Orexin 2 receptor stimulation enhances resilience, while orexin 2 inhibition promotes susceptibility, to social stress, anxiety and depression

ABSTRACT Knockdown of orexin/hypocretin 2 receptor (Orx2) in the basolateral amygdala (BLA) affects anxious and depressive behavior. We use a new behavioral paradigm, the Stress Alternatives Model (SAM), designed to improve translational impact. The SAM induces social stress in adult male mice by aggression from larger mice, allowing for adaptive decision‐making regarding escape. In this model, mice remain (Stay) in the oval SAM arena or escape from social aggression (Escape) via routes only large enough for the smaller mouse. We hypothesized intracerebroventricular (icv) stimulation of Orx2 receptors would be anxiolytic and antidepressive in SAM‐related social behavior and the Social Interaction/Preference (SIP) test. Conversely, we predicted that icv antagonism of Orx2 receptors would promote anxious and depressive behavior in these same tests. Anxious behaviors such as freezing (both cued and conflict) and startle are exhibited more often in Stay compared with Escape phenotype mice. Time spent attentive to the escape route is more frequent in Escape mice. In Stay mice, stimulation of Orx2 receptors reduces fear conditioning, conflict freezing and startle, and promotes greater attention to the escape hole. This anxiolysis was accompanied by activation of a cluster of inhibitory neurons in the amygdala. A small percentage of those Stay mice also begin escaping; whereas Escape is reversed by the Orx2 antagonist. Escape mice were also Resilient, and Stay mice Susceptible to stress (SIP), with both conditions reversed by Orx2 antagonism or stimulation respectively. Together, these results suggest that the Orx2 receptor may be a useful potential target for anxiolytic or antidepressive therapeutics. HIGHLIGHTSStress Alternatives Model (SAM) produces resilient (Escape) and susceptible (Stay) phenotypes.icv Orx2 agonist reduces anxious and depressive behaviors in Stay mice.icv Orx2 antagonist increases anxious and depressive behavior plus plasma corticosterone.Stay mice exhibit fear conditioning, reversed by an Orx2 agonist.icv Orx2 agonist increases immediate‐early gene EGR1 in BLA parvalbumin GABA interneurons.

There are existing pharmacological and behavioral treatment options for anxiety disorders, but the quality of and satisfaction with these treatments is poor to moderate (Young et al., 2001), which indicates a need for more effective treatment options. Recent examinations of classic tests for anxiety and depression using animal models suggest that the results have little translatable relevance for trials in clinical populations, with the caveat that tests including social measures have promise (Blanchard et al., 2013;Haller and Alicki, 2012;Haller et al., 2013;Holsboer and Ising, 2010;Keifer and Summers, 2016;Nestler and Hyman, 2010). Therefore, new models are necessary to specifically address the complexity of social interactions in tests of anxiety and depression (Blanchard and Blanchard, 1989a;Keifer and Summers, 2016;Robertson et al., 2015).
The Stress Alternatives Model (SAM) is designed to test simple decision-making during socially stressful situations that produce anxious and depressive behaviors (Smith et al., 2014). The SAM consists of an oval open field (OF) arena that provides two escape routes which lead to enclosed areas (Fig. 1B). In the absence of social interaction, the SAM apparatus produces OF anxiety, but when small test mice are paired with a novel larger aggressor over four days, interactions produce robust social anxiety as well (Robertson et al., 2015). The escape holes are just large enough for the smaller non-aggressive animals, but of those only a proportion of the population (usually half) ever makes use of them. Therefore, decisions made by the test mice result in two behavioral phenotypes: 1. Escape or 2. Stay. It is important to note that both phenotypes readily discover and examine the escape routes, then learn how to escape, but only some (50% is average) choose to use this option (Carpenter and Summers, 2009;Robertson et al., 2015;Smith et al., 2014). Escaping during aggressive interactions behaviorally ameliorates the stress involved, and reverses rising stress hormone and gene expression levels (Carpenter and Summers, 2009;Robertson et al., 2015Robertson et al., , 2017Smith et al., 2014Smith et al., , 2016Summers et al., 2017). Animals that choose to remain with the larger aggressor (Stay) exhibit much higher elevated plasma corticosterone, behavioral Pavlovian fear conditioning, elevated anxiolytic neuropeptide S (NPS) gene expression in the central amygdala (CeA), and enhanced gene expression of cannabinoid 2 (Cb 2 ) receptors in the hippocampus (Robertson et al., 2017;Smith et al., 2014Smith et al., , 2016 (16.75 × 20 cm) in the middle of the OF and the aggressor is placed outside this area. The cylinder is removed just prior to social interaction. C) Design of SIP apparatus; an opaque square arena (40 × 40 × 40 cm) with a clear perforated container set at the middle of one wall. Test mice are first placed in the apparatus at the opposite end from an empty container (left) for 2.5 min before being removed. The empty container is then replaced with a container that has a novel CD1 mouse strain/size matched to mice used as aggressors in SAM (right). The test mouse is placed back into the apparatus in the same manner as previously described for another 2.5 min before being removed.
are relatively stable, both reserve the requisite behavioral plasticity to reverse phenotype under the influence of experience or pharmacological treatments. For example, the anxiolytic corticotropin releasing factor type 1 receptor (CRF 1 ) antagonist, antalarmin allows escape in mice that were previously only submissive (Stay). Alternatively, the Escape phenotype can be reversed through the anxiogenic α 2 -receptor antagonist, yohimbine (Smith et al., 2016). This model (Fig. 1B) produces a gradient of anxiety, where familiar escape is least stressful, such that animals which have escaped more than once exhibit reduced apparatus and escape tunnel-related novelty stress that was expressed during the initial escape. As such, the latency to escape is dramatically reduced, which is another reliable measure of diminished anxious behavior in the SAM (Robertson et al., 2015;Smith et al., 2014). Other anxiolytic factors, such as wheel-running exercise and icv NPS, reduce the initial latency to escape (Carpenter and Summers, 2009;Robertson et al., 2015;Smith et al., 2014Smith et al., , 2016Summers et al., 2017;Yaeger et al., 2018). Alternatively, remaining submissively (Stay) leads to social defeat, which is the most stressful condition (Koolhaas et al., 1997) in our model. The pharmacological results of anxiogenic and anxiolytic agents clearly demonstrate that the stress of socially aggressive interaction is the mechanism that inhibits escape for the Stay phenotype. Recent work from our lab and others suggests that the orexin (Orx; hypocretin, Hcrt) system interacts with the circuitry producing this social stress-induced anxiety gradient (Giardino et al., 2018;Winsky-Sommerer et al., 2004, 2005, with type 1 receptors (Orx 1 ) creating anxiogenic and depressive responses, and type 2 receptors (Orx 2 ) producing contrasting anxiolytic or antidepressive results (Arendt et al., 2013(Arendt et al., , 2014. Not surprisingly, stress-reactive orexins play important roles in anxiety, fear, and depression through activation of both Orx receptor subtypes (Abbas et al., 2015;Arendt et al., 2013Arendt et al., , 2014Eacret et al., 2018;Giardino et al., 2018;Johnson et al., 2010Johnson et al., , 2015Khalil and Fendt, 2017;Lungwitz et al., 2012;Wang et al., 2017). Intracranial or icv injection of Orx A promotes anxious behavior (Lungwitz et al., 2012;Suzuki et al., 2005), and similar types of anxiogenic (including panic) physiological and behavioral responses can be reversed by blocking Orx 1 receptors (Johnson et al., 2010(Johnson et al., , 2015Staples and Cornish, 2014;Vanderhaven et al., 2015). Designer receptor activation of Orx neurons plus antagonism of Orx 2 receptors (Grafe et al., 2017) potentially increases activity only at Orx 1 receptors, or at receptors for Orx-colocalized factors such as glutamate, dynorphin, galanin or nitric oxide (Cheng et al., 2003;Chou et al., 2001;Hakansson et al., 1999;Torrealba et al., 2003), and results in increased anxiety, as well as decreased social interaction and latency to defeat, similar to Orx neuron stimulation alone (Eacret et al., 2018). We have previously suggested that Orx 1 function is diametrically opposed to that of Orx 2 (Arendt et al., 2013). We also demonstrated that knockdown of Orx 2 receptors in the basolateral amygdala (BLA) produces anxiogenic responses, suggesting that the natural function of Orx 2 is anxiolytic (Arendt et al., 2014). Amygdalar inhibition of anxious behavior is thought to be regulated through GABAergic interneurons in the BLA or intercalated region of the amygdala (ItC) (Busti et al., 2011;Lee et al., 2013;Tovote et al., 2015Tovote et al., , 2016. Recent experimentation with orexin-deficient mice resulted in increased anxiety (Abbas et al., 2015;Khalil and Fendt, 2017), which we speculate may be due to lack of Orx 2 activation. Therefore, we posited that modulation of Orx 2 function would result in modification of adaptive emotional and defensive behavior in mice. As Orx has been shown to be functionally effective following intra-nasal delivery (Baier et al., 2008;Dhuria et al., 2009;Modi et al., 2017;Van de Bittner et al., 2018), a putatively systemic brain Orx 2 -induced anxiolysis could be of therapeutic value. We hypothesized that icv delivery of an Orx 2 agonist would promote resilience, result in a reversal of submissive behavior, and decrease anxious and depressive behaviors. In relation to this agonist effect, we hypothesized that specific parvalbumin positive GABAergic neurons in the BLA or ItC would be activated by Orx 2 stimulation in submissive Stay mice. We also hypothesized that intracerebroventricular (icv) delivery of an Orx 2 antagonist would inhibit escape behavior, promote stress susceptibility, and increase anxious and depressive behaviors, such as startle responses and freezing time.

Subjects and housing
Adult (6-8 weeks) male C57BL/6N mice weighing ∼22-26 g (Envigo, Indianapolis, IN; N = 56) were group housed (4-5 per cage) for 5 days of acclimation (Fig. 1A). They were singly housed (including cage controls) following icv cannulation for the remainder of the experiment (12 days), on a 12:12 light-dark cycle (lights off at 6 P.M.) at 22°C, with ad libitum food and water. A separate cohort of male Hsd:ICR (CD1) retired breeder mice weighing ∼50 g (Envigo, Indianapolis, IN; N = 37) were similarly housed, and used to provide aggression during social interaction in the Stress Alternatives Model (SAM; Fig. 1B) and a target for the Social Interaction/Preference (SIP) tests. During the dark active, phase mice (C57BL/6N) were exposed to daily handling starting 2 days after icv cannulation, followed by behavioral aggression and testing. Handling included loosening and retightening the cannula insert for ∼1 min. All surgical and behavioral procedures were performed in a manner that minimized suffering and the number of animals used, and were in accordance with the National Institutes of Health's Guide for the Care and Use of Laboratory Animals  and approved by the Institutional Animal Care and Use Committee of the University of South Dakota.

Experimental design
Experimental groups included: home cage control (N = 10; used to derive baseline hormone and protein [immunohistochemistry, IHC] expression levels), vehicle control (N = 20), Orx 2 receptor antagonist (N = 17), and Orx 2 receptor agonist (N = 9). All treatment groups, except cage control, underwent 4 days of aggressive social interaction in the Stress Alternatives Model apparatus (SAM; Fig. 1A and B), which was preceded by fear conditioning on each day (described below; Fig. 1A). On day 5, treated animals underwent a Social Interaction/ Preference (SIP; Fig. 1C) test, followed by measurement of freezing response in the SAM to the conditioned stimulus in the absence of social interaction (CR = conditioned response to tone; Fig. 1A). Drug treatments were administered on day 3. After testing on day 5, mice were briefly anesthetized with isoflurane (5% at 1.0 min/L for ∼1.5 min) and rapidly decapitated. Whole brains and trunk blood plasma were collected and stored at −80°C for further analysis.

Stress Alternatives Model (SAM)
The SAM apparatus is constructed of a clear rectangular box (91 cm long, 22 cm wide, and 26 cm high) with curved opaque dividers (r = 10.25 cm) that partition it into 3 sections: two enclosed areas (10 × 22 × 26 cm) at each end of the oval-shaped open field (OF = 71 × 22 × 26 cm) arena for social interaction (Fig. 1B). All behavioral interactions were conducted during scotophase (dark; active period). The two enclosed areas are accessible from the OF arena through escape routes (one each at the distal ends of the oval) only large enough for a smaller C57BL/6N test mouse to pass through (Fig. 1B).
A short time before social interaction began, a novel large retired breeder CD1 mouse (aggressor) was placed into the SAM OF arena outside of a centered opaque cylindrical divider (15 cm diameter and 20 cm tall). A smaller C57BL/6N mouse (test mouse) was then placed inside the opaque cylinder (Fig. 1B). Our lab has previously shown that fear conditioning influences gene expression of Orx receptors, endocannabinoid receptors (Cb 2 ), and brain-derived neurotropic factor (BDNF) in the amygdala and hippocampus (Robertson et al., 2017;Smith et al., 2014); therefore, to incorporate a SAM-inclusive fear conditioning paradigm, after a 30 s acclimation period inside the cylindrical divider, a 5 s tone (2500 Hz at 75 dB, CS + ) was presented. Following the tone there was a 10 s trace period, after which the divider was lifted allowing test and aggressor (US + ) mice to freely interact within the SAM arena for 5 min.
Retired breeder CD1 mice are naturally aggressive toward other male mice, including smaller C57BL/6N, and interactions led to high intensity social aggression and defeat (Golden et al., 2011). Different from traditional social defeat models (Golden et al., 2011), test mice in the SAM are provided the option to either exit the arena (Escape) through one of the two escape routes or remain in the arena (Stay) with the larger aggressive mouse. Mice that escape the arena (Escape) have been shown to produce lower physiological and behavioral measurements of stress when compared to their submissive non-escaping (Stay) counterparts (Robertson et al., 2015;Smith et al., 2014Smith et al., , 2016. It is important to note, that both Escape and Stay mice experience high levels of aggression from the novel CD1 mouse, and have elevated hormonal, behavioral, and gene expression stress responses, though these measures are significantly higher in the Stay mice. Results from this and previous experiments have demonstrated that test mice generally choose a behavioral phenotype (Escape or Stay) by the second day of SAM interaction.
During interactions in which an aggressor mouse threatened the life of a test mouse, a clear, perforated divider (15 cm wide and 20 cm high) was used to briefly interrupt (5-10 s) these intensely aggressive bouts. Life threatening attacks included repeated bites to the head or neck of the test mouse. After 5 min of interaction, both mice were removed from the arena and placed back into their home cages. If test mice escaped, they were left in the enclosed area of the SAM until the end of the 5 min interaction period.
Test mice were exposed to 4 consecutive days of the SAM social aggressive interaction, following fear conditioning. Drug infusions were given prior to SAM interaction on day 3 (described below). A novel aggressor mouse was used on each day. On day 5 the Social Interaction/ Preference test (SIP) was performed (described below), followed by a final SAM exposure: placement in SAM cylinder, 5 s CS + (tone), and 10 s trace period, testing for a conditioned response (CR). However, on this day the cylinder was not lifted and there was no large mouse present, and therefore, no social aggression. All interactions were recorded manually and digitally using GoPro Hero3+ cameras under red light. Digitally recorded interactions were scored by individuals naive to the treatment of each test mouse.

Fear conditioning behavioral analysis.
Freezing behavior, commonly used to assess fear conditioning (Blanchard and Blanchard, 1988), was measured for 4 days prior to aggressive social interaction (US), and on the 5th day in the absence of the US. Freezing is the amount of time (> 1 s) that mice remained motionless, except for movements related to respiration. Freezing was measured during the time that a test mouse spent inside the opaque dividing cylinder; which included time prior to the tone (CS), during the tone, and after the tone (ending when the cylinder was removed). Percent fear freezing ratio was calculated as: [(freezing time after tone/15 s)/(freezing time before tone/30 s)] x 100; this ratio has the advantage of revealing fear conditioned freezing, as opposed to spontaneous freezing, as a number greater than 100. Significant fear conditioning should always be represented by a mean > 100.

SAM behavioral analysis.
Behavioral analysis for SAM socially aggressive interaction began following the removal of the cylinder and ended when the test mouse escaped, or at the end of the 5 min interaction time. Mice were designated as escapers (Escape) if they left the arena through an available escape tunnel on one or more of the two test days prior to drug treatment. Mice that did not escape, remained submissively (Stay) with the aggressor CD1. Interaction time was measured as the time from the removal of the cylinder until the end of the interaction (when the test mouse was removed from the arena or escaped). Freezing, startle, and time attentive to the escape route measurements were normalized by the total time spent interacting with an aggressor. Latency to escape was measured as the time starting from the removal of the cylinder and ending when the torso of the test mouse was through the escape hole. Locomotor activity was measured following SAM interactions on the day of drug treatment, in the home cage by counting line crossings of a grid with 10 cm between lines.
Freezing behavior during SAM interactions (conflict freezing) was defined in the same way as freezing during assessment of fear conditioning (before the interaction begins). Conflict freezing always occurs during social interaction, and includes freezing in anticipation of aggression, in response to aggression, and contextual freezing in response to the arena in which aggression is experienced (especially on days 2-4 of SAM interactions). Freezing time was converted into percent freezing time ([time (s) frozen/total interaction time] x 100).
Startle responses (whole-body flinching) measure an animal's reaction to sudden environmental stimuli, often acoustic (Heesink et al., 2017;Sallinen et al., 1998). We surmised, that as the intensity and amplitude of the acoustic-startle response are affected by stress-related events (Belzung and Griebel, 2001;Risbrough et al., 2004), the frequency of socially-induced startle might also be enhanced by increasing stress during aggressive interactions. With this in mind, we pirated the concept of startle for use in the SAM, because we noticed that active approach of an aggressor mouse also elicits a similar startle response. Therefore, we made use of a stimulus that exists as a part of aggressive social interaction, the threat of attack, and measured the number of flinches or recoils that were induced by the approach of the larger CD1 mouse during each entire social interaction.
The amount of time the test mouse spent in contact with the escape tunnel is an indicator of stress-sensitive novelty exploration (Eagle et al., 2013;Jacinto et al., 2013). Although the entire SAM arena is novel initially, the escape route is a unique attribute, distinctly different from the rest of the apparatus, as it clearly presents a physically different path for movement. The chamber at the other end of the escape route is also unknown, until the first passage. Thus, this indicator of novelty assessment is also specifically associated with the stress-sensitive act by which the Escape and Stay Phenotypes are demarcated. We defined the amount of time spent attentive to the hole as only including time when the test mouse was actively interested in and investigating the escape hole and tunnel directly (sniffing and placing head in the hole). This measure is a novel indicator of anxious behavior and decision-making unique to this model, which has been effectively used in SAM experiments on rainbow trout . We consider Escape and Stay behavioral outcomes to be the result of decisionmaking because early responses are variable, become stable with experience, and are modifiable by learning as well as anxiogenic or anxiolytic drugs (Carpenter and Summers, 2009;Smith et al., 2014Smith et al., , 2016Summers et al., 2017;Yaeger et al., 2018).

Social Interaction/Preference Test (SIP)
The Social Interaction/Preference (SIP) test was performed on day 5 as a reliable indicator of whether mice are resilient or susceptible to social stress following the SAM interactions and drug treatments (Arendt et al., 2014;Krishnan et al., 2007). The SIP test was conducted in a square box (40 cm 3 ) with a perforated clear container (12 cm diameter and 20 cm high) placed alongside the middle of one wall (Fig. 1C). To begin, test mice were placed into the arena at the opposite end from the empty container and allowed to explore for 2.5 min, before being removed. The empty vessel was then replaced with an identical container holding a novel CD1 mouse. Test mice were then placed back into the arena for another 2.5 min.
2.2.2.1. SIP behavioral analysis. The amount of time test mice spent in C.D. Staton et al. Neuropharmacology 143 (2018) 79-94 close proximity to (within 3 cm) the empty vessel, and the one containing a novel CD1 mouse was recorded. To determine whether animals were resilient or susceptible to social stress, the amount of time near the jar containing the aggressor mouse was divided by the time spent near the empty jar ([Time 3 cm or less from aggressor's jar/Time 3 cm or less near empty jar] x 100). Animals with values of 100% or greater were interpreted as being resilient, and those less than 100% were construed to be susceptible to social stress (Arendt et al., 2014;Krishnan et al., 2007).

icv stereotaxic surgeries
Following a 5 day acclimation period, but prior to exposure to behavioral paradigms, stereotaxic surgery for intracerebroventricular (icv) cannula implantation was performed on C57BL/6N mice that were to undergo vehicle, Orx 2 receptor agonist, or antagonist treatment (N = 46). Mice were anesthetized with isoflourane (2% at 1.0 l/min flow rate) and a guide cannula (26 ga cut to 1.5 mm) was inserted into the right lateral ventricle (Bregma; AP: −0.50 and ML: −1.0). Animals were allowed 7-10 days to recover before aggressive social interaction began.

Drugs & infusion
Drug treatments were given icv, to mice on day 3, either 1 (antagonist or vehicle) or 30 min (agonist; to allow for longer peptide diffusion) prior to being placed into the SAM arena. Drug plus placement in the SAM was followed by the CS and then social aggression interaction, similar to all SAM interaction days ( Fig. 1A and B). The SAM was designed to include both within subject and across treatment group comparisons. Injection on day 3 allows comparison of each animal to itself on days 2 and 3, before and after injection, as well as comparisons with the vehicle control. The design was intended to measure acute drug effects on day 3, just after injection. However, three kinds of longer-term effects are possible: 1) carryover effects for drugs that degrade slowly, 2) effects associated with persistent modification of stress responsiveness that yields longer-term changes in anxious or depressive behaviors, and 3) drug effects that promote neural plasticity and yield gene expression or learning.
The Orx 2 receptor antagonist, MK-1064 (0.3 nmol, MCE, Monmouth Junction, NJ), and Orx 2 receptor agonist, which is a modified Orx B peptide, [Ala 11 , D-Leu 15 ]-Orx B (0.3 nmol, TOCRIS, Minneapolis, MN), were dissolved in 0.9% saline plus 5% DMSO (= vehicle). The doses for both drugs were chosen to avoid enhanced locomotor activity or induction of sleep. Of the two drugs used, only [Ala 11 , D-Leu 15 ]-Orx B has been administered icv previously, at a dose that increased locomotion (Samson et al., 2010), and we chose a dose that was ten-fold smaller (Fig. 3D). For MK-1064, we chose a dose that was 10,000 fold smaller than that given orally, avoiding induction of sleep or acute HPA axis activity (Gotter et al., 2016;Grafe et al., 2017), but equivalent to the icv dose chosen for [Ala 11 , D-Leu 15 ]-Orx B , because the IC 50 (half-maximal inhibitory) and EC 50 (half-maximal effective) concentrations are similar for these two drugs. All treatments were directed into the lateral ventricle via micro-syringe (extended 0.2 mm below the guide cannulae) and delivered at a rate of 0.5 μl/min, with a total injection volume of 1 μl. All icv injections (1 or 30 min prior) were followed by placement into the SAM arena, plus an additional 45 s (for fear conditioning) before initiation of the aggressive interaction. As MK-1064 crosses the blood-brain barrier, and substances delivered to the cerebrospinal fluid via icv injections are transferred to the cerebral microvasculature via the choroid plexus (Pardridge, 2016), we assume this drug is delivered throughout the brain. In contrast, similar sized peptides that include [Ala 11 , D-Leu 15 ]-Orx B , Orx A , and Orx B have been administered icv previously (Flores et al., 2014;Gotter et al., 2016), producing behavioral results such as enhanced locomotion and influencing consolidation of fear memories, the latter function dependent on the BLA (Flores Fig. 2. Escape behavior is promoted by Orx 2 agonism, and limited by Orx 2 antagonism, while latency to escape is unchanged. A) Escape behavior was significantly reduced, following icv injection of Orx 2 receptor antagonist (MK-1064) on day 3 (light gray hatched bar) in mice that were previously Escape phenotype (light gray hatched bar; # indicates significance in antagonisttreated animals from Days 1&2), but remained unchanged in vehicle-treated animals (light gray bar). B) Escape behavior was promoted following day 3 icv injection of Orx 2 receptor agonist ([Ala 11 ,D-Leu 15 ]-Orexin B), on day 4 (dark gray hatched bar; not significant) in mice that were previously Stay phenotype. (C) Escape latency was significantly reduced (*) on day 4 of SAM interaction compared to day 1. There was no change seen between Escape animals treated with vehicle and animals treated with Orx 2 antagonist. et al., 2017). We followed this precedent, suggesting that the Orx peptides are functionally available following icv injections to brain regions associated with fear, anxiety and depression.

Hormone analysis
Following testing on day 5, trunk blood was taken and centrifuged for ∼5 min to separate plasma. Plasma was frozen at −80°C for further analysis. Plasma corticosterone concentrations [B] were quantified in duplicate, in a single run, using a corticosterone enzyme-linked immunosorbent assay kit (Enzo Life Sciences, Farmingdale, NY).

Immunohistochemistry
Frozen coronal sections (16 μm) from AP -0.94 to −1.70 relative to bregma were incubated with EGR 1 (1:500; rabbit monoclonal from Cell Signaling 4154S) and parvalbumin (Pv, 1:100; goat polyclonal from ABcam ab32895) primary antibody combination in solution, 5% bovine serum albumin (BSA) in phosphate buffered saline (PBS), at 4°C overnight (Sathyanesan et al., 2017). Antibodies were used as per manufacturer's instructions and specificity was tested using incubation in antibody solutions lacking primary antibody. Following primary antibody incubation, slides were washed in 1xPBS three times for 5 min each at room temperature. Slides were then incubated in fluorescent secondary antibody (1:500, Alexa-488 and 594: AF488 donkey antigoat -A11055 for Pv and AF594 chicken anti-rabbit -AF21442 for EGR 1 ; Life Technologies) in 2.5% BSA in PBS for 2 h at room temperature. The slides were then rinsed in 1× PBS three times for 5 min each and coverslipped with VECTASHIELD HardSet mounting medium with 4′,6-diamidino-2-phenylindole (DAPI). Three sections from each mouse were analyzed, from which an area (210 × 240 μm) was defined at AP -1.06, centered on ML + or -2.65 and DV 4.50 relative to bregma, in which all DAPI labeled neurons were counted and scored for EGR 1 and Pv labeling. Sections were viewed by an unbiased observer to evaluate differences and images were captured using a Nikon Eclipse Ni microscope equipped with a DS-Qi1 monochrome, cooled digital camera and NIS-AR 4.20 Elements imaging software, using 20× and 40× objective magnifications. Sections from stressed and control mice were captured using identical exposure settings.

Statistical analysis
All experimental designs and statistical analyses were based on a priori hypotheses. For conditions that changed over 4 days of SAM social interaction, we compared outcomes using a two-way repeated measures ANOVA (Orx 2 drug x behavioral phenotype x day of SAM interaction design), where phenotype was either Stay or Escape. In addition, two-way ANOVA (Orx 2 drug x Phenotype design) was utilized to determine the influence of activating Orx 2 receptors (Orx 2 Effects) relative to the expression of behavioral phenotypes (Stay vs Escape; Phenotype Effects), Phenotype by Conditioning (Interaction Effects), and analysis of BLA EGR 1 , Pv, and DAPI triple-labeling. To compare changes occurring within a treatment group across SAM interaction days, a one-way repeated measures ANOVA (Orx 2 drug x day of SAM interaction design) was performed. While the statistics for behavior did not include cage controls; these controls were necessary for interpretation of hormonal corticosterone levels, because samples from SAM treatments were compared to baseline levels determined by the mean  x 100) of Stay phenotype mice (dark gray bar) on test day (two days after vehicle injection) reflects a conditioned response (CR) which is significantly ( ‡) greater than the FFR for Escape phenotype animals. B) Treatment with an Orx 2 antagonist (hatched bars) did not significantly alter the fear freezing ratio 2 days after injection, compared with vehicle-treated mice. C) In Stay mice, which exhibit a CR (significantly elevated FFR) in vehicle-treated animals, icv injection of an Orx 2 agonist (dark gray hatched bars) blocks the fear freezing conditioned response, producing a FFR which is significantly ( §) reduced compared with vehicle controls (dark gray bars). D) Home cage activity is not altered by icv injection of Orx 2 agonist or antagonist.
[B] value of home-cage control animals. Therefore home-cage controls were added for specific one-way ANOVA comparisons. Comparison of locomotion in the home cage after drug treatment was also accomplished by one-way ANOVA. Comparisons between two treatments (Vehicle, Orx 2 receptor antagonist, or Orx 2 receptor agonist) within a given phenotype (Escape or Stay) were analyzed by Student's t-tests. To determine differences in percentage of escape, a chi-square statistical analysis was performed, where results from previous days were utilized as expected values.
Each animal provided only a singular datum for all analyses. Five assumptions of parametric statistics were applied to the data, which were transformed when necessary, but graphed in their raw form. Analyses with both non-parametric and parametric statistics (previously mentioned) were performed along with examination for multiple comparisons using the Holm-Sidak method, and when the statistical analyses match, as they do for the data herein, we report the parametric results without α adjustment (Feise, 2002;Jennions and Moller, 2003;Moran, 2003;Nakagawa, 2004;Perneger, 1998;Rothman, 1990). Significant effects between groups for one-way analyses were examined with Student-Newman-Keuls post hoc analyses (to minimize Type I error) and Duncan's Multiple Range Test (to minimize Type II error).

Stress Alternatives Model
In previous experiments using the SAM arena for mice, rats, and trout, the proportion of animals exhibiting Escape and Stay phenotypes have been approximately 50:50 (Carpenter and Summers, 2009;Robertson et al., 2015;Smith et al., 2014Smith et al., , 2016. Recent work in rainbow trout suggested that these phenotypes were fungible, even without drug treatment, and that the proportions of the two phenotypes may be altered by the degree of stress effects from the captive environment . As occasionally happens there was not an even distribution of phenotypes in these experiments, such that prior to any treatment the proportion of escaping mice was only 28.26%, leaving 71.74% submissive (Stay) mice. This ratio of phenotype distribution required non-random division of phenotypes into groups. Since the Orx 2 receptor antagonist (MK-1064) was expected to reduce escape behavior, a relatively higher proportion of Escape mice (40%) were used for this treatment. In contrast, since application of the Orx 2 receptor agonist ([Ala 11 , D-Leu 15 ]-Orx B ) was hypothesized to promote escape behavior, and the availability of this phenotype was low, we allotted 0% of the individuals displaying Escape behavior to this treatment group (100% Stay).

Escape vs. stay behavior; escape latency
As predicted, in Escape mice, escaping behavior did not change following injection of vehicle ( Fig. 2A), suggesting that the additional stress of icv injection was not enough to modify behavioral phenotype (Smith et al., 2014). Escape behavior was significantly reduced (within subject comparison, χ 2 : F 1 = 4.5, P ≤ 0.034) by injection of the Orx 2 receptor antagonist (MK-1064) on day 3 (50% Escape), in comparison with the first two days of interaction (100% Escape; Fig. 2A). In contrast, day 3 injection of Orx 2 receptor agonist in Stay animals slightly increased escape behavior one day after icv administration (11% Escape on day 4), in comparison with the first three days of interaction (0% Escape; Fig. 2B).
Escape latency was progressively reduced (Repeated measures oneway ANOVA: F 3,39 = 6.18, p ≤ 0.002; Fig. 2C) by experience in all icv injected mice taken together. As is typical of results previously reported for untreated animals (Carpenter and Summers, 2009;Robertson et al., 2015;Smith et al., 2014) the latency for days 2-4 was significantly reduced compared to day 1. However, comparison to previous studies also reveals that all latency times in icv injected animals were greater than those of untreated mice, even though the stressful icv injection occurred only on day 3. Since the availability of escaping mice was limiting, the Orx 2 agonist was not given to Escape mice. The Orx 2 receptor antagonist MK-1064 blocked escape in some animals, but did not increase the latency to escape for those that continued to do so (Fig. 2C), suggesting that future experiments should also include Orx 2 agonist treatment in Escape mice to determine if latency would be reduced.

Fear conditioning response prior to the SAM interaction
As in previous studies (Smith et al., 2014), only Stay mice reacted to fear conditioning stimuli (CS + US pairing = tone + aggression from a larger CD1 mouse) during isolation in the opaque cylinder. The Stay phenotype froze significantly (t 13 = 2.7, P ≤ 0.018) more than Escape mice on test day (5) in the absence of the US (Fig. 3A). The data are presented as a ratio of (freezing during and after the tone/time during and after tone)/(freezing prior to the tone/time prior to tone) x 100, such that significant conditioned freezing [ = CR] is greater than 100%. Importantly, the distinction between the significant CR in Stay mice and not in Escape mice was extant despite the additional stress of icv vehicle injection on day 3. Administration of the Orx 2 antagonist (MK-1064) did not change the fear freezing ratio (Two-Way ANOVA, treatment effect: F 1,19 = 0.06, P ≤ 0.817; phenotypic effect: F 1,19 = 3.28, P ≤ 0.086; interaction effect: F 1,19 = 0.86, P ≤ 0.37) when compared to vehicle treated Escape or Stay mice (Fig. 3B). However, freezing (and the fear freezing ratio) was significantly reduced in Stay mice (t 10 = 2.92, P ≤ 0.015) that received icv injection of the Orx 2 agonist ([Ala 11 , D-Leu 15 ]-Orx B ) two days before testing (Fig. 3C). For this or any behavior reported, activity changed by Orx 2 agonist or antagonist treatment is unlikely to be the cause, as home cage locomotion is unchanged (One-way ANOVA: F 2,35 = 0.391, P ≥ 0.679) immediately after icv injection and SAM social interaction on day 3 (Fig. 3D).

Freezing during SAM interaction
Conflict freezing behavior in C57BL/6N test mice during the SAM social interaction with larger CD1 aggressors, increases significantly (One-Way Repeated Measures ANOVA: F 3,30 = 5.08, P ≤ 0.006) in mice adopting the Stay phenotype after 1 day of interaction (Fig. 4A and B). This increase includes continued longer freezing time on the day of vehicle treatment (day 3, Fig. 4A and B), suggesting that the additional stress of icv injection did not increase the total duration of freezing. In addition, greater freezing time was exhibited by vehicletreated Stay compared to Escape mice on all days (Two-Way Repeated Measures ANOVA, phenotype effect: F 1,42 = 5.17, P ≤ 0.039; time effect: F 3,42 = 4.43, P ≤ 0.009; Fig. 4A), including injection day (day 3; t 16 = 2.38, P ≤ 0.03; Fig. 4A), but not day 4. Importantly, there were no significant differences between Stay mice in different icv injection groups on days 1 & 2 prior to icv injection or day 4 after treatment (Fig. 4B).

Startle responses during SAM interactions
Startle responses during SAM social interactions steadily increased over time in vehicle-injected Stay mice, significantly so by day four (One-Way Repeated Measures ANOVA: F 3,24 = 4.12, P ≤ 0.017: Fig. 5A and B). Notably, vehicle-injected Stay and Escape mice did not exhibit a significant difference in startle responses on injection day (3) compared to the first two days of interaction (1 & 2), suggesting that the additional stress from icv injection had minimal effects on startle responses (Fig. 5A). Additionally, within the vehicle-injected cohort, Stay animals exhibit a significant increase in startle responses compared to Escape animals (Two-way repeated measures ANOVA 1. phenotype effect: F 1,35 = 11.46, P ≤ 0.004; 2. time effect: F 3,35 = 3.41, P ≤ 0.028; Fig. 5A), including on the day of treatment (day 3; t 14 = 2.66, P ≤ 0.019).

Fig. 4. Freezing time is elevated in Stay mice during SAM interaction and can be increased by Orx 2 receptor antagonism and decreased by Orx 2 receptor agonism. A)
Mean ( ± SEM) contextual freezing time in vehicletreated Stay animals (solid dots) is significantly increased (*) compared to day one on days 2-4 of SAM interaction; and compared to vehicle-treated Escape mice (square ‡) on injection day (3; arrow ‡). B) Mean ( ± SEM) % freezing time increases in all groups on day 2, compared to day 1, prior to treatment (+significance in agonist treated animals, # significance in antagonist-treated animals). Injection of Orx 2 antagonist (icv; triangles) increased freezing on day 3 compared to day 1, day 2, and day 4, and icv Orx 2 agonist (diamonds, arrow) significantly decreased freezing relative to vehicle ( §) and antagonist treatments ( †) on day 3 (arrow

Time attentive to hole
Time spent attentive to the escape hole did not change across days in Stay mice injected with vehicle ( Fig. 6A and B). However, prior to (days 1 & 2) and on the day (day 3) of treatment, vehicle-treated Escape mice spent significantly more time attentive to the escape hole compared to Stay vehicle-treated mice (Two-Way Repeated Measures ANOVA, treatment effect: F 1,42 = 31.99, P ≤ 0.001, time effect: F 3,42 = 4.37, P ≤ 0.009, interaction effect: F 3,42 = 4.2, P ≤ 0.011; Fig. 6A), including treatment day (3; t 16 = 4.59, P ≤ 0.001; Fig. 6 A). Stay mice injected with Orx 2 receptor antagonist MK-1064 on day 3 of SAM social interactions exhibited no differences in time spent attentive to the escape hole on any day (Fig. 6B).

Social Interaction/Preference Test
During the SIP test on day 5, after SAM social interactions were completed, significant differences in time spent near the novel container with a social target/time near the container alone x 100 [often used as a measure of social stress resilience/susceptibility (Arendt et al., 2014;Krishnan et al., 2007) ] emerged between phenotypes. Among vehicle injected mice, the Stay phenotype were significantly less resilient than mice that escape (t 18 = 3.82, P ≤ 0.002; Fig. 7A and B). Importantly, this indicates susceptibility to stress in Stay mice that is not seen in Escape mice (Carpenter and Summers, 2009;Robertson et al., 2015;Smith et al., 2014Smith et al., , 2016. However, in Escaping mice, time spent near the target was significantly decreased by treatment with Orx 2 receptor antagonist MK-1064, when compared with vehicle-injected Escape mice (t 11 = 2.6, P ≤ 0.025; Fig. 7A and B). Conversely, icv injection of Orx 2 receptor agonist [Ala 11 , D-Leu 15 ]-Orx B to Stay mice resulted in a significant increase in social novelty seeking when compared to vehicle-treated Stay animals (t 16 = 3.71, P ≤ 0.002; Fig. 7C and D).

EGR 1 /Pv immunohistochemistry
After immunolabeling four sets of brains from each group (Vehicle-Escape, Vehicle-Stay, Orx 2 agonist-Stay, Orx 2 antagonist-Escape, Orx 2 antagonist-Stay), taken on day 5 after CR testing, for the immediate-early gene product early growth response protein 1 (EGR 1 , Zif 268 , NGF 1A , Krox 24 , or TIS 8 ), the nuclear marker 4′,6-diamidino-2phenylindole (DAPI), and the GABA neuron calcium-binding protein Parvalbumin (Pv), the images (Fig. 9) clearly indicated that a distinct cluster of cells in the BLA, or intercalated region of the amygdala (ItC) (Busti et al., 2011;Lee et al., 2013), are activated by Orx 2 agonist binding. A small tight cluster of cells located in the region AP -1.06, centered on ML + or -2.65 and DV 4.50 relative to bregma were triple labeled, but only in the Orx 2 agonist-treated Stay animals (Fig. 9). This group of cells had clear boundaries, when viewed with DAPI and/or Pv labeling, even though significant DAPI and Pv labeling was evident throughout the BLA, which were not significantly different in number based on treatment group (One-way ANOVA: F 3,21 = 1.38, P ≥ 0.28). For all vehicle and antagonist (MK-1064) treated groups, there was no significant effect on triple-labeled neuron count (Two-way ANOVA: 1.  6. Attentiveness to the escape route is low in Stay mice but can be improved with icv administration of Orx 2 agonist. A) Vehicle-treated mice exhibiting the Stay phenotype (solid dots) spend significantly ( ‡) less time attentive to the escape hole compared to Escape phenotype animals (squares), across days, and on the day of vehicle injection (arrow). B) Vehicle and Orx 2 antagonist (triangles) treated Stay mice are equally disinterested in the escape route. However, mice treated with Orx 2 agonist (diamonds) display a significant increase in % time attentive to the escape hole on days 3 and 4 when compared to day 1 (+). Specifically on day 3, Orx 2 agonist-treated mice (arrow) spend a significant more percentage of time attentive to escape hole compared to both vehicle ( §) and Orx 2 antagonist ( †).

Discussion
The primary thrust of experiments on the orexinergic system relative to affective disorders, has been to describe anxiogenic and prodepressive actions of Orx A (Eacret et al., 2018;Ito et al., 2009;Lungwitz et al., 2012;Nollet et al., 2011), through actions of Orx 1 receptors (Johnson et al., 2010;Lu et al., 2017;Ozsoy et al., 2017;Staples and Cornish, 2014), although dual Orx receptor antagonists have been shown to be antidepressive (Nollet et al., 2011(Nollet et al., , 2012. There have also been hints that Orx 1 and Orx 2 receptors have opposing actions (Flores et al., 2016;Lu et al., 2017), including stress-reduced thalamic and hypothalamic Orx 2 expression that was reversed by the antidepressant fluoxetine (Nollet et al., 2011). Our recent work suggests that activation of Orx 2 receptors may be anxiolytic and antidepressive, at least in the amygdala or hippocampus, and may have opposing actions to those of Orx 1 (Arendt et al., 2013(Arendt et al., , 2014. Here we describe, for the first time, that intracerebroventricularly activated Orx 2 receptors are anxiolytic and antidepressive. This icv Orx 2 agonist injection also produced activation, measured by labeling of the immediate-early gene EGR 1 , of a small cluster of GABAergic cells, co-expressing the calcium binding protein Pv, in the BLA or ItC, suggesting that these inhibitory cells are at least partially responsible for the anxiolysis (Fig. 9A-E). A small group of non-Pv containing neurons may also play a role (Fig. 9F). Activation of these inhibitory neurons coincident with reduced anxious

antagonism (Escape mice) and reversed by Orx 2 agonist treatment. A, B)
Vehicle-treated Stay mice (dark gray bar and dots) are significantly ( ‡) less stress-resilient compared to Escape mice (light gray bar and dots), but when Orx 2 antagonist is administered to a cohort of Escape mice (hatched light gray bar and triangles) there is a significant ( §) decrease compared to vehicle-treated Escape mice. C, D) When Orx 2 agonist is administered icv to Stay mice (hatched dark gray bar and diamonds) there is a significant ( §) increase in stress-resilience compared to vehicle-treated Stay mice (dark gray bar and dots).
behavior is supported by experiments using Orx 2 KD in the BLA (Arendt et al., 2014). This conclusion is confirmed additionally by the contrast of anxiogenic responses plus increased plasma corticosterone triggered by Orx 2 antagonist treatment. This conclusion is not based on a singular behavioral result, but includes several preclinical measures in a social stress protocol (SAM), addressing recent concerns that anxiety measures that do not include social dynamics also do not translate well to human clinical treatments (Haller and Alicki, 2012;Haller et al., 2013). Our work suggests that promoting Orx 2 receptor activity (via the Orx 2 agonist [Ala 11 , D-Leu 15 ]-Orx B ) decreases anxious and/or depressive behaviors such as fear conditioned-, as well as conflict-freezing and startle responses (Figs. 3C, 4B and 5B). Stimulation of Orx 2 receptors also increases such adaptive, anxiolytic, and antidepressive behaviors as the amount of time the test mouse spends attentive to the escape hole in the SAM arena and to a novel social target during the SIP test (resilience/susceptibility to stress). Together, these results are suggestive of an anxiolytic and antidepressive profile which stems from whole brain pro-Orx 2 treatments. The proportion of Escape to Stay mice in these experiments was particularly low (28:72) compared with the more typical (50:50) selfimposed division of phenotype from previous studies (Carpenter and Summers, 2009;Robertson et al., 2015;Smith et al., 2014Smith et al., , 2016. Among researchers using various social defeat models it is anecdotally noted that cohorts of fish, mice or rats are not equivalently partitioned into stress-susceptible or resilient animals, which may explain our skewed distribution. However, we have recently noticed in fish and mice, that the proportions of the two phenotypes may be altered by stress effects from the captive environment Yaeger et al., 2018). This recent work also suggests that these phenotypes are plastic, and may be affected by prior stress, exercise, or experience (Robertson et al., 2017;Smith et al., 2016;Summers et al., 2017;Yaeger et al., 2018). Our evidence suggests that this skewed distribution of Escape and Stay phenotypes devolve from prior nonexperimental stressors in the captive environment. This allowed greater emphasis to be placed on anxiolytic and antidepressive treatment of Stay animals. It is important to note that the mutable status of these phenotypes is most common in the first two days of the SAM, but that a proportion of the mice in either phenotype is susceptible to reversal of phenotype, if the neuromodulatory elements of the stress circuitry are altered, through pharmacological treatments or environmental conditions, such as exercise or familiarity, to reduce stress (Robertson et al., 2017;Smith et al., 2016;Summers et al., 2017;Yaeger et al., 2018). The result is, for example in Stay mice, that a subset of that phenotype is disposed to amelioration (Yaeger et al., 2018), which produces some variability in the treatment results (Fig. 7B, D).
Freezing in response to conditioned stimuli (CS) or contexts associated with fearful environments represents adaptive behavior that reduces the exposure of the quiescent animal to predators and competitors (Demuth et al., 2017). This conditioned-fear response (Blanchard and Blanchard, 1988;Iwata and LeDoux, 1988;Iwata et al., 1986) is intensity dependent, such that anxious animals develop a higher freezing response rate to a CS as the severity of the fearful conditions increase (Robertson et al., 2015;Smith et al., 2014Smith et al., , 2016. Consistent with our previous findings (Robertson et al., 2015;Smith et al., 2014Smith et al., , 2016, these results demonstrate that phenotype plays an important role in the development of classical conditioning, as Escape mice did not exhibit freezing in response to the CS, but Stay mice did. This test day result is suggestive of increased anxiety in Stay mice, which develops as a result of decision-making in the SAM arena (Smith et al., 2014). Therefore, escape is an adaptive behavior that results in reduced anxiety and stress reactivity (Robertson et al., 2015;Smith et al., 2014Smith et al., , 2016. However, when Stay mice were injected with Orx 2 agonist, freezing in response to the CS decreased to the level of vehicle-treated escape mice, suggesting that the treatment reduced anxiety. Fear conditioning has been shown to be regulated by activation of GABAergic neurons in the ItC and BLA (Busti et Fig. 8. Social stress and Orx 2 antagonist treatment increase plasma corticosterone concentrations. A) Mean ( ± SEM) plasma corticosterone concentrations are significantly (bars with differing letters above, such as A vs B) elevated by aggressive social interactions, in both Escape (light gray bar) and Stay (dark gray bar) phenotype mice compared to home-cage controls (white bar). B) Among Escape mice, icv injection with an Orx 2 antagonist 2 days prior (on day 3, light gray left hatched bar) does not affect plasma corticosterone on test (5) day. In contrast, among Stay phenotypes, icv injection on day 3 with an Orx 2 antagonist (dark gray, left hatched bar) stimulates increased plasma corticosterone measured on day 5 compared to vehicle-treated Stay mice (B vs A), antagonist-treated Escape mice (B vs A), and C) Orx 2 agonist (dark gray, right hatched bar; C vs B) Stay mice. et al., 2015, 2016), including by Orx stimulation (Flores et al., 2014(Flores et al., , 2017, comparable to those stimulated (EGR 1 labeling) in this study by Orx 2 agonist treatment in Stay mice. The agonist of Orx 2 receptors also limited the amount of conflict freezing during SAM interaction. Therefore, regardless of the cued or contextual nature of the fearful US (aggression), escape reduced this anxious behavior. Interestingly, after the acquisition of fear memory, in cue and context conditioning, Orx A impairs extinction of fear memories, which is reversed by an Orx 1 receptor antagonist (Flores et al., 2014). These effects are manifest via the basolateral amygdala (Flores et al., 2014(Flores et al., , 2017, the same brain region for which our results suggested anxiolytic actions of the Orx 2 receptor (Arendt et al., 2014). Fear conditioning also has specific effects on gene expression, including opposing actions on Orx 1 and Orx 2 receptor mRNA expression in the hippocampus, which are coincident with cannabinoid Cb 2 receptor expression changes there (Robertson et al., 2017). Anxiolytic and antinociceptive actions of phytocannabinoids are modulated via actions of Orx 2 receptors (Flores et al., 2016), a result that corroborates the conclusions of the work presented here.
Fear potentiated startle is another reliable paradigm for analyzing anxious behavior (Davis, 1979(Davis, , 1986Davis et al., 2003). We measured startle, in addition to freezing, during SAM social interactions, in which startle occurs during approach from a CD1 aggressor. In the context of repeated aggressive social interactions, which produces intensely stressful responses (Robertson et al., 2015;Smith et al., 2016), startle reactions are significantly escalated in Stay phenotype mice over time as the aggressive interactions are reiterated (Fig. 5A). Escape provides an anxiolytic behavioral solution, which nearly eliminates conflict startle responses (Fig. 5A). Injection (icv) of an Orx 2 agonist, significantly reduces the number of Stay animal startle responses (Fig. 5B), after which startle responding remains low on the fourth day of social interaction (Fig. 5B), potentially revealing a longer-term effect of the drug. In Stay mice, which don't make use of the escape route, startle responses escalate over time and freezing increases by the second day and remained high for the duration of the experiment (Fig. 4A). Although, icv Orx 2 antagonist (MK-1064) injection appeared to increase freezing to over 50% of the time in the SAM arena, startle was not similarly affected. It may be that more time freezing left less social interaction time for startle responses. As both startle and freezing anxious responses are ameliorated by Orx 2 agonism, it suggests the possibility that proactive anxiety reducing behaviors are also influenced by Orx 2 activation.
Since escape via the L-shaped tunnel at either end of the SAM arena is potently anxiolytic (Robertson et al., 2015;Smith et al., 2016), we hypothesized that time spent being attentive to the escape route would represent an anxiolytic response. Escape phenotype mice spend a significantly greater amount of time attentive to the hole (Fig. 6A). Consistent with our hypotheses, icv injection of Orx 2 agonist significantly increased attentiveness to the escape tunnel on injection day (Fig. 6B). Furthermore, similar to startle responses, attentiveness to the hole was present and significantly greater the day after injection (day 4). These results suggest longer-term effects of the drug, or alternatively, Orx 2induced neuroplastic changes. These results, in combination with Orx 2 agonism promoting escape, and Orx 2 antagonism reducing escape behavior suggests that Orx 2 activation encourages active, anxiolytic, stress-resilient behavior.
In order to validate these SAM-specific measurements of Orx 2 -induced anxiolysis we replicated the classic paradigm for indicating social stress susceptibility and resilience, the Social Interaction/Preference test (SIP) . First, we compared SIP results in vehicle-treated animals based on behavioral phenotype. The results indicated that Escape mice spent more time near the novel container occupied with a social target (CD1 mouse) than near the novel container alone (Fig. 7A and B). This result is typically interpreted as resilience to stress, anxiety and depression (Arendt et al., 2014;Berton et al., 2007;Golden et al., 2011;Krishnan et al., 2007). Escaping mice had a resilience/susceptibility ratio which is E G F Fig. 9. Anxiolysis in Stay mice is accompanied by activation of BLA or ItC inhibitory interneurons. Photomicrographs (magnification 40×) of the amygdala in the AP -1.06, centered on ML + or -2.65 and DV 4.50 relative to bregma were triple labeled with A) the immediate-early gene product early growth response protein 1 (EGR 1 , Zif 268 , NGF 1A , Krox 24 , or TIS 8 ), B) the GABA neuron calcium-binding protein Parvalbumin (Pv), and C) the nuclear marker 4′,6-diamidino-2-phenylindole (DAPI). D) The EGR 1 /Pv/DAPI triple labeling is visible in the merged image; scale bar equals 50 μm. This confluence of labeling, particularly the presence of EGR 1 , is only evident in Stay mice that have been injected icv with the Orx 2 agonist [Ala 15 , D-Leu 11 ]-Orx B . In Pv-positive neurons E) only in Stay mice treated with Orx 2 agonist [Ala 15 , D-Leu 11 ]-Orx B was the EGR 1 labeling significantly increased (dark gray right-hatched bar) compared to cage control (clear bar), vehicle (dark gray bar) and Orx 2 antagonist (MK-1064)-treated mice (dark gray left-hatched bar). F) In Pv-negative neurons, a much smaller, but similar increase in EGR 1 labeling occurred. G) Labeling of Pv was similar in all treatment groups. significantly greater than that of Stay phenotype mice. In mice with the Stay phenotype, more time was spent near the empty container than near the social target (Fig. 7A), suggesting that Stay mice are susceptible to stress, anxious and depressive behavior. Approach and avoidance behaviors are regulated by parallel reciprocal orexinergic circuits to limbic regions such as the bed nucleus of the stria terminalis (BNST), paraventricular nucleus (PVN) and amygdala (Giardino et al., 2018). Elevated plasma corticosterone from samples taken on the day of SIP and CR testing in Stay mice supports this interpretation. These results indicate that anxiety-and depression-related behaviors in the SAM paradigm are specifically associated with Escape and Stay phenotypes. That is, Escape is a stress-resilient phenotype, and Stay is associated with stress vulnerability. The effect of icv Orx 2 antagonist MK-1064 is to reverse stress-resilience in Escape animals (Fig. 7A), consistent with our previous work showing that Orx 2 gene knockdown in the BLA promoted the same anxiogenic response (Arendt et al., 2014). Here we also demonstrate the opposite effect for icv injection of the Orx 2 agonist [Ala 11 , D-Leu 15 ]-Orx B , that is, enhanced resilience as indicated by more time spent with the container holding a social target by Stay phenotype mice (Fig. 7C).

Conclusions
Consistent with our previous work, the results reported here indicate that Orx 2 receptor activity, in contrast with Orx 1 receptor actions, are anxiolytic and antidepressive (Arendt et al., 2013(Arendt et al., , 2014. Results that devolve specifically from social interactions in the SAM arena, including Orx 2 activation that reduced cued and conflict fear conditioned freezing, reduced startle responses, increased attentiveness to the escape route, and promoted stress resilience, as well as recruitment of a discrete cluster of ItC or BLA GABAergic Pv-containing inhibitory interneurons, suggest that Orx 2 receptors, putatively expressed in these cells, are well placed to promote anxiolysis and reduce depressive behavior. Corroborating evidence comes from Orx 2 antagonism which reversed the Escape phenotype and increased plasma corticosterone in Stay mice. Taken together, data from agonist and antagonist give credence in multiple formats for this assertion. What is more, the SAM-specific results are supported and validated by the classic social defeat model measure of susceptibility and resilience (SIP test). As such, we suggest that the use of the SAM can be applied to evaluate social susceptibility and resilience in a shorter period of time with more behavioral valuations to verify the outcome. Finally, given that measurements from both SAM and SIP suggest that Orx 2 receptors mediate anxiolytic and antidepressive actions, targeting these receptors may lead to new potential pharmacotherapies.

Conflicts of interest
The authors declare that they have no conflicts of interest.