The expression of Mc4r within Kiss1^ARH and Kiss1^AVPV/PeN neurons has already been described elsewhere (Cravo et al., 2011; Lam et al., 2021; Villa et al., 2024; Stephens and Kauffman, 2021), suggesting a role for MC4R in the regulation of fertility. To further investigate the specific role of Mc4r in Kiss1 neurons, we generated a Kiss1-specific Mc4r knockout mouse model (Kiss1- Mc4r KO). The specific deletion of Mc4r from Kiss1 neurons, as well as the absence of global recombination, was confirmed through RNAscope. While Kiss1 neurons of Kiss1- Mc4r KO mice lack Mc4r transcript compared to their control littermates (Figure 1—figure supplement 1A), Mc4r was detectable in the PVN of all mice, which is a major site of Mc4r expression in the brain in the regulation of metabolism, therefore supporting the specific deletion of Mc4r only within Kiss1 neurons (Figure 1—figure supplement 1B).

Puberty onset was assessed daily from weaning age through the monitoring of vaginal opening (VO) and first estrus (FE). Kiss1- Mc4r KO females showed a significant advancement in the age of VO (p = 0.0150, Kiss1- Mc4r KO: 24.67 ± 0.39 vs. controls: 26.32 ± 0.53) and FE (p = 0.0341, Kiss1- Mc4r KO: 32.63 ± 1.4 vs. controls: 37.84 ± 1.90) compared to their littermate controls (Figure 1A, B). Body weight of Kiss1- Mc4r KO females at the age of puberty onset was similar between groups (p = 0.2596, Kiss1- Mc4r KO: 12.41 ± 0.30 vs. controls: 12.89 ± 0.26) (Figure 1C). These data suggest that melanocortin signaling on Kiss1 neurons participates in the timing of puberty onset in females. Since the role of the melanocortin system on puberty onset is largely unexplored, we evaluated the expression profile of the main components of this system in the hypothalami of WT female mice at post-natal days (PND) 10, 15, 22, and 30. Interestingly, expressions of Agrp, Mc3r, and Mc4r were significantly lower at the time of puberty onset (PND30) compared to earlier developmental ages (Figure 1D). This supports our in vivo data and suggests that a decrease in hypothalamic melanocortin signaling drives puberty onset, in line with the advancement in the age of VO and FE observed in Kiss1- Mc4r KO female mice (Figure 1A, B).

Because Kiss1- Mc4r KO females showed altered puberty onset, we further investigated their reproductive phenotype. Interestingly, Kiss1- Mc4r KO female mice displayed normal BW throughout the time of the study (up to PND150) (Figure 2A); however, they presented with irregular estrous cycles with predominantly more time spent in diestrus (p = 0.0023) and less time spent in estrus (p = 0.0053) than their control littermates (Figure 2B, Figure 2—figure supplement 1A, B). The assessment of the pulsatile release of LH every 10 min over 180 min revealed no change in the total number of pulses between groups but a significant increase in basal LH release (p = 0.036) in Kiss1- Mc4r KO female mice (0.42 ± 0.03) compared to controls (0.34 ± 0.02) (Figure 2C–H). The analysis of the gene expression of the ‘KNDy’ systems in the ARH, which control the GnRH pulse generator (Goodman et al., 2022), revealed normal expression levels of Kiss1, Tac2, and Tacr3, but significantly lower expression of Pdyn (p = 0.0067, Kiss1- Mc4r KO: 0.78 ± 0.038 vs. controls: 1.00 ± 0.019) (Figure 2I–L). The conserved expression of Kiss1, Tac2, and Tacr3 correlates with the preserved LH pulse frequency and amplitude in Kiss1- Mc4r KO mice, while the lower inhibitory tone of dynorphin (Figure 2I) correlates with the higher basal LH levels (Figure 2G). However, the mRNA expression of these genes, alone, does not necessarily reflect changes in their activities. To assess the contribution of MC4R signaling in Kiss1 neurons to the induction of ovulation through the preovulatory LH surge, Kiss1- Mc4r KO females and control littermates were submitted to an LH surge induction protocol that showed the induced LH surge was significantly blunted in the Kiss1- Mc4r KO females (p = 0.0091), while the protocol clearly evoked the expected afternoon rise of LH in control mice (Figure 2M). In line with these findings, the ovaries of Kiss1- Mc4r KO females displayed fewer corpora lutea, markers of recent ovulation (p = 0.0054, Kiss1- Mc4r KO: 0.80 ± 0.37 vs. controls: 2.80 ± 0.37), in addition to increased cystic follicles (p = 0.0337, Kiss1- Mc4r KO: 1.60 ± 0.40 vs. controls: 0.40 ± 0.24) (Figure 2N–P), which correlate with decreased fertility as observed by the extended time to deliver pups (i.e., longer time to get pregnant) (p = 0.0030, Kiss1- Mc4r KO: 30.30 ± 4.60 vs. controls: 20.88 ± 0.22), and fewer pups per litter (p = 0.0095, Kiss1- Mc4r KO: 6.66 ± 0.55 vs. controls: 8.44 ± 0.24) (Figure 2S, T).

The increase in serum LH levels and decreased ovulation observed in the Kiss1- Mc4r KO females is reminiscent of PCOS mouse models (Moore et al., 2013; McCarthy et al., 2022). Thus, we investigated whether Kiss1- Mc4r KO females display a PCOS-like phenotype. We analyzed circulating levels of testosterone (T) and anti-Müllerian hormone (AMH), which are frequently elevated in PCOS models (Dewailly et al., 2020). The Kiss1- Mc4r KO females expressed normal T and AMH levels compared to control mice in diestrus (Figure 2Q, R). Thus, we can exclude a PCOS-like reproductive phenotype mediated by the lack of melanocortin signaling on Kiss1 neurons.

Kiss1- Mc4r KO females displayed reproductive abnormalities resembling those described in Mc4r KO females (Chen et al., 2017; Sandrock et al., 2009; Cui et al., 2022). Thus, we hypothesized that the reproductive defects described for the Mc4r KO mice would be, at least in part, due to the absence of MC4R signaling in Kiss1 neurons. To further investigate this hypothesis, we generated mice that do not express Mc4r anywhere (Mc4r^loxTB, i.e., Mc4r KO) or that express Mc4r only in Kiss1 neurons (Kiss1^Cre: Mc4r^loxTB). The specific re-insertion of Mc4r within Kiss1 neurons in the Kiss1^Cre: Mc4r^loxTB mice was confirmed through RNAscope (Figure 3—figure supplement 1A). Mc4r expression was not detected in the PVN of these mice (Figure 3—figure supplement 1B), and it was only detected in Kiss1 neurons in the Kiss1^Cre: Mc4r^loxTB mice. Puberty onset was assessed daily from weaning age through the monitoring of VO and FE. Mc4r^loxTB and Kiss1^Cre: Mc4r^loxTB female mice displayed normal timing of puberty onset compared to their control littermates, as assessed by VO (p = 0.104, Mc4r^loxTB: 30.86 ± 1.90 vs. Mc4r^loxTB: 30.09 ± 1.00 vs. controls: 27.57 ± 0.81), and FE (p = 0.8472, Mc4r^loxTB: 35.14 ± 2.46 vs. Kiss1^Cre: Mc4r^loxTB: 34.55 ± 1.64 vs. controls: 35.80 ± 0.84) (Figure 3A, B), despite displaying significantly higher body weight at the time of puberty onset (p = 0.0001, Mc4r^loxTB: 15.49 ± 0.50 vs. Kiss1^Cre: Mc4r^loxTB: 15.54 ± 0.40 vs. Controls: 13.29 ± 0.30) (Figure 3C). As expected, Mc4r^loxTB females (Mc4r KO) displayed increased body weight (Figure 3D) and an array of reproductive impairments that included: irregular estrous cycles with significantly more days in diestrus (p = 0.0016) and fewer days in estrus (p = 0.0317) compared to controls (Figure 3E, Figure 2—figure supplement 1D, E), significantly higher area under the curve (AUC) (Mc4r^loxTB: 109.6 ± 2.39 vs. Controls: 80.36 ± 9.05) serum LH levels characterized by higher basal (Mc4r^loxTB: 0.48 ± 0.02 vs. Controls: 0.37 ± 0.06) and amplitude levels per LH pulse (Mc4r^loxTB: 0.730 ± 0.009 vs. Controls: 0.58 ± 0.07) (Figure 3F–J), and fewer corporal lutea (p = 0.0366, Mc4r^loxTB: 0.80 ± 0.80 vs. Controls: 4.20 ± 1.15) (Figure 3K, L), recapitulating the same reproductive phenotype observed in Kiss1- Mc4r KO mice despite the differences in BW. Re-introduction of MC4R into Kiss1 neurons in Kiss1^Cre: Mc4r^loxTB mice completely recovered estrous cyclicity, which was similar to controls (diestrus: p = 0.1932, proestrus: p = 0.8262, estrus: p = 0.0547, compared to controls; Figure 3E, Figure 2—figure supplement 1C, D) despite Kiss1^Cre: Mc4r^loxTB mice showing the same degree of obesity as Mc4r^loxTB mice (Figure 3D), indicating that obesity per se was not mediating the irregular estrous cycles in Mc4r KO mice. As indicated above, Mc4r^loxTB mice display higher overall circulating LH levels, and this feature was also recovered by the re-insertion of Mc4r in Kiss1 neurons of Kiss1^Cre: Mc4r^loxTB mice (basal LH: 0.32 ± 0.04, LH amplitude: 0.48 ± 0.04, LH pulses/180 min: 2.40 ± 0.67 and AUC: 70.62 ± 7.15 compared to controls; Figure 3G–J), further confirming that the melanocortin action on Kiss1 neurons is required for the normal control of gonadotropin release. Despite these significant improvements in reproductive function, Kiss1^Cre: Mc4r^loxTB mice presented fewer corpora lutea than controls (0.80 ± 0.80), similar to Mc4r^loxTB mice, suggesting that an ovulatory impairment persists (Figure 3K, L). While tonic LH release and estrous cyclicity are predominantly controlled by Kiss1^ARH neurons, ovulation is mediated by the induction of the preovulatory LH surge by Kiss1^AVPV/PeN neurons. It is possible that the metabolic signals derived from their obese phenotype, and/or the absence of the direct action of MC4R in GnRH neurons (Israel et al., 2012; Roa and Herbison, 2012) prevents the complete recovery of ovulation in Kiss1^Cre: Mc4r^loxTB mice. Both genetic models displayed significantly lower T (Mc4r^loxTB: 28.50 ± 6.20 vs. Kiss1^Cre: Mc4r^loxTB: 31.22 ± 3.90 vs. Controls: 62.84 ± 16.53) and AMH (Mc4r^loxTB: 150.6 ± 22.0 vs. Mc4r^loxTB: 146.9 ± 10.8 vs. Controls: 321.8 ± 15.7) levels than control mice in diestrus (Figure 3M, N). Therefore, we can exclude, once again, a PCOS-like reproductive phenotype mediated by the lack of melanocortin signaling.

Based on the expression of Mc4r in Kiss1 neurons and the reproductive impairment found after Mc4r deletion within Kiss1 neurons (Figure 2), we hypothesized that Kiss1^ARH neurons are direct targets of melanocortins and would respond to MC4R activation. Initially, we did whole-cell, voltage-clamp recordings in Kiss1^ARH neurons (Figure 4A). Kiss1^Crex Ai32 or Kiss1^Cre AAV-injected mice underwent ovariectomy (OVX) and were given estradiol (E2) replacement (see Methods). We targeted fluorescent cells for recording, which were isolated synaptically by bathing the slices in tetrodotoxin (TTX, 1 μM). Focal application of the high-affinity melanocortin receptor agonist melanotan II (MTII, ~250 nM) elicited a small inward current in half of the isolated Kiss1^ARH cells (10/21) (Figure 4B). Next, we examined if the E2 state affected MTII response by recording from Kiss1^ARH neurons from OVX females and found two-thirds (6/9 cells) responded, but there was no significant difference in the average current (Figure 4C). Finally, it must be noted that at this concentration, MTII is a non-selective MCR agonist, so we followed up using a perfusion of the Mc4r-selective agonist THIQ (100 nM) to determine if activation of MC4Rs alone was sufficient to invoke a postsynaptic response in Kiss1^ARH neurons from OVX + E2 females. Similar to MTII, THIQ was able to elicit an excitatory inward current in Kiss1^ARH neurons (4/5 cells, –7.4 ± 0.5 pA), indicating MC4R activation is sufficient (Figure 4D).

Based on the pharmacological activation of MCRs, we wanted to address whether POMC^ARH neurons are the source of melanocortins to excite Kiss1^ARH neurons directly. To answer this question, we injected an AAV-EF1α-DIO-ChR2:mCherry vector into the ARH of adult Pomc-Cre female mice (Figure 4E; Dewailly et al., 2020). After 2–4 weeks we did whole-cell recordings from putative Kiss1^ARH neurons and looked for a response to high-frequency optogenetic stimulation of POMC fibers in OVX + E2-treated females (Figure 4F). We were able to confirm that these were Kiss1 neurons based on the presence of a persistent sodium current and/or post hoc identification by scRT-PCR (40/77) cells, (Figure 4G). While this current is more prevalent in the AVPV Kiss1 population, Kiss1 neurons are the only ARH neurons to display this electrophysiological ‘fingerprint’ of a pronounced I_NaP paired with a high capacitance and a low input resistance (Zhang et al., 2015). Additionally, while recording from putative Kiss1^ARH neurons, we optogenetically stimulated POMC fibers at low frequency (1 Hz, 5 ms, 470 nm light) and recorded fast postsynaptic inward currents. The fast kinetics of the EPSC are also a tell-tale sign of an ionotropic glutamatergic response (Clements and Westbrook, 1991). Previously, CNQX was sufficient to block similar excitatory ESPCs in other postsynaptic targets of POMC neurons (Stincic et al., 2018). While the consistent latency from optogenetic stimulus and current response was indicative of a single synaptic delay, we wanted to establish that it was a direct synaptic connection, as we have previously shown for output of POMC neurons (Stincic et al., 2018) and Kiss1 neurons (Qiu et al., 2016; Qiu et al., 2018; Stincic et al., 2021). First, we abrogated the optogenetic response with the addition of TTX (1 μM) to the bath, and then we were able to rescue the postsynaptic glutamate response with the addition of the K⁺ channel blockers 4-aminopyridine (4-AP; 0.5 mM) and tetraethyl ammonium (TEA; 7.5 mM) (Figure 4H). K⁺ channel block enables the calcium influx from ChR2 activation in the presynaptic terminals to be sufficient to restore vesicle fusion. Therefore, there appears to be no intervening synapses between POMC and the downstream Kiss1^ARH neurons.

Previously, we found that E2 increases glutamate release from POMC neurons by increasing Slc17a6, which encodes Vglut2 (Stincic et al., 2018). Therefore, we compared the postsynaptic glutamatergic responses following two optogenetic stimuli (50-ms interval between light flashes) in Kiss1^ARH neurons from OVX + vehicle versus OVX + E2-treated female mice (Figure 4I). As anticipated, we found that treatment with E2 increased the probability of glutamate release from POMC neurons onto Kiss1^ARH neurons based on the significant decrease in the paired pulse ratio of the two stimuli (Herman et al., 2014; Figure 4J). Although in the present study we used an in vivo treatment paradigm, we know from previous studies that this augmentation of glutamate release can happen quite rapidly after a brief exposure to E2 in vitro (within 15 min; Stincic et al., 2018). Therefore, Kiss1^ARH neurons are excited by the glutamatergic input from POMC neurons in an E2-dependent manner.

Next, we tested for a melanocortin response using a high-frequency stimulation protocol known to elicit peptide release (Qiu et al., 2018). Only cells displaying a glutamatergic response were tested. Indeed, high-frequency stimulation evoked an excitatory inward current (Figure 4K) that was blocked by the selective MC3/4R antagonist SHU 9119 (Figure 4L). However, the number of high-frequency responses seen in Kiss1 neurons was unexpectedly infrequent and relatively small (4/16, 25%, mean inward current: –1.4 ± 3.1 pA) compared to the efficacy of pharmacological MCR activation. E2 treatment enhances Pomc expression and β-endorphin labeling (Thornton et al., 1994; Petersen et al., 1993), but the effect, if any, on αMSH release or the MCRs in Kiss1 neurons is unknown. We suspected that the high-frequency stimulation was eliciting co-release of αMSH and β-endorphin, which would exert opposite effects on a postsynaptic cell. In order to isolate melanocortin signaling, we pretreated slices with the non-selective opioid antagonist naloxone (1 μM) and found an increase in the likelihood and magnitude of a response to high-frequency stimulation (9/23, 40%, mean inward current: –7.6 ± 2.9 pA) (Figure 4M).

Together, these optogenetic findings reinforced our pharmacological results showing that Kiss1^ARH neurons are excited by MCR agonists. Additional studies would need to be done to identify the cation conductance that is affected by the MC4R signaling cascade in Kiss1^ARH neurons, but clearly, we have established that there are excitatory glutamatergic and peptidergic inputs from POMC to Kiss1^ARH neurons.

Having established a direct, excitatory projection from POMC to Kiss1^ARH neurons, we next examined POMC inputs to Kiss1^AVPVPeN neurons. First, we used immunocytochemistry to label fibers expressing αMSH to demonstrate that POMC fibers densely innervated the AVPV/PeN area (Figure 5A). Next, we recorded from AVPV neurons in slices taken from OVX + E2, POMC-Cre mice injected with AAV1-DIO-YFP:ChR2 into the ARH (Figure 5B). We targeted cells along the ventricle in the AVPV and PeN, surrounded by YFP terminals, and Kiss1^AVPV/PeN neurons were identified based on the expression of several endogenous conductances (I_NaP, T-type calcium current, h-current) that are unique to these neurons (Zhang et al., 2015; Zhang et al., 2013). We recorded a direct glutamatergic synaptic response following optogenetic stimulation in six neurons, which was further verified through pharmacological ‘rescue’ of the synaptic response in the majority (5/6) neurons tested (Figure 5C), indicating that POMC neurons make direct monosynaptic connections with Kiss1^AVPV/PeN neurons.

We next hypothesized that Kiss1^AVPV/PeN neurons would be excited by exogenous application of melanocortin agonists because the receptor is typically Gs-coupled (Mountjoy et al., 1994; Yu et al., 2020). We did whole-cell, voltage-clamp recordings from Kiss1^AVPV/PeN-Cre::Ai32 neurons, which were isolated synaptically by bathing the neurons in TTX (1 μM), from OVX females. For these experiments, we tested the response to the high-affinity melanocortin receptor agonist MTII (500 nM). Surprisingly, we consistently measured an outward (inhibitory) current that could be reversed on washout of MTII (Figure 5D). This outward current was associated with an increase in K⁺ conductance based on the current-voltage plot (i.e., the outward current reversed at ~E_K+, Figure 5E). Also, the inwardly rectifying I/V could indicate that MC4R is coupled to activation of inwardly rectifying K⁺ channels, as has been previously reported for MC4R signaling in pre-autonomic parasympathetic neurons in the brainstem (Sohn et al., 2013). We hypothesized that the coupling could be modulated by E2 via a Gα_q-coupled membrane estrogen receptor (Gq-mER) as we have previously demonstrated in POMC neurons (Kelly et al., 1992; Lagrange et al., 1994; Qiu et al., 2003). For these experiments, we again utilized OVX females and targeted Kiss1^AVPV/PeN-Cre::Ai32 neurons. Once in the whole-cell voltage-clamp configuration, we perfused the slices with STX (10 nM), a selective ligand for the putative Gq-mER (Qiu et al., 2003), and tested the response to the MTII. Indeed, the outward current (inhibitory) response to MTII was completely abrogated and even reversed by STX (Figure 5F, G). The short-term (bath) treatment with STX ensured that there was no desensitization of the Gq-mER with longer-term (in vivo) treatment with E2. Therefore, estrogen receptor activation can rapidly uncouple (i.e., desensitize) the MC4R inhibitory response in Kiss1^AVPV/PeN neurons. We would predict that the intracellular signaling cascade for the heterologous desensitization is similar to what we have elucidated in POMC neurons (Qiu et al., 2003), but this will need to be determined in future experiments. We next investigated the response in vehicle-treated, OVX females. Indeed, Kiss1^AVPV/PeN neurons were even more inhibited by the same MTII exposure (Figure 5G). Therefore, in contrast to Kiss1^ARH neurons, the MC4R appears to be coupled to the opening of K⁺ channels in Kiss1^AVPV/PeN neurons. To further establish whether the MTII was having pre- or postsynaptic effects, we measured the glutamatergic mEPSCs before and after melanocortin receptor activation. We found there was no significant effect on the frequency (Figure 5H), but there was an effect on the amplitude (Figure 5I). This further supports a postsynaptic locus of STX’s effects. However, post hoc comparisons found STX increased amplitude compared to both OVX and OVX + E. This result hints at a membrane-delimited effect as STX is a much more potent agonist for Gq-mER than E2 (Qiu et al., 2006).

(1) To investigate the changes in the expression of the melanocortin genes Agrp, Pomc, Mc3r, and Mc4r in the MBH during development in prepubertal and pubertal WT females at ages P10 (n = 6), P15 (n = 6), P22 (n = 6), and P30 (n = 5), and (2) evaluate the gene expression profile of Pdyn, Kiss1, Tac2, Tacr3, in the ARH of adult intact (in diestrus) female Kiss1- Mc4r KO (n = 6) and their control Mc4r^lox/lox littermates (n = 3). The brains were removed and rapidly embedded in Tissue-Tek, frozen in −30°C isopentane solution and stored at −80°C until use. Frozen tissue punches were recovered through MBH with a 1 mm diameter canula (Naulé et al., 2015). These tissue punches encompassed the whole MBH from the WT females (Figure 1D) and the ARH from the Kiss1- Mc4r KO and their control females (Figure 2I). The tissues were homogenized, and total RNA was isolated using TRIzol reagent (Invitrogen) followed by chloroform/isopropanol extraction. RNA purity and concentration were measured via an absorbance spectrophotometer (260/280 nm >1.8; NanoDrop 1000, Thermo Fisher Scientific). Total RNA (1 μg) was reverse transcribed to cDNA using random hexamers (High-Capacity cDNA Synthesis Kit, Life Technologies). Quantitative real-time PCR assays were performed using SYBR Green RT-qPCR master mix (Applied Biosystems) and analyzed using ABI Prism 7000 SDS software (Applied Biosystems). The cycling conditions were: 2-min incubation at 95°C (hot start), 45 amplification cycles (95°C for 30 s, 60°C for 30 s, and 45 s at 75°C, with fluorescence detection at the end of each cycle), followed by melting curve of the amplified products obtained by ramped increase of the temperature from 55 to 95°C to confirm the presence of single amplification product per reaction. PCR specificity was verified by melting curve analysis and agarose gel electrophoresis. Each sample was run in duplicate to obtain an average cycle threshold (CT) value, and relative expression of each target gene was determined using the comparative Ct method (Pfaffl, 2001). The data were normalized to Hypoxanthine Guanine Phosphoribosyltransferase (Hprt) expression levels in each sample. Results were expressed as fold differences in relative gene expression with respect to (1) P10 for melanocortin genes expression analysis during development in WT mice, and (2) controls for the KNDy genes expression in female Kiss1- Mc4r KO mice. The primers used are listed in Table 1.

All behavioral analyses were performed blind to the to the genotype or group. Statistical analyses were performed using GraphPad Prism. Statistical outliers were determined using GraphPad Prism. Statistical data are expressed as means ± SEM, where n represents the number of animals in each study group. The significance of differences between groups was evaluated using unpaired Student’s t-test, or a one- or two-way ANOVA test (with post hoc comparisons). Significance level was set at p < 0.05. All analyses were performed with GraphPad Prism Software, Inc (San Diego, CA).