Heterogeneity between VIP and GRP neurons underlies AVP receptor signaling in the mouse suprachiasmatic nucleus

Abstract

Understanding the network topology of a cluster of diverse neurons acting in concert requires a detailed expression map of ligand–receptor pairs involved in cell–cell communication. The neuropeptide arginine vasopressin (AVP) and signaling mediated by its cognate receptor V1a have been implicated in dorsal-to-ventral regional communication in the suprachiasmatic nucleus (SCN), a cluster of neurons that acts in concert to generate daily rhythms in behavior and physiology. Here, we show that among vasoactive intestinal peptide (VIP)-ergic neurons in the ventral SCN only a small subpopulation expresses V1a, and we demonstrate the requirement of V1a in these VIP neurons for maintaining the robustness of the circadian clock using a jet-lag paradigm. Notably, we found that V1a expression appears to be minimal in the other major ventral neuronal population expressing gastrin-releasing peptide (GRP). The identified heterogeneity between VIP and GRP neurons, and among VIP neurons, provides a basic map for understanding the cryptic network structure from dorsal AVP neurons to receiver ventral SCN.

Heterogeneity between VIP and GRP neurons, and among VIP neurons, underlies V1a receptor–dependent vasopressin signaling in the mouse suprachiasmatic nucleus, the locus of the master circadian oscillator in the body.

Introduction

To investigate the network topography of a cluster of diverse neurons that function in concert, a precise map of expression of ligand–receptor pairs underlying cell–cell interactions is necessary and key. The neuropeptide arginine vasopressin (AVP) and its cognate G-protein-coupled receptor V1a-mediated signaling have been demonstrated to participate in cell–cell communication in the suprachiasmatic nucleus (SCN), a cluster of neurons that act in concert to generate daily 24-hour rhythms in behaviour and physiology^1–9. However, the exact role (expression map and function) of V1a receptor in specific populations of intra-SCN neurons is still unclear, resulting in a lack of necessary information for depicting the network structure of AVP-V1a receptor signaling.

The SCN is composed of ~20,000 heterogeneous neurons. Based on anatomical structure and neuropeptidergic properties, the SCN has been subclassified into two distinct regions: a ventral (core) region where vasoactive intestinal peptide (VIP)-expressing neurons and gastrin releasing peptide (GRP)-expressing neurons are located, and a dorsal (shell) region characterized by an enrichment of AVP-expressing neurons^10,11. These two regions are coupled to each other, in both core-to-shell and shell-to-core directions, for maintaining synchronous oscillations^12–15. However, receptor signaling bridging the two regions (particularly shell-to-core direction) has not been fully characterized. The expression of V1a receptor mRNA has been shown to predominantly locate in the dorsal SCN, implicating it in local circuits among AVP neurons, yet its potential role in mediating shell-to-core communication has not been systematically investigated.

Evidence suggests the existence of V1a receptor-mediated signaling from dorsal AVP neurons to ventral (core) SCN neurons. The whole-SCN genetic deletion of V1a receptors profoundly alters the resetting response of mice to shifted light-dark cycles¹⁶. In addition, ablation of the circadian clock component Bmal1 from AVP neurons lengthens the circadian period and enhances the rate of re-entrainment to light¹⁷. Notably, manipulating the cellular circadian period of AVP neurons by deleting or overexpressing CK1δ causes lengthening or shortening of the period of circadian locomotor activity rhythm, respectively¹⁸, suggesting a significant role for AVP neurons in period-setting of the SCN ensemble rhythm. Shan et al. provided more direct evidence supporting the existence of coupling signal(s) from AVP neurons to the rest of the SCN by showing weakened oscillations in non-AVP neurons after Bmal1 deletion within AVP neurons¹⁵.

However, the specific receiver neurons in the ventral SCN — and whether V1a receptor is involved in this communication — are still unclear from the previous studies. In particular, the possibility (hypothesis) of involvement of V1a receptor in dorsal (shell) to ventral (core) communication cannot be addressed by previous studies using global V1a deficient mice⁴, SCN-specific knockout mice¹⁶, or pharmacological inhibition of V1a receptor targeting the whole SCN^4,5. Earlier immunohistochemical studies provide insights into a potential network from AVP neurons to ventrally located VIP and GRP neurons by showing AVP-immunolabelled neuronal fibers ending in proximity to them^19,20. Thus, the potential involvement (or expression) of V1a receptors in these ventral populations — VIP neurons and GRP neurons — is a subject essential for exploring the network structure of AVP-V1a receptor signaling in the SCN.

In the present study, we characterized V1a receptor expression and function in specific populations of ventral SCN neurons. We found that only a specific subset of VIP neurons expresses V1a mRNA, revealing the substantial heterogeneity between VIP and GRP neurons, as well as among VIP neurons, in terms of V1a mRNA expression. We investigated the role of V1a receptors in VIP neurons by generating cell-type specific V1a-deficient mice. The data of our study provide a basis for understanding the receptor signaling underpinning the communication from dorsal AVP SCN neurons to ventral SCN.

Results

Neurites from AVP neurons to ventral VIP neurons and GRP neurons

To characterize the potential anatomical projections from dorsal (shell) SCN to ventral (core) SCN, we traced the potential circuits from AVP neurons to VIP or GRP neurons that are the two representative cell types occupying the ventral SCN (Fig. 1). To do this, we asked whether the neurites emanating from AVP neurons extend toward and form putative contacts with VIP- or GRP-expressing neurons in the SCN. We expressed membrane-bound GFP (mGFP) and synaptophysin-fused mRuby (SYP-mRuby) in AVP cells to label their projecting fibers and putative presynaptic boutons, by delivering a Cre-dependent adeno-associated virus vector AAV-DJ hSyn FLEX-mGFP 2A SYP-mRuby to the SCN of mice expressing Cre recombinase specifically in AVP neurons (AVP-Cre mice¹⁷) (see Fig. 1a). VIP and GRP neurons were each visualized by immunostaining (Fig. 1b–d for VIP; Fig. 1e–g for GRP).

Fig. 1. Neurites of AVP neurons apposed to ventral VIP- and GRP-neurons in the SCN.

a The delivery of a Cre-dependent SYP-mRuby-2A-mGFP viral vector into the SCN of AVP-Cre mice. b Representative SYP-mRuby (magenta) and mGFP (yellow) immunofluorescence images of the SCN co-stained for VIP (cyan). Scale bar, 50 µm. c Representative VIP‑positive soma receiving SYP‑mRuby‑labeled appositions. 3D reconstruction reveals punctate SYP-mRuby signals distributed along a fibrous mGFP-labeled projection. d Pie chart illustrating the proportion of VIP neurons with and without somatic appositions of SYP-positive puncta of AVP neurons. n = 12 SCN sections of 4 mice. e Representative SCN fluorescence images stained for GRP. Scale bar, 50 µm. f Representative GRP‑positive soma receiving SYP‑mRuby appositions along mGFP fibers. g Pie chart of GRP neurons with/without AVP appositions. n = 16 SCN sections of 4 mice.

We observed that a subset of VIP neurons, approximately 25% of total assessed SCN VIP neurons, received SYP-positive appositions from AVP neurons (Fig. 1d). 3D reconstruction revealed that SYP-mRuby puncta were closely apposed to VIP soma (cyan), aligned along mGFP-labeled axon (yellow), and appeared as dot-like signals adjacent to VIP cell surface (Fig. 1c; arrowheads). In comparison, more than half of GRP neurons received such appositions from AVP neurons (Fig. 1e–g), indicating that AVP neurons make putative contacts with VIP and GRP neurons at different ratios — 25.1% of VIP neurons and 53.9% of GRP neurons (Fig. 1d, g).

Optogenetic activation of AVP neurons leads to c-Fos expression in VIP neurons

To test the possibility that activating AVP neurons is accompanied by altered c-Fos expression in VIP and GRP neurons, we next examined the effect of optogenetic activation of AVP neurons (Fig. 2). We generated mice that have AVP neuron-specific expression of channelrhodopsin-2 (ChR2) by crossing AVP-Cre mice with Rosa26-floxed-STOP hChR2(H134R)-mCherry knock-in mice (see also Supplementary Method, Supplementary Figs. 1 and 2a, and Fig. 2a; hereafter referred to as AVP-ChR2 mice). Immunolabeling revealed a high degree of colocalization of mCherry (ChR2) and AVP (97.1% of AVP⁺ neurons were labeled by the Cre-driven mCherry expression, and 94.1% of the labeled cells were AVP⁺; counts of five independent SCN slices) in AVP-ChR2 mice (Supplementary Fig. 2b). In comparison, there was little or no appreciable expression of ChR2 in either VIP or GRP neurons (Supplementary Fig. 2c; 0% of GRP neurons and only 1.8% of VIP neurons were labeled by the Cre-mediated mCherry expression; counts from five independent SCN slices) verifying the specificity of Cre-dependent ChR2 expression¹⁷. Under these conditions, we performed optogenetic stimulation of AVP neurons and tested for its potential effect on VIP neurons or GRP neurons.

Fig. 2. Optogenetic activation of AVP neurons leads to increased c-Fos-positive VIP neurons in the SCN.

a Experimental design. Mice were enucleated to eliminate potential retinal light input. Optogenetic stimulation was applied to the SCN of AVP-ChR2 mice. AVP-Cre mice without ChR2 were subjected to the same procedures. b Representative SCN sections from AVP-ChR2 (left) and AVP-Cre control (right) mice. Sections were stained for VIP (cyan), c-Fos (yellow), and ChR2-mCherry (magenta). c Enlarged view of the ventral region outlined by the dashed box in (b). A solid arrowhead indicates a c-Fos-immunopositive soma of VIP-labeled neuron; open arrowheads point to c-Fos-positive somas that were immunonegative to VIP and express ChR2-mCherry. d Three-dimensional reconstruction of the c-Fos/VIP double-positive neuron from c. Arrows indicate mCherry-labeled AVP projections apposing to the VIP-positive cell body. e–g same as a–c, but sections were stained for GRP. Solid arrowheads on the images in (g) indicate GRP-positive/c-Fos-negative neurons; open arrowheads point to c-Fos-positive somas that express ChR2-mCherry without GRP expression. h Percentages of c-Fos-positive neurons per VIP neurons (left) and GRP neurons (right) in AVP-ChR2 or AVP-Cre mice. VIP: n = 4 mice for AVP-ChR2, n = 5 for AVP-Cre; GRP: n = 4 for AVP-ChR2, n = 4 for AVP-Cre. Values are mean ± SEM. **P < 0.01, two-tailed unpaired t-test. ns, not significant.

Optogenetic stimulation was conducted by delivering blue light (450 nm, 15 Hz, 15-ms pulse duration, 51-ms interval, 1 h) through an optic fiber implanted above the SCN of AVP-ChR2 mice (Fig. 2a). AVP-Cre mice lacking ChR2 expression were similarly examined as controls. All experiments were also performed using enucleated mice, which enables us to exclude potential light-induced c-Fos expression by ambient light or any light source derived from optogenetic stimulation (Fig. 2a). As expected, immunohistochemistry confirmed selective activation of c-Fos expression in the SCN of AVP-ChR2 mice but not in AVP-Cre control mice after optogenetic stimulation (Fig. 2b). The majority of c-Fos induction (yellow) occurred in mCherry-positive (thus directly activated) AVP neurons (Fig. 2b). Under these conditions, we observed that a subset of VIP neurons located in the ventral SCN was also marked by positive c-Fos expression following AVP neuronal activation (Fig. 2b, dashed boxed area). Importantly, these cells were c-Fos and VIP-positive, but were mCherry-negative, indicating that their c-Fos expression was not the result of direct optogenetic stimulation (Fig. 2c). 3D reconstruction confirmed nuclear c-Fos expression (yellow) in a VIP-positive but mCherry-negative soma (cyan) that received putative appositions from AVP neurons (magenta, pointed by arrows, Fig. 2d).

On the other hand, to our surprise, c-Fos-positive cells that we identified in the ventral SCN were rarely overlapped with GRP neurons (see Fig. 2e-g; GRP-immunopositive, c-Fos immunonegative cells are indicated by solid arrowheads in Fig. 2g), which contrasts with activated c-Fos expression in mCherry-positive cells (open arrowheads) nearby GRP neurons. We quantified the number of c-Fos expressing VIP or GRP neurons throughout the SCN (see Fig. 2h). The average of c-Fos/VIP double positive cells per SCN was calculated as 9.0 ± 1.1 (± s.e.m) in optogenetically activated AVP-ChR2 mice (n = 4 mice), being higher than 1.5 ± 0.7 cells in similarly treated control AVP-Cre mice (n = 5 mice). By comparison, the number of c-Fos/GRP double positive cells was minimal (0 or 1) in AVP-ChR2 mice as well as in AVP-Cre control mice (n = 4 mice for each group), indicating the difference between VIP neurons and GRP neurons in the capacity of c-Fos expression upon AVP neuron stimulation. The values shown in Fig. 2h are the percentages of c-Fos positive VIP or GRP neurons with respect to the total number of corresponding neurons detected, with statistical comparison (VIP: P = 0.0015, GRP: P = 0.94, AVP-ChR2 vs. AVP-Cre mice, unpaired t-test). Taken together, our data suggest that although both VIP neurons and GRP neurons received appositions from AVP neurons (see Fig. 1) and are located similarly in the ventral SCN, they differ from each other: a small subpopulation of VIP neurons (but not of GRP neurons) possesses the ability to express activated c-Fos expression in response to AVP neuron stimulation.

V1a mRNA distribution profiles: VIP neurons vs. GRP neurons

To investigate the potential underlying molecular mechanism of the communication from AVP neurons to the small subset of VIP neurons, we hypothesized that AVP receptor V1a might be a mediator for this signaling. V1a has been described to be a major AVP-receptor subtype in the SCN^6,21 and it couples to Gq^22,23, thus linked to neuronal activation²⁴; however, whether V1a is expressed in ventrally located VIP-expressing neurons (or GRP neurons) is not reported. To test this idea, we first examined V1a mRNA distribution in the SCN by using digoxigenin in situ hybridization (DIG-ISH) in combination with anti-VIP immunostaining. As many studies reported^4,6,21, V1a signals were predominantly identified in the dorsal region of the SCN (Fig. 3). Nevertheless, some signals were also identified in the ventral SCN, where we noted a small but clear overlap of V1a and VIP expression signals (see Fig. 3a, lower panels: V1a, green, pseudo-colored; VIP, magenta; although semi-quantitative due to our staining methodology, the overlap appeared in ~10% of VIP cells counted, n = 242 from 6 independent SCN sections). To validate this finding and gain quantitative data regarding the V1a mRNA expression in VIP neurons, in comparison with those in GRP neurons, we next employed confocal microscopy-based comparison of V1a mRNA fluorescence in situ hybridization (SABER-FISH²⁵) signals in VIP neurons and GRP neurons, each marked by VIP-Cre or GRP-Cre dependent GFP expression (Fig. 3b, upper panels, VIP; lower panels, GRP).

Fig. 3. AVP receptor mRNA is highly expressed in a subset of VIP neurons but minimal in GRP neurons.

a Double labeling of V1a transcript and VIP protein in the SCN. Upper panels: DIG in situ hybridization of V1a (left) and immunofluorescence of VIP (right). Scale bar, 50 µm. Lower panels: enlarged view of the boxed region from the upper panels. A merged image of V1a (green, pseudo-colored) and VIP is shown, with a higher-magnification view shown to the right. Arrowheads indicate cells that have overlapping signals of V1a and VIP. Scale bars, 20 µm (left), 10 µm (right). b SABER-FISH imaging of V1a expression in SCN-VIP or -GRP neurons. VIP and GRP neurons were detected using their respective Cre-reporter mice. DAPI staining identifies all cell nuclei. The dotted boxes indicate the areas of high magnification. Closed and open arrowheads indicate V1a-positive and -negative neurons, respectively. Scale bars, 50 µm (left), 15 µm (right). c Percentages of VIP and GRP neurons that express V1a mRNA puncta in the SCN. d Histogram comparing the distribution of the number of V1a puncta per cell between VIP and GRP neurons. Values are expressed as percentage of VIP neurons (total 368 cells from n = 7 SCN sections of 3 mice) and GRP neurons (total 344 cells from 6 sections of 2 mice). Values are mean ± SEM. ****P < 0.0001, two-way analysis of variance (ANOVA) followed by Bonferroni’s post hoc test.

Notably, these approaches allowed us to detect a previously unrecognized difference between VIP neurons and GRP neurons in terms of V1a mRNA expression: V1a transcripts (represented by red puncta in SABER-FISH labeling) exhibit a remarkably skewed distribution toward VIP neurons compared to GRP neurons (Fig. 3b–d). V1a mRNA-positive cells were observed for 7.1 ± 1.5% (mean ± s.e.m.) of VIP neurons and 1.9 ± 0.5% of GRP neurons (Fig. 3c, P = 0.011, VIP vs. GRP, unpaired t-test). Cells with more than seven V1a mRNA puncta were observed only for a subpopulation of VIP neurons (1.4 ± 0.4%), whereas such cells were not detected among GRP neurons (Fig. 3d). These data are aligned with a recently reported SCN single-cell RNA-seq dataset²⁶, by which we were able to reproduce the greater V1a mRNA expression in VIP neurons than in GRP neurons (Supplementary Fig. 3).

Deficient optogenetic activation of c-Fos expression in VIP neurons in V1a^–/– mice

If V1a is a mediator is a mediator for activating the subset of VIP neurons, deletion of V1a would result in altered up-regulation of c-Fos expression after AVP neuron activation. We therefore generated V1a receptor-deficient AVP-ChR2 mice (hereafter V1a^–/–; AVP-ChR2 mice) by crossing AVP-ChR2 mice (i.e., AVP-Cre; floxed-STOP ChR2(H134R)-mCherry mice) with V1a receptor knockout mice^4,27 (Fig. 4a). Optogenetic stimulation was provided to V1a^–/–; AVP-ChR2 mice as well as to V1a^+/+; AVP-ChR2 mice for comparison (Fig. 4b, left vs. right). As was observed in (V1a-wildtype) AVP-ChR2 mice (see Fig. 2), clear c-Fos-positive VIP neurons were observed in V1a^+/+; AVP-ChR2 mice after optogenetic stimulation (Fig. 4c) with a number of 9.6 ± 0.8 cells per SCN (Fig. 4e, n = 8 mice). In contrast, V1a^–/–; AVP-ChR2 mice were found to express a remarkably reduced c-Fos⁺/VIP⁺ cell number with an average of 1.9 ± 0.3 cells per SCN (n = 7 mice; Fig. 4d,e, P < 0.0001 vs. V1a^+/+; AVP-ChR2 mice), indicating an engagement of V1a receptor in the up-regulation of c-Fos expression in VIP neurons triggered by AVP neuron activation. The values shown in Fig. 4e are the percentages of c-Fos expressing VIP neurons with respect to the total number of VIP neurons detected. Both V1a^–/–; AVP-ChR2 mice and V1a^+/+; AVP-ChR2 mice showed comparable c-Fos expression among mCherry-positive (i.e. ChR2-expressing) AVP neurons (80.5 ± 1.4% and 83.5 ± 2.5% for V1a^–/– and V1a^+/+, respectively; P = 0.33, n = 7 mice, V1a^–/–; n = 8 mice, V1a^+/+), verifying equivalent effectiveness of AVP neuron activation between genotypes (Fig. 4b).

Fig. 4. Deficient optogenetic activation of c-Fos expression in VIP neurons in V1a^–/– mice.

a Experimental design. V1a wildtype (V1a^+/+) or V1a-deficient (V1a^–/–) AVP-ChR2 mice were used. b Representative images of SCN from V1a^+/+; AVP-ChR2 (left) and V1a^–/–; AVP-ChR2 (right) mice. Sections were stained for VIP, c-Fos, and ChR2-mCherry. Scale bars, 50 µm. c, d Enlarged views of the ventral SCN region of V1a^+/+ (c) or V1a^–/– (d) AVP-ChR2 mice. Filled arrowheads indicate c-Fos/VIP double-positive neurons; shaded arrowheads, c-Fos-positive but VIP-negative neurons; open arrowheads, c-Fos-negative VIP neurons. Scale bars, 15 µm. e Percentage of c-Fos-positive VIP neurons per SCN in V1a^+/+ (n = 8), and V1a^–/– (n = 7) AVP-ChR2 mice. Values are means ± SEM. ****P < 0.0001, two-tailed unpaired t-test.

Targeted V1a deletion in VIP neurons affects adaptability of animal behavior to LD cycle

V1a receptors in the SCN have been demonstrated to play a key role in forming the resilience of the circadian clock to abrupt environmental light-dark (LD) cycle shifts — experimental conditions that are analogous to jet lag or shift work^4,16. However, whether the V1a receptors that are expressed by VIP neurons contribute to this process remains unknown. We therefore finally inquired the potential functional contribution of the V1a receptors of VIP neurons. We generated mice lacking V1a receptors specifically in VIP neurons (hereafter referred to as VIP-V1a^–/– mice) by crossing V1a-floxed mice (V1a^flox/flox mice)¹⁶ with hemizygous VIP-Cre mice.

As expected from V1a global knockout mice⁴ as well as from SCN-wide knockout mice¹⁶, the circadian period length of VIP-V1a^–/– mice in constant darkness was indistinguishable from that of their littermate control mice (V1a^flox/flox mice), verifying the dispensability of V1a in period determination (Supplementary Fig. 4). Then, VIP-V1a^–/– and control V1a^flox/flox mice, both male and female, were subjected to simulated jet-lag conditions involving an 8-hour advance or delay of LD cycle (Fig. 5). Following an 8-hour LD advance (Fig. 5a, b), VIP-V1a^–/– mice exhibited significantly accelerated behavioral re-entrainment compared to V1a^flox/flox controls in both sexes (Fig. 5a). The rate of re-entrainment assessed by 50% phase-shift (PS50) — half of the number of days required for complete re-entrainment — was 5.5 ± 0.6 days in control males vs. 3.2 ± 0.5 days in VIP-V1a^–/– males (PS50, means ± s.e.m, n = 10 for V1a^flox/flox and n = 9 for VIP-V1a^–/–, P = 0.012, unpaired t-test), and was 3.5 ± 0.5 days in control females vs. 2.0 ± 0.5 days in VIP-V1a^–/– females (n = 10 for V1a^flox/flox and n = 11 for VIP-V1a^–/–, P = 0.036).

Fig. 5. Targeted V1a ablation in VIP neurons alters adaptation speed to abrupt shifts of the environmental LD cycle.

a Representative double-plotted actograms of V1a^flox/flox and VIP-Cre;V1a^flox/flox (VIP-V1a^–/–) mice before and after an 8-h phase advance in LD cycles for male (upper) and female (lower). Arrows indicate the day of complete reentrainment. b Activity onsets (left) and PS₅₀ values (right) of (a). c Representative actograms before and after an 8-h phase delay in LD cycles. d Activity onsets (left) and PS₅₀ values (right) of (c). Values in (b) and (d), are means ± SEM. V1a^flox/flox mice: n = 10 for male, n = 10 for female; VIP-V1a^−/− mice: n = 9 for male, n = 11 for female. *P < 0.05, **P < 0.001, two-tailed unpaired t-test.

As was reported in previous studies^28,29, it is worth noticing that female mice needed fewer days to re-entrain to new LD cycles than male mice did, while the phenotype of faster re-entrainment in VIP-V1a^–/– was conserved in both sexes. Following an 8-hour delay regimen, on the other hand, male VIP-V1a^–/– mice exhibited more rapid re-entrainment than controls (1.1 ± 0.3 days for VIP-V1a^–/– vs. 2.2 ± 0.2 days for controls; n = 9 and 10, respectively, P = 0.007), whereas no statistically significant difference was observed for females, where both genotypes exhibited similar PS50 values as fast as approximately 1 day (1.0 ± 0.1 days for VIP-V1a^–/– and 1.2 ± 0.1 days for controls; n = 11 and 10) (Fig. 5c, d). These results indicate that V1a receptors in VIP neurons are indispensable for keeping normal phase response of the circadian clock, while it appears that the extent of the faster re-entrainment phenotype differs slightly between sexes.

Discussion

The previous studies investigating the role of V1a receptors in the SCN have been conducted using global V1a deficient mice (V1a^–/–)⁴ as well as using the SCN-ubiquitous, Slc32a1-Cre dependent tissue-specific knockout mice¹⁶, and by pharmacologically administering V1a receptor blockers to the whole SCN^2,4–6. However, whether and to what extent the V1a receptors present in VIPergic neurons subserve a relevant role has remained unexplored. Within this research framework, in this study, we have demonstrated that the mice lacking V1a receptors in VIP neurons (VIP-V1a^–/– mice) re-entrained more rapidly to shifted light-dark cycles compared to control mice (Fig. 5). These data provide functional evidence for the role of V1a receptors in VIPergic neurons and suggest the existence of V1a receptor-mediated signaling from AVP neurons to VIP neurons in maintaining normal phase modulation of the circadian clock against abrupt shifts in light-dark cycle. We observed that V1a receptors are required for inducing c-Fos expression in VIP neurons in response to optogenetic activation of AVP neurons (Fig. 4); we also found that V1a receptors are distributed in a small subpopulation of VIP neurons while only minimally in GRP neurons (Fig. 3), a finding compatible with undetectable c-Fos expression in GRP neurons upon optogenetic activation of AVP neurons (Fig. 2). Anatomically, synaptic appositions derived from AVP neurons were present on both GRP neurons and VIP neurons, suggesting additional communication modes involving neither V1a nor c-Fos induction exist, to be investigated in the future (Fig. 1). Our data about V1a receptors therefore not only provide evidence for the functional relevance of V1a receptors in VIP neurons but also reveal a previously uncharacterized heterogeneity between VIP and GRP neurons — the two dominant cell groups in the ventral SCN.

Besides the VIP neurons-versus-GRP neurons difference, our data revealed an additional layer of diversity within the population of VIP-expressing neurons. We found that a relatively small subpopulation of VIP neurons expresses V1a receptors (Fig. 3), indicating the presence of at least two subtypes of VIP neurons, one with V1a receptors and the other without — which is compatible with the subclassification based on recent single-cell RNA-seq data^26,30. In this context, it is interesting to reiterate that although few in cell number, deletion of V1a receptors specifically in VIP neurons was able to affect the rate of reentrainment in mice to shifted light-dark cycles (Fig. 5). Additionally, it is our surprise that the magnitude of the re-entrainment speed (determined by PS50) in these mice (Fig. 5) was almost comparable to that reported for the whole SCN V1a deficient mice¹⁶, suggesting the possibility that the defined small cell population conveying V1a receptor-signaling from AVP neurons to VIP neurons may have a critical role in the whole SCN function. However, a caveat exists in this speculation. Because we merely used VIP-V1a^–/– mice in which V1a receptors were removed from VIP neurons, we cannot exclude the possibility of secondary effects derived from the genetic deletion, such as those due to developmental and/or transcriptional defects affecting the SCN network structure.

It is also possible that AVP signals originating outside the SCN, including those from retinal ganglion cells as suggested by Tsuji et al.³¹, may act in concert with the intra-SCN AVP circuit. Thus, the in vivo photic entrainment alteration of VIP-V1a^–/– mice may be a mixed phenotype. Our data also do not negate the importance of V1a in other neuron types such as AVP neurons.

Post-developmental deletion and inhibition of circuit-specific V1a receptor signal in vivo as well as ex vivo will therefore be necessary for addressing the hypothesis assuming the importance of V1a receptor-mediated signaling in a small subset of VIP neurons. Our current findings on the heterogeneity between VIP neurons and GRP neurons, and among VIP neurons, in terms of V1a receptor expression, within the SCN therefore provide a basis for further investigation of AVP-V1a circuits contributing to the organization of the central clock function.

A more general question raised by our data is how AVP is signaled to receiver neurons, either synaptically or non-synaptically. Although optogenetic activation of AVP neurons can elicit c-Fos expression in VIP neurons in a V1a receptor-dependent manner, the mode(s) of AVP secretion (transmission) remains uncertain. It has been demonstrated that paracrine action of AVP is required for restoring circadian oscillation of clock gene expression in the SCN by co-culturing SCN slices with pharmacological inhibition of AVP receptor signaling in the absence of parallel VIP signaling². This suggests the possibility of paracrine signaling of AVP to V1a receptor-expressing VIP neurons. However, besides this hypothesis, signaling through non-paracrine mechanisms may also be possible (Supplementary Fig. 5)³². Two other potential mechanisms include i) trans-synaptic transmission, where AVP release at synapses (synaptic cleft) signals upon postsynaptic receptors on the target neuron, and ii) extra-synaptic local diffusion, where AVP is released from non-synaptic axonal sites and locally diffuses to the receptors on the postsynaptic receiver neuron (see Supplementary Fig. 5, models 1, 2 and 3).

The reported presence of dense-core vesicles (DCV) in the synapses of SCN neurons^33,34 and our observation of anatomical appositions from AVP neurons to VIP neurons are compatible with these two additional models. It should be mentioned that the results of our current study cannot distinguish between these possibilities and it is possible that they are not mutually exclusive, perhaps acting in concert over different spatial and temporal frameworks as previously discussed (or proposed) by other researchers^2,7.

A comprehensive mapping (and understanding) of connections between AVP neuron and GRP neuron as well as those between AVP neuron and VIP neuron requires more in-depth analysis. Particularly, in our study, because of the limitation of dendrite visualization in our anti-VIP and anti-GRP immunostaining, we only evaluated the number of appositions (SYP-positive) from AVP neurons to each neuronal cell body (either VIP-positive or GRP-positive) as performed in previous studies^19,35,36. Consequently, there is a critical underestimation of AVP innervation to both VIP and GRP neurons in our present dataset. To investigate axodendritic contacts, more precise methods such as electron microscopy analysis are required^36–38. Moreover, we only used c-Fos expression as a marker to assess AVP-V1a receptor signaling. However, although V1a receptor couples to Gq and has been linked to neuronal activation^22–24, we do not rule out the possibility of V1a receptor signaling that does not involve c-Fos expression. Thus, the potential V1a receptor signaling from AVP neurons to V1a-positive VIP neurons may be underestimated.

This may explain the observation that AVP neurons establish putative synaptic inputs with >25% of VIP neurons (Fig. 1d), yet only <10% of these cells coexpress V1a (Fig. 3d) and a few exhibit detectable c-Fos expression in response to optogenetic activation of AVP neurons (Fig. 2h, 1.5%). In addition, to understand the functional connections that are not mediated by the V1a receptor, further investigation utilizing e.g. calcium and electrophysiological recording will be required³⁹. Given that almost all SCN neurons are GABAergic, inhibitory signals that involve neither V1a nor c-Fos induction may mediate the multiple connections from AVP neurons to VIP and GRP neurons, which require further investigation.

Several lines of evidence strongly suggest the existence of signaling from dorsal SCN AVP neurons to ventral SCN^15,17. Shan et al. reported that a functional circadian clock component Bmal1 in AVP-neurons is essential for synchronized oscillations of the entire SCN, which was demonstrated by dual simultaneous recording of circadian rhythms in AVP neurons and those in the rest of the SCN using two different PER2:luc reporters, green and red, respectively¹⁵. Another study from Edwards et al. reported that pharmacological blockade of AVP receptors interfered with the development of dorsal-to-ventral spatiotemporal wave of circadian clock gene expression in Cry1/2-deficient SCN slices upon introduction of rhythmic Cry1 expression⁵. These studies suggest the presence of coupling signals mediated by AVP receptors from AVP-expressing neurons to the rest of SCN, including the ventral area of the SCN. In this context, we featured the presence and function of V1a receptors in a subset of VIP neurons located in the ventral SCN, revealing the heterogeneity between VIP and GRP neurons. Our data thus add two previously hidden facets to the AVP signaling: the identification of V1a receptor’s function in a subset of VIP neurons and the absence of V1a in GRP neurons and the other (remaining) VIP neurons. These may help to provide a blueprint for understanding and studying the still cryptic network structure from AVP neurons to receiver ventral SCN neurons. The role of V1a receptor-positive VIP neurons in the ventral SCN in terms of organization of GRP neurons as well as V1a receptor-absent VIP neurons will require further study. It is tempting to speculate that multiple mechanisms involving, for example, gap junctions, synaptic GABAergic signaling, and other neuropeptidergic cues may act as synchronizers for the dorsal-to-ventral communication which underpins the robustness of the circadian clock.

Methods

Animals

AVP-ChR2 mice were generated by intercrossing AVP-Cre mice (Mouse Genome Informatics, Tg(Avp-icre)#Meid, ID: 5697941) and Rosa26-floxed-STOP hChR2 (H134R)-mCherry mice (RIKEN BioResource Research Center, accession No. RBRC11087; Supplementary Method). V1a receptor-deficient AVP-ChR2 mice were similarly generated using V1a^–/– mice⁴ that we developed previously²⁷. VIP-Cre;EGFP and GRP-Cre;EGFP mice were generated by crossing VIP-ires-Cre mice (The Jackson Laboratory, stock No: 010908) and Grp-iCre mice⁴⁰ (RIKEN BRC, accession No: RBRC10694) with Rosa26^fsTRAP mice (The Jackson Laboratory, stock No: 022367). VIP-V1a^–/– mice were generated by crossing V1a^flox/flox mice¹⁶ (Mouse Genome Informatics, ID: 7639761) with VIP-ires-Cre mice. To avoid the altered locomotor phenotype of homozygous VIP-ires-Cre mutants⁴¹, only heterozygous mutant mice were used. Moreover, we confirmed the specificity of VIP-ires-Cre mice in labeling VIP neurons (in Fig. 3b, 80.9% of VIP⁺ neurons were labeled by the VIP-Cre-driven EGFP reporter expression and 98.3% of the labeled cells were VIP⁺), suggesting that V1a deletion happens in VIP neurons. Mice used in this study were all bred on a C57BL/6 J background. Animals were housed under a regular 12-h light/12-h dark (LD) cycle, maintained at 22 ± 2°C, with free access to food and water. All animals were group-housed (3–4 per cage) unless otherwise mentioned. All animal experiments were conducted in compliance with ethical regulations of Kyoto University and performed under protocols approved by the Animal Care and Experimentation Committee of Kyoto University. We have complied with all relevant ethical regulations for animal use. Regarding animal care, no expected or unexpected adverse events were observed throughout the study.

Introduction of SYP-mRuby into SCN AVP neurons by AAV delivery

AVP-Cre mice were anesthetized with isoflurane and positioned in a stereotaxic apparatus (Kopf). A viral vector solution of AAV-DJ hSyn FLEX-mGFP 2A Synaptophysin-mRuby (Stanford Gene Vector and Virus Core, GVVC-AAV-100, titer 4.7 ×10¹² GC/ml) was delivered into the SCN at the speed of 0.1 µl/min for a total amount of 0.4 µl using a Hamilton Syringe (Neuros Model 1701). Stereotaxic coordinates for SCN injection were anteroposterior (AP) 0.0 mm, mediolateral (ML) 0.0 mm, and dorsoventral (DV) − 5.7 mm. Following surgery, mice were singly housed in individual cages. Four weeks after AAV delivery, mice were processed for immunofluorescence staining.

Immunofluorescence

Free-floating immunofluorescence staining was performed using 30-µm-thick serial coronal brain sections as described previously^42,43. The primary antibodies used were: chicken polyclonal anti-GFP antibody (Abcam, ab13970, RRID: AB_300798, 1:2 K dilution), rat monoclonal anti-mCherry (Thermo Fisher Scientific, M11217, RRID: AB_253611, 1:1 K), rabbit polyclonal anti-AVP (Millipore, ab1565, RRID: AB_90782, 1:5 K), rabbit polyclonal anti-VIP (1:3K)⁴⁴, rabbit polyclonal anti-GRP (ImmunoStar, 20073, RRID: AB_572221, 1:1 K), and guinea pig polyclonal anti-c-Fos antibody (Synaptic Systems, 226-004, RRID: AB_2619946, 1:500). The secondary antibodies used were: Alexa 488-conjugated goat anti-chicken IgG (Thermo Fisher Scientific, A11039, RRID: AB_2534096), Alexa 568 donkey anti-rat IgG (Abcam, ab175475, RRID: AB_2636887), Alexa 647 donkey anti-rabbit IgG (Abcam, ab150075, RRID: AB_2752244), and Alexa 488 donkey anti-guinea pig IgG (Jackson ImmunoResearch, 706-545-148, RRID: AB_2340472), each in 1:500 dilution. To minimize technical variations in immunostaining, brain sections from different genotypes were processed simultaneously. Sections were stained with DAPI before imaging. Images were obtained using a Nikon A1 confocal microscopy system with a ×40 objective lens. Imaging settings—including laser power, HV, pinhole size, and scan speed—were kept consistent across samples in each experimental condition. To analyze and gain 3D images of cells with synaptic appositions, z-stack images were collected under a ×60 objective lens using ×2 zoom, with a 512 × 512 pixel resolution and a z-step of 0.2 µm, resulting in a 0.2 × 0.2 × 0.2 μm voxel dimension in x, y, and z directions; the images were then deconvoluted with Huygens software (Scientific Volume Imaging) and 3D-reconstructed with Imaris software (Oxford Instruments) to obtain 3D images. For determination of appositions, each VIP or GRP neuronal cell body was examined for the presence of SYP-mRuby⁺/mGFP⁺ double-positive puncta derived from AVP axons. A cell was classified as SYP-positive if its soma contained SYP-mRuby⁺/mGFP⁺ double-positive puncta whose edge was within 1 pixel of the soma border. There are caveats to the present methods of evaluating contacts, including (1) appositions less than 1 pixel in distance are not differentiated, (2) axodendritic contacts were not assessed and (3) over or underestimation of the number of appositions due to immuno-histochemistry procedures, or (4) to missing fibers cut during sectioning, may occur as previously acknowledged^19,35,36.

Optogenetic activation of SCN AVP neurons and c-Fos evaluation in VIP/GRP neurons

AVP-ChR2, AVP-Cre, and V1a-deficient (V1a^–/–) AVP-ChR2 mice were stereotaxically implanted with an optical fiber (Doric Lenses, 400-µm core diameter, 5-mm length, 0.53 numerical aperture, 18–20-mW light intensity) targeting to the SCN using the coordinates of AP: –0.4 mm, ML: 0.0 mm, DV: –5.0 mm for the fiber tip. To avoid retinal stimulation by ambient light, both eyeballs of mice were surgically removed under anesthesia. After at least 3 days of recovery, mice were connected to a fiber-optic patch cord (Doric Lenses, 400-µm core diameter, 25-cm length, 0.53 numerical aperture) attached to the implanted fiber-optic cannula and individually housed in constant darkness (DD). The mice were allowed to accommodate for at least 10 days. Optogenetic stimulation was performed at CT14 for 60 min at 15 Hz (15-ms duration, 51-ms interval) using a 450-nm blue LED (Doric Lenses, CLED_450) controlled by an LED driver (Doric Lenses, LEDD1). Immediately after stimulation, mice were sacrificed for immunofluorescence staining. Serial coronal brain sections, covering from the rostral-most to the caudal-most ends of the SCN, were examined. For quantification, an observer blinded to mouse genotypes manually counted VIP/GRP and c-Fos double-positive cells in the SCN using ImageJ software. Brightness and contrast settings were identical across samples. VIP neurons that were immunopositive for mCherry (i.e., co-expressing AVP-ChR2, accounting for 1.8% of VIP neurons) were excluded from analysis to avoid counting c-Fos expression directly induced by optogenetic stimulation.

DIG in situ hybridization followed by immunostaining – for V1a mRNA in VIP neurons

DIG in situ hybridization was performed as described⁴⁵ using the mouse V1a antisense cRNA probe targeting the V1a nucleotides 1292–2627 (NM_016847). Coronal brain sections (30-µm thick) were prepared from mice sacrificed at ZT16 and processed for free-floating DIG in situ hybridization. After the enzymatic colorimetric detection of V1a mRNA using NBT/BCIP, the same sections were subjected to anti-VIP immunostaining. VIP was visualized by immunofluorescence using Alexa-647-labeled secondary antibody. Images were acquired using a BZX-710 microscope (Keyence) for both V1a (colored) and VIP (fluorescence) signals.

SABER-FISH for V1a mRNA expression in VIP and GRP neurons

SABER-FISH was performed using brain sections from Cre-dependent Rosa26-EGFP-L10a mice crossed with either VIP- or GRP-Cre mice. A set of gene-specific probes for the mouse V1a mRNA (NM_016847) was designed using the OligoMiner pipeline⁴⁶ followed by BLAST homology search to ensure minimal sequence-similarity to other transcripts (probe sequences are available in Supplementary Data 2). ssDNA probes were extended by primer exchange reaction (PER)²⁵, using BST DNA polymerase large fragment (New England Biolabs), dHTP (dATP, dTTP and dCTP), DNA hairpin oligos, and Clean.G oligo, according to a protocol by Kishi et al²⁵. Secondary DIG-labeled PER concatemers targeting the primary probes were extended using DIG-11-dUTP (Roche Diagnostics) for anti-DIG-POD-based detection using Styramide™ Signal Amplification system (AAT Bioquest) (see Supplementary Data 2 for sequences of oligos used for SABER-FISH). Serial coronal brain sections were processed for SABER-FISH as reported in earlier studies^25,47,48 except that, after the hybridization of secondary DIG-labeled concatemers, signals were acquired using POD-conjugated anti-DIG antibody (Roche) and iFluor® 568 dye-labeled Styramide™ conjugate (AAT Bioquest). To visualize VIP/GRP neurons, EGFP expression was immunodetected using the following antibodies: chicken polyclonal anti-GFP (Abcam, ab13970) and Alexa 488-conjugated goat anti-chicken IgG (Thermo Fisher Scientific, A11039). Z-stack image sequences of the SCN were acquired using Leica TCS SP8 confocal microscope under a ×40 objective lens, and the expression of V1a mRNA in VIP/GRP neurons was assessed using ImageJ software.

Locomotor activity recording and data analysis

Age-matched VIP-V1a^–/– and V1a^fl/fl mice (3–6 months old) were individually housed in light-tight, ventilated closets in a temperature- and humidity-controlled facility with ad libitum access to food and water. The animals were entrained to a 12-h light (~200 lux fluorescent light)/12-h dark cycle for at least two weeks to synchronize (entrain) the circadian clock of the mice to the ambient light-dark cycle, and then light-dark cycles were either 8-h phase-advanced or 8-h phase-delayed as previously reported^4,16,49. Locomotor activity was monitored using a passive (pyroelectric) infrared sensor (FA-05 F5B; Omron), and the obtained data were analyzed with ClockLab software (Actimetrics) running on MATLAB (MathWorks)⁵⁰. Onset and offset time points of locomotor activity under the jet lag conditions were acquired by visually inspecting activity data, referencing gradual behavioral transitions across days. The speed of behavioral re-entrainment was evaluated using 50% phase-shift values (PS50)^4,16,49. PS50 values were calculated by fitting sigmoidal dose–response curves with variable slope, Y = Bottom + (Top – Bottom)/(1 + 10^{(log PS50 – X) HillSlope}), to daily onset or offset time points in GraphPad Prism. Free-running period was determined using mice kept under constant dark (DD) conditions. Period measurements were based on a 10-day interval taken after 2 d of a DD regime and were executed with a χ² periodogram as described previously^51,52.

Analysis of publicly available scRNA-seq data

Relative mRNA levels of V1a in AVP, VIP, CCK, and GRP neurons were estimated from a publicly available single-cell RNA-seq dataset by Wen et al.²⁶ (Gene Expression Omnibus accession number: GSE117295). This dataset comprises SCN cells isolated from adult mice across 12 circadian timepoints (every 4 h for 48 h). In order to exclude low-quality cells, the cells were filtered by gene count, removing cells with less than 200 or more than 800 genes, and by mitochondrial gene detection ratio, with the threshold set to 10% — a total of 45,515 cells met these criteria and were included in the analysis. Cells were classified into a group of marker-positive cells (UMI value ≥ 1) or negative cells (UMI = 0) for Avp, Vip, Cck, and Grp. For each cell group, percent V1a expression was calculated relative to the total values of V1a expression in cells that were included. The percent values were calculated for each timepoint, and equivalent circadian timepoints were plotted in parallel with their average (i.e., n = 2).

Statistics and reproducibility

All experimental conditions include a minimum of three mice, and mice were randomly assigned to experimental groups, ensuring that each litter was distributed across groups. Animal cage locations were also randomized. The sample size was determined based on previous studies conducted in similar analyses^4,16,19,39 No animals were excluded during the experiments or data analysis. Inclusion and exclusion criteria were not predefined. Statistical analysis and graph plotting were conducted using Microsoft Excel and GraphPad Prism 8 with the statistical tests shown in the figure legends. All data are expressed as means ± SEM. Experiments with one variable were analyzed by Student’s t-test, and experiments involving two variables were analyzed by two-way ANOVA with Bonferroni’s multiple comparison test. A P-value of 0.05 or higher indicated a non-significant (ns) result.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Author contributions

M.D., H.Z. and D.M. designed the research with Y.Y. and W.J.S.; H.Z. and D.M. performed experiments in collaboration with S.-W.H., Y.Y., M.A., T.K., K.I., K.S., H.O. and E.H.; M.D. and H.Z. wrote the paper with input from all authors; M.D. supervised the project.

Peer review

Peer review information

Communications Biology thanks Mike Ludwig, Chris Coyle and the other, anonymous, reviewer(s) for their contribution to the peer review of this work. Primary Handling Editors: Shani Stern and Benjamin Bessieres. [A peer review file is available].

Data availability

The numerical source data for graphs and charts are provided in Supplementary Data 1. All other data are available from the corresponding author upon reasonable request.

Code availability

This paper does not report original code.

Materials availability

Rosa26-floxed-STOP hChR2(H134R)-mCherry mouse line was generated by Kenji Sakimura and Keiichi Itoi and deposited to RIKEN BioResource Research Center (RIKEN BRC No. 11087, https://knowledge.brc.riken.jp/resource/animal/). All other reagents generated in this study are available from the lead contact with a completed materials transfer agreement.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s42003-026-09694-9.