaviors .................................................................................... 4
1.2 Brain structures involved in defensive behaviors ............................................................. 7
1.2.1 Midbrain and brainstem nuclei – output ......................................................................... 8
1.2.2 Hypothalamus and Amygdala –integration .................................................................... 9
1.2.3 Cortex and thalamus – input ......................................................................................... 13
1.3 Flexible, context-dependent defensive networks ............................................................ 16
1.3.1 Freezing or flight .......................................................................................................... 17
1.3.2 Context-dependent modulation of defensive behaviors ............................................... 19
1.3.3 Experience-dependent modulation of defensive behavior ............................................ 21
Chapter 2: Differential Circuit Control of Visual Defense Behaviors in Superior Colliculus
....................................................................................................................................................... 24
2.1 Introduction ....................................................................................................................... 24
2.2 Results ................................................................................................................................. 27
2.2.1 Depth-specific response pattern exists in the SC ......................................................... 27
2.2.2. Cortical inputs regulate the visual responses of the intermediate SC ......................... 31
2.2.3. Cortico-recipient SC Neurons Mediate Looming-Evoked Freezing Responses ......... 34
2.2.4. Retina-recipient SC Neurons Mediate Looming-Evoked Flight Responses ............... 36
2.2.5. Distinct Behavioral Functions for SC Downstream Targets ....................................... 37
2.3 Discussion ........................................................................................................................... 42
2.4 Material and Methods ....................................................................................................... 46
2.4.1 Animals, stereotaxic injection and imaging ................................................................. 46
2.4.2 Visual stimulation and in vivo electrophysiology ........................................................ 49
2.4.3 Behavioral tests ............................................................................................................ 51
2.4.4 Data analysis and statistics ........................................................................................... 52
Chapter 3: A Cross-modality Enhancement of Defensive Flight via ..................................... 55
Parvalbumin Neurons in Zonal Incerta .................................................................................... 55
3.1 Introduction ....................................................................................................................... 55
3.2 Results ................................................................................................................................. 57
3.2.1 Tactile stimulation enhances sound-induced flight response via SSp .......................... 57
iv
3.2.2 The SSp-ZIv projection mediates the tactile enhancement of sound-induced flight .... 59
3.2.3 PV+ neurons in ZIv mediate the tactile enhancement of flight behavior ..................... 62
3.2.4 The projection of ZIv PV+ neurons to POm enhances sound-induced flight .............. 66
3.3 Discussion ........................................................................................................................... 69
3.4 Material and methods ....................................................................................................... 74
2.4.1 Animals, viral and reagent injections ........................................................................... 74
3.4.2 Histology, imaging and quantification ......................................................................... 75
3.4.3 Optogenetic preparation and stimulation ...................................................................... 76
3.4.4 Behavioral tests ............................................................................................................ 77
3.4.5 Slice preparation and recording .................................................................................... 78
3.4.6 Optrode recording and spike sorting ............................................................................ 79
3.4.6 Data Processing and statistics ....................................................................................... 80
References .................................................................................................................................... 82

v
Table of Figures

Figure 1. The Threat-Imminence Model of Defensive Behaviors ................................................... 3
Figure 2. A dark expanding disc in the upper visual field triggers flight and freezing ................... 6
Figure 3. Noise can evoke robust flight responses .......................................................................... 7
Figure 4. A schematic diagram showing the hypothalamic nuclei involved in processing
predatory and conspecific threats .................................................................................................. 10
Figure 5. The amygdala ................................................................................................................ 12
Figure 6. Schematic representation of the neural circuits mediating innate fear to different
threats ............................................................................................................................................ 15
Figure 7. Schematic representation of the freezing pathway ......................................................... 18
Figure 8. The SC is essential for looming-evoked defensive behaviors ....................................... 26
Figure 9. Depth-specific response pattern exists in the SC ........................................................... 29
Figure 10. Basic response properties in the SC changes over depths ........................................... 31
Figure 11. Cortical inputs regulate the visual responses of the intermediate SC .......................... 33
Figure 12. Cortico-recipient SC Neurons Mediate Looming-Evoked Freezing Responses .......... 35
Figure 13. Silencing AM did not affect looming-evoked defensive behaviors ............................. 37
Figure 14. Retina-recipient SC neurons mediate looming-evoked flight responses ..................... 38
Figure 15. Distinct Behavioral Functions for SC Downstream Targets ........................................ 40
Figure 16. Silence dlPAG could potentiate freezing responses .................................................... 42
Figure 17. A Circuit model for laminar controls of looming-evoked defensive behavior ............ 46
Figure 18. Tactile stimulation enhances sound-induced flight response via SSp ......................... 58
Figure 19. Flight speed decreased without whiskers in two-chamber test .................................... 59
Figure 20. The SSp-ZIv projection mediates the tactile enhancement of sound-induced flight ... 60
Figure 21. Optogenetic stimulations could effectively activate/suppress labeled cells ................ 62
Figure 22. PV+ neurons in ZIv mediate the tactile enhancement of flight behavior .................... 63
Figure 23. Optogenetic stimulation of ZI did not alter baseline speed ......................................... 65
Figure 24. DREADDi manipulation were effective ...................................................................... 66
Figure 25. The projection of ZIv PV+ neurons to POm enhances sound-induced flight .............. 67
Figure 26. Mapping of axonal outputs for ZIv PV+ neurons ........................................................ 69
vi

Table of Tables
Table 1. Analysis of temporal profiles of speed traces in different sets of experiments ............... 73

vii
Abstract
Defensive behaviors are essential for animal survival. In the natural environment, there are
sensory stimuli that innately represent threats and trigger stereotyped behaviors such as freezing
and flight. Although considered to be hardwired, innate defensive behavior is also flexible and
subject to modulation by dynamic environmental contexts and internal physiological states. For
example, the properties and intensity of the sensory cues, satiety states, and past experience are
all shown to contribute to defensive behaviors. Aided by powerful genetic tools, remarkable
advances have been made in recent years in our understanding of innate defensive behaviors and
the underlying neural circuits. However, the brain circuits allowing flexible and context-
dependent defensive behaviors are not fully understood. Here, I will present two studies
investigating part of the neural circuits for control and modulation of innate defensive behaviors.
In the first study, we investigated the circuit mechanisms that allow distinct defensive
reactions in response to a single sensory modality. Survival in threatening situations depends
on the rapid execution of an appropriate active or passive defensive response. Animals show
both freezing and flight innately in respond to looming visual stimuli. How visual
information converge into the superior colliculus (SC) to evoke dimorphic divergent
responses are not well-understood. Here we identify the retinal-recipient SC neurons and
cortical-recipient SC neurons are located in different laminae of the SC. Silencing the
primary visual cortex (V1) mainly effect the visual responses in the lower superficial and
intermediate layers but not the most superficial layer of SC. In addition, the retinal-recipient
SC neurons are necessary for flight responses by preferentially target the deep layer of SC,
while the cortical-recipient SC neurons are critical for the freezing response by preferentially
viii
target the lateral posterior nucleus of the thalamus. Together, our data define the differential
circuit control of visual defense behaviors in the superior colliculus.
Secondly, we studied the neural circuits that modulate defensive responses in complex
sensory environment. The ability to adjust defensive behavior is critical for animal survival in
dynamic environments. However, neural circuits underlying the modulation of innate
defensive behavior remain not well-understood. In particular, environmental threats are
commonly associated with cues of multiple sensory modalities. It remains to be investigated
how these modalities interact to shape defense behavior. In this study, we report that
auditory-induced defensive flight can be facilitated by somatosensory input in mice. This
cross-modality modulation of defensive behavior is mediated by the projection from the
primary somatosensory cortex (SSp) to the ventral sector of zona incerta (ZIv). Parvalbumin-
positive neurons in ZIv, receiving direct input from SSp, mediate the enhancement of the
flight behavior via their projections to the medial posterior complex of thalamus (POm).
Thus, defensive flight behavior can be enhanced in a somatosensory context-dependent
manner via recruiting PV neurons in ZIv, which may be important for increasing survival of
prey animals.

1
Chapter 1: Introduction

In natural environment, animals are exposed to a wide range of threats, including predators,
attacks by conspecifics and dangerous features of the environment (Blanchard and Blanchard,
2008). Amongst the most intensely investigated of these are predators, with prey animals
exhibiting a variety of passive adaptations to avoid capture as well as active defend themselves
from predators (Bolles, 1970; Curio, 1993; Fanselow, 1994; Tovote, Fadok and Lüthi, 2015).
Defensive behaviors are characterized as a set of responses to threat stimuli and situations that
have evolved on the basis of their adaptiveness in reducing harm to the threatened organism
(Blanchard and Blanchard, 2008). They can be innate or learned. Lots of efforts have been spent
to investigate the brain circuits that control different types of defensive behaviors (Fanselow,
1994; Yilmaz and Meister, 2013; Liang et al., 2015; Tovote, Fadok and Lüthi, 2015; Wang,
Chen and Lin, 2015; Xiong et al., 2015; Fadok et al., 2017), but our understanding of the defense
circuits remains incomplete due to the specific structure or the specific behavior each study was
focusing on. In fact, survival in the face of predation in a dynamic environment often requires
brain mechanisms that instantaneously select more “adaptive” responses, which may recruit
multiple circuits to govern different defensive responses.
In this chapter, I will overview the different types of defensive behaviors as well as the
underlying neural circuits, with a focus on innate defensive behaviors. I will then introduce
current understandings of the neural circuits that allow switches between different defensive
responses. Finally, I will briefly introduce the impact of multisensory integration on defensive
behaviors.
2
1.1 Overview of defensive behaviors
1.1.1 Different types of defensive behaviors
Animal studies suggest that defensive behaviors are organized along a dimension depending on
the discreteness and ambiguity of the threat, defense distance and presence of particular enabling
stimuli (Fanselow, 1994; Blanchard and Blanchard, 2008). Defensive behaviors include
vocalizations warning conspecifics, avoidance of the threats, orienting towards safe zone,
freezing, escape and defensive attacks. These behaviors are associated with changes in
autonomic tone and endocrine activity (Fanselow, 1994; Blanchard and Blanchard, 2008;
Adolphs, 2013). Although the repertoire of defensive behaviors is relatively limited, defense is
considered as general form of behaviors across species and individuals (Figure 1).
When the animal is in an environment or context in which a threat has been encountered before
but not yet detected (pre-encounter, see Fanselow, 1994), a series of adaptive defensive
responses is engaged. It includes reduce activity, increase vigilance, or move to a safe zone in
normal speed (Fanselow, 1994; Eilam, 2005). In human, novel and ambiguous environments in
which danger may occur will induce feeling of anxiety (Davis et al., 2010). As long as the threat
is detected but still discrete, post-encounter defense replace pre-encounter defense behaviors. It
is characterized by an increase in attention to the threat accompanied by heart rate deceleration,
quickly enter a “shelter” if accessible, travel quickly in short segments, or temporary episodes of
motor “freezing” (Campbell, Wood and McBride, 1997; Eilam, 2005; Eilam, Izhar and Mort,
2011; Tovote et al., 2016). In human, a closing-in threat often triggers feeling of fear or stress
(Davis et al., 2010).
3

Figure 1. The Threat-Imminence Model of Defensive Behaviors
Transdiagnostic dimensional model of defensive behaviors suggesting that defensive reactivity is
dynamically organized along a continuum depending on the imminence or proximity of the
threat (Adapted from Hamm, 2019).

With the increasing proximity of the treat (contacting or contact is inevitable), animals often
show circa-strike defense (Fanselow, 1994). Defensive responses changes from passive into
active actions depends on behavioral options, which includes undirected flight, fight, and
freezing (Eilam, 2005; Tovote et al., 2016). Human subjects frequently report feeling of panic
(Mobbs et al., 2009) and discharge of the sympathetic nervous system during contacting threats
(Adolphs, 2013). Increase of adrenaline in bloodstream could result in accelerating heart rate and
sweating, and modulating formation and consolidation of emotional memory (McGaugh, 2004).
The dynamic organization of defensive behaviors are reviewed by many researchers. Popular
model includes threat-imminence model (Fanselow, 1994; Eilam, 2005) and the action-action
4
tendency model (Adolphs, 2013). All in all, there are different types of defensive behaviors,
which are integrated into a complex yet flexible “program” to ensure survival.

1.1.2 Learning of defensive behaviors
Defensive behaviors are a class of innate response dispositions retained by natural selection,
because they promote survival in the face of threats (Lüthi and Lüscher, 2014; McCullough,
Morrison and Ressler, 2016). For example, chicks respond to alarm calls, and strong order can
evoke avoidance behaviors in flies. In order to defend, animals must successfully recognize and
then respond appropriately to the danger the first time it is encountered. This has led to general
acceptance of the idea that most fears are innately programmed. However, there is a very
complex relationship between defensive behaviors and learning.
The actions involved in a particular defensive behavior may be acquired through associative
learning either from past experience or observation of others performing the same actions. One
famous example is Pavlovian conditioning, which is widely used to study defensive behaviors in
lab (Blanchard and Blanchard, 2008; Fanselow and Wassum, 2016). It pairs a neutral stimulus
with pain during training, and later animals would show robust fear responses to the neutral
stimulus. However, the development of defensive behaviors is not solely dependent on reward or
punishment contingencies during experience of pain or danger. One interesting experiment
performed by Mineka group (Mineka et al., 1984; Cook and Mineka, 1989) showed that only
rhesus monkeys born in wild show initial fear responses to toy snakes, but lab born rhesus
monkeys could acquire fear of snakes quickly by observing other monkeys’ fearful reactions
through video. Similarly, lab born rat pups (P18) did not show avoidance to cat odor while adult
5
rats showed strong avoidance, indicating that the specific behaviors comprising defense may be
learned in early development via interactions with other conspecifics (Hubbard et al., 2004).
While fear profoundly constricts the behavioral repertoire of defense, learning could increase the
flexibility of defensive behaviors by engaging active risk assessment. For example, mice could
easily memorize the shelter location in a single, short-lived visit. They can then rapidly update
their defensive actions from freezing to escape to shelter accurately (Vale, Evans and Branco,
2017). In addition, mice can quickly learn to stay in “insulating zone” to avoid shocking during
Pavlovian conditioning (Kong et al., 2014). These results demonstrate that mice have very
adaptable risk assessment mechanisms they can incorporate into defense, and the degree to
which the form and patterning of defensive responses may be dependent of learning based on
response contingencies or observation learning.

1.1.3 Animal models to study innate defensive behaviors
Certain sensory cues representing characteristics of predators are used in laboratory to study
innate defensive behaviors, particularly the flight and freezing behaviors. Here I will briefly
introduce several popular animal models to study innate defensive behaviors.
Many animals rely on their sense to smell. Predator orders can induce an array of different anti-
predatory responses (Dielenberg and McGregor, 2001; Takahashi et al., 2005). 2, 3, 5-trimethyl-
3-thiazoline (TMT) is a compound isolated from fox feces that commonly used in lab to study
defensive behaviors. Rodents displays strong avoidance, hiding and freezing behaviors upon
exposed to TMT, and their normal behaviors such as grooming and locomotor activity are
suppressed (Wallace and Rosen, 2000; Endres, Apfelbach and Fendt, 2005; Fendt et al., 2005).
6

Figure 2. A dark expanding disc in the upper visual field triggers flight and freezing
Schematic of the experimental setup: a box with a display monitor on the ceiling and an opaque
nest in a corner. Expansion of the looming stimulus in time from 2 degrees to 20 degrees. Upon
stimulus, animals could show freezing (green), flight (blue) or no response (grey). (Adapted from
Daviu et al., 2020)

Previous studies have also shown a looming visual stimulus, characterized as an expanding dark
disk in upper visual field, could mimic approaching predators and evoke defensive behaviors in
rodents and drosophila in an open field (Maimon, Straw and Dickinson, 2008; Yilmaz and
Meister, 2013; Wei et al., 2015; De Franceschi et al., 2016; Ache et al., 2019). In rodents,
looming stimuli could evoke robust flight and freezing responses, and the bias of adopting flight
response is dependent on the presence of a hiding shelter (Yilmaz and Meister, 2013; Shang,
Chen, Liu, Li, Zhang, Qu, Yan, Zhang, Liu, Liu, Guo, Li, Wang and Cao, 2018).
7
Besides looming stimuli, flight behaviors invoked by sensory stimuli have been observed in
various species (Eilam, 2005). Previous studies have found that a loud white noise or ultrasound
Figure 3. Noise can evoke robust flight responses
White noise (80 db) could evoke robust flight responses in rodents. (Adapted from Xiong et al., 2015).

is aversive to rodents, and can evoke flight responses in both head-fixed (running on plate) and
freely-moving animals (one-chamber or two-chamber) (Mongeau et al., 2003; Xiong et al., 2015;
Zingg et al., 2017; Evans et al., 2018; Dong et al., 2019). The behavioral responses towards
these stimuli are usually innate, but the level of the responses can exhibit adaptation after
repeated exposures without the occurrence of actual harm (Yilmaz and Meister, 2013; Xiong et
al., 2015). This phenomenon supports the idea that active risk assessment can modulate the hard-
wired innate defense circuits.

1.2 Brain structures involved in defensive behaviors
When a threat approaching, a wave of neural activities cascades through the nervous system.
Many cortical regions, hypothalamic regions together with midbrain and brainstem nuclei
participate in fear responses. A great part of the neurobiology of defensive behaviors still
remains unclear. Although defensive behaviors are conserved across different species, here I will
8
briefly outline some well-studied circuits using rodents, which is the most commonly used
organisms for the study of fear circuits.
1.2.1 Midbrain and brainstem nuclei – output
Midbrain and brainstem nuclei are broadly involved in controlling locomotion planning and
execution due to their extensive connection with the spinal cord (Ferreira-Pinto et al., 2018).
Amongst the periaqueductal gray (PAG) is considered a final common path for all types of
defensive responses. Recording data have shown that PAG neurons respond to predator order,
aggressive conspecific, loud noise/ultrasound and electric foot-shock (Cezario et al., 2008; Motta
et al., 2009; Xiong et al., 2015). Different subsets of PAG neurons are activated during risk
assessment, flight and freezing responses (Deng, Xiao and Wang, 2016). In addition, direct
stimulation of PAG neurons could induce various defensive behaviors, including jump, freezing,
flight and flight (Xiong et al., 2015; Deng, Xiao and Wang, 2016; Hashikawa et al., 2017; Y. Li
et al., 2018a), while impairment of PAG diminishes the expression of these behaviors (B.A. et
al., 2013; Silva, Gross and Gräff, 2016; Tovote et al., 2016). The PAG can be further divided
into dorsolateral (dlPAG) and ventrolateral (vlPAG) sectors based on their functionality. dlPAG
receives strong inputs from the hypothalamus, superior colliculus (SC) and inferior colliculus
(IC) to mediate active defensive responses, include flight, jumping and fight (Canteras, 2002;
Wang, Chen and Lin, 2015; Xiong et al., 2015; Evans et al., 2018). The vlPAG receives inputs
from the amygdala and mediate passive defensive responses (Tovote et al., 2016).
Besides the PAG, several other midbrain and brainstem structures were shown to be involved in
defensive behaviors. The SC receives direct visual inputs from the retina and cortex (Shi et al.,
2017; Zingg et al., 2017). When a threatening visual stimulus is presented, the SC could projects
to dlPAG and the parabigeminal nucleus (PBN) to mediate flight responses (Shang et al., 2015;
9
Evans et al., 2018), and project to the lateral posterior complex of the thalamus (LP) to mediate
freezing and “arrest” responses (Liang et al., 2015; Wei et al., 2015; Fang et al., 2020). In
addition, the midbrain locomotor region (MLR) includes cuneiform nucleus (CnF) and
pedunculopontine nucleus (PPN) participates in controlling the escape speed (Capelli et al.,
2017; Ferreira-Pinto et al., 2018). Both the PAG and MLR descend to the magnocellular nucleus
(Mc) and gigantocellular nucleus (Gi), which directly control freezing and locomotion (speed
and gait) (Esposito, Capelli and Arber, 2014; Tovote et al., 2016; Ferreira-Pinto et al., 2018).
In summary, innate defensive behaviors recruit many neuronal subpopulations in the mid- and
hindbrain. However, we should note that these structures are heavily intertwined and may not be
specific for defensive behaviors but more closely related to the locomotion execution. Revealing
the precise contribution of each structure to the different behavioral outcomes in different
contexts remains to be determined.

1.2.2 Hypothalamus and Amygdala –integration
The vertebrate hypothalamus regulates a variety of behaviors and physiological stats to promote
survival (Sternson, 2013). Direct activation of distinct subregions or genetically specified
subpopulations within the hypothalamus can drive different behaviors, including food-seeking,
water-seeking, fighting, parental behaviors and defensive behaviors (Canteras, 2002; Dietrich et
al., 2012; Wang, Chen and Lin, 2015; Allen et al., 2017; Hashikawa et al., 2017).
The medial hypothalamic defensive system (MHDS) is important for integrating fear to
predatory threats (Canteras, 2002; Motta et al., 2009). This system consists of a set of nuclei
located in the hypothalamic medial zone: the anterior hypothalamic nucleus (AH), the
10
dorsomedial portion of the ventromedial hypothalamus (VMHdm), and the dorsal
premammillary nucleus (PMD). These structures are highly interconnected nuclei are selectively
Figure 4. A schematic diagram showing the hypothalamic nuclei involved in processing predatory
and conspecific threats
Abbreviations: AHN, anterior hypothalamic nucleus; BMAp, basomedial amygdalar nucleus,
posterior part; LA, lateral amygdalar nucleus; MEAad, -pd, and -pv, medial amygdalar nucleus,
anterodorsal, posterodorsal, and posteroventral parts; MPO, medial preoptic area; PAGdl, -dm,
and -l, periaqueductal gray, dorsolateral, dorsomedial, and lateral parts; PMDdm and -vl, dorsal
premammillary nucleus, dorsomedial and ventrolateral parts; PMV, ventral premammillary
nucleus; VMHdm and -vl, ventromedial nucleus, dorsomedial and ventrolateral parts (Adapted
from Motta et al., 2009)

recruited by predator exposures (Canteras, 2002). Activation of VMHdm is sufficient to drive
multiple defensive responses, including avoidance and escape jumping through its projections to
AN and freezing through its projections to PAG (Kunwar et al., 2015; Wang, Chen and Lin,
2015). In addition, the MHDS also encodes motivational states of fear that are associated with
predatory threats (Kunwar et al., 2015). Surprisingly, the defensive responses to an aggressive
conspecific require several hypothalamic nuclei that do not overlap with the ones recruited by
exposure to a predator (Figure 4). This set of nuclei includes the medial preoptic nucleus (MPN),
11
ventrolateral portion of the ventromedial hypothalamus (VMHvl), the ventral premammillary
nucleus (PMV) and dorsomedial portion of the dorsal premammillary nucleus (PMDdm)
(Canteras, 2002; Motta et al., 2009; Lin et al., 2011; Wang et al., 2019). Lesion of VMHvl and
PMDdm decreases defensive responses to an aggressive conspecific (Motta et al., 2009), and
direct activation of VMHvl could induce defensive attacks to conspecifics through its projections
to the PAG (Wang et al., 2019).
Besides the above structures, several other hypothalamic nuclei also contribute to defensive
behaviors. The lateral hypothalamus area (LH) drives evasion and attacks via PAG and escape
behaviors via lateral habenula (LHb) (Lecca et al., 2017; Y. Li et al., 2018). Moreover, the
dorsomedial hypothalamus (DMH) and paraventricular hypothalamus (PVH) could trigger
escape or jumping behaviors (Ullah et al., 2015; Mangieri et al., 2019). In summary, the
hypothalamus plays a fundamental role in integrating signals from external threats, orchestrating
defensive responses and adjusting internal motivational states to promote survival. It is important
to note that the hypothalamus is composed of many neuropeptidergic cell populations that direct
different behaviors (Lovett-Barron et al., 2020). A better understanding of how different sub-
populations cooperate in different contests in addition to population-level analysis is important
for determining how animals select or switch behavioral strategies during agonistic encounters.
The amygdala is another important integration center for fear responses. It is necessary for
association of neutral stimuli with an appetitive or aversive outcome, which is crucial for
conditioned fear behaviors. The amygdala consists of several interconnected subnuclei with
distinct anatomical and physiological features (Figure 5). The lateral amygdala (LA) and
basolateral amygdala (BLA) receives most of the sensory inputs that specify fear associations
(Fanselow, 1994). The BLA has strong reciprocal connections with the cortex, especially the
12
medial prefrontal cortex (mPFC), and is considered as the main site for synaptic plasticity during
acquisition and extinction of conditioned fear (Shaban et al., 2006; Ehrlich et al., 2009).
Activation of BLA is sufficient to associate sensory cues with unconditioned fear responses
Figure 5. The amygdala
Some of the main amygdala nuclei and their inputs and outputs, emphasizing the complex
internal architecture of this structure (Adapted from Adolphs, 2013).

(Johansen et al., 2010), and block BLA could impair the fear extinction process through its
projections to mPFC (Mao, Hsiao and Gean, 2006; Quirk and Mueller, 2008). The BLA projects
strongly to the central nucleus of the amygdala (CeA), in which distinct subpopulations could
mediate different fear responses. For instance, the inhibitory output from CeA to vlPAG induce
freezing responses generated by Pavlovian conditioning (Kim et al., 2013; Tovote et al., 2016).
Moreover, the intricate inhibitory microcircuits within CeA gates the expression and selection of
defensive responses (Haubensak et al., 2010; Fadok et al., 2017). Specifically, the somatostatin
positive (SOM+) and protein kinase C (PKC-δ) positive neurons inhibits each other to generate
13
opposite behavioral outcomes, and PKC-δ could inhibit the output neurons in medial sector of
CeA to control the level of conditioned freezing.
In addition to the learned fear responses, some studies suggest that the amygdala might also have
a role in innate defense circuits by relaying aversive sensory stimuli to the midbrain or
hypothalamus (Mongeau et al., 2003; Miller et al., 2019; Silva, Burns and Gräff, 2019). Overall,
the amygdala plays an important role in mediating the defensive behaviors induced by learned
cues through conditioning. However, to achieve a better understanding of the mechanisms
underlying fear processing within amygdala, further detailed mapping of the circuits based on the
molecular profile or connectivity of amygdalar neurons are required.

1.2.3 Cortex and thalamus – input
Three main classes of threats inducing predators, attacks by conspecifics and danger signals in
the environment are detected by the brain via different sensory modalities, including vision,
audition, olfaction and nociception. In general, sensory cues representing threats are relayed
from the thalamus to cortex then fed into higher association cortices or integration centers to
generate fear responses (Figure 5).
Animals frequently depend on their sense of hearing to detect threats, especially nocturnal
animals. As a result, auditory cues have been widely used to evoke defensive behaviors in
associated learning, where animals learn to paired a painful stimulus with neutral auditory cues
(Fanselow and Ledoux, 1999; Fanselow and Wassum, 2016). The primary auditory cortex (A1)
is recruited during the conditioning (Peter et al., 2012), and silence which could impair fear
acquisition, extinction and retrieval (Weible et al., 2014; Nomura et al., 2015). In addition, both
14
loud noise and ultrasound are shown to trigger robust flight responses, which is mediated by the
A1-IC pathway (Mongeau et al., 2003; Xiong et al., 2015).
Visual cues could also evoke several innate defensive behaviors. For example, the primary visual
cortex (V1) could excite SC to induce a temporary “arrest” behavior induced by presenting a
flash light in front of moving animals (Liang et al., 2015). Moreover, rodents display defensive
responses including flight, shelter seeking and freezing to looming shadows in the upper visual
field (Maimon, Straw and Dickinson, 2008; Yilmaz and Meister, 2013). Although the circuits
underlying these behaviors are still under investigation, several circuits centered with the SC
have been proposed for the processing of looming stimuli-induced defensive responses. Amongst
the V1 is shown to amplify the SC’s responses to looming stimuli, which may modulate the
magnitude of the defensive responses (Zhao, Liu and Cang, 2014).
In contrast to humans, rodents rely on their sense of smell to collect information about the
environment. The detection of odorants signaling danger relies on two main systems: the main
olfactory system (MOS) and the accessory olfactory system (AOS). The MOS is capable of
sensing a wide range of volatile molecules conveying information about the environment,
whereas the AOS appears to be responsible for sensing other individuals through pheromones
(Dulac and Torello, 2003; Carvalho et al., 2020). Information then converges into the
hypothalamus and amygdala to elicit different defensive behaviors (Figure 6).
Although painful stimuli like an electrical foot shock is often used in fear conditioning, the
circuits underlying fear responses to noxious stimuli have not been investigated in depth. The
pain pathway shares many common nodes with the defensive networks. The nociceptors convey
noxious information from the spinal cord. The projection neurons then transmit information to
the somatosensory cortex (SSp) via brainstem nuclei such as the rostral ventromedial medulla
15
(RVM) and PAG and the thalamus (Basbaum et al., 2009). In addition, painful stimuli can be
directly perceived by the CeA (Wilson et al., 2019), which could drive various defensive
behaviors through the midbrain.
Figure 6. Schematic representation of the neural circuits mediating innate fear to different threats
Acoustic inputs (such as ultrasounds) are processed by the auditory cortex (AuC), which in turn
projects to the inferior colliculus (IC) that sends afferents to the dorsal periaqueductal gray
(PAGd). Moving visual stimuli in the upper visual field are processed by the superior colliculus
(SC), which receives inputs from the retinal ganglion cells (RGN) and primary visual cortex
(V1) and mediates fear responses through targeting the amygdala and brainstem. Olfaction plays
a crucial role in the detection of both predator (orange) and conspecific (yellow) signals. The
main olfactory system (MOS) mediates defensive responses to the predator odor via projections
to the cortical amygdala (CoA), but the outputs of this structure mediating behavioral responses
remain unclear. The accessory olfactory system (AOS) signals conspecific cues to the posterior
dorsal portion of the medial amygdala (MEAdd) and predator cues to its posterior ventral portion
(MEApv). These two medial amygdalar nuclei project to the conspecific and predator integration
circuits in the hypothalamus. The predator fear circuit also receives polymodal sensory
information about the threat via a basolateral amygdala (BLA)-basomedial amygdala (BMA)
circuit. The hypothalamic integration unit processing conspecific fear includes four highly
16
interconnected nuclei: the medial preoptic nucleus (MPN), the ventrolateral portion of the
ventromedial hypothalamic nucleus (VMHvl), ventral premammillary nucleus (PMV), and
dorsomedial portion of the dorsal premammillary nucleus (PMDdm). The conspecific fear circuit
mediates defensive responses through its projections to the PAGd. The predator fear circuit
consists of the anterior hypothalamic nucleus (AH), the VMHdm, and the PMD and mediates
defensive responses through projections to the PAGd. Importantly, both the conspecific and
predator hypothalamic circuits receive nociceptive information from the parabrachial nucleus
(PB). Defense to painful stimuli (blue lines) such as an electrical footshock is mediated by
activation of the ventrolateral periaqueductal gray (vlPAG) via the central nucleus of the
amygdala (CEA). The CEA receives noxious information from the parabrachial nucleus (PB).
The basolateral amygdala complex (BLA) plays a major role in footshock-induced fear through
its projections to the CEA. The BLA integrates nociceptive information from the PAG via
midline thalamic nuclei (MTN). (Adapted from Silva, Gross and Gräff, 2016).

To sum up, the involvement of cortices and thalamus in defensive behaviors is highly dependent
on the specific sensory cues and the detailed circuits remains further explorations. It is important
to note that cortex could also effectively modulate defensive behaviors (Mobbs et al., 2009;
Courtin et al., 2013), which I will briefly cover in next section. In Chapter 3, I will present a
study showing the contribution of multimodal integration in defensive behaviors.

1.3 Flexible, context-dependent defensive networks
From what I have reviewed in previous sessions, defensive networks seem overly complex and
somewhat “redundant”. However, it is the complexity allows animals more flexibility in the
selection of defensive strategies to ensure survival. The imminence to threats in natural
environment is always dynamic, and animals need to instantly weigh different options. For
instance, fight or flight when the threat is at proximity, hide, freezing or flight when the predator
is in distance. Other factors such as past experiences, metabolic needs, familiarity with the
environment and possible conspecifics nearby could also contribute to the defense selection
(Warden et al., 2012; Sternson, 2013; Mobbs et al., 2018). Here I will introduce the flexibility of
17
defense system from three aspects, switches between strategies, context-dependent modulation of
defensive behaviors, and experience-dependent modulation of defensive behaviors.

1.3.1 Freezing or flight
Extensive studies on fear circuits have focused on learned fear through conditioning (Fanselow,
1994; LeDoux, 2012). However, the predominant behavioral response to conditioned stimuli is
freezing. Both freezing and flight can be elicited from exposure to a natural predator, and a
combination of both defense strategies can be observed (Mongeau et al., 2003; Yilmaz and
Meister, 2013; Wei et al., 2015). Survival in a volatile world is dependent on rapid selection of
appropriate actions. Therefore, understanding how distinct neuronal circuit elements interface to
generate switches between behavioral states is important.
A few circuits have been proposed to control defense action selection. As outlined earlier, the
PAG drives a variety of defensive responses (Xiong et al., 2015; Tovote et al., 2016).
Specifically, the excitatory neurons in dlPAG control flight whereas the excitatory neurons in
vlPAG control freezing. Interestingly, the excitatory neurons in dlPAG could excite inhibitory
neurons in vlPAG to suppress the “freezing” neurons to facilitate the switch from freezing to
flight as shown in the circuit model (Figure 7). Same functional motif exists in another important
defense integration center, the amygdala. The excitatory responses to a fear cue in the BLA
activate PKCδ- cells, some of which are SOM
+
in CeA, to initiate freezing. By contrast,
excitatory responses in the BLA activate PKCδ
+
CeA cells to promote anxiolysis (Janak and Tye,
2015). In addition, corticotropin-releasing factor (CRF) positive cells in CeA mediate flight, and
SOM
+
CeA neurons initiate passive freezing. More importantly, the balance between flight and
freezing behaviors is regulated by means of local inhibitory connections between CRF
+
and
18
SOM
+
neurons (Fadok et al., 2017). Together, these results suggest that the interactions of
distinct types of neurons within some highly organized nuclei are critical for any higher brain
function. However, these experiments were carried out in specific context and effects of other
neuropeptidergic populations were ignored. Future studies synthesize molecules, circuits and
behavioral contexts are necessary to achieve a better understanding of the circuits underlying
defense strategy selection.
Figure 7. Schematic representation of the freezing pathway
The excitatory neurons in vlPAG could disinhibit excitatory neurons in dlPAG to allow switch
between freezing and flight. PN, projection neuron; IN, interneuron; MN, motor neuron
(Adapted from Tovote et al., 2016).

In summary, defensive behaviors, and specifically freezing and flight, are not mutually exclusive
responses. Rather, they are integrated into a more complex and flexible, context-dependent
19
defensive scheme. Although several local inhibitory microcircuits are shown to balance freezing
and flight, the circuits allow dynamically switches between them based on the environmental
context remain largely unclear. Looming stimulus is innately aversive and evokes a mix of
freezing and flight responses. Despite many efforts on the downstream mediating different
actions, little is known about how visual information processed in the SC to covey the
threatening information. In chapter 2, I will introduce a study showing differential laminae
control of visual defense behaviors in the SC.

1.3.2 Context-dependent modulation of defensive behaviors
Features of the threatening stimulus and the situations in which it is encountered modulate the
specific defensive behavior (Blanchard, Blanchard and Rodgers, 1991). For example, the
presence of an escape route of a refuge will promote flight (Blanchard, 2004). If these
environmental features are not accessible, freezing might be the dominant response. Active
evaluation of the environment is crucial for animal survival.
However, a limited number of studies have examined how similar threatening stimuli can elicit
different defensive responses depending on the context. A recent study identified that rapid
spatial learning could bias freezing towards escape in respond to looming stimuli (Vale, Evans
and Branco, 2017). Specifically, the authors showed that mice could memorize the shelter
location in a single trial. They could then instinctively escape to the shelter without any related
cues, instead of freezing for the most time. Moreover, mice could rapidly update the changing
environment (shelter location) and incorporate the information into action. This study provides
some insights on how the brain engages risk assessment with behavioral choice. In addition,
Mongeau and his colleges (Mongeau et al., 2003) showed that context could modulate animals’
20
defense strategies. In particular, animals show freezing by recruiting cortical-amygdalo-striatal
pathway when exposed to an aversive ultrasound in novel environment, whereas same stimulus
evokes flight in home cage by recruiting the ventral lateral septum and periventricular zone of
the hypothalamus.
Nature environment is filled with a full spectrum of physical characteristics, including
luminosity, sound, scents and temperatures in a dynamic and interactive way. However, how
information about the environment is integrated to elicit appropriate defense remains largely
unknown. Multisensory integration could affect defensive behaviors in several ways. One study
has shown that learned avoidance to aversive context (chamber with foot-shock) requires
integration of olfactory and visual pathway (Masuda et al., 2013). Moreover, aversive stimuli
cause animals to associate their environmental context, allowing defensive behaviors during
future exposure. This process requires the brain to incorporate the multisensory features of the
context but exclude the aversive sensory cue. Previous study showed that subcortical multimodal
inputs drive the hippocampus CA1 interneurons to selectively inhibit integration of the excitatory
inputs carrying neutral information, which allows the hippocamps process the aversive sensory
cues by excluding sensory features of the context (Lovett-Barron et al., 2014). Finally,
multisensory integration could affect emotional states (Collignon et al., 2008; Taffou et al.,
2017), which is important for the expression of defensive behaviors.
Taken together, the learning and expression of defensive behaviors are highly dependent on the
context. An important issue is that studies of the neural correlates of fear in rodents have been
limited to single-modality-motor processing of defensive responses to a threat and only few
studies have tried to uncover the brain circuits integrating multimodal processing of fear. In
21
chapter 3, I will introduce a study showing cross-modality modulation of the level of defensive
behavior through a cortico-subthalamic-thalamic pathway.

1.3.3 Experience-dependent modulation of defensive behavior
Another major type of modulation of defensive behaviors is experience dependent. After
repetitive exposure to the threatening stimuli without actual harm, animals could adapt their
defensive responses. For instance, the looming-evoked defensive responses showed a quick
adaptation after several representations of the stimuli (Yilmaz and Meister, 2013; Wei et al.,
2015). The sound-evoked defensive behavior showed limited adaptation, which is highly
dependent on the sound intensity (Xiong et al., 2015). It is probably due to cues associated less
certain threats could be adapted to increase energy efficiency to ensure survival.
Many efforts have been made to study the neuronal circuits that engaged for evaluating and
storing fear memories and enable animals to adapt to changing contingencies, mostly with
conditioning paradigms (Fanselow, 1994; Fanselow and Ledoux, 1999; Barad, Gean and Lutz,
2006; Courtin et al., 2013; Xu and Südhof, 2013; Chou et al., 2018). Extinction is characterized
as a decreased in the amplitude and frequency of a conditioned response when the conditioned
stimulus that elicits it is repeatedly nonreinforced (Myers and Davis, 2002). The amygdala and
mPFC have emerged as key structures for fear extinction (Herry et al., 2010; Marek et al., 2013).
Inhibition of the neuronal activity or disruption of the synaptic plasticity in the BLA impair
extinction (Sotres-Bayon, Bush and LeDoux, 2007; Amano et al., 2011). The plasticity within
the BLA is essential for forming long-term extinction memories (Mao, Hsiao and Gean, 2006).
The BLA neurons are mutually connected with different subregions of the mPFC (ILA,
22
infralimbic; PL, prelimbic). The PL-projecting neurons transmit fear signals and the ILA-
projecting neurons convey extinction signals (Senn et al., 2014). This is supported by a series
studies showing that the ILA neurons exhibit increased activity during extinction retrieval and
silence which could suppress extinction (Milad and Quirk, 2002; Courtin et al., 2013).
Apart from the amygdala and mPFC, several other structures are important for fear extinction.
One important feature of fear extinction is that it is heavily dependent on the context. The
contextual information for the acquisition and retrieval of extinction involves the hippocampus,
which is well connected with the amygdala and mPFC (Hoover and Vertes, 2007; Herry et al.,
2008; Bissiere et al., 2011). Our lab previously identified that GABAergic neurons in the rostral
zona incerta (ZI), a major subthalamic nucleus, could relay information from the mPFC. Its
activity increases during fear extinction and suppress which could impair the extinction process
(Chou et al., 2018). Moreover, the PV
+
neurons in the ZI receive inhibitory input from the SOM
+

neurons in the CeA, and are required for acquisition and remote recall of the fear memory (Zhou
et al., 2018). Interestingly, the ZI is highly connected with different cortical regions and are
shown to be able to integrate multiple sensory modalities (Mitrofanis, 2005; Zhao et al., 2019).
Given its role in defense modulation and multimodal capacity, the ZI is a good candidate for
both context and experience-dependent modulation of defensive behaviors. In chapter 3, I will
elaborate a study showing the ZI integrate auditory and somatosensory information to change the
level of defensive flight responses.
Overall, defensive behaviors can be modulated or adapted after repeated exposure to the same
experience that are uncorrelated with danger. Extinction involves intricate functional changes in
defined long-range circuits that link to the amygdala and mPFC. These circuits work in parallel
with the defense “driver” circuits to make the animals better adjust with the changing
23
environments. However, the mechanisms by which they interact with other brain structures are
not fully understood. In addition, the circuits underlying adaption in innate defensive responses
remain largely unclear. Further studies characterizing these could lead to a better understanding
of fear and extinction.

24
Chapter 2: Differential Circuit Control of Visual Defense
Behaviors in Superior Colliculus

2.1 Introduction
The superior colliculus (SC) is a laminated midbrain structure important for multimodal
integration and sensorimotor transformation. It mediates a variety of behaviors ranging from
cognition to motor functions, such as attention, orientation, navigation and defense (Alex
Meredith and Stein 1983; Gandhi and Katnani 2011; May 2006; Meredith and Stein 1986; Wurtz
and Albano 1980; Zingg et al. 2017). In order for the SC to coordinate different behaviors, it
requires integration of top-down sensory information. In mammals, the SC is composed of three
superficial (zo, zonal; SuG, superficial grey; op, optic), two intermediate (Ig, intermediate grey;
Iw, intermediate white), and two deep layers (dg, deep grey; dw, deep white) (May, 2006). The
superficial layers mainly receive visual information (Hofbauer and Holländer, 1986; Tardif et al.,
2005). Nonvisual information including somatosensory and auditory maps also converge into
intermediate and deep layers of the SC (Drager and Hubel, 1975; Edwards et al., 1979). Neurons
in the SC are able to alter their sensitivity to external events by integrating multimodal sensory
inputs, and many of them have descending projections to the premotor and motor areas of the
brainstem and spinal cord (Drager and Hubel, 1975; Alex Meredith and Stein, 1983; Meredith
and Stein, 1986; Liang et al., 2015). Therefore, it is important to understand the how the sensory
information is integrated and channeled to downstream motor systems to regulate a diverse
repertoire of behaviors.
The superficial SC (sSC) is an important structure for visual information processing and
mediating visuomotor behavior. It receives strong inputs from retina ganglion cells (RGCs) as
well as primary visual cortex (V1) and higher visual cortical areas (Chalupa and Rhoades 1977;
25
Feig, Van Lieshout, and Harting 1992; Hofbauer and Holländer 1986; McLaughlin, Hindges, and
O’Leary 2003; Q. Wang and Burkhalter 2013; Zhao, Liu, and Cang 2014; Zingg et al. 2017).
Anatomical results show zo and SuG receive direct inputs from RGCs axons that course in the
op, following a retinotopic organization. The majority of V1 afferents to sSC terminate ventrally
to the retinal afferents in lower SuG and op, with some dSC projections. In addition, the ventral
visual stream preferably target the intermediate layers (iSC), and the dorsal visual stream mostly
project to dSC (Wang et al., 2010; Wang and Burkhalter, 2013). Such an anatomical pattern
suggests depth-specific function organizations in the sSC. Indeed, recording results shows that
neurons in the most superficial SC (~50 µm) have strong direction selectivity but less
orientation-selective, and it is consistent with the differential projection pattern of different
RGCs subtypes (Gabriel et al., 2012; Dhande and Huberman, 2014; Inayat et al., 2015).
sSC neurons could contribute to visually guided behavior in several different ways. The can
product to the lateral posterior nucleus of the thalamus (LP) to affect sensory processing in visual
cortex (May, 2006; Fang et al., 2020). They can also projects to the deep layers of SC (dSC) and
other downstream targets to mediate visually-guided behaviors (Gandhi and Katnani, 2011;
Mysore and Knudsen, 2011). One prominent example of visuomotor behaviors in rodents is the
defensive reaction towards an approaching aerial predator. Freezing or flight responses can be
triggered by an overhead display of looming stimuli (Yilmaz and Meister, 2013b; De Franceschi
et al., 2016a). Previous studies have revealed that sSC neurons display robust responses and
speed tuning to looming stimuli (Zhao, Liu and Cang, 2014), and the SC controls different
looming-evoked defensive response through divergent pathways (Shang et al., 2015; Shang,
Chen, Liu, Li, Zhang, Qu, Yan, Zhang, Liu, Liu, Guo, Li, Wang and Cao, 2018). However,
whether different laminae and visual channels contribute to looming-induced dimorphic
26
responses remains unclear. In the present study, we systematically examined the roles of retinal-
recipient and cortical-recipient SC neurons’ responses properties, and their role in looming-
evoked defensive behaviors. Our data indicate that the retinal-recipient and cortical-recipient
neurons mediate dimorphic defensive behaviors by preferentially target different downstream.
Figure 8. The SC is essential for looming-evoked defensive behaviors
(A) Upper left, experiment paradigm for open arena, scale: 10 cm. Upper right, movement
tracking for 2 example mice in an open arena while presenting looming stimuli. Blue, freezing;
red, flight route. Lower left, experiment paradigm for arena with hiding nest, scale: 10 cm. Right,
movement tracking for 2 example mice in arena with nest while presenting looming stimuli. Red,
flight route; green, nest. (B) Upper, speed trace of an example mouse during visual stimuli in an
open arena. Lower, speed trace of an example mouse during visual stimuli in an open arena. (C)
nest
stim
0
30
Speed(cm/s)
Time(s)
stim
0
50
Speed(cm/s)
flight
0
Flight Prob. (%)
50
Ctr. SC
100
11/11
5/10
Ctr. SC
0
50
100
Freezing (%)
✱✱✱
✱✱
Ctr. SC
0
5
Flight Latency (s)
10 + muscimol
SC
SC
SC
0 10
1
5
2
3
4
Animal#
0 10
0 10 0 10
1
5
2
3
4
freezing
0 10
Time(s)
Animal#
stim
0
50
Speed(cm/s)
Time(s)
0 10
ns
Ctr. Stim.
0
50
100
Freezing (%)
ns
0
Peak speed (cm/s)
30
Ctr. Stim.
90
60
A
B C
D
E F G H I
J K L M
1
5
2
3
4
1
5
2
3
4
Time(s)
0 10
freezing flight
1
5
2
3
4
Animal#
Time(s)
27
Comparison of two example conditions. Upper panel reveals open field, lower panel reveals
arena with hiding nest. For each condition, an ethogram indicates occurrences of three behaviors
after stimulus onsite. Green, nest; red, flight; blue, freezing. (D) Experiment paradigm, muscimol
were injected into the SC bilaterally. (E) Ethograms showing behaviors in an open arena after SC
silence. (F) Summary of freezing percentage after silencing SC in an open arena compared to
control. ***p < 0.001, two-sided unpaired t-test. (G) Ethograms showing behaviors in arena
with nest after SC silence. (H) Summary of probability to flight after silencing SC in arena with
nest compare to control. (I) Summary of response latency after silencing SC in arena with nest
compare to control. ** p<0.01, two-sided unpaired t-test. (J) Left, Experiment paradigm; right,
speed trace of an example animal in respond to receding stimuli in an open arena. (K) Ethograms
showing behaviors in respond to receding stimuli in an open arena. (L) Summary of freezing
percentage in respond to receding stimuli compared to control. n.s., two-sided unpaired t-test.
(M) Summary of peak movement speed in respond to receding stimuli compared to control.
n.s., two-sided unpaired t-test. Error bars showing S.E.M..

2.2 Results
2.2.1 Depth-specific response pattern exists in the SC
Previous studies have demonstrated that rodents display defensive responses including flight,
shelter seeking and freezing to looming shadows in the upper visual field, and this behavior is
mediated by the SC (Yilmaz and Meister, 2013c; Wei et al., 2015; De Franceschi et al., 2016b).
We repeated this experiment by placing wild-type animals into a behavioral arena with a display
monitor covering the ceiling (Figure 8, upper). In another setting, a nest was placed in one corner
of the arena offering hiding place from visual stimuli (Figure8, lower) (Yilmaz and Meister,
2013b). The animals were allowed 10 min of habituation prior to testing. The looming stimuli
were presented as an expanding black disc appeared directly above the animal on a grey
background from 2 degrees to 20 degrees visual angle in 250 ms and remained at 20 degrees for
250 ms. The stimulus was presented 10 times with 500 ms interval. In an open arena without
hiding nest, looming stimuli could trigger one or both flight and freezing responses. While in
arena with hiding nest, looming stimuli could trigger robust flight responses, which is consistent
28
with previous findings (Figure 8B-C) (Yilmaz and Meister, 2013b; Shang, Chen, Liu, Li, Zhang,
Qu, Yan, Zhang, Liu, Liu, Guo, Li, Wang and Cao, 2018). We next confirmed SC’s role in
looming-evoke defensive behaviors by silencing the SC through infusion of a GABA agonist
muscimol bilaterally (Figure 8D). We observed that in open arena, silence of SC significantly
impaired both flight and freezing responses (Figure 8E-F). In arena with hiding nest, silence of
SC would decrease animals’ probability of flight and increase their response latency to flight
(Figure 8H-I).
The SC is a complex visual center. In rodents, it inherited retinotopic visual information from
retina (McLaughlin, Hindges and O’Leary, 2003) and cortex, whose projections are aligned with
the retinocollicular map (Chalupa and Rhoades, 1977) (Figure 9A). To visualize different visual
inputs to SC, we first injected anterograde transmittable AAV1-Cre in retina, V1 or V2 in Ai14
animals (Zingg et al., 2017a). We observed the majority of tdTomato+ cell bodies were located
in most superficial SC (zo and upper SuG) in retina-injected animals (Figure 9B left). In V1-
injected animals, the tdTomato+ somas were mostly located in lower SuG and op, with sparse
labeling in InW (Figure 9B middle). In V2-injected animals, the tdTomoto+ somas were sparsely
located in the iSC and dSC (Figure 9B right). Our results showed a spatial segregation of
different visual inputs in the SC (Figure 9C). By reconstructing the morphology of the recipient
neurons, we found that the V1 and V2 recipient neurons have larger dendritic arbors (Figure 8A
lower), which indicate they have bigger receptive field (RF). To understand the overall
responsiveness of SC neurons in the superficial and intermediate layers, we first presented 200-
ms visual noise flashes and found these neurons mostly showed robust visually evoked responses
(Figure 10A). In addition to the noise flashes, we presented patterned stimuli, drifting gratings of
29

Figure 9. Depth-specific response pattern exists in the SC
(A) Schematic illustration. Different visual channels project to SC. (B) Left, AAV1-hSyn-Cre
was injected into retina of Ai14 mice. Middle, AAV1-hSyn-Cre was injected into V1 of Ai14
mice. Right, AAV1-hSyn-Cre was injected into V2 of Ai14 mice. Upper panel reveal the
injection sites, scale: 500 µm and td-tomato-labeled SC neurons, scale: 200 µm. Lower panel
reveal the reconstructed recipient-neurons for each type of injections. Blue, Nissl staining. (C)
Quantification of percentage of observed labeled-SC neurons across different depth in retina
(grey), V1(blue) and V2 (red) injected animals. (D) Spatial-temporal RF of SC neurons in most
superficial (left), intermediate (middle) and deep (right) layers. Upper panel reveal the RF size,
lower panel reveal the RF strength. (E) Summary of RF size across different depths. (F) RF
p1/p2 ratio across different depths. (G) Raster plot of example SC neurons response to looming
(left) and receding (right) visual stimuli. (H) Summary of evoked-firing rate of SC neurons to
looming and receding stimuli across different depths. (I) Summary of rise time to peak responses
to looming and receding stimuli across different depths. (J) Summary of responses duration to
looming and receding stimuli across different depths.
% of total labeled cells
Depth (µm)
50
0
100
0–200
200–400
400–600
600–800
800–1000
Retina
V1
V2
0–100
100– 200
200– 300
RF size (deg)
0
20
30
10
400–500
500–600
300–400
0–200 μm 200–400 μm 400–600 μm
ON OFF ON OFF
8°
-1
1
Norm. resp
ON OFF
p2
ON RF
OFF RF
1
0.5
0
0 50 100 150
Norm. strength
Time (ms)
p1
0 50 100 150
Time (ms)
0 50 100 150
Time (ms)
0–200
200–400
400–600
p1/p2 ratio
0
4
6
2
V1 V2 Retina
V2 V1 Retina
SC SC SC
Recon.
B C
D E F
0
40
80
receding
500
0
50
80
FR (Hz)
0 250
Time (ms)
looming
40
40
25
0–200 µm
400–600 µm
0
30
60
500 0 250
Time (ms) 0–200
200–400
400–600
0
80
160
Evoked peak FR (Hz)
Rise time (ms)
0
200
400
Duration(ms)
0
120
240
looming
receding
0–200
200–400
400–600
0–200
200–400
400–600
G
H I J
A
Retina
Vis. Cortex
SC
Defense
?
30

different directions. Consistent with previous results, neurons with strong direction selectivity
(quantified with the direction selectivity index, DSI, see Methods) to moving gratings were
largely seen in the most superficial layer of SC (Figure 10B). In addition, SC neurons were
generally not orientation tuned or weakly tuned, showing relatively low orientation selectivity
index (OSI, see Methods) values (Figure 10C). Furthermore, for neurons tuned to direction or
orientation information (DSI > 0.3 or OSI > 0.3), we observed an over-representation of cardinal
versus oblique angles (Figure 9B-C, insets). These results were consistent with previous SC
studies.
Next, we tested the receptive field (RF) structure of SC neurons by presenting sparse white. The
spatial-temporal RF was determined by spike-triggered average of stimuli and reverse
correlation. The neurons in the most superficial SC tended to exhibit small RF, while deeper SC
neurons come with increasingly larger RF (Figure 9D, upper panels; Figure 9E). Surprisingly, we
found a two-peak pattern of ON RF strength across time (Figure 9D, lower panels), which is
reminiscent of the two-peak responses to noise flashes. Similarly, the p1/p2 ratios were the
largest in the most superficial layer and decreased with depth (Figure 9F), indicating the relative
strength of retinal versus cortical inputs across SC depth. Together, these data suggested some
fundamental laminar differences in visual responses that superficial SC is dominated by retinal
inputs and mostly exhibits high direction selectivity and small RF, while intermediate SC is more
strongly innervated by cortical inputs and shows weak feature selectivity and large RF.
Finally, we examined the neuronal responses to looming versus receding dark disks of SC. We
found that both looming and receding stimuli could trigger robust visual-evoked responses
31
(Figure 9G-H). In the most superficial layer, the peak time of looming evoked responses was
similar to that of receding evoked responses, while in deeper layers, looming responses ramped
up significantly more slowly than the receding responses (Figure 9I). Moreover, the duration of
looming responses was significantly longer than that of receding responses (Figure 9J). These
data suggested that looming stimuli generate relatively slow but more sustained response,
whereas receding stimuli caused fast and transient neural activity, which may serve as neural
bases for the differential behavioral outcomes to looming versus receding stimuli (Figure
9G). To explore behavioral outcome for receding visual stimuli, we placed wild type animals in
an open arena described above without hiding nest (Figure 9J). The animals did not show any
defensive responses when receding dark disks were displayed (Figure 9K-M). In summary, these
experiments confirmed that the SC is essential for looming-evoked defensive behaviors.

Figure 10. Basic response properties in the SC changes over depths
(A) Visual noise-evoked responses in different SC depths. (B) Direction selectivity of SC
neurons across different depths. (L) Orientation selectivity of SC neurons cross different depths.

2.2.2. Cortical inputs regulate the visual responses of the intermediate SC
To further determine the response properties of SC neurons across depths, we presented 1-s long-
term noise patterns (Figure 11A). Interestingly, SC neurons typically exhibited 2-peak responses
0–200
200–400
400–600
1
0.5
0
DSI
(17%)
0–200
200–400
400–600
1
0.5
0
OSI
(33%)
B C
0–200
200–400
400–600
80
40
0
Vis noise
evoked FR (HZ)
A
32
(Figure 11B, gray bars). We quantified the time from stimulus onset to the peaks and found that
the first peak (p1) mostly reached around 40 ms, while the second peak (p2) was around 95 ms
(Figure 11C). The short peak time of p1 suggested it originated from direct feedforward retina-
SC inputs. Indeed, we found that the p1 evoked firing rates were the highest in the most
superficial layer and decreased in the deeper layers (Figure 11D), which is in line with the
anatomical results that retinal inputs preferentially target the most superficial SC layer (Figure
9C, blue line). To examine the relative contribution of feedforward retinal versus feedback
cortical inputs in the response pattern of SC neurons, we silenced V1 with muscimol and
compared visually evoked responses before and after V1 silencing. As expected, the p1 values
were not changed after V1 silencing in all SC layers (Figure 11B-G). However, we observed
laminar-specific heterogenous changes in p2 values: p2 was increased in the most superficial
layer but decreased in deeper layers (Figure 11F). The changes were more prominent when we
quantified the p1/p2 ratios. Before V1 silencing, the most superficial SC was dominated by
retinal inputs (high p1/p2 ratio), while the ratios were reduced with the depth, suggesting a
gradient transition of retina-l to cortical-dominated inputs (Figure 11G, grey bars), which is
again consistent with our anatomical findings (Figure 9C, gray versus blue line). After V1
silencing, the unchanged p1 and changed p2 together caused reduced p1/p2 ratios in the most
superficial SC but enhanced ratios in deeper layers (Figure 11G, red bars). Pervious study has
shown that the SC exhibits robust responses towards looming stimuli(Zhao, Liu and Cang, 2014;
Lee et al., 2020). We next tested the contribution of cortical inputs to SC’s neuronal in respond
to looming stimuli. Similar to the noise responses, silencing V1 increased the looming-evoked
responses in the most superficial SC and decreased the evoked-responses in intermediate and
deep layers (Figure 11H-I). Together, these results further suggest that cortical inputs mainly
33
contribute to the visual responses of intermediate layers of the SC, and a depth-specific
functional organization exists within the SC.

Figure 11. Cortical inputs regulate the visual responses of the intermediate SC
(A) Experimental paradigm. Muscimol was injected into V1 of the same site for recording. (B)
Noise-evoked firing rate of the superficial (left), intermediate (middle), deep (right) layers of the
SC before (grey) and after (red) silencing V1 with muscimol. (C) Rise to peak time for p1 and p2
responses. (D) The p1-evoked firing rate across different depths. (E)-(F) p1 (E) and p2 (F)
responses relative change after silencing V1 with muscimol. (G) The p1/p2 ratio before (grey)
and after (red) silencing V1 with muscimol. (H) Raster plot of an example cell in respond to
looming stimuli before (upper) and after (lower) silencing V1 with muscimol across different
P1/p2 ratio
0
4
6
2
0–200
200–400
400–600
Ctrl
ΔV1
Rise time (ms)
0
50
100
150
0–200
200–400
400–600
50 ms
0
60
120
FR (Hz)
p1
p2
0–200 μm
0
50
100
200–400 μm
0
40
80
400–600 μm
0–200
200–400
400–600
% change ΔV1
-100
0
100
0–200
200–400
400–600
% change ΔV1
100
50
50
100
0
50
-50
p1 p2
p2
p1
p1 evoked FR (Hz)
0
50
100
150
0–200
200–400
400–600
A B C
D
E
F
G
Ctrl
ΔV1
I
0-200
200-400
400-600
% change ΔV1
-100
0
100
-50
50
H
0
60
120
0
60
30
0
60
120
FR (Hz)
500 ms
looming stim
0
60
120
0
60
120
0
60
30
Ctr.
ΔV1.
0-200 μm 200-400 μm 400-600 μm
34
depths. Left, 0-200 µm; middle, 200-400 µm; right, 400-600 µm. (I) Summary of evoked-FR
fold change after silencing V1.

2.2.3. Cortico-recipient SC Neurons Mediate Looming-Evoked Freezing Responses
Previous study has shown that directly activate V1-SC pathway could induce freezing responses
(Zingg et. 2017), and our recording data suggested that the retinal and cortical visual channels
contribute to different laminae of the SC. We next investigated how cortical visual inputs impact
the looming-evoked defensive responses. To test this, we performed paired cortical injections of
AAV1-Cre and flex-hM4D(Gi) in SC to enable Cre-dependent expression of hM4D(Gi) in SC
neurons of wild-type mice. We could then inhibit the SC neurons in awake, freely moving mice
by intraperitoneally (IP) injected the hM4D(Gi) agonist, clozapine-N-oxide (CNO) (Figure 12A)
(Zhu and Roth, 2014). We tested looming-evoked responses in both open arena and arena with a
hiding nest. Mice expressing hM4D(Gi) in V1-recipient SC neurons showed no changes in flight
responses (Figure 12B-D), but significant reduction of their freezing time (Figure 12E-F).
Mouse higher visual areas also project strongly to SC (Meredith and Stein, 1986; Wang and
Burkhalter, 2013). Specifically, neurons in the posterior area (P/POR) projects to the medial
parts of sSC, which respond well to dorsal receptive fields. While neurons in anteromedial area
(AM) project sparsely to iSC, which respond well to central receptive fields (Wang et al., 2010;
Glickfeld, Reid and Andermann, 2014; Cang et al., 2018). We then tested P/POR and AM’s
effects on looming-evoked defensive responses respectively. Infusing muscimol into P/POR
bilaterally (Figure 12G) did not affect looming-evoked flight responses (Figure 12H-J) but
significantly suppressed looming-evoked freezing responses (Figure 12K-I). On the contrary,
infusing muscimol into AM did not alter either looming-evoked defensive responses (Figure
35
12M-N, Figure 13). In addition, we specifically activated the V1-SC and V2-SC pathways by
directly optogenetically stimulating the axon terminals over the SC in animals injected with
ChR2 in V1 and V2 respectively (Figure 12O). Activation of both pathways could trigger strong
freezing responses (Figure 12P-Q). Together, our data demonstrated cortico-recipient SC
neurons are essential for looming-evoked freezing responses.

Figure 12. Cortico-recipient SC Neurons Mediate Looming-Evoked Freezing Responses
(A) Experimental paradigm. AAV-Cre was injected bilaterally into the V1 and cre-dependent
DREADDi was injected bilaterally into the SC. (B) Ethograms showing behaviors in an arena
with nest after silencing V1-recipient SC neurons. (C) Summary of flight latency after silencing
V1-recipient SC neurons compared to control. n.s., two-sided unpaired t-test. (D) Summary of
peak flight speed after silencing V1-recipient SC neurons compared to control. n.s., two-sided
unpaired t-test. (E) Ethogram showing behaviors in an open arena after silencing V1-recipient
SC neurons. (F) Summary of freezing percentage after silencing V1-recipient SC neurons
compared to control. ***p<0.001, two-sided unpaired t-test. (G) Experimental paradigm.
Muscimol was bilaterally injected into the P/POR. (H) Ethograms showing behaviors in an arena
1
5
2
3
4
✱✱✱
0
50
100
Ctr. V1-SC
Freezing (%)
✱✱✱
0
50
100
Ctr. POR
Freezing (%)
ns
0
Latency (s)
6
Ctr. POR
3
ns
0
Latency (s)
6
Ctr. V1-SC
3
ns
0
50
Ctr. POR
100
Peak Speed (cm/s)
ns
0
50
Ctr. V1-SC
100
Peak Speed (cm/s)
P/POR
V1
muscimol
SC V1
AAV-Cre
AAV-flex-hDm4(Gi)
1
5
2
3
4
1
5
2
3
4
1
5
2
3
4
100
0
50
Ctr.
Freezing (%)
V1-SC
V2-SC
✱✱✱
✱✱✱
✱✱
Speed (cm/s)
SC
V1/V2
470 nm
AAV-ChR2
A B C D E F
G H I J K L
O P Q
Time(s)
0 10
Time(s)
0 10
12
V1
V2
0
8
0
Time(s)
0 10
Time(s)
0 10
Time(s)
0 10
AM
V1
muscimol
M
ns
0
50
100
Ctr. AM
Freezing (%)
N
36
with nest after silencing P/POR-recipient SC neurons. (I) Summary of flight latency after
silencing P/POR-recipient SC neurons compared to control. n.s., two-sided unpaired t-test. (J)
Summary of peak flight speed after silencing P/POR-recipient SC neurons compared to control.
n.s., two-sided unpaired t-test. (K) Ethogram showing behaviors in an open arena after silencing
P/POR-recipient SC neurons. (L) Summary of freezing percentage after silencing P/POR-
recipient SC neurons compared to control. ***p<0.001, two-sided unpaired t-test. (M)
Experimental paradigm. Muscimol was bilaterally injected into the AM. (Q) Summary of
freezing percentage after silencing AM-recipient SC neurons compared to control. n.s., two-
sided unpaired t-test. (O) Experimental paradigm. ChR2 was bilaterally injected into the V1 or
V2, and fibers were implanted over SC bilaterally. (P) Speed trace of example mice after
activating V1-SC (upper) and V2-SC (lower) pathways. (Q) Summary of freezing percentage
after activating V1-SC or V2-SC pathways compared to control. **p<0.01, ***p<0.001, two-
sided unpaired t-test. Error bars showing S.E.M..

2.2.4. Retina-recipient SC Neurons Mediate Looming-Evoked Flight Responses
To confirm our cortical silencing results, we further blocked the visual thalamic inputs to cortex
by injecting flex-hM4D(Gi) into the dorsal lateral geniculate body (dLGN) of Vglut2-Cre
animals (Figure 14A). Consistent with previous results, silencing dLGN during looming stimuli
with CNO significantly suppressed freezing responses (Figure 13B-F) but did not change their
flight responses (Figure 14B-D).
The SC is one of the most prominent retinal targets in the mouse. The majority of mouse RGCs
project to the most superficial sSC (Wang et al., 2010; Cang et al., 2018). Their axons are
organized in a retinotopic manner and provide a source of direction selectivity to the SC neurons
(Wang et al., 2010; Shi et al., 2017), which is consistent with our previous recording results. We
next explored retinal inputs’ role in the looming-evoked defensive responses. To test this, we
first performed paired retinal injections of AAV1-Cre and flex-hM4D(Gi) in sSC of WT mice
(Figure 14G). After infusing CNO intraperitoneally, mice expressing hM4D(Gi) in retina-
recipient sSC neurons showed increase latency to fight and reduced peak flight speed (Figure
14G-J), without significant changes of their freezing responses (Figure 14K-L). We then
37
specifically activated the retina projection to SC by injecting ChR2 into retina and
optogenetically stimulate the axon terminals over SC (Figure 14M), which could directly induce
locomotion in the animals (Figure 14N). More specifically, activation of retina-SC terminals
increased the peak moving speed of the animals and biased the animal towards locomotion
(Figure 4O-Q). Taken together, our results demonstrated retina-recipient SC neurons are
essential for looming-evoked flight response.

Figure 13. Silencing AM did not affect looming-evoked defensive behaviors
(A) Ethogram showing behaviors in an open arena after silencing AM-recipient SC neurons. (B)
Summary of flight latency after AM-recipient SC neurons compared to control. n.s., two-sided
unpaired t-test. (C) Summary of peak flight speed after silencing AM-recipient SC neurons
compared to control. n.s., two-sided unpaired t-test. Error bars showing S.E.M..

2.2.5. Distinct Behavioral Functions for SC Downstream Targets
Previous studies have identified LP as a major downstream of sSC and directly activate SC-LP
pathway could evoke immediate freezing responses (Shang, Chen, Liu, Li, Zhang, Qu, Yan,
Zhang, Liu, Liu, Guo, Li, Wang, Cao, et al., 2018; Fang et al., 2020). dSC also receives direct
inputs from sSC (Zingg et al. 2017). Activity of the excitatory neurons in dSC can gate the
activity of excitatory neurons in the dorsolateral periaqueductal grey (dlPAG), which directly
ns
Latency (s)
6
Ctr. AM
0
3
ns
0
50
Ctr. AM
100
Peak Speed (cm/s)
A B C
1
5
2
3
4
Time(s)
0 10
38
controls flight behaviors (Xiong et al., 2015; Tovote et al., 2016; Evans et al., 2018). Our lab
previously have identified that retina-SC and V1-SC projections share common downstream
pathways (Zingg et al. 2017). Here, we hypothesized that different visual channels might drive
distinct downstream target to mediate divergent looming-evoked responses. To test this, we first

Figure 14. Retina-recipient SC neurons mediate looming-evoked flight responses
(A)Experiment paradigm. Cre-dependent DREADDi was injected into dLGN of Vglut2-Cre
mice bilaterally. (B) Ethograms showing behaviors in an arena with nest after silencing dLGN
neurons. (C) Summary of flight latency after silencing dLGN neurons compared to control. n.s.,
two-sided unpaired t-test. (D) Summary of peak flight speed after silencing dLGN neurons
compared to control. n.s., two-sided unpaired t-test. € Ethogram showing behaviors in an open
arena after silencing dLGN neurons. (F) Summary of freezing percentage after silencing dLGN
neurons compared to control. **p<0.01, two-sided unpaired t-test. (G) Experimental paradigm.
AAV-Cre was injected bilaterally into the retina and cre-dependent DREADDi was injected
bilaterally into the SC. (H) Ethograms showing behaviors in an arena with nest after silencing
retina-recipient SC neurons. (I) Summary of flight latency after silencing retina-recipient SC
0
50
100
Ctr. ret.-SC
Freezing (%)
ns
0
5
Latency (s)
10
Ctr. ret.-SC
✱✱
0
30
60
Ctr. ret.-SC
90
Peak Speed (cm/s)
✱✱
retina
SC
AAV-Cre
0
50
100
Ctr. dLGN
Freezing (%)
✱✱
Ctr. dLGN
ns
0
Latency (s)
10
5
ns
0
Ctr. dLGN
80
Peak Speed (cm/s)
40
dLGN
Vglut2-Cre mice
0
50
100
Ctr. reti.-SC
Freezing (%)
ns
Norm. Peak Speed
0
1
2
3
Ctr. reti.-SC
✱✱
LP
A B C D E F
G H I J K L
M N O P
Q
1
5
2
3
4
0
40
Speed (cm/s)
0 10
Time (s)
0
25
50
Percentage (%)
6 18 30
Speed (cm/s)
1
5
2
3
4
Time(s)
0 10
Time(s)
0 10
1
5
2
3
4
Time(s)
0 10
1
5
2
4
3
Time(s)
0 10
ChR2
retina
SC
flex-hDm4(Gi)
flex-hDm4(Gi)
39
neurons compared to control. **p<0.01, two-sided unpaired t-test. (D) Summary of peak flight
speed after silencing retina-recipient SC neurons compared to control. **p<0.01, two-sided
unpaired t-test. € Ethogram showing behaviors in an open arena after silencing retina-recipient
SC neurons. (F) Summary of freezing percentage after silencing dLGN neurons compared to
control. n.s., two-sided unpaired t-test. (M) Experiment paradigm. ChR2 was injected bilaterally
into the retina and fibers were bilaterally implanted over the SC (N) Speed trace of example mice
after activating retina-SC pathway. (O) Summary of freezing percentage after activating retina-
SC pathway. n.s., two-sided unpaired t-test. (P) Summary of normalized speed after activating
retina-SC pathway. **p<0.01, two-sided unpaired t-test. (Q) Summary of speed distributions of
animals with/without retina-SC activation. Error bars showing S.E.M..

performed paired injections of AAV1-Cre in either retina or V1 and Cre-dependent GFP in SC of
WT mice to visualize their axonal projections respectively (Figure 15A left). We found that
retina-recipient SC neurons preferably target dSC but not LP, whereas V1-recepient SC neurons
preferably target LP instead of dSC (Figure 15A right). We next injected a retrograde tracer,
cholera toxin B (CTB), into LP and dSC respectively (Figure 15B left). The results were
consistent with our antegrade tracing, showing that LP predominantly receives inputs from lower
SuG and op whereas dSC predominantly receives inputs from zo and upper SuG (Figure 15B
right). Together, the anatomical data suggests different visual channels preferentially target
different downstream.
In order to test the role of different downstream targets in looming-evoked defensive behaviors,
we measured the freezing and flight responses in Vglut2-Cre animals injected with flex-
hM4D(Gi) in LP and dSC respectively. All mice were tested in both open arena and arena with
nest as previously described (Figure 15C, I). In LP-injected animals, we found a significant
reduction in their freezing responses in respond to looming stimuli after infusion of CNO (Figure
15D-E), but there were no significant changes of their flight responses (Figure 15F-H).
Oppositely, in dSC-injected animals, we did not observe significant changes in their freezing
responses in respond to looming stimuli (Figure 15J-K), but a notable increase in their flight
40

Figure 15. Distinct Behavioral Functions for SC Downstream Targets
(A) Left, experimental paradigm, AAV1-hSyn-Cre was injected into retina, V1 and V2 of WT
mice, followed by a second injection of AAV1-CAG-FLEX-GFP into SC. Middle, GFP-labeled
neurons in SC, scale: 500 µm. Right two panels, GFP-labeled axons in LP and dSC, scale: 200
µm. Blue, Nissl staining. (B) Left, experimental paradigm, CTB was injected into LP and dSC in
WT mice. Middle, injection sites, scale: 500 µm. Right, CTB-labeled somas in sSC. scale: 200
µm. Blue, Nissl staining. (C) Experiment paradigm. Cre-dependent DREADDi was injected into
the LP of Vglut2-Cre mice. (D) Ethograms showing behaviors in an open arena after silencing
ChR2
0
150
Speed (cm/s)
Pre. Stim.
0
60
120
Peak Speed (cm/s)
180
✱✱✱
0 10
Time (s)
0
40
80
20 60 100
Speed (cm/s)
Percentage (%)
SCd
SC
SC
V1
retina
AAV-CRE
AAV-CRE
flex-GFP
flex-GFP
A B
Retina
V1
SCd
SCd
LP SC
SCd SC
SC
LP
LP
CTB
dSC
CTB
LP
0
50
100
Ctr. LP
Freezing (%)
0
50
100
Ctr. SCd
Freezing (%)
ns
SCd
LP
flex-hDm4(Gi)
Ctr. SCd
0
Flight latency(s)
4
8
Ctr. LP
0
Flight latency(s)
2
4
✱✱
Ctr. SCd
0 Avg. Flight Speed (cm/s)
10
20
ns
Ctr. LP
0 Avg. Flight Speed (cm/s)
10
20
✱✱✱
C D E F G H
I J K L M N
O P Q
R
1
5
2
3
4
Time(s)
0 10
Time(s)
0 10
Time(s)
0 10
LP
1
5
2
3
4
Time(s)
0 10
flex-hDm4(Gi)
✱✱✱
ns
41
LP neurons. (E) Summary of freezing percentage after silencing LP neurons compared to
control. ***p<0.001, two-sided unpaired t-test. (F) Ethogram showing behaviors in an arena with
nest after silencing LP neurons. (G) Summary of flight latency after silencing LP neurons
compared to control. n.s., two-sided unpaired t-test. (H) Summary of peak flight speed after
silencing LP neurons compared to control. n.s., two-sided unpaired t-test. (I) Experimental
paradigm. Cre-dependent DREADDi was injected into the LP of Vglut2-Cre mice. (J) Ethogram
showing behaviors in an open arena after silencing dSC neurons. (K) Summary of freezing
percentage after silencing dSC neurons compared to control. n.s., two-sided unpaired t-test. (L)
Ethograms showing behaviors in an arena with nest after silencing dSC neurons. (M) Summary
of flight latency after silencing dSC neurons compared to control. **p<0.01, two-sided unpaired
t-test. (N) Summary of peak flight speed after silencing dSN neurons compared to control.
**p<0.01, two-sided unpaired t-test. (O) Experiment paradigm. Chr2 was injected into the dSC
of WT mice bilaterally and optic fibers were implanted over the injection sites. Scale: 500 µm.
(P) Speed trace of example mice after activating dSC neurons. (Q) Summary of peak flight speed
after activating dSC neurons. ***p<0.001, two-sided unpaired t-test. (Q) Summary of speed
distributions of animals with/without dSC activation. Error bars showing S.E.M..

latency and decrease in flight peak speed (Figure 15L-N), indicating an impairment of the flight
response. We further investigated dSC’s function by performing direct optogenetic stimulation of
dSC neurons in WT animals expressing ChR2 in the dSC (Figure 15O). Upon activation, we
found a significant increase in their locomotion (Figure 15P), and the activation biased testing
animals towards locomotor compared to control animals (Figure 15Q-R).
It is known that dSC innervates dlPAG, the commanding center for flight responses (Xiong et
al., 2015; Evans et al., 2018). We next inhibited dlPAG by injecting hM4D(Gi) into dlPAG
bilaterally (Figure 16A). Upon IP infusion of CNO, all tested animals showed prolonged freezing
(Figure 16B). Overall, our results indicated that retina-recipient SC neurons preferentially target
dSC, which is essential for looming-evoked flight responses. Overall, our results indicated that
retina-recipient SC neurons preferentially target dSC, which is essential for looming-evoked
flight responses. V1-recipient SC neurons preferentially target LP, which is essential for
looming-evoked freezing response. These findings are supported with their previously
demonstrated roles in mediating fear responses.
42
Figure 16. Silence dlPAG could potentiate freezing responses
(A) Experiment paradigm. (B) The freezing time significantly increase after silencing PAG
compared to sham. ***p<0.001, two-tail unpaired t-test. Error bar showing S.E.M..

2.3 Discussion
The ability to detect and respond to rapidly approaching predators is an innate defensive
behavior essential for animals’ survival. The superior colliculus (SC) is an important interface
between sensory integration and motor control. Previous studies have demonstrated the SC is
critical for triggering looming-evoked defensive responses (Yilmaz and Meister, 2013b; Shang et
al., 2015a; Wei et al., 2015; Shang, Chen, Liu, Li, Zhang, Qu, Yan, Zhang, Liu, Liu, Guo, Li,
Wang, Cao, et al., 2018). In this study, we identified the retinal-recipient and cortical-recipient
SC neurons are located within different laminae of SC and play distinct roles in looming-evoked
defensive behavior.
In mice, retina input to the contralateral SC terminates in sSC, especially the most superficial
layers – zo and upper SuG (Godement, Salaün and Imbert, 1984; Hong, Kim and Sanes, 2011;
Dhande and Huberman, 2014; Kay and Triplett, 2017). SuG is further subdivided into an upper
layer that receives input from direction-selective retinal ganglion cells (Rivlin-Etzion et al.,
2011; Kay and Triplett, 2017). On the other hand, V1’s projections to SC mostly terminate
PAG
AAV-flex-hDm4(Gi)
A B
✱✱✱
sham PAG
0
40
80
Freezing (s)
43
within lower SuG and upper op and V2’ projections to SC expand from SuG to dSC (Wang and
Burkhalter, 2013). Using a trans-synaptic approach (Zingg et al. 2017) we showed that retina
projects strongly to zo and upper SuG, V1 projects strongly to lower SuG, op and InG, and V2
mostly targets InG, InWh and some deep layers (Figure 8B). This is consistent with previous
study using anterograde approach. In addition, we found that silencing V1 inputs significantly
reduce the p2 peak responses in 200-600 µm but not the most superficial layers (Figure 9), which
further supports that V1 visual information is mostly processing in lower SuG, op and InG.
Conversely, Previous study (Zhao, Liu and Cang, 2014) demonstrated that V1 input exserts a
gain-control to looming-evoked visual responses. This may be due to their recording depth was
restricted to the most superficial SC. Finally, we found that the dendritic arbors of visual-
recipient SC neurons expand along the dorsal-ventral axis, along with their decreased DSI, which
further suggests the origin of visual information moves from retina to more abstract cortical
areas. Our data combined with previous studies suggests depth-specific function organizations
may exist in sSC.
Next, we identified divergent visual inputs converge into different laminae of SC to control
different phases of looming-induced defensive behavior. Specifically, retinal-recipient SC
neurons are necessary for looming-evoked flight responses and cortical-recipient SC neurons are
critical for looming-evoked freezing responses (Figure 9-10). It is support by a recent study
showing that retinal interneurons are essential for looming-evoked defensive behaviors (Kim et
al., 2020). Interestingly, a recent study (Hoy, Bishop and Niell, 2019) showed that the wide-field
neurons are mostly located in op and control for prey detection and approach initiation, while the
narrow-field neurons are sparsely located in sSC and control accurate and continuous
approaches. Moreover, large-scale electrophysiology results showed that sSC is tuned to sensory
44
signals and dSC is more relevant to behavioral signals (Lee et al., 2020). It is possible that
laminar controls of behaviors may be a general functional motif implemented by the SC.
Finally, we demonstrated that the retinal-recipient SC neurons preferentially target dSC to drive
the flight responses and the cortical-recipient SC neurons preferentially target LP to mediate
freezing responses (Figure 5). In align with their previous identified functions (Wei et al., 2015;
Evans et al., 2018; Fang et al., 2020), we showed that dSC is essential for looming-induced
flight responses and LP is necessary for looming-induced freezing responses. The SC is a
complex structure that connected with a variety of brain areas including the cortices, limbic
areas, midbrain and thalamic region. Previous study showed that the projections from SC
parvalbumin+ neurons to the parabigeminal nucleus trigger flight responses (Shang et al. 2015).
In addition, the SC projections to the GABAergic neurons in the ventral tegmental area, an
intermediate station to the central amygdala (CeA), to drive flight responses (Zhou et al., 2019).
Interestingly, stress could activate the tyrosine hydroxylase+ neurons in locus coeruleus, which
projects the SC to potentiate the looming-evoked flight responses (L. Li et al., 2018). Many of
these regions are interconnected and share common pathways. For instance, LP projects to the
basolateral amygdala (BLA), which relays information through CeA to the periaqueductal gray
(PAG) and mediates freezing responses (Wei et al., 2015; Tovote et al., 2016). When exposed to
threat-relevant visual, different circuits may coordinate together to generate an optimal
behavioral output to ensure survival.
How do retinal-recipients and cortical recipient SC neurons interact to allow switches between
different behavioral states? Freezing is an evolutionarily conserved passive fear response, and
flight is a more active defense response (Fanselow, 1994; Tovote et al., 2016). Previously we
showed that V1 projects to both inhibitory and executory neurons in SC (Zingg et al., 2017). One
45
possibility is that the retina provides a strong signal to the ‘superficial’ lamina, which drives dSC
for flight. It is known that dSC implement a threshold mechanism to gate dPAG responses
(Evans et al., 2018), which directly trigger flight. The cortical input may provide an inhibitory
input to dSC and ‘pull’ the defense response towards freezing. How this intra-SC circuits
implement need further investigation. In addition, different subpopulations in CeA or PAG could
mediate freezing and flight respectively (Tovote et al., 2016; Fadok et al., 2017). The freezing
and flight pathways both descend to the magnocellular nucleus (Tovote, Fadok and Lüthi, 2015;
Ferreira-Pinto et al., 2018). Therefore, the switch between two behaviors to allow proper defense
in respond to threat signal can happen at many levels. How theses downstream targets interact
with each other under different environment or context need further investigation.
To summary, our study demonstrated a differential circuit control of visual defense behaviors in
the SC. The retinal-recipient and cortical-recipient SC neurons are located within different
laminae of SC and drive flight and freezing through dSC and LP respectively (Figure 8). Similar
organizational principles and functional motifs may be generally implemented by the SC to
better adapt to dynamic environment and context changes.
46

Figure 17. A Circuit model for laminar controls of looming-evoked defensive behavior
(A) The retinal visual inputs mainly target zo and upper SuG and relays through dSC to dlPAG
to drive flight responses. The cortical visual input mainly targets lower SuG, op and InG, and
project to LP to mediate freezing responses.

2.4 Material and Methods

2.4.1 Animals, stereotaxic injection and imaging
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 wild-type (C57BL/6) and
transgenic (Vglut2-Cre and Ai14) mice aged 8–16 weeks were obtained from the Jackson
Laboratory. Animal sample sizes were determined by the estimated variances of the experiments
and previous experience from similar experiments and were sufficient for all the statistical testes.
V1
V2
retina
LGn
zo
SuG
Flight
LP
Freezing
dlPAG
A
op
InG
InWh
DpG
47
Mice were housed on 12 h light/dark cycle, with food and water provided ad libitum.
Randomization methods were used to allocate experimental groups.
Viral and reagent injections were carried out as we previously described (Ibrahim et al., 2016;
Zingg et al., 2017). Mice were anesthetized using 4.0% isoflurane–oxygen mixture for
induction 1.5% throughout the surgery procedure. For brain injections, stereotaxic coordinates
were based on the Allen Reference Atlas (www.brain-map.org). A small incision was made on
the skin after shaving to expose the skull. A 0.2 mm craniotomy was made, and the virus was
delivered through a pulled glass micropipette with beveled tip (~15 mm diameter) by pressure
injection. Retina injections was carried out as previously described (Hombrebueno et al., 2014).
A 33-gauge needle (Hamilton Bonaduz AG, Bonaduz, Switzerland) was inserted from the limbus
with a 45° injection angle into the vitreous. The direction and location of the needle was
monitored through a microscope. Virus (800nl-1000nl) was injected using a repeating dispenser
(PB-600-1; Hamilton Bonaduz) at a depth of 0.7 mm.
For anterograde tracing, AAV2/1-hSyn-Cre-WPRE-hGH (UPenn Vector Core, 2.5 ×
10
13
GC/mL) was injected into retina (800 nl total volume), V1 (60 nl total volume; AP 4.0 mm,
ML +3.0 mm, DV 0.5 mm) and V2 (60 nl total volume; AP 4.0 mm, ML +3.5 mm, DV 0.5 mm)
in Ai14 animals. For retrograde tracing, cholera toxin B (CTB-488; Invitrogen) was injected into
dSC (30 nl total volume; AP 4.0 mm, ML +0.5 mm, DV 2.7 mm) and LP (30 nl total volume; AP
2.0 mm, ML +2.7 mm, DV 2.3 mm) in wild-type animals respectively. For two-step tracing,
AAV2/1-hSyn-Cre-WPRE-hGH was injected into V1 (60 nl) or contralateral retina (800 nl).
Following 2–7 days, a second injection of AAV2/1-CAG-FLEX-eGFP-WPRE-bGH (UPenn
vector core, 1.7 × 10
13
GC/mL) was made into the ipsilateral SC (60 nL). The spacing of the two
injections over several days was selected to allow sufficient time for the clearance of any
48
residual AAV-Cre virus that may have spread across the pial surface. Animals were euthanized
3–4 weeks following the injection for examination.
For activation experiments, AAV2/1-pEF1a-hChR2-eYFP (UPenn Vector Core, 1.6 1013 GC/
ml) was injected bilaterally into V1 (60 nl), V2 (60 nl), retina (800 nl) and dSC (60 nl) in wild-
type animals. For inhibition experiments, pAAV-hSyn-DIO-hM4D(Gi)-mCherry (Addgene, 3
1012 VC/ml) was injected bilaterally into dLGN (60 nl total volume; AP 2.0 mm, ML 1.7 mm,
DV 2.3 mm) and LP (60 nl) in Vglut2-Cre animals. For two-step inhibition, AAV2/1-hSyn-Cre-
WPRE-hGH was injected into V1 (60 nl) or retina (800 nl) bilaterally. Following 2–7 days, a
second injection of pAAV-hSyn-DIO-hM4D(Gi)-mCherry was made into the SC (60 nL)
bilaterally in wild-type animals. For cortical silencing studies, muscimol (M23400;
ThermoFisher) was injected bilaterally into P/POR (200 nl total volume; AP 4.0 mm, ML +3.5
mm, DV 0.5 mm) and AM (200 nl total volume; AP 4.0 mm, ML +3.5 mm, DV 0.5 mm) in
wild-type animals. For dlPAG silencing, pAAV-hSyn- hM4D(Gi)-mCherry (Addgene, 3 1012
VC/ml) was injected bilaterally into dlPAG (60 nl total volume; AP 4.0 mm, ML +0.2 mm, DV
2.4 mm). Viruses were expressed for at least three weeks.
Animals were deeply anesthetized and transcardially perfused with phosphate buffered saline
(PBS) followed by 4% paraformaldehyde. Brains were post-fixed at 4 ˚C overnight in 4%
paraformaldehyde and then sliced into 150 mm sections using a vibratome (Leica, VT1000s). To
reveal the cytoarchitectural information, brain slices were first rinsed three times with PBS for
10 min, and then incubated in PBS containing Nissl (Neurotrace 620, ThermoFisher, N21483)
and 0.1% Triton-X100 (SigmaAldrich) for 2 hr. All images were acquired using a confocal
microscope (Olympus FluoView FV1000). To quantify the cell body distributions for different
visual inputs, serial sections across the whole brain were collected. Regions of interest were
49
imaged at 10X magnification across the depth of the tissue (15 mm z-stack interval) were
collected. For each brain, images were taken using identical laser power, gain and offset values.
Quantifications of tdTomato+ expressing soma were performed by ImageJ. The soma counts for
each animal was normalized.

2.4.2 Visual stimulation and in vivo electrophysiology
Visual stimuli were generated with Matlab (MATLAB) with the Psychophysics Toolbox Version
2 (Peirce, 2007) and presented on a ViewSonic VA705b monitor (1920 × 1440 pixels, 33.9 cm
wide, 27.2 cm high, 60 Hz refresh rate, mean luminance 41 cd/m
2
) mounted on a flexible arm.
The monitor was adjusted to be 20 cm away from the contralateral eye, at 45° azimuth, 25°
elevation, and thus subtending 80° azimuth × 70° elevation of mouse’s visual field. The monitor
was gamma-corrected to achieve linear luminance. To measure the overall visual response level,
a set of noise flashes stimuli were presented. Each flash pattern was a checkboard of 20 × 20
white (58 cd/m
2
) and black (24 cd/m
2
) squares (each square 4° × 3°). Each flash was presented
for 200 ms or 1000 ms and 20–50 patterns were presented. To measure orientation and direction
tuning, a set of drifting sinusoidal gratings were presented in a pseudorandom order. The
stimulus set consisted of gratings of 12 directions (0°–330°, 30° per step, 6 orientations). The
spatial frequency of the gratings was 0.04 cycles per second (cpd), and temporal frequency was 2
cycles per second (Hz) based on previous research (Niell and Stryker, 2008). Each grating drifted
for 1.5 s and another grating appeared, which remained to be static for 3 s before moving. To
map fine-scale spatial RFs, sparse noise stimuli, composed of static bright and dark squares (each
square 4° × 4°) was used. The squares were presented individually on a gray screen within a 16 ×
16 grid according to an m-sequence. Each square was displayed for 32 ms (i.e. updated every
other frame) without an inter-stimulus interval between two consecutive squares. The looming
50
stimulus was an expanding black disk presented in the upper visual field. It changed from 2˚ to
20˚ (in diameter) within 250 ms. The largest black disk after the expansion stayed for another
250 ms, and the interstimulus interval was 250 ms. Conversely, the receding stimulus was a disk
receding from 20˚ to 2˚, with all other parameters remained the same as the looming stimulus.
The looming and receding stimuli were usually presented repeatedly for 15 cycles.
Multichannel recordings were carried out by lowering a 64-channel silicone probe (NeuroNexus)
into the targeted region. Signals were recorded by an Open-Ephys system (Open Ephys) at 30
kHz sampling rate. Raw unfiltered traces during stimulus presentation were saved for offline
spike sorting and analysis.
One week before the behavioral tests, animals were prepared as previously described (Xiong et
al., 2015). Briefly, to optogenetically manipulate terminals in SC, mice were implanted with
fiber optic cannulas (200 mm ID, Thorlabs) two weeks after injecting ChR2 virus (Boyden et al.,
2005; Chow et al., 2010). The animal was anesthetized and mounted on a stereotaxic apparatus
(Stoelting co.). Small holes (500 mm diameter) were drilled at a 20-degree angle relative to the
vertical plane above SC (AP 3.8 mm, ML ±0.5 mm, DV 0.5 mm). The cannulas were lowered to
the desired depth and fixed in place using dental cement. In the meantime, a screw for head
fixation was mounted on the top of the skull with dental cement. Light from a blue LED source
(470 nm, 10 mW, Thorlabs) was delivered at a rate of 20 Hz (20 ms pulse duration) via the
implanted cannulas using a bifurcated patch cord (Ø200 mm, 0.22 NA SMA 905, Thorlabs) for
ChR2 or GFP control animals. The plastic sleeve (Thorlabs) securing the patch cord and cannula
was wrapped with black tape to prevent light leakage. Following testing sessions, animals were
euthanized, and the brain was imaged to verify the locations of viral expression and implanted
51
optic fibers. Mice with mistargeted viral injections or misplaced fibers were excluded from data
analysis.
2.4.3 Behavioral tests
All behavioral tests were conducted during the same circadian period. After virus injection or
fiber implantation, the animals were handled daily by the experimenters for at least 3 days before
the behavioral tests. On the day of the behavioral test, the mice were transferred to the testing
room and were habituated to the room conditions for 3 h before the experiments started. The
apparatus was cleaned with 20% ethanol to eliminate odor from other mice.
Looming evoked behavioral tests were conducted in an arena (34 cm × 34 cm square open field)
with regular mouse bedding, with or without a hiding nest. A display monitor was positioned
above the arena for presentation of overhead looming visual stimuli. Cameras mounted on top
and side of the arena to record the animals’ behaviors. The animals were allowed 10 min of
habituation prior to visual stimuli. The stimulus was triggered by the experimenter when the
animal entered a black square in the middle of the arena. The looming stimuli were presented as
an expanding black disc appeared directly above the animal on a grey background from 2
degrees to 20 degrees visual angle in 250 ms and remained at that size for 250 ms. The stimulus
was presented 10 times with 500 ms interval.
The video recordings were analyzed with a custom-written python program. The velocity was
calculated and smoothed with a median filter. Freezing is defined as the episodes of 1 s or more
where the velocity is less than 10% of its average value over the 10 s interval prior to the
stimulus onset. Flight is defined as episodes where the velocity is greater than 3 times the
average and the animal’s final position is in the nest. For DREADDi experiments, animals
52
infected with AAV-hM4Di(Gi)-mCherry (Zhu and Roth, 2014) received an intraperitoneal (IP)
injection of clozapine-N-oxide (CNO) (1 mg/kg) or saline 45 minutes prior to the behavioral
session. For silencing experiments using muscimol, animals were infused with muscimol 30
minutes prior to the behavior session through an implanted canula. Each animal was tested for
consecutive 2 days.
For optogenetic experiments, animals were tested in an arena (34 cm × 34 cm square open field).
No hiding nests were provided. Animals were allowed 10 min to acclimate to the arena. A 10-s
blue LED (20 Hz) was triggered when the animal entered the center zone. Total 5 trials were
given for each animal, with 5 min intervals. Their behaviors were recorded through cameras
mounted on top and side of the arena. Videos were analyzed using a custom-made python
program.

2.4.4 Data analysis and statistics
Data analysis was performed using customized scripts written in Matlab (MATLAB). Raw
waveforms were saved online, and spikes were detected and sorted offline.
Overall response level. To quantify the overall response level of SC neurons, the number of
spikes evoked by noise flashes and looming/receding stimuli was counted. The spike number
was averaged across trials to derive the peri-stimulus time histogram (PSTH). The spontaneous
firing rate was calculated from the time window 500 ms before the stimulus onset and then
subtracted from the evoked firing rate.
Orientation and direction tuning. To derive tuning curves, the spike number evoked by drifting
gratings was counted within the 70–1570 ms window after stimulus onset. The responses were
organized according to the direction of moving stimuli and averaged across repetitions.
53
Spontaneous firing rate calculated within a 500 ms window before the stimulus onset was
subtracted. Orientation tuning curves were obtained by averaging responses to two opposite
directions (i.e. 0° and 180°, 30° and 210°, etc). The orientation selectivity index (OSI) was
calculated as:
𝑂𝑆𝐼 =
𝑅
!"#$
−𝑅
%"&'
𝑅
!"#$
+𝑅
%"&'
,
where 𝑅
!"#$
and 𝑅
%"&'
are the responses to the preferred and the orthogonal orientation (90°
from the preferred orientation), respectively. Similarly, the direction selectivity index (DSI) was
quantified as:
𝐷𝑆𝐼 =
𝑅
!"#$
−𝑅
()**
𝑅
!"#$
+𝑅
()**
,
where 𝑅
!"#$
and 𝑅
()**
are the responses to the preferred and the null direction (180° from the
preferred direction), respectively.
Spatiotemporal receptive field (STRF). To map the fine-scale structure of RF, the spike train
evoked by the sparse noise stimuli was reversely correlated with the stimulus sequence to derive
the spike-triggered average response of the stimuli (Jones and Palmer, 1987). A 2D standard
deviation was calculated at each time lag from 0–200 ms as the strength of RFs. The RF showing
strongest response was used to determine the size of RF. The background response level was
determined by averaging the spike numbers at the 4 borders of RF and subtracted from RF. The
RF map was further divided by its 2D standard deviation to obtain Z scores. Pixel value with a Z
score < 4 was set to be 0. The area of connected pixels was calculated and transformed to a circle
with the same area. The radius of the circle was defined as the RF size.
54
Spike sorting. The raw signals from the 64-channel silicon probe were filtered through a 300-
6000 Hz band-pass filter. The spatially varying motion artifacts were removed by applying a
local common average referencing (L-CAR) scheme. The nearby four channels of the silicon
probe were grouped as tetrodes, and semi-automatic spike detection and sorting was performed
using the Plexon offline sorter (Dallas, Texas). Clusters with isolation distance > 20 was
considered as separate clusters. Spike clusters would be classified as single units only if the
waveform SNR (Signal Noise Ratio) exceeded 4 (12 dB) and the inter-spike interval was longer
than 1.2 ms for more than 99.5% of the spikes.
Shapiro–Wilk test was first applied to examine whether samples had a normal distribution. In the
case of a normal distribution, two-tailed t-test was applied. Otherwise, a two-tailed non-
parametric test (Wilcoxon signed-rank test or Mann-Whitney U test in this study) was applied.
Statistical analysis was conducted using SPSS (IBM) and Prism (Graphpad).
All data and custom written code supporting the findings of this study are available upon
reasonable request from the corresponding authors.

55
Chapter 3: A Cross-modality Enhancement of Defensive Flight via
Parvalbumin Neurons in Zonal Incerta

3.1 Introduction
Defensive behaviors are critical for animal survival. They are dynamic and adaptive, as
environmental contexts, properties and intensity of threats, as well as expectations from past
experiences can all modulate the form as well as the magnitude of defensive behaviors
(Fanselow, 1994; Gross and Canteras, 2012; Tovote et al., 2016). Threat signals in the external
environment are detected by different sensory modalities through distinct sensory pathways to
initiate appropriate defense behaviors. Previous studies have mostly been focused on defense
behaviors evoked by an individual sensory modality (Fanselow and Ledoux, 1999; Yilmaz and
Meister, 2013; Xiong et al., 2015). However, a danger signal may be associated with cues of
multiple sensory modalities arriving at the same time, and the integration of information of these
different modalities may profoundly influence the behavioral output. Intuitively, the presence of
multisensory signals is helpful for strengthening defense responses. However, neural circuit
bases for the potential cross-modality interactions in defense behaviors are largely unknown. In
the present study, we designed experiments to specifically examine whether tactile input can
affect a well-established auditory-induced defensive behavior (Fanselow and Ledoux, 1999;
Tovote et al., 2016). The vibrissal system is crucial to behaviors such as navigation and
exploration (Carvell and Simons, 1990; Diamond et al., 2008), and rodents frequently collect
information from surroundings using their whiskers (Prigg et al., 2002). We reason that it may be
common for animals to use both tactile and auditory systems in sensing environmental dangers
and navigate to the safe zone.
56
Zona incerta (ZI) is a major GABAergic subthalamic nucleus consisting of heterogeneous groups
of cells. In rodents, four (rostral, ventral, dorsal, caudal) sectors of ZI can be loosely defined
based on the cytoarchitecture and neurochemical expression patterns (Ma, Johnson and Hoskins,
1997; Kolmac and Mitrofanis, 1999; Mitrofanis et al., 2004), and it has been suggested that
different sectors might be involved in different circuits and functions (Plaha, Khan and Gill,
2008; Liu et al., 2017). For instance, our recent study has shown that GABAergic neurons in the
rostral sector of ZI (ZIr) play a role in reducing defensive behavior in an experience-dependent
manner (Chou et al., 2018). Previous anatomical studies have shown that ZI receives dense
inputs from structures involved in different sensory pathways (Roger and Cadusseau, 1985;
Mitrofanis, 2005). The multisensory convergence may provide ZI with versatile response
characteristics. Since ZI projects strongly to many hypothalamic, higher-order thalamic and
brainstem regions considered as behavioral commanding centers, is it possible that ZI outputs
play a broader role in modulating defensive behavior in complex sensory environment. ZI
receives inputs from various cortical areas including the primary somatosensory cortex (SSp)
(Shammah-Lagnado et al., 1985; Kolmac, Power and Mitrofanis, 1998) as well as from the
brainstem trigeminal nucleus that relays vibrissal information (Smith, 1973; Roger and
Cadusseau, 1985). In addition, a recent study has demonstrated that deflecting whiskers directly
induces neuronal activity in the ventral sector of ZI (ZIv)(Urbain and Deschênes, 2007), where
parvalbumin (PV) positive neurons are enriched(Kolmac and Mitrofanis, 1999). In the present
study, we investigated whether somatosensory input through whisker stimulation could modulate
auditory-induced flight responses via recruiting ZIv PV+ neurons.
57
3.2 Results
3.2.1 Tactile stimulation enhances sound-induced flight response via SSp
To test whether tactile input can affect defensive behavior, we employed a simple behavioral
test, sound-induced flight, following our previous studies (Xiong et al., 2015; Zingg et al.,
2017a). Such innate behavior has been observed in both freely moving and head-fixed conditions
(Xiong et al., 2015; Zingg et al., 2017a). In our first set of experiments, animals were head-fixed
and placed on a smoothly rotatable plate (Liang et al., 2015; Chou et al., 2018). Loud noise
sound (80 dB sound pressure level or SPL) would trigger robust running, and the running speed
was recorded in real time (Figure 6A, left). Tactile stimulation was applied by deflecting
whiskers unilaterally with a cotton stick controlled by a piezo actuator. In our control
experiments, the whisker deflection per se did not elicit significant locomotion of animals. Trials
without and with tactile stimulation were interleaved. We found that tactile stimulation enhanced
the running induced by noise sound (Figure 18A, right), as demonstrated by the increased peak
speed (Figure 18B) and total travel distance (Figure 18C). The temporal protable of the
behavioral response was not significantly affected, as shown by the quantifications of onset
latency and time to peak (Supplemental table 1). Silencing the SSp contralateral to the whiskers
being deflected by infusing a GABA receptor agonist, muscimol (Figure 18D, left), removed the
difference in speed between conditions without and with whisker stimulation (Figure 18D-F),
without altering the response temporal protable (Supplemental table 1). This suggests that the
tactile enhancement of running is mediated mainly through SSp. To further demonstrate the
tactile effect on flight behavior in freely moving animals, we used a two-chamber test following
our previous study. When the mouse was exposed to noise applied in one chamber, it quickly
escaped to the other chamber by crossing through a narrow channel (Figure 19A). Trimming of
58
all whiskers of the animal significantly decreased the average speed of the flight through the
channel (Figure 19B), suggesting that tactile information through whiskers can indeed enhance
flight behavior in a more natural condition.

Figure 18. Tactile stimulation enhances sound-induced flight response via SSp
(A) Left, illustration of the head-fixed animal behavioral paradigm. Right, plots of running speed
under noise presentation without (black) and with (red) concurrent whisker stimulation for an
example animal. Red line marks the duration of noise/whisker stimulation. (B) Summary of
peak noise-induced running speed in the absence and presence of whisker stimulation. **p =
0.0011, two-sided paired t-test, n = 7 animals. (C) Summary of total travel distance. **p =
0.0072, two-sided paired t-test, n = 7 animals. (D) Left, illustration of the experimental
paradigm: SSp was silenced with infusion of muscimol (red) as shown in the confocal image
(upper left, scale: 500 µm). Right, plots of speed without (black) and with (red) whisker
stimulation for an example animal. (E) Summary of peak speed in the absence and presence of
whisker stimulation. n.s., not significant, two-sided paired t-test, n = 5 animals. (F) Summary of
total travel distance. n.s., not significant, two-sided paired t-test, n = 5 animals.

SSp
Normal Trimmed

59
Figure 19. Flight speed decreased without whiskers in two-chamber test
(A) Illustration of the two-chamber flight test. (B) Summary of average speed for flight crossing
the channel before and after whisker trimming. Data points for the same animal are connected
with a line. **p < 0.01, two-sided paired t-test, n = 7 animals. Error bars represent mean ± s.d.

3.2.2 The SSp-ZIv projection mediates the tactile enhancement of sound-induced flight
Previous studies have suggested that SSp projects to ZIv (Shammah-Lagnado et al., 1985;
Kolmac, Power and Mitrofanis, 1998), and that ZIv neurons respond to whisker deflections
(Urbain and Deschênes, 2007). To confirm this projection, we injected AAV1-CamKII-hChR2-
eYFP into SSp of PV-ires-Cre crossed with Ai14 (Cre-dependent tdTomato) reporter mice
(Figure 20A), in which PV+ neurons could be labeled with tdTomato. We found profuse
fluorescence-labeled axons in ZIv, but few in other ZI sectors (Figure 20B), which is consistent
with previous findings. We next directly examined the effect of stimulating the SSp projections
to ZIv, by placing optic fibers over the top of ZIv to deliver LED light pulses (20 Hz train for 5
sec) bilaterally (Figure 20C). The optogenetic activation of the SSp axons in ZIv enhanced noise-
induced running (Figure 20C-E) without affecting the response temporal profile (Supplement
Table 1), but by itself had no effect on the baseline locomotion speed. In addition, infusing a
GABA agonist, muscimol into ZIv bilaterally abolished the enhancement of flight response by
Ave. Speed (cm/s)
0
15
30
Normal Trimmed
**
50 cm
camera
A B
Figure1 - figure supplement 2
60
Figure 20. The SSp-ZIv projection mediates the tactile enhancement of sound-induced flight
(A) Illustration of the injection paradigm. (B) Anterogradely labeled axon terminals in rostral
(left), dorsal and ventral (middle), as well as caudal (right) sectors of ZI. Scale bar, 200 µm. Blue
AAV1-CamKII-hChR2-eYFP
PV-ires-Cre::Ai14
SSp
ZI
A B
Speed (cm/s)
0
10
20
5 10 15
Time (s)
20
OFF ON
Peak Speed (cm/s)
0
15
30
Travel Dist. (cm)
OFF ON
0
50
100
** *
C D E
AAV1-CamKII-hChR2-eYFP
470 nm
wild type
SSp
ZI
LED ON
LED OFF
ZIr
ZIc
5 10 15 20
Time (s)
Speed (cm/s)
0
10
20
contr whisk
Peak Speed (cm/s)
0
15
30
n.s.
Travel Dist. (cm)
contr whisk
0
25
75
n.s.
50
muscimol
ZI
whisk
contr
F G H
Figure 2
ZId
ZIv
PV-ires-Cre::Ai14
AAV1-CamKII-hChR2-eYFP
+ TTX & 4-AP
+ CNQX
8/10)
0
200
400
EPSC Amplitude (pA)
I J K
61
shows Nissl staining; red shows PV neuron distribution. (C) Left, illustration of the experimental
paradigm: optic fibers were implanted to stimulate ChR2-expressing SSp axons in ZI. Right,
plots of speed without (black) and with (blue) LED stimulation for an example animal. (D)
Summary of peak noise-induced speed in the absence and presence of LED stimulation of SSp-
ZI terminals. **p = 0.003195, two-sided paired t-test, n = 7 animals. (E) Summary of the travel
distance. *p = 0.01854, two-sided paired t-test, n = 7 animals. (F) Left, ZI was silenced with
muscimol (red) as shown in the confocal image (lower, scale: 500 µm). Right, plots of speed
without (black) and with (red) whisker stimulation for an example animal. (G) Summary of peak
speed in the absence and presence of whisker stimulation. n.s., not significant, two-sided paired
t-test, n = 5 animals. (H) Summary of total travel distance. n.s., not significant, two-sided paired
t-test, n = 5 animals. (I) Experimental paradigm for slice recording. (J) Average LED-evoked
EPSC in an example ZIv+ PV neuron before and after (lower) perfusion of CNQX. Arrow points
to the onset of LED light. Recording was made in the presence of TTX and 4-AP. Scale: 25 pA,
25 ms. (K) Amplitudes of LED-evoked EPSCs of 8 responding neurons out of 10 recorded ZIv
PV+ cells. Bars represent s.d..

whisker stimulation (Figure 20F-H) without affecting the response temporal profile
(Supplemental table 1). Together, these results suggest that activation of the SSp-ZIv projection
is sufficient and necessary for the tactile enhancement of auditory-induced flight response.
Immuno-histological studies have suggested that PV+ neurons are a major cell type in the ventral
sector of ZI (Kolmac and Mitrofanis, 1999). To test whether SSp axons innervate PV+ neurons,
we performed slice whole-cell recording from ZIv PV+ neurons (labeled by tdTomato expression
in PV-Cre::Ai14 animals) while optically activating ChR2-expressing SSp axons in ZI (Figure
20I). We observed that blue light pulses evoked monosynaptic excitatory postsynaptic currents
(EPSCs) in most ZIv PV+ neurons recorded with tetrodotoxin (TTX) and 4-aminopyridine (4-
AP) present in the bath solution. The EPSC could be blocked by an AMPA receptor blocker, 6-
62
cyano-7-nitroquinoxaline-2,3-dione (CNQX) (Figure 7J-K). These results indicate that ZIv PV+
neurons receive direct excitatory input from SSp.
Figure 21. Optogenetic stimulations could effectively activate/suppress labeled cells
(A) Confocal images showing the expression of ChR2 (green) and tdTomato (red) in ZIv in a
PV-Cre animal. White arrows point to cells showing colocalization of ChR2 and tdTomato.
Scale: 50 µm. (B) Upper, blue light induced spiking in an example ChR2-expressing neuron.
Each blue dot represents a blue light pulse. Scale: 20 mV, 100 ms. Lower, summary of
probability of spiking in response to 10 pulses (at 20 Hz) of blue light stimulation (n= 5
neurons). (C) Upper, green light (500 ms duration) induced hyperpolarization in an example
ArchT-expressing neuron. Lower, summary of maximum level of hyperpolarization for 4
recorded neurons.

3.2.3 PV+ neurons in ZIv mediate the tactile enhancement of flight behavior
To investigate whether PV+ neurons play a role in the tactile modulation of flight response, we
injected AAV encoding Cre-dependent ChR2 or ArchT into ZI of PV-Cre::Ai14 mice (Figure
8A, D). The viral expression of opsins co-localized well with Cre-dependent tdTomato
Figure 3 - figure supplement 1
Norm. Peak Speed
0
1.0
1 2 3
4 5 6
Animal Number
0.5
*
**
*
* *
Norm. Travel Dist.
0
2
1 2 3 4 5 6
Animal Number
1
**
*
*
* *
OFF
ON
OFF
ON
ChR2 Ai14 Merge
A
F G
D E
1 3 4 5
Animal Number
2
0
1.0
0.5
1 3 4 5
Animal Number
2
Norm. Peak Speed
0
1.0
0.5
** *
*
*
*
*
*
Norm. Travel Dist.
ChR2
ArchT
0 50 100
Probability (%)
0 -20 -40
Hyperpolarization (mV)
-80
-70
-60
Vm (mV)
B C
63
Figure 22. PV+ neurons in ZIv mediate the tactile enhancement of flight behavior
(A) Left, experimental paradigm. Right, Plots of speed without (black) and with (blue) LED
stimulation for an example animal. Blue line marks the duration of noise/LED stimulation. (B)
Summary of peak noise-induced speed in the absence and presence of LED stimulation of ZIv
PV+ neurons. ***p = 0.0009, two-sided paired t-test, n = 6 animals. (C) Summary of total travel
distance. **p = 0.0042, two-sided paired t-test, n = 6 animals. (D) Left, experimental paradigm.
64
Right, plots of speed without (black) and with (green) LED stimulation for an example animal.
Green line marks the duration of noise/LED stimulation. (E) Summary of peak noise-induced
speed in the absence and presence of LED inhibition. ***p = 0.0004, two-sided paired t-test, n =
5 animals. (F) Summary of total travel distance. *p = 0.0136, two-sided paired t-test, n = 5
animals. (G) Left, expressing DREADDi in ZIv PV+ neurons. Right, plots of speed without
(black) and with (red) whisker stimulation for an example animal. (H) Summary of peak noise-
induced speed in the absence and presence of whisker stimulation with ZIv PV+ neurons
inhibited by CNO. n.s., not significant, two-sided paired t-test, n = 8 animals. (I) Summary of
total travel distance. n.s., not significant, two-sided paired t-test, n = 8 animals. Open symbols
represent mean ± s.d. (J) Upper, optrode recording in the head-fixed animal. Lower, raster plot
of an example ZIv PV+ neuron to 20-Hz LED stimulation in 7 trials. Scale: 50 ms. (K) Peri-
stimulus spike time histogram for an example PV+ neuron in response to whisker (red), noise
(yellow) and whisker plus noise (black). Bin size = 100 ms. (L) Summary of evoked firing rates
of recorded PV+ neurons (within the stimulation window). ***p < 0.0001, one-way ANOVA
with post hoc test, n = 22 cells.

expression, indicating PV-specific expression of opsins. Optogenetic manipulation of ZI PV+
neuron activity with blue (for the ChR2 group to activate) or green (for the ArchT group to
suppress) LED light was interleaved with control trials in which only sound was delivered. The
efficacies of ChR2 and ArchT were confirmed by slice whole-cell recordings showing that blue
LED evoked robust spiking in ChR2-expressing neurons and green LED induced a strong
hyperpolarization of the membrane potential in ArchT-expressing cells (Figure 21). We found
that activation of ZI PV+ neurons enhanced flight response induced by noise stimulation (Figure
22A-C), whereas suppression of these neurons reduced the flight response (Figure 22D-F). None
of the manipulations affected the temporal protable of the behavioral response (Supplemental
table 1). As a control, neither activation nor suppression of ZIv PV+ neurons alone significantly
affected the baseline locomotion (Figure 23). We next expressed Cre-dependent inhibitory
designer receptors exclusively-activated by designer drugs (DREADDi) (Zhu and Roth, 2014),
hM4D(Gi), in ZI of PV-Cre mice, and intraperitoneally injected the DREADDi agonist,
clozapine-N-oxide (CNO) (Figure 22G). The efficacy of DREADDi inhibition was confirmed by
65
slice recording showing that CNO increased the threshold for spiking and decreased the number
of spikes evoked by current injections (Figure 24). The chemogenetic silencing of ZIv PV+

Figure 23. Optogenetic stimulation of ZI did not alter baseline speed
(A-B) Summary of average speed without and with LED simulation alone for the ChR2 (C, n = 5
animals) and ArchT (D, n = 5 animals) group. n.s., non-significant, two-sided paired t-test. Error
bars represent mean ± s.d.

neurons prevented the enhancement of noise-induced flight response by whisker stimulation
(Figure 22G-I) without affecting the response temporal protable (Supplemental table 1).
We next performed awake single-unit optrode recordings in ZI, following our previous study
(Zhang et al., 2018). ZIv PV+ neurons were optogenetically identified by their time-locked spike
responses to blue laser pulses (Figure 22J). We found that these neurons responded to both noise
sound and whisker deflections and that concurrent whisker deflections increased the response
level to noise (Figure 8K-L). This result indicates that ZIv PV+ neurons can integrate tactile and
auditory inputs and that tactile input plays a faciliatory role, consistent with the behavioral
results. Altogether, our results strongly suggest that the tactile enhancement of flight behavior is
mediated primarily by ZIv PV+ neurons.

n.s.
Ave. Speed (cm/s)
OFF ON
0
5
10
A
ZI PV ChR2
n.s.
Ave. Speed (cm/s)
OFF ON
0
5
10
B
ZI PV ArchT
Figure 3 - figure supplement 2
66

Figure 24. DREADDi manipulation was effective
(A) Current-clamp recording traces for an example hM4D(Gi)-expressing neuron in response to
a series of current injections (500 ms duration) with amplitude ranging from 0 to 120 pA with a
step of 20 pA, before (left) and after (middle) CNO infusion and after washing out CNO (right).
Scale: 20 mV. Red color labels the trace at the maximum level of current injection. (B)
Summary of minimum amplitude of current injection needed to induce spiking before and after
CNO infusion and after washing out CNO. (C) Summary of number of spikes induced by
current injection at the same level (which was the threshold level after CNO infusion) before and
after CNO infusion (n = 6 neurons).

3.2.4 The projection of ZIv PV+ neurons to POm enhances sound-induced flight
To identify which downstream target nuclei of ZIv PV+ neurons is involved in the behavioral
modulation, we traced the projections from ZIv PV+ neurons by injecting AAV encoding Cre-
dependent GFP in PV-Cre mice (Figure 25A). Consistent with previous results (Barthó, Freund
and Acsády, 2002; Trageser, 2004), we found two targets, the medial posterior complex of
thalamus (POm) and the superior colliculus (SC), received the strongest projections from ZIv
PV+ neurons (Figure 26). Both structures were reported to contribute to sensorimotor
Figure 3 - figure supplement 3
before after
wash
Minimal injected
current to induce AP (pA)
150
100
50
0
before after
Spike #
40
20
0
A
B C
CNO CNO
before CNO wash out
-70
-60
-50
-40
-30
-20
-10
0
10
20
-70
-60
-50
-40
-30
-20
-10
0
10
20
-70
-60
-50
-40
-30
-20
-10
0
10
20
67
Figure 25. The projection of ZIv PV+ neurons to POm enhances sound-induced flight
(A) Illustration of injection paradigm. (B) Confocal images showing GFP expression at the
injection site (left; scale: 500 µm) and in major target regions (middle and right; scale: 200 µm).
Blue shows Nissl staining. SC, superior colliculus; POm, posterior medial nucleus of thalamus.
(C) Left, stimulating ChR2-expressing ZI PV+ neuron axons in SC. Right, plots of speed without
(black) and with (blue) LED stimulation for an example animal. (D) Summary of peak noise-
A
H G F
E D C
AAV1-Flex-GFP-WPRE
SC
SC
ZI
ZI
ZI POm
POm
470 nm
PV-ires-Cre
PV-ires-Cre
AAV1-DIO-hChR2-eYFP
n.s. n.s.
** *
Speed (cm/s)
0
10
20
30
5 10 15 20
Time (s)
Speed (cm/s)
0
5 10 15 20
Time (s)
15
30
OFF ON OFF ON
OFF ON OFF ON
Travel Dist. (cm)
Peak Speed (cm/s)
0
20
40
40
80
120
0
Peak Speed (cm/s)
0
30
60
Travel Dist. (cm)
0
75
150
ZI SC POm
AAV1-DIO-hChR2-eYFP
LED ON
LED OFF
LED ON
LED OFF
PV-ires-Cre
470 nm
B
Figure 4
PV-ires-Cre
CNO
AAV-hM4D(Gi)-mCherry
I
ZI
POm
Speed (cm/s)
0
15
30
5 10 15 20
Time (s)
whisk
contr
contr whisk
Peak Speed (cm/s)
0
20
40
n.s.
J
Travel Dist. (cm)
75
150
0
n.s.
K
contr whisk
68
induced speed in the absence and presence of LED inhibition of ZIv-SC axons. n.s., not
significant, two-sided paired t-test, n = 5 animals. (E) Summary of total travel distance. Two-
sided paired t-test, n = 5 animals. (F) Left, stimulating ChR2-expressing ZI PV+ neuron axons in
POm. Right, plots of speed without (black) and with (blue) LED stimulation for an example
animal. (G) Summary of peak noise-induced speed in the absence and presence of LED
inhibition of ZI-POm axons. *p = 0.0198, two-sided paired t-test, n = 5 animals. (H) Summary of
total travel distance. **p = 0.0034, two-sided paired t-test, n = 5 animals. (I) Left, silencing
DREADDi-expressing ZI PV+ neuron axons in POm. Right, plots of speed without (black) and
with (red) whisker stimulation after local infusion of CNO for an example animal. (J) Summary
of peak noise-induced speed in the absence and presence of whisker stimulation when silencing
ZIv-POm axons. n.s., not significant, two-sided paired t-test, n = 7 animals. (K) Summary of
total travel distance. n.s., not significant, two-sided paired t-test, n = 7 animals. Open symbols
represent mean ± s.d. for all panels.

transformation (Gandhi and Katnani, 2011; Watson, Smith and Alloway, 2015; Cang et al.,
2018). We then specifically activated the ZIv PV+ projections to POm or SC by bilaterally
placing optic fibers on top of POm or SC, respectively, in PV-Cre mice injected with AAV
encoding Cre-dependent ChR2 in ZI (Figure 25C, F). To our surprise, activation of the ZIv-SC
projection did not significantly change the flight speed (Figure 25C-E), but activation of the ZIv-
POm projection significantly increased the flight speed (Figure 25F-H), similar to the results of
activation of ZIv PV+ neuron cell bodies. As a control, activation of the ZIv-POm projections
alone did not change the baseline locomotion speed.
To confirm that the ZIv-POm projection is indeed necessary for the tactile modulation, we then
expressed Cre-dependent hM4D(Gi) in ZI of PV-Cre mice and locally infused CNO into POm
through implanted cannulas (Figure 25I). The chemogenetic silencing of the ZIv-POm projection
prevented the enhancement of flight speed by whisker stimulation (Figure 25I-K). It is important
to mention that none of the manipulations affected the latency of flight response. Taken together,
our results demonstrate that the projection of ZIv PV+ neurons to POm primarily mediates the
enhancement of sound-induced flight behavior by tactile stimulation.
69

Figure 26. Mapping of axonal outputs for ZIv PV+ neurons
Quantification of relative fluorescence density of GFP-labeled processes in different downstream
regions of ZIv PV+ neurons (n = 4 animals). Bar = s.d. Abbreviations: vlPAG, ventrolateral
periaqueductal gray; dlPAG, dorsolateral periaqueductal gray; IC, inferior colliculus; MRN,
midbrain reticular nucleus; RN, red nucleus; APN, anterior pretectal nucleus; PRN, pontine
reticular nucleus; M1, primary motor cortex; S1, primary somatosensory cortex; V1, primary
visual cortex; A1, primary auditory cortex; LA, lateral amygdalar nucleus; BLA, basolateral
amygdalar nucleus; CEA, central amygdalar nucleus.

3.3 Discussion
In this study, we demonstrate that additional tactile stimulation enhances flight behavior
triggered by aversive auditory stimulus – loud noise. Both SSp and its downstream, ZIv PV+
neurons, are necessary for this modulation, and activation of the SSp-ZIv projection is sufficient
for driving the enhancement. We also demonstrate that activation of ZIv PV+ neurons alone can
enhance the flight behavior and that inactivation of the PV+ neurons or their projections to POm
blocks the tactile enhancement of the flight behavior. Together, our data suggest that tactile input
through whiskers can modulate defensive flight via the SSp-ZIv-POm pathway.
Rodents frequently use their whiskers to locate and identify objects (O’Connor et al., 2013;
Pammer et al., 2013). In complex sensory environments, whiskers are essential for them to
vlPAG dlPAG POm SC MRN M1 RN PRN APN
0
8
16
24
% of Total Fluorescence
IC S1 V1 A1 CEA BLA LA
Figure 4 - figure supplement 1
70
gather information from surroundings as to guide their behaviors during exploration and
navigation (Ahl, 1986; Diamond et al., 2008; Sofroniew et al., 2014; Yu et al., 2016). When
escape behavior is concerned, specific somatosensory input plus loud noise may indicate the
immediate proximity of a predator, and enhancement of flight at such moments will greatly
increase survival chances of prey animals. In addition, information conveyed by the
somatosensory system about the environment could be extremely useful for the prey animal to
quickly choose and navigate through an effective escape route (Diamond et al., 2008; Douglass
et al., 2008). Therefore, the ability to integrate somatosensory input and modulate defensive
flight behavior accordingly is beneficial for animals to avoid dangers. Here, we show that
somatosensory input from whiskers can enhance auditory-induced flight response. Indeed, in
freely moving mice, trimming of whiskers reduces the efficiency of their escape from a source of
loud noise by crossing through a channel, indicating a faciliatory role of the tactile input.
The tactile-auditory cross-modality modulation relies on conveying somatosensory information
primarily from SSp to ZI. ZI has been implicated in normal posture and locomotor functions
(Edwards and Isaacs, 1991), as it sends dense projections to motor-related thalamic and
brainstem nuclei (Kolmac, Power and Mitrofanis, 1998; Shaw and Mitrofanis, 2002). The
somatosensory input to ZI thus has a potential to influence motor functions (Supko, Uretsky and
Wallace, 1991; Perier et al., 2002). In this study, we showed that SSp projections to ZI are
mainly concentrated in ZIv, where PV+ neurons are a major cell type (Mitrofanis, 2005; Zhou et
al., 2018b). Consistent with the anatomical data, PV+ neurons in ZIv receive direct excitatory
input from SSp and respond to whisker deflections. Concurrent whisker deflections also increase
their responses to sound, indicating that the tactile-auditory integration takes place in ZIv PV+
neurons. Activating SSp-ZIv axon terminals or ZIv PV+ neurons directly enhance auditory-
71
induced flight, while silencing the PV+ neurons abolishes the enhancement of flight by tactile
stimulation. Therefore, our data demonstrate that via the SSp-ZIv pathway mediated mainly by
ZIv PV+ neurons, somatosensory input can modulate motor functions in defensive behavior.
Whether ZIv PV+ neurons are involved specifically in tactile-auditory integration or
multisensory integration in general remains to be further investigated.
Different ZI sectors are dominated by distinct cell types (Ricardo, 1981; Mitrofanis, 2005). It has
been suggested that different ZI cell types or sectors may exhibit different connectivity patterns
(Mitrofanis, 2005), contributing to ZI’s multiplex roles in various physiological functions. For
example, it has been shown that activation of GABAergic neurons in the rostral sector of ZI
(ZIr), which are mostly PV-negative, can induce binge-like eating via its projections to the
periventricular nucleus of thalamus (Zhang and van den Pol, 2017), while Lhx6-expressing
neurons in ZIv, which are also PV-negative, can regulate sleep through their projections to
hypothalamic areas(Liu et al., 2017). Different sectors or cell types may also play different roles
in defensive behavior. Indeed, we have previously shown that activation of ZIr GABAergic
neurons reduces noise-induced flight via their projections to the periaqueductal gray
(PAG)(Chou et al., 2018). This effect is opposite to that of activating ZIv PV+ neurons, which
have few projections to PAG. More recently, using conditioned freezing response as a model, a
study of ZIv PV+ neurons has shown that both silencing the PV+ neuron output and silencing the
amygdala inhibitory input to the PV+ neurons disrupt fear memory acquisition as well as recall
of remote fear memory (Zhou et al., 2018b). In the current study, the behavior we examined is an
innate defensive behavior. Therefore, ZIv PV+ neurons can play a role in both innate and learned
defensive behavior, which are generated under different contexts and likely engage different
upstream pathways. It would be interesting to investigate in the future how ZI, through
72
interactions among its different cell-types/subdivisions, regulates behaviors in complex sensory
and behavioral environments.
We have identified POm as a target of ZIv PV+ neurons that is mainly responsible for the tactile
enhancement of flight behavior. Silencing of the projection from ZI PV+ neurons to POm
prevents the faciliatory effect of tactile stimulation. PV+ neurons in ZI are GABAergic(Barthó,
Freund and Acsády, 2002) and provide inhibition to their target neurons. To achieve the effect of
enhancing the behavioral output, disinhibitory circuits may be involved. POm is known to
project to the striatum to modulate locomotion (Ohno et al., 2012; Smith, Mowery and Alloway,
2012). The inhibitory nature of striatal neurons makes them a good candidate for engaging
disinhibition of distant output responses (Grillner et al., 2005; Kreitzer and Malenka, 2008).
Furthermore, we have shown previously that the noise-induced flight behavior depends on a
pathway from the auditory cortex (AC) to the cortex of inferior colliculus (ICx) and then to PAG
(Xiong et al., 2015). It is possible that the ZIv-POm pathway directly or indirectly connect to
midbrain areas downstream of the AC-ICx-PAG pathway (Marchand and Hagino, 1983;
Roseberry et al., 2016). As such, somatosensory information carried by the ZIv-POm pathway
can modulate the auditory-induced behavior mediated by the AC-ICx-PAG pathway. It would be
interesting to investigate in the future whether and how the POm-striatal circuit is involved in
this modulation.
Overall, ZI has complex input and output connectivity patterns(Roger and Cadusseau, 1985;
Shammah-Lagnado et al., 1985; Nicolelis, Chapin and Lin, 1992; Chou et al., 2018; Zhou et al.,
2018b). Via convergent and divergent connectivity with various brain areas, ZI may be able to
carry out multiple physiological and behavioral functions synergistically.

73

Table 1. Analysis of temporal profiles of speed traces in different sets of experiments
Data are presented as mean ± SD. Two-sided paired t-test were performed to compared values between
control and manipulation conditions. The type of experiment is shown by the corresponding figure
number in main figures.

Response Latency Time to Peak
Control (s) Manipulation
(s)
p-value Control (s) Manipulation
(s)
p-value
Figure 18A-C 0.171±0.049 0.157±0.054 0.604 1.843±0.730 1.814±0.876 0.937
Figure 18D-F 0.400±0.255 0.520±0.683 0.569 1.940±0.428 2.040±0.230 0.528
Figure 20C-E 0.544±0.341 0.600±0.386 0.955 1.900±0.753 2.481±0.964 0.129
Figure 20F-H 0.720±0.192 0.560±0.270 0.327 1.700±0.579 1.920±0.482 0.585
Figure 22A-C 0.443±0.237 0.491±0.221 0.066 2.514±1.320 2.757±1.655 0.716
Figure 22D-F 0.700±0.430 0.740±0.365 0.597 2.940±1.161 2.800±0.656 0.786
Figure 22G-I 0.675±0.362 0.550±0.359 0.499 1.738±0.537 1.388±0.564 0.224
Figure 22C-E 0.800±0.430 0.840±0.261 0.862 2.260±0.796 2.400±0.946 0.505
Figure 25F-H 0.400±0.187 0.700±0.367 0.259 3.760±1.240 2.860±0.602 0.061
Figure 25I-K 0.500±0.141

0.5200±0.239 0.799 2.500±0.593 2.2830±0.568 0.734
74
3.4 Material and methods
2.4.1 Animals, viral and reagent injections
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 wild-type (C57BL/6) and
transgenic (PV-ires-Cre; Ai14-tdTomato) mice aged 8–16 weeks were obtained from the Jackson
Laboratory. Mice were housed on 12h light/dark cycle, with food and water provided ad libitum.
Viral injections were carried out as we previously described (Ibrahim et al., 2016; Zingg et al.,
2017a). Stereotaxic coordinates were based on the Allen Reference Atlas (www.brain-map.org).
Mice were anesthetized using 1.5% isoflurane throughout the surgery procedure. A small
incision was made on the skin after shaving to expose the skull. A 0.2 mm craniotomy was made,
and virus was delivered through a pulled glass micropipette with beveled tip (~15 µm diameter)
by pressure injection. For anterograde tracing, AAV2/1-CamKII-hChR2-eYFP-WPRE-hGh
(UPenn Vector Core, 1.6×1013 GC/ml) was injected into the SSp barrel field (30-nl total
volume; AP -1.1 mm, ML +3.5 mm, DV -0.6 mm) of PV-ires-Cre::Ai14. AAV1-CAG-FLEX-
eGFP-WPRE-bGH (UPenn Vector Core, 1.7×1013 GC/ml) was injected into the ZI (30-nl total
volume; AP -2.1 mm, ML +1.5 mm, DV -4.3 mm) of PV-ires-Cre mice. Animals were
euthanized 3-4 weeks following the injection for examination.
For activity manipulations, AAV2/1-pEF1α-DIO-hChR2-eYFP (UPenn Vector Core, 1.6×10
13

GC/ml), AAV1-CAG-FLEX-ArchT-GFP (UNC Vector Core, 1.6×10
13
GC/ml), and pAAV-
hSyn-hM4D(Gi)-mCherry (Addgene, 3×10
12
VC/ml) was injected bilaterally into ZI (100 nl for
each site; AP -2.1 mm, ML +1.5 mm, DV -4.3 mm) of PV-ires-Cre mice. AAV1-CamKII-
hChR2(E123A)-eYFP-WPRE-hGh (UPenn Vector Core, 1.6×10
13
GC/ml) was injected into SSp
75
(30-nl total volume; AP +1.1 mm, ML -3.5 mm, DV -0.6 mm) of wild-type C57BL/6 mice.
Viruses were expressed for at least three weeks. For silencing studies, muscimol (M23400;
ThermoFisher) was injected unilaterally into SSp (100-nl total volume; AP +1.1 mm, ML -3.5
mm, DV -0.6 mm) or bilaterally into ZI (100-nl total volume; AP -2.1 mm, ML +1.5 mm, DV -
4.3 mm) of wild-type mice.
3.4.2 Histology, imaging and quantification
Animals were deeply anesthetized and transcardially perfused with phosphate buffered saline
(PBS) followed by 4% paraformaldehyde. Brains were post-fixed at 4˚C overnight in 4%
paraformaldehyde and then sliced into 150-μm sections using a vibratome (Leica, VT1000s). To
reveal the cytoarchitectural information, brain slices were first rinsed three times with PBS for
10 min, and then incubated in PBS containing Nissl (Neurotrace 620, ThermoFisher, N21483)
and 0.1% Triton-X100 (Sigma-Aldrich) for 2h. All images were acquired using a confocal
microscope (Olympus FluoView FV1000). To quantify the relative strength of axonal
projections of ZIv PV+ neurons in downstream structures, serial sections across the whole brain
were collected. Regions of interest were imaged at 10X magnification across the depth of the
tissue (15 μm z-stack interval). For each brain, images were taken using identical laser power,
gain and offset values. Fluorescence quantifications were performed by converting the images
into monochromatic so that each pixel had a grayscale ranging from 0 to 255. Intensity value of
the region of interest (200×200 pixel) was normalized to the baseline value. For each region of
interest, three or more sections were imaged and averaged. The fluorescence density for each
target structure was normalized for each animal and averaged across the animal group.
76
3.4.3 Optogenetic preparation and stimulation
One week before the behavioral tests, animals were prepared as previously described(Xiong et
al., 2015). Briefly, to optogenetically manipulate ZI neuron cell bodies, or ZI-POm, ZI-SC or
SSp-ZI axon terminals, mice were implanted with fiber optic cannulas (200 µm ID, Thorlabs)
two weeks after injecting ChR2 or ArchT virus(Boyden et al., 2005; Chow et al., 2010). The
animal was anaesthetized and mounted on a stereotaxic apparatus (Stoelting co.). Small holes
(500 µm diameter) were drilled at a 20-degree angle relative to the vertical plane above ZI (AP -
2.1 mm, ML ±1.5 mm, DV -4.3 mm), POm (AP -2.0 mm, ML ±1.5 mm, DV -3.0 mm) or SC
(AP -4.0 mm, ML ±1.5 mm, DV -2.0 mm). The cannulas were lowered to the desired depth and
fixed in place using dental cement. In the meantime, a screw for head fixation was mounted on
the top of the skull with dental cement. Light from a blue LED source (470 nm, 10 mW,
Thorlabs) was delivered at a rate of 20 Hz (20-ms pulse duration) via the implanted-cannulas
using a bifurcated patch cord (Ø200 µm, 0.22 NA SMA 905, Thorlabs) for ChR2 or GFP control
animals. The plastic sleeve (Thorlabs) securing the patch cord and cannula was wrapped with
black tape to prevent light leakage. Light from a green LED source (530 nm, 10 mW, Thorlabs)
for ArchT animals was delivered continuously for 5 s. Animals were allowed to recover for one
week before behavioral tests. During the recovery period, they were habituated to the head
fixation on the running plate. The head screw was tightly fit into a metal post while the animal
could run freely on a flat rotating plate. Following testing sessions, animals were euthanized, and
the brain was imaged to verify the locations of viral expression and implanted optic fibers. Mice
with mistargeted viral injections or misplaced fibers were excluded from data analysis.
77
3.4.4 Behavioral tests
Head-fixed Flight Response The test was conducted in a sound-attenuation booth (Gretch-Ken
Industries, Inc). Sound stimulation, LED stimulation and data acquisition software were custom
developed in LabVIEW (National Instruments). Each mouse was tested for one session per day
which lasted no longer than two hours. During the behavioral session, the animal was head-fixed,
and the speed of the running plate was detected with an optical shaft encoder (US Digital) and
recorded in real time(Zhou et al., 2014; Xiong et al., 2015; Zhang et al., 2018). A 2-s or 5-s
noise sound at 80 dB SPL (Scan-speaker D2905) was presented to trigger flight response as
previously described. The stimulus was repeated for about 20 trials per session at an irregular
interval ranging from 120 - 180 s. Little adaptation was observed(Xiong et al., 2015). Whisker
stimulation (for 2-s) was delivered through a cotton stick controlled by a piezo actuator
(Thorlabs). The stimulation frequency was 5 Hz and the vibration range was 4 mm. For
optogenetic experiments, the blue or green LED stimulation (lasting for the entire 5-s duration of
noise presentation) was randomly co-applied in half of the trials. LED-On and LED-Off trials
were interleaved. The exact sequence, “On-Off-On-Off…” or “Off-On-Off-On…”, was
randomized for animals in the same group, or between different test sessions. Whisker
stimulation was applied on the same side of auditory stimulation during the 2-s noise
presentation without or with muscimol infusions into the contralateral SSp or bilateral ZI. For
DREADDi experiments, animals infected with AAV-hM4Di(Gi)-mCherry(Zhu and Roth, 2014)
received either an intraperitoneal (IP) injection of clozapine-N-oxide (CNO) (1 mg/kg), or a local
infusion of CNO (3 μM, 100 nl)(Zhu et al., 2016) or saline (100 nl) through implanted cannulas
into the POm. For the LED-only or whisker stimulation only control experiments, LED or
78
whisker stimulation was given in the same way but without noise stimulation. Each animal was
tested for consecutive 2 days and data were averaged across days for each animal.
Two-Chamber Flight C57LB/6 mice were placed inside a two-chamber test box (25 cm × 40 cm
× 25 cm for each chamber). The two chambers were connected by a 50-cm long and 4-cm wide
channel. Animals were allowed to habituate in the arena for 10 min. 10-s 80 dB SPL noise was
delivered in one of the chambers. Animals flee to the other chamber by crossing the channel,
which was video recorded. Each animal was tested for two consecutive days (two trials per day).
On day two, 5h before the testing session, all whiskers were trimmed under anesthesia using
1.5% isoflurane throughout the procedure.
3.4.5 Slice preparation and recording
To confirm the connectivity between SSp axons and ZI PV+ neurons. PV-ires-Cre::Ai14 mice
injected with AAV2/1-pEF1α-DIO-hChR2-eYFP in SSp were used for slice recording. Three
weeks following the injections, animals were decapitated following urethane anesthesia and the
brain was rapidly removed and immersed in an ice-cold dissection buffer (composition: 60 mM
NaCl, 3 mM 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). Coronal slices at 350
µm thickness were sectioned by a vibrating microtome (Leica VT1000s) and recovered for 30
min in a submersion chamber filled with warmed (35 °C) ACSF (composition:119 mM NaCl,
26.2 mM NaHCO3, 11 mM glucose, 2.5 mM KCl, 2 mM CaCl2, 2 mM MgCl2, and 1.2
NaH2PO4, 2 mM Sodium Pyruvate, 0.5 mM VC). ZIv neurons surrounded by EYFP
+
fibers were
visualized under a fluorescence microscope (Olympus BX51 WI). Patch pipettes (~4 -5 MΩ
resistance) filled with a cesium-based internal solution (composition: 125 mM cesium gluconate,
5 mM TEA-Cl, 2 mM NaCl, 2 mM CsCl, 10 mM HEPES, 10 mM EGTA, 4 mM ATP, 0.3 mM
79
GTP, and 10 mM phosphocreatine; pH = 7.25; 290 mOsm) were used for whole-cell recordings.
Signals were recorded with an Axopatch 700B amplifier (Molecular Devices) under voltage
clamp mode at a holding voltage of –70 mV for excitatory currents, filtered at 2 kHz and
sampled at 10 kHz(Ji et al., 2016). Tetrodotoxin (TTX, 1 μM) and 4-aminopyridine (4-AP, 1
mM) were added to the external solution for recording monosynaptic responses only(Petreanu et
al., 2009) to blue light stimulation (5 ms pulse, 3 mW power, 10-30 trials). CNQX (20 μM,
Sigma-Aldrich) was added to the external solution to block glutamatergic currents.
For testing the efficacies of ChR2, ArchT and DREADDi, brain slices were prepared similarly,
and whole-cell current-clamp recordings were made in neurons expressing ChR2, ArchT or
DREADDi. A train of blue light pulses (20 Hz, 5-ms pulse duration) was applied to measure
spike responses of ChR2- expressing neurons. Green light stimulation (500 ms duration) was
applied to measure hyperpolarizations in ArchT-expressing neurons. For neurons expressing
DREADDi receptors, a series of 500-ms current injections with amplitude ranging from 0 to
200 pA in 20 pA steps were applied before and after perfusion of CNO (5 μM) and after washing
out CNO.
3.4.6 Optrode recording and spike sorting
The mouse was anesthetized with isoflurane (1.5%–2% by volume), and a head post for fixation
was mounted on top of the skull with dental cement and a craniotomy was performed over ZI
(AP -2.0 ~ -2.2 mm, ML +1.4 ~ +1.6 mm) three days before the recording. Silicone adhesive
(Kwik-Cast Sealant, WPI Inc.) was applied to cover the craniotomy window until the recording
experiment. Recording was carried out with an optrode (A1x16-Poly2-5mm-50 s-177-OA16LP,
16 contacts separated by 50 μm, the distance between the tip of the optic fiber and the probes is
200 μm, NA 0.22, Neuronexus Technologies) connected to a laser source (473 nm) with an optic
80
fiber. The optrode was lowered into the ZIv region, and data were acquired with the Plexon
recording system. The PV+ neurons were optogenetically tagged by injecting floxed AAV-ChR2
in PV-Cre animals. To identify PV+ neurons, 20-Hz (20-ms pulse duration, 500-ms total
duration) laser pulse trains were delivered intermittently. Signals were recorded and filtered
through a bandpass filter (0.3 - 3 kHz). The nearby four channels of the probe were grouped as
tetrodes, and semiautomatic spike sorting was performed by using Offline Sorter (Plexon).
Semiautomated clustering was carried out on the basis of the first three principal components of
the spike waveform on each tetrode channel using a T-Dist E-M scan algorithm (scan over a
range of 10-30 degree of freedom) and then evaluated with sort quality metrics. Clusters with
isolation distance < 20 and L-Ratio > 0.1 were discarded(Zhang et al., 2018). Spike clusters were
classified as single units only if the waveform SNR (Signal Noise Ratio) exceeded 4 (12 dB) and
the inter-spike intervals exceeded 1.2 ms for >99.5% of the spikes. To assess whether these units
were driven directly by ChR2 or indirectly by synaptic connections, we analyzed the onset
latency relative to each light stimulation. Only spikes with latency < 3 ms were considered as
being directly stimulated in this study. The whisker, noise or LED stimulation was given in a
pseudorandom order for 7 to 12 trials. The evoked firing rate was calculated within the
stimulation time window, subtracting the spontaneous firing rate.
3.4.6 Data Processing and statistics
For the head-fixed running test, running speed was recorded at 10-Hz sampling rate. For each
animal, trials were excluded if the peak noise-induced speed did not exceed the baseline speed by
3 standard deviations. Peak speed was determined as the maximum running speed after
averaging all running trials. Total travel distance was calculated as the integral of running speed
within a 5-s window after the onset of noise. Significance was tested between two conditions for
81
all running trials, considering the trial-by-trial variation. For the two-chamber flight test, flight
speed was calculated as the length of the channel divided by the total time animal spent in it.
Shapiro–Wilk test was first applied to examine whether samples had a normal distribution. In the
case of a normal distribution, two-tailed t-test or one-way ANOVA test was applied. Statistical
analysis was conducted using SPSS (IBM) and Excel (Microsoft).
All data and custom written code supporting the findings of this study are available upon
reasonable request from the corresponding authors.

82
References
Ache, J. M. et al. (2019) ‘Neural Basis for Looming Size and Velocity Encoding in the
Drosophila Giant Fiber Escape Pathway’, Current Biology. doi: 10.1016/j.cub.2019.01.079.
Adolphs, R. (2013) ‘The biology of fear’, Current Biology. doi: 10.1016/j.cub.2012.11.055.
Ahl, A. S. (1986) ‘The role of vibrissae in behavior: A status review’, Veterinary Research
Communications. doi: 10.1007/BF02213989.
Alex Meredith, M. and Stein, B. E. (1983) ‘Interactions among converging sensory inputs in the
superior colliculus’, Science. doi: 10.1126/science.6867718.
Allen, W. E. et al. (2017) ‘Thirst-associated preoptic neurons encode an aversive motivational
drive’, Science. doi: 10.1126/science.aan6747.
Amano, T. et al. (2011) ‘The fear circuit revisited: Contributions of the basal amygdala nuclei to
conditioned fear’, Journal of Neuroscience. doi: 10.1523/JNEUROSCI.3410-11.2011.
B.A., S. et al. (2013) ‘Independent hypothalamic circuits for social and predator fear’, Nature
Neuroscience.
Barad, M., Gean, P.-W. and Lutz, B. (2006) ‘The Role of the Amygdala in the Extinction of
Conditioned Fear’, Biological Psychiatry, 60(4), pp. 322–328. doi:
10.1016/j.biopsych.2006.05.029.
Barthó, P., Freund, T. F. and Acsády, L. (2002) ‘Selective GABAergic innervation of thalamic
nuclei from zona incerta’, European Journal of Neuroscience, 16(6), pp. 999–1014. doi:
10.1046/j.1460-9568.2002.02157.x.
Basbaum, A. I. et al. (2009) ‘Cellular and Molecular Mechanisms of Pain’, Cell. doi:
10.1016/j.cell.2009.09.028.
Bissiere, S. et al. (2011) ‘Electrical synapses control hippocampal contributions to fear learning
and memory’, Science. doi: 10.1126/science.1193785.
Blanchard, D. C. (2004) ‘Stimulus, environmental, and pharmacological control of defensive
behaviors.’, in Learning, motivation, and cognition: The functional behaviorism of Robert C.
Bolles. doi: 10.1037/10223-014.
Blanchard, D. C. and Blanchard, R. J. (2008) ‘Chapter 2.4 Defensive behaviors, fear, and
anxiety’, Handbook of Behavioral Neuroscience. doi: 10.1016/S1569-7339(07)00005-7.
Blanchard, D. C., Blanchard, R. J. and Rodgers, R. J. (1991) ‘Risk Assessment and Animal
Models of Anxiety’, in Animal Models in Psychopharmacology. doi: 10.1007/978-3-0348-6419-
0_13.
Bolles, R. C. (1970) ‘Species-specific defense reactions and avoidance learning’, Psychological
83
Review. doi: 10.1037/h0028589.
Boyden, E. S. et al. (2005) ‘Millisecond-timescale, genetically targeted optical control of neural
activity’, Nature Neuroscience, 8(9), pp. 1263–1268. doi: 10.1038/nn1525.
Campbell, B. A., Wood, G. and McBride, T. (1997) ‘Origins of orienting and defensive
responses: An evolutionary perspective.’, in Attention and Orienting: sensory and motivational
processes.
Cang, J. et al. (2018) ‘Visual function, organization, and development of the mouse superior
colliculus’, Annual Review of Vision Science. doi: 10.1146/annurev-vision-091517-034142.
Canteras, N. S. (2002) ‘The medial hypothalamic defensive system: Hodological organization
and functional implications’, Pharmacology Biochemistry and Behavior. doi: 10.1016/S0091-
3057(01)00685-2.
Capelli, P. et al. (2017) ‘Locomotor speed control circuits in the caudal brainstem’, Nature. doi:
10.1038/nature24064.
Carvalho, V. M. de A. et al. (2020) ‘Representation of Olfactory Information in Organized
Active Neural Ensembles in the Hypothalamus’, Cell Reports. doi:
10.1016/j.celrep.2020.108061.
Carvell, G. E. and Simons, D. J. (1990) ‘Biometric analyses of vibrissal tactile discrimination in
the rat.’, The Journal of neuroscience : the official journal of the Society for Neuroscience, 10(8),
pp. 2638–2648. doi: 10.1523/jneurosci.5995-10.2011.
Cezario, A. F. et al. (2008) ‘Hypothalamic sites responding to predator threats - The role of the
dorsal premammillary nucleus in unconditioned and conditioned antipredatory defensive
behavior’, European Journal of Neuroscience. doi: 10.1111/j.1460-9568.2008.06392.x.
Chalupa, L. M. and Rhoades, R. W. (1977) ‘Responses of visual, somatosensory, and auditory
neurones in the golden hamster’s superior colliculus’, The Journal of Physiology. doi:
10.1113/jphysiol.1977.sp011971.
Chou, X. L. et al. (2018) ‘Inhibitory gain modulation of defense behaviors by zona incerta’,
Nature Communications, 9(1), pp. 1–12. doi: 10.1038/s41467-018-03581-6.
Chow, B. Y. et al. (2010) ‘High-performance genetically targetable optical neural silencing by
light-driven proton pumps’, Nature, 463(7277), pp. 98–102. doi: 10.1038/nature08652.
Collignon, O. et al. (2008) ‘Audio-visual integration of emotion expression’, Brain Research.
doi: 10.1016/j.brainres.2008.04.023.
Cook, M. and Mineka, S. (1989) ‘Observational Conditioning of Fear to Fear-Relevant Versus
Fear-Irrelevant Stimuli in Rhesus Monkeys’, Journal of Abnormal Psychology. doi:
10.1037/0021-843X.98.4.448.
84
Courtin, J. et al. (2013) ‘Medial prefrontal cortex neuronal circuits in fear behavior’,
Neuroscience. doi: 10.1016/j.neuroscience.2013.03.001.
Curio, E. (1993) ‘Proximate and Developmental Aspects of Antipredator Behavior’, in Advances
in the Study of Behavior. doi: 10.1016/S0065-3454(08)60407-6.
Davis, M. et al. (2010) ‘Phasic vs sustained fear in rats and humans: Role of the extended
amygdala in fear vs anxiety’, Neuropsychopharmacology. doi: 10.1038/npp.2009.109.
Daviu, N. et al. (2020) ‘Paraventricular nucleus CRH neurons encode stress controllability and
regulate defensive behavior selection’, Nature Neuroscience. doi: 10.1038/s41593-020-0591-0.
Deng, H., Xiao, X. and Wang, Z. (2016) ‘Periaqueductal gray neuronal activities underlie
different aspects of defensive behaviors’, Journal of Neuroscience. doi:
10.1523/JNEUROSCI.4425-15.2016.
Dhande, O. S. and Huberman, A. D. (2014) ‘Retinal ganglion cell maps in the brain:
Implications for visual processing’, Current Opinion in Neurobiology. doi:
10.1016/j.conb.2013.08.006.
Diamond, M. E. et al. (2008) ‘“Where” and “what” in the whisker sensorimotor system’, Nature
Reviews Neuroscience, 9(8), pp. 601–612. doi: 10.1038/nrn2411.
Dielenberg, R. A. and McGregor, I. S. (2001) ‘Defensive behavior in rats towards predatory
odors: A review’, Neuroscience and Biobehavioral Reviews. doi: 10.1016/S0149-
7634(01)00044-6.
Dietrich, M. O. et al. (2012) ‘AgRP neurons regulate development of dopamine neuronal
plasticity and nonfood-associated behaviors’, Nature Neuroscience, 15(8), pp. 1108–1110. doi:
10.1038/nn.3147.
Dong, P. et al. (2019) ‘A novel cortico-intrathalamic circuit for flight behavior’, Nature
Neuroscience. doi: 10.1038/s41593-019-0391-6.
Douglass, A. D. et al. (2008) ‘Escape Behavior Elicited by Single, Channelrhodopsin-2-Evoked
Spikes in Zebrafish Somatosensory Neurons’, Current Biology, 18(15), pp. 1133–1137. doi:
10.1016/j.cub.2008.06.077.
Drager, U. C. and Hubel, D. H. (1975) ‘Physiology of visual cells in mouse superior colliculus
and correlation with somatosensory and auditory input’, Nature. doi: 10.1038/253203a0.
Dulac, C. and Torello, A. T. (2003) ‘Molecular detection of pheromone signals in mammals:
From genes to behaviour’, Nature Reviews Neuroscience. doi: 10.1038/nrn1140.
Edwards, D. A. and Isaacs, S. (1991) ‘Zona incerta lesions: effects on copulation, partner-
preference and other socio-sexual behaviors’, Behavioural Brain Research, 44(2), pp. 145–150.
doi: 10.1016/S0166-4328(05)80019-1.
85
Edwards, S. B. et al. (1979) ‘Sources of subcortical projections to the superior colliculus in the
cat’, Journal of Comparative Neurology. doi: 10.1002/cne.901840207.
Ehrlich, I. et al. (2009) ‘Amygdala Inhibitory Circuits and the Control of Fear Memory’, Neuron,
62(6), pp. 757–771. doi: 10.1016/j.neuron.2009.05.026.
Eilam, D. (2005) ‘Die hard: A blend of freezing and fleeing as a dynamic defense - Implications
for the control of defensive behavior’, in Neuroscience and Biobehavioral Reviews. doi:
10.1016/j.neubiorev.2005.03.027.
Eilam, D., Izhar, R. and Mort, J. (2011) ‘Threat detection: Behavioral practices in animals and
humans’, Neuroscience and Biobehavioral Reviews, 35(4), pp. 999–1006. doi:
10.1016/j.neubiorev.2010.08.002.
Endres, T., Apfelbach, R. and Fendt, M. (2005) ‘Behavioral changes induced in rats by exposure
to trimethylthiazoline, a component of fox odor’, Behavioral Neuroscience. doi: 10.1037/0735-
7044.119.4.1004.
Esposito, M. S., Capelli, P. and Arber, S. (2014) ‘Brainstem nucleus MdV mediates skilled
forelimb motor tasks’, Nature. doi: 10.1038/nature13023.
Evans, D. A. et al. (2018) ‘A synaptic threshold mechanism for computing escape decisions’,
Nature. doi: 10.1038/s41586-018-0244-6.
Fadok, J. P. et al. (2017) ‘A competitive inhibitory circuit for selection of active and passive fear
responses’, Nature. doi: 10.1038/nature21047.
Fang, Q. et al. (2020) ‘A Differential Circuit via Retino-Colliculo-Pulvinar Pathway Enhances
Feature Selectivity in Visual Cortex through Surround Suppression’, Neuron. doi:
10.1016/j.neuron.2019.10.027.
Fanselow, M. S. (1994) ‘Neural organization of the defensive behavior system responsible for
fear’, Psychonomic Bulletin & Review, 1(4), pp. 429–438. doi: 10.3758/BF03210947.
Fanselow, M. S. and Ledoux, J. E. (1999) ‘Why we think Pavlovian fear conditioning occurs in
the basolateral amygdala’, Neuron, 23, pp. 229–232. doi: 10.1016/S0896-6273(00)80775-8.
Fanselow, M. S. and Wassum, K. M. (2016) ‘The origins and organization of vertebrate
pavlovian conditioning’, Cold Spring Harbor Perspectives in Biology. doi:
10.1101/cshperspect.a021717.
Feig, S., Van Lieshout, D. P. and Harting, J. K. (1992) ‘Ultrastructural studies of retinal, visual
cortical (area 17), and parabigeminal terminals within the superior colliculus of Galago
crassicaudatus’, Journal of Comparative Neurology. doi: 10.1002/cne.903190109.
Fendt, M. et al. (2005) ‘TMT-induced autonomic and behavioral changes and the neural basis of
its processing’, in Neuroscience and Biobehavioral Reviews. doi:
10.1016/j.neubiorev.2005.04.018.
86
Ferreira-Pinto, M. J. et al. (2018) ‘Connecting Circuits for Supraspinal Control of Locomotion’,
Neuron. doi: 10.1016/j.neuron.2018.09.015.
De Franceschi, G. et al. (2016a) ‘Vision Guides Selection of Freeze or Flight Defense Strategies
in Mice’, Current Biology. doi: 10.1016/j.cub.2016.06.006.
De Franceschi, G. et al. (2016b) ‘Vision Guides Selection of Freeze or Flight Defense Strategies
in Mice’, Current Biology. doi: 10.1016/j.cub.2016.06.006.
Gabriel, J. P. et al. (2012) ‘Layer-Specific Targeting of Direction-Selective Neurons in the
Zebrafish Optic Tectum’, Neuron. doi: 10.1016/j.neuron.2012.12.003.
Gandhi, N. J. and Katnani, H. A. (2011) ‘Motor functions of the superior colliculus’, Annual
Review of Neuroscience. doi: 10.1146/annurev-neuro-061010-113728.
Glickfeld, L. L., Reid, R. C. and Andermann, M. L. (2014) ‘A mouse model of higher visual
cortical function’, Current Opinion in Neurobiology. doi: 10.1016/j.conb.2013.08.009.
Godement, P., Salaün, J. and Imbert, M. (1984) ‘Prenatal and postnatal development of
retinogeniculate and retinocollicular projections in the mouse’, Journal of Comparative
Neurology. doi: 10.1002/cne.902300406.
Grillner, S. et al. (2005) ‘Mechanisms for selection of basic motor programs - Roles for the
striatum and pallidum’, Trends in Neurosciences, 28(7), pp. 364–370. doi:
10.1016/j.tins.2005.05.004.
Gross, C. T. and Canteras, N. S. (2012) ‘The many paths to fear’, Nature Reviews Neuroscience,
13(9), pp. 651–658. doi: 10.1038/nrn3301.
Hamm, A. O. (2019) ‘Experimental validation of psychopathology in personalized psychiatry’,
in Personalized Psychiatry. doi: 10.1016/B978-0-12-813176-3.00007-9.
Hashikawa, K. et al. (2017) ‘Esr1+ cells in the ventromedial hypothalamus control female
aggression’, Nature Neuroscience. doi: 10.1038/nn.4644.
Haubensak, W. et al. (2010) ‘Genetic dissection of an amygdala microcircuit that gates
conditioned fear’, Nature. doi: 10.1038/nature09553.
Herry, C. et al. (2008) ‘Switching on and off fear by distinct neuronal circuits’, Nature. doi:
10.1038/nature07166.
Herry, C. et al. (2010) ‘Neuronal circuits of fear extinction’, European Journal of Neuroscience,
31(4), pp. 599–612. doi: 10.1111/j.1460-9568.2010.07101.x.
Hofbauer, A. and Holländer, H. (1986) ‘Synaptic connections of cortical and retinal terminals in
the superior colliculus of the rabbit: an electron microscopic double labelling study’,
Experimental Brain Research. doi: 10.1007/BF00243837.
87
Hombrebueno, J. R. et al. (2014) ‘Intravitreal Injection of Normal Saline Induces Retinal
Degeneration in the C57BL/6J Mouse’, Translational Vision Science & Technology. doi:
10.1167/tvst.3.2.3.
Hong, Y. K., Kim, I. J. and Sanes, J. R. (2011) ‘Stereotyped axonal arbors of retinal ganglion
cell subsets in the mouse superior colliculus’, Journal of Comparative Neurology. doi:
10.1002/cne.22595.
Hoover, W. B. and Vertes, R. P. (2007) ‘Anatomical analysis of afferent projections to the
medial prefrontal cortex in the rat’, Brain Structure and Function. doi: 10.1007/s00429-007-
0150-4.
Hoy, J. L., Bishop, H. I. and Niell, C. M. (2019) ‘Defined Cell Types in Superior Colliculus
Make Distinct Contributions to Prey Capture Behavior in the Mouse’, Current Biology. doi:
10.1016/j.cub.2019.10.017.
Hubbard, D. T. et al. (2004) ‘Development of defensive behavior and conditioning to cat odor in
the rat’, Physiology and Behavior. doi: 10.1016/j.physbeh.2003.10.006.
Ibrahim, L. A. et al. (2016) ‘Cross-Modality Sharpening of Visual Cortical Processing through
Layer-1-Mediated Inhibition and Disinhibition’, Neuron, 89(5), pp. 1031–1045. doi:
10.1016/j.neuron.2016.01.027.
Inayat, S. et al. (2015) ‘Neurons in the most superficial lamina of the mouse superior colliculus
are highly selective for stimulus direction’, Journal of Neuroscience. doi:
10.1523/JNEUROSCI.0173-15.2015.
Janak, P. H. and Tye, K. M. (2015) ‘From circuits to behaviour in the amygdala’, Nature. doi:
10.1038/nature14188.
Ji, X. Y. et al. (2016) ‘Thalamocortical Innervation Pattern in Mouse Auditory and Visual
Cortex: Laminar and Cell-Type Specificity’, Cerebral Cortex, 26(6), pp. 2612–2625. doi:
10.1093/cercor/bhv099.
Johansen, J. P. et al. (2010) ‘Optical activation of lateral amygdala pyramidal cells instructs
associative fear learning’, Proceedings of the National Academy of Sciences of the United States
of America. doi: 10.1073/pnas.1002418107.
Kay, R. B. and Triplett, J. W. (2017) ‘Visual neurons in the superior colliculus innervated by
Islet2+or Islet2- retinal ganglion cells display distinct tuning properties’, Frontiers in Neural
Circuits. doi: 10.3389/fncir.2017.00073.
Kim, E. J. et al. (2013) ‘Dorsal periaqueductal gray-amygdala pathway conveys both innate and
learned fear responses in rats’, Proceedings of the National Academy of Sciences of the United
States of America. doi: 10.1073/pnas.1310845110.
Kim, T. et al. (2020) ‘Dendritic and parallel processing of visual threats in the retina control
defensive responses’, Science advances. doi: 10.1126/sciadv.abc9920.
88
Kolmac, C. I., Power, B. D. and Mitrofanis, J. (1998) ‘Patterns of connections between zona
incerta and brainstem in rats’, Journal of Comparative Neurology, 396(4), pp. 544–555. doi:
10.1002/(SICI)1096-9861(19980713)396:4<544::AID-CNE10>3.0.CO;2-G.
Kolmac, C. and Mitrofanis, J. (1999) ‘Distribution of various neurochemicals within the zona
incerta: An immunocytochemical and histochemical study’, Anatomy and Embryology, 199(3),
pp. 265–280. doi: 10.1007/s004290050227.
Kong, E. et al. (2014) ‘Learning not to fear: Neural correlates of learned safety’,
Neuropsychopharmacology. doi: 10.1038/npp.2013.191.
Kreitzer, A. C. and Malenka, R. C. (2008) ‘Striatal Plasticity and Basal Ganglia Circuit
Function’, Neuron, 60(4), pp. 543–554. doi: 10.1016/j.neuron.2008.11.005.
Kunwar, P. S. et al. (2015) ‘Ventromedial hypothalamic neurons control a defensive emotion
state’, eLife. doi: 10.7554/eLife.06633.
Lecca, S. et al. (2017) ‘Aversive stimuli drive hypothalamus-to-habenula excitation to promote
escape behavior’, eLife. doi: 10.7554/eLife.30697.
LeDoux, J. (2012) ‘Rethinking the Emotional Brain’, Neuron, 73(4), pp. 653–676. doi:
10.1016/j.neuron.2012.02.004.
Lee, K. H. et al. (2020) ‘The sifting of visual information in the superior colliculus’, eLife. doi:
10.7554/eLife.50678.
Li, L. et al. (2018) ‘Stress Accelerates Defensive Responses to Looming in Mice and Involves a
Locus Coeruleus-Superior Colliculus Projection’, Current Biology. doi:
10.1016/j.cub.2018.02.005.
Li, Y. et al. (2018a) ‘Hypothalamic Circuits for Predation and Evasion’, Neuron. doi:
10.1016/j.neuron.2018.01.005.
Li, Y. et al. (2018b) ‘Hypothalamic Circuits for Predation and Evasion’, Neuron. doi:
10.1016/j.neuron.2018.01.005.
Liang, F. et al. (2015) ‘Sensory Cortical Control of a Visually Induced Arrest Behavior via
Corticotectal Projections’, Neuron, 86(3), pp. 755–767. doi: 10.1016/j.neuron.2015.03.048.
Lin, D. et al. (2011) ‘Functional identification of an aggression locus in the mouse
hypothalamus’, Nature. doi: 10.1038/nature09736.
Liu, K. et al. (2017) ‘Lhx6-positive GABA-releasing neurons of the zona incerta promote sleep’,
Nature, 548(7669), pp. 582–587. doi: 10.1038/nature23663.
Lovett-Barron, M. et al. (2014) ‘Dendritic inhibition in the hippocampus supports fear learning’,
Science. doi: 10.1126/science.1247485.
89
Lovett-Barron, M. et al. (2020) ‘Multiple convergent hypothalamus–brainstem circuits drive
defensive behavior’, Nature Neuroscience. doi: 10.1038/s41593-020-0655-1.
Lüthi, A. and Lüscher, C. (2014) ‘Pathological circuit function underlying addiction and anxiety
disorders’, Nature Neuroscience. doi: 10.1038/nn.3849.
Ma, T. P., Johnson, J. C. and Hoskins, G. A. (1997) ‘Organization of the zona incerta in the
macaque: An electron microscopic study’, Anatomical Record, 249(2), pp. 259–275. doi:
10.1002/(SICI)1097-0185(199710)249:2<259::AID-AR14>3.0.CO;2-N.
Maimon, G., Straw, A. D. and Dickinson, M. H. (2008) ‘A Simple Vision-Based Algorithm for
Decision Making in Flying Drosophila’, Current Biology. doi: 10.1016/j.cub.2008.02.054.
Mangieri, L. R. et al. (2019) ‘Defensive behaviors driven by a hypothalamic-ventral midbrain
circuit’, eNeuro. doi: 10.1523/ENEURO.0156-19.2019.
Mao, S. C., Hsiao, Y. H. and Gean, P. W. (2006) ‘Extinction training in conjunction with a
partial agonist of the glycine site on the NMDA receptor erases memory trace’, Journal of
Neuroscience. doi: 10.1523/JNEUROSCI.0365-06.2006.
Marchand, J. E. and Hagino, N. (1983) ‘Afferents to the periaqueductal gray in the rat. A
horseradish peroxidase study’, Neuroscience. doi: 10.1016/0306-4522(83)90049-0.
Marek, R. et al. (2013) ‘The amygdala and medial prefrontal cortex: Partners in the fear circuit’,
Journal of Physiology. doi: 10.1113/jphysiol.2012.248575.
Masuda, A. et al. (2013) ‘Multisensory interaction mediates the social transmission of avoidance
in rats: Dissociation from social transmission of fear’, Behavioural Brain Research. doi:
10.1016/j.bbr.2013.06.011.
May, P. J. (2006) ‘The mammalian superior colliculus: Laminar structure and connections’,
Progress in Brain Research. doi: 10.1016/S0079-6123(05)51011-2.
McCullough, K. M., Morrison, F. G. and Ressler, K. J. (2016) ‘Bridging the Gap: Towards a
cell-type specific understanding of neural circuits underlying fear behaviors’, Neurobiology of
Learning and Memory. doi: 10.1016/j.nlm.2016.07.025.
McGaugh, J. L. (2004) ‘Memory reconsolidation hypothesis revived but restrained: Theoretical
comment on Biedenkapp and Rudy (2004)’, Behavioral Neuroscience. doi: 10.1037/0735-
7044.118.5.1140.
McLaughlin, T., Hindges, R. and O’Leary, D. D. M. (2003) ‘Regulation of axial patterning of the
retina and its topographic mapping in the brain’, Current Opinion in Neurobiology. doi:
10.1016/S0959-4388(03)00014-X.
Meredith, M. A. and Stein, B. E. (1986) ‘Visual, auditory, and somatosensory convergence on
cells in superior colliculus results in multisensory integration’, Journal of Neurophysiology. doi:
10.1152/jn.1986.56.3.640.
90
Milad, M. R. R. and Quirk, G. J. J. (2002) ‘Neurons in medial prefrontal cortex signal memory
for fear extinction’, Nature, 420(6911), pp. 70–74. doi: 10.1038/nature01144.1.
Miller, S. M. et al. (2019) ‘Divergent medial amygdala projections regulate approach–avoidance
conflict behavior’, Nature Neuroscience. doi: 10.1038/s41593-019-0337-z.
Mineka, S. et al. (1984) ‘Observational conditioning of snake fear in rhesus monkeys’, Journal
of Abnormal Psychology. doi: 10.1037/0021-843X.93.4.355.
Mitrofanis, J. et al. (2004) ‘Chemoarchitectonic heterogeneities in the primate zona incerta:
Clinical and functional implications’, Journal of Neurocytology, 33(4), pp. 429–440. doi:
10.1023/B:NEUR.0000046573.28081.dd.
Mitrofanis, J. (2005) ‘Some certainty for the “zone of uncertainty”? Exploring the function of the
zona incerta’, Neuroscience, 130(1), pp. 1–15. doi: 10.1016/j.neuroscience.2004.08.017.
Mobbs, D. et al. (2009) ‘From threat to fear: The neural organization of defensive fear systems
in humans’, Journal of Neuroscience. doi: 10.1523/JNEUROSCI.2378-09.2009.
Mobbs, D. et al. (2018) ‘Foraging for foundations in decision neuroscience: Insights from
ethology’, Nature Reviews Neuroscience. doi: 10.1038/s41583-018-0010-7.
Mongeau, R. et al. (2003) ‘Neural correlates of competing fear behaviors evoked by an innately
aversive stimulus’, Journal of Neuroscience. doi: 10.1523/jneurosci.23-09-03855.2003.
Motta, S. C. et al. (2009) ‘Dissecting the brain’s fear system reveals the hypothalamus is critical
for responding in subordinate conspecific intruders’, Proceedings of the National Academy of
Sciences of the United States of America. doi: 10.1073/pnas.0900939106.
Myers, K. M. and Davis, M. (2002) ‘Behavioral and neural analysis of extinction’, Neuron,
36(4), pp. 567–584. doi: 10.1016/S0896-6273(02)01064-4.
Mysore, S. P. and Knudsen, E. I. (2011) ‘The role of a midbrain network in competitive stimulus
selection’, Current Opinion in Neurobiology. doi: 10.1016/j.conb.2011.05.024.
Nicolelis, M. A. L., Chapin, J. K. and Lin, R. C. S. (1992) ‘Somatotopic maps within the zona
incerta relay parallel GABAergic somatosensory pathways to the neocortex, superior colliculus,
and brainstem’, Brain Research, 577(1), pp. 134–141. doi: 10.1016/0006-8993(92)90546-L.
Niell, C. M. and Stryker, M. P. (2008) ‘Highly selective receptive fields in mouse visual cortex’,
Journal of Neuroscience. doi: 10.1523/JNEUROSCI.0623-08.2008.
Nomura, H. et al. (2015) ‘Memory formation and retrieval of neuronal silencing in the auditory
cortex’, Proceedings of the National Academy of Sciences of the United States of America. doi:
10.1073/pnas.1500869112.
O’Connor, D. H. et al. (2013) ‘Neural coding during active somatosensation revealed using
illusory touch.’, Nature neuroscience, 16(7), pp. 958–65. doi: 10.1038/nn.3419.
91
Ohno, S. et al. (2012) ‘A morphological analysis of thalamocortical axon fibers of rat posterior
thalamic nuclei: A single neuron tracing study with viral vectors’, Cerebral Cortex, 22(12), pp.
2840–2857. doi: 10.1093/cercor/bhr356.
Pammer, L. et al. (2013) ‘The Mechanical Variables Underlying Object Localization along the
Axis of the Whisker’, Journal of Neuroscience, 33(16), pp. 6726–6741. doi:
10.1523/JNEUROSCI.4316-12.2013.
Peirce, J. W. (2007) ‘PsychoPy-Psychophysics software in Python’, Journal of Neuroscience
Methods. doi: 10.1016/j.jneumeth.2006.11.017.
Perier, C. et al. (2002) ‘Behavioral consequences of bicuculline injection in the subthalamic
nucleus and the zona incerta in rat’, Journal of Neuroscience, 22(1529-2401 (Electronic)), pp.
8711–8719. doi: 22/19/8711 [pii].
Peter, M. et al. (2012) ‘Induction of immediate early genes in the mouse auditory cortex after
auditory cued fear conditioning to complex sounds’, Genes, Brain and Behavior. doi:
10.1111/j.1601-183X.2011.00761.x.
Petreanu, L. et al. (2009) ‘The subcellular organization of neocortical excitatory connections’,
Nature, 457(7233), pp. 1142–1145. doi: 10.1038/nature07709.
Plaha, P., Khan, S. and Gill, S. S. (2008) ‘Bilateral stimulation of the caudal zona incerta nucleus
for tremor control’, Journal of Neurology, Neurosurgery & Psychiatry, 79(5), pp. 504–513. doi:
10.1136/jnnp.2006.112334.
Prigg, T. et al. (2002) ‘Texture discrimination and unit recordings in the rat whisker/barrel
system’, Physiology and Behavior, 77(4–5), pp. 671–675. doi: 10.1016/S0031-9384(02)00917-4.
Quirk, G. J. and Mueller, D. (2008) ‘Neural Mechanisms of Extinction Learning and Retrieval’,
Neuropsychopharmacology, 33(1), pp. 56–72. doi: 10.1038/sj.npp.1301555.
Ricardo, J. A. (1981) ‘Efferent connections of the subthalamic region in the rat. II. The zona
incerta’, Brain Research, 214(1), pp. 43–60. doi: 10.1016/0006-8993(81)90437-6.
Rivlin-Etzion, M. et al. (2011) ‘Transgenic mice reveal unexpected diversity of on-off direction-
selective retinal ganglion cell subtypes and brain structures involved in motion processing’,
Journal of Neuroscience. doi: 10.1523/JNEUROSCI.0564-11.2011.
Roger, M. and Cadusseau, J. (1985) ‘Afferents to the zona incerta in the rat: A combined
retrograde and anterograde study’, Journal of Comparative Neurology, 241(4), pp. 480–492. doi:
10.1002/cne.902410407.
Roseberry, T. K. et al. (2016) ‘Cell-Type-Specific Control of Brainstem Locomotor Circuits by
Basal Ganglia’, Cell. doi: 10.1016/j.cell.2015.12.037.
Senn, V. et al. (2014) ‘Long-range connectivity defines behavioral specificity of amygdala
neurons’, Neuron. doi: 10.1016/j.neuron.2013.11.006.
92
Shaban, H. et al. (2006) ‘Generalization of amygdala LTP and conditioned fear in the absence of
presynaptic inhibition’, Nature Neuroscience. doi: 10.1038/nn1732.
Shammah-Lagnado, S. J. et al. (1985) ‘Afferent connections of the magnocellular reticular
formation: A horseradish peroxidase study in the rat’, Soc. Neurosci. Abs, 11(1), p. 1026.
Shang, C. et al. (2015) ‘A parvalbumin-positive excitatory visual pathway to trigger fear
responses in mice’, Science. doi: 10.1126/science.aaa8694.
Shang, C., Chen, Z., Liu, A., Li, Y., Zhang, J., Qu, B., Yan, F., Zhang, Y., Liu, W., Liu, Z., Guo,
X., Li, D., Wang, Y. and Cao, P. (2018) ‘Divergent midbrain circuits orchestrate escape and
freezing responses to looming stimuli in mice’, Nature Communications. doi: 10.1038/s41467-
018-03580-7.
Shang, C., Chen, Z., Liu, A., Li, Y., Zhang, J., Qu, B., Yan, F., Zhang, Y., Liu, W., Liu, Z., Guo,
X., Li, D., Wang, Y., Cao, P., et al. (2018) ‘Freezing Responses To Looming Stimuli in Mice’,
Nature Communications.
Shaw, V. and Mitrofanis, J. (2002) ‘Anatomical evidence for somatotopic maps in the zona
incerta of rats’, Anatomy and Embryology, 206(1–2), pp. 119–130. doi: 10.1007/s00429-002-
0280-7.
Shi, X. et al. (2017) ‘Retinal origin of direction selectivity in the superior colliculus’, Nature
Neuroscience. doi: 10.1038/nn.4498.
Silva, B. A., Burns, A. M. and Gräff, J. (2019) ‘A cFos activation map of remote fear memory
attenuation’, Psychopharmacology. doi: 10.1007/s00213-018-5000-y.
Silva, B. A., Gross, C. T. and Gräff, J. (2016) ‘The neural circuits of innate fear: Detection,
integration, action, and memorization’, Learning and Memory. doi: 10.1101/lm.042812.116.
Smith, J. B., Mowery, T. M. and Alloway, K. D. (2012) ‘Thalamic POm projections to the
dorsolateral striatum of rats: potential pathway for mediating stimulus-response associations for
sensorimotor habits’, Journal of Neurophysiology, 108(1), pp. 160–174. doi:
10.1152/jn.00142.2012.
Smith, R. L. (1973) ‘The ascending fiber projections from the principal sensory trigeminal
nucleus in the rat’, Journal of Comparative Neurology, 148(4), pp. 423–445. doi:
10.1002/cne.901480403.
Sofroniew, N. J. et al. (2014) ‘Natural Whisker-Guided Behavior by Head-Fixed Mice in Tactile
Virtual Reality’, Journal of Neuroscience, 34(29), pp. 9537–9550. doi:
10.1523/JNEUROSCI.0712-14.2014.
Sotres-Bayon, F., Bush, D. E. A. and LeDoux, J. E. (2007) ‘Acquisition of fear extinction
requires activation of NR2B-containing NMDA receptors in the lateral amygdala’,
Neuropsychopharmacology. doi: 10.1038/sj.npp.1301316.
93
Sternson, S. M. (2013) ‘Hypothalamic survival circuits: Blueprints for purposive behaviors’,
Neuron. doi: 10.1016/j.neuron.2013.02.018.
Supko, D. E., Uretsky, N. J. and Wallace, L. J. (1991) ‘Activation of AMPA/kainic acid
glutamate receptors in the zona incerta stimulates locomotor activity’, Brain Research, 564(1),
pp. 159–163. doi: 10.1016/0006-8993(91)91367-A.
Taffou, M. et al. (2017) ‘Multisensory aversive stimuli differentially modulate negative feelings
in near and far space’, Psychological Research. doi: 10.1007/s00426-016-0774-1.
Takahashi, L. K. et al. (2005) ‘The smell of danger: A behavioral and neural analysis of predator
odor-induced fear’, in Neuroscience and Biobehavioral Reviews. doi:
10.1016/j.neubiorev.2005.04.008.
Tardif, E. et al. (2005) ‘Intrinsic connectivity of human superior colliculus’, in Experimental
Brain Research. doi: 10.1007/s00221-005-2373-z.
Tovote, P. et al. (2016) ‘Midbrain circuits for defensive behaviour’, Nature, 534(7606), pp. 206–
212. doi: 10.1038/nature17996.
Tovote, P., Fadok, J. P. and Lüthi, A. (2015) ‘Neuronal circuits for fear and anxiety’, Nature
Reviews Neuroscience, 16(6), pp. 317–331. doi: 10.1038/nrn3945.
Trageser, J. C. (2004) ‘Reducing the Uncertainty: Gating of Peripheral Inputs by Zona Incerta’,
Journal of Neuroscience, 24(40), pp. 8911–8915. doi: 10.1523/JNEUROSCI.3218-04.2004.
Ullah, F. et al. (2015) ‘Relevance of dorsomedial hypothalamus, dorsomedial division of the
ventromedial hypothalamus and the dorsal periaqueductal gray matter in the organization of
freezing or oriented and non-oriented escape emotional behaviors’, Behavioural Brain Research.
doi: 10.1016/j.bbr.2015.07.013.
Urbain, N. and Deschênes, M. (2007) ‘Motor Cortex Gates Vibrissal Responses in a
Thalamocortical Projection Pathway’, Neuron, 56(4), pp. 714–725. doi:
10.1016/j.neuron.2007.10.023.
Vale, R., Evans, D. A. and Branco, T. (2017) ‘Rapid Spatial Learning Controls Instinctive
Defensive Behavior in Mice’, Current Biology. doi: 10.1016/j.cub.2017.03.031.
Wallace, K. J. and Rosen, J. B. (2000) ‘Predator odor as an unconditioned fear stimulus in rats:
Elicitation of freezing by trimethylthiazoline, a component of fox feces’, Behavioral
Neuroscience. doi: 10.1037/0735-7044.114.5.912.
Wang, L. et al. (2010) ‘Visual receptive field properties of neurons in the superficial superior
colliculus of the mouse’, Journal of Neuroscience. doi: 10.1523/JNEUROSCI.3305-10.2010.
Wang, L. et al. (2019) ‘Hypothalamic Control of Conspecific Self-Defense’, Cell Reports. doi:
10.1016/j.celrep.2019.01.078.
94
Wang, L., Chen, I. Z. and Lin, D. (2015) ‘Collateral Pathways from the Ventromedial
Hypothalamus Mediate Defensive Behaviors’, Neuron. doi: 10.1016/j.neuron.2014.12.025.
Wang, Q. and Burkhalter, A. (2013) ‘Stream-related preferences of inputs to the superior
colliculus from areas of dorsal and ventral streams of mouse visual cortex’, Journal of
Neuroscience. doi: 10.1523/JNEUROSCI.3067-12.2013.
Warden, M. R. et al. (2012) ‘A prefrontal cortex-brainstem neuronal projection that controls
response to behavioural challenge.’, Nature, 492, pp. 428–32. doi: 10.1038/nature11617.
Watson, G. D. R., Smith, J. B. and Alloway, K. D. (2015) ‘The Zona Incerta Regulates
Communication between the Superior Colliculus and the Posteromedial Thalamus : Implications
for Thalamic Interactions with the Dorsolateral Striatum’, The Journal of Neuroscience, 35(25),
pp. 9463–9476. doi: 10.1523/JNEUROSCI.1606-15.2015.
Wei, P. et al. (2015) ‘Processing of visually evoked innate fear by a non-canonical thalamic
pathway’, Nature Communications, 6, pp. 1–12. doi: 10.1038/ncomms7756.
Weible, A. P. et al. (2014) ‘Auditory cortex is required for fear potentiation of gap detection’,
Journal of Neuroscience. doi: 10.1523/JNEUROSCI.3408-14.2014.
Wilson, T. D. et al. (2019) ‘Dual and Opposing Functions of the Central Amygdala in the
Modulation of Pain’, Cell Reports. doi: 10.1016/j.celrep.2019.09.011.
Wurtz, R. H. and Albano, J. E. (1980) ‘Visual-motor function of the primate superior
colliculus.’, Annual review of neuroscience. doi: 10.1146/annurev.ne.03.030180.001201.
Xiong, X. R. et al. (2015) ‘Auditory cortex controls sound-driven innate defense behaviour
through corticofugal projections to inferior colliculus’, Nature Communications, 6, p. 7224. doi:
10.1038/ncomms8224.
Xu, W. and Südhof, T. C. (2013) ‘A Neural CIrcuit for Memory Specificty and Generalization’,
Science, 339(6125), pp. 1290–1295. doi: 10.1016/j.biotechadv.2011.08.021.Secreted.
Yilmaz, M. and Meister, M. (2013) ‘Rapid innate defensive responses of mice to looming visual
stimuli’, Current Biology. doi: 10.1016/j.cub.2013.08.015.
Yu, Y. S. W. et al. (2016) ‘Whiskers aid anemotaxis in rats’, Science Advances, 2(8), pp. 1–8.
doi: 10.1126/sciadv.1600716.
Zhang, G. W. et al. (2018) ‘Transforming Sensory Cues into Aversive Emotion via Septal-
Habenular Pathway’, Neuron. doi: 10.1016/j.neuron.2018.07.023.
Zhang, X. and van den Pol, A. N. (2017) ‘Rapid binge-like eating and body weight gain driven
by zona incerta GABA neuron activation’, Science, 356(6340), pp. 853–859. doi:
10.1126/science.aam7100.
Zhao, X., Liu, M. and Cang, J. (2014) ‘Visual cortex modulates the magnitude but not the
95
selectivity of looming-evoked responses in the superior colliculus of awake mice’, Neuron. doi:
10.1016/j.neuron.2014.08.037.
Zhao, Z. dong et al. (2019) ‘Zona incerta GABAergic neurons integrate prey-related sensory
signals and induce an appetitive drive to promote hunting’, Nature Neuroscience. doi:
10.1038/s41593-019-0404-5.
Zhou, M. et al. (2014) ‘Scaling down of balanced excitation and inhibition by active behavioral
states in auditory cortex.’, Nature neuroscience, 17(6), pp. 841–50. doi: 10.1038/nn.3701.
Zhou, M. et al. (2018) ‘A central amygdala to zona incerta projection is required for acquisition
and remote recall of conditioned fear memory’, Nature Neuroscience. doi: 10.1038/s41593-018-
0248-4.
Zhou, Z. et al. (2019) ‘A VTA GABAergic Neural Circuit Mediates Visually Evoked Innate
Defensive Responses’, Neuron. doi: 10.1016/j.neuron.2019.05.027.
Zhu, H. and Roth, B. L. (2014) ‘Silencing Synapses with DREADDs’, Neuron, 82(4), pp. 723–
725. doi: 10.1016/j.neuron.2014.05.002.
Zhu, Y. et al. (2016) ‘A thalamic input to the nucleus accumbens mediates opiate dependence’,
Nature. doi: 10.1038/nature16954.
Zingg, B. et al. (2017) ‘AAV-Mediated Anterograde Transsynaptic Tagging: Mapping
Corticocollicular Input-Defined Neural Pathways for Defense Behaviors’, Neuron, 93(1), pp. 33–
47. doi: 10.1016/j.neuron.2016.11.045.

Abstract (if available)

Abstract Defensive behaviors are essential for animal survival. In the natural environment, there are sensory stimuli that innately represent threats and trigger stereotyped behaviors such as freezing and flight. Although considered to be hardwired, innate defensive behavior is also flexible and subject to modulation by dynamic environmental contexts and internal physiological states. For example, the properties and intensity of the sensory cues, satiety states, and past experience are all shown to contribute to defensive behaviors. Aided by powerful genetic tools, remarkable advances have been made in recent years in our understanding of innate defensive behaviors and the underlying neural circuits. However, the brain circuits allowing flexible and context-dependent defensive behaviors are not fully understood. Here, I will present two studies investigating part of the neural circuits for control and modulation of innate defensive behaviors. ❧ In the first study, we investigated the circuit mechanisms that allow distinct defensive reactions in response to a single sensory modality. Survival in threatening situations depends on the rapid execution of an appropriate active or passive defensive response. Animals show both freezing and flight innately in respond to looming visual stimuli. How visual information converge into the superior colliculus (SC) to evoke dimorphic divergent responses are not well-understood. Here we identify the retinal-recipient SC neurons and cortical-recipient SC neurons are located in different laminae of the SC. Silencing the primary visual cortex (V1) mainly effect the visual responses in the lower superficial and intermediate layers but not the most superficial layer of SC. In addition, the retinal-recipient SC neurons are necessary for flight responses by preferentially target the deep layer of SC, while the cortical-recipient SC neurons are critical for the freezing response by preferentially target the lateral posterior nucleus of the thalamus. Together, our data define the differential circuit control of visual defense behaviors in the superior colliculus. ❧ Secondly, we studied the neural circuits that modulate defensive responses in complex sensory environment. The ability to adjust defensive behavior is critical for animal survival in dynamic environments. However, neural circuits underlying the modulation of innate defensive behavior remain not well-understood. In particular, environmental threats are commonly associated with cues of multiple sensory modalities. It remains to be investigated how these modalities interact to shape defense behavior. In this study, we report that auditory-induced defensive flight can be facilitated by somatosensory input in mice. This cross-modality modulation of defensive behavior is mediated by the projection from the primary somatosensory cortex (SSp) to the ventral sector of zona incerta (ZIv). Parvalbumin-positive neurons in ZIv, receiving direct input from SSp, mediate the enhancement of the flight behavior via their projections to the medial posterior complex of thalamus (POm). Thus, defensive flight behavior can be enhanced in a somatosensory context-dependent manner via recruiting PV neurons in ZIv, which may be important for increasing survival of prey animals.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Neural circuits underlying the modulation and impact of defensive behaviors

PDF

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

PDF

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

PDF

Contextual modulation of sensory processing via the pulvinar nucleus

PDF

The behavioral and neural bases of tactile object localization

PDF

Tactile object localization: behavioral correlates, neural representations, and a deep learning hybrid model to classify touch

PDF

Axon guidance cues in development of the mammalian auditory circuit

PDF

Sensory information processing by retinothalamic neural circuits

PDF

Functional properties of the superficial cortical interneurons

PDF

Subnetwork organization of the superior colliculus and visual system in the mouse brain

PDF

Functional synaptic circuits in primary visual cortex

PDF

Quantitative analysis of mouse medial prefrontal cortex whole brain structural connectivity

PDF

Exploring sensory responses in the different subdivisions of the visual thalamus

PDF

Synaptic circuits for information processing along the central auditory pathway

PDF

Synaptic integration in dendrites: theories and applications

Asset Metadata

Creator Wang, Xiyue (author)

Core Title Neural circuits control and modulate innate defensive behaviors

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Neuroscience

Publication Date 02/22/2021

Defense Date 01/14/2021

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

Tag defensive behaviors,Flight,freezing,OAI-PMH Harvest,superior colliculus,zona incerta

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Bonnin, Alexandre (committee chair), Hires, Samuel Andrew (committee member), Tao, Huizhong Whit (committee member), Zhang, Li I. (committee member)

Creator Email xiyuewan@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-421659

Unique identifier UC11668614

Identifier etd-WangXiyue-9282.pdf (filename),usctheses-c89-421659 (legacy record id)

Legacy Identifier etd-WangXiyue-9282.pdf

Dmrecord 421659

Document Type Dissertation

Rights Wang, Xiyue

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