Optogenetic strategies to investigate neural circuitry engaged by stress

Abstract

Optogenetic techniques have given researchers unprecedented access to the function of discrete neural circuit elements and have been instrumental in the identification of novel brain pathways that become dysregulated in neuropsychiatric diseases. For example, stress is integrally linked to the manifestation and pathophysiology of neuropsychiatric illness, including anxiety, addiction and depression. Due to the heterogeneous populations of genetically and neurochemically distinct neurons in areas such as the bed nucleus of the stria terminalis (BNST), as well as their substantial number of projections, our understanding of how neural circuits become disturbed after stress has been limited. Using optogenetic tools, we are now able to selectively isolate distinct neural circuits that contribute to these disorders and perturb these circuits in vivo, which in turn may lead to the normalization of maladaptive behavior. This review will focus on current optogenetic strategies to identify, manipulate, and record from discrete neural circuit elements in vivo as well as highlight recent optogenetic studies that have been utilized to parcel out BNST function.

Introduction

Stress is defined as the body’s response to any demands for change¹. Although stress can be positive, as it forces an organism to adapt in order to survive, the neurophysiological components of the stress response often times become disturbed in neuropsychiatric illnesses, such as depression, generalized anxiety disorder, post traumatic stress disorder and drug and alcohol addiction^2,3,4,5. Thus, in order to develop novel and effective treatments, a critical understanding of how discrete neural circuit elements are altered through stress-producing stimuli is essential.

The extended amygdala has been implicated in rodent and human studies as a crucial mediator of the behavioral effects of aversive stimuli and both acute and chronic stress^6,7,8,9,10. The region is comprised of the bed nucleus of the stria terminalis (BNST), the central (CeA) and medial (MeA) nucleus of the amygdala, and the shell of the nucleus accumbens (NAc)¹¹. Since the extended amygdala encapsulates many different structures and cell types, which are intermingled within these regions, the neurophysiological properties and behavioral importance of genetically distinct neural circuits mediating stress and anxiety remain elusive. This review will focus on current optogenetic strategies that can be used to examine neural circuit function from synapse to behavior. In addition, this review will also highlight recent optogenetic strategies that have been used to dissect the BNST, a region that has been implicated in the integration and processing of the stress response¹⁰.

Optogenetics

Introduction of the light-gated cation channel channelrhdopsin-2 (ChR2) or chloride and proton pumps such as halorhodpson (NpHR) and archaerhodopsin (Arch) into genetically defined neural tissues has revolutionized neuroscience^12,13,14,15. Opsin proteins can be delivered into mammalian brain tissue through viral vectors that contain cell specific promoters, such as calcium-calmodulin dependent protein kinase IIα (CaMKIIα)^16,17,18. Additionally, transgenic animals that express cre-recombinase in defined populations of neurons can be injected with viral vectors encoding a cre-inducible opsin protein, to express opsins in only neurons that contain cre-recombinase^19,20,21. This approach has been used to target discrete neuronal subtypes in the ventral tegmental area (VTA). For example, infusion of a cre-inducible ChR2 into the VTA of Vgat-cre mice transduced only GABAergic neurons²¹. Thus, promoter driven and cre-dependent viral strategies allow for the precise control of neurochemically discrete cell bodies and terminals in heterogeneous brain tissue. Unlike other neuroscience approaches such as lesions and pharmacology, optogenetics allows for more temporally controlled perturbations of distinct neural circuit pathways in awake and behaving animals. Furthermore, the combination of optogenetic strategies and traditional pharmacological techniques can also increase the precision of neural circuit manipulations²². Importantly, optogenetic techniques can be utilized for long term behavioral manipulations, which is critical for identifying novel neural circuits that are involved in chronic stress. For a more thorough review of optogenetic procedures and principles see^23,14,22,24. This review will now focus on several new strategies that utilize optogenetic techniques to assist in the identification and modulation of discrete neural circuits in brain slices and behaving animals.

Optogenetic circuit mapping

ChR2-assisted circuit mapping allows for the characterization of functional connectivity between multiple neural substrates in heterogeneous tissue. With ChR2-assisted circuit mapping, axons and terminals from an input brain region can be optically stimulated even when excised from cell bodies, thus allowing for the examination of circuit connectivity in brain slices. In one of the first studies to utilize this procedure, Petreanu et al.²⁵ transduced pyramidal neurons in layers 2/3 of the somatosensory cortex with ChR2. Presynaptic terminals from this region were then optically stimulated in various cortical output areas to examine their synaptic input onto targeted postsynaptic neurons using slice electrophysiology. In these experiments, the authors found that layers 2/3 pyramidal neurons synapse onto pyramidal neurons in layers 2/3/5/6, a finding that would have been difficult to obtain using traditional electrophysiological techniques. In an elegant series of experiments, Atasoy et al.¹⁹, used ChR2-assisted circuit mapping to functionally dissect subcircuits within the arcuate nucleus of the hypothalamus (ARC) and their projection neurons in order to deconstruct neural circuits that control hunger and feeding. Initially, the authors transduced two populations of neurons: agouti-related peptide (AGRP)- and pro-opiomelanocortin (POMC)-containing cells, within the ARC with a cre-inducible ChR2 to test their functional connectivity. Using optogenetics in conjunction with patch-clamp electrophysiology, the authors demonstrated that AGRP-containing neurons formed functional synapses on POMC-containing neurons within the ARC and that optical stimulation of the ARC^AGRP-ARC^POMC circuit produced robust inhibitory post-synaptic currents. Furthermore, no synaptic responses were seen in ARC^AGRP-ARC^AGRP, ARC^POMC-ARC^AGRP, or ARC^POMC-ARC^POMC circuits. Additionally, the authors examined the behavioral consequences of stimulating AGRP-containing ARC neurons on two output areas: the paraventricular nucleus of the hypothalamus (PVH) and the parabrachial nucleus (PBN). In these studies, AGRP neurons were transduced with a cre-inducible ChR2 with optical fibers implanted above the PVH and PBN within the same animal. Optical stimulation of the ARC^AGRP-PVH pathway produced robust increases in food intake, whereas optogenetic activation of the ARC^AGRP-PBN pathway had no effect.

Although the previous studies examined the functional connectivity of neural circuit output regions, optogenetic techniques can also be used to examine synaptic input onto a brain region of interest. For example, viral vectors such as rabies, which allow for the retrograde transport of proteins, can be employed²⁶. Watabe-Uchida et al.,²⁷ used an EnvA-pseudotyoed, G-deleted rabies virus to visualize monosynaptic inputs to genetically defined neurons within the VTA in an exquisite series of experiments. In these studies, an AAV coding a cre-inducible avian receptor TVA, required for initial infection, and an AAV encoding rabies virus envelop glycoprotein (RG), necessary for transsynaptic spread, was injected into the VTA or substantia nigra reticulata (SNr) in dopamine transporter (DAT) -cre mice, to transduce only dopaminergic neurons expressing TVA. Approximately 2 weeks after the initial viral cocktail injection, the same mice were injected with the modified rabies virus. Since only neurons that expressed TVA were transduced by the rabies viral construct, inputs to dopamine neurons were visualized and quantified, providing a comprehensive map of midbrain dopaminergic circuitry. Furthermore, rabies viral vectors can also be packaged with cre and used in combination with optogenetic strategies in order to selectively control presynaptic inputs into a target region. Utilizing these techniques, Chaudhury et al.²⁸, injected a pseudorabies virus expressing cre (PRV-cre) into the NAc and an AAV coding a cre-inducible ChR2 into the VTA. Consequently, only neurons that expressed cre expressed a functional ChR2. Thus, the authors were able to optically stimulate VTA neurons that project to the NAC to manipulate behavior during a social-interaction task. Opsins can also be tethered to rabies viral vectors for trans-synaptic transduction. Lammel et al.²⁹ injected a rabies virus coding a cre-inducible ChR2 into the VTA of tyrosine hydroxylase (TH) -cre mice and implanted fibers into two VTA input structures, the laterodorsal tegmentum (LDT) and the lateral habenula (LHb). Due to the retrograde insertion of ChR2, the authors were able to control two distinct inputs onto VTA dopamine neurons. The authors found that optical stimulation of the LDT caused a conditioned place preference, whereas optogenetic activation of the LHb resulted in a conditioned place aversion, consistent with previous data³⁰. One caveat in using a rabies viral vector for retrograde transsynaptic labeling, however, is that the virus will cause cell death approximately 2 weeks post transduction²⁶. Thus, long-term behavioral manipulations are often not feasible. Herpes simplex viral vectors (HSV), which can also retrogradely transport proteins, such as opsins, may be beneficial for long term in vivo applications since cell death from viral infection is significantly lowered when compared to rabies. Additionally, HSV allows for rapid transduction (approx. 24 hr) into cell bodies, whereas AAV vectors take approximately 10–14 days for maximal expression. Figure 1 depicts a schematic summary detailing various strategies discussed for optogenetic circuit mapping.

Figure 1. Optogenetic strategies for circuit mapping.

a. Schematic detailing how to employ ChR2 assisted circuit mapping in slice electrophysiology. Presynaptic neurons are transduced with ChR2-eYFP. Optical stimulation (473 nm light to optimally activate ChR2) of transduced presynaptic terminals results in the detection of ChR2 mediated currents in the patch clamped postsynaptic neuron. Additionally, the recording pipette can be filled with dye to identify its neurochemical makeup. b. Schematic detailing how to employ a rabies viral delivery strategy in order to manipulate VTA neurons terminating in the NAc. Pseudorabies (PRV) that express a functional cre protein is injected into the NAc. PRV-cre is taken up by presynaptic terminals in the NAc. Infusion of a cre-inducible ChR2 into the VTA. Only neurons that express cre will express ChR2. An optical fiber is implanted above the VTA in order to manipulate VTA neurons that project to the NAc.

Optogenetic manipulation of projection neurons in vivo

Although a majority of optogenetic experiments have focused on controlling cell bodies to measure a behavioral phenotype^{31,32,33,34,35,36,20}, opsins can be trafficked from the soma to axon terminals for pathway specific stimulation or inhibition, as mentioned previously in the experiments conducted by Atasoy et al¹⁹. In another example, Stamatakis et al.³⁰ globally transduced lateral habenula (LHb) neurons with ChR2 and implanted a fiber optic above the rostromedial tegmental nucleus (RMTg) in order to stimulate the LHb-RMTg pathway during behavioral tasks. The authors found that optical stimulation of LHb-RMTg neural circuit resulted in active, passive, and conditioned behavioral avoidance, indicating that this pathway is aversive. One drawback for stimulating axon terminals in vivo using optogenetic manipulations is that ChR2-activation of axon terminals produces back-propagating action potentials, thereby activating transduced cell bodies, which in turn can activate axons to other brain regions. To minimize the potential contribution of network activity, pharmacological inactivation of ChR2-infected cell bodies, in conjunction with optical stimulation, during behavioral assays can be employed. In a study by Stuber et al.¹⁷, BLA glutamatergic efferents were transduced with ChR2 and a guide cannula was placed above the NAc for optogenetic activation of the BLA-NAc pathway. In these studies, mice reliably nose poked to receive optogenetic stimulation of BLA glutamatergic fibers terminating within the NAc. To further test the specificity of the BLA-NAc pathway in promoting reward seeking, the authors injected lidocaine into the BLA to inactivate ChR2-expressing cell bodies before the start of the nose-poking task. Infusions of lidocaine did not alter the acquisition or expression of reward-seeking behavior, indicating that the effects observed were not due to the influence of network mediated activity through the BLA. Another limitation of using optogenetic terminal stimulation in vivo is that fibers of passage from the transduced brain region may also be activated if they pass through the output target area. Therefore, to demonstrate that the behavioral effects of ChR2 terminal stimulation are specific to the circuit, and not from fibers of passage, traditional pharmacological manipulations can be incorporated with optogenetic strategies. As previously discussed, Atasoy et al.,¹⁹ optically stimulated inhibitory AGRP fibers originating in the ARC within the PVH, which increased food intake. In a separate series of experiments, the authors infused a GABA antagonist into the PVN during ARC^AGRP-PVN optical stimulation in a feeding paradigm. This resulted in a significant reduction of food intake, indicating the behavioral necessity of ARC^AGRP signaling within the PVN in promoting feeding.

Optogenetic identification of neurons in vivo

By combining optogenetics techniques and in vivo electrophysiology, real-time identification and recordings of distinct, genetically identified neurons is possible in heterogeneous brain tissue. In a recent study, Cohen et al.³⁷, recorded the activity of neurochemically distinct neurons in the VTA using optogenetic strategies in vivo. Here, the authors utilized the cre-dependent viral strategy in order to selectively target either dopaminergic or GABAergic neurons with ChR2 in DAT-cre and Vgat-cre mice, respectively. By coupling a fiber optic to a multielectrode array, which was implanted in the VTA, the authors revealed the identity of the recorded neurons using ChR2-mediated spiking during a classical conditioning task.

A recent study used a similar optical identification technique in order to identify genetically distinct circuit projection neurons. In these studies, Jennings, Sparta et al.³⁸ transduced glutamate and GABA neurons with ChR2 in the BNST of Vglut2-cre and Vgat-cre mice, respectively. Additionally, a multielectrode array was implanted in the BNST with an optical fiber above the VTA. By photostimulating terminals in the VTA, antidromic spiking was reliably detected within the BNST based on back propagating action potentials. Latencies in spike detection were then utilized to filter possible transsynaptic spikes from responses due to antidromic activation. Importantly, these experiments were completed in awake and behaving animals, which allowed for the real-time measurement of neuronal activity in optically identified VTA-projecting BNST glutamate and GABA during behavioral assays.

Strategies for optogenetic manipulation in vivo

Acutely implanted optical fibers were initially used to optogenetically perturb neural circuits in vivo. For this procedure, a fiber optic coupled to a laser was inserted into a guide cannula before any behavioral manipulations. However, these optical fibers often broke inside the cannula and repeated insertion of the optical fiber led to tissue damage. Therefore, the development of implantable optical fibers that can manipulate neural circuit elements with minimal tissue damage or change in light output over time has optimized optogenetic tools with long-term behavioral experimentation. These implanted optical fibers readily interface with in vivo electrophysiological arrays or electrochemical detection electrodes to provide a more thorough probing of neural circuits in vivo³⁹. Importantly, implantable optical fibers can penetrate into deep brain structures such as the VTA²¹. Light emitting diodes (LEDs) can also be used to deliver light to mammalian tissue. These devices have been used to photostimulate neurons transduced with ChR2 in the barrel cortex during a detection task⁴⁰. Recently, wireless LEDs have been developed, which allow for the transmission of light into the brain without the need to tether an animal to an optical patch cable⁴¹. These devices weigh between 2–3 g, allow the animal a full range of movement, and can deliver up to 2 W of power. Wentz et al.⁴¹ used wireless LEDs to deliver light in vivo to activate ChR2-transuced neurons within the motor cortex during a locomotor assay. One caveat with LEDs is the limited level of depth that light can penetrate into brain tissue. Studies using LEDs for optical stimulation in vivo have focused primarily on cortical regions although a recent study has utilized an injectable wireless LED in order to perturb deep brain structures such as the VTA⁴². These microLEDs are approximately 1000 × smaller than traditional LEDs and can provide better spatial resolution. Additionally, these devices can be coupled with electrode arrays that allows for concurrent collection of photostimulation and neuronal spiking from the same brain region. This review will now focus on optogenetic strategies used to dissect the BNST, an integral structure of the stress response.

BNST

The BNST is considered to be a connective locus between amygdaloid stress regions including the BLA and CeA and brain reward centers such as the VTA and NAc^{11,43,44,45,46,10}. The BNST can be subdivided into many sub-nuclei, including the oval, juxtacapsular, and rhomboid nuclei⁴⁷, yet consists primarily of GABAergic neurons⁴⁸. However, neurons expressing vesicular glutamate transporters (Vglut2 and Vglut3) are also evident^46,49,50. Neuropeptides including corticotropin releasing factor (CRF), neuropeptide Y (NPY), enkephalin, substance P, neurotensin, and dynorphin also comprise many BNST neurons^{51,52,53,54,55,56,50,57}. These neuropeptides can be used as selective markers for small subpopulations of cells. However, their role is not well-understood in the regulation of the stress response.

BNST and stress

The BNST has been implicated in integrating and processing stress as well as pathological anxiety states, such as post traumatic stress disorder (PTSD), that result from chronic stress exposure and/or aversive stimuli^{58,59,60,61,62,63,64}. The BNST has been hypothesized to mediate “sustained” fear and/or anxiety after repeated stressors, as well as contextual cues that predict aversive and/or stressful stimuli⁶⁵. Information pertaining to acute or phasic fear responses, on the other hand, is thought to be mediated by other amygdala nuclei, such as the BLA and CeA^66,64. Additionally, the BNST has been linked as a key structure in modulating stress-induced reinstatement of drug seeking⁶⁷. Inactivation of the BNST, using muscimol and baclofen, can block cocaine reinstatement following injections of the pharmacological stressor yohimbine⁶⁸. Interestingly, BNST efferents to brain reward centers such as the VTA can encode stressful stimuli, as VTA-projecting BNST neurons exhibit increased c-Fos protein levels following cocaine reinstatement and a forced swim stressor^69,70.

One intriguing signaling molecule within the BNST that plays a key role in modulating stress is CRF. Stress-producing stimuli can elevate CRF expression in the dorsolateral and ventrolateral regions of the BNST^71,72,73,74. Additionally, overexpression of CRF by a lentivirus in the BNST enhances a conditioned fear response⁷⁵. Consistent with the BNST’s role in promoting sustained fear/anxiety, administration of a CRF antagonist into the BNST blocked sustained, but not phasic, fear behaviors⁶⁵. Moreover, intra-BNST injections of CRF antagonists can block both stress-induced reinstatement of cocaine seeking and morphine conditioned place preference^76,77. Interestingly, acute inhibition of the BNST by tetrodotoxin (TTX) injection during stress exposure prevents the expression of stress-induced anxiety⁵⁸ and activation of CRF receptors directly in the BNST induces social anxiety⁷⁸. In a recent study, Gafford et al.,⁷⁹ knocked out GABA_A α1 subunits in CRF-containing neurons within in the BNST, causing increases in anxiety-like behavior as well as impairments in the extinction of conditioned fear. These behavioral deficits were rescued with an intra-BNST infusion of a CRF antagonist.

Since the BNST comprises many different subnuclei and genetically distinct cell types, parceling out specific neural circuit elements that are altered during or following stress exposure has been extremely challenging. Moreover, little is known about specific BNST projection targets since electrical stimulation of this structure often results in indirect stimulation of other brain regions and fibers of passage. However, by utilizing optogenetic strategies, we now have a contemporary toolkit that will provide a circuit-aided design to determine how genetically defined BNST neural circuits process and respond to stress-producing stimuli.

Optogenetic manipulations of BNST neural circuits

Within the BNST, this type of precise functional circuit mapping described above will be essential for dissecting the plethora of genetically distinct neurons and neural circuit elements that are involved in the stress response, which traditional anatomical tracing and electrophysiology techniques cannot address. Conventional circuit tracing experiments cannot differentiate between distinct cell types in heterogeneous tissue. Additionally, electrophysiology experiments, which utilize electrical stimulation, do not provide precise synaptic connectivity due to stimulation of all cell types as well as fibers of passage. Two recent studies have utilized optogenetic strategies to parcel out distinct BNST circuits involved in reward, stress, and anxiety-related behaviors. Kim et al.⁸⁰ globally transduced BNST neurons with NpHR in order to inhibit cell bodies within this region. The authors demonstrated that photoinhibition of the BNST increased open-arm time in an elevated plus maze (EPM), indicative of anxiolysis. Conversely, photostimulation of the BNST with ChR2 resulted in anxiogenesis, as mice spent less time in the open arm of the EPM. In the same study, the authors then examined the role of BLA inputs into the BNST on anxiety-like behaviors. In these experiments, separate groups of mice were injected with a viral vector containing either ChR2 or NpHR into the BLA. Additionally, mice were implanted with optical fibers above the BNST in order to stimulate or inhibit BLA terminals within the BNST. Surprisingly, the authors found that photostimulation of the BLA-BNST pathway produced anxiolysis, whereas photoinhibition of this pathway resulted in anxiogenesis in both the EPM and respiration rate. In a final series of experiments, the authors examined the behavioral ramifications of stimulating three discrete BNST output regions: LH, VTA, and PBN. Here, the authors reported that photostimulation of BNST efferents in each of the target output regions resulted in a unique behavioral phenotype of the anxiety-like state. Jennings, Sparta et al.³⁸, demonstrated that parallel yet opposing BNST subcircuits modulate divergent motivational states by utilizing optogenetic techniques. The authors transduced both glutamatergic and GABAergic BNST neurons with ChR2 in Vglut2-cre and Vgat-cre mouse lines, respectively. By antidromically stimulating BNST terminals within the VTA, the authors were able to identify VTA-projecting BNST glutamate and GABAergic neurons during a foot-shock stress paradigm. The authors found that aversive foot shock caused a net enhancement of activity of optically identified VTA-projecting BNST glutamate neurons. In contrast, aversive foot shock resulted in the reduction of activity in optically identified VTA-projecting BNST GABA neurons. In a separate series of experiments, the authors found that photostimulation of BNST-VTA glutamatergic fibers resulted in aversion and anxiety-like behavior, whereas photostimulation of BNST-VTA GABAergic terminals promoted reward-related behaviors and buffered stress induced anxiety. The results from both of these studies indicate that the BNST has a more complex role than what is canonically thought in regard to stress, anxiety, and reward. Figure 2 depicts a schematic of potential BNST output stress neural circuits, which can be manipulated by optogenetic strategies discussed above.

Figure 2. Potential BNST stress neural circuits.

The BNST sends projections to the VTA, PAG, CeA, PBN, and LH. These potential stress neural circuits can be analyzed using optogenetic strategies detailed above.

Conclusions

Using optogenetic strategies, we now have the ability to parcel out neural circuits within heterogeneous structures, such as the BNST, in order to examine their contribution to the stress response. With these new techniques, we will come to a greater understanding of BNST neural circuits that are recruited during exposure to stressful stimuli and potentially identify new pharmacological targets for the treatment of neuropsychiatric illness.

Highlights.

Review of current techniques for optogenetic neural circuit mapping.

Strategies for optogenetic manipulation of neural circuit elements in vivo.

Review of optotgenetic techniques used to dissect the BNST.

List of acronyms

AAV

adeno-associated virus

AGRP

agouti-related peptide

ARC

arcuate nucleus of the hypothalamus

Arch

archaerhodopsin

BLA

basolateral amygdala

BNST

bed nucleus of the stria terminalis

CaMKIIα

calcium-calmodulin dependent protein kinase IIα

CeA

central nucleus of the amygdala

ChR2

channelrhodopsin-2

CRF

corticotropin releasing factor

DAT

dopamine transporter

GABA

γ-Aminobutyric acid

HSV

herpes simplex virus

LED

light emitting diode

LDT

laterodorsal tegmentum

LH

lateral hypothalamus

LHb

lateral habenula

MeA

medial nucleus of the amygdala

NAc

nucleus accumbens

NpHR

halorhodopsin

NPY

neuropeptide Y

PAG

periaqueductal gray

PBN

parabrachial nucleus

POMC

pro-opiomelanocortin

PRV

pseudorabies virus

PTSD

post traumatic stress disorder

PVH

paraventricular nucleus of the hypothalamus

RG

rabies virus envelop glycoprotein

RMTg

rostromedial tegmental nucleus

SNr

substantia nigra reticulata

TH

tyrosine hydroxylase

TTX

tetrodotoxin

TVA

avian retroviral receptor

Vgat

vesicular GABA transporter

Vglut

vesicular glutamate transporters

W

watts