A neuronal coping mechanism linking stress-induced anxiety to motivation for reward Authors: Paul M. Klenowski, Rubing Zhao-Shea, Timothy G. Freels, Susanna Molas, Max Zinter, Peter M’Angale, Cong Xiao, Leonora Martinez-Núñez, Travis Thomson, Andrew R. Tapper Published: 2023 ABSTRACT Stress coping involves innate and active motivational behaviors that reduce anxiety under stressful situations. However, the neuronal bases directly linking stress, anxiety, and motivation are largely unknown. Here, we show that acute stressors activate mouse GABAergic neurons in the interpeduncular nucleus (IPN). Stress-coping behavior including self-grooming and reward behavior including sucrose consumption inherently reduced IPN GABAergic neuron activity. Optogenetic silencing of IPN GABAergic neuron activation during acute stress episodes mimicked coping strategies and alleviated anxiety-like behavior. In a mouse model of stress-enhanced motivation for sucrose seeking, photoinhibition of IPN GABAergic neurons reduced stress-induced motivation for sucrose, whereas photoactivation of IPN GABAergic neurons or excitatory inputs from medial habenula potentiated sucrose seeking. Single-cell sequencing, fiber photometry, and optogenetic experiments revealed that stress-activated IPN GABAergic neurons that drive motivated sucrose seeking express somatostatin. Together, these data suggest that stress induces innate behaviors and motivates reward seeking to oppose IPN neuronal activation as an anxiolytic stress-coping mechanism. INTRODUCTION Stressful situations coordinate complex biological mechanisms and neural circuits that detect homeostatic challenge and imminent threat, resulting in behavioral outputs that minimize homeostatic disruption and increase survival (1, 2). Cognitive processing of stressful stimuli requires subjective appraisal to guide behavioral responses appropriate to the level of stress (3). Responses inconsistent with stress levels are maladaptive and underlie the development of neuropsychiatric disorders including anxiety-related illness and depression (3, 4). However, the processing and integration of neuronal activity in response to stress that motivates behavior to cope with the consequences of stress is not fully understood. Stress induces anxiety-like behavior via brain areas that respond to aversive and stressful stimuli, including the prefrontal cortex (5–8), amygdala (9–11), bed nucleus of the stria terminalis (12–14), and ventral hippocampus (15–17). In addition, stress can modulate the activity of dopaminergic neurons within motivational circuits that include the ventral tegmental area (VTA) and/or the nucleus accumbens (NAc) to mediate responses that are driven by changes in motivation to either avoid or seek out predictors of stress or reward (4, 18–23). However, the brain areas and neuronal mechanisms that directly control motivated coping behaviors during anxiogenic conditions remain unclear. An understudied brain area involved in negative and positive reinforcement of emotional and social behavior is the interpeduncular nucleus (IPN), located in the ventral midbrain (24–27). The IPN, which forms part of the medial habenula (MHb)-to-IPN axis (MHb-IPN) is an inhibitory brain area consisting predominantly of γ-aminobutyric acid (GABA)-releasing (GABAergic) neurons (28). The IPN is innervated by the prefrontal cortex, hypothalamus, raphe and septal nuclei, the lateral dorsal tegmental nucleus (LDTg), and the VTA, while the IPN projects to the median and dorsal raphe and the LDTg, among other brain regions (29–38) and is therefore well positioned to integrate emotional and/or motivational stimuli. Previous work has shown that heightened activity within the MHb-IPN pathway controls motivated nicotine consumption (39) and aversion (40–42). Moreover, GABAergic neurons within the IPN are activated by nicotine withdrawal (38, 43), a known stressor in mice (44–46), leading to increased somatic and affective withdrawal behavior (38, 42, 47, 48). The release of corticotropin-releasing hormone (CRH) into the IPN binds to CRH1 receptors to induce anxiety during nicotine withdrawal, suggesting that the IPN forms part of the extrahypothalamic stress circuitry in the brain (48). Increased IPN GABAergic activity also induces avoidance and motivational drive of novel versus familiar stimuli and is controlled by optogenetic manipulation of IPN GABAergic activity (26). Moreover, we recently showed that IPN GABAergic neurons exhibit heightened activity during nicotine withdrawal, which controls withdrawal-induced anxiety and the expression of somatic behaviors including self-grooming and scratching, that opposes IPN activation (43). Thus, the IPN can influence behavioral reactivity in response to emotional states including nicotine withdrawal and highly salient stimuli. Because the neuronal activity dynamics of the IPN have not been elucidated in detail, whether the IPN responds to acute stressors other than nicotine withdrawal is unknown. How stress-induced anxiety may drive motivated behavior and whether this involves changes in IPN neuronal activity remain to be explored. RESULTS Stress increases IPN GABAergic neuron activity To determine in vivo responses of IPN GABAergic neuron activity during acute stress, we expressed a Cre-dependent genetically encoded calcium sensor, GCaMP6m (49), into the IPN of glutamic acid decarboxylase-2 (GAD2)-Cre mice via adeno-associated virus (AAV) serotype 5 (AAV5)–mediated gene delivery and used fiber photometry to record bulk calcium transients during restraint stress (Fig. 1A and fig. S1, A and B). Mice were exposed to a brief period of restraint (5 min), which was sufficient to induce a significant increase in blood plasma corticosterone (CORT) compared to nonstressed controls (unpaired two-tailed t test [t5.53 = 6.26, P = 0.0011], n = 6; Fig. 1, B and C). Recordings of GCaMP fluorescence were made before, during, and after restraint (Fig. 1, D and E) and were analyzed and quantified to determine area under the curve (AUC), maximum change in fluorescence (dF/F0), and mean peak amplitude (Fig. 1, F to H). Analysis of recordings (n = 20) revealed significant increases in GCaMP signal from IPN GAD2+ neurons during and after stress, compared to no-stress baseline activity determined by AUC [repeated-measures one-way analysis of variance (ANOVA) [F1.49,28.3 = 14.31, P = 0.0002]; posttests: no stress versus stress, P < 0.0001; no stress versus after stress, P = 0.0001; and stress versus after stress, P = 0.9772; Fig. 1F], maximum dF/F0 (repeated-measures one-way ANOVA [F1.234,23.45 = 18.34, P = 0.0001]; posttests: no stress versus stress, P = 0.0001; no stress versus after stress, P < 0.0001; and stress versus after stress, P = 0.0343; Fig. 1G), and mean peak amplitude (repeated-measures one-way ANOVA [F1.333,25.32 = 13.35, P = 0.0005]; posttests: no stress versus stress, P = 0.0005; no stress versus after stress, P < 0.0001; and stress versus after stress, P = 0.2901; Fig. 1H). To rule out an involvement of VTA GABAergic neurons, we implanted optic fibers in the VTA at sites that were dorsal to IPN recording locations and measured GAD2+ neuron activity (fig. S1, C to H). No difference in AUC [repeated-measures one-way ANOVA [F1.712,10.27 = 0.4779, P = 0.6048]; posttests: no stress versus stress, P = 0.9987; no stress versus after stress, P = 0.6790; and stress versus after stress, P = 0.7536; fig. S1F), maximum dF/F0 (repeated-measures one-way ANOVA [F1.412,8.473 = 2.224, P = 0.1716]; posttests: no stress versus stress, P = 0.5717; no stress versus after stress, P = 0.9844; and stress versus after stress, P = 0.4590; fig. S1G), and mean peak amplitude (repeated-measures one-way ANOVA [F1.322,7.932 = 1.567, P = 0.2566]; posttests: no stress versus stress, P = 0.1606; no stress versus after stress, P > 0.9999; and stress versus after stress, P > 0.9999; fig. S1H) was detected in GCaMP signal (n = 7) recorded from VTA GAD2+ neurons in response to stress. To confirm that stress activated neurons within the IPN, brain slices from separate C57BL/6J restraint-stressed or unstressed mice were immunolabeled for c-fos, a marker of neuronal activation. We observed a significant increase in c-fos expression within the IPN of mice (n = 3) following restraint stress compared to no-stress controls (unpaired two-tailed t test [t8.04 = 4, P = 0.01]; Fig. 1, I and J). We also measured GCaMP activity of IPN GAD2+ neurons in response to an additional acute stressor, an aversive foot shock (0.5 mA, 1-s duration). We recorded GCaMP activity (n = 6) and compared z-scores of 2-s epochs before and following onset of each foot-shock delivery (fig. S1, I to K). We observed a significant increase in GCaMP signal before versus after onset of foot shocks (paired two-tailed t test [t5 = 2.878, P = 0.03]; fig. S1K). Together, these data show that the IPN is a stress-responsive brain area, where stress enhances GABAergic neuronal activity. Restraint stress increases IPN neuronal activity. (A) Experimental design for recording GCaMP activity from IPN neurons in GAD2-Cre mice during no-stress and stress conditions. (B) Strategy for measuring plasma CORT. ELISA, enzyme-linked immunosorbent assay. (C) Plasma CORT measurements taken from no-stress (control) and restraint-stressed (5 min) mice (n = 6). (D) Recording protocol to determine changes in GCaMP fluorescence before, during, and after restraint stress. (E) Representative GCaMP recordings from pre-stress (left), stress (middle), and post-stress (right) conditions. Analysis of area under the curve (AUC) (F), maximum dF/F0 (G), and mean peak amplitude (H) from GCaMP recordings (n = 20). (I) Representative images of IPN c-fos expression from no-stress (left) and restraint-stressed (right) mice (n = 3; bregma, −3.4 mm). (J) Analysis of c-fos expression across treatment groups expressed as number of neurons/mm2 of IPN tissue. IPR, IPN rostral; IPL, IPN lateral; IPI, IPN intermediate; IPC, IPN caudal. Scale bars, 100 μm. Data presented as means ± SEM. Unpaired t tests or repeated-measures one-way ANOVAs with Tukey’s posttests. *P < 0.05, **P < 0.01, ***P < 0.001, and ****P < 0.0001. Optical inhibition of GABAergic IPN neurons alleviates stress-induced anxiety To test whether stress activation of IPN GABAergic neurons is necessary for anxiety-like behavior, we used AAV2-mediated gene delivery to express a Cre-dependent halorhodopsin (NpHR) or enhanced yellow fluorescent protein (eYFP) as a control into the IPN of GAD2-Cre mice (Fig. 2, A and B, and fig. S2, A and B) as previously described (26, 43). Anxiety-like behavior was measured in the elevated plus maze (EPM; Fig. 2, B and C) and open-field test (Fig. 2, B and D to G) in NpHR- and eYFP-expressing mice without stress or following 5 min of restraint during light delivery to the IPN (593-nm constant light, 20 s on, 5 s off). In the EPM, two-way ANOVA of open-arm time revealed a significant genotype (genotype [F1,27 = 6.26, P = 0.01] and genotype × stress interaction [F1,27 = 7.54, P = 0.01]). Posttests revealed that restraint stress significantly decreased open-arm exploration in eYFP mice compared to unstressed eYFP and NpHR mice (no-stress eYFP versus stress eYFP, P = 0.02, n = 9; no-stress NpHR versus stress eYFP, P = 0.04, n = 7 of 9; Bonferroni’s posttests; Fig. 2C), while optical inhibition of IPN GAD2+ neurons increased open-arm exploration in stressed NpHR mice compared to stressed eYFP mice (stress eYFP versus stress NpHR, P = 0.006, n = 7 of 9; Bonferroni’s posttests; Fig. 2C). No difference in total arm entries was observed across groups (two-way ANOVA with Bonferroni’s posttests; Fig. 2C). Two-way ANOVA of open-field center and periphery time revealed a significant genotype (genotype [F1,24 = 5.17, P = 0.03] and stress interaction [F1,24 = 8.35, P = 0.008]). Posttest analysis revealed that stressed eYFP mice spent significantly less time in the center of the arena (no-stress eYFP versus stress eYFP, P = 0.008, n = 8; no-stress NpHR versus stress eYFP, P = 0.008, n = 6 of 8; Bonferroni’s posttests; Fig. 2, D and E) and more time in the periphery (no-stress eYFP versus stress eYFP, P = 0.008, n = 8; no-stress NpHR versus stress eYFP, P = 0.008, n = 6 of 8; Bonferroni’s posttests; Fig. 2, D and F) compared to no-stress eYFP and NpHR mice. Under light delivery, stressed NpHR mice spent significantly more time in the center of the arena (stress eYFP versus stress NpHR, P = 0.045, n = 7 of 9; Bonferroni’s posttests; Fig. 2, D and E) and less time in the periphery (stress eYFP versus stress NpHR, P = 0.045, n = 7 of 9; Bonferroni’s posttests; Fig. 2, D and F) compared to stressed eYFP mice, while there was no difference in distance traveled between groups (Fig. 2, D and G). These data show that optogenetic inhibition of IPN GABAergic neurons is anxiolytic and sufficient to block stress-induced anxiety. Optical inhibition of IPN GAD2+ neurons reduces anxiety-like behavior following stress. (A) NpHR or eYFP was injected into the IPN of GAD2-Cre mice. (B) Under no stress or following 5 min of restraint stress, NpHR and eYFP mice spent 5 min exploring the elevated plus maze (EPM) followed by 10 min exploring the open-field arena during light delivery into the IPN. (C) Time spent in open arms (top) and total arm entries (bottom) of the EPM (n= 7 to 9 per group). (D) Representative heatmaps showing location of eYFP and NpHR mice under no stress (left) and following restraint stress (right) in the open-field arena. (E) Time spent in the center and periphery (F) of the open-field arena (n = 6 to 8 per group). (G) Total distance traveled in the open-field arena. Scale bars, 100 μm. Data presented as means ± SEM. Two-way ANOVAs with Bonferroni’s posttests. *P < 0.05 and **P < 0.01. Stress-induced grooming decreases IPN GABAergic neuron activity Acute stress is a robust mediator of self-grooming in rodents (50), which is an inherent behavioral response associated with positive valence (51, 52). Our acute stress–restraint paradigm (Fig. 3A) triggered a robust increase in time spent self-grooming (paired two-tailed t test [t19 = 8.79, P < 0.0001]; Fig. 3B) compared to before restraint (n = 20). To investigate the influence of self-grooming on IPN neuron activity under no-stress and stress conditions, we analyzed GCaMP activity from IPN GAD2+ neurons during grooming bouts that were time-locked to the photometry recordings (Fig. 3, C to L; and figs. S1, A and B, and S3, G, H). Under no stress, analysis of the average z-score for all grooming events from each mouse (n = 12; Fig. 3, E and F) revealed a nonsignificant trend toward decreased IPN GAD2+ neuron GCaMP signal following onset compared to before onset of grooming (paired two-tailed t test [t11 = 2.187, P = 0.051]; Fig. 3G). Restraint stress significantly reduced the latency to groom compared to the no-stress condition ([t30 = 9.542, P < 0.0001]; Fig. 3H). Analysis of GCaMP signal from IPN GAD2+ neurons during grooming bouts following restraint stress (n = 20; Fig. 3, I to K) revealed a significant decrease in GCaMP activity following onset compared to before onset of grooming (paired two-tailed t test [t19 = 4.706, P = 0.0002]; Fig. 3L]. In addition, the average z-scored reduction in the GCaMP signal after onset of grooming was significantly higher in mice following stress compared to no stress [t30 = 2.451, P = 0.02]. To exclude the potential of reduced movement influencing the neural signal during grooming, we compared neural activity before and after onset of periods of immobility where mice displayed limited movement. Immobility events were generated from 5-s epochs where mice exhibited a significant reduction in the average distance traveled (paired two-tailed t test [t11 = 23.06, P < 0.0001], n = 12) and average velocity (paired two-tailed t test [t11 = 23.06, P < 0.0001], n = 12) compared to before onset (fig. S3, A and B). Analysis of time-locked photometry recordings (fig. S3, C and D) showed that there was no difference in IPN GAD2+ neuron GCaMP signal before versus after onset of immobility events (paired two-tailed t test [t11 = 1.285, P = 0.23], n = 12; fig. S3, E and F). These results suggest that reduced movement during grooming events does not influence the neuronal signal. To validate that stress-induced grooming represents a motivational coping strategy that engages midbrain VTA dopamine (DA) activity, we selectively expressed GCaMP in the VTA of dopamine transporter (DAT)-Cre mice, and signals from VTA DAT+ neurons were recorded following restraint (Fig. 3, M to O, and fig. S3I). During bouts of post-stress grooming, we detected significant increases in calcium signal following onset compared to before onset (paired two-tailed t test [t10 = 4.424, P = 0.0013]; Fig. 3, P to R), suggesting that DA neurons in the VTA are engaged during post-stress grooming. These data show that increased self-grooming in response to stress can oppose increases in IPN activity. Post-stress self-grooming behavior decreases activity of IPN GAD2+ neurons and increases activity of VTA DAT+ neurons. (A) Experimental strategy for recording GCaMP activity from IPN neurons in GAD2-Cre mice during self-grooming behavior under no stress or stress conditions. (B) Time spent grooming before and following restraint stress (n = 20). (C) Representative GCaMP expression and recording location in the IPN (D) with GCaMP fluorescence and time-locked grooming bouts depicted in pink-shaded areas under no stress. (E) Z-score of GCaMP signal before and after onset of grooming bouts from all mice (n = 12) with arrowhead and blue area indicating onset of grooming. (F) Heatmap showing z-score of GCaMP signal before and after onset of grooming bouts from individual mice (one mouse per column). (G) Comparative analysis of z-scores before versus after onset of grooming bouts calculated from the average per mouse. (H) Restraint stress significantly reduced the latency to groom compared to the no-stress condition. (I) Representative photometry recording following restraint stress with time-locked grooming bouts depicted in pink-shaded areas. (J) Z-score of GCaMP signal before and after onset of grooming bouts from all mice (n = 20). (K) Heatmap showing z-score of GCaMP signal before and after onset of grooming bouts from individual mice. (L) Comparative analysis of z-scores before versus after onset of grooming bouts. (M) Strategy for recording GCaMP activity from VTA neurons in DAT-Cre mice (n = 11). (N) Representative GCaMP expression and fiber location in the VTA (O) with photometry recording and time-locked grooming bouts depicted in blue-shaded areas. (P) Z-score of GCaMP signal before and after onset of grooming bouts with arrowhead and pink area indicating the grooming onset. (Q) Heatmap and comparative analysis (R) of z-scores before versus after onset of grooming bouts in. Scale bars, 100 μm. Data presented as means ± SEM. Unpaired or paired two-tailed t tests. **P < 0.01 and ****P < 0.0001. Self-grooming behavior is modulated by optogenetic manipulation of IPN GABAergic neuron activity Recent work has implicated self-grooming as a behavioral mechanism of self-soothing that occurs via activity displacement and reduction of a negative affective state (51). We hypothesized that optogenetic inhibition of IPN GABAergic neurons, which is anxiolytic under conditions of stress, would also reduce self-grooming. We measured grooming time following restraint stress in NpHR (n = 9) and eYFP mice (n = 6) in the absence and presence of light (Fig. 4, A and B, and fig. S4, A and B). Two-way ANOVA of grooming time revealed a significant light condition [F1,13 = 7.115, P = 0.019] and virus interaction [F1,13 = 10.49, P = 0.007]. Posttests revealed that, during light delivery, NpHR mice spent significantly less time grooming in light-on compared to light-off conditions (light-off NpHR versus light-on NpHR, P = 0.03; Bonferroni’s posttests; Fig. 4B), while no difference in grooming behavior was observed in eYFP mice in the absence or presence of light (Fig. 4B). We also measured grooming time in channelrhodopsin (ChR2)–expressing mice (n = 6) and eYFP controls (n = 7), with and without light delivery (Fig. 4, A and C, and fig. S4, C and D). Two-way ANOVA of grooming time revealed a significant light [F1,11 = 12.87, P = 0.004], virus [F1,11 = 21.51, P = 0.0007], and light × virus interaction [F1,11 = 17.59, P = 0.0015]. When IPN GAD2+ neurons were photoactivated with ChR2, we observed a significant increase in grooming time compared to when light was absent (light-off ChR2 versus light-on ChR2, P = 0.0002; Bonferroni’s posttests; Fig. 4C). Control eYFP mice exhibited no change in grooming behavior in the absence or presence of light (Fig. 4C). These data show that IPN activation induces self-grooming behavior to potentially reduce anxiety-like behavior that is mediated during periods of heightened IPN activity. Optogenetic manipulation of IPN GAD2+ neuron activity influences self-grooming behavior. (A) Grooming behavior was measured in NpHR or eYFP mice (n = 6 to 9 per group) during light-off and light-on conditions after stress and in ChR2 or eYFP mice (n = 6 to 7 per group) during delivery of blue light into the IPN. (B) Time spent grooming in NpHR and eYFP mice. (C) Time spent grooming in ChR2 and eYPF mice. Scale bars, 100 μm. Data presented as means ± SEM. Two-way ANOVAs with Bonferroni’s posttests or unpaired t test. **P < 0.01 and ***P < 0.001. Sucrose consumption reduces IPN GABAergic neuron activity in a stress-dependent manner Accumulating evidence suggests that natural rewards including consumption of palatable foods such as sucrose (53, 54) can act to blunt stress responses and reduce anxiety (55–57). However, the direct influence of sucrose reward consumption on IPN GABAergic activity has not been determined. Thus, we measured GCaMP fluorescence from IPN GAD2+ neurons during recording sessions where ad libitum mice (n = 6) had access to sucrose pellets in their home cage (Fig. 5A). We observed a nonsignificant trend toward decreased GCaMP signal from IPN GAD2+ neurons before versus after onset of sucrose consumption bouts (paired two-tailed t test [t5 = 2.418, P = 0.0603]; Fig. 5A and fig. S5, A and C). Because previous studies have shown that acute stress can facilitate changes in reward sensitivity (21, 58–60), we tested whether the response to sucrose consumption on IPN activity was enhanced under conditions of stress including food restriction (mild stress) and following restraint that increased stress level, as determined by plasma CORT (one-way ANOVA [F2,15 = 29.11, P < 0.0001]; posttests: no-food restriction versus food restriction, P = 0.0245; no-food restriction versus restraint stress, P < 0.0001; and food restriction versus restraint stress, P = 0.0003; Fig. 5, A to C, and fig. S5B). During food restriction (paired two-tailed t test [t5 = 4.292, P = 0.0078]; Fig. 5B and fig. S5A) and following restraint stress (paired two-tailed t test [t5 = 7.057, P = 0.0009]; Fig. 5C and fig. S5A), we detected significant decreases in GCaMP signal from IPN GAD2+ neurons after onset of sucrose consumption compared to prior to consumption. Comparison of event plots across treatments (Fig. 5D) revealed that the magnitude of the sucrose response increased in a manner that was dependent on stress level (repeated-measures one-way ANOVA [F1.156,5.778 = 20.88, P = 0.0037]; posttests: no-food restriction versus food restriction, P = 0.0041; no-food restriction versus restraint stress, P = 0.0099; and food restriction versus restraint stress, P = 0.0406; Fig. 5E). These results demonstrate that inhibitory responses of IPN GABAergic neurons mediated by consumption of natural reward emerge during periods of stress. Responses of IPN GAD2+ neurons during sucrose consumption in mice without food restriction, during food restriction and following restraint stress. (A) GCaMP fluorescence of IPN neurons was recorded from GAD2-Cre mice without food restriction, during food restriction (B) and following restraint stress (C). [(A) to (C), middle panels] Representative GCaMP recordings with time-locked periods of sucrose consumption depicted in blue-shaded areas. [(A) to (C), right panels] Z-score of GCaMP signal before and after onset of sucrose consumption bouts from all mice (n = 6) under each condition. (D) Event plot showing z-scores across treatments. (E) Comparative analysis of sucrose responses across conditions. Data presented as means ± SEM. Repeated-measures one-way ANOVAs with Tukey’s posttests. *P < 0.05 and **P < 0.01. Optical silencing of IPN GABAergic neurons inhibits stress-induced increases in motivated sucrose seeking The graded inhibitory response of IPN GABAergic neurons to sucrose with increasing stressor magnitude suggests that stress activation of the IPN may motivate reward seeking to reduce IPN activity as a coping mechanism to alleviate anxiety. To test this hypothesis, we measured motivated sucrose seeking in an operant sucrose self-administration paradigm in stressed and control mice using optogenetics to silence stress induced activity of IPN GABAergic neurons. Following 2 weeks of habituation to sucrose and successful acquisition of operant responding (see Materials and Methods), mice underwent progressive ration (PR) sessions to determine breakpoint, which was used as a measure of motivation (Fig. 6A). We determined breakpoint in nonstressed NpHR control mice (n = 9) and following restraint stress with or without light illumination (Fig. 6, B to D, and fig. S6D). Sucrose breakpoint increased in mice following restraint stress compared to the breakpoint determined without stress (Fig. 6C). In contrast, photoinhibited mice did not exhibit stress-induced increases in breakpoint (repeated-measures one-way ANOVA [F1.198,23.45 = 15.85, P = 0.0021]; posttests: no stress versus stress light off, P = 0.0044; no stress versus stress light on, P = 0.1035; and stress light off versus stress light on, P = 0.0088; Fig. 6C). Light delivery did not change breakpoint under no-stress conditions (paired two-tailed t test [t6 = 0.2634, P = 0.8011], n = 7; Fig. 6D) or in eYFP mice following stress (paired two-tailed t test [t5 = 0.9608, P = 0.3808], n = 6; Fig. 6E and fig. S6F). Next, we tested whether activation of IPN GAD2+ neurons was sufficient to increase motivation for sucrose. To this end, we measured sucrose breakpoint in GAD2-Cre mice infected with AAV2 Cre-dependent ChR2 or eYFP as a control (Fig. 6F and fig. S6E). Two-way ANOVA of sucrose breakpoint revealed a light × virus interaction [F1,12 = 8.453, P = 0.01]. Posttests revealed that, during light delivery (473 nm, 20 Hz, 25% duty cycle, 10 min on, 5 min off), ChR2-expressing mice (n = 8) exhibited increased breakpoint compared to the no-light condition (light-off ChR2 versus light-on ChR2, P = 0.007; Bonferroni’s posttests; Fig. 6G). Conversely, in eYFP-expressing mice (n = 6), light delivery did not alter breakpoint (light-off eYFP versus light-on eYFP, P = 0.96; Bonferroni’s posttests; Fig. 6G). To control for light spillover activating GABAergic neurons in the VTA, which are in close proximity to the IPN and also play a role in motivation for sucrose reward (61), we stimulated MHb-IPN inputs that provide the primary source of acetylcholine/glutamate specifically to the IPN. We injected MHb of ChAT-Cre mice with AAV2 Cre-dependent ChR2 or eYFP that allowed for direct stimulation of MHb-IPN terminals. We found that light stimulation of MHb-IPN terminals [473 nm, 20 Hz, 25% duty cycle; (26)] in ChAT-Cre mice expressing ChR2 increased IPN c-fos expression compared to no-light controls (unpaired two-tailed t test [t5.991 = 19.87, P < 0.0001], n = 4; fig. S6, A and B] and increased self-grooming in light versus no-light conditions (paired one-tailed t test [t3 = 3.086, P = 0.0269], n = 4; fig. S6C). We also photostimulated MHb-IPN terminals during sucrose PR sessions with and without stress (Fig. 6, H to J, and fig. S6G). In ChR2-expressing ChAT-Cre mice (n = 10), two-way ANOVA analysis revealed a significant light [F1,9 = 23.27, P = 0.0053] and stress interaction [F1,9 = 19.67, P = 0.0016]. Posttests revealed that optical stimulation of MHb-IPN terminals increased breakpoint following restraint stress (stress light off versus stress light on, P = 0.044; Fig. 6I) but did not change breakpoint under no-stress conditions (no-stress light off versus no-stress light on, P > 0.9999; Fig. 6I). In eYFP-expressing ChAT-Cre mice (n = 7), two-way ANOVA analysis revealed a significant stress interaction [F1,6 = 6.805, P = 0.0402], while posttests showed no differences in breakpoint in the absence or presence of light under no-stress (no-stress light off versus no-stress light on, P > 0.9999; Fig. 6J and fig. S6H) or stress conditions (stress light off versus stress light on, P > 0.9999; Fig. 6J and fig. S6H). These results show that heightened IPN activity motivates reward-seeking, while optical inhibition of the IPN can block stress-induced increases in motivation for sucrose. Optical silencing of IPN GAD2+ neurons blocks stress-induced increases in motivated sucrose seeking, and optical activation of IPN GAD2+ neurons and MHb-IPN inputs enhances motivated sucrose seeking. (A) Experimental timeline for sucrose drinking in the dark, operant training, and PR testing of sucrose motivation under no-stress (NS) or stress (S) conditions. (B) Motivated sucrose-seeking (PR) sessions were conducted during light-off and light-on conditions in GAD2-Cre mice injected with NpHR or eYFP. (C) Sucrose breakpoint measured in NpHR mice without stress and following restraint stress in the absence or presence of light delivered to the IPN (n = 9). (D) Sucrose breakpoint measured in NpHR mice with no stress in light-off and light-on conditions (n = 7). (E) Sucrose breakpoint measured in eYFP mice following stress with and without light (n = 6). (F) Motivated sucrose-seeking sessions were conducted during light-off and light-on conditions in the IPN of GAD2-Cre mice injected with ChR2 or eYFP. (G) Sucrose breakpoint measured in ChR2 and eYFP mice in light-off and light-on conditions (n = 6 to 8 per group). (H) PR sessions were performed in ChAT-Cre mice with ChR2 or eYFP injected into the MHb and during no-light or light delivery into the IPN with or without stress exposure. (I) Sucrose breakpoints measured during no-light or light stimulation of MHb-IPN terminals in ChAT-Cre ChR2-expressing mice with and without stress (n = 10). (J) Sucrose breakpoints measured during light-off and light-on conditions in ChAT-Cre eYFP-expressing mice with and without stress (n = 7). Scale bars, 100 μm. Data presented as means ± SEM. Paired t test, repeated-measures one-way ANOVAs with Tukey’s posttests, or repeated measures two-way ANOVAs with Bonferroni’s posttests. *P < 0.05 and **P < 0.01. Somatostatin-expressing IPN neurons are required for stress-induced increases in motivated sucrose seeking Because IPN GABAergic neurons are a heterogenous population (35, 38, 62), we used single-cell RNA sequencing (scRNA-seq) to identify neuronal subpopulations within the IPN that exhibit increased early immediate gene expression in mice (n = 5) exposed to stress or nonstressed conditions (Fig. 7A). We identified 18 cell clusters that were then annotated and assessed for expression of canonical markers of the mammalian central nervous system (63, 64) by the Wilcoxon rank sum test (fig. S7A). Reclustering of cells based on neuronal identity markers and visualization by Uniform Manifold Approximation and Projection (UMAP) identified six discrete neuron clusters (Fig. 7B). The neuron clusters were then analyzed for distinct neuronal subtypes by visualization of canonical markers (63) and for differential expression of the immediate early gene c-fos across the treatment groups. From this, we identified an enriched GAD2+ neuronal cluster (cluster 4; fig. S7B) that showed significantly increased c-fos expression (fold change > 0.25) in response to stress when normalized c-fos counts across conditions were analyzed (Wilcoxon rank sum test, P = 0.0402; Fig. 7C). All other neuronal clusters did not show enrichment of c-fos in response to restraint stress (Wilcoxon rank sum test, P = 0.1565; Fig. 7C). The GAD2+ cluster was then analyzed for differentially expressed genes in comparison to all other neuronal clusters. Notably, we identified that the GAD2+ cluster showed a significant enrichment of somatostatin (SST; Fig. 7D). To probe in vivo responses of SST+ IPN neurons to stress, we used AAV5 to deliver Cre-dependent GCaMP into the IPN of SST-Cre mice (Fig. 7E), which allowed for GCaMP expression within SST+ IPN neurons (fig. S7C). We then recorded GCaMP fluorescence (n = 10) before, during, and after restraint (Fig. 7, E to I, and fig. S8A). We detected significant increases in AUC (repeated-measures one-way ANOVA [F1.898,17.08 = 7.296, P = 0.0056]; posttests: no stress versus stress, P = 0.0159; no stress versus after stress, P = 0.03; and stress versus after stress, P = 0.9642; Fig. 7G), maximum dF/F0 (repeated-measures one-way ANOVA [F1.427,12.85 = 4.635, P = 0.0402]; posttests: no stress versus stress, P = 0.0352; no stress versus after stress, P = 0.0431; and stress versus after stress, P > 0.9999; Fig. 7H), and mean peak amplitude (repeated-measures one-way ANOVA [F1.372,12.35 = 4.964, P = 0.0362]; posttests: no stress versus stress, P = 0.0439; no stress versus after stress, P = 0.0187; and stress versus after stress, P > 0.9999; Fig. 7I) of GCaMP fluorescence from IPN SST+ neurons recorded during and after stress compared to no-stress baseline fluorescence. To assess the role of IPN SST+ neuronal subtypes on motivated sucrose seeking, we measured sucrose breakpoint in SST-Cre mice (NpHR, n = 8; and eYFP, n = 7) under no-stress and stress conditions (Fig. 7J and fig. S8, B and C). We found that optical inhibition of IPN SST+ neurons was sufficient to block stress-mediated increases in motivated sucrose seeking (repeated-measures one-way ANOVA [F1.868,13.08 = 14.21, P = 0.0006]; posttests: no stress versus stress light off, P = 0.0096; no stress versus stress light on, P = 0.3527; and stress light off versus stress light on, P = 0.0055; Fig. 7K), while light delivery in eYFP control mice did not change breakpoints following stress (repeated-measures one-way ANOVA [F1.98,11.88 = 9.034, P = 0.0042]; posttests: no stress versus stress light off, P = 0.0134; no stress versus stress light on, P = 0.0496; and stress light off, versus stress light on, P = 0.5581; Fig. 7L). These results identify a stress-responsive IPN GABAergic subpopulation that is enriched for SST and controls stress-induced changes in motivated sucrose seeking. A subpopulation IPN GABAergic neurons containing SST is activated by restraint stress and control stress-mediated increases in motivated sucrose seeking. (A) Experimental design for scRNA-seq of IPN neurons from no-stress and stressed mice. (B) UMAP of IPN neuronal populations. (C) Comparison of c-fos expression across neuronal subclusters from control (no restraint) and restraint-stressed mice (n = 5). (D) Volcano plot identifying enriched SST expression within the neuronal cluster exhibiting increased c-fos expression in stressed mice. (E) Experimental strategy for recording GCaMP signal from IPN neurons of SST-Cre mice. (F) Representative GCaMP recordings from SST-Cre mice (n = 10) during no-stress (left), stress (middle), and post-stress (right) conditions. Analysis of AUC (G), maximum dF/F0 (H), and mean peak amplitude (I) from GCaMP recordings. (J) PR sessions were performed during light-off and light-on conditions in SST-Cre mice injected with NpHR or eYFP. (K) Sucrose breakpoint measured in NpHR mice (n = 8) without stress and following restraint stress in the absence or presence of light delivered to the IPN. (L) Sucrose breakpoint measured in eYFP mice (n = 7) with no stress and following restraint stress during light-off and light-on conditions. Scale bars, 100 μm. Data presented as means ± SEM. Wilcoxon rank sum test, unpaired t test, or repeated-measures one-way ANOVAs with Tukey’s posttests. *P < 0.05 and **P < 0.01. DISCUSSION Behavioral coping mechanisms that displace the effects of stress protect against exaggerated responses that can precipitate into emotional states underlying neuropsychiatric disorders, including depression and anxiety-related illness (3, 4). Our data have identified the IPN as a stress-responsive brain area and mediator of anxiety-like behavior. We also found that optogenetic inhibition of IPN GABAergic neurons is anxiolytic and mimicked by stress-mediated self-grooming and sucrose-consumption, suggesting that these behavioral responses emerge in response to stress as a coping mechanism that limits IPN overactivation to reduce anxiety-like behavior. Our data revealing heightened activity of IPN GABAergic neurons in response to restraint stress are in line with previous reports demonstrating activation of IPN neurons during nicotine withdrawal (38, 43, 65, 66), a known stressor (44–46). In addition, we found that optical inhibition of the IPN reduced the expression of anxiety-like behavior that was mediated by restraint stress and was sufficient to block stress-mediated self-grooming. Prior work suggests that the expression of the physical and affective symptoms of nicotine withdrawal involves glutamate release from MHb-IPN inputs (38, 48). Here, we found that optical activation of either GAD2+ IPN neurons or MHb-IPN terminals increased self-grooming behavior and that MHb-IPN terminal photoactivation increased IPN c-fos expression, in a manner similar to restraint stress. This finding is consistent with prior work showing that optogenetic activation of IPN GABAergic neurons elicits somatic withdrawal behavior in mice (38). We also observed reduced IPN GABAergic excitability during self-grooming following stress that was independent of reduced movement and/or velocity. The magnitude of this response was increased compared to grooming during no-stress conditions, which is similar to our previous observations in mice undergoing nicotine withdrawal (43), suggesting that these behaviors emerge as a coping mechanism to reduce increases in IPN activity that mediate the expression of anxiety-like behavior. Recent studies have provided evidence to support the emergence of self-grooming as an emotional response to stress (51, 52). This work has identified a limbic circuit consisting of the hippocampus ventral subiculum, ventral lateral septum, and the lateral hypothalamus tuberal nucleus that is activated during grooming behavior induced by stress (51). Similarly, activation of a circuit that triggers DA release in the NAc via an excitatory SST input from the paraleminiscal nucleus to VTA DA neurons encodes reward, is anxiolytic, and is also engaged during self-grooming behavior (52). Previous studies indicate that SST can mediate inhibitory responses primarily at dendritic sites within excitatory neuronal networks (67–69) and also at axonal sites within the MHb-IPN pathway (62). In addition, SST function is implicated in anxiety-like behavior (70, 71) and behavioral responses mediated by chronic stress (72–74). Our scRNA-seq data identified IPN stress-activated neurons as highly enriched for SST, suggesting that the IPN is part of the SSTergic network underlying self-grooming. Collectively, these results support self-grooming as an inherent behavioral response that displaces stress-induced activity, provides positive valence, and reduces anxiety-like behavior. Further investigation is required to identify the synaptic targets of IPN SST+ neurons that mediate stress-induced behavioral responses observed in this study. Although motivational responses to stress are mixed in animal models (75–78), likely due to the type and severity of the stressor (20, 54, 79), acute stress has been shown to increase “wanting” behavior for reward in humans (60, 80) and animals (59, 76, 80, 81), and this response appears more prominent in subjects with diets high in palatable foods (54, 76, 82). We found that the sensitivity of the inhibitory response on IPN GABAergic neuron activity during sucrose consumption increased with greater stress level. In addition, we found that, in sucrose-consuming mice, significant increases in motivated sucrose seeking occurred following acute restraint compared to motivation without stress, a response that was inhibited by photoinactivation of GABAergic IPN neurons and increased during optical stimulation of IPN GABAergic neurons even in the absence of stress. These data support previous studies demonstrating that activation of IPN neurons controls motivated nicotine consumption (39) and motivational drive of novel versus familiar stimuli (26) but indicate that stress activation of the IPN and stress state likely determine the level of motivation for reward consumption. In addition, these data suggest that the IPN may contribute to the motivational drive of other drugs of abuse during stressful situations such as during withdrawal or stress-induced reinstatement. One limitation of our study is that, while neuronal responses to stress and coping behavior did not grossly differ between sexes, subsequent optogenetic experiments were only performed in male mice. Previous work suggests that sex-specific differences within the IPN may contribute to enhanced anxiety-like behavior in female versus male rodents during nicotine withdrawal (83, 84). However, further investigation is required to determine whether IPN contributions to stress-induced anxiety and coping behavior including motivated reward seeking differ between females and males. Despite the defined role of stress in regulating appetite and food preference via the hypothalamic-pituitary-adrenal axis and the endocrine response (53), our results showing that IPN neuron excitability is reduced during reward consumption under stress and motivates sucrose seeking support previous findings that suggest that natural rewards can regulate emotional responses to stress and act as a coping strategy (54). Thus, our data provide a neuronal mechanism underlying how stress motivates reward seeking to reduce anxiety-like behavior: Stress increases activation of IPN GABAergic neurons triggering anxiety-like behavior while also driving reward seeking via SST-expressing neurons because reward consumption reduces IPN neuronal activity, thereby opposing the physiological response to stress and alleviating anxiety. Aberrant responses to stress are maladaptive and underlie behavioral symptoms of neuropsychiatric diseases including depression and anxiety disorders (3, 4). Our findings indicate stress-induced activation of the IPN is a critical mediator of anxiety-like behavior. We have also identified a neuronal mechanism whereby innate self-grooming and natural reward consumption reduces IPN neuron activity following stress exposure, providing a mechanistic link underlying how stress motivates coping responses to reduce anxiety-like behavior. Taken together our data indicate the IPN is a critical node balancing integrated emotional and motivated behavior triggered by stress. We anticipate that chronic stress may dysregulate this balance to elicit persistent anxiety-like behavior and inability to cope with stress, similar to what is observed in anxiety disorders and/or depression. MATERIALS AND METHODS Animals Animal experiments were performed in accordance with the Guide for the Care and Use of Laboratory Animals provided by the National Research Council using animal protocols approved by the Institutional Animal Care and Use Committee of the University of Massachusetts Chan Medical School (UMMS). Male and female GAD2-Cre mice (GAD2-Cre mice; stock no. 010802) were used for IPN photometry experiments, and no significant differences were observed. Male SST-Cre mice (SST-Cre mice; stock no. 013044) were used for IPN photometry experiments, and male GAD2-Cre and DAT-Cre mice (DAT-Cre mice; stock no. 006660) were used for VTA photometry experiments. Optogenetic experiments were performed in male GAD2-Cre, ChAT-Cre (ChAT-Cre mice; stock no 006410), and SST-Cre mice. Immunohistochemistry and CORT data were obtained from male C57Bl/6J mice (C57Bl/6J mice; stock no. 000664). All mice were obtained from the Jackson Laboratory (West Grove, PA, USA) and bred at the UMMS animal facility. Cre mice were crossed with C57Bl/6J mice, and only heterozygous offspring were used for experiments. Mice were housed at a maximum of 5 per cage and kept on a standard 12-hour light/dark cycle (light on at 7 a.m.) with access to standard chow and water ad libitum. Mice that received viral injections were given overnight to recover then transferred to a reverse 12-hour light/dark cycle (light on at 7 p.m.). Following fiber implantation, mice were given at least 1 week to recover before starting experiments. For stress photometry experiments, EPM and open-field testing, mice remained group-housed. For experiments involving food restriction, mice were single-housed following recovery from fiber implantation. All experiments were performed during the dark cycle. Viral preparation The plasmids pAAV-Ef1a-DIO-eNpHR3.0-eYFP, pAAV-Ef1a-DIO-ChR2-eYFP, and pAAV-Ef1a-DIO-eYFP were packaged into AAV2 viral particles at the UMMS Viral Vector Core. The titrations were 8.5 × 1012 genome copies/ml for pAAV-Ef1a-DIO-eNpHR3.0-eYFP, 2.5 × 1012 viral particles/ml for pAAV-Ef1a-DIO-ChR2-eYFP, and 5 × 1012 viral particles/ml for pAAV-Ef1a-DIO-eYFP. Cre-dependent AAV-Flex-GCaMP6m packaged in AAV5 particles (pAAV.CAG.Flex.GCaMP6m.WPRE.SV40) was purchased from Addgene (100839-AAV5). Viral injections were performed 4 to 6 weeks before the start of experiments to allow for adequate transgene expression. Stereotaxic surgeries Stereotaxic viral injections and fiber placements were performed using aseptic technique as described previously (43). Briefly, mice at 6 to 8 weeks of age were anaesthetized via intraperitoneal administration of ketamine (100 mg/kg) and xylazine (10 mg/kg) (VEDCO). Surgeries were performed with the aid of a stereotaxic frame (Stoelting Co.) and by using bregma and lambda as landmarks. Viral preparations were infused into the IPN using a Hamilton 10-μl neurosyringe (1701RN; Hamilton) and a microsyringe pump (Stoelting Co.) at the following coordinates in millimeters from bregma: anteroposterior (AP) at −3.4, mediolateral (ML) at ±0.5, and dorsoventral (DV) at 4.8, and 6° angle. The following coordinates in millimeters were used to target the MHb: AP at −1.7, ML at ±1.0, and DV at 2.9, and 10° angle; or the VTA that was injected bilaterally at AP at −3.4, ML at ±0.5, and DV at 4.2. Between 0.3 and 0.5 μl of each virus was delivered at a constant flow rate of between 30 and 50 nl/min. Following injection, the needle was left in place for 10 min before being slowly retracted. The skin was then closed and glued with Vetbond (3M). Post-surgery procedures Mice recovered in a clean cage on a heating pad and were given a subcutaneous injection of ketoprofen (Zoetis; 1 mg/kg) and monitored until sternal. Mice were given a minimum of 1 week to recover before experiments were performed. Following each experiment, mice were euthanized by sodium pentobarbital overdose (200 mg/kg), perfused with 4% paraformaldehyde (PFA), and the brains were removed for verification of virus expression and fiber placement. Verification was performed by experimenters blinded to behavioral outcome or treatment. Animals with off-target viral expression or optic fiber placement were excluded from analysis. Restraint stress All photometry experiments were performed during the dark cycle under dim red light. To record fluorescence from a genetically encoded calcium sensor (GCaMP6m), a Doric Instruments Fiber Photometry System connected to an light-emitting diode (LED) driver was used to deliver excitation light from LEDs at 465 nm and at 405 nm (~30- to 60-mW output at the fiber tip), which was used as an isosbestic wavelength for the calcium sensor (Doric Instruments). A 200-μm 0.48–numerical aperture optic fiber patch cord was used to deliver the light via a Dual Fluorescence Minicube (Doric Instruments), and emissions were detected with a femtowatt photoreceiver (Model 2151, Newport) and amplified by transimpedance amplification. The patch cord was connected to the optic fiber before the start of the experiment, and each recording session was performed within the home cage of each mouse. Mice were habituated to the handling and recording procedures for a minimum of 3 days before the start of the experiment. Before restraint, GCaMP fluorescence was recorded for 5 min to determine baseline activity. Each mouse was then restrained in a disposable plastic cone (Model MDC-200, Braintree Scientific), which was taped and cut above the head to allow for connection of the optic fiber to the patch cord. During restraint, mice were placed in their home cage, and GCaMP fluorescence was recorded for 5 min. Mice were then released from restraint, and GCaMP signal was recorded for an additional 10 min. Foot shock Mice were habituated to a fear conditioning cage and handling procedures for 3 days before the start of the experiment. On the test day, GCaMP fluorescence from IPN GABAergic neurons was recorded for 5 min before the first shock. During a 5-min foot-shock period, five shocks (0.5 mA, 1-s duration) were delivered at random intervals. During the foot-shock period, randomized foot shocks were time-stamped into the photometry recording via a transistor-transistor logic (TTL) pulse from the fear conditioning system. Post-stress behaviors To assess GCaMP activity during grooming behavior and sucrose consumption, a video camera was synchronized to the photometry system that time-locked the video and photometry recordings. Bouts of grooming and sucrose consumption were tallied using criteria previously described (43) from videos in a blinded fashion. Analysis of the GCaMP signal was done using the time-locked photometry recording in a manner similar to previous reports (43). Bouts of sucrose consumption were analyzed in mice without stress, during food restriction in which mice had their food intake restricted until they reached 90 to 95% of their free feeding weight and following 5 min of restraint. Sucrose was provided in the form of 20-mg pellets (Test Diet). Mice were preexposed to sucrose pellets for a minimum of 3 days before the start of experiments to reduce neophobia. For the no-food restriction and food restriction conditions, GCaMP signal from the IPN of GAD2-Cre mice was measured across two sessions of 15 min or until 10 sucrose bouts were recorded, whichever came first. For the restraint stress group, GCaMP signal was measured across one session for up to 30 min or until 10 sucrose bouts were recorded. Grooming behavior and GCaMP signal were recorded in mice without stress across two sessions of up to 30 min each or for 10 min in mice following 5 min of restraint stress exposure. Immobility events Immobility events were scored from a video recording using EthoVision XT 11.5 (Noldus Apparatus). Immobility events were determined using a MATLAB script that identified timestamps where the distance traveled was less than 1 cm. Events that were shorter than 5 s, were less than 5 s from the previous event, and/or contained grooming were excluded from analysis. This resulted in immobility events where mice traveled a distance less than 0.26 cm and exhibited a velocity of less than 3.1 cm/s in 5 s after onset of the event, which was primarily due to movement in the vertical direction. Behavioral assays Grooming behavior All behavioral assays were performed during the dark cycle under dim red light. Mice were habituated to handling and recording procedures for 3 days before the start of experiments. To assess grooming behavior, GAD2-Cre mice expressing either NpHR or eYFP in the IPN were restrained for 5 min as described above. Grooming behavior was then video recorded for 10 min after restraint. Each mouse underwent two test sessions a week apart, and light delivery was counterbalanced across both sessions. An LED and an LED driver (Thorlabs) were used to deliver yellow light (593-nm constant light, 20 s on, 5 s off) in a manner similar to previous reports (26, 43). Light delivery began as soon as the mouse was secured in restraint and returned to the home cage and continued throughout the test session. Grooming behavior was also assessed in GAD2-Cre mice expressing either ChR2 or eYFP during delivery of blue light into the IPN [473 nm, 20 Hz, 25% duty cycle; (38)]. For behavioral experiments, a light output at the end of the fiber tip was between 2 and 5 mW. Grooming behavior was tallied offline by an independent observer in a blinded manner. Elevated plus maze Before EPM exploration, mice were placed in restraint for 5 min, and yellow light was delivered to the IPN as above. After restraint, mice were placed in the EPM apparatus that consisted of a central junction (5 cm by 5 cm) and four arms elevated 45 cm above the floor with each arm positioned at 90° relative to the adjacent arms. The two closed arms encompassed high walls (30 cm by 5 cm by 15 cm), and the open arms were exposed (30 cm by 5 cm by 0.25 cm). No-stress control mice were placed in the EPM without restraint. Each mouse was positioned at the center of the maze with their head facing one of the open arms and was given 5 min to freely explore under light delivery. Total entries into the open and closed arms and time spent in the open arms were measured by MED-PC IV software (MED Associates Inc.). The time spent in the open arms was used as an index of anxiety-like behavior. The total number of entries to open and closed arms was considered as the index of locomotor activity. The apparatus was cleaned thoroughly between trials. Open-field test Following the EPM test, no-stress and restraint-stressed mice were then placed facing a wall of an open-field chamber (42 cm by 38 cm by 30 cm). Each mouse was given 10 min to explore, and the time spent in the center and outer parts of the chamber was tracked from a video recording using EthoVision XT 11.5 (Noldus Apparatus). Light delivery continued throughout the entire session using the same parameters described above. Operant chamber conditioning Before operant training, mice were pretrained to self-administer sucrose solution in their home cage for 2 weeks (10 sessions) using a drinking-in-the-dark paradigm (85). Briefly, mice were given access to a bottle containing 5% (v/v) sucrose solution for 2 hours, 3 hours into the dark cycle from Monday to Friday. Solution was presented in 50-ml tubes fitted with rubber stoppers and a stainless-steel sipper tube containing a ball bearing. Before the first operant session, mice were also restricted to ~90 to 95% of their free feeding weight and were given access to sucrose pellets in their home cage for a minimum of 2 days. Mice were then trained in operant chambers (17.8 cm by 15.2 cm by 18.4 cm; Med Associates) that contained a house light, pellet receptacle, and two nose-poke ports. Mice learned to nose-poke an active portal to receive a sucrose pellet (20 mg; Test Diet). Each session began with illumination of the house light and training consisted of 1 session per day for a minimum of six consecutive days, whereby the number of nose pokes required to earn a reward started at a fixed ratio of 1:1 (FRt1) and increased up to FRt5 (86) until inclusion criteria was met, which included higher than 75% discrimination of the active portal over the inactive portal and greater than 100 active nose pokes over the last 3 days of training (87). Mice were then tested using a progressive ratio 3 (PR3) schedule that determined breakpoint and was used as a measure of motivation. Following delivery of the first reinforcer, mice were given 15 min to complete each ratio, and breakpoint was defined as the last schedule completed within 15 min (88). All session parameters and responses were controlled and recorded with Med-PC IV software (Med Associates). In GAD2-Cre or SST-Cre mice expressing either NpHR or eYFP in the IPN, sucrose breakpoint was determined without stress. Mice then underwent two additional PR3 sessions where they were restrained for 5 min and then released directly into the operant chamber. Light was delivered during and after stress using the same parameters as above and was counterbalanced across PR sessions. Separate groups of GAD2-Cre mice expressing either NpHR or ChR2 in the IPN underwent the same training protocol as above and completed 2× PR3 sessions without stress in the absence or presence of yellow light delivered for NpHR-expressing mice or blue light for ChR2-expressing mice as described above. Light delivery was counterbalanced across the sessions. Control eYFP-expressing mice were also tested in the absence and presence of blue light under the same parameters. A group of ChAT-Cre mice expressing either ChR2 or eYFP in MHb-IPN axon terminals underwent PR3 sessions in the absence or presence of stress/light delivery, which was counterbalanced under each condition. PR sessions were conducted a day apart or 7 days apart for sessions following stress exposure. Mice were maintained on FRt5 schedule between PR sessions. CORT assay A commercially available enzyme-linked immunosorbent assay plasma CORT test kit (Arbor Assays, Ann Arbor, MI) was used to quantify CORT from blood plasma according to the manufacturer’s protocol. Plasma CORT was quantified in C57Bl/6J mice that were exposed to 5 min of restraint stress and in control mice that were handled and returned to the home cage. Mice were euthanized 10 min later, and trunk blood was collected. Plasma CORT was also determined in food restricted and age-matched mice without food restriction. Immunostaining and microscopy Mice were euthanized and transcardially perfused with ice-cold 0.1 M phosphate-buffered saline (PBS; pH 7.4) followed by 10 ml of cold 4% (w/v) PFA in 0.1 M PBS. Brains were postfixed in 4% PFA for 2 hours and then submerged in 30% sucrose solution overnight. Brains were sliced to coronal sections (25 μm) using a freezing microtome (HM430, Thermo Fisher Scientific). To assess virus expression and fiber locations, sections were washed in 0.1 M PBS, mounted, air-dried, and coverslipped with Vectashield mounting medium (Vector Laboratories). Slices were imaged using a fluorescence microscope (Zeiss, Carl Zeiss MicroImaging Inc.). Immunostaining for c-fos expression was done in three of four mice per group. For restraint stress, C57Bl/6J mice received 5 min of restraint in the home cage, and the controls were handled and immediately put back in home cage. For light stimulation, ChAT-Cre mice with ChR2 injected in the MHb and optic fibers implanted in the IPN were split into two groups. One group received light (473 nm, 20 Hz, 25% duty cycle, 5 min on, 5 min off) over a 30-min period, while the other group were handled and used as a no-light control. Brain sections from each group were permeabilized with 0.2% Triton X-100 in 0.1 M PBS for 5 min, blocked with 2% bovine serum albumin (BSA) in 0.1 M PBS for 30 min, and then incubated overnight in primary antibody (rabbit anti–c-fos, 1:800; Santa Cruz Biotechnology, sc-52) in 2% BSA at 4°C. Slices were subsequently washed in 0.1 M PBS, blocked with 2% donkey serum (Sigma-Aldrich) for 30 min, and then incubated in secondary antibody for 1 hour (donkey anti-rabbit 594, 1:800; Life Technologies, A21207). Slices were washed in 0.1 M PBS, and sections were mounted, air-dried, and coverslipped. Three parallel sets of IPN slices (25 μm) were collected from bregma −3.28 to −3.88. Images were obtained on a fluorescence microscope (Zeiss, Carl Zeiss MicroImaging Inc.). The number of c-fos immunoreactive neurons in the IPN was counted from one set of the slices (six to seven slices per mouse) using Fiji (ImageJ) and is presented as the average number/mm2 of IPN tissue. Quantification of SST and GCaMP colocalization was performed in Fiji (ImageJ). Single-cell RNA sequencing Tissue preparation For scRNA-seq, one group of C57Bl/6J mice received 5 min of restraint in the home cage, and the control group was handled and immediately put back in the home cage. Mice were euthanized 30 min later, and the brains were swiftly removed and placed in ice-cold N-methyl-d-glucamine–artificial cerebrospinal fluid (NMDG-aCSF) that contained 92 mM NMDG, 2.5 mM KCl, 1.2 mM NaH2PO4, 30 mM NaHCO3, 20 mM Hepes, 25 mM glucose, 5 mM sodium ascorbate, 2 mM thiourea, 3 mM sodium pyruvate, 10 mM MgSO4·7H2O, 0.5 mM CaCl2·2H2O (osmolarity, ∼300 mosmol). The NMDG-aCSF solution was continuously bubbled with 95% O2–5% CO2 to maintain pH at 7.4. Actinomycin D (45 μM; Sigma-Aldrich, cat. no. A1410) was added during slicing and subsequent tissue preparation steps to minimize artificially induced activation of early immediate genes (89). IPN tissue punches were obtain from 300-μm sections prepared on a vibratome (Leica VT 1200 S, Leica Biosystems). Tissue dissociation IPN tissue punches from five brains were pooled and transferred to Dulbecco’s PBS (DPBS), supplemented with d-glucose (1 g/liter), sodium pyruvate (36 mg/liter; Gibco), and actinomycin D (Sigma-Aldrich). A neural tissue dissociation kit (Miltenyi Biotec) was used to generate a single-cell suspension following the manufacturer’s protocol for manual dissociation. Brain punches were rinsed in DPBS with calcium and magnesium (Sigma-Aldrich), and tissue was transferred into 1960 μl of enzyme mix 1 supplemented with actinomycin D. The tissue was incubated for 15 min at 37°C under slow rotation after which 30 μl of enzyme mix 2 was added. The tissue was dissociated with a wide bore pipette by gentle mixing and then incubated for another 10 min at 37°C under slow rotation after which 15 μl of enzyme mix 2 was added. The tissue was dissociated by pipetting slowly until no tissue pieces were observed. The cell suspension was filtered through a 70-μm magnetic-activated cell sorting (MACS) SmartStrainer (Miltenyi Biotec) and topped up with 10 ml of DPBS, supplemented with 0.5% BSA (Sigma-Aldrich) and actinomycin D (DPBS-BSA-aD). The cell suspension was centrifuged at 300g for 10 min at 4°C, and the supernatant was removed and discarded. The cell pellet was resuspended in 1550 μl of DPBS-BSA-aD, and 450 μl of debris removal solution was added to the mixture and overlaid with 2 ml of DPBS. The cell mixture was centrifuged at 3000g and 4°C for 10 min. The top two phases were carefully aspirated, and the bottom layer containing the cells was topped up with DPBS-BSA-aD. The cell suspension was centrifuged at 1000g and 4°C for 10 min, and the supernatant was discarded. The cell pellet was resuspended in 0.5 ml of 1× Red Blood Cell Removal Solution and incubated for 10 min at 4°C after which 5 ml of DPBS-BSA-aD was added and the cell suspension was centrifuged at 300g and 4°C for 10 min. The supernatant was discarded, and the pellet was resuspended in DPBS-BSA-aD and passed through a 30-μm MACS SmartStrainer cell strainer. The cells were counted in a Countess II FL (Thermo Fisher Scientific), and, on achieving a cell viability of at least 70%, the samples were prepared for Gel Beads-in-Emulsion (GEMs; 10x Genomics) generation and barcoding. Preparation of 10x Genomic library for scRNA-seq The single-cell suspension was used to generate GEMs using the Chromium Next GEM Single Cell 3′ v3.1 kit (10x Genomics) following the manufacturer’s protocol. Briefly, the barcoded Single Cell 3′ v3.1 gel beads were combined with the cell suspension in master mix and partitioning oil, and the GEMs were generated by the Chromium Controller. The GEMs were transferred from the chip into a tube strip and then reverse-transcribed in a thermal cycler to generate cDNA. Post–GEM-Reverse Transcription cleanup was performed, and the cDNA was amplified and then cleaned. After cDNA cleanup, 1 μl of the sample was run on a Bioanalyzer High Sensitivity chip (Agilent) for quality control and quantification. The cDNA was used to generate 3′ gene expression libraries using the Chromium Next GEM Single Cell 3′ Library kit v3.1, which were then shipped for sequencing. scRNA-seq data analysis Clustering, differential gene expression analysis, and integrative analysis of control and stress conditions were performed using the Seurat V4 package (90). Low-quality cells (genes < 200 or genes > 7000, or percentage of mitochondrial genes > 20%) were removed before downstream analysis. A total of 3863 cells across control and stress conditions were then used for integrative analysis as described in the Seurat V3 package (91). Briefly, gene counts were scaled by the cellular sequencing depth (total unique molecular identifiers) with a constant scale factor (10,000) and then transformed by natural log (log1p). A total of 2000 highly variable genes were selected in each sample based on a variance stabilizing transformation. Anchors between individual data were identified, and correction vectors were calculated to generate an integrated expression matrix, which was used for subsequent clustering. Integrated expression matrices were scaled and centered followed by principal components analysis (PCA) for dimensional reduction. PC1 to PC20 were used to construct nearest neighbor graphs in the PCA space followed by Louvain clustering to identify clusters (resolution of 0.5). For visualization of clusters, UMAP was generated using the same PC1 to PC20. Markers for each cluster were then identified by comparing gene expression values within an integrated cluster against all other cells using a Wilcoxon rank sum test. Genes with log fold change > 0.25 and adjusted P value < 0.05 were considered significantly enriched. Significantly enriched genes from each cluster were then compared against canonical markers of cell types in the mammalian central nervous system for subsequent cluster annotation. Expression of Thy1, Stmn2, and Map2 identified three neuronal clusters that were then subset from the dataset for reclustering using the integrative analysis described above. Reclustering resulted in six neuronal clusters, and c-fos gene counts from each cluster across control and stress conditions were then extracted from the Seurat object and scaled by cellular sequencing depth and log1p as described above. Cell counts for c-fos were then passed through a Grubbs’ test (α = 0.05) to test for outlier cells. Grubbs’ test removed two outlier cells based on c-fos expression (one from cluster 3 and one from cluster 4, both in the stress condition). Clusters were then analyzed for significant enrichment of c-fos (fold change > 0.25 and adjusted P value < 0.05) in the stress condition compared to control cells within the same cluster by a Wilcoxon rank sum test to identify neuronal populations activated by restraint stress. A single cluster demonstrated a significant enrichment of c-fos in the stress condition, which was then analyzed for differentially expressed genes against all other neuronal clusters. Fiber photometry analysis GCaMP recordings were analyzed in a manner identical to previous reports (43). Briefly, demodulated fluorescence signals were low-pass–filtered (3 Hz) using the photometry analysis module of the Doric Neuroscience Studio software, MATLAB scripts were used to scale the 405-nm channel to the 465-nm channel by applying a least mean squares linear regression, and scaled signals were used to calculate the dF/F0, where dF/F0 = (465-nm signal − fitted 405-nm signal)/fitted 405-nm signal. The AUC of the dF/F0 curves was calculated in GraphPad Prism using the trapezoidal rule. Maximum dF/F0 was scored from each recording. Mean peak amplitude was calculated in MATLAB. Quantitative measures were determined from each 5-min recording under no-stress and stress conditions. A per–5-min average was calculated from quantification of the 10-min post-stress recording. Z-scores were calculated using the average dF/F0 from 5 s of recording before the event being analyzed. Statistical methods Means and SEM were calculated for each dataset. Where indicated, unpaired t tests, paired two-tailed t tests, repeated-measures one-way ANOVAs with Tukey’s posttests, or two-way ANOVAs, or repeated measures two-way ANOVAs, with Bonferroni’s posttests were conducted for all analyses involving the comparison of group means. Grubbs’ test was applied to check for outliers, and one mouse was removed from the eYFP group in the EPM assay. Z-scores are presented as means ± SEM for all bouts of grooming, sucrose consumption, foot shock, or immobility. Comparisons of z-scores were made using the calculated average for each animal. All other data in the results and figures are presented as means ± SEM. All analyses were performed using Prism 8 (GraphPad, San Diego, CA) or MATLAB with custom scripts (The MathWorks Inc., Natick, MA). Statistical significance was accepted at P < 0.05.