University of Southern California Dissertations and Theses

Neural circuits underlying the modulation and impact of defensive behaviors

(USC Thesis Other)

Neural circuits underlying the modulation and impact of defensive behaviors

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

NEURAL CIRCUITS UNDERLYING THE
MODULATION AND IMPACT OF DEFENSIVE BEHAVIORS

by

Xiaolin Chou

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 2020

ii

Acknowledgements
Firstly, I would like to express my deepest gratitude to my mentors, Dr. Huizhong Whit Tao and
Dr. Li I Zhang, for their exceptional guidance, systematic training and long-lasting support
throughout my graduate study. I could not have achieved anything without their help and
supervision.
Secondly, I would like to thank my colleagues for their help in discussing and collaborating on
numerous experiments and their moral support to help me through many difficulties. Dr. Lukas
Mesik and Dr. Leena Ali Ibrahim taught me a lot and helped me through my early years in my
graduate school. I will also want to thank my good colleague and friends, Xiyue Wang and
Junxiang Huang, for their continuous help in both research and daily life. I would also like to thank
all my other lab members who have helped me with all kinds of experiments and made my stay in
the lab a nice experience.
Thirdly, I would also like to thank my committee members, Dr. Alexandre Bonnin and Dr. Samuel
Andrew Hires for their thoughtful suggestions and guidance throughout the dissertation process.
Finally, I would like to thank my family for their continuous understanding and support, which
have always been a great treasure of my life.

iii

Table of Contents

Table of Figures ............................................................................................................................. vi
Abstract ........................................................................................................................................ viii
Chapter 1: Introduction ................................................................................................................... 1
1.1 Different types of defensive behaviors ................................................................................. 1
1.1.1 Innate defensive behaviors ............................................................................................. 2
1.1.2 Learned defensive behaviors .......................................................................................... 4
1.2 Brain structures involved in defensive behaviors ................................................................. 6
1.2.1 Midbrain ......................................................................................................................... 6
1.2.2 Amygdala ....................................................................................................................... 8
1.2.3 Hypothalamus .............................................................................................................. 10
1.2.4 Cortex and thalamus .................................................................................................... 12
1.3 Modulation of defensive behaviors ..................................................................................... 15
1.3.1 Contextual modulation of defensive behaviors ............................................................ 15
1.3.2 Experience-dependent modulation of defensive behaviors ......................................... 17
1.4 Impact of defense circuits on sensory processing ............................................................... 19
Chapter 2: Cortical control of defensive behaviors through superior colliculus .......................... 21
2.1 Introduction ......................................................................................................................... 21
2.2 Results ................................................................................................................................. 22
2.2.1 Mapping outputs of input-defined SC neurons ............................................................ 22
2.2.2 SC neurons receiving A1 inputs control flight behaviors ............................................ 25
2.2.3 SC neurons receiving V1 inputs control freezing behaviors........................................ 27
2.3 Discussion ........................................................................................................................... 30
2.4 Material and methods .......................................................................................................... 32
2.4.1 Animal preparation and stereotaxic surgery ................................................................ 32
2.4.2 Injection of viruses for anterograde labeling ............................................................... 33
2.4.3 Histology and imaging ................................................................................................. 34
2.4.4 Optogenetic preparation and stimulation ..................................................................... 35
2.4.5 Behavioral testing and analysis .................................................................................... 36
2.4.6 Statistical analysis ........................................................................................................ 37
Chapter 3: Modulation of defensive behaviors by Zona Incerta ................................................... 38
iv

3.1 Introduction ......................................................................................................................... 38
3.2 Results ................................................................................................................................. 40
3.2.1 Inhibitory projection from ZIr to PAG ........................................................................ 40
3.2.2 ZIr modulates innate flight response ............................................................................ 43
3.2.3 ZIr modulates conditioned freezing response .............................................................. 48
3.2.4 ZIr is engaged in extinction of conditioned freezing ................................................... 49
3.2.5 Inhibitory effects of ZIr input to PAG ......................................................................... 54
3.3 Discussion ........................................................................................................................... 57
3.3.1 Function roles of different subpopulation of ZI neurons ............................................. 58
3.3.2 Top-down regulation of behaviors depending on contexts .......................................... 59
3.3.3 Gain modulation of ZIr on defensive behaviors .......................................................... 60
3.4 Material and methods .......................................................................................................... 61
3.4.1 Animal model and stereotaxic injection....................................................................... 61
3.4.2 Histology, imaging and quantification ......................................................................... 63
3.4.3 Optogenetic preparation and stimulation ..................................................................... 63
3.4.4 Behavioral tests and analysis ....................................................................................... 64
3.4.5 Slice preparation and recording ................................................................................... 67
3.4.6 In vivo recording in head-fixed animals ....................................................................... 68
3.4.7 Data processing and statistics ...................................................................................... 69
Chapter 4: Modulation of auditory cortical processing by lateral posterior nucleus of thalamus 71
4.1 Introduction ......................................................................................................................... 71
4.2 Results ................................................................................................................................. 72
4.2.1 Bidirectional modulation of frequency tuning and SNR by LP ................................... 72
4.2.2 LP exerts a thresholding effect on A1 L2/3 responses................................................. 77
4.2.3 LP’s modulatory effect is mediated by the its A1 projection ...................................... 78
4.2.4 LP axons generate a disynaptic inhibitory effect on A1 L2/3 neurons ........................ 81
4.2.5 SC can provide input to drive LP-mediated modulation of A1 responses ................... 82
4.2.6 LP plays a role in noise-related contextual modulation of A1 responses .................... 85
4.3 Discussion ........................................................................................................................... 88
4.3.1 Contextual modulation role of LP in different background noise levels ..................... 88
4.3.2 Parallel and multimodal pathway from SC to LP ........................................................ 89
4.4 Material and methods .......................................................................................................... 90
4.4.1 Animal model, stereotaxic injection and imaging ....................................................... 90
4.4.2 Optogenetic, pharmacological manipulation and slice recording ................................ 91
v

4.4.3 Sound stimulation ........................................................................................................ 92
4.4.4 In vivo electrophysiology ............................................................................................. 92
4.4.5 Data analysis and statistics ........................................................................................... 93
References ..................................................................................................................................... 96

vi

Table of Figures
Figure 1: Defensive behaviors under increasing threat ................................................................... 3
Figure 2: Diagram showing the auditory conditioning process ...................................................... 6
Figure 3: Microcircuits in PAG for flight and freezing .................................................................. 7
Figure 4: Microcircuits in amygdala for different behaviors ........................................................ 10
Figure 5: Brian circuits for innate defensive behaviors to predator odor ..................................... 12
Figure 6: Neural circuits for mediating defensive behaviors to different threats ......................... 13
Figure 7: Different types of defensive behaviors depending on contexts ..................................... 17
Figure 8: Mapping of Axonal Outputs of Input-Defined Neuronal Populations in SC ................ 24
Figure 9: A1-Recipient SC Neurons Drive an Innate Escape Behavior ....................................... 27
Figure 10: V1-Recipent SC Neurons Drive Freezing Behavior ................................................... 28
Figure 11: Efficiency of anterograde labeling within the V1-SC-LP pathway ............................ 30
Figure 12. A distinct GABAergic projection from ZIr to PAG. ................................................... 42
Figure 13. Projection patterns of ZIr and ZIv/ZId. ....................................................................... 42
Figure 14. ZIr bi-directionally modulates noise-induced flight response. .................................... 43
Figure 15. GFP control and effects of optogenetic manipulation on flight latency. ..................... 45
Figure 16. ZIr do not affect baseline locomotion and balance beam performance. ...................... 46
Figure 17. ZIr activity exerts a gain modulation........................................................................... 48
Figure 18. ZIr bidirectionally modulates conditioned freezing response. .................................... 49
Figure 19. Effect of inactivation of ZIr on extinction retrieval. ................................................... 50
Figure 20. ZIr activities during fear conditioning and fear extinction. ......................................... 51
Figure 21. mPFC contributes to ZIr activity increase during extinction. ..................................... 52
Figure 22. Silencing of mPFC impairs fear extinction. ................................................................ 54
vii

Figure 23. GABAergic output of ZIr suppresses PAG activity. .................................................... 54
Figure 24. ZIr innervates excitatory but not inhibitory neurons in PAG ..................................... 56
Figure 25. ZIr-PAG projection mediates the modulatory effects on defensive behaviors. .......... 57
Figure 26. Effects of LP manipulations on A1 response properties. ............................................ 73
Figure 27. LED illumination had no effect on auditory response................................................. 75
Figure 28. LP manipulation did not change CF values. ................................................................ 76
Figure 29. Optogenetic manipulation of LP had no effect on A1 L4 auditory response. ............. 77
Figure 30. Bidirectional activity manipulations of LP-A1 projection. ......................................... 80
Figure 31. Biased projection from caudal LP to A1. .................................................................... 81
Figure 32. Effects of LP terminal manipulation on TRF of A1 neurons. ..................................... 81
Figure 33. Cell types innervated by LP-A1 axons. ....................................................................... 82
Figure 34. SC provides input to drive LP-mediated modulation of A1 responses. ...................... 83
Figure 35. SC projections to LP with a caudal bias. ..................................................................... 85
Figure 36. LP plays a role in noise-related contextual modulation of A1 responses. ................... 86

viii

Abstract
Defensive behaviors are critical for animals’ survival. Animals need to adopt the proper type as
well as the proper level of defensive behaviors when encountering any specific danger. Previous
studies have already shown many brain structures involved in controlling the defensive behaviors
in various conditions. However, the understanding of detailed neural circuits governing the cortical
control and modulation of defensive behaviors is still lacking. In addition, it is also unknown how
the sensory processing would be influenced by the defense circuits. Here, I will present three
studies demonstrating part of the neural circuits for the control and modulation of defensive
behaviors and its impact on sensory processing in mice.
In the first study, we investigated how superior colliculus (SC) neurons control different types of
defensive behavior given a specific sensory cortical input. We combined anterograde transsynaptic
tagging and optogenetics to manipulate the SC neurons receiving inputs from either the primary
auditory cortex (A1) or the primary visual cortex (V1). We found that SC neurons receiving A1
inputs mediated the flight behavior, while SC neurons receiving V1 inputs controlled the freezing
behavior through its downstream target, lateral posterior nucleus of the thalamus (LP). These
results suggested that different subpopulations of SC neurons receiving specific cortical inputs
could control different types of defensive behaviors.
In the second study, we used techniques including neural tracing, optogenetics and in-vivo
electrophysiological recordings to investigate a potential inhibitory nucleus for modulating the
level of defensive behaviors. Here, we discovered that GABAergic neurons in the rostral sector of
ZI (ZIr) directly innervated the excitatory but not the inhibitory neurons in both the dorsolateral
and ventrolateral compartments of periaqueductal gray (PAG), which can drive flight or freezing
ix

behaviors respectively. Optogenetic activation of ZIr neurons or their projections to PAG reduced
both sound-induced innate flight response and conditioned freezing response, while optogenetic
suppression of these neurons enhanced these defensive behaviors, likely through a mechanism of
gain modulation. In addition, ZIr activity progressively increased during extinction of conditioned
freezing response and suppressing ZIr activity impaired the expression of fear extinction.
Furthermore, ZIr was innervated by the medial prefrontal cortex (mPFC), and silencing mPFC
prevented the increase of ZIr activity during extinction and the expression of fear extinction. These
results suggest that ZIr is engaged in modulating the level of various types of defensive behaviors.
In the third study, we explored the potential role of LP in auditory processing in A1. Here, we
modulated LP activity or its projection to primary auditory cortex (A1) in awake mice and found
that LP improved auditory processing in A1 supragranular-layer neurons by sharpening their
frequency selectivity and increasing the signal-to-noise ratio (SNR) of their responses. This was
achieved through a universal suppression of cortical responses to both tone and background noise,
which is mediated largely by LP’s projection to inhibitory neurons in superficial layers of A1.
Providing the major bottom-up input, SC can drive the LP-mediated modulation of A1 responses,
which alleviates the deterioration of A1 processing by increasing background noise levels. Our
results suggest that the SC-LP-A1 pathway may play a role in modulation of auditory cortical
processing during SC-LP related defensive behaviors.
Together, these studies demonstrate the neural circuits for the control, modulation as well as the
potential sensory impact of defensive behaviors. They fill the gap in our understanding of the
overall defense circuits and provide many potential directions to further investigate the defensive
behaviors from different aspects.
1

Chapter 1: Introduction
Defensive behaviors are described 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). Though the threats vary greatly for different species, the defensive behaviors
surprisingly exhibit similar patterns (Blanchard and Blanchard 2008). To minimize the potential
harm caused by the threat, the type and level of defensive behaviors are usually well controlled
based on the feature of both the threat stimuli and the surrounding environments. Lots of efforts
have been spent to investigate diverse types of defensive behaviors under various threat conditions
(Adolphs 2013; Fanselow 2018; Headley et al. 2019; Mongeau et al. 2003; Pérez-Gómez et al.
2015; Silva et al. 2016; Silva, Gross, and Gräff 2016; L. Wang, Chen, and Lin 2015; Yu et al.
2016), but our understanding of the highly conserved defense circuits remains incomplete due to
the specific brain structures or different defensive behaviors that each study focused on.
In this chapter, I will briefly overview the different types of defensive behaviors as well as its
neural circuits, with a focus on the two common types, flight and freezing. I will also introduce
the modulation of defensive behaviors and the impact of the defense circuit on sensory perception.
1.1 Different types of defensive behaviors
In natural environments, animals would adopt different types of defensive behaviors depending on
the specific danger signal and surroundings. Mice, as a prey animal, exhibit various types of
defensive behaviors (Bray 2016; Eilam 2005; Fanselow 2018; Headley et al. 2019; Tovote et al.
2016) and serve as a good model to study the underlying defense circuits. Under different
conditions, the types of defensive behaviors are highly dependent on the properties of the
2

dangerous signal and the previous experience. In general, animals would exhibit vigilant to violent
defensive behaviors as the danger moves closer (Adolphs 2013; Eilam 2005; Eilam, Izhar, and
Mort 2011; Fanselow 2018; Headley et al. 2019; Silva, Gross, and Gräff 2016). In addition,
animals could also learn to display defensive responses after learning the association of neutral
stimuli with aversive ones, which is the classical fear conditioning model (Tovote et al. 2016;
Tovote, Fadok, and Lüthi 2015). I will introduce the defensive behaviors induced by different
conditions in the following parts.
1.1.1 Innate defensive behaviors
Animals will display robust defensive behaviors in face of immediate danger, and the approaching
of predators is one of the most common threatening situations. For mice, they could exhibit various
types of defensive behaviors including increased vigilance, flight to shelter, hiding and freezing,
and defensive attack and jump (Blanchard and Blanchard 2008; Eilam 2005; Silva, Gross, and
Gräff 2016). The specific type of defensive behaviors chosen largely depends on the distance of
the predators and the availability of the escape route (Figure 1). When the predator is far away,
animals will be vigilant and move towards shelter at a normal speed. Both the extra vigilance and
moving towards the shelter of the animals protect them from the direct encounter of the predator.
As the predator comes closer, flight and freezing are the two most common types of defensive
behaviors observed. Flight can help animals to escape from the potential predators and freezing
can help animals avoid the detection by the predators. We should notice that flight is usually
dominant if an exit or safety spot is available (Adolphs 2013; Eilam 2005; Yilmaz and Meister
2013), and the switch between flight and freezing is instant according to the rapidly changing
contexts. When the predator almost contacts the animals, they would display more violent types
of defensive behaviors like jumping and attack. These forms of violent defensive behaviors are the
3

last option to scare away or stop the predator to pursue the prey. All these types of defensive
behaviors work together to facilitate the survival of animals and need to be accurately chosen and
controlled. Furthermore, the defense strategy for predators is quite similar in a variety of species
(Adolphs 2013; Blanchard and Blanchard 2008), possibly suggesting a highly conserved neural
circuits for defense.

Figure 1: Defensive behaviors under increasing threat
Diagram of different types of defensive behaviors exhibited by prey animals with an approaching
predator at each stage and the many transitions that may occur between these defensive behaviors.
Such versatility is vital in the arms race between predator and prey, in order to prevent predators
from learning the defensive response. Adapted from Eilam (2005).

4

In addition to the actual presence of predators, certain sensory cues representing characteristics of
predators can also elicit robust defensive behaviors, particularly the flight and freezing behaviors.
Previous studies have shown that the “looming” stimuli, which is an overhead expanding dark disk
visually mimicking the approach of the predators from sky, could induce robust flight and freezing
behaviors, and the bias of adopting flight behavior is dependent on the presence of a shelter (Evans
et al. 2018; Shang et al. 2015, 2018; Wei et al. 2015; Yilmaz and Meister 2013), which is quite
similar to the animals’ behavior pattern towards to a natural predator. Besides the looming stimuli,
other sensory cues like loud noise (Xiong et al. 2015), aversive ultrasound (Mongeau et al. 2003),
and predator odor (Pérez-Gómez et al. 2015; Yang et al. 2016) could also drive defensive flight or
freezing responses. The behavioral responses towards these stimuli are usually innate, but the level
of the responses can exhibit adaptation (Wei et al. 2015; Yilmaz and Meister 2013) after repeated
exposures to the stimuli without the occurrence of actual harm. This phenomenon indicates that
the experience dependent top-down inputs can probably modulate the hard-wired innate defense
circuits. I will elaborate on the neural circuits and modulation of defensive behaviors in more detail
in the following sections.

1.1.2 Learned defensive behaviors
Apart from the aversive stimuli that can induce defensive behaviors without learning, animals can
develop defense responses towards original neutral stimuli or the context by pairing it with the
unconditioned aversive stimuli. This associative learning process is termed conditioning. Figure 2
shows a paradigm that animals would freeze towards the neutral sound cue or the conditioning
contexts after pairing the stimuli with foot shock. Sensory cues from various modalities could all
5

serve as the conditioned stimulus including light, sound and olfactory cues. The defensive
behaviors induced by the conditioning are usually freezing, and sometimes flight depending on the
experiment design (Ciocchi et al. 2010; Ehrlich et al. 2009; Letzkus et al. 2011; Tovote et al. 2016;
Tovote, Fadok, and Lüthi 2015), and these behaviors are thought to highly rely on the recruitment
of the amygdala complex (Canteras, Pavesi, and Carobrez 2015; Duvarci, Popa, and Paré 2011;
Ehrlich et al. 2009; Fanselow 1994; Headley et al. 2019; Kim and Jung 2006; Pape and Pare 2010;
Romanski and LeDoux 1992; Tovote et al. 2016; Yu et al. 2016). The ability of animals to learn
the association of neutral and aversive stimuli helps them to quickly adapt to the changing
environments and more efficiently avoid the threatening situations. The defensive behaviors
induced by these learned cues will possibly require certain level of independent neural circuits
from the innate defense circuits, but the two-parallel defense circuits should finally converge
together to generate the proper behavioral outcome.

6

Figure 2: Diagram showing the auditory conditioning process
After pairing the conditioned sound cue with the unconditioned aversive stimuli, foot shock,
animals would exhibit freezing behaviors to both the sound cue and the conditioning context. The
freezing response can be reduced after the extinction training. Adapted from Tovote et al. (2015)

1.2 Brain structures involved in defensive behaviors
Lots of previous studies have shown the contribution of many brain structures in various types of
defensive behaviors. In this section, I will summarize the specific role of each brain regions in
defensive behaviors covering midbrain, amygdala, hypothalamus, and cortex.
1.2.1 Midbrain
Midbrain, also called mesencephalon, is the relative upper part of brainstem including tectum,
tegmentum, and substantia nigra. It is largely involved in motor movements due to its extensive
reciprocal connections with the spinal cord (H. Liang, Paxinos, and Watson 2011, 2012). As for
the defensive behaviors, several midbrain structures were reported to directly control the
expression of flight, freezing or jumping behaviors. The most studied midbrain structure for
defensive behaviors is periaqueductal grey (PAG). Different sub-divisions of PAG were shown to
be responsible for different types of defensive behaviors, with the dorsolateral part (dlPAG) for
flight and ventrolateral part (vlPAG) for freezing responses respectively (Deng, Xiao, and Wang
2016; Misslin 2003; Morgan, Whitney, and Gold 1998; Xiong et al. 2015). In addition, the
excitatory projection neurons were shown to be the commanding center for both behaviors, and
the excitatory neurons in dlPAG could excite inhibitory neurons in vlPAG to facilitate the switch
from freezing to flight as shown in the circuit model in Figure 3 (Tovote et al. 2016). In the
meanwhile, the contribution of other parts of PAG in defensive behaviors remains elusive. Several
reports have conflicting results regarding the specific behaviors induced by different sub-regions
7

of PAG. More studies with detail characterization of the specific contribution of different sectors
and cell types in PAG are necessary for better interpreting its role in controlling various types of
defensive behaviors.

Figure 3: Microcircuits in PAG for flight and freezing
Circuit diagram showing the inhibitory neuron in vlPAG receive inhibition from CEA and
excitation from dlPAG to control the switch of freezing and flight behaviors. BLA, basolateral
amygdala; CEA, central nucleus of amygdala; PN, pyramidal neruons; IN, inhibitory neurons.
Adapted from Tovote et al. (2016).

Besides PAG, several other midbrain structures were also shown to be involved in the control of
defensive behaviors. The superior colliculus (SC) and inferior colliculus (IC) are the two structures
closely related to transform aversive sensory stimuli to defensive behaviors. Both SC and IC could
send direct excitation to dlPAG to trigger flight behaviors (Evans et al. 2018; Xiong et al. 2015).
In addition, SC could initiate freezing responses through lateral posterior nucleus of the thalamus
(Wei et al. 2015). The pavalbumin positive (PV+) cells in the superficial and intermediate layers
8

seem to largely contribute to the defensive behaviors evoked by threatening visual stimuli, and its
projection to the parabigeminal nucleus (PBN) is responsible for the flight responses (Shang et al.
2015, 2018). Midbrain locomotion region (MLR) or cuneiform nucleus was also reported to be
engaged in controlling the flight or escape behaviors (Capelli et al. 2017; Gatto and Goulding
2018), while lateral dorsal tegmentum nucleus (LDT) was shown to mediate olfactory induced fear
response, like freezing (Yang et al. 2016).
With all these findings, it is with little doubt that various midbrain structures are indeed involved
in controlling the defensive behaviors. However, we should be cautious that these midbrain
structures might not be specific for defensive behaviors. All these midbrain structures are closely
related with movement control, and the expression of defensive behaviors would naturally recruit
them. It is important to identify under what circumstances these regions would be activated, and
how the danger signals would be relayed to these structures to initiate defensive behaviors. The
input circuits to these midbrain structures for controlling defensive behaviors would be covered in
the latter sections.
1.2.2 Amygdala
Amygdala is an almond shape nucleus located in the medial temporal lobe of the brain. It is part
of the limbic system and mainly associated with the processing of memory and emotions,
particularly the fear conditioning process. It consists of two major parts, the basolateral part (BLA)
including lateral, basal and basomedial subregions, and the central nucleus of amygdala (CeA)
including the lateral and medial subdivisions. In general, sensory information from both cortical
and subcortical regions feeds into BLA and then relay to output center, CeA, to generate proper
behaviors (Janak and Tye 2015). During conditioning, the inhibitory output from CeA to vlPAG
9

are believed to induce defensive freezing response (Ciocchi et al. 2010; Kim and Jung 2006;
Letzkus et al. 2011; Pape and Pare 2010; Tovote et al. 2016). Apart from freezing, the conditioned
flight response was shown to follow a similar circuit pathway with the output from CeA to
disinhibit dlAPG (Tovote et al. 2016). Inside CeA, studies found two populations of inhibitory
neurons, which are somatostatin positive (SOM+) and protein kinase C (PKC) δ positive neurons,
inhibits each other to generate opposite behavioral control (Ciocchi et al. 2010; Ehrlich et al. 2009;
Janak and Tye 2015; Tovote, Fadok, and Lüthi 2015). Detail circuits model are shown in Figure
4. The complicated inner circuits of amygdala suggest its differential role in controlling the
defensive behaviors depending on the inputs from diverse cortical or subcortical regions under
different scenarios.
This amygdala mediated pathway for defensive behaviors usually converge into the midbrain
nuclei, especially PAG, to manifest its influence. This potentially demonstrates that the midbrain
is the converging point for both learned and innate defensive behaviors, and the amygdala is critical
for the association of neutral stimuli with negative valence. However, recent findings indicate that
amygdala might not only convey negative valence signal for fear conditioning, but also the reward
signal (Baxter and Murray 2002; Murray 2007; Wassum and Izquierdo 2015). In addition,
amygdala might also have a role in innate defense circuits by directly relaying aversive sensory
stimuli to midbrain region or hypothalamic regions, which is independent of the learning or
conditioning process, as shown by some studies (Miller et al. 2019; Pérez-Gómez et al. 2015;
Silva, Gross, and Gräff 2016) Different subregions and different cell types within amygdala could
be the key to dissolve the multifarious functions reported. It will be quite beneficial to pay attention
to both the cell type and subdivision specificity in amygdala when interpreting these results.

10

Figure 4: Microcircuits in amygdala for different behaviors
Different populations of BLA neurons are proposed to activate distinct populations of lateral
central nucleus of the amygdala (CeL) neurons to either promote or reduce fear related behaviors.
DVC, dorsal vagal complex; PVT, paraventricular nucleus of the thalamus; HYP, hypothalamus.
Adapted from Janak & Tye (2015).

Overall, the amygdala complex play an important role in mediating the defensive behaviors
induced by learned cues through conditioning, while its contribution in conveying unconditioned
stimulus to trigger defensive behaviors, like flight or freezing by predator odor, sound or visual
cues, needs further exploration to reach a comprehensive understanding.

1.2.3 Hypothalamus
Hypothalamus is a relatively small region located at the base of the brain. It is mainly responsible
for the control of hormone release to maintain the internal balance including heart rate, blood
11

pressure, body temperature, breathing, food and drink intake, sleep etc. Defensive behaviors, as an
acute response towards the danger, would certainly involve the changes in these internal states, so
hypothalamus is naturally an important part of the defense circuits to direct corresponding changes
and behaviors according to the immediate danger signal.
Numerous studies have indeed shown the involvement of hypothalamus in mediating the defensive
behaviors in various contexts. We should note that there are still lots of sub-regions within
hypothalamus, and the function of each could differ. Among these sub-regions, ventromedial
hypothalamus (VMH) seems to be closely related to the control of defensive behaviors. The
activation level of VMH is correlated with the type of defensive behaviors exhibited (Kunwar et
al. 2015), and its projection to PAG induce freezing, while its projection to anterior hypothalamic
nucleus (AHN) induce avoidance and jumping (L. Wang, Chen, and Lin 2015). There are also
reports showing that estrogen receptor α (Esr1) positive cells in ventrolateral part of the
ventromedial hypothalamus (VMHvl) contribute to the conspecific defensive behaviors like attack
(L. Wang et al. 2019). Besides VMH, other hypothalamic nuclei were involved in the control of
defensive behaviors as well. Lateral hypothalamus (LHA) was shown to induce attack or evasion
depending on its inhibitory or excitatory inputs to PAG (Li et al. 2018), and it could also induce
escape behavior by sending inputs to lateral habenula (LHb) (Lecca et al. 2017). In addition,
dorsomedial hypothalamus (DMH) and paraventricular hypothalamus (PVH) could also trigger
escape or jumping behaviors by their projection to midbrain structures (Mangieri et al. 2019; Ullah
et al. 2015). The hypothalamic circuit for defensive behaviors is mostly activated by odor signals
either from the predator or the conspecific competitor, and this olfactory fear signal could also be
relayed by dorsal premammillary nucleus (PMd) to PAG apart from the direct input pathway from
12

VMH to PAG (Canteras, Pavesi, and Carobrez 2015) to control the defensive behaviors, as shown
in Figure 5.

Figure 5: Brian circuits for innate defensive behaviors to predator odor
Signal of predator odor could follow from amygdala to hypothalamus then to midbrain structures
to initiate defensive behaviors. MEApv, medial amygdalar nucleus posteroventral part; AHN,
anterior hypothalamic nucleus; PMd, dorsal premammillary nucleus; VMHdm, ventromedial
hypothalamic nucleus, dorsomedial part. Adapted from Canterous et al. (2015)

To sum up, hypothalamus can directly initiate various types of defensive behaviors, and different
subdivisions could have differential roles depending on its cell type and downstream circuits
engaged. The involvement of hypothalamus in defensive behaviors reflects that the changes in all
kinds of physiological states like heart rate, breathing or even hormonal level, are necessary for
the expression of the corresponding behaviors, and hypothalamus sits at the right position to
reconcile the two needs for initiating behaviors and adjusting states.

1.2.4 Cortex and thalamus
After the previous discussion for the defense circuits regarding the integration and output part, the
circuit for detecting the threat is still missing. Various sensory cortices and thalamus are important
for coding the external sensory stimuli, but how the danger signal is specifically passed on to the
downstream defense circuit remains unclear. Different sensory modalities might follow distinct
13

pathways for relaying threat signal to trigger defensive behaviors. A simple diagram showing the
circuits for processing different types of threats is shown in Figure 6 (Silva, Gross, and Gräff
2016).
Figure 6: Neural circuits for mediating defensive behaviors to different threats
A detection unit (upper plane), an integration unit (middle plane), and an output unit (lower plane)
consist the circuit. Information about the threat is collected through different sensory modalities.
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 PAGd. Moving visual stimuli in
the upper visual field are processed by SC, which receives inputs from the retinal ganglion cells
(RGN) and V1, 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
14

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 MEApd and
predator cues to 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 BLA circuit. The hypothalamic integration unit
processing conspecific fear includes four highly interconnected nuclei: the medial preoptic nucleus
(MPN), VMHvl, ventral premammillary nucleus (PMV), and PMDdm. The conspecific fear circuit
mediates defensive responses through its projections to the PAGd. The predator fear circuit
consists of the AHN, 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 vlPAG via CEA. The
CEA receives noxious information from PB. The BLA plays a major role in foot shock-induced
fear through its projections to the CEA and integrates nociceptive information from the PAG via
midline thalamic nuclei (MTN). Adapted from Silva, Gross, et al. (2016)

In general, sensory information about the danger could relay from the thalamus to the cortex, and
then it could feed into the higher association cortices or BLA to generate the output to midbrain
structures. For the auditory modality, primary auditory cortex (A1) was shown to provide
excitatory inputs to the external shell part of inferior colliculus (ICx) to trigger the flight response
given a loud noise (Xiong et al. 2015). There is an additional recent study demonstrating a direct
connection from auditory cortex to dlPAG to trigger flight (H. Wang et al. 2019). For the visual
modality, primary visual cortex (V1) was also shown to excites SC to induce a temporary arrest
behavior given a flash light (F. Liang et al. 2015). However, the situation for transmitting visual
threat signals is a bit more complicated given that the retinal ganglion cells (RGC) could directly
drive SC neurons, which is important for the looming induced defensive behaviors (Shang et al.
2015, 2018; Wei et al. 2015; Yilmaz and Meister 2013). In this scenario, the contribution of the
visual cortex might just be to amplify the magnitude of the looming responses in SC (Zhao, Liu,
and Cang 2014). We should also noticed that the lateral posterior nucleus of the thalamus (LP), a
secondary visual thalamus, plays an important role in mediating the freezing response initiated by
SC (Shang et al. 2018; Wei et al. 2015). Apart from the sensory cortices, another recent study
15

showed that cingulate cortex could drive the limbic thalamic reticular nucleus (TRN) to inhibit
intermediodorsal thalamic nucleus (IMD) to induce flight (Dong et al. 2019). This possibly
indicates a pathway for risk assessment of sensory information by higher order cortices.
With these results, we should know that the involvement of cortices and thalamus in defensive
behaviors varies a lot depending on the specific sensory stimulus and its related circuits. The only
possible common feature is that these circuits would finally converge onto the midbrain nuclei to
initiate the output of the behaviors. Therefore, it would be reasonable to identify the role of various
cortices and thalamic nuclei in defensive behaviors case by case based on the specific scenario and
sensory stimuli. In Chapter 2, I will present a study showing the contribution of cortical-collicular
pathways in direct control of defensive behaviors.

1.3 Modulation of defensive behaviors
In the previous sections, I have introduced the neural circuits directly controlling the expression
of the defensive behaviors. Meanwhile, both environmental contexts and previous experience of
the threat stimuli can both modulate the level and type of the exhibited defensive behaviors, which
adds another layer of complexity to the circuits involved. Here, I will briefly review the modulation
of defensive behaviors under different conditions.
1.3.1 Contextual modulation of defensive behaviors
As mentioned earlier, animals would choose to flight in response to the looming stimulus if exit
routes or safe shelters are available. If not, animals would choose to freeze (Yilmaz and Meister
2013). This is a good example for the adjustment of defensive behaviors based on the
16

environmental contexts. There are also many other situations when the animals would adjust their
defense behaviors according to the environmental cues, like the location of the shelter, presence
of obstacle or company of groups (Evans et al. 2019), as shown in Figure 7. Such modulation
process could happen real time for animals to choose the most proper defense strategy to ensure
survival. However, the underlying circuits for this type of modulation are largely unknown, and it
would require the involvement of higher order cortical areas, possibly the medial prefrontal cortex
(mPFC), for risk assessment and decision making. One recent report showing the contribution of
cingulate cortex in flight behavior (Dong et al. 2019) shed light on the potential circuits involved
in integrating environmental cues to guide proper behaviors. It still needs further investigation
whether mPFC or other cortical or thalamic nuclei could mediate the regulation of defensive
behaviors in various contexts and scenarios.

17

Figure 7: Different types of defensive behaviors depending on contexts
Escape directionality depends on the presence and location of shelter (left panels). (Top) When an
animal has knowledge that a refuge is not available, the predominant response to threat switches
from flight to freezing. (Bottom) The presence of a close refuge in the environment guides precise
escape trajectories to its location. If a refuge becomes unavailable and the animal finds a new one,
flight paths are modified accordingly. The physical and social environment modulates escape
(right panels). (Top) Flight trajectories depends on the presence of obstacles in the environment.
Fish would flee away from an approaching predator in different routes if an obstacle occludes the
optimal escape path. (Bottom) Solitary fish can initiate escapes at various angles, whereas
schooling fish escape in straight and uniform trajectories owing to the spatial constraints imposed
by the group. Adapted from Evans, Stempel, Vale, & Branco. (2019)

1.3.2 Experience-dependent modulation of defensive behaviors
Another major type of modulation of defensive behaviors is experience dependent. After repetitive
presentation of the threatening stimuli without actual harm, animals would adapt their defensive
responses towards it. For naturally aversive stimuli, the adaptation could probably happen at
multiple levels along the signaling pathway. For example, the visual adaptation could happen at
retina level as well as the cortical level (Carandini 2000; Demb 2008; Ferreira-Netto, Genaro
Borelli, and Lira Brandão 2005; Kohn 2007; LeDue et al. 2013; Wark, Fairhall, and Rieke 2009).
We should notice that stimulus of different modalities or features would have different levels of
adaptation. The visual looming responses showed a very quick adaptation after two or three
representation of the stimuli (Wei et al. 2015), while sound stimuli including strong noise and
ultrasound or predator odor had limited level of adaptation (Mongeau et al. 2003; Xiong et al.
2015; Yang et al. 2016). This probably indicates the necessity of robust responses towards the
highly conservative and dangerous signal, but those cues associated with less certain threat could
be adapted to increase energy efficiency.
Apart from the adaptation of innate aversive signal, the defensive responses towards learned cues
after fear conditioning could also be down-regulated by repetitive exposure of the conditioned cues
18

without pairing with the unconditioned stimulus, a process termed fear extinction. In general,
amygdala still plays a central role in extinction, while the specific inputs to amygdala during the
extinction process is critical to understand the overall circuit. It is believed that the extinction
circuits are composed of other separate upstream pathways to amygdala, not simply reverse the
learning process (Barad, Gean, and Lutz 2006; Headley et al. 2019; Herry et al. 2010; Myers and
Davis 2002; Pape and Pare 2010; Quirk and Mueller 2008). The medial prefrontal cortex (mPFC),
especially the prelimbic (PL) and infralimbic (ILA) areas, send strong projection to BLA, which
has differential roles in the extinction process (M. R. Milad and Quirk 2002; Peters et al. 2010;
Sierra-Mercado, Padilla-Coreano, and Quirk 2011; Sotres-Bayon et al. 2012). This pathway
possibly contributes to the dissociation of the conditioned cue with negative valence. Besides the
input from mPFC, subiculum in hippocampus and the entorhinal cortex were also reported to be
involved in extinction by a direct projection to BLA (Baldi and Bucherelli 2014; L. R. Bevilaqua
et al. 2006; L. R. M. Bevilaqua et al. 2008; Herry et al. 2010; J. Ji and Maren 2007, 2008; Kim and
Jung 2006; Sotres-Bayon et al. 2012), which might carry the contextual information about the fear
memories. By changing the connection strength from these inputs to BLA during the extinction
training, the output of amygdala could be down-regulated to suppress the defensive behaviors.
Since the expression of the defensive behaviors engage many nodes along the pathway, the
modulation could happen at multiple stages not limited to amygdala.
Overall, defensive behaviors towards both the innate aversive stimuli and learned cues can be
modulated or adapted after repetitive experience that they are uncorrelated with danger. This
modulation process seems to be opposite to the conditioning process, in which animals learn the
association of neutral cues with danger and develop defensive behavior towards them. Yet, the
underlying circuits for the two differ a lot. The experience dependent modulation of defensive
19

behaviors would require the coordination of many brain structures from the sensory input pathways
involving sensory cortices to the integration nuclei, including mPFC and amygdala, and then to
the behavioral control midbrain structures like PAG. These modulation circuits work in parallel
with the basic circuits for the initiation of defensive behaviors to make the animals cope with the
changing environments much better. In Chapter 3, I would present a study to demonstrate an
inhibitory sub-thalamic pathway for modulating defensive behaviors according to experience.

1.4 Impact of defense circuits on sensory processing
The execution of defensive behaviors is not only the immediate responses towards the dangerous
stimuli, but also a long-lasting process that persists until the disappearance of threats. External
cues keep changing during the defense process, and it is also necessary to modulate the sensory
responses to the changing sensory cues to prioritize the execution of defensive behaviors. The
circuits underlying the impact of defensive behaviors on sensory processing could work together
with the commanding circuits to achieve the best performance for survival.
Previously, I have reviewed the neural circuits for both control and modulation of defensive
behaviors, but the circuit mechanism for its impact on sensory processing is not well understood.
However, we could still get some clues for the potential circuits from the studies showing the
modulation role of different brain states on sensory processing in cortices. For auditory processing,
active brain states including locomotion could down-regulate the cortical responses in layer 2/3
(L2/3) neurons in A1 by scaling down the balanced excitation and inhibition (M. Zhou et al. 2014).
This process probably involves the recruitment of layer 1 inhibitory neurons. Another study
showed that sound-guided behavior could suppress the activity of SOM+ cells to cancel the
20

habituation effects (Kato, Gillet, and Isaacson 2015). For somatosensory processing, active
whisking induced a desynchronized cortical activity in L2/3 neurons in barrel cortex (Poulet and
Petersen 2008). In addition, the non-fast-spiking GABAergic neurons would fire more in active
wakefulness compared to quiet state, which lead to the reduction in synchrony of membrane
potential in L2/3 excitatory neurons in the barrel cortex (Gentet et al. 2010). For the visual
modality, locomotion would cause an increase in firing rates and depolarization of the membrane
potential in L2/3 neurons in V1 (Niell and Stryker 2010; Polack, Friedman, and Golshani 2013),
while isolated arousal state could suppress the neural activity and enhance the visual processing
(Bennett, Arroyo, and Hestrin 2013; Vinck et al. 2015). Midbrain locomotor region (MLR)
responsible for the flight responses was reported to project to basal forebrain to exert the influence
on V1 (A. M. Lee et al. 2014). These results suggest that sensory processing in cortical areas are
highly likely to be modulated by the defensive behaviors as well.
Among the various structures involved in the defense circuits, the thalamic nuclei are in good
positions to exert such modulation effects on sensory processing. As mentioned above, LP is
responsible for the visual induced freezing response and sends direct projection to the cortical area
at the same time. This suggest that LP could be a possible candidate for modulating cortical activity
induced by the defense circuits. In Chapter 4, I will present a study investigating role of LP in
auditory processing in A1, which reflects the potential circuit for the impact of defense on sensory
processing.

21

Chapter 2: Cortical control of defensive behaviors through
superior colliculus
2.1 Introduction
Defensive behaviors are critical for animals’ survival, especially for prey animals like mice.
Animals exhibit different types of defensive behaviors depending on both internal states and
external contexts. When animals are in the state of anxiety or depression, their fear responses
towards dangerous signal might be amplified. Meanwhile, sensory signals bearing distinct
threatening information could also trigger different types or levels of defensive behaviors. It is
well known that stimulus from various sensory modalities excite sensory cortices through separate
bottom-up pathways, but it is still unclear how different sensory cortices control the expression of
certain defensive behaviors.
Superior colliculus (SC) is a highly conservative midbrain structure across species, and it is known
to be engaged in many behaviors like orienting, pursuing, defending, and arousal (Cang et al. 2018;
Dean, Redgrave, and Westby 1989; Ito and Feldheim 2018). In mice, SC receives divergent inputs
from multiple cortical regions and projects to various brainstem structures (Dean, Redgrave, and
Westby 1989; Fanselow 1994; Huerta and Harting 1984; Oh et al. 2014; Zingg et al. 2017). This
specific connectivity pattern of SC makes it a good candidate for mediating the control of
behaviors induced by distinct cortical activities. In this study, we take advantage of the anterograde
transsynaptic feature of a certain type of AAV virus and mapped the organization of
subpopulations of SC neurons receiving different cortical inputs. We found that different cortical
areas target distinct neuronal populations in SC, which have different downstream outputs.
22

Activation of these specific corticollicular pathway from either auditory or visual cortex generates
flight or freezing behaviors, respectively.

2.2 Results
2.2.1 Mapping outputs of input-defined SC neurons
The anterograde transsynaptic property of AAV-Cre allows its application in mapping the axonal
projections of neuronal populations with a specific presynaptic input source (i.e., input defined).
To take advantage of this property, we used a two-step viral injection procedure and selected SC
as the target structure because it receives inputs from the retina as well as a variety of cortical
areas, which terminate within distinct layers of SC (Comoli et al. 2012; Huerta and Harting 1984).
We first injected AAV1-hSyn-Cre in V1 of Ai14 mice to label SC neurons that receive V1 input.
Several days later, a second injection of AAV1-CAG-FLEX-GFP was made in SC (Figure 8A, left
panel). After 4 weeks post-injection survival time, we observed robust GFP expression in
tdTomato+ neurons specifically within the superficial gray layer of SC (i.e., SC-sg; Figure 8A,
middle panel). In total, 100% of GFP+ neurons were also tdTomato+, consistent with a result of
co-transduction of Cre-dependent GFP and anterogradely transported Cre viruses. We examined
the long-range axonal projections of this specific set of SC superficial neurons to various target
structures. The most prominent axonal labeling was found in the lateral posterior nucleus (LP) of
the thalamus, with additional labeling seen in pretectal regions and PBG (Figure 8A, right panel).
Next, we examined SC neurons receiving retina ganglion cell projections, which also target the
superficial layer of SC. As with V1, paired injections in the retina and SC yielded a group of GFP+
neurons specifically within SC-sg (Figure 8B). Their axonal projection targets were found to
23

closely match those of the V1 input-defined population (compare Figures 8A-B), suggesting that
the retina and V1 may target a similar group of SC neurons.
We further examined SC neuron groups specifically receiving inputs from the primary auditory
cortex (A1) and primary motor cortex (M1). Following paired injections into A1 and SC, we
observed numerous GFP+ cells within the medial aspect of deep gray layer (SC-dg), with a few
scattered in overlying intermediate gray layer (SC-ig) (Figure 8C). This labeling pattern is
consistent with the known distribution of A1 projections to SC (Xiong et al. 2015). Axonal outputs
of A1 input-defined SC neurons were apparently different from those of superficial SC neurons,
as prominent projections were seen in the PBG, periaqueductal gray (PAG), cuneiform nucleus
(CUN), rostral pontine reticular nucleus (PRNr), and among other regions (Figures 8C, E). Only
very sparse input to LP was observed as compared with the strong projection to LP from superficial
SC neurons. In comparison, paired injections in M1 and SC labeled a population of neurons
restricted to the lateral aspect of SC-ig (Figure 8D). Axonal outputs from this group of SC neurons
differed further from those defined by A1 and V1/retina inputs (Figure 8D).
To summarize the axonal output profiles of input-defined SC neuron subpopulations, we examined
all brain sections containing GFP+ axons and plotted those regions containing observable synaptic
boutons on corresponding atlas sections (Figure 8E). SC neurons receiving V1 and retinal inputs
exhibited very similar output profiles (Figures 8A, B, E [blue]), while those defined by A1 and
M1 inputs showed profiles distinct from each other and from the V1/retina-defined population
(Figures 8C, D [red], E [yellow]). For example, unlike sub-populations receiving A1 or V1 input,
M1-defined SC neurons exhibited prominent projections to some contralateral targets, especially
in the posterior part of the brainstem, such as the tegmental reticular nucleus (TRN),
gigantocellular reticular nucleus (GRN), and inferior olive (IO) (Figures 8D-E). Together, these
24

results demonstrate that different subpopulation of SC neurons receiving inputs from different
cortical areas have distinct output projection patterns, indicating their differential roles in
mediating various types of behaviors.

Figure 8: Mapping of Axonal Outputs of Input-Defined Neuronal Populations in SC
(A) Left, AAV1-hSyn-Cre was injected into V1 of Ai14 mice, followed by a second injection of
AAV1-CAG-FLEX-GFP into SC. Middle, GFP-labeled neurons in SC-sg were also tdTomato+.
Right four panels, GFP-labeled axons in various regions (LP, pretectal area, PBG, and cuneiform
nucleus [CUN]) downstream of SC. High-magnification images (bottom) reveal ramified axons
and their terminal and bouton structures. Blue, Nissl staining. (B) Axonal outputs of SC-sg neurons
that receive input from the contralateral retina. Data are displayed in a similar way as in A. (C)
Axonal outputs of SC neurons that receive input from A1, which are located mainly in SC-dg and
25

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

2.2.2 SC neurons receiving A1 inputs control flight behaviors
The SC integrates diverse sensory and motor information and has been implicated in controlling
orienting behaviors as well as innate defense behaviors, such as freezing and escape (Dean,
Mitchell, and Redgrave 1988; F. Liang et al. 2015; Sahibzada, Dean, and Redgrave 1986;
Schenberg et al. 2005; Shang et al. 2015; Wei et al. 2015). Since SC neuron subpopulations
receiving V1 and A1 inputs have different downstream target profiles, we speculated that each
might participate in driving a distinct SC-mediated behavior. To test this, we first performed paired
injections of AAV1-Cre in A1 and AAV1-EF1α-DIO-ChR2-EYFP in SC to enable Cre-dependent
expression of ChR2 in deep-layer SC neurons (Figures 9A). We could then activate the SC neurons
in awake, freely moving mice using pulses of blue LED light (at 20 Hz for 5 s) delivered through
an implanted optical fiber (see 2.4 Material and methods). We tested both freezing and escape
responses using a two-chamber set up in which the mouse was acclimated to a “home” chamber
on one side for 10 min, upon which a door was then removed to allow its exploration of the
adjoining, “novel” chamber (Figure 9B). As the mouse was exploring the novel chamber, SC was
optically activated and subsequent escape to home chamber or freezing was monitored (Figure
9B). Mice expressing ChR2 in A1-recipient, deep-layer SC neurons demonstrated a robust escape
response following LED light activation (Figures 9C-D). Such LED-induced escape behavior was
26

not observed in sham mice receiving injection of only GFP-expressing virus (Figures 9C-D). The
effect of optically activating A1-recipient SC neurons was like applying a loud noise sound (5 s,
70 dB sound pressure level), which robustly drove an escape response in most of the trials (Figures
9C-D, noise), consistent with our previous study (Xiong et al. 2015). No freezing response was
observed for these mice on any trials (Figure 9E). Together, these results demonstrate that SC
neurons receiving A1 inputs mediate the flight but not the freezing responses.

27

Figure 9: A1-Recipient SC Neurons Drive an Innate Escape Behavior
(A) Schematic illustration of paired injections in A1 and SC, as well as LED illumination applied
(A1-SC). Right panels, images of injection sites (red for tdTomato; green for ChR2) in an example
animal. Scale bar, 500 mm. (B) Schematic illustration of two-chamber behavior setup for testing
freezing or escape. (C) Movement tracking for an example A1-SC mouse under LED stimulation
(left), A1-SC mouse under noise stimulation (middle), and sham mouse under LED stimulation
(right) in the novel chamber during 5 s LED activation or 5 s noise stimulation. Each curve
represents one trial. Blue dot indicates the starting location at the initiation of LED or noise
stimulus, and red dot indicates the location at the end of the stimulus. Red dot beyond the novel
chamber boundary indicates that the animal has returned to the adjacent home chamber within 5 s
(bottom left). For ‘‘sham,’’ AAV1-FLEX-GFP was injected in SC. (D) Summary of percentage of
trials that induced escape behavior (n = 7 mice for A1-SC group; n = 5 mice for sham). Error bar
= SD. ***p < 0.001, t test. (E) Percentage of trials that induced freezing behavior. Error bar = SD.

2.2.3 SC neurons receiving V1 inputs control freezing behaviors
We similarly performed paired injections of AAV1-Cre in V1 and AAV1-EF1α-DIO-ChR2-EYFP
in SC and tested the behavior in the same setting as previously described. Mice expressing ChR2
in V1-recipient, superficial-layer SC neurons (Figure 10A), all demonstrated freezing response
following LED light activation (Figures 10B-C), which was not observed for sham mice (Figures
10B-C). None of the mice exhibited escape behavior following LED activation (Figure 10G).
These results thus reveal distinct functional roles of superficial- versus deep-layer SC neurons in
controlling two different types of defensive behaviors.
As shown previously, LP is a major axonal target of V1-recipient SC neurons (Figure 10A). In
addition, combining anterograde labeling with AAV-Cre virus from V1 and retrograde labeling
using CTB from LP, we found a high percentage of co-labelled cells, which confirmed that a large
fraction of V1-recipient SC neurons project to LP (Figure 11). We thus explored whether the
freezing behavior evoked by SC-sg activation might be mediated through this structure. To test
this, we first optogenetically activated the axon terminals of SC-sg neurons in LP (Figure 10D) in
freely behaving mice, which resulted in similar, though somewhat weaker, expression of freezing
28

(Figures 10E-F; V1-SC-LP), implicating LP’s role in driving SC-mediated freezing behavior. In
contrast, optically activating SC axon terminals in PBG did not produce any freezing behavior
(Figures 10E-F; V1-SC-PBG), suggesting that PBG does not play a role in this behavior. None of
these activations induced escape behavior (Figure 10G).

Figure 10: V1-Recipent SC Neurons Drive Freezing Behavior
(A) Paired injections labeling SC neurons receiving V1 input. LED illumination was applied to
cell bodies in SC (V1-SC). Scale bar, 500 mm. (B) Percentage of time spent freezing within the
time window of LED illumination (n = 5 mice for each group). Error bar = SD. ***p < 0.001, t
test. (C) Percentage of trials that induced freezing. ***p < 0.001, t test. (D) LED illumination was
29

applied to ChR2+ SC axon terminals in either LP (V1-SC-LP) or PBG (V1-SC-PBG). Right,
images showing ChR2 labeled SC axons in LP or PBG. Scale bar, 250 mm. (E) Percentage of time
spent freezing within the time window of LED illumination (n = 5 mice for each group). ***p <
0.001, t test. (F) Percentage of trials that induced freezing. ***p < 0.001, t test. (G) Percentage of
trials that induced escape behavior. (H) Paired injections labeling LP neurons that receive input
from SC. LED illumination was applied to LP (SC-LP). Right panel, images showing injection
sites in SC and LP. Scale bar, 500 mm. (I) Percentage of time spent freezing within the time
window of LED illumination (n = 6 mice for SC-LP; n = 5 mice for sham). Error bar = SD. **p <
0.05, t test. (J) Percentage of trials that induced freezing behavior. Error bar = SD. ***p < 0.001,
t test.

With terminal activation, nonspecific antidromic stimulation of collateral targets could occur,
resulting in activation of other undesired SC targets. Moreover, LED light from the optic fiber
might also activate ChR2-labeled axons passing through LP to other targets. Both scenarios would
confound interpretation of results. To overcome these limitations, we used AAV-Cre to express
ChR2 specifically in LP neurons that receive input from SC (Figure 10H) by injecting AAV1-Cre
into SC and AAV1-DIO-ChR2 into LP. Activation of these LP neurons in freely moving animals
induced freezing behaviors (Figure 10I-J). This result provides a strong evidence that LP acts as a
downstream mediator of the freezing response generated by the activation of superficial SC
neurons. Together, these results demonstrate that SC neurons receiving V1 input mediate freezing
responses through its downstream target LP.
30

Figure 11: Efficiency of anterograde labeling within the V1-SC-LP pathway
(A) Schematic diagram of injections. AAV1-hSyn-Cre was injected into V1 in Ai14 tdTomato
mice and fluorescently conjugated cholera toxin subunit b (CTB-488, green) was injected into LP
to retrogradely label cell bodies in SC that project to LP. Numerous LP-projecting cell bodies were
observed in the deepest part of superficial SC (middle and right panel, green) co-mingled with
Cre+/Tom+ V1-recipient SC neurons (red). At 40X magnification, extensive co-labeling was
observed. Scale bars, 250 μm, left panel, 25 μm, bottom right panel. (B) Quantification of CTB
co-labeled cells within the region of anterograde transneuronal labeling. An average of 66% of
CTB+ cells were co-labeled with tdTomato (n =3 mice).

2.3 Discussion
Taking advantage of the anterograde transsynaptic ability of the AAV-Cre virus, we accessed
discrete populations of neurons residing in superficial, deep, and lateral aspects of SC. This
enabled us to reveal that each subpopulation, defined by its specific cortical input, has a unique
divergent axonal targeting profile. We further showed that two of these subpopulations—those in
superficial layer SC receiving input from V1, and those in deep layers of SC receiving input from
A1—mediate distinct freezing and escape behaviors, respectively. Together, these results reveal
that SC functions as a control center for different types of defensive behaviors given distinct
sensory cortical inputs.
Here, we clearly showed the contribution of the projection from A1 or V1 to SC in mediating
either escape or freezing behavior. We noticed that different cortical areas target SC neurons in
separate locations, and the output patterns of these neurons also differ. Since the lateral SC is
reported to be involved in orienting and saccade control (Dean, Mitchell, and Redgrave 1988;
Dorris, Paré, and Munoz 1997; Hafed 2018; Hanes and Wurtz 2001; Kopec et al. 2015; Krauzlis,
Lovejoy, and Zénon 2013; C. Lee, Rohrer, and Sparks 1988; McPeek and Keller 2004; Sahibzada,
Dean, and Redgrave 1986; Stubblefield, Costabile, and Felsen 2013; White and Munoz 2012), the
31

SC neurons receiving M1 or S1 inputs could control the related behaviors through their specific
downstream pathways. Given that SC is implicated in a variety of behaviors (Basso and May 2017;
Dean, Redgrave, and Westby 1989; Fanselow 1994; Ito and Feldheim 2018; Shang et al. 2018;
White and Munoz 2012), it is quite possible that different SC subpopulations can drive other types
of behaviors depending its diverse cortical inputs.
Additionally, we examined the potential role of downstream structures in mediating the freezing
behavior observed from activation of SC neuron receiving V1 input and implicated LP as an
important downstream structure, which was also suggested by a previous report (Wei et al. 2015).
To fully confirm the contribution of LP in the freezing circuit, we went one step further and directly
activated SC-recipient cell bodies in LP and found similar freezing behavior. Employing an
anterograde transsynaptic labeling approach to this functional test allowed us to bypass potential
problems associated with optogenetic activation of axon terminals. These could include activation
of axon collaterals of SC neurons and activation of fibers of passage in LP. Instead, directly
activating the LP neurons that receive SC input provides more conclusive evidence implicating
this structure’s role in mediating the freezing response evoked by SC-sg activation.
Overall, our data suggest that SC serve as a control center for both defensive freezing and flight
depending on its input from V1 and A1. This provides a neural circuit mechanism for cortical
activity to directly control defensive behaviors. LP, as the downstream target of SC-sg for
controlling freezing behavior, could influence cortical activities in turn, given its widespread
projection to various cortices (Beltramo and Scanziani 2019; Bennett et al. 2019; Cappe et al.
2009; Hackett, Stepniewska, and Kaas 1998; Juavinett et al. 2019; Kaas and Lyon 2007; Nakamura
et al. 2015; Oh et al. 2014; Roth et al. 2016; S Shipp 2001; Stewart Shipp 2007; Stitt et al. 2018;
32

Wong et al. 2009; N. Zhou et al. 2018). The potential function role of LP in sensory processing
will be investigated in Chapter 4.

2.4 Material and methods
2.4.1 Animal preparation and stereotaxic surgery
All experimental procedures used in this study were approved by the Animal Care and Use
Committee at the University of Southern California. Male and female C57BL/6J and Ai14 (Cre-
dependent tdTomato reporter, RRID: IMSR_JAX:007914) mice (Jackson Laboratories) aged 2-6
months were used in this study. Mice were group housed in a light controlled (12 hr light: 12 hr
dark cycle) environment with ad libitum access to food and water.
Stereotaxic injection of viruses was carried out as we previously described (Ibrahim et al. 2016; F.
Liang et al. 2015; Xiong et al. 2015). Mice were anesthetized initially in an induction chamber
containing 5% isoflurane mixed with oxygen and then transferred to a stereotaxic frame equipped
with a heating pad. Anesthesia was maintained throughout the procedure using continuous delivery
of 2% isoflurane through a nose cone at a rate of 1.5 liters/min. The scalp was shaved, and a small
incision was made along the midline to expose the skull. After leveling the head relative to the
stereotaxic frame, injection coordinates based on the Allen Reference Atlas were used to mark the
location on the skull directly above the target area and a small hole (0.5mm diameter) was drilled.
Viruses were delivered through pulled glass micropipettes with a beveled tip (inner diameter of
tip: ~20 μm) using pressure injection via a micropump (World Precision Instruments). Total
injection volume was 50 to 100 nl, at 15 nl/min. Following injection, the micropipette was left in
place for approximately 5 mins to minimize diffusion of virus into the pipette path. After
33

withdrawing the micropipette, the scalp was sutured closed and animals were administered ketofen
(5mg/kg) to minimize inflammation and discomfort. Animals were recovered from anesthesia on
a heating pad and then returned to their home cage. For virus injection into the retina, animals were
anesthetized and positioned in a stereotaxic frame as described above. Using a pulled glass
micropipette with a beveled tip (40 μm inner tip diameter), 800 nl of virus was injected into the
left eye at a rate of 20 nl/min at a depth of 0.7 mm from the surface of the lateral most curvature
of the eye using a 40º angle of approach relative to the optic axis.

2.4.2 Injection of viruses for anterograde labeling
For mapping the axonal outputs of subpopulations of neurons in SC, AAV2/1-hSyn-Cre-WPRE-
hGH (UPenn Vector Core, 2.5 × 10
13
GC/ml) was injected into V1 (60 nl total volume; 3.9 mm
posterior and 2.6 mm lateral to bregma and 0.5 mm ventral from the cortical surface), contralateral
retina (800 nl total volume), A1 (3.1 mm posterior and 4.5 mm lateral to bregma and 0.75 mm
ventral from the cortical surface) or M1 (0.5 mm anterior and 1.5 mm lateral to bregma and 0.5
mm ventral from the cortical surface) of Ai14 mice (60 nl total volume). Following 2 to 7 days, a
second injection of AAV2/1-CAG-FLEX-eGFP-WPRE-bGH (UPenn vector core, 1.7 × 10
13

GC/ml, originally created by Allen Institute) was made into the ipsilateral SC (3.9 mm posterior
and 0.8 mm lateral to bregma and 1.5 mm ventral from the cortical surface; 60 nl total volume).
The spacing of the two injections over several days was enough for the clearance of any residual
AAV-Cre virus that may have spread across the pial surface to eliminate any local contamination
of the Cre-dependent virus injection site. Animals would recover for at least 4 weeks following
the second injection and be euthanized to examine the viral transport and transgene expression.
34

For behavioral testing, AAV2/1-hSyn-Cre-WPRE-hGH was injected into V1, A1, or SC, as
described above. Following 2 to 7 days, a second injection of AAV2/1-EF1a-DIO-hChR2-eYFP
(UPenn vector core, 1.6 × 10
13
GC/ml) was made into SC (for A1 and V1 injections of AAV-Cre)
or LP (for SC injections of AAV-Cre; 2.4 mm posterior and 1.6 mm lateral to bregma and 2.6 mm
ventral to the cortical surface). Animals were then prepared for optogenetic tests 4 weeks after the
second injection. To estimate the fraction of LP-projecting SC cells that are labeled following V1
injections of AAV-Cre, AAV2/1-hSyn-Cre-WPRE-hGH was injected into V1 of Ai14 tdTomato
mice as described above, and LP was injected with cholera toxin subunit B, Alexa 488 (CTB-488,
100 nl injection volume, 0.5% solution in PBS, ThermoFisher) using coordinates described above.
Animals were euthanized 4 weeks after injection.

2.4.3 Histology and imaging
Following desired post-injection survival time, animals were deeply anesthetized and
transcardially perfused with 4% paraformaldehyde. Brains were extracted and post-fixed for 24
hours at 4 ˚C in 4% paraformaldehyde and then sliced into 150 μm sections using a vibratome
(Leica, VT1000s). The sections were serially mounted onto glass slides. A fluorescent Nissl stain
was added (Neurotrace 640, ThermoFisher, N21483) to reveal cell body location and
cytoarchitectural information.
All images were generated using a confocal microscope (Olympus FluoView FV1000). Serial
sections across the whole brain were collected and examined. Regions with labeled cells were
imaged at 10X magnification across the depth of the tissue (150 μm thickness, 15 μm z-stack
interval) over all sections containing labeling. To quantify the percentage of LP-projecting SC
35

cells labeled with AAV-Cre injections in V1 (Figure S6), 40X magnification images were taken
across all sections of SC containing Tomato+ cells. All CTB+ and CTB+/Tomato+ cells were
quantified manually within the local region of Tomato+ labeling.

2.4.4 Optogenetic preparation and stimulation
To examine whether SC neuron subgroups mediate different behavioral responses, mice were
implanted with an optical fiber (200 μm diameter, Thorlabs) three weeks after secondary injection
of AAV2/1-EF1a-DIO-hChR2-eYFP in SC, following our previous study (Xiong et al. 2015).
Briefly, mice were anesthetized with isoflurane and mounted into a stereotaxic apparatus. A small
hole (~500 μm diameter) was drilled in the skull directly above the targeted region and the optical
fiber was lowered to the desired depth and fixed in place using dental cement. For activating ChR2-
expressing neurons in superficial SC, the fiber was positioned 3.9 mm posterior and 0.6 mm lateral
to bregma, and 0.9 mm ventral from the cortical surface. For activating SC neurons in deep layers,
the fiber was positioned as above, but lowered to a depth of 1.6 mm below the cortical surface.
For activating the SC axonal projection to LP, or LP neurons directly, the optical fiber was
positioned at 2.4 mm posterior and 1.6 mm lateral to bregma and 2.1 mm ventral to the cortical
surface. Animals would recover for 5-7 days prior to behavioral testing. During test sessions, the
implanted optical fiber was connected to a patch cord fiber secured with a plastic sleeve (Thorlabs,
200 μm Core, 0.22 NA (numerical aperture)). The latter fiber was equipped with an integrated
rotary joint (Thorlabs) and was supported from above to allow the animal to move freely in the
chamber without being hindered. To activate ChR2-expressing neurons or axons, optical
stimulation was delivered using a blue LED source (470 nm, 5 mW, Thorlabs) at a rate of 20 Hz
36

(20 ms pulse) for a duration of 5 seconds. Following testing sessions, animals were euthanized,
and each brain was sectioned and imaged to verify the specificity of ChR2 expression and location
of the implanted fiber.

2.4.5 Behavioral testing and analysis
Escape behaviors were tested in a box containing two chambers connected by a small opening, as
we previously described (Xiong et al. 2015). Mice (n = 7 mice for A1-SC, n = 5 mice for V1-SC)
could acclimate to one chamber for 10 minutes prior to behavioral testing. During this time, the
opening connecting the two chambers was blocked with a removable door. After 10 mins the door
was removed, and the mouse was free to explore the adjacent, novel chamber. Following complete
entry into the novel chamber, 5-s 20-Hz blue LED stimulation or 5-s 70 dB SPL white noise was
applied. The animal behavior was recorded with a camera mounted above the box. If the mouse
went back to the home chamber within 5 s of noise or LED stimulation, it was considered as a
successful escape trial. Test process was repeated for 4 to 6 times depending on the willingness of
the animal to enter the novel chamber after the door removal. Between trials, animals could rest
for at least 5 mins. Each animal was tested for 2 sessions separated by one day. Escape rate was
calculated as the fraction of total trials for each animal that evoked escape.
Freezing behavior was tested in a single or double chamber box with a camera on top recording
the whole process (n = 5 mice for V1-SC, n = 6 mice for SC-LP). After the animal explored the
single chamber or the novel chamber for 5 minutes, 5-s 20-Hz blue LED stimulation was applied.
Each session contained 4-6 trials, with an interval of at least 5 mins. Each animal was tested for 2
sessions separated by one day. The trial number depended on the activity level of the mouse, and
37

a session would end if the animal stayed unmoved in the corner for more than 1 min. If the animal
stayed motionless for more than 1.5 s after the onset of stimulation, it was considered as a
successful freezing response (Shang et al. 2015; Tovote et al. 2016; Wei et al. 2015; Wolff et al.
2014). Freezing time was quantified for each animal as the fraction of time spent freezing during
the optogenetic stimulation (total freezing time/ total LED stimulation time). Freezing rate was
calculated as the fraction of total trials that evoked freezing.
Control subjects for each behavioral test (sham, n = 5 each) received a sham injection of AAV2/1-
CAG-FLEX-GFP and were then implanted with an optical fiber and tested with the same
stimulation parameters. In this study, the optogenetic stimulation was unilateral, but was enough
for inducing defense behaviors such as flight and freezing, which is consistent with previous
studies (Comoli et al. 2012; Dean, Mitchell, and Redgrave 1988; F. Liang et al. 2015; Sahibzada,
Dean, and Redgrave 1986).

2.4.6 Statistical analysis
Samples were first determined to have normal distribution using the Shapiro-Wilk test. In the case
of a normal distribution, we performed paired t-test. Otherwise, we performed a non-parametric
test (Wilcoxon signed-rank test in this study). No statistical method was used to pre-determine
sample sizes, but our sample sizes were comparable to those reported in previous publications in
the field.

38

Chapter 3: Modulation of defensive behaviors by Zona
Incerta
3.1 Introduction
In natural environments, it is equally important for the animals to choose the right type of defensive
behavior as well as to modulate the level of specific defensive behavior. From the previous chapter,
we already know that the type of defensive behavior can be directly controlled by the cortical
inputs to SC. To achieve the modulation of defensive behaviors, certain inhibition circuits are
needed to properly adjust the activation level of the defense circuit. We searched major inhibitory
nucleus in mouse brain and found Zona incerta (ZI) as a candidate to perform this task based on
its cell type and connectivity pattern.
Zona incerta (ZI), first described more than a century ago by Auguste Forel (Forel 1877) as a “zone
of uncertainty”, is a major subthalamic inhibitory structure, functions of which remain largely
unclear. Recently, it has become a region of interest and studies have revealed some important
features of this region (K. Liu et al. 2017; J. Mitrofanis 2005; Moon and Park 2017; X. Zhang and
van den Pol 2017). First, ZI has extensive efferent and afferent projections in connection with
almost the entire neuroaxis, from cerebral cortices to the spinal cord (Barthó, Freund, and Acsády
2002; Nicolelis, Chapin, and Lin 1992; Shammah-Lagnado et al. 1985; Watson, Smith, and
Alloway 2015). This widespread connectivity may allow ZI to be involved in various physiological
functions, such as feeding, sleeping, sensory-motor integration, maintenance of posture and
locomotion, as well as regulation of pain (K. Liu et al. 2017; M. Liu et al. 2011; Moon and Park
2017; Perier et al. 2002; Urbain and Deschênes 2007; X. Zhang and van den Pol 2017). ZI is also
39

a clinically relevant structure since it has been implicated in alleviating symptoms of Parkinson’s
disease by deep brain stimulation (Blomstedt, Sandvik, and Tisch 2010; Khan et al. 2011; P Plaha,
Khan, and Gill 2008; Puneet Plaha et al. 2006). These findings raise an interesting hypothesis that
ZI can serve as an important hub to coordinate and modulate various behaviors, including
defensive behaviors.
In addition, ZI consists of heterogenous groups of cells, cytoarchitecture of which loosely divide
the structure into multiple sectors (Ma, Johnson, and Hoskins 1997; John Mitrofanis et al. 2004;
Nicolelis, Chapin, and Lin 1992). In rodents, four sectors (rostral, ventral, dorsal, caudal) of ZI
can be defined based on the neurochemical expression pattern(Kolmac and Mitrofanis 1999).
Given the various functional roles of ZI mentioned above, it would be interesting to investigate
whether these different sectors might contribute to different aspects of ZI function.
In this study, we find that GABAergic neurons in the rostral sector of ZI (ZIr) project to the
periaqueductal gray (PAG) in the midbrain. Many previous studies have suggested that PAG is an
important commanding center to produce various types of defensive behaviors (Bandler and
Shipley 1994; Tovote et al. 2016; Xiong et al. 2015). We thus test whether ZIr activity could
modulate these behaviors. Using optogenetic methods, we find that activation and suppression of
ZIr reduces and enhances both innate and learned defensive behaviors, respectively. Consistent
with these behavioral effects, we find that ZIr directly inhibits excitatory neurons in both the
dorsolateral and ventrolateral compartments of PAG. In addition, we provide evidence that ZIr is
involved in extinction of conditioned fear response via the mPFC-ZIr connection. Together, our
data suggest that ZIr plays a role in modulating defensive behaviors depending on specific
experience or contexts.
40

3.2 Results
3.2.1 Inhibitory projection from ZIr to PAG
Although ZI is known as an inhibitory nucleus (Chen and Kriegstein 2015; J. Mitrofanis 2005),
diverse cell types have been reported in this structure (Kolmac and Mitrofanis 1999; Oertel et al.
1982; Swanson, Sanchez-Watts, and Watts 2005). We first sought to understand the proportion of
GABAergic neurons within ZI. For this, brain slices from the GAD67-GFP mice, in which all
GABAergic cells are labeled with GFP (Tamamaki et al. 2003), were stained with NeuN to label
neuronal cell bodies. Consistent with previous reports (K. Liu et al. 2017; J. Mitrofanis 2005; X.
Zhang and van den Pol 2017), we found that the majority of neurons in ZI, in particular the rostral
part of ZI (ZIr), were GABAergic cells (Figure 12A). To understand where these neurons project
to, we performed focal injections of adeno-associated virus (AAV) encoding Cre-dependent GFP
into inhibitory neuron specific Cre lines. GAD2, a GABAergic cell marker, is expressed
throughout the ZI (Kolmac and Mitrofanis 1999), but in GAD2-Cre mice we limited our injections
in ZIr (Figure 13A, inset). Parvalbumin (PV) is most strongly expressed in the ventral sector of ZI
(ZIv) and less in its dorsal sector (ZId) (J. Mitrofanis 2005). Accordingly, we made injections in
the more caudal part of ZI in PV-Cre mice. Comparison of efferent projection patterns from the
injections in these two different Cre lines revealed a clear difference. For the ZIr injection in
GAD2-Cre mice, we found that profuse GFP-labeled axons in PAG, including both its dorsolateral
and ventrolateral compartments (dlPAG and vlPAG) (Figure 12B), consistent with previous
observations (Beitz 1989; Grofová, Ottersen, and Rinvik 1978). In contrast, there were few axonal
projections to PAG for the ZIv/ZId injection in PV-Cre mice (Figure 12C). Other than this, the ZIr
and ZIv/ZId injections revealed similar axonal labeling patterns in the midbrain (such as in SC,
RN and MRN), hindbrain (such as PRN) and thalamus (such as PO) (Figure 12C & Figure 13A-
41

B), while few projections were found in cortical regions or the amygdaloid complex (Figure 12D).
We also performed retrograde labeling experiment by injecting rabies virus in either dlPAG or
vlPAG and found that most PAG-projecting ZI neurons located in ZIr rather than other
compartments of ZI (Figure 13C-D). From these results, we have identified a distinct GABAergic
inhibitory projection from ZIr to PAG.
42

Figure 12. A distinct GABAergic projection from ZIr to PAG
(A) Left three panels, co-localization of GFP (green) signal and NeuN (red) staining in the ZIr of
a GAD67-GFP mouse. Scale: 200 µm. Right panel, percentage of GFP
+
neurons in the rostral
sector (ZIr) and more caudal part (ZIv/ZId) of ZI. (B) Injection of Cre-dependent GFP virus into
ZIr of GAD2-Cre mice. Confocal images show GFP expression in the injection site (upper middle;
scale: 500 µm) and in several target structures (upper right and lower; scale: 200 µm). Blue shows
Nissl staining. ZIr, zona incerta, rostral; PAG, periaqueductal grey; SC, superior colliculus; PRNr,
pontine reticular nucleus, rostral; RN, red nucleus. (C) Injection of Cre-dependent GFP virus into
ZIv/ZId of PV-Cre mice. Images show GFP expression in the injection site (upper middle; scale:
500 µm) and in several target structures (upper right and lower; scale: 200 µm). (D) Summary of
target areas of GAD2
+
ZIr (left panel) and PV
+
ZIv/ZId (right panel) neurons. PO, Posterior
complex of the thalamus; APN, Anterior pretectal nucleus; MRN, Midbrain reticular nucleus;
MARN, Magnocellular reticular nucleus.

Figure 13. Projection patterns of ZIr and ZIv/ZId
(A) Injection of Cre-dependent GFP virus in ZIr of an example GAD2-Cre mouse. Three panels
show GFP expression at different coronal levels. Inset of the left panel, an image at the level of
ZIv/ZId; note that florescence signals are mainly from projecting axons instead of labeled cell
bodies. Scale bars, 500 µm) . (B) Injection of Cre-dependent GFP virus in ZIv/ZId of an example
PV-Cre mouse. Three panels show GFP expression at different coronal levels. Scale bar, 500 µm.
(C) Injection of rabies-ΔG-eGFP into the ventrolateral PAG (vlPAG, upper panel) or dorsolateral
PAG (dlPAG, lower panel) of wild-type mice. Left, GFP-labeled neurons around the injection site.
43

Right, retrogradely labeled neurons in ZIr. Blue shows Nissl staining. Scale: 200 µm. (D)
Comparison of relative numbers of labeled neurons (expressed as the percentage of total labeled
neurons in ZI) in different ZI subdivisions between dlPAG and vlPAG injections (n = 3 mice for
each). Abbreviations: LD, Lateral dorsal nucleus of the thalamus; PO, Posterior complex of the
thalamus; RT, Reticular nucleus of the thalamus; VM, Ventral medial nucleus of the thalamus;
VAL, Ventral anterior-lateral complex of the thalamus; ZIr, Zona incerta, rostral; LH, Lateral
habenula; LP, Lateral posterior nucleus of the thalamus; LG, Lateral geniculate complex; PF,
Parafascicular nucleus; VPM, Ventral posteromedial nucleus of the thalamus; VPL, Ventral
posterolateral nucleus of the thalamus; VPMpc, Ventral posteromedial nucleus of the thalamus,
parvocellular part; ZId, Zona incerta, dorsal; ZIv, Zona incerta, ventral; SC-ig, Superior colliculus,
intermediate gray; SC-dg, Superior colliculus, deep gray; SC-sg, Superior colliculus, superficial
gray; PAG, Periaqueductal gray; MRN, Midbrain reticular nucleus; APN, Anterior pretectal
nucleus; RN, Red nucleus; VTA, Ventral tegmental area; SN, Substantia nigra; TH, Thalamus;
NOT, Nucleus of the optic tract; PRNr, Pontine reticular nucleus, rostral; IC, Inferior colliculus.
Solid symbol represents mean ± s.e.m. for D.

3.2.2 ZIr modulates innate flight response

Figure 14. ZIr bi-directionally modulates noise-induced flight response
(A) Upper, illustration of the experimental paradigm. Lower, image showing ChR2 expression
(green) and placement of two optic fibers over ZIr. Scale: 1000 µm. (B) Normalized average speed
induced by noise sound for an example ChR2-expressing animal with (blue) and without (grey)
blue LED stimulation. Two dash lines mark the duration of noise. (C) Summary of noise-induced
peak speed without (OFF) and with (ON) activation of ZIr neurons. ***p = 0.001, two-sided paired
t-test, n = 11 animals. (D) Summary of total travel distance during the 5-s noise stimulation without
and with activation of ZIr neurons. ***p = 0.003, two-sided Wilcoxon signed-rank test, n = 11.
(E) Upper, illustration of the experimental paradigm. Lower, image showing ArchT expression
44

(green) and placement of two optic fibers over ZIr. Scale: 1000 µm. (F) Normalized average speed
induced by noise for an example ArchT-expressing animal with (green) and without (grey) green
LED stimulation. (G) Summary of noise-induced peak speed without and with suppressing ZIr
neurons. *p = 0.019, two-sided paired t-test, n = 9 animals. (H) Summary of total travel distance
without and with suppressing ZIr. *p = 0.032, two-sided paired t-test, n = 9 animals. Solid symbol
represents mean ± s.d. for all panels.

PAG is known to be a commanding center to produce various types of defense behaviors (Bandler
and Shipley 1994; Tovote et al. 2016; Xiong et al. 2015), and previous studies have suggested that
dlPAG and vlPAG drive defensive flight and freezing behaviors respectively (Tovote et al. 2016;
Xiong et al. 2015). Since both dlPAG and vlPAG receive GABAergic projections from ZIr, we
speculated that ZIr might be able to modulate both flight and freezing types of PAG-mediated
behaviors (Vianna and Brandão 2003). Previously, we have reported that loud noise can trigger
flight response in naïve freely moving or head-fixed mice, as manifested by a robust increase of
their running speed, and that this behavior is mediated by dlPAG (Xiong et al. 2015). We first
tested whether ZIr could modulate this innate flight behavior in head-fixed mice. To this end, we
optogenetically targeted GABAergic neurons in ZIr by focal injections of AAV encoding Cre-
dependent ChR2 for activation or ArchT for suppression (Boyden et al. 2005; Chow et al. 2010)
in GAD2-Cre mice. Blue or green LED light was delivered bilaterally through implanted optic
fibers (Figure 14A, D). LED light was applied during noise stimulation (80 dB sound pressure
level) in trials interleaved with control LED off trials. We found that both the peak running speed
and travel distance of noise-induced running were significantly reduced in trials when ZIr neurons
were activated as compared with control trials (Figure 14B-D). Opposite effects were observed
when ZIr neurons were suppressed: the peak speed and travel distance both increased (Figure 14F-
H).

45

Figure 15. GFP control and effects of optogenetic manipulation on flight latency
(A) Normalized average speed traces for an example ZIr GFP-expressing animal with and without
LED stimulation. Two dash lines mark the duration of noise presentation. Solid bar indicates the
duration of LED stimulation (20-ms pulses at 20 Hz). (B) Summary of noise-induced peak speed
with (ON) and without (OFF) LED stimulation for control animals injected with AAV-GFP in ZIr.
N.S., not significant, p = 0.196, two-sided paired t-test, n = 6 animals. (C) Summary of total travel
distance during the 5-s noise stimulation with and without LED stimulation for GFP control
animals. N.S., not significant, p = 0.687, two-sided Wilcoxon-signed rank test, n = 6 animals. (D-
E) Summary of onset latency of noise induced flight response for ZIr ChR2 and ArchT expressing
groups with and without LED stimulation in ZIr. N.S., not significant, p(f) = 0.157, p(g) = 0.683,
two-sided paired t-test, n = 10 animals for ChR2; 9 for AchT. (F) Onset latency of noise induced
flight response for the experimental group in which ZIr-PAG axonal terminals were stimulated by
LED. p = 0.317, two-sided Wilcoxon-signed rank test, n = 8 animals. Solid symbol represents
mean ± s.e.m. for all panels.

Changes in running were not observed in GFP-expressing control animals (Figure 15A-C). Despite
the changes in behavioral amplitude, the onset latency of the flight response was not affected by
either type of optogenetic manipulation (Figure 15D-E). The LED stimulation alone, either with
blue or green light, did not change the baseline locomotion in the absence of noise stimulation, or
46

balance beam performance (Figure 16). Together, these results demonstrate that ZIr activity can
bidirectionally modulate the magnitude of innate flight response.

Figure 16. ZIr do not affect baseline locomotion and balance beam performance
(A) Normalized travel distance in the open field locomotion test with and without LED stimulation.
N.S., non-significant, p > 0.05, Paired t-test, Black, ChR2-expressing group (n = 5 animals); gray,
ArchT-expressing group (n = 6). (B-D) Summary of average running speed in the absence of noise
with and without LED stimulation for GFP, ChR2 and ArchT injected group respectively. N.S.,
not significant, p(b) = 0.789, p(c) = 0.556, p(d) = 0.271, two-sided paired t-test, n = 9 animals for
GFP, 7 for ChR2, 6 for ArchT group respectively. (E) Experimental paradigm for balance beam
test. A mouse needs to move cross the beam to reach a larger safer platform. (E-H) Summary of
average speed of crossing the balance beam with and without LED stimulation for different
experimental groups. N.S., not significant, p(f) = 0.777, p(g) = 0.322, p(h) = 0.846, two-sided
paired t-test, n = 11 animals for ZIr ChR2, 9 for ZIr AchT, and 8 for ZIr-PAG ChR2 groups
respectively. Solid symbol represents mean ± s.e.m. for all panels.
47

As previously reported (Xiong et al. 2015), the magnitude of innate flight response increases with
increasing noise intensities. In a separate cohort of animals, we examined how ZIr activity affected
running speeds under different noise intensities (Figure 17A, C, black). We found that optical
manipulations of ZIr activity using ChR2 or ArchT reduced or enhanced flight responses across
effective noise intensities (Figure 17A, C, color). Importantly, the peak running speed in the LED
on condition was linearly correlated with that in the LED off condition in each examined animal
(Figure 17B, D). These data demonstrate that the level of flight response under different sound
intensities is divisively modulated by ZIr activity. Specifically, under our current optical
stimulation condition (5 mW on one side), activation of ZIr reduced the flight speeds across
different noise intensities by ~20%, whereas suppression of ZIr increased the flight speeds by
~20%. Our data thus suggest that ZIr can exert a gain control function in modulating innate flight
response.
48

Figure 17. ZIr activity exerts a gain modulation
(A) Average peak speeds induced by noise of different intensities without (black) and with (blue)
activation of ZIr neurons for an example animal. Inset, illustration of experimental setup. (B) Left,
peak speed in the LED on condition versus that in the LED off condition plotted for 3 animals.
Each color represents one animal. Colored lines represent the corresponding linear regression line.
Black dash line is the unity line. Right, correlation coefficient calculated for each animal (upper)
and the slope for the best-fit linear regression line (lower). (C) Average peak speeds induced by
noise of different intensities without (black) and with (green) suppression of ZIr neurons for an
example animal. (D) Speed with ZIr suppression versus that without suppression plotted for 4
animals.

3.2.3 ZIr modulates conditioned freezing response
Since vlPAG has been implicated in mediating conditioned freezing response (Tovote et al. 2016),
we reasoned that ZIr might be able to modulate this behavior as well. To test this possibility, we
applied similar optogenetic approaches during the classical fear conditioning test. The animal was
first exposed to a conditioned stimulus (CS, 20-s 5kHz tone) for 5 times in a test box during day 1
(Figure 18A). In day 2, it was exposed to the CS paired with a foot shock (1s) for 5 times in a
conditioning chamber. In day 3, the CS was applied without a foot shock for 6 trials in the test box
in order to measure cued conditioned freezing response. In half of these trials, blue or green LED
stimulation was applied during the CS presentation in a randomized order. We found that freezing
time during CS presentation was significantly reduced in trials when ZIr neurons were activated
as compared with control trials (Figure 18B). On the other hand, freezing time was increased when
ZIr neurons were suppressed (Figure 18C). These data demonstrate that increasing or decreasing
ZIr activity can suppress or enhance conditioned freezing response, respectively. Therefore, ZIr
can bidirectionally modulate the level of conditioned freezing behavior as well.
49

Figure 18. ZIr bidirectionally modulates conditioned freezing response
(A) Experimental paradigm for fear conditioning and testing of learned freezing response. (B)
Percentage of time freezing during presentations of CS alone without and with activation of ZIr.
***p < 0.001, two-sided paired t-test, n = 10 animals. (C) Percentage of time freezing during
presentations of CS alone without and with suppressing ZIr. **p = 0.005, two-sided paired t-test,
n = 7 animals. (D) Percentage of time freezing at each presentation of CS during extinction training
with (green) and without (black) suppressing ZIr. Bar = s.e., n = 5 for ArchT and 7 for control.
(E) Modulation index calculated for each animal (as the ratio of freezing time between the last and
first trial) for the GFP control and ArchT suppression group. ***p = 0.004, two-sided unpaired t-
test, n = 5 for ArchT and 7 for control. Solid symbol represents mean ± s.d. for all panels.

3.2.4 ZIr is engaged in extinction of conditioned freezing
It is known that conditioned freezing response naturally diminishes with repeated presentations of
CS alone, a process termed “extinction” (Herry et al. 2010; Myers and Davis 2002). In other words,
learned fear response is naturally modulated in an experience-dependent manner. Since ZIr
activation could reduce the freezing response, we wondered whether ZIr might be involved in its
extinction. To test this possibility, we expressed ArchT in ZIr neurons using GAD2-Cre mice. We
50

conditioned the mice similarly with 5 pairings of CS and foot shock. On the test day, the animal
was presented with CS alone for 10 trials in order to obtain an extinction curve(Peters et al. 2010;
Sierra-Mercado, Padilla-Coreano, and Quirk 2011). In GFP-expressing control mice, freezing was
significantly reduced after the first 5 trials (Figure 18D, black), consistent with previous
observations(Peters et al. 2010; Sierra-Mercado, Padilla-Coreano, and Quirk 2011). For the
experimental group in which green LED light was paired with each CS presentation to suppress
ZIr activity, the reduction of freezing over repeated CS presentations was much slower (Figure
18D, green), as shown by the quantification of a modulation index (Figure 18E), indicating that
extinction was impaired. In addition, one day after the normal extinction training, suppressing ZIr
activity also impaired the expression of extinction retrieval (Monfils et al. 2009; Quirk and Mueller
2008) (Figure 19). These data further demonstrate that ZIr activity regulates the expression of
conditioned fear response.

Figure 19. Effect of inactivation of ZIr on extinction retrieval
Percentage of time freezing during 10 presentations of CS on the day following conditioning (i.e.
extinction training) and during 5 presentations of CS (i.e. extinction retrieval) on the day following
extinction training. Arrows point to the trials in which LED was applied. N = 3 animals.

51

We next monitored spiking activity in ZIr by performing in vivo single-unit recordings (Figure
20A). The animal was either conditioned similarly or exposed to CS without foot shocks (control).
Spikes of ZIr neurons were recorded during presentations of CS either on the conditioning day, or
in the following day when conditioned freezing exhibited extinction over repeated CS
presentations. ZIr activity did not change significantly during conditioning (Figure 20B-C).
Interestingly, in the following day, ZIr activity was found to increase progressively with repeated
CS presentations in the conditioned but not control mice (Figure 20D-F). This observed increase
of ZIr activity is consistent with the above result that enhancing ZIr activity led to reduced freezing.
These results suggest that ZIr may be naturally engaged in extinction.

Figure 20. ZIr activities during fear conditioning and fear extinction
(A) Left, experimental setup for extracellular recording. Right, principal component analysis for
an example recording session. The graph shows the projection of principal components to the
PC1-PC3 plane. Colors mark different clusters of the spikes. Dashed circles denote the boundary
of well separated spikes. (B) Raster plot of spikes for an example ZIr unit during CS presentation
(19s, just before the foot shock). Scale: 2 s. (C) Average normalized spike rate of ZIr units during
52

fear induction, n = 12 units from 4 animals. (D) Raster plot of spikes for example units during the
first and last 3 presentations of CS alone in an example conditioned and non-conditioned control
mouse respectively. Scale: 0.5s. Inset, superimposed 50 individual spikes and their average (black)
for the corresponding unit. Scale: 20 µV, 0.5 ms. (E) Average normalized spike rate of ZIr units
at each CS presentation in conditioned (black, n = 11 units from 6 animals) and control (grey, n
=18 units from 6 animals) mice. ***p(trial5) < 0.001, *p(trial6) = 0.031, ***p(trial8) = 0.003,
***p(trial9) < 0.001, **p(trial10) = 0.006, two-sided Mann-Whitney U test. (F) Modulation Index
calculated for each unit for the conditioned and control groups. ***p < 0.001, two-sided unpaired
t-test. Solid symbol represents mean ± s.e.m. for all panels.

Figure 21. mPFC contributes to ZIr activity increase during extinction
(A) Injection of rAAV2-retro-syn-Cre into ZIr of Ai14 mouse. Left, tdTomato expression around
the injection site. Right, retrogradely labeled presynaptic neurons (red) in mPFC. Blue shows Nissl
staining. ILA and PL subregions are marked. Scale: 500 µm. (B) Injection of GFP virus into mPFC
of wild-type mice. Left, GFP expression at the injection site. Middle, GFP-labeled axons in ZIr.
Right, no GFP-labeled axons in the more caudal part of ZI. Scale: 500 µm. (C) Left, experimental
paradigm. Right, average EPSC amplitudes of 15 out of 25 recorded ZIr neurons in response to
LED activation of mPFC axons. Recording was made in the presence of TTX and 4-AP. Top inset,
average EPSC trace for an example ZIr neuron. Arrow points to the onset of LED. Scale: 40 pA,
10 ms. (D) Average normalized spike rate of ZIr units during extinction training for saline control
(black, n = 21 units from 7 animals) and mPFC muscimol injected (green, n = 16 units from 6
animals) mice. ***p(trial 8) < 0.001, **p(trial9) = 0.002, ***p(trial10) < 0.001, two-sided Mann-
Whitney U test. (E) Modulation index calculate for each unit in the saline and muscimol group.
***p < 0.001, two-sided unpaired t-test. Solid symbol represents mean ± s.e.m. for all panels.

53

Previously the medial prefrontal cortex (mPFC) has been implicated in extinction (Laurent and
Westbrook 2009; M. R. R. Milad and Quirk 2002), and its projection to ZI has been reported
(Hurley et al. 1991; John Mitrofanis and Mikuletic 1999). We wondered whether mPFC might
play a role in driving the activity increase in ZIr during extinction. We first performed retrograde
labeling by injecting rAAV2-retro-syn-Cre into ZIr of Ai14 mice (Tervo et al. 2016) (Figure 21A),
and found robust labeling in mPFC subregions such as the infralimbic area (ILA), prelimbic area
(PL), as well as part of the anterior cingulate area (ACA). To confirm the functional connectivity,
we injected AAV-ChR2-eYFP into mPFC, mainly in ILA and PL subregions (Figure 21B, left).
Dense axonal projections were found in ZIr, but not in the more caudal part of ZI (Figure 21B,
middle and right). In brain slices, we performed whole-cell recordings and found that a majority
of recorded ZIr cells exhibited monosynaptic excitatory responses to the optogenetic activation of
mPFC axons (Figure 21C), confirming the excitatory innervation from mPFC to ZIr. We next
injected muscimol into mPFC to silence its local spiking activity (Figure 22A). This prevented the
increase of ZIr activity during extinction, whereas the increase was still observed in saline-injected
control animals (Figure 21D-E). In addition, silencing mPFC with muscimol impaired extinction
of freezing response behaviorally (Figure 22B-C). Our data thus suggest that ZIr could mediate
freezing reduction by direct mPFC inputs during extinction training.
54

Figure 22. Silencing of mPFC impairs fear extinction
(A) Left, image showing site for fluorescent muscimol injection in mPFC. Scale bar, 500 µm.
Right, multiunit recording in mPFC before and after the muscimol injection, n = 6 animals. (B)
Percentage of time freezing during 10 presentations of CS for control and mPFC muscimol injected
animals, n = 4 animals for muscimol and 5 animals for control group. (C) Modulation index
quantified for muscimol and control group. ***, p < 0.001, two-sided unpaired t-test, n = 4 animals
for muscimol and 5 animals for control group. Solid symbol represents mean ± s.e.m. for B, C.

3.2.5 Inhibitory effects of ZIr input to PAG
Figure 23. GABAergic output of ZIr suppresses PAG activity
(A) Left three panels, co-localization of GFP (green) with CTB (red) retrogradely transported
from PAG in the ZIr of a GAD67-GFP mouse. Scale: 200 µm. Right panel, percentage of GAD-
GFP
+
neurons in total CTB labeled neurons in ZIr. (B) Left, slice recording paradigm. Excitatory
55

neurons are labeled by red color in vGLTU2-Cre::Ai14 mice. Right, average EPSC (recorded at -
70mV) and IPSC (recorded at 0mV) traces of a vGLUT2
+
neuron in PAG. The IPSC was
blocked by Gabazine (bottom). Blue arrow points to the onset of LED. Scale: 50 pA, 50 ms. (C)
Average amplitudes of LED-evoked IPSCs and EPSCs in excitatory (left) and inhibitory (right)
PAG neuron populations. (D) Left, average local field potential recorded in PAG in response to
noise without (red) and with (blue) LED stimulation in a ZIr ChR2 expressing mouse. Right,
summary of peak amplitude of LFP recorded in PAG without (OFF) and with (ON) activation of
ZIr. ***p = 0.004, two-sided paired t-test, n = 6 sites from 2 animals. (E) Summary of noise-
evoked spike rate in PAG without and with ZIr activation. *p = 0.028, two-sided Wilcoxon
signed-rank test, n = 6 sites from 2 animals. (F) Summary of spontaneous firing rate in PAG
without and with ZIr activation. *p = 0.03, two-sided paired t-test, n = 8 sites from 2 animals.
Solid symbol represents mean ± s.d. for D, E, F, and represents mean ± s.e.m. for A, C.

To further elucidate the nature of connectivity between ZIr and PAG, we injected a retrograde dye,
CTB, into PAG of GAD67-GFP mice. Nearly all retrogradely labeled neurons in ZIr colocalized
with GFP (Figure 23A), confirming that PAG-projecting ZIr neurons are GABAergic. In slice
preparations, we made whole-cell recordings from PAG neurons, with TTX and 4AP present in
the extracellular solution to block polysynaptic responses. To label excitatory or inhibitory PAG
neurons, we used vGLUT2-Cre or GAD2-Cre crossed with Ai14 mice, respectively. We found
that only excitatory but not inhibitory PAG neurons received monosynaptic inhibitory input from
ZIr, which could be blocked by a GABAA receptor blocker, Gabazine (Figure 23B). Similar results
were obtained for recording in both dlPAG and vlPAG (Figure 24). On the other hand, none of the
recorded neurons exhibited monosynaptic excitatory responses (Figure 23B-C), as expected from
the GABAergic cell type of ZIr neurons projecting to PAG. We further carried out multiunit and
local field potential (LFP) recordings in PAG (Xiong et al. 2015) in vivo, and found that the peak
amplitude of LFP and multiunit spike rate evoked by noise sound were reduced by optogenetic
activation of ZIr (Figure 23D-E). In addition, the spontaneous spike rate in PAG was also reduced
by the activation of ZIr (Figure 23F). All these data demonstrate that the ZIr to PAG projection is
56

inhibitory and can directly suppress PAG activity, which is consistent with the above results
showing a suppressive effect of ZIr on PAG-mediated defensive behaviors.

Figure 24. ZIr innervates excitatory but not inhibitory neurons in PAG
(A) Left, experimental paradigm: AAV encoding Cre-dependent ChR2 injected into ZIr of GAD2-
Cre::Ai14 mice, and whole-cell recording made from excitatory (white, unlabeled) or inhibitory
(red, labeled by tdTomato) neurons in PAG. Blue LED was applied to stimulate ZIr axons in PAG.
Right, LED evoked excitatory (EPSC) and inhibitory postsynaptic currents (IPSC) in example
excitatory and inhibitory neurons in dlPAG and vlPAG respectively. Arrow points to the onset of
LED. Scale: 40 pA, 100 ms. (B) Average amplitudes of LED-evoked IPSCs in excitatory (black)
and inhibitory (grey) PAG neurons. Numbers in parentheses indicate the number of neurons
exhibiting evoked responses out of the total number of recorded cells. (C) No EPSC was observed
in any of the recorded cells. Solid symbol represents mean ± s.d. for B.

As reported previously (Xiong et al. 2015), optogenetic activation of dlPAG neurons directly
induced flight response. To further confirm whether ZIr regulates the magnitude of defense
behaviors by directly modulating PAG activity, we optically activated ChR2-expressing ZIr axons
in PAG bilaterally while silencing ZIr neuronal cell bodies with muscimol to prevent antidromic
spikes (Figure 25A). Results similar as activating ZIr cell bodies were obtained: the peak speed
and travel distance of noise-induced flight response were significantly reduced as compared with
control trials (Figure 25B-D), and the conditioned fear response was also reduced (Figure 25E).
These results indicate that the inhibitory ZIr-PAG projection can mediate the modulatory effects
of ZIr on both types of defensive behaviors.
57

Figure 25. ZIr-PAG projection mediates the modulatory effects on defensive behaviors
(A) Experimental paradigm. LED was applied to PAG, and muscimol was injected into ZIr as to
prevent antidromic stimulation. (B) Normalized average speed responding to noise without (grey)
and with (blue) LED stimulation plotted for an example animal. (C) Summary of noise-induced
peak speed without and with activation of ZIr-PAG projections. ***p =0.003, two-sided paired t-
test, n = 8 animals. (D) Summary of noise-induced travel distance without and with activation of
ZIr-PAG projections. *p = 0.021, two-sided paired t-test, n = 8 animals. (E) Percentage of time
freezing during presentations of CS alone without and with activation of ZIr-PAG projections in
conditioned mice. ***p < 0.001, two-sided paired t-test, n = 10 animals. Solid symbol represents
mean ± s.d. for all panel.

3.3 Discussion
In this study, by using combined anatomical tracing, electrophysiological recording and
bidirectional optogenetic manipulations, we demonstrate that the rostral part of ZI can
bidirectionally modulate defensive behaviors, including both innate flight and learned freezing.
By testing flight responses under different sound intensities, we further demonstrate that the
modulation by ZIr is likely through a gain control mechanism. Moreover, for a natural form of
58

experience-dependent modulation of defense behavior, fear extinction, we demonstrate that ZIr
activity increase is correlated with the expression of extinction, i.e. reduced freezing over repeated
presentations of CS alone. This finding further supports the engagement of ZIr in modulating the
level of defensive behaviors based on experience or contexts. Finally, we provide evidence that
the modulatory effects of ZIr are achieved through its inhibitory projection to PAG. Our results
together have unveiled a novel function of ZIr in regulating defensive behaviors.
3.3.1 Function roles of different subpopulation of ZI neurons
ZI is a relatively large inhibitory subthalamic nucleus containing multiple sectors and cell types
(Kolmac and Mitrofanis 1999; Ma, Johnson, and Hoskins 1997; John Mitrofanis et al. 2004;
Nicolelis, Chapin, and Lin 1992). Previous studies have also demonstrated that ZI receives a rich
variety of inputs from many brain regions, including multisensory inputs and neuromodulatory
inputs that carry information about brain states (J. Mitrofanis 2005; Paré et al. 1988; Trageser
2004; Watson, Smith, and Alloway 2015). This cytoarchitectural and connectivity complexity may
be the basis for the ZI to be involved in various aspects of animals’ physiological and behavioral
functions. For example, clinically, ZI has been implicated in alleviating symptoms of Parkinson’s
disease as well as essential tremor by deep brain stimulation (Blomstedt, Sandvik, and Tisch 2010;
Khan et al. 2011; P Plaha, Khan, and Gill 2008; Puneet Plaha et al. 2006). More recently, it has
been reported that stimulation of ZI GABAergic neurons evokes binge-like eating, resulting in
body weight gain (X. Zhang and van den Pol 2017). A subpopulation of GABAergic neurons in
the ventral ZI, the Lhx6-positive neuron, has also been shown to promote sleep (K. Liu et al. 2017).
In the current study, we have discovered a new functional role of ZIr in modulating defense
behaviors. All these findings help us to better understand the overall function of ZI.
59

PAG is a commanding center responsible for initiating various defense behaviors (Bandler and
Shipley 1994; Tovote et al. 2016; Xiong et al. 2015). Our data demonstrate that ZIr sends
GABAergic projections to both dlPAG and vlPAG. This allows ZIr to be able to modulate various
types of defense behaviors. It is worth noting that other brain regions containing GABAergic
neurons such as the hypothalamus also project to both compartments of PAG (Allen Brain Atlas).
These areas may also be able to exert similar regulatory functions on defense behaviors. Besides
PAG, ZI projects broadly to other motor-related midbrain and hindbrain areas such as the superior
colliculus (SC), midbrain reticular nucleus (MRN), red nucleus (RN) and pontine reticular nucleus
(PRN) (Bernays et al. 1988; Ricardo 1981; Watson, Smith, and Alloway 2015). The GABAergic
projections of ZI to these areas may contribute to the regulation of locomotor activities and posture
as well. It would be interesting to distinguish different cell types within ZI sectors and examine
whether they have distinct projection patterns. Such information may provide insights into unique
functional roles of subpopulations of ZI neurons.

3.3.2 Top-down regulation of behaviors depending on contexts
Our anatomical results and data from online resources for connectome (Allen Brain Atlas) indicate
that ZI receives input from mPFC subregions including ACA, PL and ILA. It is thus reasonable to
postulate that by bridging higher cortical areas and midbrain/hindbrain nuclei, ZI may serve as a
regulatory hub to mediate the top-down regulation of various motor behaviors, in addition to that
mediated by direct projections from cortical areas (Beitz 1989; Hurley et al. 1991). When the
environment becomes safer or when tangible threats are removed, signals may be transmitted to
60

ZI to reduce defense and promote eating or sleeping. Whether some other inhibitory nuclei can
serve a similar function is worth further explorations.
We demonstrate that ZIr activity increase correlates with a reduction of conditioned fear response
during extinction, which supports the notion that ZIr is naturally engaged in defense modulation.
We have also identified mPFC as a potential input source that drives the engagement of ZIr. There
have been a vast number of studies on the circuits underlying fear extinction (Monfils et al. 2009;
Quirk and Mueller 2008). Based on the observed effect of ZIr suppression on fear extinction, we
postulate that the identified ZIr-PAG pathway could serve in parallel with the classic amygdala-
PAG pathway (Barad, Gean, and Lutz 2006; Paré et al. 1988) to reduce fear expression during
extinction. While it is generally accepted from previous studies that PL and ILA have opposite
effects on fear expression (Laurent and Westbrook 2009; Sierra-Mercado, Padilla-Coreano, and
Quirk 2011), it would be interesting to further investigate how each subregion of mPFC exerts the
top-down influence to ZIr during extinction.

3.3.3 Gain modulation of ZIr on defensive behaviors
It is an intriguing finding that the ZIr’s effect on defensive behavior is through a gain modulation
mechanism, i.e. ZIr modulates the amplitude of behavioral output. Such gain modulation may have
a great advantage. In a natural environment, danger signals may vary greatly in strength, from
mildly intimidating to life threatening, and there could be hidden unperceived threats. Maintaining
certain levels of defense is protective and beneficial, while prolonged or intensified defense may
cause a failure of the animal to adapt to the changing environment timely (Eilam, Izhar, and Mort
2011; Fanselow 1994; LeDoux 2012). A gain control mechanism allows adjustment of the
61

magnitude of ongoing defense behavior in accordance with the current danger level. In our
experiments, the observed modulation of behavioral response was moderate. However, it is
unknown what the upper limit of ZIr modulation could be, since the activation or suppression of
ZIr neurons in our experiments was unlikely complete. It remains unclear how the gain modulation
function of ZIr is achieved. Our data have implied that the level of spiking activity in its
downstream structure, PAG, correlates with the magnitude of defense behavior. Does ZIr
activation suppress spike rates of PAG neurons proportionally under different danger levels? This
question awaits further investigations in the future.
In summary, we have discovered a role of ZI, a major inhibitory subthalamic nucleus, in
modulating defense behaviors. We propose that ZI can serve as a bridge between higher cortical
areas and midbrain/hindbrain nuclei for the top-down regulation of motor behaviors. This
inhibitory nucleus mediated gain modulation may be a common strategy implemented by the
mammalian brain circuits.

3.4 Material and methods
3.4.1 Animal model and stereotaxic injection
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 (GAD2-Cre, vGLUT2-Cre, PV-Cre and Ai14) 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.
62

Viral injections were carried out similarly as described in Chapter 2. Stereotaxic coordinates were
based on the Allen Reference Atlas (www.brain-map.org). For anterograde tracing, AAV1-CAG-
FLEX-eGFP-WPRE-bGH (UPenn Vector Core, 1.7×10
13
GC/ml) was injected into the rostral
sector of ZI (30nl total volume; AP -1.2 mm, ML +1.6 mm, DV -4.4 mm) of GAD2-Cre mice.
The same virus was also injected into ZIv/ZId (30nl total volume; AP -1.5 mm, ML +2.2 mm, DV
-4.2 mm) of PV-Cre mice. Animals were euthanized 3-4 weeks following the injection for
examination. For retrograde tracing, EnvA G-deleted Rabies-eGFP (Addgene# 32635) was
injected into dlPAG (30nl total volume; AP -4.4 mm, ML +0.6 mm, DV -2.4 mm) and vlPAG
(30nl total volume; AP -4.4 mm, ML +0.5 mm, DV -2.7 mm) of wild-type C57BL/6 mice,
respectively. CTB-584 was injected into PAG (30nl total volume; AP -4.4 mm, ML +0.5 mm, DV
-2.5 mm). rAAV2-retro-syn-Cre (Vigene, custom order) was injected into ZIr (30nl total volume;
AP -1.2 mm, ML +1.6 mm, DV -4.4 mm) of Ai14 mice. Animals were sacrificed one week after
the injection.
For behavioral and electrophysiological assessments, 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), or AAV1-CAG-FLEX-eGFP-WPRE-bGH (UPenn Vector Core, 1.7×10
13
GC/ml, as
control) was injected bilaterally into ZIr (100 nl for each site) of GAD2-Cre::Ai14 mice. For the
vGLUT2-Cre::Ai14 mice, AAV1-Syn-Cre mixed with AAV2/1-pEF1α-DIO-hChR2-eYFP was
injected bilaterally into ZIr (100 nl for each site). AAV1-CamKII-hChR2(E123A)-eYFP-WPRE-
hGh (UPenn Vector Core, 1.6×10
13
GC/ml) was injected into dlPAG or mPFC (30nl total volume;
AP +1.6 mm, ML -0.4 mm, DV -1.9 mm) of wild-type C57BL/6 mice. Viruses were expressed for
at least three weeks.

63

3.4.2 Histology, imaging and quantification
Brain slices for imaging were prepared similarly as described in Chapter 2. All images were
acquired using a confocal microscope (Olympus FluoView FV1000). For NeuN staining, sections
were blocked with 5% normal goat serum/0.05% Triton X-100 for 2 h at room temperature, then
incubated with mouse IgG monoclonal anti-NeuN antibody (MAB377; Millipore, Billerica, MA)
at 1:2000 overnight at 4 °C. Sections were washed three times in PBS for 10 min and exposed to
a secondary anti-mouse IgG conjugated with Alexa fluor 549 (115-505-003; Jackson, Cambridge,
MA) at 1:1000 for 2 hours at room temperature. In quantification of retrograde labeling in ZI, cell
bodies labeled with GFP were counted manually across all sections containing ZI, and the relative
number of labeled cells in different compartments of ZI was calculated as the percentage of total
labeled cells across ZI. The values were then averaged across animals. To obtain the anterograde
projection patterns of ZIr or ZIv/ZId, serial sections across the whole brain were collected and
imaged under 4X objective. Regions with axonal labeling were then imaged under 10X objective
across the depth of the tissue (15 μm z-stack interval). Each image was taken using identical laser
power, gain and offset values.

3.4.3 Optogenetic preparation and stimulation
One week before the behavioral tests, animals were prepared as previously described(Xiong et al.
2015) in Chapter 2. Briefly, to optogenetically manipulate ZIr cell bodies or axon terminals, mice
were implanted with fiber optic cannula (200 µm ID, Thorlabs) two weeks after injecting ChR2,
ArchT or GFP virus (Boyden et al. 2005; Chow et al. 2010). The implantation was made while the
animal was anaesthetized and mounted on stereotaxic apparatus. Small holes (500 µm diameter)
64

were drilled at a 20-degree angle relative to the vertical plane above ZIr (AP -1.2 mm, ML ±2.2
mm, DV -4.4 mm) or PAG (AP -4.4 mm, ML ±1.5 mm, DV -2.2 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 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 in a similar way. Animals could recover for at least
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 specificity of virus expression and the locations of implanted-
fibers. Mice with mistargeted injections or misplacement of optic fibers were excluded from data
analysis.

3.4.4 Behavioral tests and analysis
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 sensor and recorded in real time (M. Zhou et al.
65

2014). A 5-s noise sound at 80 dB sound pressure level (Scan-speaker D2905) was applied to
trigger flight response as previously described (Xiong et al. 2015). The stimulus was repeated for
20 trials per session at an irregular interval ranging from 120 - 180 s. The blue or green LED light
(lasting for the entire duration of noise presentation) was randomly co-applied in half of the trials.
For testing the gain control modulation, noise at 20, 40, 60, 80, and 100 dB sound pressure level
with or without coupled LED stimulation were presented in a randomized order for 10 repetitions
for each condition. For the LED-only control experiments, LED was given in the same way but
without noise stimulation. For the PAG activation experiment, 5-s long LED stimuli with different
powers (1, 4, 7 and 10 mW) were applied for 10 trials. Each animal was tested for consecutive 3
days and data were averaged across days for each animal.
Conditioned fear response Mice underwent auditory fear conditioning in a custom made
conditioning chamber and tested in a test box in a sound-attenuation booth (Gretch-Ken Industries,
Inc). The conditioning chamber and test box were cleaned with 70% ethanol before and after each
session. The bedding material in the test box was replaced before each test session. On the first
day, the animals were exposed to five tones (5kHz tone, 80 dB SPL, duration = 20 s) after 10 min
habituation in the test box. On the following day (conditioning), they were exposed to the 20-s 5
kHz tone co-terminated with a 0.75-mA foot shock (5Hz for 1 s with the duration of each pulse =
100 ms) for 5 times in the conditioning chamber. For testing conditioned fear response, mice (in
ChR2, ArchT and GFP control group respectively) were placed in the test box and given 6 tone
presentations in the absence of foot shocks, with half of the trials (in a randomized order) paired
with LED stimulation. Inter-trial interval was randomly chosen from a range of 120 s to 240 s
(mean = 180 s). For fear extinction, mice were given 10 tone presentations in the absence of foot
shocks. A subset of these mice underwent extinction training with pairing optogenectic
66

suppression and tone presentations. For testing extinction recall, mice which had undergone
extinction training were put in the same test box on the following day and 5 tone presentations
were given. Extinction was also tested on head-fixed animals habituated on the same plate used
for recording, with the speed of the plate recorded during the whole process. For silencing mPFC,
muscimol (200 nl, 1.5 mM) mixed with Alexa Fluor 488 conjugated Dextran (Life Technologies)
was pressure injected into ILA/PL regions right before the extinction test. All the behavioral assays
were video recorded, and a blind procedure was implemented for analysis. Fear response was
scored as the percentage of time freezing during the 20-s presentation of 5 kHz tone. The freezing
of the animal was scored if no movement was detected (except for respiratory movements) for at
least 1s, and the total freezing time during a tone presentation was counted based on the video
analysis. Animals were excluded if they failed to exhibit freezing upon the first CS representation
one day following the conditioning, as defined by less than 30% of time freezing.
Open field locomotion test Mice were placed inside the same test box (25 cm × 25 cm × 50 cm)
for testing the baseline locomotion activity. They could habituate to the arena for 10 min. Each
animal was tested for 2 sessions per day and each session lasted 15 min, during which blue or
green LED stimulation (5 s On /5 s Off) was applied. The animal’s movements were recorded with
an infra-red camera mounted on the top center of the arena. The mouse position was determined
by using custom made semi-automated MATLAB-based tracking software.
Balance beam test Mice were trained to walk along a 70-cm long and 2-cm wide beam elevated
30 cm above the bench. The beam was connected to an enclosed goal box. Following the training,
the animal was placed at the other end and allowed to pass the beam to reach the goal box. Each
animal was tested 6 – 8 trials per day and blue or green LED applied was applied randomly in half
67

of these trials. In the stimulation trial, the LED stimulation lasted the entire duration of the animal’s
walking. The time for the animal to cross the beam was recorded and averaged across trials.

3.4.5 Slice preparation and recording
To confirm the input and output connectivity of ZIr, GAD2::Ai14 mice injected with AAV2/1-
pEF1α-DIO-hChR2-eYFP in ZIr, vGLUT2::Ai14 mice injected with AAV1-Syn-Cre mixed with
AAV2/1-pEF1α-DIO-hChR2-eYFP in ZIr, or wild-type C57BL/6 mice injected with AAV1-
CamKII-hChR2(E123A)-eYFP-WPRE-hGh in mPFC 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). PAG and ZIr 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 GTP, and 10 mM
phosphocreatine; pH = 7.25; 290 mOsm) were used for whole-cell recordings. Signals were
recorded with an Axopatch 200B amplifier (Molecular Devices) under voltage clamp mode at a
holding voltage of –70 mV for excitatory currents or 0 mV for inhibitory currents, filtered at 2 kHz
68

and sampled at 10 kHz (X. Y. 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. 2009a) to blue light stimulation (10 ms pulse, 3 mW power, 10-30 trials,
delivered via a mercury Arc lamp gated with an electronic shutter). Gabazine (4 μM) was added
to the external solution to block GABAergic currents.

3.4.6 In vivo recording in head-fixed animals
One week before electrophysiological recordings, mice were anaesthetized using 1.5% isoflurane
and a head post was attached as described previously(F. Liang et al. 2015). For recording during
conditioning, the animal was head-fixed on the shock plate and a 16-channel silicon probe
(NeuroNexus) was lowered into ZIr. It was exposed to the 20-s 5 kHz tone co-terminated with a
0.75-mA foot shock (5Hz for 1 s with the duration of each pulse = 100 ms) for 5 times. Spikes
were recorded for the first 19 sec after the onset of the tone. For recording during extinction, one
day before the recording session, animals went through the auditory fear conditioning as described
above. On the day of recording, the animal was head-fixed on the running plate (to which it had
been habituated) in a sound-attenuation booth. A parylene-coated tungsten electrode (2 MΩ, FHC)
or a 16-channel silicon probe (NeuroNexus) was lowered into ZIr. The animal was exposed to the
5-kHz tone for 10 times. Spikes during tone presentations were recorded. For silencing mPFC,
muscimol (200 nl, 1.5 mM) mixed with Alexa Fluor 488 conjugated Dextran (or saline) was
pressure injected into mPFC right before the recording session. To confirm the silencing effect, a
tungsten electrode was lowered into the injection area to record spikes before and after the
injection. For recording in PAG, GAD2-Cre mice injected with AAV2/1-pEF1α-DIO-hChR2-
69

eYFP in ZIr and implanted with optic cannula above ZIr were head-fixed on the running plate.
Recording with a tungsten electrode was carried out at two sites in dlPAG. The animal was exposed
to a 50-ms noise (80 dB SPL) for 50 trials to record the sound-evoked responses. Half of these
trials were coupled with 50-ms LED stimulation. To examine effects on spontaneous activity in
PAG, 25-ms on, 25-ms off, 500 ms LED stimulation was applied without sound stimulation for 50
repetitions. Signals were amplified (Plexon) and recorded with custom made LabVIEW software.
The spike timing was analyzed offline. Recording sites were marked by DiI staining (2mg/ml).
Mice were perfused 4% paraformaldehyde right after the recording session to examine the
recording site.

3.4.7 Data processing and statistics
For the flight test, running speed was recorded at 10 Hz sampling rate. The onset latency of flight
response was defined by the time point at which running speed exceeded the average baseline
speed (measured within the 10-s window preceding the noise onset) by 3 standard deviations of
baseline fluctuations. Animals were excluded if they did not show robust flight response at the
beginning, as defined by noise-induced speed not exceeding baseline by 3 standard deviations.
Running traces were normalized based on the peak speed of flight in the absence of optogenetic
manipulation. Peak speed was determined as the maximum running speed after averaging 20
running traces for a session. Travel distance was calculated as the integral of running speed within
the 5-s stimulation window. For the open field test, normalized travel distance was calculated as
the travel distance during the optogenetic stimulation over the baseline travel distance within the
same length of time window.
70

For in vivo extracellular recording, signals were amplified by a preamp (Plexon) at 30 kHz
sampling rate. Spike signals were filtered with a 300–3,000 Hz band-pass filter. The nearby four
channels of the silicon probe were grouped as tetrodes and semi-automatic spike sorting was
performed using the offline sorter of Plexon (Dallas, Texas). Clusters with isolation distance > 20
were considered as separate clusters(Harris et al. 2001). Spike clusters were 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 >99.5% of the spikes. Spike rate was normalized to the average
firing rate of the first 5 trials for each animal. All data analysis performers were blind to the
allocation of the experimental groups. The modulation index was calculated by the average firing
rate of last three trails divided by that of the first three trials.
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 Excel (Microsoft).

71

Chapter 4: Modulation of auditory cortical processing by
lateral posterior nucleus of thalamus
4.1 Introduction
Thalamus is generally considered as a gate to the cerebral cortex. First-order thalamic nuclei, such
as the dorsal lateral geniculate nucleus (dLGN) and ventral medial geniculate body (MGBv), send
bottom-up sensory information to primary sensory cortices, visual and auditory, respectively
(Kremkow and Alonso 2018; Winer et al. 2005). They serve as the major driver of sensory
responses in the cortex for each sensory modality (Guillery and Sherman 2002; Halassa and
Sherman 2019). Compared to first-order thalamic nuclei, the functional roles of higher-order
thalamic nuclei, many of which have broad connections with both primary and secondary cortices,
in sensory processing are much less well understood. The lateral posterior nucleus (LP) of the
thalamus is the rodent homologue of the primate pulvinar nucleus (Harting et al. 1973; Harting,
Hall, and Diamond 1972), with the latter considered the largest thalamic complex (Harting et al.
1973). LP is shown to be involved in the defense circuit, particularly the control of freezing
responses, in Chapter 2. During the manifestation of defensive behaviors, sensory processing could
also be modulated for animals to properly avoid dangers. The aim of this study is to explore the
potential impact of LP, which is part of the defense circuit, on auditory processing in A1.
Previous studies on LP/pulvinar have mostly been focused on its involvements in visual-related
functions, such as visual attention, visually guided behaviors and control of eye movements
(Dominguez-Vargas et al. 2017; Saalmann et al. 2012; Soares et al. 2017; Stitt et al. 2018; H.
Zhou, Schafer, and Desimone 2016), largely due to its extensive reciprocal connections with visual
72

cortical areas (Beltramo and Scanziani 2019; Bennett et al. 2019; Juavinett et al. 2019; Kaas and
Lyon 2007; Nakamura et al. 2015; Oh et al. 2014; Roth et al. 2016; S Shipp 2001; Stitt et al. 2018;
Wong et al. 2009; N. Zhou et al. 2018). Besides visual cortices, LP/pulvinar has also connections
with other sensory cortices including primary and secondary auditory cortices (Cappe et al. 2009;
Hackett, Stepniewska, and Kaas 1998; Nakamura et al. 2015; Oh et al. 2014; Stewart Shipp 2007),
and contains auditory responsive neurons which exhibit short-latency responses (Chalupa and Fish
1978; Gattass, P.B. Sousa, and Oswaldo-Cruz 1978; Magariños-Ascone, Buño, and García-Austt
1988; Woody, Melamed, and Chizhevsky 1991; Yirmiya and Hocherman 1987). These findings
suggest a potential role of LP in auditory processing by its direct connection with auditory cortices.
In addition, the projections from LP to the primary sensory cortex mainly terminates in layer 1
(L1). As L1 contains predominantly inhibitory neurons (Jiang et al. 2013), it is quite likely that LP
could have a suppressing effect on sensory responses in A1.
In this present study, we investigated the effect of LP activity on auditory responses of neurons in
superficial layers of A1 by using bidirectional optogenetic manipulations. We found that LP
activity improved auditory processing in A1 in that it helped to sharpen frequency tuning of
neurons and enhance the signal-to-noise ratio (SNR) of auditory responses, through suppressing
A1 responses via a thresholding mechanism. Moreover, such modulatory effect could contribute
to the relatively stable SNR of A1 responses with varying noise background.

4.2 Results
4.2.1 Bidirectional modulation of frequency tuning and SNR by LP
73

Figure 26. Effects of LP manipulations on A1 response properties
(A) Left, schematic experimental condition. AAV-ArchT was injected into LP and green LED was
applied via an implanted optic fiber to silence LP neurons. Loose-patch recordings were made
74

from A1 L2/3 neurons on the same side. Right, image showing the expression of ArchT in LP,
boundary of which is marked by the white dashed line. Scale bar, 500 m. (B) Reconstructed tonal
receptive field (TRF) of an example A1 L2/3 neuron without (left) and with (right) optogenetic
silencing of LP. Red double arrows mark the frequency tuning bandwidth at 60 dB SPL. Red
arrowhead marks the characteristic frequency (CF). (C) Normalized frequency tuning curves (at
60 dB SPL) without (black) and with (green) LP silencing. (D-G) Normalized spontaneous firing
rate (D), evoked firing rate (E), tuning bandwidth at 60 dB SPL (F) and signal-to-noise ratio (G)
of A1 L2/3 neurons without (OFF) and with (ON) LP silencing. **p < 0.01, *p < 0.05, paired t-
test, n = 8 cells in 7 animals. Data points for the same cell are connected with a line. (H) Left,
experimental condition. AAV-ChR2 was injected into LP and blue LED was delivered to activate
LP neurons. Right, example image showing the expression of ChR2 within LP. Scale bar, 500 m.
(I) TRF of an example A1 L2/3 neuron without (left) and with (right) optogenetic activation of LP
neurons. (J) Normalized frequency tuning curves without (black) and with (blue) LP activation.
(K-N) Normalized spontaneous firing rate (K), evoked firing rate (L), tuning bandwidth at 60 dB
SPL (M), and SNR (N) of A1 L2/3 neurons without and with LP activation. ***p < 0.001, **p <
0.01, paired t-test, n = 13 cells in 10 animals. (O-P) Plot of normalized firing rates evoked by
effective tones (at 60 dB SPL) with vs. without LP silencing (O) or activation (P) for an example
cell. Green and blue lines are the best fit linear regression line. Gray dashed line is the identity
line. (Q-S) Summary of parameters of linear regression for all neurons in LP silencing (ArchT)
and activation (ChR2) groups, respectively. R
2
(Q): 0.84 ± 0.09 (mean ± SD, n = 8 cells) vs. 0.88
± 0.08 (n = 13 cells); slope (R): 1.05 ± 0.17 (not significantly different from 1, p = 0.41, t test, n =
8 cells) vs. 0.94 ± 0.17 (not significantly different from 1, p = 0.26, Z-test, n = 13 cells); y-intercept
(S): 0.23 ± 0.11 (significantly > 0, p < 0.001, Z-test, n = 8 cells) vs. -0.19 ± 0.12 (significantly <
0, p < 0.001, Z-test, n = 13 cells); Bar = SEM.

To study whether LP can influence auditory cortical processing, we manipulated LP activity in
awake mice using optogenetic approaches and examined functional response properties of A1
neurons by single-cell loose-patch recording (see 4.4 Material and methods). To reversibly and
temporarily silence LP spikes, we injected AAV expressing ArchT into LP and implanted an optic
fiber on top of the LP to deliver green LED light (Figure 26A). To investigate auditory information
processing, we presented a repertoire of tone pips of varying intensities and frequencies (50-ms
duration, 10 – 70 dB sound pressure level or SPL, 5 – 45 kHz) and recorded tone-evoked spike
responses from L2/3 pyramidal neurons (see 4.4 Material and methods). A typical L2/3 pyramidal
neuron exhibited a V-shaped tonal receptive field (TRF) with a distinct characteristic frequency
(CF) (Figure 26B, left). We interleaved LED-on and LED-off trials so that TRF without and with
75

LP silencing could be compared in the same neuron. When LP was silenced, we observed an
overall increase in response level and receptive field size as compared to the LED-off condition
(Figure 26B, right; Figure 27, left). Nevertheless, the overall shape of frequency tuning (Figure
26C) and the CF of TRF (Figure 28, left) remained similar. Analyzing all the recorded L2/3
neurons, we found significant increases in the spontaneous firing rate (FR), evoked FR and
bandwidth of frequency tuning (as measured at an intensity level of 60 dB SPL) when silencing
LP (Figure 26D-F). More importantly, the signal-to-noise ratio (SNR), as measured by the ratio of
evoked to spontaneous FR, was decreased (Figure 26G). These results indicated that when LP
activity was suppressed, auditory information processing in A1 might be compromised due to
broadened frequency tuning and a reduced SNR.

Figure 27. LED illumination had no effect on auditory response
Summary of cumulative distribution of the percentage changes of auditory responses in A1
neurons with and without green (left, n = 12 cells) and blue LED stimulation (right, n = 13 cells).
p < 0.001, two-sample Kolmogorov–Smirnov test.

76

Figure 28. LP manipulation did not change CF values
Summary of the CF values of A1 L2/3 neurons with and without optic LP silencing (left, n = 8
cells) and activation (right, n = 13 cells). N.S., not significant, p = 1.0 (left) and 1.0 (right),
respectively, two-sided paired t-test.

Next, we optogenetically activated LP by injecting AAV expressing channelrhodopsin2 (ChR2)
into LP and delivering blue LED light pulses (Figure 26H). As shown by an example L2/3 neuron,
activation of LP decreased the amplitude of tone-evoked responses and shrank the tonal receptive
field (Figure 26I-J; Figure 27, right), without changing the CF of TRF (Figure 28, right). Opposite
to the effects of silencing LP, activation of LP reduced the spontaneous and evoked FR, narrowed
the frequency tuning bandwidth, while enhanced the SNR (Figure 26K-N). Therefore, the
frequency tuning and SNR of A1 L2/3 pyramidal neurons could be bidirectionally modulated by
manipulating LP activity. Increasing LP activity generally improves auditory cortical processing
by enhancing frequency selectivity and the SNR of auditory responses. As a control, we performed
similar experiments in GFP-expressing animals and did not observe any significant changes in
auditory response level by either green or blue LED light delivery (Figure 27). In L4, we did not
observe changes of either spontaneous or evoked firing rates induced by optogenetic manipulations
of LP (Figure 29). Therefore, the LP’s modulatory effect is specific to superficial layers.
77

Figure 29. Optogenetic manipulation of LP had no effect on A1 L4 auditory response
(A-B) Summary of the normalized spontaneous and evoked firing rates in A1 L4 neurons without
and with LP silencing. N = 5 cells. (C-D) Similar as (A-B) but plotted for A1 L4 neurons without
and with LP activation. N = 12 cells. N.S., not significant, p = 0.76 (A), 0.82 (B), 0.38 (C), and
0.55 (D), respectively, two-sided paired t-test.

4.2.2 LP exerts a thresholding effect on A1 L2/3 responses
Comparing frequency tuning curves without and with LP manipulations (Figure 26C, J), it
appeared that LP activity just shifted the A1 frequency tuning curve up or down, without changing
the tuning preference. To further elucidate the nature of the LP modulatory effect, we plotted firing
rates evoked by effective tones (see 4.4 Material and methods) with versus without manipulating
LP activity and then performed the linear regression analysis. As shown by the two example A1
neurons (the same as shown in Figure 26B, I), data points were distributed along a line which had
a positive y-intercept for LP silencing (Figure 26O) and a negative y-intercept for LP activation
78

(Figure 26P), consistent with the notion that all tone responses were elevated or reduced by a
certain amount when LP activity was manipulated. We did a similar analysis for all the recorded
neurons and found that the linearity was generally good within the recorded L2/3 population, as
shown by the close-to-1 R
2
(Figure 26Q). The slope of the best fit line was close to 1 for both LP
silencing and activation (Figure 26R). The y-intercept was all positive for LP silencing but
negative for LP activation (Figure 26S). These results strongly suggest a subtractive modulation
of A1 responses by LP activity without changing the response gain, since the slope was 1, similar
to a thresholding effect produced by increasing the level of background noise (F. Liang et al. 2014).
Thus, LP can bidirectionally modulate auditory responses in superficial layers of A1 through a
thresholding mechanism, i.e. by shifting A1 frequency tuning up or down.

4.2.3 LP’s modulatory effect is mediated by the its A1 projection
LP may modulate A1 responses directly through the LP to A1 projection, or indirectly through the
LP projections to secondary cortices (Arend et al. 2008; Cappe et al. 2009; Hackett, Stepniewska,
and Kaas 1998; De La Mothe et al. 2006; Nakamura et al. 2015; Oh et al. 2014). We wondered
whether the direct LP-A1 projection could mediate the observed modulatory effect of LP. By
injecting a retrograde dye, CTB-488, in A1, we examined the distribution of LP neurons projecting
to A1 (Figure 30A). Numerous retrogradely labelled neurons were found in LP (Figure 30A, right),
with a bias towards its caudal part (Figure 31). Furthermore, injection of AAV-GFP into LP
revealed densely labeled axon terminals particularly in L1 of A1 (Figure 30B), confirming the
direct projection from LP to A1 (De La Mothe et al. 2006; Nakamura et al. 2015). To test whether
the LP-A1 projection could account for the LP modulatory effect in A1, we optogenetically
79

silenced the LP-A1 axon terminals by injecting AAV-eNpHR3.0 in LP and placing an optic fiber
on the exposed A1 surface (Figure 30C). We observed an effect on A1 L2/3 neurons like silencing
LP neurons: the spontaneous and evoked FR was increased, the frequency tuning bandwidth was
broadened, and the SNR was decreased (Figure 30D-G; Figure 32A). Conversely, we
optogenetically activated the LP-A1 axon terminals by expressing AAV-ChR2 in LP and shining
blue light on the exposed A1 surface (Figure 30H). In line with the effects caused by LP activation,
the activation of LP-A1 axon terminals reduced spontaneous and evoked firing rates, sharpened
frequency tuning, and enhanced the SNR in A1 L2/3 neurons (Figure 30I-L; Figure 32A).

80

Figure 30. Bidirectional activity manipulations of LP-A1 projection
(A) Left, injection of retrograde tracer CTB-488 into A1. Middle, image showing spread of CTB-
488 at the injection site. Scale bar, 500 m. Right, retrogradely labelled neurons in LP. Scale bars,
200 m. (B) Left, injection of AAV-GFP into LP. Middle, expression of GFP at the injection site.
Scale bar, 500 m. Right, GFP-labeled LP axons in A1. Scale bar, 100 m. L1 is marked. (C)
Injection of AAV-eNpHR3.0 into LP and optical silencing of LP-A1 axon terminals by placing
the optic fiber over A1. (D-G) Normalized spontaneous firing rate (D), evoked firing rate (E),
tuning bandwidth (F), and SNR (G) of A1 L2/3 neurons without and with LP-A1 axon terminal
silencing. *p < 0.05, paired t-test, n = 9 cells in 4 animals. (H) Injection of AAV-ChR2 into LP
and optical activation of LP-A1 axon terminals. (I-L) Normalized spontaneous firing rate (I),
evoked firing rate (J), tuning bandwidth (K), and SNR (L) of A1 L2/3 neurons without and with
LP-A1 axons terminal activation. **p < 0.01, *p < 0.05, paired t-test, n = 10 cells in 4 animals.
(M-N) Plot of normalized firing rates evoked by effective tones with vs. without LP-A1 axon
terminal silencing (M) or activation (N) for an example cell. (O-Q) Summary of parameters of
linear regression for the terminal silencing (eNpHR3.0) and activation (ChR2) group, respectively.
R
2
(O): 0.87 ± 0.098 (n = 9 cells) vs. 0.84 ± 0.091 (n = 10 cells); slope (P): 1.05 ± 0.22 (not
significantly different from 1, p = 0.48, Z-test, n = 9 cells) vs. 0.95 ± 0.18 (not significantly
different from 1, p = 0.42, Z-test, n = 10 cells); y-intercept (Q): 0.19 ± 0.077 (significantly > 0, p
< 0.001, Z-test, n = 9 cells) vs. -0.19 ± 0.083 (significantly < 0, p < 0.001, Z-test, n = 10 cells). Bar
= SEM.

We performed similar linear fitting on tone-evoked firing rates with and without terminal
manipulations (Figure 30M-N) and observed a thresholding effect like manipulating LP neurons.
The linearity was high (Figure 30O), the slope of the best fit line was close to 1 (Figure 30P), and
the y-intercept was a positive value for the terminal silencing while a negative value for the
terminal activation (Figure 30Q). These data suggest that the LP to A1 projection can largely
mediate the LP modulatory effect on A1 responses.

81

Figure 31. Biased projection from caudal LP to A1
Images showing retrogradely labelled A1-prejecting neurons in LP. Scale bar = 200 m.

Figure 32. Effects of LP terminal manipulation on TRF of A1 neurons
The TRF of two example A1 L2/3 neurons without and with LP terminal silencing (A) and
activation (B), respectively.

4.2.4 LP axons generate a disynaptic inhibitory effect on A1 L2/3 neurons
As the LP projection to the cortex was shown to be excitatory (Roth et al. 2016; N. Zhou et al.
2018), an immediate question is how LP activity exerts a net inhibitory effect on A1 L2/3 neurons.
Since LP axons project to L1 (Figure 30B), which contains predominantly inhibitory neurons
(Jiang et al. 2013), it is possible that LP axons can indirectly suppress L2/3 pyramidal neurons via
L1 inhibitory neurons (Jiang et al. 2013). To test this idea, we injected AAV-ChR2 into LP and
made whole-cell voltage-clamp recordings from A1 neurons in slice preparations (Figure 33A).
TTX and 4AP were present in the bath solution to ensure that only monosynaptic responses were
recorded (Petreanu et al. 2009b). We performed whole-cell recordings from several types of
neurons in A1: inhibitory neurons in L1 as labeled by crossing GAD2-Cre with the Ai14 Cre-
dependent tdTomato reporter, pyramidal neurons identified as tdTomato-negative cells in GAD2-
Cre::Ai14 animals, parvalbumin (PV) and somatostatin (SOM) positive inhibitory neurons as
82

labeled by crossing PV-Cre or SOM-Cre with Ai14, respectively. As shown by two example cells,
blue light activation of LP-A1 axons resulted in a robust excitatory postsynaptic current (EPSC)
in the L1 inhibitory neuron, whereas no EPSC was observed in the pyramidal (PYR) cell (Figure
33B). Overall, none of the PYR neurons we recorded across L2-4 received direct input from LP,
whereas more than 50% of L1 inhibitory neurons received the direct LP input (Figure 33C, E). A
lower fraction of PV neurons in L2/3 received direct input from LP, whereas none of the recorded
SOM neurons did so (Figure 33D-E). Together, these results indicate that LP-A1 axons
preferentially innervate L1 inhibitory neurons and PV inhibitory neurons in superficial layers,
which may then provide disynaptic inhibition to L2/3 pyramidal neurons.

Figure 33. Cell types innervated by LP-A1 axons
(A) Slice recording paradigm. AAV-ChR2 was injected into LP. Blue LED was applied to A1 to
active LP-A1 axon terminals. Whole-cell recordings were made from A1 neurons. (B) Light-
evoked monosynaptic EPSC (average trace) recorded in an example pyramidal (PRY, top) and L1
inhibitory neuron (bottom). TTX and 4AP were present in the bath solution. Arrowhead marks the
onset of 5-ms blue light pulse. Scale bars: 200 ms and 20 pA. (C) Plot of average amplitude of
light-evoked monosynaptic EPSCs vs. the cell’s cortical depth for the recorded L1 inhibitory
neurons (red, GAD2+) and pyramidal cells (black). Dashed line marks the boundary between L1
and L2/3. (D) Plot of average amplitude of light-evoked monosynaptic EPSCs vs. the cell’s cortical
depth for the recorded PV (magenta) and SOM (blue) neurons. (E) Summary of connection
probability between LP-A1 axons and L1 GAD2+ neurons, as well as pyramidal, PV and SOM
neurons in different layers.

4.2.5 SC can provide input to drive LP-mediated modulation of A1 responses
83

Figure 34. SC provides input to drive LP-mediated modulation of A1 responses
(A) Injection of AAV-GFP into deep layers of SC. Left, expression of GFP in SC. Boundaries
between the superficial layer, intermediate layer and deep layer are marked by dashed lines. Right,
GFP labeled SC axons in LP. Scale bar, 500 m. (B) Left, slice recording paradigm: expressing
ChR2 in deep layers of SC and whole-cell recording from LP neurons. Right, light-evoked
monosynaptic EPSC (average trace) in an example LP neuron before and after perfusing in CNQX,
indicating that the EPSC was glutamatergic in nature. Scale bars: 100 pA and 50 ms. (C) Average
amplitude of light-evoked monosynaptic EPSCs in LP neurons. Neurons that did not show light-
evoked responses were excluded. (D) Transsynaptic labeling of SC-recipient LP neurons with
ChR2 by first injection of AAV-Cre in SC and second injection of Cre-dependent ChR2 virus in
LP. Right, imaging showing LP neurons expressing ChR2-EYFP. Scale bar, 200 m. (E-G)
Normalized spontaneous firing rate (E), evoked firing rate (F), and SNR (G) of A1 L2/3 neurons
without and with activation of SC-recipient LP neurons. ***p < 0.001, **p < 0.01, paired t-test, n
= 9 cells in 5 animals. (H) Experimental paradigm: silencing SC by infusing bupivacaine and
recording in A1. (I-L) Normalized spontaneous firing rate (I), evoked firing rate (J), tuning
bandwidth at 60 dB SPL (K) and SNR (L) of A1 L2/3 neurons before and after SC silencing. ***p
< 0.001, **p < 0.01, *p < 0.05, paired t-test, n = 10 cells in 7 animals.

84

It has been shown that the caudal part of LP receives strong input from the superior colliculus (SC)
in the midbrain (Beltramo and Scanziani 2019; Bennett et al. 2019; Gale and Murphy 2014;
Stepniewska, Ql, and Kaas 2000; Wei et al. 2015; Zingg et al. 2017). We thus wondered whether
SC, which does not project to auditory cortices (Basso and May 2017; Cang et al. 2018; Ito and
Feldheim 2018), could provide direct input to drive the LP-mediated modulation of A1 responses.
To confirm the connectivity from SC to LP, we injected AAV-GFP in deep layers of SC, the
auditory related part of SC (Bednárová, Grothe, and Myoga 2018; Drager and Hubel 1975; King
and Palmer 1985; Meredith and Stein 1986; Wise and Irvine 1983; Zingg et al. 2017), and found
abundant GFP-labeled axons in LP (Figure 34A), with a bias towards its caudal part (Figure 35).
Expressing ChR2 in SC and then performing whole-cell slice recoding from caudal LP neurons in
the presence of TTX and 4AP further confirmed direct innervations of LP neurons by SC axons
(Figure 34B-C). We also expressed ChR2 in SC-recipient LP neurons by first injection of AAV-
Cre in SC and second injection of AAV encoding Cre-dependent ChR2 in the caudal LP (Zingg et
al., 2017) (Figure 34D). In A1, optogenetic activation of SC-recipient LP neurons produced similar
effects to the activation of the general LP population: spontaneous and evoked firing rates were
reduced, and SNR was increased (Figure 34E-G). Conversely, if SC drives LP to modulate A1
responses, then silencing SC may produce a similar effect as silencing LP. We next silenced SC
by infusing bupivacaine (S. Lee, Carvell, and Simons 2008; Moraga-Amaro et al. 2014) (Figure
34H). This resulted in increases of spontaneous and evoked firing rates, broadening of frequency
tuning and a reduction of SNR in A1 L2/3 neurons (Figure 34I-L), like the effects of silencing LP.
Together, these results provide evidence that SC can provide input to drive LP-mediated
modulation of A1 responses.
85

Figure 35. SC projections to LP with a caudal bias
Images showing anterogradely labelled SC terminals in LP. Scale bar = 200 m.

4.2.6 LP plays a role in noise-related contextual modulation of A1 responses
Previously, in the visual system, it has been proposed that LP may provide contextual information
to visual cortex (Roth et al. 2016). Whether LP could play a similar role in auditory processing has
been unknown. In an acoustic environment, one common contextual factor is the background
noise. Previously, it has been shown that elevating the background noise level results in narrowing
of frequency tuning of A1 neurons (without changing tuning preference) through a thresholding
effect (F. Liang et al. 2014). Since LP manipulations also produce a thresholding effect, we
wondered whether LP could contribute to noise-related contextual modulation. We noted that LP
neurons responded robustly to white-noise sound, with the response amplitude increasing with
increasing noise levels (Figure 36A). Comparing the onset latency of noise responses in LP, SC
and A1 L4, we found that it was the shortest in SC, while similar between LP and A1 L4 (Figure
36B). In addition, silencing SC greatly reduced the amplitude of noise responses in LP (Figure
36C). These results further support the idea that the LP auditory responses are primarily driven by
the bottom-up input from SC.
86

Figure 36. LP plays a role in noise-related contextual modulation of A1 responses.
(A) Left, peristimulus spike time histogram (PSTH) of responses of an example caudal LP neuron
to white noise sound, duration of which is marked by a thick line. Right, dependence of evoked
firing rate on the intensity of noise (n = 10 LP neurons). Bar = SD. (B) Left, onset latencies of
noise responses at 60 dB SPL for SC (n = 17 cells), LP (n = 17 cells) and A1 L4 (n = 8 cells)
neurons. ***p < 0.001, One-way ANOVA with Bonferroni's multiple comparisons test. (C)
Normalized noise-evoked firing rates of LP neurons before (Ctrl) and after silencing SC (ΔSC).
***p < 0.001, paired t-test, n = 12 cells in 2 animals. (D) Left, spectrogram of an example stimulus:
a 50-ms tone (60 dB SPL) embedded in low-level noise (0 dB SPL, 250 ms). Right, PSTHs for
87

responses of an example A1 L2/3 neuron to CF tone embedded in 0 dB noise before (black) and
after (red) silencing LP with bupivacaine. Thick line marks the tone duration. (E) Response of the
same cell to the same tone (60 dB SPL) embedded in high-level noise (45 dB SPL) before and
after silencing LP. (F) Upper, summary of evoked firing rates of A1 neurons at different noise
levels before (black) and after (red) silencing LP. Lower, suppression of evoked firing rate by LP
at different noise levels. ***p < 0.001, **p < 0.01, *p < 0.05, compared to the values under 0 dB
noise condition. Bar = SD. n = 9 cells from 4 animals. (G) Summary of spontaneous firing rate
before and after silencing LP. (H) Summary of SNR before and after silencing LP. (I) Summary
of tuning bandwidth before and after silencing LP.

To test how noise-driven LP activity affects A1 frequency processing, we applied tones (of varying
frequencies) embedded in broadband noise of different levels (Figure 36D-E). Frequency tuning
was compared before and after silencing LP with bupivacaine. As shown by an example A1 L2/3
neuron (Figure 36D-E, gray), increasing the noise level reduced the amplitude of the response
evoked by the preferred tone. Silencing LP resulted in a general increase in the tone-evoked
response, regardless of noise levels, but to different degrees (Figure 36D-E, red). Summarizing all
recorded neurons, we found that silencing LP universally enhanced tone-evoked responses in A1
across different noise levels (Figure 36F, upper). Notably, the enhancement was larger at higher
noise levels. This is consistent with the notion that LP neuron responses increase with increasing
noise levels (Figure 36A) and therefore exert a larger suppressive effect in A1 (Figure 36F, lower).
Silencing LP also elevated the spontaneous firing rate in a noise-level dependent manner so that
the elevation was larger at higher noise levels (Figure 36G). The SNR decreased with increasing
noise levels in the control condition (Figure 36H, upper), indicating that high-level noise has a
detrimental effect on SNR, thus deteriorating auditory processing. Silencing LP not only reduced
SNR, but also accelerated the detrimental effect of background noise (Figure 36I, upper). Again,
the modulatory effect on SNR was larger at higher noise levels (Figure 36I, lower). Finally, the
frequency tuning bandwidth was reduced with increasing noise levels (Figure 36H, upper),
consistent with previous studies (F. Liang et al. 2014). Silencing LP not only broadened the tuning
88

bandwidth, but also slowed down the modulatory effect on bandwidth by increasing the noise level
(Figure 36H). Together, these results suggest that LP plays a role in contextual modulation of A1
frequency processing by noise background.

4.3 Discussion
LP is considered a higher-order thalamic nucleus. In general, the influence the higher-order
thalamus has on A1 processing has remained obscure. In the present study, our results demonstrate
that LP activity can modulate A1 processing in superficial layers. The general outcome of
increasing LP activity is to improve auditory processing in A1 by sharpening frequency tuning and
increasing the SNR of auditory evoked responses. This is achieved by a linear subtractive
suppression of auditory evoked responses together with a suppression of spontaneous activity. We
also demonstrate that such modulatory effect is largely mediated by the direct projection of LP to
A1, where LP axons preferentially innervate L1 inhibitory neurons and PV inhibitory neurons in
superficial layers, producing a disynaptic inhibitory effect on L2/3 pyramidal neurons.
4.3.1 Contextual modulation role of LP in different background noise levels
The LP-mediated modulatory effect on A1 processing may be particularly pronounced and
beneficial when there is a high noise background. Our previous study has demonstrated that
increasing background noise levels is equivalent to lowering the intensity of test (tone) stimuli, i.e.
shifting up the tonal receptive field by a certain Δthreshold value (F. Liang et al. 2014). In the
current study, our results suggest that the noise effect on A1 frequency tuning can be at least
partially achieved through LP, activity of which is modulated by noise levels, as LP activity also
exerts a similar thresholding effect on A1 frequency tuning. LP contributes to noise-related
89

contextual modulation of A1 frequency processing in a positive manner, in that it helps not only
to prevent the SNR of auditory responses from being quickly deteriorated by high-level noise but
also to accelerate the sharpening of frequency tuning with increasing noise levels, which may
compensate somehow for the detrimental effect on SNR.

4.3.2 Parallel and multimodal pathway from SC to LP
It is known that LP has extensive reciprocal connectivity with cortical areas including visual and
auditory cortices (Hackett et al., 1998; De La Mothe et al., 2006; Nakamura et al., 2015; Oh et al.,
2014; Tohmi et al., 2014; Zhou et al., 2018; Bennett et al., 2019). Previously, it has been suggested
that LP/pulvinar serves in a cortico-thalamo-cortical (“transthalamic”) indirect route for
information transfer from one cortical area to another (Guillery and Sherman 2002; Sherman
2016). Besides cortical inputs, LP also receive strong inputs from SC (Beltramo and Scanziani
2019; Bennett et al. 2019; Gale and Murphy 2014; Stepniewska, Ql, and Kaas 2000; Wei et al.
2015; Zingg et al. 2017). In this study, we demonstrate that the caudal LP receives direct input
from the auditory related part of SC and projects to A1, and that silencing SC produces similar
changes in A1 functional properties as silencing LP. This suggests that SC can relay bottom-up
auditory information to LP to drive its modulation of A1 processing. Thus, our results suggest that
a previously unrecognized pathway, the SC-LP-A1 pathway, can provide bottom-up modulation
of A1 processing to enhance salience of auditory information. How this pathway may interact with
the canonical geniculo-thalamo-cortical auditory pathway remains to be investigated. Finally, SC
is multisensory (Drager and Hubel 1975; King and Palmer 1985; Meredith and Stein 1986). The
retino-recipient part of SC, i.e. the superficial layer of SC, also projects strongly to the caudal LP
90

(Beltramo and Scanziani 2019; Bennett et al. 2019). Therefore, it is likely that caudal LP neurons
are multi-modal as well and that specific visual inputs may be able to cross-modally modulate A1
processing via LP. This interesting possibility awaits to be further tested in the future.

4.4 Material and methods
4.4.1 Animal model, 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/6J) and
transgenic (GAD2-Cre, PV-Cre, SOM-Cre, and Ai14) mice aged 8-12 weeks weighed 18-28g were
obtained from the Jackson Laboratory. Mice were housed under a 12 h light/dark cycle. Food and
water were provided ad libitum.
Viral injections were performed as described in previous chapters. For optogenetic silencing and
activating LP and its terminals, AAV1-CAMKII-hChR2-eYFP (UPenn Vector Core, 1.7×10
13

GC/ml), AAV1-CAG-ArchT-GFP (UPenn Vector Core, 1.7×10
13
GC/ml), and AAV1-hSyn-
eNpHR3.0-mCherry (UPenn Vector Core, 1.7×10
13
GC/ml) was injected into LP (50 nl total
volume, AP -2.4 mm, ML +1.6 mm, DV -2.4 mm) of wild type animals, respectively. For
anterograde tracing of LP projection pattern, AAV2/1-CB7-Cl-eGFP-WPRE-rBG (UPenn Vector
Core, 1.7×10
13
GC/ml) was injected into LP. For retrograde tracing the afferent inputs, we injected
CTB488 into A1 (AP -2.6 mm, ML +4.4 mm, DV +0.6 mm). For optogenetic activating SC axonal
terminals in LP for slice recording, AAV1-CAMKII-hChR2-eYFP (UPenn Vector Core, 1.7×10
13

GC/ml) was injected into SC (AP -3.75 mm, ML +0.6 mm, DV -1.45 mm). For transsynaptic
labelling from SC to LP, AAV2/1-hSyn-Cre-WPRE-hGH (UPenn Vector Core, 2.5×10
13
GC/mL)
91

was injected into SC (AP -3.75 mm, ML +0.6 mm, DV -1.45 mm), and AAV2/1-EF1a-DIO-
hChR2-eYFP (UPenn Vector Core, 1.6×10
13
GC/mL) was injected into LP. Animals could recover
for at least 3 weeks following the injection for the expression of viruses.
After experiments, animals were perfused and imaged as described in the previous chapters (3.4.2).
Nissil stain was also performed as previously described (3.4.2).

4.4.2 Optogenetic, pharmacological manipulation and slice recording
For optogenetic manipulation, optic fibers were implanted as previously described (3.4.3). For
manipulation of terminals from LP to A1, optic cannula connected with a blue or an amber LED
(589 nm, 10 Mw, Thorlabs) source was placed on the surface of A1. For drug infusions during
recording, an injector (100 µm ID, RWD) was inserted into microinjection tube, and 200 nl 0.5%
bupivacaine mixed with DiI (2 mg/ml) were slowly injected into SC (AP -3.75 mm, ML +0.6 mm,
DV -1.45 mm) by a micro-syringe. Animals could recover for one week before recording session.
During the recovery period, they were habituated to the head fixation on the running plate.
Following recording sessions, animals were euthanized, and the brain was imaged to verify the
specificity of virus expression and the locations of implantations. Mice with mistargeted injections
or misplacement of drug infusion tubes or optic fibers were excluded from data analysis.
Slice recordings were also performed as previously described (3.4.5). PV+, SOM+ or L1 GAD2+
neuron cell types were determined by the tdTomato expression. Laminar location of the recorded
cell was determined by its depth from the pial surface. 6-cyano-7-nitroquinoxaline-2,3-dioneis
(CNQX) was added to block AMPA receptors for glutamate transmission.
92

4.4.3 Sound stimulation
White noise (50-ms, 70 dB sound pressure level or SPL) or tone pips (100-ms duration, 3-ms ramp)
of various frequencies (2–45.25 kHz, 0.1 octave interval) and intensities (10–70 dB SPL, at 10-dB
interval) were generated by custom software (LabView, National Instruments) through a 16-bit
National Instruments interface, and delivered through a calibrated speaker (Tucker-Davis
Technologies) to the contralateral ear. The 322 testing stimuli were presented in a pseudorandom
sequence. The inter-stimulus interval of noise stimulation or two-tone pips was 1 s. For auditory
stimuli embedded in noise of different levels, 50 ms tone pips at different intensities and
frequencies as described above were embedded in white noise of different intensity levels (0, 15,
30, 45 dB SPL). The white noise was 250 ms in duration and preceded the tone presentation by
100 ms.

4.4.4 In vivo electrophysiology
One week after the preparation, animals were head-fixed on the running plate, and
electrophysiology recording with either optogenetic or pharmacological manipulation was carried
out in a sound-attenuation booth. Loose-patch recordings were performed as previously described
(Ibrahim et al. 2016; F. Liang et al. 2018; M. Zhou et al. 2014), with a patch pipette filled with an
artificial cerebral spinal fluid (ACSF; 126 mM NaCl, 2.5 mM KCl, 1.25 mM Na2PO4, 26 mM
NaHCO3, 1 mM MgCl2, 2 mM CaCl2 and 10 mM glucose). A loose seal (0.1–0.5 GΩ) was made
on the cell body, allowing spikes only from the patched cell to be recorded. Signals were recorded
with an Axopatch 200B amplifier (Molecular Devices) under voltage-clamp mode, with a
93

command voltage applied to adjust the baseline current to near zero. Loose-patch recording signals
were filtered with a 100–5,000 Hz band-pass filter. The depths of the recorded neurons were
determined based on the micromanipulator reading. Spikes could be detected without ambiguity
because their amplitudes were normally higher than 100 pA, while the baseline fluctuation was <5
pA. Multichannel recordings were carried out by lowering a 64-channel silicone probe
(NeuroNexus) into the target region. Signals were recorded by an Open-Ephys system. Multi-unit
signals during sound stimulation were recorded and saved for offline analysis.

4.4.5 Data analysis and statistics
Noise-driven and tone-driven spike rates were analyzed within a 10- to 60-ms time window after
the onset of tones. Spontaneous firing rate were analyzed within 50-ms before the onset of sound
stimulation. TRFs were reconstructed according to the array sequence. Boundaries of spike TRF
were determined following previous descriptions (F. Liang et al. 2014; Schumacher, Schneider,
and Woolley 2011; Sutter and Schreiner 1991; Xiong et al. 2013). The frequency–intensity space
was first up sampled 3 times along the frequency and intensity dimensions to increase tuning
resolution. A threshold at the value equal to the spontaneous spike rate plus 20% of the peak-
evoked rate was then used to define significant evoked responses. Responses to frequency–
intensity combinations that met this criterion were considered to fall within the TRF of the neuron,
which generated the contour of the TRF (Xiong et al. 2013). For optogenetic or pharmacological
manipulation, the tone evoked response was calculated by average tone-evoked firing rates across
the responsive frequency–intensity combinations under both control and manipulation conditions.
Characteristic frequency (CF) was defined as the frequency (Hz) at which the lowest sound
94

pressure level was necessary to evoke a significant excitatory response. Bandwidth of TRF was
determined as the total frequency range for effective tones at 60 dB SPL. Onset latency of spike
response was determined from the generated peristimulus spike-time histogram (PSTH) as the lag
between the stimulus onset and the time point where spike rate exceeded the average baseline by
2 standard deviations of baseline fluctuations. Signal to noise ratios (SNR) for auditory responses
were calculated as the evoked firing rate divided by the spontaneous firing rate.
For frequency tuning curve, the tone-evoked responses at 60 dB SPL for different frequencies were
used with and without manipulation, and they were normalized to the highest evoked firing rate in
the control condition then averaged for all the recorded neurons. Linear regression fitting for the
normalized frequency tuning curve with and without manipulation was also performed for all
recorded neurons in each group, and R
2
, slope, and intercept of the best fit line for each cell were
determined.
For spike sorting in multichannel recording, spike signals were filtered with a 300–3,000 Hz band-
pass filter. The nearby four channels of the silicon probe were grouped as tetrodes and semi-
automatic spike sorting was performed using the offline sorter of Plexon (Dallas, Texas), following
our previous study (G.-W. Zhang et al. 2018). Clusters with isolation distance > 20 were
considered as separate clusters. Spike clusters were classified as single units only if the waveform
SNR (signal-to-noise ratio) exceeded 4 (12 dB) and the inter-spike interval was longer than1.2 ms
for >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, Z-test or two-sided paired t-test was applied. Two-sample
95

Kolmogorov-Smirnov test was used to determine whether the data from two groups were from the
same distribution. Statistical analysis was conducted with Matlab.

96

References
Adolphs, Ralph. 2013. “The Biology of Fear.” Current Biology 23(2): R79–93.
Arend, Isabel et al. 2008. “The Role of the Human Pulvinar in Visual Attention and Action:
Evidence from Temporal-Order Judgment, Saccade Decision, and Antisaccade Tasks.” Progress
in brain research 171: 475–83.
Baldi, Elisabetta, and Corrado Bucherelli. 2014. “Entorhinal Cortex Contribution to Contextual
Fear Conditioning Extinction and Reconsolidation in Rats.” Neurobiology of Learning and
Memory.
Bandler, Richard, and Michael T. Shipley. 1994. “Columnar Organization in the Midbrain
Periaqueductal Gray: Modules for Emotional Expression?” Trends in Neurosciences 17(9): 379–
89.
Barad, Mark, Po-Wu Gean, and Beat Lutz. 2006. “The Role of the Amygdala in the Extinction of
Conditioned Fear.” Biological Psychiatry 60(4): 322–28.
Barthó, P., T. F. Freund, and László Acsády. 2002. “Selective GABAergic Innervation of Thalamic
Nuclei from Zona Incerta.” European Journal of Neuroscience 16(6): 999–1014.
Basso, Michele A, and Paul J May. 2017. “Circuits for Action and Cognition: A View from the
Superior Colliculus.” Annual review of vision science 3: 197–226.
Baxter, Mark G., and Elisabeth A. Murray. 2002. “The Amygdala and Reward.” Nature Reviews
Neuroscience.
Bednárová, Veronika, Benedikt Grothe, and Michael H. Myoga. 2018. “Complex and Spatially
Segregated Auditory Inputs of the Mouse Superior Colliculus.” The Journal of Physiology
596(21): 5281–98.
Beitz, Alvin J. 1989. “Possible Origin of Glutamatergic Projections to the Midbrain Periaqueductal
Gray and Deep Layer of the Superior Colliculus of the Rat.” Brain Research Bulletin 23(1–2): 25–
35.
Beltramo, Riccardo, and Massimo Scanziani. 2019. “A Collicular Visual Cortex: Neocortical
Space for an Ancient Midbrain Visual Structure.” Science 363(6422): 64–69.
Bennett, Corbett et al. 2019. “Higher-Order Thalamic Circuits Channel Parallel Streams of Visual
Information in Mice.” Neuron 102(2): 477-492.e5.
Bennett, Corbett, Sergio Arroyo, and Shaul Hestrin. 2013. “Subthreshold Mechanisms Underlying
State-Dependent Modulation of Visual Responses.” Neuron.
97

Bernays, R. L., L. Heeb, M. Cuenod, and P. Streit. 1988. “Afferents to the Rat Red Nucleus Studied
by Means of D-[3H] Aspart Ate, [3H]Choline and Non-Selective Tracers.” Neuroscience 26(2):
601–19.
Bevilaqua, Lia R. et al. 2006. “The Entorhinal Cortex Plays a Role in Extinction.” Neurobiology
of Learning and Memory.
Bevilaqua, Lia R.M. et al. 2008. “The Role of the Entorhinal Cortex in Extinction: Influences of
Aging.” Neural Plasticity.
Blanchard, D. Caroline, and Robert J. Blanchard. 2008. “Chapter 2.4 Defensive Behaviors, Fear,
and Anxiety.” Handbook of Behavioral Neuroscience 17(07): 63–79.
Blomstedt, Patric, Ulrika Sandvik, and Stephen Tisch. 2010. “Deep Brain Stimulation in the
Posterior Subthalamic Area in the Treatment of Essential Tremor.” Movement Disorders 25(10):
1350–56.
Boyden, Edward S et al. 2005. “Millisecond-Timescale, Genetically Targeted Optical Control of
Neural Activity.” Nature Neuroscience 8(9): 1263–68.
Bray, Natasha. 2016. “Behavioural Neuroscience: On the Defensive.” Nature Reviews
Neuroscience 17(8): 463.
Cang, Jianhua, Elise Savier, Jad Barchini, and Xiaorong Liu. 2018. “Visual Function,
Organization, and Development of the Mouse Superior Colliculus.” Annual review of vision
science 4: 239–62.
Canteras, Newton Sabino, Eloisa Pavesi, and Antonio Padua Carobrez. 2015. “Olfactory
Instruction for Fear: Neural System Analysis.” Frontiers in Neuroscience 9(JUL): 1–10.
Capelli, Paolo, Chiara Pivetta, Maria Soledad Esposito, and Silvia Arber. 2017. “Locomotor Speed
Control Circuits in the Caudal Brainstem.” Nature.
Cappe, Céline, Anne Morel, Pascal Barone, and Eric M. Rouiller. 2009. “The Thalamocortical
Projection Systems in Primate: An Anatomical Support for Multisensory and Sensorimotor
Interplay.” Cerebral Cortex 19(9): 2025–37.
Carandini, Matteo. 2000. “Visual Cortex: Fatigue and Adaptation.” Current Biology.
Chalupa, Leo M., and Stephen E. Fish. 1978. “Response Characteristics of Visual and Extravisual
Neurons in the Pulvinar and Lateral Posterior Nuclei of the Cat.” Experimental Neurology 61(1):
96–120.
Chen, Jiadong, and Arnold R Kriegstein. 2015. “A GABAergic Projection from the Zona Incerta
to Cortex Promotes Cortical Neuron Development.” Science (New York, N.Y.) 350(6260): 554–58.
98

Chow, Brian Y. et al. 2010. “High-Performance Genetically Targetable Optical Neural Silencing
by Light-Driven Proton Pumps.” Nature 463(7277): 98–102.
Ciocchi, Stephane et al. 2010. “Encoding of Conditioned Fear in Central Amygdala Inhibitory
Circuits.” Nature 468(7321): 277–82.
Comoli, Eliane et al. 2012. “Segregated Anatomical Input to Sub-Regions of the Rodent Superior
Colliculus Associated with Approach and Defense.” Frontiers in Neuroanatomy.
Dean, P., I. J. Mitchell, and P. Redgrave. 1988. “Responses Resembling Defensive Behaviour
Produced by Microinjection of Glutamate into Superior Colliculus of Rats.” Neuroscience.
Dean, P., P. Redgrave, and G. W.M. Westby. 1989. “Event or Emergency? Two Response Systems
in the Mammalian Superior Colliculus.” Trends in Neurosciences.
Demb, Jonathan B. 2008. “Functional Circuitry of Visual Adaptation in the Retina.” In Journal of
Physiology,.
Deng, Hanfei, Xiong Xiao, and Zuoren Wang. 2016. “Periaqueductal Gray Neuronal Activities
Underlie Different Aspects of Defensive Behaviors.” Journal of Neuroscience.
Dominguez-Vargas, Adan-Ulises, Lukas Schneider, Melanie Wilke, and Igor Kagan. 2017.
“Electrical Microstimulation of the Pulvinar Biases Saccade Choices and Reaction Times in a
Time-Dependent Manner.” The Journal of neuroscience : the official journal of the Society for
Neuroscience 37(8): 2234–57.
Dong, Ping et al. 2019. “A Novel Cortico-Intrathalamic Circuit for Flight Behavior.” Nature
Neuroscience 22(6): 941–49.
Dorris, Michael C., Martin Paré, and Douglas P. Munoz. 1997. “Neuronal Activity in Monkey
Superior Colliculus Related to the Initiation of Saccadic Eye Movements.” Journal of
Neuroscience.
Drager, U. C., and D. H. Hubel. 1975. “Responses to Visual Stimulation and Relationship between
Visual, Auditory, and Somatosensory Inputs in Mouse Superior Colliculus.” Journal of
Neurophysiology 38(3): 690–713.
Duvarci, Sevil, Daniela Popa, and Denis Paré. 2011. “Central Amygdala Activity during Fear
Conditioning.” Journal of Neuroscience 31(1): 289–94.
Ehrlich, Ingrid et al. 2009. “Amygdala Inhibitory Circuits and the Control of Fear Memory.”
Neuron 62(6): 757–71.
Eilam, David. 2005. “Die Hard: A Blend of Freezing and Fleeing as a Dynamic Defense -
Implications for the Control of Defensive Behavior.” Neuroscience and Biobehavioral Reviews
29(8): 1181–91.
99

Eilam, David, Rony Izhar, and Joel Mort. 2011. “Threat Detection: Behavioral Practices in
Animals and Humans.” Neuroscience and Biobehavioral Reviews 35(4): 999–1006.
Evans, Dominic A. et al. 2018. “A Synaptic Threshold Mechanism for Computing Escape
Decisions.” Nature.
Evans, Dominic A., A. Vanessa Stempel, Ruben Vale, and Tiago Branco. 2019. “Cognitive
Control of Escape Behaviour.” Trends in Cognitive Sciences 23(4): 334–48.
Fanselow, Michael S. 1994. “Neural Organization of the Defensive Behavior System Responsible
for Fear.” Psychonomic Bulletin & Review 1(4): 429–38.
———. 2018. “The Role of Learning in Threat Imminence and Defensive Behaviors.” Current
Opinion in Behavioral Sciences 24: 44–49.
Ferreira-Netto, Cristina, Karina Genaro Borelli, and Marcus Lira Brandão. 2005. “Neural
Segregation of Fos-Protein Distribution in the Brain Following Freezing and Escape Behaviors
Induced by Injections of Either Glutamate or NMDA into the Dorsal Periaqueductal Gray of Rats.”
Brain Research 1031(2): 151–63.
Forel, August. 1877. “Untersuchungen Über Die Haubenregion Und Ihre Oberen Verknüpfungen
Im Gehirne Des Menschen Und Einiger Säugethiere, Mit Beiträgen Zu Den Methoden Der
Gehirnuntersuchung.” Archiv für Psychiatrie und Nervenkrankheiten 7(3): 393–495.
Gale, Samuel D, and Gabe J Murphy. 2014. “Distinct Representation and Distribution of Visual
Information by Specific Cell Types in Mouse Superficial Superior Colliculus.” The Journal of
neuroscience : the official journal of the Society for Neuroscience 34(40): 13458–71.
Gattass, Ricardo, Aglai P.B. Sousa, and Eduardo Oswaldo-Cruz. 1978. “Single Unit Response
Types in the Pulvinar of the Cebus Monkey to Multisensory Stimulation.” Brain Research 158(1):
75–87.
Gatto, Graziana, and Martyn Goulding. 2018. “Locomotion Control: Brainstem Circuits Satisfy
the Need for Speed.” Current Biology.
Gentet, Luc J. et al. 2010. “Membrane Potential Dynamics of GABAergic Neurons in the Barrel
Cortex of Behaving Mice.” Neuron.
Grofová, Irena, Ole Petter Ottersen, and Eric Rinvik. 1978. “Mesencephalic and Diencephalic
Afferents to the Superior Colliculus and Periaqueductal Gray Substance Demonstrated by
Retrograde Axonal Transport of Horseradish Peroxidase in the Cat.” Brain Research 146(2): 205–
20.
Guillery, R. W., and S. Murray Sherman. 2002. “Thalamic Relay Functions and Their Role in
Corticocortical Communication: Generalizations from the Visual System.” Neuron 33(2): 163–75.
100

Hackett, T.A., I. Stepniewska, and J.H. Kaas. 1998. “Thalamocortical Connections of the Parabelt
Auditory Cortex in Macaque Monkeys.” The Journal of Comparative Neurology 400(2): 271–86.
Hafed, Ziad M. 2018. “Superior Colliculus: A Vision for Orienting.” Current Biology.
Halassa, Michael M., and S. Murray Sherman. 2019. “Thalamocortical Circuit Motifs: A General
Framework.” Neuron 103(5): 762–70.
Hanes, Doug P., and Robert H. Wurtz. 2001. “Interaction of the Frontal Eye Field and Superior
Colliculus for Saccade Generation.” Journal of Neurophysiology.
Harris, Kenneth D. et al. 2001. “Temporal Interaction between Single Spikes and Complex Spike
Bursts in Hippocampal Pyramidal Cells.” Neuron 32(1): 141–49.
Harting, J. K., K. K. Glendenning, I. T. Diamond, and W. C. Hall. 1973. “Evolution of the Primate
Visual System: Anterograde Degeneration Studies of the Tecto-Pulvinar System.” American
Journal of Physical Anthropology 38(2): 383–92.
Harting, J.K., W.C. Hall, and I.T. Diamond. 1972. “Evolution of the Pulvinar (Part 1 of 2).” Brain,
Behavior and Evolution 6(1–6): 424–38.
Headley, Drew B., Vasiliki Kanta, Pinelopi Kyriazi, and D. Paré. 2019. “Embracing Complexity
in Defensive Networks.” Neuron 103(2): 189–201.
Herry, Cyril et al. 2010. “Neuronal Circuits of Fear Extinction.” European Journal of
Neuroscience 31(4): 599–612.
Huerta, Michael F., and John K. Harting. 1984. “Connectional Organization of the Superior
Colliculus.” Trends in Neurosciences.
Hurley, Karen M., Horst Herbert, Margaret M. Moga, and Clifford B. Saper. 1991. “Efferent
Projections of the Infralimbic Cortex of the Rat.” Journal of Comparative Neurology 308(2): 249–
76.
Ibrahim, Leena A. et al. 2016. “Cross-Modality Sharpening of Visual Cortical Processing through
Layer-1-Mediated Inhibition and Disinhibition.” Neuron 89(5): 1031–45.
Ito, Shinya, and David A. Feldheim. 2018. “The Mouse Superior Colliculus: An Emerging Model
for Studying Circuit Formation and Function.” Frontiers in Neural Circuits 12(February): 1–11.
Janak, Patricia H., and Kay M. Tye. 2015. “From Circuits to Behaviour in the Amygdala.” Nature
517(7534): 284–92.
Ji, Jinzhao, and Stephen Maren. 2007. “Hippocampal Involvement in Contextual Modulation of
Fear Extinction.” Hippocampus.
101

———. 2008. “Lesions of the Entorhinal Cortex or Fornix Disrupt the Context-Dependence of
Fear Extinction in Rats.” Behavioural Brain Research.
Ji, Xu Ying et al. 2016. “Thalamocortical Innervation Pattern in Mouse Auditory and Visual
Cortex: Laminar and Cell-Type Specificity.” Cerebral Cortex 26(6): 2612–25.
Jiang, Xiaolong et al. 2013. “The Organization of Two New Cortical Interneuronal Circuits.”
Nature neuroscience 16(2): 210–18.
Juavinett, Ashley L., Euiseok J. Kim, Hannah C. Collins, and Edward M. Callaway. 2019. “A
Systematic Topographical Relationship between Mouse Lateral Posterior Thalamic Neurons and
Their Visual Cortical Projection Targets.” Journal of Comparative Neurology: cne.24737.
Kaas, Jon H, and David C Lyon. 2007. “Pulvinar Contributions to the Dorsal and Ventral Streams
of Visual Processing in Primates.” Brain research reviews 55(2): 285–96.
Kato, Hiroyuki K., Shea N. Gillet, and Jeffry S. Isaacson. 2015. “Flexible Sensory Representations
in Auditory Cortex Driven by Behavioral Relevance.” Neuron 88(5): 1027–39.
Khan, Sadaquate et al. 2011. “Outcomes from Stimulation of the Caudal Zona Incerta and
Pedunculopontine Nucleus in Patients with Parkinson’s Disease.” British Journal of Neurosurgery
25(2): 273–80.
Kim, Jeansok J., and Min Whan Jung. 2006. “Neural Circuits and Mechanisms Involved in
Pavlovian Fear Conditioning: A Critical Review.” Neuroscience and Biobehavioral Reviews
30(2): 188–202.
King, A. J., and A. R. Palmer. 1985. “Integration of Visual and Auditory Information in Bimodal
Neurones in the Guinea-Pig Superior Colliculus.” Experimental Brain Research 60(3): 492–500.
Kohn, Adam. 2007. “Visual Adaptation: Physiology, Mechanisms, and Functional Benefits.”
Journal of Neurophysiology 97(5): 3155–64.
Kolmac, Christian, and John Mitrofanis. 1999. “Distribution of Various Neurochemicals within
the Zona Incerta: An Immunocytochemical and Histochemical Study.” Anatomy and Embryology
199(3): 265–80.
Kopec, Charles D. et al. 2015. “Cortical and Subcortical Contributions to Short-Term Memory for
Orienting Movements.” Neuron.
Krauzlis, Richard J., Lee P. Lovejoy, and Alexandre Zénon. 2013. “Superior Colliculus and Visual
Spatial Attention.” Annual Review of Neuroscience.
Kremkow, Jens, and Jose-Manuel Alonso. 2018. “Thalamocortical Circuits and Functional
Architecture.” Annual Review of Vision Science 4(1): 263–85.
102

Kunwar, Prabliat S. et al. 2015. “Ventromedial Hypothalamic Neurons Control a Defensive
Emotion State.” eLife.
De La Mothe, Lisa A., Suzanne Blumell, Yoshinao Kajikawa, and Troy A. Hackett. 2006.
“Thalamic Connections of the Auditory Cortex in Marmoset Monkeys: Core and Medial Belt
Regions.” The Journal of Comparative Neurology 496(1): 72–96.
Laurent, V., and R. F. Westbrook. 2009. “Inactivation of the Infralimbic but Not the Prelimbic
Cortex Impairs Consolidation and Retrieval of Fear Extinction.” Learning & Memory 16(9): 520–
29.
Lecca, Salvatore et al. 2017. “Aversive Stimuli Drive Hypothalamus-to-Habenula Excitation to
Promote Escape Behavior.” eLife 6: 1–16.
LeDoux, Joseph. 2012. “Rethinking the Emotional Brain.” Neuron 73(4): 653–76.
LeDue, Emily E., Jillian L. King, Kurt R. Stover, and Nathan A. Crowder. 2013. “Spatiotemporal
Specificity of Contrast Adaptation in Mouse Primary Visual Cortex.” Frontiers in Neural Circuits.
Lee, A. Moses et al. 2014. “Identification of a Brainstem Circuit Regulating Visual Cortical State
in Parallel with Locomotion.” Neuron.
Lee, Choongkil, William H. Rohrer, and David L. Sparks. 1988. “Population Coding of Saccadic
Eye Movements by Neurons in the Superior Colliculus.” Nature.
Lee, SooHyun, George E Carvell, and Daniel J Simons. 2008. “Motor Modulation of Afferent
Somatosensory Circuits.” Nature Neuroscience 11(12): 1430–38.
Letzkus, Johannes J. et al. 2011. “A Disinhibitory Microcircuit for Associative Fear Learning in
the Auditory Cortex.” Nature 480(7377): 331–35.
Li, Yi et al. 2018. “Hypothalamic Circuits for Predation and Evasion.” Neuron.
Liang, Feixue et al. 2014. “Thresholding of Auditory Cortical Representation by Background
Noise.” Frontiers in Neural Circuits 8.
———. 2015. “Sensory Cortical Control of a Visually Induced Arrest Behavior via Corticotectal
Projections.” Neuron 86(3): 755–67.
———. 2018. “Sparse Representation in Awake Auditory Cortex: Cell-Type Dependence,
Synaptic Mechanisms, Developmental Emergence, and Modulation.” Cerebral cortex (New York,
N.Y. : 1991).
Liang, Huazheng, George Paxinos, and Charles Watson. 2011. “Projections from the Brain to the
Spinal Cord in the Mouse.” Brain Structure and Function.
103

———. 2012. “Spinal Projections from the Presumptive Midbrain Locomotor Region in the
Mouse.” Brain Structure and Function.
Liu, Kai et al. 2017. “Lhx6-Positive GABA-Releasing Neurons of the Zona Incerta Promote
Sleep.” Nature 548(7669): 582–87.
Liu, M. et al. 2011. “Orexin Gene Transfer into Zona Incerta Neurons Suppresses Muscle Paralysis
in Narcoleptic Mice.” Journal of Neuroscience 31(16): 6028–40.
Ma, Terence P., James C. Johnson, and Glenn A. Hoskins. 1997. “Organization of the Zona Incerta
in the Macaque: An Electron Microscopic Study.” Anatomical Record 249(2): 259–75.
Magariños-Ascone, C., W. Buño, and E. García-Austt. 1988. “Monkey Pulvinar Units Related to
Motor Activity and Sensory Response.” Brain Research 445(1): 30–38.
Mangieri, Leandra R. et al. 2019. “Defensive Behaviors Driven by a Hypothalamic-Ventral
Midbrain Circuit.” Eneuro 6(4): ENEURO.0156-19.2019.
McPeek, Robert M., and Edward L. Keller. 2004. “Deficits in Saccade Target Selection after
Inactivation of Superior Colliculus.” Nature Neuroscience.
Meredith, M A, and Barry E Stein. 1986. “Visual, Auditory, and Somatosensory Convergence on
Cells in Superior Colliculus Results in Multisensory Integration.” Journal of Neurophysiology
56(3): 640–62.
Milad, M.R. R, and G.J. J Quirk. 2002. “Neurons in Medial Prefrontal Cortex Signal Memory for
Fear Extinction.” Nature 420(6911): 70–74.
Milad, Mohammed R., and Gregory J. Quirk. 2002. “Neurons in Medial Prefrontal Cortex Signal
Memory for Fear Extinction.” Nature.
Miller, Samara M., Daniele Marcotulli, Angela Shen, and Larry S. Zweifel. 2019. “Divergent
Medial Amygdala Projections Regulate Approach–Avoidance Conflict Behavior.” Nature
Neuroscience.
Misslin, René. 2003. “The Defense System of Fear: Behavior and Neurocircuitry.”
Neurophysiologie Clinique.
Mitrofanis, J. 2005. “Some Certainty for the ‘Zone of Uncertainty’? Exploring the Function of the
Zona Incerta.” Neuroscience 130(1): 1–15.
Mitrofanis, John, Keyoumars Ashkan, Bradley A. Wallace, and Alim Louis Benabid. 2004.
“Chemoarchitectonic Heterogeneities in the Primate Zona Incerta: Clinical and Functional
Implications.” Journal of Neurocytology 33(4): 429–40.
Mitrofanis, John, and Lilijana Mikuletic. 1999. “Organisation of the Cortical Projection to the
104

Zona Incerta of the Thalamus.” Journal of Comparative Neurology 412(1): 173–85.
Monfils, M.-H., K. K. Cowansage, E. Klann, and J. E. LeDoux. 2009. “Extinction-Reconsolidation
Boundaries: Key to Persistent Attenuation of Fear Memories.” Science 324(5929): 951–55.
Mongeau, Raymond, Gabriel A. Miller, Elizabeth Chiang, and David J. Anderson. 2003. “Neural
Correlates of Competing Fear Behaviors Evoked by an Innately Aversive Stimulus.” Journal of
Neuroscience 23(9): 3855–68.
Moon, H.C., and Y.S. Park. 2017. “Reduced GABAergic Neuronal Activity in Zona Incerta Causes
Neuropathic Pain in a Rat Sciatic Nerve Chronic Constriction Injury Model.” Journal of Pain
Research 10: 1125–34.
Moraga-Amaro, Rodrigo, Andrés Cortés-Rojas, Felipe Simon, and Jimmy Stehberg. 2014. “Role
of the Insular Cortex in Taste Familiarity.” Neurobiology of learning and memory 109: 37–45.
Morgan, M. M., P. K. Whitney, and M. S. Gold. 1998. “Immobility and Flight Associated with
Antinociception Produced by Activation of the Ventral and Lateral/Dorsal Regions of the Rat
Periaqueductal Gray.” Brain Research.
Murray, Elisabeth A. 2007. “The Amygdala, Reward and Emotion.” Trends in Cognitive Sciences.
Myers, Karyn M., and Michael Davis. 2002. “Behavioral and Neural Analysis of Extinction.”
Neuron 36(4): 567–84.
Nakamura, Hisashi, Hiroyuki Hioki, Takahiro Furuta, and Takeshi Kaneko. 2015. “Different
Cortical Projections from Three Subdivisions of the Rat Lateral Posterior Thalamic Nucleus: A
Single-Neuron Tracing Study with Viral Vectors.” European Journal of Neuroscience 41(10):
1294–1310.
Nicolelis, Miguel A L, John K. Chapin, and Rick C S Lin. 1992. “Somatotopic Maps within the
Zona Incerta Relay Parallel GABAergic Somatosensory Pathways to the Neocortex, Superior
Colliculus, and Brainstem.” Brain Research 577(1): 134–41.
Niell, Cristopher M., and Michael P. Stryker. 2010. “Modulation of Visual Responses by
Behavioral State in Mouse Visual Cortex.” Neuron.
Oertel, W. H., M. L. Tappaz, A. Berod, and E. Mugnaini. 1982. “Two-Color
Immunohistochemistry for Dopamine and GABA Neurons in Rat Substantia Nigra and Zona
Incerta.” Brain Research Bulletin 9(1–6): 463–74.
Oh, Seung Wook et al. 2014. “A Mesoscale Connectome of the Mouse Brain.” Nature 508(7495):
207–14.
Pape, Hans Christian, and Denis Pare. 2010. “Plastic Synaptic Networks of the Amygdala for the
Acquisition, Expression, and Extinction of Conditioned Fear.” Physiological Reviews 90(2): 419–
105

63.
Paré, D., Y. Smith, A. Parent, and M. Steriade. 1988. “Projections of Brainstem Core Cholinergic
and Non-Cholinergic Neurons of Cat to Intralaminar and Reticular Thalamic Nuclei.”
Neuroscience 25(1): 69–86.
Pérez-Gómez, Anabel et al. 2015. “Innate Predator Odor Aversion Driven by Parallel Olfactory
Subsystems That Converge in the Ventromedial Hypothalamus.” Current Biology 25(10): 1340–
46.
Perier, C, L Tremblay, J Feger, and E C Hirsch. 2002. “Behavioral Consequences of Bicuculline
Injection in the Subthalamic Nucleus and the Zona Incerta in Rat.” Journal of Neuroscience
22(1529-2401 (Electronic)): 8711–19.
Peters, J., L. M. Dieppa-Perea, L. M. Melendez, and G. J. Quirk. 2010. “Induction of Fear
Extinction with Hippocampal-Infralimbic BDNF.” Science 328(5983): 1288–90.
Petreanu, Leopoldo, Tianyi Mao, Scott M. Sternson, and Karel Svoboda. 2009a. “The Subcellular
Organization of Neocortical Excitatory Connections.” Nature 457(7233): 1142–45.
Petreanu, Leopoldo, Tianyi Mao, Scott M Sternson, and Karel Svoboda. 2009b. “The Subcellular
Organization of Neocortical Excitatory Connections.” Nature 457(7233): 1142–45.
Plaha, P, S Khan, and S S Gill. 2008. “Bilateral Stimulation of the Caudal Zona Incerta Nucleus
for Tremor Control.” Journal of Neurology, Neurosurgery & Psychiatry 79(5): 504–13.
Plaha, Puneet, Y. Ben-Shlomo, Nikunj K. Patel, and Steven S. Gill. 2006. “Stimulation of the
Caudal Zona Incerta Is Superior to Stimulation of the Subthalamic Nucleus in Improving
Contralateral Parkinsonism.” Brain 129(7): 1732–47.
Polack, Pierre Olivier, Jonathan Friedman, and Peyman Golshani. 2013. “Cellular Mechanisms of
Brain State-Dependent Gain Modulation in Visual Cortex.” Nature Neuroscience.
Poulet, James F.A., and Carl C.H. Petersen. 2008. “Internal Brain State Regulates Membrane
Potential Synchrony in Barrel Cortex of Behaving Mice.” Nature.
Quirk, Gregory J, and Devin Mueller. 2008. “Neural Mechanisms of Extinction Learning and
Retrieval.” Neuropsychopharmacology 33(1): 56–72.
Ricardo, Juarez A. 1981. “Efferent Connections of the Subthalamic Region in the Rat. II. The Zona
Incerta.” Brain Research 214(1): 43–60.
Romanski, L. M., and J. E. LeDoux. 1992. “Equipotentiality of Thalamo-Amygdala and Thalamo-
Cortico-Amygdala Circuits in Auditory Fear Conditioning.” Journal of Neuroscience 12(11):
4501–9.
106

Roth, Morgane M et al. 2016. “Thalamic Nuclei Convey Diverse Contextual Information to Layer
1 of Visual Cortex.” Nature neuroscience 19(2): 299–307.
Saalmann, Yuri B et al. 2012. “The Pulvinar Regulates Information Transmission between Cortical
Areas Based on Attention Demands.” Science (New York, N.Y.) 337(6095): 753–56.
Sahibzada, N., P. Dean, and P. Redgrave. 1986. “Movements Resembling Orientation or
Avoidance Elicited by Electrical Stimulation of the Superior Colliculus in Rats.” Journal of
Neuroscience.
Schenberg, L. C. et al. 2005. “Functional Specializations within the Tectum Defense Systems of
the Rat.” In Neuroscience and Biobehavioral Reviews,.
Schumacher, J. W., D. M. Schneider, and S. M. N. Woolley. 2011. “Anesthetic State Modulates
Excitability but Not Spectral Tuning or Neural Discrimination in Single Auditory Midbrain
Neurons.” Journal of Neurophysiology 106(2): 500–514.
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): 1026.
Shang, Congping et al. 2015. “A Parvalbumin-Positive Excitatory Visual Pathway to Trigger Fear
Responses in Mice.” Science.
———. 2018. “Divergent Midbrain Circuits Orchestrate Escape and Freezing Responses to
Looming Stimuli in Mice.” Nature Communications 9(1): 1–17.
Sherman, S Murray. 2016. “Thalamus Plays a Central Role in Ongoing Cortical Functioning.”
Nature neuroscience 19(4): 533–41.
Shipp, S. 2001. “Corticopulvinar Connections of Areas V5, V4, and V3 in the Macaque Monkey:
A Dual Model of Retinal and Cortical Topographies.” The Journal of comparative neurology
439(4): 469–90.
Shipp, Stewart. 2007. “Structure and Function of the Cerebral Cortex.” Current Biology 17(12):
R443–49.
Sierra-Mercado, Demetrio, Nancy Padilla-Coreano, and Gregory J Quirk. 2011. “Dissociable
Roles of Prelimbic and Infralimbic Cortices, Ventral Hippocampus, and Basolateral Amygdala in
the Expression and Extinction of Conditioned Fear.” Neuropsychopharmacology 36(2): 529–38.
Silva, Bianca A. et al. 2016. “The Ventromedial Hypothalamus Mediates Predator Fear Memory.”
European Journal of Neuroscience 43(11): 1431–39.
Silva, Bianca A., Cornelius T. Gross, and Johannes Gräff. 2016. “The Neural Circuits of Innate
Fear: Detection, Integration, Action, and Memorization.” Learning and Memory 23(10): 544–55.
107

Soares, Sandra C et al. 2017. “Fast Detector/First Responder: Interactions between the Superior
Colliculus-Pulvinar Pathway and Stimuli Relevant to Primates.” Frontiers in neuroscience 11: 67.
Sotres-Bayon, Francisco, Demetrio Sierra-Mercado, Enmanuelle Pardilla-Delgado, and Gregory
J. Quirk. 2012. “Gating of Fear in Prelimbic Cortex by Hippocampal and Amygdala Inputs.”
Neuron.
Stepniewska, I, H X Ql, and J H Kaas. 2000. “Projections of the Superior Colliculus to
Subdivisions of the Inferior Pulvinar in New World and Old World Monkeys.” Visual
neuroscience 17(4): 529–49.
Stitt, Iain, Zhe Charles Zhou, Susanne Radtke-Schuller, and Flavio Fröhlich. 2018. “Arousal
Dependent Modulation of Thalamo-Cortical Functional Interaction.” Nature Communications
9(1): 2455.
Stubblefield, Elizabeth A., Jamie D. Costabile, and Gidon Felsen. 2013. “Optogenetic
Investigation of the Role of the Superior Colliculus in Orienting Movements.” Behavioural Brain
Research.
Sutter, M L, and C E Schreiner. 1991. “Physiology and Topography of Neurons with Multipeaked
Tuning Curves in Cat Primary Auditory Cortex.” Journal of neurophysiology 65(5): 1207–26.
Swanson, Larry W., Graciela Sanchez-Watts, and Alan G. Watts. 2005. “Comparison of Melanin-
Concentrating Hormone and Hypocretin/Orexin MRNA Expression Patterns in a New Parceling
Scheme of the Lateral Hypothalamic Zone.” Neuroscience Letters 387(2): 80–84.
Tamamaki, Nobuaki et al. 2003. “Green Fluorescent Protein Expression and Colocalization with
Calretinin, Parvalbumin, and Somatostatin in the GAD67-GFP Knock-In Mouse.” Journal of
Comparative Neurology 467(1): 60–79.
Tervo, D. Gowanlock R. et al. 2016. “A Designer AAV Variant Permits Efficient Retrograde
Access to Projection Neurons.” Neuron 92(2): 372–82.
Tohmi, Manavu et al. 2014. “The Extrageniculate Visual Pathway Generates Distinct Response
Properties in the Higher Visual Areas of Mice.” Current Biology 24(6): 587–97.
Tovote, Philip et al. 2016. “Midbrain Circuits for Defensive Behaviour.” Nature 534(7606): 206–
12.
Tovote, Philip, Jonathan Paul Fadok, and Andreas Lüthi. 2015. “Neuronal Circuits for Fear and
Anxiety.” Nature Reviews Neuroscience.
Trageser, J. C. 2004. “Reducing the Uncertainty: Gating of Peripheral Inputs by Zona Incerta.”
Journal of Neuroscience 24(40): 8911–15.
Ullah, Farhad et al. 2015. “Relevance of Dorsomedial Hypothalamus, Dorsomedial Division of
108

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 293: 143–52.
Urbain, Nadia, and Martin Deschênes. 2007. “Motor Cortex Gates Vibrissal Responses in a
Thalamocortical Projection Pathway.” Neuron 56(4): 714–25.
Vianna, D. M L, and M. L. Brandão. 2003. “Anatomical Connections of the Periaqueductal Gray:
Specific Neural Substrates for Different Kinds of Fear.” Brazilian Journal of Medical and
Biological Research 36(5): 557–66.
Vinck, Martin, Renata Batista-Brito, Ulf Knoblich, and Jessica A. Cardin. 2015. “Arousal and
Locomotion Make Distinct Contributions to Cortical Activity Patterns and Visual Encoding.”
Neuron.
Wang, Haitao et al. 2019. “Direct Auditory Cortical Input to the Lateral Periaqueductal Gray
Controls Sound-Driven Defensive Behavior.” PLOS Biology.
Wang, Li et al. 2019. “Hypothalamic Control of Conspecific Self-Defense.” Cell Reports.
Wang, Li, Irene Z. Chen, and Dayu Lin. 2015. “Collateral Pathways from the Ventromedial
Hypothalamus Mediate Defensive Behaviors.” Neuron 85(6): 1344–58.
http://dx.doi.org/10.1016/j.neuron.2014.12.025.
Wark, Barry, Adrienne Fairhall, and Fred Rieke. 2009. “Timescales of Inference in Visual
Adaptation.” Neuron.
Wassum, Kate M., and Alicia Izquierdo. 2015. “The Basolateral Amygdala in Reward Learning
and Addiction.” Neuroscience and Biobehavioral Reviews.
Watson, Glenn D. R., Jared B. Smith, and Kevin D. Alloway. 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):
9463–76.
Wei, Pengfei et al. 2015. “Processing of Visually Evoked Innate Fear by a Non-Canonical
Thalamic Pathway.” Nature Communications 6: 1–12.
White, Brian J., and Douglas P. Munoz. 2012. “The Superior Colliculus.” In The Oxford Handbook
of Eye Movements,.
Winer, Jeffery A., Lee M. Miller, Charles C. Lee, and Christoph E. Schreiner. 2005. “Auditory
Thalamocortical Transformation: Structure and Function.” Trends in Neurosciences 28(5): 255–
63.
Wise, L. Z., and D. R. Irvine. 1983. “Auditory Response Properties of Neurons in Deep Layers of
109

Cat Superior Colliculus.” Journal of Neurophysiology 49(3): 674–85.
Wolff, Steffen B.E. et al. 2014. “Amygdala Interneuron Subtypes Control Fear Learning through
Disinhibition.” Nature.
Wong, Peiyan, Christine E. Collins, Mary K.L. Baldwin, and Jon H. Kaas. 2009. “Cortical
Connections of the Visual Pulvinar Complex in Prosimian Galagos ( Otolemur Garnetti ).” The
Journal of Comparative Neurology 517(4): 493–511.
Woody, CD, O Melamed, and V Chizhevsky. 1991. “Patterns of Activity Coding Discrimination
of Auditory Stimuli Differ between Mid- and Posterolateral Thalamus of Cats.” The Journal of
Neuroscience 11(11): 3379–87.
Xiong, Xiaorui R. et al. 2013. “Interaural Level Difference-Dependent Gain Control and Synaptic
Scaling Underlying Binaural Computation.” Neuron 79(4): 738–53.
———. 2015. “Auditory Cortex Controls Sound-Driven Innate Defense Behaviour through
Corticofugal Projections to Inferior Colliculus.” Nature Communications 6: 7224.
Yang, Hongbin et al. 2016. “Laterodorsal Tegmentum Interneuron Subtypes Oppositely Regulate
Olfactory Cue-Induced Innate Fear.” Nature Neuroscience 19(2): 283–89.
Yilmaz, Melis, and Markus Meister. 2013. “Rapid Innate Defensive Responses of Mice to
Looming Visual Stimuli.” Current Biology.
Yirmiya, Raz, and Shraga Hocherman. 1987. “Auditory- and Movement-Related Neural Activity
Interact in the Pulvinar of the Behaving Rhesus Monkey.” Brain Research 402(1): 93–102.
Yu, Kai, Pedro Garcia da Silva, Dinu F. Albeanu, and Bo Li. 2016. “Central Amygdala
Somatostatin Neurons Gate Passive and Active Defensive Behaviors.” Journal of Neuroscience
36(24): 6488–96.
Zhang, Guang-Wei et al. 2018. “Transforming Sensory Cues into Aversive Emotion via Septal-
Habenular Pathway.” Neuron 99(5): 1016-1028.e5.
Zhang, Xiaobing, and Anthony N. van den Pol. 2017. “Rapid Binge-like Eating and Body Weight
Gain Driven by Zona Incerta GABA Neuron Activation.” Science 356(6340): 853–59.
Zhao, Xinyu, Mingna Liu, and Jianhua Cang. 2014. “Visual Cortex Modulates the Magnitude but
Not the Selectivity of Looming-Evoked Responses in the Superior Colliculus of Awake Mice.”
Neuron.
Zhou, Huihui, Robert John Schafer, and Robert Desimone. 2016. “Pulvinar-Cortex Interactions in
Vision and Attention.” Neuron 89(1): 209–20.
Zhou, Mu et al. 2014. “Scaling down of Balanced Excitation and Inhibition by Active Behavioral
110

States in Auditory Cortex.” Nature neuroscience 17(6): 841–50.
Zhou, Na et al. 2018. “The Mouse Pulvinar Nucleus Links the Lateral Extrastriate Cortex,
Striatum, and Amygdala.” The Journal of neuroscience : the official journal of the Society for
Neuroscience 38(2): 347–62.
Zingg, Brian et al. 2017. “AAV-Mediated Anterograde Transsynaptic Tagging: Mapping
Corticocollicular Input-Defined Neural Pathways for Defense Behaviors.” Neuron 93(1): 33–47.

Abstract (if available)

Abstract Defensive behaviors are critical for animals’ survival. Animals need to adopt the proper type as well as the proper level of defensive behaviors when encountering any specific danger. Previous studies have already shown many brain structures involved in controlling the defensive behaviors in various conditions. However, the understanding of detailed neural circuits governing the cortical control and modulation of defensive behaviors is still lacking. In addition, it is also unknown how the sensory processing would be influenced by the defense circuits. Here, I will present three studies demonstrating part of the neural circuits for the control and modulation of defensive behaviors and its impact on sensory processing in mice. ❧ In the first study, we investigated how superior colliculus (SC) neurons control different types of defensive behavior given a specific sensory cortical input. We combined anterograde transsynaptic tagging and optogenetics to manipulate the SC neurons receiving inputs from either the primary auditory cortex (A1) or the primary visual cortex (V1). We found that SC neurons receiving A1 inputs mediated the flight behavior, while SC neurons receiving V1 inputs controlled the freezing behavior through its downstream target, lateral posterior nucleus of the thalamus (LP). These results suggested that different subpopulations of SC neurons receiving specific cortical inputs could control different types of defensive behaviors. ❧ In the second study, we used techniques including neural tracing, optogenetics and in-vivo electrophysiological recordings to investigate a potential inhibitory nucleus for modulating the level of defensive behaviors. Here, we discovered that GABAergic neurons in the rostral sector of ZI (ZIr) directly innervated the excitatory but not the inhibitory neurons in both the dorsolateral and ventrolateral compartments of periaqueductal gray (PAG), which can drive flight or freezing behaviors respectively. Optogenetic activation of ZIr neurons or their projections to PAG reduced both sound-induced innate flight response and conditioned freezing response, while optogenetic suppression of these neurons enhanced these defensive behaviors, likely through a mechanism of gain modulation. In addition, ZIr activity progressively increased during extinction of conditioned freezing response and suppressing ZIr activity impaired the expression of fear extinction. Furthermore, ZIr was innervated by the medial prefrontal cortex (mPFC), and silencing mPFC prevented the increase of ZIr activity during extinction and the expression of fear extinction. These results suggest that ZIr is engaged in modulating the level of various types of defensive behaviors. ❧ In the third study, we explored the potential role of LP in auditory processing in A1. Here, we modulated LP activity or its projection to primary auditory cortex (A1) in awake mice and found that LP improved auditory processing in A1 supragranular-layer neurons by sharpening their frequency selectivity and increasing the signal-to-noise ratio (SNR) of their responses. This was achieved through a universal suppression of cortical responses to both tone and background noise, which is mediated largely by LP’s projection to inhibitory neurons in superficial layers of A1. Providing the major bottom-up input, SC can drive the LP-mediated modulation of A1 responses, which alleviates the deterioration of A1 processing by increasing background noise levels. Our results suggest that the SC-LP-A1 pathway may play a role in modulation of auditory cortical processing during SC-LP related defensive behaviors. ❧ Together, these studies demonstrate the neural circuits for the control, modulation as well as the potential sensory impact of defensive behaviors. They fill the gap in our understanding of the overall defense circuits and provide many potential directions to further investigate the defensive behaviors from different aspects.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Neural circuits control and modulate innate defensive behaviors

PDF

Contextual modulation of sensory processing via the pulvinar nucleus

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

Sensory information processing by retinothalamic neural circuits

PDF

Axon guidance cues in development of the mammalian auditory circuit

PDF

Functional properties of the superficial cortical interneurons

PDF

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

PDF

Synaptic circuits for information processing along the central auditory pathway

PDF

Functional synaptic circuits in primary visual cortex

PDF

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

PDF

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

PDF

Cell-type specialization of layer 5 excitatory neuron functions in tactile behavior underlying object localization

PDF

The behavioral and neural bases of tactile object localization

PDF

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

PDF

Defining the circuits and mechanisms mediating a pacemaker-controlled behavior in C. elegans

PDF

Mapping multi-scale connectivity of the mouse posterior parietal cortex

PDF

Spatial and temporal precision of inhibitory and excitatory neurons in the murine dorsal lateral geniculate nucleus

PDF

Quantitative analysis of mouse medial prefrontal cortex whole brain structural connectivity

Asset Metadata

Creator Chou, Xiaolin (author)

Core Title Neural circuits underlying the modulation and impact of defensive behaviors

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Neuroscience

Publication Date 01/27/2020

Defense Date 11/22/2019

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

Tag defensive behaviors,lateral posterior nucleus of thalamus,neural circuits,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 (committee member), Zhang, Li (committee member)

Creator Email xchou@usc.edu,xiaolinchou@gmail.com

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c89-263130

Unique identifier UC11673179

Identifier etd-ChouXiaoli-8133.pdf (filename),usctheses-c89-263130 (legacy record id)

Legacy Identifier etd-ChouXiaoli-8133.pdf

Dmrecord 263130

Document Type Dissertation

Rights Chou, Xiaolin

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