Down-regulation of habenular calcium-dependent secretion activator 2 induces despair-like behavior

Calcium-dependent secretion activator 2 (CAPS2) regulates the trafficking and exocytosis of neuropeptide-containing dense-core vesicles (DCVs). CAPS2 is prominently expressed in the medial habenula (MHb), which is related to depressive behavior; however, how MHb neurons cause depressive symptoms and the role of CAPS2 remains unclear. We hypothesized that dysfunction of MHb CAPS neurons might cause defects in neuropeptide secretion and the activity of monoaminergic centers, resulting in depressive-like behaviors. In this study, we examined (1) CAPS2 expression in the habenula of depression animal models and major depressive disorder patients and (2) the effects of down-regulation of MHb CAPS2 on the animal behaviors, synaptic transmission in the interpeduncular nucleus (IPN), and neuronal activity of monoamine centers. Habenular CAPS2 expression was decreased in the rat chronic restraint stress model, mouse learned helplessness model, and showed tendency to decrease in depression patients who died by suicide. Knockdown of CAPS2 in the mouse habenula evoked despair-like behavior and a reduction of the release of DCVs in the IPN. Neuronal activity of IPN and monoaminergic centers was also reduced. These results implicate MHb CAPS2 as playing a pivotal role in depressive behavior through the regulation of neuropeptide secretion of the MHb-IPN pathway and the activity of monoaminergic centers.


Results
Down-regulation of CAPS2 mRNA seen in the habenula of the animal models of depression. We previously measured the mRNA levels of the target genes in the habenula of chronic restraint stress (CRS) animal model of depression and postmortem brain samples of MDD patients 24 . We found that CAPS2 mRNA expression was significantly reduced in the habenula of CRS rats compared with those of non-stressed (NS) rats (n = 4 for each group, Mann-Whitney U-test; NS vs. CRS expressed as fold change: 1.00 ± 0.14 vs. 0.61 ± 0.09, P = 0.029; Fig. 1a) and CAPS2 protein levels in the IPN, a direct downstream of the MHb, also showed a tendency to decrease, being 53.8% of control, as shown via Western blotting ( Supplementary Figure S1). CAPS2 mRNA level was also decreased in mice exposed to inescapable electric foot shocks inducing learned helplessness (LH) as compared with control mice (n = 6 for each group, Student's t-test; CON vs. LH expressed as fold change: 1.00 ± 0.46 vs. 0.52 ± 0.38, T (14) = 2.285, P = 0.038; Fig. 1b). In addition, CAPS2 mRNA levels in postmortem habenula samples of depressive patients tended to decrease, being 73% of nondepressed controls (CON, n = 11; MDD, n = 12; Student's t-test; fold change for CON vs. MDD suicides: for CAPS2, 1.00 ± 0.73 vs. 0.73 ± 0.42, T (21) = 1.111, P = 0.279; for CAMK2B, 1.00 ± 0.38 vs. 1.11 ± 0.38, T (21) = -0.714, P = 0.483; Fig. 1c and d).
MHb CAPS2 knockdown evokes despair-like behavior. We next investigated whether the selective suppression of the MHb CAPS2 expression induced depression-related behaviors. We generated viral vectors to express small interfering RNA (siRNA) targeting the CAPS2 transcript. After testing three CAPS2 knockdown viral vectors in vitro, we selected a sh-CAPS2 that reduced endogenous CAPS2 expression by 50% (Supplementary Table S1). We then injected either adeno-associated virus 2/9 (AAV2/9) containing sh-CAPS2 (KD) or control with empty target sequence (CON) into the mice MHb by stereotaxic injection and checked for injection site (Fig. 2b-d). After 2 weeks of injection, behavior tests were done and the expression levels of CAPS2 in the region of interest were assessed (Fig. 2a). Compared with control, the siRNA specifically targeting the CAPS2 transcript effectively reduced CAPS2 protein and mRNA levels (Supplementary Figure S2).
We then evaluated the effects of a MHb-specific CAPS2 knockdown on depression-and anxiety-like behavioral phenotypes. Interestingly, the MHb CAPS2 knockdown resulted in an increased immobility time in both the tail-suspension test (TST) and FST (CON, n = 26; KD, n = 29; Student's t-test; CON Fig. 2e and f). However, there was no effect in daily fluid consumption normalized to body weight (CON, n = 18; KD, n = 20; Student's t-test; water plus sucrose solution (g/day) /body weight(g): 0.20 ± 0.028 vs. 0.20 ± 0.018, T (36) = -0.475, P = 0.637; Fig. 2g) and sucrose preference (CON, n = 18; KD, n = 20; Student's t-test; percentage of the sucrose solution consumed in total fluid intake: 0.83 ± 0.08 vs. 0.83 ± 0.17, T (36) = 0.137, P = 0.892; Fig. 2h) in SPT, showing no effect on anhedonia-like behavior. Reduced expression of CAPS2 in the MHb did not alter the total distance moved in the open field test (OFT) and elevated zero maze (EZM), implicating no changes in locomotion (Supplementary Figure S3) Figure S3). To further examine the stress-induced anxiety levels, the novelty-suppressed feeding test was performed (Supplementary Figure S4) www.nature.com/scientificreports/ pellets and the amount of food consumed after the test did not differ between the groups. As CAPS2 KO mice have manifested autism-like behavior 8 , we next examined the social interaction in the MHb CAPS2 knockdown mice. There was also no significant difference in the social interaction ratio between control and MHb CAPS2 knockdown mice (Supplementary Figure S4). Together, these data suggest that selective reduction of MHb CAPS2 protein induces despair-like behavior, but not anxiety, anhedonia, and social dysfunction.
MHb CAPS2 knockdown leads to accumulation of presynaptic dense-core vesicles in the IPN. It is known that the major output of the MHb connects to the IPN through fasciculus retroflexus 15,26 .
Since CAPS2 is well-known in regulating DCV exocytosis, we examined the IPN to observe the ultrastructural effect of MHb CAPS2 knockdown using transmission electron microscopy. Areas of presynaptic bouton and length of postsynaptic density did not differ between the two groups (n = 4 for each group; Mann-Whitney U-test; area of presynaptic bouton: 0.91 ± 0.25 μm 2 vs. 1.07 ± 0.13 μm 2 , P = 0.486; postsynaptic density length: 0.60 ± 0.07 μm vs. 0.62 ± 0.08 μm, P = 1.000; Fig. 3a Fig. 3a, d, and e). The number of synaptic vesicles (SV) and density of SV showed a slight tendency to increase, but without a significant difference between the two groups (n = 4 for each (a-d) qRT-PCR analysis showed that the CAPS2 mRNA expression level decreased in the habenula of rats exposed to CRS (a), mice exposed to electric shock (b). CAPS2 mRNA expression showed a tendency to decrease in the habenula of MDD patients who died by suicide (c and d). CAMK2B levels did not show any difference between the two groups. The relative abundances of mRNA were normalized to the amount of GAPDH using the comparative threshold cycle method.   Fig. 3a, f, and g). This implied that the CAPS2 reduction impairs DCV exocytosis and leads to accumulation of DCVs in the MHb terminals in the IPN.

Discussion
In this study, we demonstrated that a down-regulation of CAPS2 protein in the MHb leads to an increase in the despair-like symptom, one of the core symptoms in the diagnosis of major depressive disorder 35 . CAPS2 expression level is decreased in the CRS rat model, LH mice model of depression, and tended to decline in human MDD, implicating a link for CAPS2 levels in depression. The IPN in MHb CAPS2 knockdown mice manifested an increase in DCV in the presynaptic bouton and decreased neuronal activity. Notably, c-Fos was reduced in monoaminergic centers, including the VTA dopamine neurons and the DRN glutamatergic neurons in MHb CAPS2 knockdown mice. CAPS2 plays an important role in calcium-dependent synaptic neuropeptide release, and CAPS2 KO mice have shown decreased release of BDNF and NT-3 in the cerebellum, accumulation of DCVs and decreased SVs near active zone in the hippocampus, leading to abnormal synaptic structure and function 7,11,12 . In the MHb, CAPS2-expressing cells co-express TAC1 and TAC2 ( Fig. 1e and g), which encode DCV contents such as substance P, neurokinin A, neuropeptide K, and neurokinin B 34 . In addition, a recent peptidogenomic study has reported MHb producing 262 neuropeptides generated from 27 prohormones, such as somatostatin (SST), neuropeptide Y (NPY), and secretogranin 1-3, etc 20 .
The IPN receives the majority of afferents from the MHb 15,17 , and we confirmed the widespread en passant projection of MHb AAV-sh-CAPS2 infected axon terminals to the IPN (Fig. 2c). This finding suggests that the accumulation of DCVs at the MHb axon terminals is mediated by CAPS2 deficiency in the MHb (Fig. 3a). Considering that neuropeptides are abundant in the MHb, DCV accumulation at the presynaptic terminals due to the down-regulation of MHb CAPS2 would eventually interfere with the secretion of numerous neuropeptides at the MHb-IPN synapses. On the other hand, SV number and density remain unchanged in CAPS2 KD mice, indicating that the MHb CAPS2 knockdown does not affect exocytosis of SV. As evidenced by the decreased level of p-ERK, which is a downstream molecule in neuropeptide signaling 36,37 , MHb CAPS2 knockdown-mediated reduction in neuropeptide signaling led to decreased neuronal activity in the IPN.
Xu et al. has reported that the substance P content in rat IPN tissue exposed to chronic mild stress (CMS) increases 38 . This CMS-mediated increase of substance P in IPN tissues may be a result of accumulation due to impaired DCV exocytosis caused by CAPS2 reduction. In the same study, the firing rate of IPN neurons in response to substance P perfusion was recorded, and the spontaneous activity after substance P perfusion was increased by approximately 41% of the baseline 38 . Since TAC1, a gene producing substance P, is enriched in MHbD and the activity of IPN neurons is augmented by substance P, the reduction in c-Fos expression in IPN neurons by MHb CAPS2 knockdown (Fig. 4) may be partially related to the impaired exocytic release of substance P, which is consistent with the results of Xu et al.. Although their roles in the habenula are not known for other neuropeptides, they have been reported to be involved in mood changes such as depression. For example, SST is significantly decreased in CSF of depressed patients 39,40 , and SST expression is reduced in the anterior cingulate cortex of MDD patients 41 . Also, the disinhibition of SST positive interneurons leads to reduced anxiety in the elevated plus maze and despair-like behavior in FST 42 . In addition, intracerebroventricular injection of NPY and NPY1R agonist reduced despair-like symptoms in FST 43,44 .
The CAPS2 gene has also been reported to be associated with neuropsychiatric symptoms. Aberrant CAPS2 splicing, leading to lack of the complete sequence of exon 3, has been detected in a subgroup of autistic patients. CAPS2 KO mice also manifested autistic phenotypes, including impaired social interaction, decreased exploratory behavior and increased anxiety in a novel environment, plus increased immobility time in FST 8,12,45 . Nonetheless, the reduction in CAPS2 expression in the habenula did not show any defects in social interaction (Supplementary Figure S4), and habenula lesions, including both MHb and LHb, also did not affect social interaction 46 . We assume that the despair-like symptom in CAPS2 KO mice is partially due to the lack of habenular CAPS2 function, and the reduction of calcium-dependent docking of DCVs to the plasma membrane in the MHb-IPN synapses might have prevented neuropeptide signaling and led to decreased neuronal activity in the IPN and the behavioral change.
It is realized that monoaminergic centers, including VTA and DRN, are important structures in mood-related behaviors. Inhibition of VTA dopamine neurons using optogenetic tools induce increased despair-like symptom and stimulation of VTA dopamine neurons rescues despair-like symptom in the mouse CMS model 29,47 . In recent studies, the release of dual serotonin and glutamate by optical activation from DRN serotonergic neurons projecting to the VTA led to the release of dopamine in the nucleus accumbens and established conditioned place preference 30 . In addition, optogenetic activation of DRN non-serotonergic glutamatergic neurons elicits rewardrelated behavior, suggesting that it is glutamatergic transmission that makes up the majority of the DRN-VTA pathway producing reward-seeking behavior 31,48 .
Although there was no significant activity change in DRN serotonergic neurons in MHb CAPS2 knockdown mice (c-Fos expression of TPH2-positive neurons, Fig. 5d and f), it is notable that glutamatergic neuronal activity was significantly reduced by MHb CAPS2 knockdown (Fig. 5g and h). Recently, the afferent connections of the DRN, conveyed by the IPN, were systemically investigated with sensitive tracers, and demonstrated that the GABAergic neurons of the IPN input into DRN GAD67 expressing neurons 15 . Since GABAergic interneurons of DRN form an internal circuit and modulate DRN serotonergic neurons [49][50][51] , it is reasonable to assume that DRN glutamatergic neurons are also regulated by DRN GABAergic interneurons. One plausible possibility is that the reduced activity of IPN GABAergic neurons by MHb CAPS2 knockdown may increase the activity of DRN GABAergic neurons and lead to disinhibition of DRN glutamatergic neurons. Furthermore, since IPN mainly forms connections with GABAergic neurons in various brain regions such as the median raphe, nucleus incertus, Scientific Reports | (2021) 11:3700 | https://doi.org/10.1038/s41598-021-83310-0 www.nature.com/scientificreports/ supramammillary nucleus, septum, and laterodorsal tegmental nucleus 15 , IPN GABAergic neurons could be indirectly regulating the neuronal activity of the DRN and the VTA through various pathways. Many studies have reported that MHb is closely linked to various neuropsychiatric symptoms 25,52 . We have also previously reported that reduced cholinergic signaling by MHb CHAT knockdown leads to anhedonia but not despair-like behavior 24 . Also, genetic ablation of a key transcription factor leading to the lesion of MHbD, which is abundant in substance P-expressing neurons, resulted in anhedonia-like behavior 22 . It is interesting to note that MHb CAPS2 knockdown induces only despair-like behavior as opposed to that of CHAT knockdown. The results of the above studies suggest that the MHb is a remarkably complex structure, and the functions of dorsal and ventral parts of the MHb act independently or in coordination in affecting emotional behavior. The details of the close relationship between neuronal signaling in the MHbD and MHbV in the MHb-IPN pathway are currently not known. The IPN also has a complex structure compared to its size with an efferent pathway to various brain regions. As such, further research on defining the effects of various neuropeptides and neurotransmitters on synaptic plasticity concerning the IPN neurons and their impact on emotion-related behaviors is warranted.
In this study, we showed that reduced neuropeptide signaling in the MHb-IPN circuit induces despair-like symptoms and changes in neuronal activity in monoaminergic centers. These findings bring to focus the mechanistic importance of the MHb and originating neuropeptide secretion on depression circuitry and functional pathways. Animal model of depression and human subject samples. Animals were randomly divided into experimental and control groups. We used cDNA library generated from mRNA of rats exposed to chronic restrain stress and human patients with MDD 24 . LH model was generated as previously reported 53 . Detailed methods for developing rat CRS and mice LH depression animal models are described in supplementary information. Post-mortem habenular tissue was obtained by the Douglas-Bell Canada Brain Bank (www.dougl asbra inban k.ca; Douglas Institute, McGill University, Canada) following banking guidelines of the Fonds de Recherche du Québec Santé and with approval of the Douglas Institute Ethics Research Board. Brains were collected after obtaining informed consent from next-of-kin. For demographic characterization of human subjects, refer to our prior study 24 . qRT-PCR. qRT-PCR was done as previously described 24 . The habenula was isolated from the animal brain immediately after decapitation and placed in TRIzol solution (Ambion, Austin, TX, USA). Total RNA sample (2 μg) was reverse-transcribed into cDNA using Moloney Murine Leukemia Virus reverse-transcriptase (M-MLV RT; Promega, Madison, WI, USA) and oligo (dT) primer (Novagen, Milwaukee, WI, USA). qRT-PCR was performed with 0.5 μg of the RT product in presence of specific primer sets (Supplementary Table S2). PCR amplification with iQ SYBR Green Supermix was performed in triplicate using the CFX96 Touch-Time System (Bio-Rad, Hercules, CA, USA). Final products of qPCR were electrophoresed on 2% agarose gels and visualized with SafeView Nucleic Acid Stain (G108, Applied Biological Materials, Canada). The cycle numbers (C t ) of the critical point at which the fluorescent signal exceeded the background were determined by qRT-PCR, and expression values for each gene were normalized to expression values of GAPDH, the endogenous control within each sample. Relative quantification used to calculate the fold change was performed using the comparative C t method (ΔΔC t ).

Behavioral paradigms.
Depression behavior studies were done as previously described 24 www.nature.com/scientificreports/ Sucrose preference test. Single-housed mice were habituated with two identical water bottles for a day and then were exposed to two bottles for three days, one with 1% sucrose and the other with tap water. Sucrose and water consumption were recorded daily by re-weighing the two bottles. Sucrose preference was calculated as a relative ratio of mass of sucrose solution intake/total fluid intake. Tail-suspension test. Tail-suspension test was conducted in a 4-chamber apparatus divided by acrylic partitions, and the mice were suspended in each chamber by the tail. A video was recorded for 6 min, and the last 4 min were scored for immobility time.
Forced swim test. Mice were exposed to a clear cylinder filled with 24-25 °C water for 6 min and immobility time was scored for the last 4 min. The cylinder was 45 cm in diameter and 60 cm high.

Social interaction test.
One day before the test, mice were allowed to explore the open arena, in which two cages were placed for acclimatization to the novelty of the environment. On the test day, a cage with a social object (a mouse) was placed on one side of the arena and another cage with a non-social object (marbles) on the other side. The level of sociality was assessed by measuring the frequency and the time mice engaged in interaction with each cage.
Novelty suppressed feeding test. Each mouse was weighed and deprived of food for 24 h. In the plastic cage (30 cm × 50 cm × 20 cm) with standard bedding, the food pellet is placed on a white filter paper. The latency for the mouse to start eating the pellet under bright pin-point light is recorded. After the test, the mice were moved to their home cages but pre-weighed to measure the food amount eaten.
RNAscope assay. Frozen brain sections (14 μm thick) were cut coronally. Sections were then thaw-mounted onto Superfrost Plus Microscope Slides (Fisher Scientific, Waltham, MA, USA). The slides were post-fixed in 4% paraformaldehyde (PFA) and dehydrated in ascending concentration of ethanol, then treated with protease. For RNA detection, RNAscope fluorescent Multiplex detection reagents (ACDBio, Newark, CA, US) were used as in previous studies and following the manufacture's recommendations 54 . Detailed methods and probes used for RNAscope are described in the supplementary information.
Electron microscopy. The animal was anesthetized with alfaxalone and xylazine and transcardially perfused with 0.9% NaCl before 2% PFA, and 2.5% glutaraldehyde in 0.1 M phosphate buffer (pH 7.4). The brain was removed and post-fixed overnight at 4 °C. The IPN was dissected from the fixed brain and washed with the same buffer two times, and treated with 1% osmium tetroxide for 90 min. The tissues were dehydrated with an ethanol series in ascending concentration, propylene oxide, and embedded in epon mixture (Oken Shoji, Tokyo, Japan). Polymerized blocks were trimmed, and the region of interest was selected. Thin sections (70 nm) were made using Leica EM UC6 ultramicrotome (Leica Microsystems), mounted on 200 mesh copper grids, stained with 2% uranyl acetate and 1% lead citrate for 5 min each. For each sample, 50 presynaptic boutons with post-synaptic density were randomly selected and observed under a transmission electron microscope (Hitachi H-7650; Hitachi, Tokyo, Japan) at the accelerating voltage of 80 kV.
Immunohistochemistry. Mice were anesthetized and perfused transcardially with heparinized 0.9% NaCl, then with 4% PFA in phosphate-buffered saline (PBS). Mouse brain was harvested and post-fixed in 4% PFA, then in 30% sucrose in PBS solution at 4 °C. After being cryosectioned to 40 μm thick sections, it was blocked for 1 h in 0.2% Triton X-100, 3% bovine serum albumin in PBS. The sections were incubated with primary antibody overnight at 4℃, and the secondary antibody was applied for 1 h at room temperature. After mounting the slices, the sections were observed on Zeiss LSM700 (Zeiss, Oberkochen, Germany) confocal microscope.
Antibodies. The antibodies for experiments were as follows; CAPS2 (ab69794), TPH2 (ab111828), CAMK2B Western blot. Habenula and IPN tissues were obtained from a cryo-section containing the habenula region using a punch. The tissues were lysed with sodium dodecyl sulfate (SDS) lysis buffer containing 4% SDS, 125 mM