Insulin and Leptin Signaling Interact in the Mouse Kiss1 Neuron during the Peripubertal Period

Reproduction requires adequate energy stores for parents and offspring to survive. Kiss1 neurons, which are essential for fertility, have the potential to serve as the central sensors of metabolic factors that signal to the reproductive axis the presence of stored calories. Paradoxically, obesity is often accompanied by infertility. Despite excess circulating levels of insulin and leptin, obese individuals exhibit resistance to both metabolic factors in many neuron types. Thus, resistance to insulin or leptin in Kiss1 neurons could lead to infertility. Single deletion of the receptors for either insulin or the adipokine leptin from Kiss1 neurons does not impair adult reproductive dysfunction. However, insulin and leptin signaling pathways may interact in such a way as to obscure their individual functions. We hypothesized that in the presence of genetic or obesity-induced concurrent insulin and leptin resistance, Kiss1 neurons would be unable to maintain reproductive function. We therefore induced a chronic hyperinsulinemic and hyperleptinemic state in mice lacking insulin receptors in Kiss1 neurons through high fat feeding and examined the impact on fertility. In an additional, genetic model, we ablated both leptin and insulin signaling in Kiss1 neurons (IR/LepRKiss mice). Counter to our hypothesis, we found that the addition of leptin insensitivity did not alter the reproductive phenotype of IRKiss mice. We also found that weight gain, body composition, glucose and insulin tolerance were normal in mice of both genders. Nonetheless, leptin and insulin receptor deletion altered pubertal timing as well as LH and FSH levels in mid-puberty in a reciprocal manner. Our results confirm that Kiss1 neurons do not directly mediate the critical role that insulin and leptin play in reproduction. However, during puberty kisspeptin neurons may experience a critical window of susceptibility to the influence of metabolic factors that can modify the onset of fertility.


Introduction
Obesity and hyperinsulinemia are associated with infertility [1]. Hypogonadotrophic hypogonadism, a state in which the release of hormones from all levels of the reproductive axis is reduced, is the most common reproductive consequence of type 2 diabetes and obesity [2,3]. In obese women, LH [4,5], LH pulse amplitude [6], follicle-stimulating hormone (FSH) [2,4,5], and progesterone [4][5][6] are significantly lower than in normal weight women. LH pulse amplitude is also lower in obese men [7]. Similarly, more than 25% of men with type 2 diabetes have hypothalamic hypogonadism [8]. While obesity and insulin resistance are each independently associated with lower plasma testosterone levels [9,10], obese men with T2DM appear to be at particularly high risk [11,12].
These metabolic factors have an impact on GnRH release. Obesity is associated with high levels of leptin production by adipocytes. Circulating leptin is required for maintaining GnRH release [13][14][15]. Interestingly, GnRH neurons themselves do not sense leptin, implicating upstream neurons in that role [16][17][18]. Insulin signaling also plays a role in central reproductive control. Female mice with disruption of the insulin receptor in the brain have lower LH levels and subfertility [19]. Moreover, the hypogonadism of sheep or rats lacking pancreatic insulin secretion can frequently [20][21][22][23][24] but not always [25] be ameliorated by central administration of insulin. Therefore, the preponderance of evidence suggests that central insulin signaling promotes GnRH release. While hyperinsulinemia and hyperleptinemia would therefore be expected to increase GnRH production, it does not. While the cause of low GnRH release in obese patients is unclear, insulin and leptin resistance develops in some neuronal circuits during obesity in a manner similar to insulin resistance in peripheral tissues [26][27][28][29][30]. Evidence supporting leptin resistance in neurons controlling GnRH in obese animals comes from hyperleptinemic mutant mice, which develop hypogonadotropic hypogonadism during middle age [31]. However, whether insulin resistance develops in neurons controlling GnRH release is not yet known.
It has been hypothesized that obesity acts via Kiss1 neurons to cause secondary hypogonadism [32]. Kisspeptin (a product of the Kiss1 gene) is a key hypothalamic neuropeptide that stimulates GnRH release in mice and humans [33][34][35][36][37]. Body weight, nutrition, metabolism, and hormone levels may influence the activity of Kiss1 neurons [38,39]. Tissue specific knockout studies indicate that absence of the insulin or leptin receptor in kisspeptin neurons does not alter LH levels in adult lean mice [40][41][42][43]. The function of these neurons in a chronic state of hyperinsulinemia and hyperleptinemia, however, has not been tested. Previous studies have demonstrated that intracellular signaling pathways activated by insulin and leptin overlap and may thereby exert similar effects [44][45][46][47], but it is unclear whether insulin and leptin play a cooperative role in maintaining Kiss1 neuron function. To determine the role of insulin receptors in the reproductive neuroendocrine dysfunction associated with obesity, we used a mouse model of insulin receptor (IR) deletion in kisspeptin neurons [42] and induced hyperinsulinemia with high fat feeding. Then, using transgenic KO mice with Kiss1-specific deletion of both the insulin and leptin receptor, we investigated the effects of simultaneously impaired leptin and insulin signaling in Kiss1 neurons on fertility and energy metabolism.

Animals and genotyping
To generate mice with the IR specifically deleted in Kiss1 neurons (IR Kiss mice), Kiss1-Cre mice [48] were crossed with insulin receptor floxed mice [19] and bred to homozygosity for the floxed allele only. The IR flox/flox mice were designed with loxP sites flanking exon 4. Excision of exon 4 in the presence of Cre recombinase results in a frameshift mutation and produces a premature stop codon. Littermates lacking Cre expression served as controls (IR flox/flox ). All mice were on a mixed C57BL/6J-129S6/SvEv background.
To generate mice with the IR and leptin receptor (LepR) specifically deleted in Kiss1 neurons (IR/LepR Kiss mice), Kiss1-Cre mice [48] were first crossed with insulin receptor floxed mice [19] and bred to homozygosity for the floxed allele only. These mice were then crossed with mice carrying a loxp flanked LepR to produce IR/LepR Kiss mice. Exon 17 of LepR was flanked by loxP sites, containing the Box 1 motifs crucial for leptin signal transduction through JAK2 and STAT3 pathways [49]. Cre-mediated recombination produces a LepR that lacks exon 17 and has a truncated missense exon 18, which has previously been reported to recapitulate phenotypes found in the LepR db/db1J mutation [49][50][51]. All mice were on a mixed C57BL/ 6J-129S6/SvEv background.
Mice were housed in the University of Toledo College of Medicine animal facility at 22-24°C on a 12h light/12h dark cycle and were fed standard rodent chow 2016 Teklad Global 16% Protein Rodent Diet (12% fat by calories, Harlan Laboratories, Indianapolis, Indiana). On postnatal day (PND) 22, mice were weaned. Mice were used only if litter size was within 5 to 10 pups to prevent birth size effects on body weight. At the end of the study, all animals were sacrificed by CO 2 asphyxiation or by cardiac puncture under 2% isoflurane anesthesia to draw blood. All procedures were reviewed and approved by University of Toledo College of Medicine Animal Care and Use Committee. Mice were genotyped as previously described [46,48] ( Table 1). Additional genotyping was carried out by Transnetyx, Inc. (Cordova, Tennessee) using a real-time PCR-based approach.

Electrophysiology
Kiss1-cre mice were crossed with an enhanced green fluorescent protein (EGFP) reporter strain (B6;129-Gt(ROSA)26Sortm2Sho/J; Jackson Laboratory) to create mice that produce EGFP in cre-expressing neurons. Male offspring 6-8 weeks in age were anesthetized with ketamine (80 mg/kg) and xylazine (20 mg/kg) and decapitated. The forebrain was rapidly removed into ice-cold artificial cerebrospinal fluid (ACSF) containing (in mM) 127.3 NaCl, 1.3 KCl, 1.3 KH2PO4, 21.4 NaHCO3, 5 glucose, 2.5 CaCl2 and 1.3 MgSO4. Coronal slices 250-300 μm thick containing hypothalamus were obtained using a vibrating microtome (Vibratome, St. Louis MO) and placed in an incubation beaker with room-temperature ACSF continuously perfused with 95% O2-5% CO2. Following dissection, the incubation temperature was slowly raised to the recording temperature of 32°C. Slices remained in incubation at least 1 hour before being transferred to the recording chamber and were superfused with the 0.5-2 ml/min ACSF maintained at 32°C and continuously perfused with 95% O2-5% CO2 throughout all recordings.
Viable arcuate or periventricular neurons were visualized with infrared video microscopy on an Olympus BX51W1 fixed-stage upright microscope with a 40X, 0.8 N.A. water immersion Recording pipettes were pulled from borosilicate glass capillary tubing (KG33, Garner Glass, Claremont CA) and have a final resistance of 8-12 MΩ. Recordings were deemed acceptable if Vm < -50 mV (corrected for-10 mV junction potential) and access resistance < 25 MΩ immediately after the whole-cell configuration was obtained. If needed, small amounts of hyperpolarizing holding current (typically less than 200 pA) were applied to prevent spontaneous action potentials (APs). Insulin was applied using 1-5 MΩ pressure pipettes filled with 200 nM insulin in ACSF. A sequence of 1, 2, 3 or 4 x 20 ms pressure pulses was applied during traces obtained every 10 sec and interspersed with control traces in which no pressure pulses were applied. Responses were averaged across 6-12 traces obtained from an individual cell and statistically significant changes in membrane potential produced by 1-4 insulin pulses relative to control traces without insulin application were identified by one-way analysis of variance.

Diet-induced obesity (DIO) Studies
One group of IR Kiss mice and littermate controls was placed on high fat diet with 60% kcal from fat (5.24kcal/g; Research Diets, Inc.; New Brunswick, NJ) at 8 weeks old for 5 months. During the feeding, the mice were measured for body weight every week. After 3 months of high fat diet treatment, NMR was performed to assess body composition. Fasting serum was obtained for insulin and leptin measurements, and non-fasting serum was obtained for LH, and FSH measurements. At the age of 4-5 months, mice were paired with established breeders for at least 1 month. If a mouse did not produce a new litter after 2 months of breeding, it was considered to be sterile. Date of birth of new litters and litter size were recorded if a litter was produced. A second group of IR Kiss mice and littermate controls received high fat diet from the day of weaning (PND 21). Body weight was measured every 3 days until 8 weeks old. Vaginal opening and cytology were checked every day until 8 weeks old.

Puberty and reproductive phenotype assessment
Balanopreputial separation was checked daily from weaning by manually retracting the prepuce with gentle pressure [52]. Singly housed female mice were checked daily for vaginal opening after weaning at 3 weeks of age. Vaginal lavages from female mice were collected from the day of vaginal opening for at least 3 weeks. Stages were assessed based on vaginal cytology [53,54]; predominant cornified epithelium indicated the estrous stage, predominant nucleated cells indicated the proestrus stage, and predominant leukocytes indicated the diestrus stage. At the age of 33 days, each male mouse was paired with 1 wild-type female mouse of proven fertility for 4 weeks or until the female mouse was obviously pregnant. Then the paired mice were separated, and the delivery date was recorded. The age of sexual maturation was estimated from the birth of the first litter minus average pregnancy duration for mice (20 days). At 4 to 6 months of age, animals were again paired with wild-type adult breeders to collect additional data on litter size and intervals between litters.

Metabolic phenotype assessment
Body weight was measured weekly in a single-occupant cage with ALPHA-dri bedding. Body composition in 4 month old mice was assessed by nuclear magnetic resonance (minispec mq7.5, Bruker Optics, Billerica, Massachusetts) to determine the percentage of fat mass, as previously described [55]. Glucose tolerance tests (GTT) and insulin tolerance tests (ITT) were done as previously described [46]. Briefly, following overnight fast, tail blood glucose was measured using a mouse-specific glucometer (AlphaTRAK, Abbott, Abbott Park, Illinois) before and 15, 30, 60, 90 and 120 min after dextrose (2g/kg, i.p.) injection. For ITT, following a 4 hour fast, mice were injected with recombinant insulin (0.75U/kg, i.p.). Tail blood glucose was measured again at specified time points. Food intake was measured every week until mice were 4 months old.

Quantitative real-time PCR
Mice were decapitated after isoflurane anesthesia and brains and other tissues were removed. Total RNA were extracted from dissected tissues by RNeasy Lipid Tissue Mini Kit (Qiagen) and single-strand cDNA was synthesized by High-Capacity cDNA Reverse Transcription kit (Applied Biosystems) using random hexamers as primers. 10μM cDNA template was used in a 25μl system in 96-well plates by SYBR green qPCR SuperMix/ROX (Smart Bioscience). Each sample was analyzed in triplicate to measure gene expression level. The reactions were run in ABI PRISM 7000 sequence detection system (PE Applied Biosystems) and analyzed using the comparative Ct method (2 -ΔΔCt ) with GAPDH as the normalizer.

Hormone assays
Submandibular blood was collected at 10:00-11:00AM to detect basal LH and FSH levels using the rat pituitary panel (Millipore), performed by the University of Virginia Center for Research in Reproduction. We chose this time point to avoid the LH surge in randomly cycling mice. The assay for LH had sensitivity of detection of 3.28 pg/ml. The intra-assay and inter-assay coefficients of variance (CV) were 6.9% and 17.2%, respectively. For FSH the lower limit of detection was 7.62pg/ml, with intraassay and interassay CV 6.7% and 16.9%, respectively. Serum estradiol was measured by ELISA (Calbiotech, Spring Valley, California) with sensitivity <3pg/ ml, and intra-assay CV 3.1% and interassay CV 9.9%. Serum testosterone was also measured by ELISA (Calbiotech). Serum collected after an overnight fast was used for measurement of insulin and leptin (Crystal Chem Inc. Downers Grove, Illinois). The leptin ELISA has a sensitivity range of 0.2 to 12.8ng/ml, with both intraassay and interassay CV10%. The insulin ELISA kit had a sensitivity range of 50-3200pg/ml, with both intraassay and interassay CV10%.

Perfusion and immunohistochemistry
Adult male mice and female mice at diestrus at the age of 3-6 months were deeply anesthetized by ketamine and xylazine (100 mg/kg and10 mg/kg, respectively). After briefly perfusing with a saline rinse, mice were perfused transcardially with 10% formalin for 10 minutes and the brain was removed. The brain was post-fixed in 10% formalin at 4°C overnight, and then immersed in 20% sucrose at 4°C for 48 hours. 25μm sections were cut with a sliding microtome into 5 equal serials. Sections were treated with 3% hydrogen peroxide for 30 minutes to quench endogenous peroxidase activity. After rinsing in phosphate-buffered solution (PBS), sections were blocked 2h in PBS-azide-T (PBS-azide; Triton X; 3% normal donkey serum). Then, samples were incubated for at least 48 h at 4°C in PBS-azide-T containing rabbit anti-kisspeptin IgG (1:1000, Millipore), which has been tested for specificity [46,56]. After several washes in PBS, sections were incubated in PBS-T (Triton X, 3% donkey serum) containing biotinylated anti-rabbit IgG (1:1000, Vector Laboratories, Burlingame, California), followed by incubation in ABC reagent (Vector Laboratories) for 60 minutes at room temperature. Sections were washed and immunoreactivity was visualized by 0.6mg/ml diaminobenzidine hydrochloride (Sigma) in PBS with hydrogen peroxide. Finally, sections were washed, mounted on slides, dried overnight, dehydrated, cleared and coverslipped. Kiss1-immunoreactive neurons in the AVPV/PeN and ARC were quantified as previously described [56,57].

Histology
Ovaries and testes were collected from mice and fixed in 10% formalin overnight. Tissues were embedded in paraffin and cut into 5-8μm sections by the University of Toledo Pathology Department. Sections were stained by hematoxylin and eosin.

Western blotting
Adult mice were sacrificed, and the hypothalamus, liver, muscle, visceral adipose tissues, and gonads were harvested. Tissues were snap frozen in liquid nitrogen and stored in -80°C until homogenized in radioimmunoprecipitation assay lysis buffer (Millipore, Billerica, Massachusetts) supplemented with protease inhibitor and phosphatase inhibitor (Thermo Scientific, Waltham, Massachusetts). After centrifugation, supernatant protein concentrations were determined by BCA protein assay (Thermo Scientific). 30μg or 50 μg of denatured samples were subjected to SDS-PAGE electrophoresis and western blotting using IR β subunit (1:1000, Santa Cruz Biotechnology Inc. Santa Cruz, California). β-actin (1:1000, Sigma, St Louis, Missouri) or α-tubulin (1:1000, Cell Signaling, Danvers, Massachusetts) was used as a loading control.

Statistical analysis
Data are presented as the mean ± SEM. Two-tailed unpaired t testing served as the main statistical method. Mann-Whitney U test was used if the data did not assume a normal distribution. One-way analysis of variance (ANOVA) was performed to compare three independent groups, followed by Sidak's multiple comparison test. Paired t testing was employed in the case of single repeated measures. For body weight, GTT and ITT, repeated measures ANOVAs were used to compare changes over time between two genotypes. P<0.05 was considered to be statistically significant.

Electrophysiological response to insulin
We previously reported 22% of Kiss1 neurons in the ARC and 5% of Kiss1 neurons in the AVPV expressed IR [42]. To determine if these neurons are in fact insulin-responsive, we first examined the responsiveness of Kiss1 neurons to insulin in mice that produce EGFP in cre-expressing neurons. Electrophysiological responses during current clamp recordings showed hyperpolarizing responses to pressure application of insulin in 26.6% (4 out of 15) of Kiss1 neurons from 9 animals ( Table 2).The majority of responsive neurons were located in the ARC (Fig. 1E). In responsive neurons, sequential pressure pulses of 200 nM insulin in ACSF elicited inhibitory postsynaptic potentials (IPSPs) observed in a dose-dependent manner (Fig. 1A). Other Kiss1 neurons were not responsive to insulin (Fig. 1B), nor were any Kiss1 neurons responsive to pressure application of ACSF without insulin (not shown).
Average membrane potential (Vm) changes from 15 Kiss1 neurons in hypothalamic slices obtained from 9 animals in response to pressure application of 1, 2, 3 or 4 x 20 ms pulses of 200 nM insulin. Insulin-induced Vm changes are relative to the average Vm observed during control traces without insulin application. Averages were obtained from 6-12 repetitions obtained every 10 sec. Statistically significant Vm changes produced by 1-4 insulin pulses relative to control traces without insulin application were identified by one-way ANOVA. Significant responses to insulin application were observed in the first 4 Kiss1 neurons obtained from 3 animals (P < 0.05); remaining 11 Kiss1 neurons from 6 animals did not show any significant responses (P > 0.05).
In addition to traces showing changes in resting Vm in response to pressure application of insulin (e.g. Fig. 1A and B), APs were elicited by brief (0.2 ms) depolarizing current pulses during pressure application of insulin to examine the effects of insulin on voltage-dependent responses. Outside of the superposition of APs upon insulin-evoked IPSPs (Fig. 1C), no substantial changes to the magnitude of APs were observed in any insulin-responsive neurons (Fig. 1D). APs elicited during the insulin-dependent IPSP appeared to repolarize more slowly (Fig. 1D, solid lines), which could be attributed to the reduction in voltage gated K + channel activity at the more hyperpolarized Vm produced by the IPSP.

Impact of high-fat feeding on body weight and fertility of IR Kiss mice
While insulin sensing by Kiss1 expressing cells has been reported to have minor effects on puberty and serum insulin levels [42,43], its overall physiological relevance remains unclear. To test whether Kiss1 insulin sensitivity has heightened importance in an obese, hyperinsulinemic milieu, we treated IR Kiss mice with high fat diet in both peripubertal period and adulthood. In male and female mice treated with high-fat diet (HFD) from 8 weeks old, obesity, hyperinsulinemia and hyperleptinemia developed in control and IR Kiss mice ( Fig. 2A-H). However, no difference was detected in these parameters between control and IR Kiss mice. This finding suggests that insulin secretion is normal in this model and that insulin signaling in Kiss1 neurons does not contribute to the development of obesity.
We also examined the impact of obesity on the fertility of these mice. After 3 months HFD treatment, more than half of female mice had ceased to exhibit estrous cycles (defined as constant estrus or diestrus vaginal cytology of 10 days or more) in both groups (Table 3). After 4 months HFD treatment, 60-70% female mice had stopped cycling regardless of their genotypes (not shown) and failed to become pregnant when paired with a fertile male (Table 4). However, most male mice in both groups retained their fertility (Table 4). Since in some reports, obesity can increase testosterone levels in female mice [58], we examined its levels in both sexes. Control and IR Kiss mice displayed similar estradiol and testosterone levels ( Fig. 3A  and 3B). Likewise, GnRH expression did not differ significantly in males and females of either genotype, although a trend toward lower GnRH expression was seen in IR Kiss mice (Fig. 3C). Control and IR Kiss mice displayed similar LH and FSH levels ( Fig. 3D and 3E). Finally, given our previous finding that insulin signaling in kisspeptin neurons contributes to the timing of puberty in lean mice, we examined whether exposure to HFD from the age of weaning would exacerbate this effect. HFD advanced the age of vaginal opening in all mice, although the timing of first estrus was unaffected. As previously reported, IR Kiss mice exhibited a significant delay in the timing of vaginal opening and first estrus. However, the timing of vaginal opening and first estrus was similar in both groups following high-fat feeding (Fig. 3, F-H). Therefore, (emerging) obesity can advance puberty in females without insulin sensing by kisspeptin neurons. It remains to be determined whether a stronger challenge, for example with maternal high-fat feeding, would show similar results.

Kisspeptin immunoreactivity in IR/LepR Kiss mice
These foregoing results suggest that in an obese state, insulin signaling does not alter the actions of kisspeptin neurons. However, it is possible that leptin signaling could obscure the effects of insulin signaling in some Kiss1 neurons, given their partially overlapping downstream signaling cascades [44][45][46][47]. We therefore generated mice with the IR and LepR specifically  deleted in Kiss1 neurons. To do so, we crossed LepR flox/flox mice with IR Kiss mice carrying the Cre recombinase gene driven by the Kiss1 promoter as previously described [42,59] (Fig. 4A).
To verify that floxed IR and LepR gene was excised in Kiss1 neurons, PCR was performed on DNA from different tissues. As expected, a 500bp band indicating IR and LepR gene deletion was produced from the hypothalamus and other tissues expressing Kiss1, including cerebrum, ovary and testis (Fig. 4B) [48,59,60]. IR protein levels were similar between wildtype and targeted-knockout liver, muscle, visceral fat tissue, testis and ovary (Fig. 4C). Consistent with restriction of IR inactivation to a defined subpopulation of hypothalamic neurons, western blot analysis revealed no alteration of IR expression in hypothalamus (Fig. 4C) [61]. Thus, insulin sensing appears to be intact in the majority of cells in the ovary and other tissues. Since leptin and insulin may alter neuronal development [62][63][64][65][66], we examined the number of Kiss1 neurons and fibers in IR/LepR Kiss mice. Normal patterns of kisspeptin immunoreactivity were seen in both AVPV and ARC in adult mice in both sexes (Fig. 5A, 5B). We counted the Kiss1 neuron number in the AVPV/PeN and measured the Kiss1 immunoreactive areas in the ARC from  Insulin and Leptin Signaling in the Kiss1 Neuron adult mice. There was no difference in either parameter after quantification (Fig. 5C, 5D). Thus, leptin and insulin signaling are not required for adult Kiss1 gene expression or neuron survival.

Puberty in IR/LepR Kiss mice
Since we have demonstrated a delay in puberty compared to wild-type controls after deleting IRs from Kiss1 neurons, we tested whether the additional LepR deletion would modify the pubertal phenotype of the mice. In males, we saw no alteration in the age of balanopreputial separation among the three groups, although we may have failed to detect a significant difference due to low statistical power. The age at which males were able to impregnate a female was delayed by approximately 6 days in the IR Kiss mice, while the deletion of LepRs from Kiss1 neurons resulted in an intermediate phenotype (Fig. 6A). Interestingly, the LH levels of males on postnatal day 31 was significantly decreased in IR Kiss males and restored in IR/LepR Kiss males (Fig. 6C). In contrast, FSH levels were unaffected by insulin receptor deletion, but significantly decreased in IR/LepR Kiss males (Fig. 6D). Overall, the additional deletion of leptin receptors restored LH levels, suppressed FSH levels and tended to decrease the age of puberty in males. In females, vaginal opening was delayed by nearly 3 days in IR Kiss mice, while IR/LepR Kiss females exhibited an intermediate phenotype (Fig. 6B). The first day of estrus was significantly delayed in the double knockout mice, while IR Kiss females exhibited an intermediate phenotype between the controls and IR/LepR Kiss mice (Fig. 6B). Similar to the males, IR Kiss females showed a significant reduction in LH levels that was restored in the double knockout animals Fig. 6C). While FSH levels were unaffected by deletion of insulin receptors from Kiss1 neurons, deletion of both IRs and LepRs suppressed FSH levels (Fig. 6D). Overall, the additional deletion Insulin and Leptin Signaling in the Kiss1 Neuron of leptin receptors restored LH levels, suppressed FSH levels and tended to decrease the age of puberty in females.

Adult fertility of IR/LepR Kiss mice
Adult female IR/LepR Kiss mice showed normal estrous cyclicity, characterized by normal cycle length and progression ( Fig. 7A and 7B). Female knockout mice also showed no difference in circulating levels of estradiol, LH, and FSH on diestrus (Fig. 7C-E). Female IR/LepR Kiss mice had similar ovarian weights (Fig. 7F). Despite a trend toward increased preantral follicles in the ovaries of IR/LepR Kiss mice, a comparable number of corpora lutea and follicles at all stages of maturation were found in control and IR/LepR Kiss females (Fig. 7G). For analysis of fertility, mice were paired with established wild-type male breeders. The latency to birth was comparable between the two groups (Fig. 7H). Thus, fertility appears to be normal in these animals. Male IR/LepR Kiss mice showed no significant difference in levels of testosterone, LH or FSH in adulthood (Fig. 8A-C). Furthermore, testis weight was similar to controls in adult IR/LepR Kiss mice (Fig. 8D). Testis histology showed all stages of spermatogenesis in seminiferous tubules and interstitial Leydig cells with normal morphology in IR/LepR Kiss mice (Fig. 8E). Fertility, characterized by litter size and days required to impregnate a female, was not different between control and IR/LepR Kiss males (Fig. 8F-G).

Metabolic status of IR/LepR Kiss mice
Due to the potential interaction between Kiss1 neurons and metabolic circuits in the hypothalamus [67], we examined the metabolic phenotype of IR/LepR Kiss mice. The body weight of both females and males showed no significant difference between control and knockout mice ( Fig. 9A and 9B). Likewise, average daily food intake was comparable between mice 2-3 months old of both sexes (Fig. 9C and 9D). Fat mass, lean mass and fat percentages were comparable in both females and males ( Fig. 9E and 9F). Leptin levels were also similar between the two groups (Table 5).
Since Kiss1 expressing cells in the hypothalamus, liver, or pancreas have the potential to influence glucose homeostasis, we examined relevant markers of glucose disregulation in IR/ LepR Kiss mice. We found no change in fasting glucose levels because of LepR deletion from Kiss1 neurons (Fig. 10A and 10B). Overnight fasting serum insulin was unchanged in female IR/LepR Kiss mice, but slightly higher in males (Table 5). Finally, glucose tolerance and insulin tolerance were normal in both sexes of IR/LepR Kiss mice (Fig. 10A-10D).

Discussion
Insulin appears to play a role in the early phases of puberty. The number of kisspeptin neurons in the AVPV in females increases sevenfold across puberty [68]. Insulin appears to contribute to this increase [42]. This action may involve the PI3K pathway since female mice lacking p110 isoforms in kisspeptin neurons also show lower numbers of Kiss1 neurons in the AVPV. These changes are likely to promote the establishment of normal preovulatory surges in females.
Recently, Evans and colleagues reported that only 5% of kisspeptin-immunoreactive cells in the hypothalamus showed co-labeling with IR-β [43]. This percentage is substantially lower than the 22% co-localization of IR mRNA and kisspeptin immunoreactivity we previously reported in the ARC [42]. Further, they found that a 200mU subcutaneous injection of porcine insulin to C57BL/6J mice did not cause Akt phosphorylation in Kiss1 neurons. We therefore sought further evidence of Kiss1 responsiveness to insulin. Using whole-cell recordings in the AVPV and ARC, we found that while the majority of Kiss1 neurons were unresponsive to insulin, roughly a quarter (26.6%) of cells hyperpolarized in response to a 200nM concentration of insulin. Evans and coworkers also found that high concentrations of insulin (10U, i.p.) activated all AVPV Kiss1 neurons and 17% of ARC neurons in an IR independent manner, possibly through IGF receptors. We believe that the electrophysiological responses we have observed are specific, because the concentrations used were physiological and in line with previously used concentrations inducing IR activation [69,70].
Insulin has been shown to hyperpolarize several hypothalamic cell types, including leptininhibited and leptin-excited neurons, through activation of K ATP channels [69][70][71][72]. Our results demonstrate that a subpopulation of Kiss1 neurons in the AVPV and ARC are similarly hyperpolarized in response to insulin. Recently, Qiu and coworkers reported that guinea pig Kiss1 neurons and pro-opiomelanocortin (POMC) neurons are depolarized by guinea pig insulin application [73]. It should be noted that guinea pig insulin is unique; of 51 residues, the A and B chains of guinea pig insulin differ from human insulin at 18 positions while most other mammalian insulins differ from each other at only 1-3 sites [74,75]. In addition, Kiss1 neuronal populations of guinea pigs and other rodents exhibit significant functional differences [76,77], so insulin may act differently on Kiss1 neurons in this species. Qiu and colleagues also suggested that zinc contamination of insulin preparations leads to hyperpolarization of POMC neurons in mice. This assertion was supported by experiments demonstrating hyperpolarization of mouse POMC neurons in response to zinc-containing Humulin and Novolin but not "pure" bovine or human recombinant insulin from Sigma-Aldrich. However, the zinc content of the Sigma-Aldrich formulations they used is 0.4-0.7% dry basis (Sigma-Aldrich, personal communication), comparable to the zinc content of Humulin. Using pharmacological and genetic techniques, we and others have demonstrated that insulin hyperpolarization of POMC neurons is dependent on the IR and its signaling pathways [70][71][72]78]. In these studies, we therefore attribute the hyperpolarization of Kiss1 neurons in response to insulin to a direct effect of the peptide.
Obesity involves multiple metabolic changes, including hyperleptinemia and hyperinsulinemia, which may contribute to the impaired fertility seen in the obese state. High-fat diet feeding, commonly used to induce obesity and hyperinsulinemia in animal models, is known to impair fertility. For instance, high fat feeding reduces testosterone, LH, and Kiss1 levels and inhibits LH responses to kisspeptin in male rats [79]. Others have shown that insulin signaling in the pituitary gonadotroph or ovarian theca cell contributes to infertility in female DIO mice [80,81]. We therefore tested the possibility that insulin signaling in Kiss1 neurons impacts LH secretion and fertility in the obese state. Our findings argue against insulin playing a role in the hypogonadism associated with obesity. The reproductive phenotype of IR Kiss mice on a high fat diet matched that of wild-type littermates, who were equally obese and hyperinsulinemic.
Chronic subnutrition has been shown to arrest puberty and reduce Kiss1 mRNA levels in the arcuate nucleus (ARC) in female rats [82]. Repeated central injections of kisspeptin to these rats can restore pubertal progression despite the persistent caloric restriction [83]. In support of the idea that low insulin levels may serve as one of the signals of such chronic subnutrition, IR ablation from Kiss1 neurons caused a moderate delay of puberty onset during the pubertal transition ( [42], current study). Our current findings suggest that the effects of insulin on pubertal timing are modified by the actions of leptin. While IR ablation reduced LH levels midpuberty and left FSH levels unaffected, the additional deletion of LepRs restored LH levels and suppressed FSH.
Numerous reports have provided evidence that leptin is a significant upregulator of hypothalamic Kiss1 expression [25,[83][84][85][86][87][88]. Although one study demonstrated that up to 40% of Kiss1 neurons in the ARC of male mice appear to have functional LepRs [85], subsequent studies challenged this finding by contending that LepRs were rarely colocalized with Kiss1 cell [59,89]. We found no leptin receptor expression in Kiss1 neurons in female mice on days P21-25 (several days prior to any sign of puberty) [41]. On P60-70, however, 8-15% of ARC Kiss1 neurons coexpressed the LepR, depending on estrogen levels [41,48]. However, there was no examination of when leptin receptor expression arose between these two time points. It is reasonable to think this occurred at some point during puberty. We suggest that leptin receptor expression is released from repression in the kisspeptin neuron at the initiation of puberty, as are other genes [90], allowing it to alter Kiss1 function as puberty progresses.
Single deletion and reexpression studies clearly showed that puberty can proceed without leptin signaling in Kiss1 neurons and demonstrated conclusively that mice lacking leptin fail to develop sexually due to leptin actions elsewhere in the brain [41,59]. In other words, any effects of leptin in Kiss1 neurons are not sufficient to induce puberty. The current results augment these findings by showing that leptin plays a nonessential role in modifying the timing of puberty and interacts with the actions of insulin.
It is unclear whether these effects are due to the activation of intracellular signaling pathways that alter gene expression or due to direct actions on the excitability of Kiss1 neurons. If the former, it is plausible that leptin inhibits transcription of genes that promote GnRH release (or vice versa) specifically in Kiss1 neurons, although no evidence for a candidate gene has yet been identified (e.g. [59,91]). Such an action could occur via STAT3 mediated gene transcription and be opposed by insulin signaling. Alternatively, the opposing effects of insulin and leptin on Kiss1 neuronal excitability may underlie their contrasting actions on pubertal timing. We have previously shown that leptin raises the membrane potential and increases the activity of a subset of kisspeptin neurons [41]. In contrast, the current study shows that insulin hyperpolarizes Kiss1 neurons. In both the case of leptin and insulin, their actions on Kiss1 neuron excitability have little functional relevance in adulthood, as they do not affect adult reproductive function; nevertheless, Kiss1 neurons may be particularly sensitive to these effects during the pubertal transition.
Our original assumption was that depolarization should increase kisspeptin release and hyperpolarization decrease it. However, recently an interesting alternative interpretation has been proposed [92]. While single-action potential-generated calcium influx is ideal for the release of classical neurotransmitters, burst firing promotes the release of neuropeptides [93][94][95].
The "pacemaker" current is required for rhythmic firing, and virtually all Kiss1 neurons in both the ARC and RP3V express this current [96][97][98]. The pacemaker current can only depolarizes neurons from hyperpolarized states [99,100]. Thus, inhibitory synaptic inputs may be necessary for burst firing to occur. We have demonstrated that prepubertal Arc Kiss1 neurons exhibit higher inhibitory presynaptic activity, but an action potential frequency significantly higher compared to that of Arc Kiss1 neurons recorded from adult females. Therefore, it is possible that insulin-induced hyperpolarlization promotes burst-firing while leptin inhibits it. However, it is unclear why, in this scenario, leptin receptor deletion alone did not advance puberty by increasing Kiss1 firing rates.
These finding raise the interesting possibility that these classic metabolic factors influence the frequency of GnRH pulsatile release during puberty by altering Kiss1 neuronal activity. The frequency and amplitude of GnRH pulses varies developmentally and determines the relative proportions of LH and FSH synthesis and secretion [101,102]. Increased frequency of pulsatile hypothalamic GnRH release favors LHβ gene transcription over FSHβ and increases the ratio of secreted LH to FSH, while a decreased GnRH pulse frequency favors FSHβ [101,[103][104][105][106]. The idea that leptin and insulin alter GnRH pulse frequency via actions on Kiss1 neurons during puberty remains to be tested. Despite this influence on LH and FSH secretion, neither metabolic factor is essential for puberty. Indeed, genetic factors may mask the role of insulin sensing by Kiss1 neurons during the pubertal transition [43]. In its absence, compensatory mechanisms complete the maturation of the reproductive axis. The Kiss1 neuron may sense other circulating metabolic factors with important roles in this regard [107][108][109]. Indeed, ghrelin appears to modulate arcuate Kiss1 activity in an estradiol dependent manner [108]. In addition, the Kiss1 neuron may play a key role in transmitting metabolic information from upstream neurons, such as those that release NPY, α-MSH, and melanin-concentrating hormone. These functions remain to be elucidated.
Kiss1 neurons synapse onto neurons such as POMC neurons that powerfully regulate glucose homeostasis and energy expenditure [88,110,111]. Indeed, kisspeptin receptor knockout females exhibit obesity, reduced energy expenditure and impaired glucose tolerance [112]. However, IR/LepR Kiss mice exhibited normal weight gain, body composition, and glucose and insulin tolerance. Thus, deletion of Kiss1 LepR and IRs did not alter any influence that Kiss1 neurons have on melanocortin and other neuronal circuits controlling glucose and energy regulation. Kiss1 is also reportedly expressed in several peripheral tissues, including the liver and α and β cells of the pancreas [60,[113][114][115][116]. Pancreatic α-cell targeted disruption of insulin receptor expression in mice results in glucose intolerance and hyperglycemia [117], not seen in these animals. Specific deletion of the β cell insulin receptor causes a loss of insulin secretion in response to glucose and fasting hyperinsulinemia and impaired glucose tolerance by 6 months of age. Given that we saw an isolated increase in fasting serum insulin levels in males, we conclude that if β cells expressing Kiss1 and cre recombinase exist in this line, their number is too small to greatly impact glucose homeostasis.
In summary, in the presence of genetic or obesity-induced concurrent insulin and leptin resistance, Kiss1 neurons are able to maintain reproductive function. In contrast, LepR and IR deletion in Kiss1 neurons altered pubertal timing as well as LH and FSH levels in mid-puberty in a reciprocal manner. Our results confirm that Kiss1 neurons do not directly mediate the critical role that insulin and leptin play in reproduction. However, kisspeptin neurons may experience a critical window of susceptibility to the influence of metabolic factors during puberty that can modify the onset of fertility.