Orexin prevents depressive-like behavior by promoting stress resilience

Animals

Adult male Sprague-Dawley rats (8 weeks, 200–250 g) were housed under standard animal housing conditions except for a specific description. All experimental procedures were approved by the Experimental Animal Care and Use Committee of Nanjing University.

Retrograde tracings

The retrograde tracer Fluoro-Gold (FG) was micro-electrophoresed in VP (A 0.0 to −0.3, L 2.5, and H 7.6) following our previous reports [23, 24]. The immunohistochemical experiment was performed 2 weeks later to determine the injection site and location of retrogradely labeled cells.

Immunohistochemistry

Sections with FG were immunolabelled with primary antibodies against orexin-A (1:200; R&D, Minneapolis, MN; Cat# MAB763, RRID:AB_2117627) and FG (1:2000; Millipore, Boston, MA; Cat# AB153, RRID:AB_90738) as we described previously [23, 24]. Primary antibodies to OX1R (1:100; Abcam; Cat# ab68718, RRID: AB_1269637) and OX2R (1:100; Everest, Waltham, MA; Cat# EB08124, RRID:AB_2117786) were used for detecting the expression of two orexin receptors in VP. For immunohistochemical identification of the GABAergic VP neurons recorded in the voltage and current clamp, brain slices containing the recorded neurons filled with biocytin were fixed, dehydrated, resectioned, and then incubated with primary antibodies against GABA (1:1000; Sigma, St. Louis, MO; Cat# A0310, RRID: AB_476667).

Stereotactic microinjection and lentiviruses

Rats were submitted to stereotaxic surgery and bilateral implantation of guide cannula 2 mm above VPs (A 0.0 to −0.3, L 2.5, and H 7.6) for microinjection of saline, orexin-A (one of the endogenous orexin peptides; 1 μM, 0.5 μl/lateral), or TCS1102 (a potent dual orexin receptor antagonist; 20 nM, 0.5 μl/lateral). shRNA lentiviruses targeting OX1R and OX2R mRNAs (LV-shOX1R-eGFP and LV-shOX2R-eGFP) were produced by the lentiviral vector GV248 (pFU-GW-007-hU6-Ubiquitin-EGFP-IRES-puromycin; GeneChem, Shanghai, China) and stereotaxically microinjected (1 × 10⁹ TU/ml, 1 μl/lateral) into the bilateral VPs for knockdown of orexin receptors. The non-targeting shRNA lentiviral vector (LV-CON-eGFP) was used as control. After 2 weeks, the orexin receptor mRNA and protein levels in the VP were assessed by quantitative real-time RT-PCR and western blot [24–26].

Behavioral assessments

Depressive-like behaviors were assessed by forced swim test (FST), sucrose preference test (SPT), and social interaction test (SIT) using Clever TopScan [27]. Social interactions between novel partners were measured as social proximity. The tested rat was placed in the center of a plastic open field apparatus (50 × 50 cm) simultaneously with the other novel partner, which was put in an iron cage (16 × 22 cm) in one corner of the arena. Time of head approaching and staying close to the cage (≤5 cm) and the times of forepaws climbing the cage in 10 min were recorded and calculated to reflect social proximity. Interactions of the subordinate with the dominator ranked by a tube test for social dominance [28, 29] were measured as social avoidance under an acute psychosocial stress. Open field test was applied to examine general locomotor activity.

Whole-cell patch clamp recordings

Whole-cell patch clamp recordings in brain slices were applied to assess the effect of orexin on VP neurons and the underlying receptor and ionic mechanisms as previously reported [23, 25, 26, 30]. Briefly, recordings of whole-cell currents were low-pass filtered at 2 kHz and digitized at 10 kHz and recordings of membrane potentials were low-pass filtered at 5 kHz and digitized at 20 kHz.

Enzyme-linked immunosorbent assay (ELISA)

Serum corticosterone level and orexin concentration in VP were quantified by ELISA (corticosterone kit and orexin-A kit), following the manufacturer’s instructions. After averaging the results of duplicate wells, corticosterone value of each sample was calculated as ng/ml serum and orexin value as pg/mg wet tissue.

Data analysis

All data were analyzed with SPSS 17.0 and presented as median (horizontal bar) with 25th–75th (box) and 10th–90th (whiskers) percentiles. Two-tailed unpaired and paired t test, one-way, two-way, and repeated measures two-way analysis of variance (ANOVA), and post hoc Bonferroni-corrected t test was employed for statistical analysis. P-values of <0.05 were considered to be significant. The exact sample size, the statistical test used, and the results of the tests for each experiment were included in results and figure legends.

Pharmacological blockade or knockdown of orexin receptors in VP induces depression-like behaviors

Two orexin receptor subtypes, OX1R and OX2R [31, 32], are known to mediate orexin functions. In the present study, we first identify the expression of orexin receptors in VP in rats. Double immunostaining results showed that both OX1R and OX2R were co-localized in VP neurons (Fig. 1a1–a4).

Next, we blocked orexin receptors in VP to determine the role of endogenous orexinergic system in the depressive-like behaviors in FST and SPT. Both of these tests are classical depressive paradigms related to two core symptom domains, despair and anhedonia, in depression, respectively [33]. Pharmacological blockade of both OX1R and OX2R by microinjection of TCS1102, a potent dual orexin receptor antagonist, into bilateral VPs, led to a significant increase in immobility time in the FST (P < 0.05; Fig. 1b) and a remarkable decline in sucrose intake in the SPT (P < 0.05; Fig. 1c), but did not influence the total distance traveled (P = 0.612; Fig. 1d) and number of rearing in the open field test (P = 0.261; Fig. 1d).

To further assess the contribution of each subtype of orexin receptors to the orexin-induced reduction of depressive-like behaviors, we generated lentiviral vectors carrying shRNA targeting OX1R (LV-shOX1R-eGFP) or OX2R mRNAs (LV-shOX2R-eGFP) to separately knockdown the expression of OX1R and OX2R in VP. The extent of lentivirus infection in VP was identified by detecting a fluorescent tag eGFP coexpressed with shRNA (Fig. 1e), and the effectiveness of knockdown was examined by qPCR (Supplementary Figure 1a) and western blot (Fig. 1f and Supplementary Figure 1b). OX1R/OX2R shRNA lentivirus significantly decreased the level of OX1R and OX2R mRNAs to 42.04 ± 0.02% and 65.36 ± 0.01% (P < 0.001, respectively; Supplementary Figure 1a), and accordingly reduced the level of OX1R and OX2R proteins to 46.20 ± 0.04% and 32.88 ± 0.04% (P < 0.01 and P < 0.001; Supplementary Figure 1b), respectively. In line with the above pharmacological results, we found that either OX1R or OX2R knockdown in VP significantly prolonged the immobility time of rats in the FST (P < 0.05, respectively; Fig. 1g), and maximally suppressed the sucrose preference at 3 (P < 0.05, respectively) and 2 weeks (P < 0.05, respectively) after lentivirus administration, followed by a low plateau (Fig. 1h). However, knockdown of OX1R or OX2R had no effect on general locomotor activity in the open field (P = 0.340; Fig. 1i). These behavioral results from pharmacological and molecular manipulations collectively suggest that blockage of endogenous orexinergic inputs in VP results in depressive-like behaviors, and OX1R and OX2R may have the homogeneous contribution to the protective effect of orexin on depression.

Orexin depolarizes GABAergic VP neurons and prevents depressive-like behaviors

To investigate whether orexinergic neurons directly project to VP, we micro-electrophoresed FG into VP (Fig. 2a), and found that a number of orexin-immunopositive neurons in the perifornical/lateral hypothalamic area (PFA/LHA), the origin of the central orexinergic system, were retrogradely labeled with FG (Fig. 2b). Moreover, the orexin-immunoreactive fibers with bead-like varicosities were scattered throughout the VP (Fig. 2c, d). Although the density of orexinergic fibers was slightly higher in the posterior VP, there is no statistically significant difference between anterior and posterior VP (P = 0.145; Fig. 2d). These results indicate a direct projection of PFA/LHA orexinergic neurons to VP and a possible modulation of orexin released by the orexinergic terminals on VP neuronal activities.

Since GABAergic neurons are the majority and principle neurons in VP [34, 35], we used whole-cell patch clamp recordings on brain slices to assess the effect of orexin on VP GABAergic neurons. We filled all recorded neurons with biocytin after recordings and then immunostained them with GABA. The recorded GABAergic VP neurons co-labeled by biocytin and GABA in this study had small sized (somata diameters around 15 μm, Fig. 2e1–e4), and characteristically exhibited a prominent h-current, a slowly hyperpolarization activated inwardly rectifying current (Fig. 2f). These morphological and electrophysiological properties are consistent with previous reports [35, 36]. We found that orexin-A evoked a strong depolarization on GABAergic VP neurons (10/11, 90.9%), which were silent at rest, and sufficiently brought up the neurons firing (Fig. 2g, left panel). In addition, orexin induced a depolarization in the presence of TTX (Fig. 2g, right panel) and there is no significant difference between the amplitude of the orexin-induced depolarization on GABAergic neurons of anterior and posterior VP (P = 0.668; Fig. 2h), suggesting a direct postsynaptic excitatory effect of orexin on GABAergic VP neurons. Consistently, double immunostaining results showed that OX1R is expressed on the GABA-positive neurons in VP (Fig. 2i1–i3).

Next, we determined the receptor and ionic mechanisms underlying the orexin-induced excitation on GABAergic VP neurons. Selective OX1R antagonist SB334867 attenuated the orexin-induced inward current (P < 0.01; Supplementary Figure 2), which was mimicked by selective OX2R agonist [Ala¹¹, D-Leu¹⁵]-orexin B (Supplementary Figure 2) and nearly totally blocked by the potent dual orexin receptor antagonist TCS1102 (P < 0.01; Supplementary Figure 2). Moreover, OX1R or OX2R knockdown markedly lessened the orexin-induced excitation on GABAergic VP neurons (P < 0.01, respectively; Fig. 3a1–b2). These results suggest that both OX1R and OX2R are involved in the excitatory effect of orexin on VP neurons. Furthermore, KB-R7943, a selective blocker for Na⁺–Ca²⁺ exchanger (NCX), or CdCl₂, a broad-spectrum blocker for voltage-gated Ca²⁺ channels, partly blocked the orexin-induced inward currents (P < 0.01, respectively; Fig. 3c, d and Supplementary Figure 3a). Combined application of KB-R7943 and CdCl₂ nearly totally blocked the orexin-induced excitation (P < 0.001; Fig. 3c, d, and Supplementary Figure 3a), implying a dual downstream ionic mechanism involving both NCX and voltage-gated Ca²⁺ channels. Besides, orexin-A significantly enhanced the high-voltage-activated (HVA) (P < 0.01; Supplementary Figure 3b) rather than low-voltage-activated (LVA) calcium currents (P = 0.458; Supplementary Figure 3b), indicating an involvement of HVA but not LVA Ca²⁺ channels. Since HVA Ca²⁺ channels include L-, P/Q-, and N-subtypes, we further examined the HVA Ca²⁺ channel subtypes responsible for the excitation of orexin on VP neurons. As shown in Fig. 3e, f and Supplementary Figure 4b and c, the orexin-induced inward current was partially suppressed by nifedipine (selective blocker for L-type Ca²⁺ channel, P < 0.01) rather than ω-CTX GVIA (selective blocker for N-type Ca²⁺ channel, P = 0.483) or ω-CTX MVIIC (blocker for N- and P/Q- type Ca²⁺ channel, P = 0.687), suggesting that L-, rather than N- and P/Q-subtype of HVA Ca²⁺ channel contributes to the orexin-induced excitation. Furthermore, combined application of nifedipine and KB-R7943 totally blocked the excitation of orexin-A (P < 0.01; Fig. 3f and Supplementary Figure 4a). All these results strongly suggest that NCX together with L-type Ca²⁺ channel may be coupled to orexin receptors and mediate the excitatory effect of orexin on GABAergic VP neurons.

Given that orexin directly excites VP neurons, we microinjected orexin-A into the bilateral VPs to investigate the effect of orexin in VP on depressive-like behaviors. As shown in Fig. 3g, h, orexin significantly reduced the immobility time of rats in the FST (P < 0.05; Fig. 3g) and markedly increased the preference of rats for sucrose solution (P < 0.05; Fig. 3h). In addition, orexin did not influence the distance traveled (P = 0.619; Fig. 3i) and rearing numbers in the open field (P = 0.134; Fig. 3i). Thus, we speculate that the orexinergic afferent inputs may directly excite GABAergic VP neurons and subsequently prevent depressive-like behaviors.

Endogenous orexin in VP does not affect novel social interaction

Social avoidance is one of the crucial symptoms of depression. Therefore, we employed a social proximity test to assess the effect of orexin on social interaction between novel partners. As shown in Fig. 4, microinjection of orexin-A into bilateral VPs significantly increased the time of rats spent in the area close to a caged novel rat (P < 0.01; Fig. 4a, b) and the times of climbing the cage (P < 0.01; Fig. 4c). Nevertheless, neither the time spent close to the cotton (P = 0.084; Fig. 4a, b) nor the times of climbing the cage loaded with cotton (P = 0.280; Fig. 4c) was changed by orexin. These results, together with the above-mentioned result of open field test (Fig. 3i), indicate that exogenously injected orexin in VP may enhance the willingness of rats to interact with others, without change in general exploratory and locomotor activity.

To further determine the role of endogenous orexinergic inputs in VP in social interaction, we knocked down OX1R in VP by shRNA lentivirus. Unexpectedly, blockage of endogenous orexinergic inputs in VP did not change either the time rats spent in the area close to the caged novel rats (P = 0.325; Fig. 4d, e) or the times of climbing (P = 0.806; Fig. 4f). It suggests that endogenous orexin in VP may actually have no effect on novel social proximity, although exogenous orexin significantly promotes it.

Endogenous orexin in VP alleviates social avoidance under acute psychosocial stress triggered by social rank

The consistent effects of endogenous and exogenous orexin on forced swimming and sucrose preference, with the paradoxical effects of them on novel social interaction, impel us to reconsider the exact pathophysiological function of orexinergic modulation on VP in depression. Considering both FST and SPT are under acute stress (forced swimming and food/water deprivation) whereas novel SIT involves no apparent stress, we speculated that endogenous orexin in VP may be more critical for alleviation of depressive behaviors under stress than non-stress conditions. Given that social subordinance causes psychosocial stress and influences mental health [37, 38], we designed a novel paradigm by combination of SIT with the tube test (Fig. 5a), which is usually applied to assess social hierarchy [29, 39]. Pairs of dominator and subordinate rats competed from the tube test were used in the following SIT (Fig. 5a), during which the dominator exerted an acute psychosocial stress to the subordinate. Lentivirus was injected into VP 14 days before the tube test. Performances between OX1R knockdown rats, normal rats (LV-CON-eGFP), and OX1R knockdown rats vs normal rats in the tube test have no significant difference (P = 0.768; Supplementary Figure 5), indicating orexin in VP is not involved in the regulation of social hierarchy. Notably, we found that blockage of endogenous orexinergic inputs in VP by knockdown of OX1R significantly suppressed the social proximity of subordinates to caged dominators (P < 0.01; Fig. 5b–d), but had no influence on the social interaction between dominators and caged subordinates (time: P = 0.843; times of climbing: P = 0.344; Fig. 5b–d). The result implicates that endogenous orexin in VP may prevent social avoidance under psychosocial stress but not non-stress conditions.

Intriguingly, in normal rats (LV-CON-eGFP), no significant difference was found between social interaction of subordinates with caged dominators and that of dominators with caged subordinates (time: P = 0.073; times of climbing: P = 0.910, Fig. 5b–d). However, in OX1R knockdown rats (LV-shOX1R-eGFP), willingness of subordinates to interact with caged dominators was significantly lower than that of dominators to interact with caged subordinates (P < 0.01; Fig. 5b–d). These results indicate that psychosocial stress may induce social avoidance more easily in rats with knockdown of OX1R in VP. Therefore, we propose that the endogenous orexinergic system is likely to help to cope with acute stressful social experiences, while dysfunction of orexinergic inputs in VP may result in susceptibility to psychosocial stress and social avoidance in depression.

To confirm this hypothesis, we assessed serum corticosterone level, a marker for stress, and orexin concentration in VP, in rats facing the above different behavioral tasks. We found that serum corticosterone level was elevated with a significant increase in orexin level in VP 15 min after finishing exposure to acute stresses, including forced swimming, food/water deprivation and social rank stress (P < 0.05; Fig. 5e). However, both serum corticosterone level and orexin level in VP did not change in rats in non-stress situations, such as home cage and social interaction without psychosocial stress (orexin: P = 0.941; corticosterone: P = 0.519; Fig. 5e). Considering the stress-dependent elevation of orexin level in VP and the orexin-produced preventing effect in depressive paradigms under stress, we suggest that the direct orexinergic modulation on GABAergic VP neurons may prevent depressive-like behaviors especially by improving the capacity for resilience to stress.

Conflict of interest

The authors declare that they have no conflict of interest.