|Home | About | Journals | Submit | Contact Us | Français|
Studies in humans and rodents indicate that a minimum amount of stored energy is required for normal pubertal development. The adipocyte-derived hormone leptin is a key metabolic signal to the neuroendocrine reproductive axis. Humans and mice lacking leptin or the leptin receptor (LepR) (ob/ob and db/db mice, respectively) are infertile and fail to enter puberty. Leptin administration to leptin-deficient subjects and ob/ob mice induces puberty and restores fertility, but the exact site or sites of leptin action are unclear. Here, we found that genetic deletion of LepR selectively from hypothalamic Kiss1 neurons in mice had no effect on puberty or fertility, indicating that direct leptin signaling in Kiss1 neurons is not required for these processes. However, bilateral lesions of the ventral premammillary nucleus (PMV) of ob/ob mice blunted the ability of exogenous leptin to induce sexual maturation. Moreover, unilateral reexpression of endogenous LepR in PMV neurons was sufficient to induce puberty and improve fertility in female LepR-null mice. This LepR reexpression also normalized the increased hypothalamic GnRH content characteristic of leptin-signaling deficiency. These data suggest that the PMV is a key site for leptin’s permissive action at the onset of puberty and support the hypothesis that the multiple actions of leptin to control metabolism and reproduction are anatomically dissociated.
The existence of a fundamental link between nutrition and reproduction is well established. Early studies in humans and rodents suggested that a minimum amount of stored energy is required for normal pubertal development and to maintain the tone of the reproductive system (1, 2). This concept is based on the idea that when survival is threatened by scarcity of food or increased energy demands, males and females of most species divert energy away from reproduction. This includes sexual maturation, the production of reproductive hormones and gametes, and the maintenance of pregnancy and lactation. On the other hand, excess energy may have a negative impact on the reproductive physiology. For example, elevated adiposity in women aggravates polycystic ovarian syndrome and ovulatory dysfunctions and may induce hypothalamic hypogonadism (3, 4). Moreover, the increasing rates of childhood obesity have been associated with the advance in the timing of pubertal maturation and its deleterious consequences (5–8). Earlier menarche in girls is correlated with increased risk of adult obesity, type 2 diabetes, and breast cancer (9, 10). Thus, changing levels of key metabolic cues is an essential signal for the onset of puberty. But the assessment of the mechanisms underlying puberty initiation has been obstructed by the lack of information on the brain sites in which this event is integrated.
It is now well known that the adipocyte-derived hormone leptin signals the amount of energy stored to the neuroendocrine reproductive axis. Mice lacking leptin (ob/ob) or the leptin receptor (LepR) (db/db) develop hyperphagic obesity as well as diabetes and are infertile (11–13). These mice exhibit low luteinizing hormone (LH) levels and incomplete development of reproductive organs and do not undergo puberty. Leptin administration to ob/ob mice, but not weight loss alone, induces pubertal development and maturation of reproductive organs, increases LH secretion, and restores fertility (14–16). The leptin-deficient reproductive phenotype is recapitulated in humans with monogenic forms of leptin deficiency (17). In leptin-deficient subjects, leptin treatment induces an increase in the levels of gonadotropins and sex steroids as well as enlargement of the gonads and normal pubertal development (18, 19).
LepR is expressed in many organs and tissues (13, 20), but studies using genetically engineered mouse models have determined that the brain plays a major role (21–23). Compelling evidence supports the concept that leptin acts indirectly on gonadotropin-releasing hormone–secreting (GnRH-secreting) cells via actions on leptin-responsive afferent neurons (24, 25). Thus, efforts have been made to identify the key site(s) or neuronal populations that mediate leptin’s effects on GnRH secretion.
Significant attention has been given to neurons from the arcuate nucleus (Arc), where LepR is colocalized with POMC, AgRP/NPY, and Kiss1 neurons (26–28). A number of studies have indicated that melanocortin neurons are not the key link between leptin and the reproductive axis (29–32). But, alternatively, a series of findings have suggested that Kiss1 neurons are potential candidates to mediate leptin’s effect on reproduction. Leptin-deficient ob/ob male mice show decreased expression of Kiss1 in the Arc, which is increased by leptin treatment (28). Hypothalamic Kiss1 mRNA levels are decreased in male rats made diabetic by administration of streptozotocin (33). Intracerebroventricular administration of leptin normalizes Kiss1 gene expression and the levels of LH and androgens. However, the requirement of leptin signaling on Kiss1 neurons for normal pubertal development has not been directly tested.
We recently showed that adult female rats with bilateral lesions of the ventral premammillary nucleus (PMV) are unresponsive to leptin’s effect to induce LH secretion during fasting (34). However, this experimental model was not suitable to assess whether leptin signaling in PMV neurons is required and/or sufficient for leptin’s effects on pubertal development. In the current study, we have used multiple genetic mouse models to determine the specific brain site or sites underlying leptin’s permissive action in the onset of puberty.
To determine whether LepR signaling in Kiss1 neurons is required for pubertal development, we used the Cre/loxP system to delete LepR selectively from Kiss1 neurons. Mice lacking LepR only in Kiss1 neurons were obtained by crossing Kiss1-Cre mice (line J2–4, pure C57BL/6 background) with mice homozygous for a loxP-modified Lepr allele (LepRlox/lox, mixed 129-C57BL/6 background). These mouse models have been previously verified and validated (31, 35, 36). Two cohorts of Kiss1-Cre LepRflox/flox mice (n = 13 males and n = 8 females) and LepRflox/flox littermates (n = 13 males and n = 11 females) were evaluated. In parallel, Kiss1-Cre mice (lacking the loxP-modified Lepr allele) and WT littermates (lacking both the Kiss1-Cre and the loxP-modified Lepr allele) (n = 4 per sex and per genotype) were also monitored as additional controls. Groups were defined by tail genotyping, and deletion of LepR from Kiss1 neurons was assessed at the end of the experiment. The selective deletion of LepRs from Kiss1 neurons in Kiss1-Cre LepRflox/flox mice was validated by identification of DNA recombination using a PCR strategy in which the presence of full-length and Cre-mediated deletion of the Lepr gene was distinguished (n = 4 per sex and per genotype). We found that the Lepr gene sequence flanked by loxP sites was successfully deleted from the genome only in tissues that express Kiss1, including the hypothalamus and gonads (Figure (Figure1A).1A). To further validate the loss of functional LepRs in Kiss1 neurons, mice were fasted for 24 hours and treated with i.p. leptin (5 μg/g) 45 minutes before perfusion (n = 4 per sex and per genotype). Brain sections were processed for immunohistochemistry to detect leptin-induced phosphorylation of the STAT3 (pSTAT3) as a marker for leptin signaling (37, 38). Mice used in this histological procedure were previously gonadectomized in order to intensify the expression of Kiss1 mRNA in the Arc (39, 40). The expression of functional LepR in Kiss1 neurons was determined by colocalization of pSTAT3 immunoreactivity in Kiss1 neurons identified by in situ hybridization (Figure (Figure1B).1B). Quantification of pSTAT3 immunoreactive Kiss1 neurons revealed a reduction of 90%–95% in the number of leptin-responsive Kiss1 neurons (1 or 2 dual-labeled neurons/brain) in male and female Kiss1-Cre LepRflox/flox mice compared with LepRflox/flox littermates (Figure (Figure1C). 1C).
The deleterious effect of increased adiposity on reproductive physiology is well described (4, 41, 42). Thus, given the established role of leptin in regulating energy balance (43–46), we initially assessed whether deletion of LepR from Kiss1 neurons disrupts body weight regulation. No difference in body weights between Kiss1-Cre LepRflox/flox and LepRflox/flox male (Figure (Figure1D)1D) and female mice was observed across the experiment (18 weeks). After weaning (28 days old), female mice were monitored for vaginal opening. All Kiss1-Cre LepRflox/flox (n = 8) and LepRflox/flox (n = 11) mice showed vaginal opening at 32 to 35 days of age (no difference between groups). At 60 days of age, females were housed with 1 sexually experienced C57BL/6 male and were monitored for fertility and fecundity. All mice were fertile and produced comparable litter sizes (Figure (Figure1,1, E and F). No difference in the time for pregnancy was observed between groups. At 60 days of age, male Kiss1-Cre LepRflox/flox (n = 13) and LepRflox/flox mice (n = 13) were housed with 2 FVB females of proven fertility. All males impregnated both females except for one male Kiss1-Cre LepRflox/flox that did not impregnate the females after 8 weeks (endpoint of the experiment). No apparent deficit was observed in the reproductive tract of this male, which suggests a random phenomenon, as a 10%–20% infertility rate is observed in male mice with the C57BL/6 genetic background (47). Comparison among fertile males from both genotypes showed no difference in the mean time for pregnancy and mean number of offspring (Figure (Figure1,1, E and F).
In order to assess the role of PMV neurons in leptin’s effects on pubertal development, we used an established experimental design in which puberty is induced and fertility rescued by giving leptin to the leptin-deficient ob/ob mice (14, 15). Prior to leptin treatment, we performed bilateral lesions of the PMV in ob/ob female mice. Lesions were accomplished by stereotaxic delivery of N-methyl-D-aspartate (NMDA) (0.15 M, 100 nl) bilaterally into the PMV of 16-week-old ob/ob female mice (n = 24). NMDA is effective in inducing excitotoxic neuronal lesions without affecting fibers of passage (48). We produced 5 cases with complete bilateral lesion of the PMV, classified as “PMV-lesion” cases (Figure (Figure2,2, A and B), and 12 animals with small or no lesion of the PMV, classified as “PMV-nonlesion” cases. Two rostro-to-caudal sections (125 μm interval) of the PMV, comprising the entire nucleus, were examined. We defined PMV-lesion cases as those in which fewer than 10 pSTAT3/LepR cells were identified throughout the nucleus. We also assessed possible lesions of leptin-responsive neurons in adjacent nuclei including the ventromedial nucleus of the hypothalamus (VMH), 3 rostro-to-caudal levels of the Arc, and the dorsomedial nucleus of the hypothalamus (DMH). Small or virtually no lesion of adjacent nuclei was noticed. Additionally, specifically for the Arc, we evaluated the putative lesion of Kiss1 neurons. Mice classified as PMV-lesion showed numbers of Kiss1 neurons and expression of Kiss1 mRNA/cell comparable to those classified as PMV-nonlesion (Figure (Figure2,2, C–E).
We assessed food intake and body weight in PMV-lesion and PMV-nonlesion mice before and after surgery and following leptin treatment. Initial body weight and food intake were comparable between groups. After surgery, mice from both groups exhibited an expected and transient decrease in body weight (Figure (Figure2F).2F). After full surgical recovery (18 days), weight gain was restored in both groups and the leptin regimen was initiated. Leptin induced comparable changes in body weight in both groups (Figure (Figure2F).2F). After 3 weeks of leptin treatment, all mice had achieved body weights comparable to age-matched WT (26.56 ± 1.6 g, injected animals vs. 25.22 ± 0.88 g, WT). At this point, the daily dose of leptin was adjusted to 2 μg/g. Only a transient difference in body weight between groups was noticed during the fifth week of leptin treatment. Lesions of the PMV caused no changes in food intake. Likewise, both groups showed similar decreases in food intake in response to leptin administration (Figure (Figure2G).2G). These comparable changes in body weight and food intake also produced a comparable change in body composition (Figure (Figure2H).2H). Our findings indicate that the PMV is not required for leptin’s effects on food intake and body weight in ob/ob female mice.
Leptin induced vaginal opening in all females during the first week of treatment. After 3 weeks on the leptin regimen, a sexually experienced male was introduced in the cage to test the progression of sexual maturation and the ability of the mice to copulate. Mice were monitored daily for the presence of copulatory plugs, as a sign of sexual maturity (49). After 3 weeks of mating (6 weeks of leptin treatment), most of the control PMV-nonlesion mice had shown copulatory plugs (67%; Figure Figure3A).3A). This was is in agreement with previous studies that reported copulatory plugs after 4-6 weeks of leptin treatment (15). However, no PMV-lesion mice had shown copulatory plugs at the same time point. Following an additional period, PMV-lesioned mice displayed a significant delay for the presence of copulatory plugs (35.8 ± 3.1 days) compared with nonlesioned mice (Figure (Figure3,3, A and B). Of note, 1 PMV-lesioned mouse failed to show a copulatory plug after 6 weeks of mating (endpoint). Because the sizes of PMV lesions were variable and it is not possible to determine a priori the number of neurons that are necessary to maintain a specific function, we used general regression models to assess whether the time for copulatory plugs was correlated with either pSTAT3-immunoreactivity in the PMV (a measure of the number of leptin-responsive neurons remaining in the lesioned PMV), pSTAT3-immunoreactivity in the Arc or number of Kiss1 neurons in the Arc. Notably, we found that when other putative lesions are included in the multivariate model, only lesion of the PMV shows a significant coefficient (Figure (Figure3C).3C). Our results indicate that PMV neurons are required for leptin-induced sexual maturation of ob/ob female mice.
We also assessed the ability of leptin to induce LH secretion in leptin-deficient ob/ob female mice with lesions of the PMV. After detecting the copulatory plugs, leptin regimen was ended. Following 24 to 36 hours — the period after which low to undetectable levels of circulating leptin are achieved in ob/ob mice (49) — PMV-lesion and PMV-nonlesion mice received a bolus of leptin i.p. (2.5 μg/g) at 3–5 pm and blood was collected 40 minutes later, before perfusion. A group of intact ob/ob female mice (n = 4) was treated with saline to serve as an additional control. Following acute leptin administration, control PMV-nonlesion ob/ob mice displayed higher levels of LH compared with intact saline-treated ob/ob mice (Figure (Figure3D).3D). However, in mice with complete bilateral lesions of the PMV, leptin treatment failed to induce LH secretion above that observed for saline-treated ob/ob mice. The PMV-lesioned mice also showed low levels of progesterone compared with nonlesioned mice (Figure (Figure3E). 3E).
In order to determine whether leptin signaling in PMV neurons is sufficient for leptin’s effects on puberty, we employed the Flp/FRT system to selectively restore LepR in PMV neurons in mice otherwise null for LepRs. The LepRneo/neo mice are homozygous for a neomycin cassette flanked by FRT sites targeted to the Lepr locus (35, 50). These mice recapitulate the LepR-deficient phenotypes and therefore are obese, diabetic, and infertile. We performed unilateral stereotaxic injection of an adeno-associated virus vector expressing Flp recombinase (AAV-Flp) into the PMV of adult LepRneo/neo male and female mice. The viral vector is an AAV-Flp-IRES-eGFP construct designed so that Flp and GFP are driven by CMV regulatory elements, allowing the visualization of the injection sites using GFP immunoreactivity (50).
Prior to brain injections, LepRneo/neo male (n = 21) and female (n = 23) mice and WT littermates (n = 6 males and n = 6 females) were housed with opposite sex, sexually experienced WT C57BL/6 mice, starting at 4–5 weeks of age. After 6 weeks, none of the LepRneo/neo male mice showed fertility and none of the female mice displayed progressive vaginal opening at 67–72 days of age. In comparison, all WT littermates were sexually mature and fertile at the same age. The AAV-Flp injections were performed in adult (67–72 days old) LepRneo/neo mice (n = 21 males and n = 23 females) and sexually naive WT littermates (n = 4 males and n = 6 females). Additional control groups consisted of LepRneo/neo mice that had received stereotaxic injections of AAV-GFP into the PMV.
In order to identify the exact sites of functional LepR reactivation, before sacrifice all animals were fasted for 24 hours and treated with leptin i.p. Sites of reactivation were identified by the presence of GFP and pSTAT3 immunoreactive neurons. We produced 11 males and 10 females with injections of AAV-Flp centered in the PMV (classified as PMV-hits; Figure Figure4).4). In 10 males and 13 females, the injections were not located in the PMV but in adjacent nuclei including the DMH (n = 2 males and n = 3 females), the Arc (n = 3 males and n = 4 females), and surrounding regions of the hypothalamus that do not express LepR (n = 5 males and n = 6 females). These mice were classified as PMV-misses and were considered anatomic controls.
As described in previous experiments, we also assessed food intake and body weight in PMV-hits compared with PMV-misses and WT littermates. The PMV-misses with contamination of the Arc were not included in our analysis, as these mice showed the expected decrease in food intake and body weight (50, 51). We found that selective reexpression of LepRs in the PMV (PMV-hits, n = 11 males and n = 7 females) induced no changes in food intake and body weight compared with PMV-misses (Figure (Figure5,5, A and F). In this analysis, PMV-misses consisted of mice with AAV-Flp injected in areas that do not express LepR (n = 5 males and n = 6 females). PMV-misses with contamination of the DMH were analyzed separately, as the role played by leptin-responsive cells in the DMH is not fully understood. We also found no difference between PMV-hits and DMH-injected mice (n = 3 males and n = 3 females).
Ten weeks following AAV-Flp injections, the female mice with AAV-GFP injections in the PMV (n = 3) or those with AAV-Flp injections that missed the PMV displayed virtually no or very small and rounded vaginal opening. In marked contrast, we found that 9 of 10 PMV-hits displayed an unambiguous and progressive vaginal opening. The specific timing of vaginal opening varied from case to case. Two mice showed vaginal opening at 3 weeks after injection, 3 mice at 4 weeks, 2 mice at 5 weeks, and 2 mice at 6 weeks. The vaginal cytology assessed for 2 weeks also varied from case to case. For example, 2 mice showed a persistent diestrus (10–12 days), 2 mice a persistent estrus (8–10 days), and the others showed variable and inconsistent cycles. All PMV-reactivated mice (PMV-hits) showed at least 2 episodes of vaginal cornification during the period analyzed (2 weeks), indicating progression of sexual maturation (52).
To further characterize the sexual maturity and changes in fertility induced by endogenous LepR reexpression in the PMV, females were housed with sexually experienced C57BL/6 male mice. Remarkably, during a period of 6 weeks, we found that 5 of the mice classified as PMV-hits became pregnant. These mice displayed an increase in body weight and abdominal dilation 4–5 weeks after the introduction of the male in the cage. However, 4 of these 5 cases did not carry pups to term. One case successfully delivered pups (5 in the litter) that died soon after birth without milk in their stomachs. Three of the mice that did not deliver pups were perfused as soon as consistent weight loss due to pregnancy failure was observed. We confirmed pregnancy by collecting the uteri and analyzing the occurrence of implantations (Figure (Figure6A).6A). The number of implantations in these cases was comparable to that observed in pregnant WT C57BL/6 females (6 to 8).
In order to assess changes in the maturation of reproductive organs, we additionally collected the uteri and ovaries of intact LepRneo/neo female mice (n = 6, 18 to 20 weeks of age) and of WT littermates on diestrus (n = 7, 18–20 weeks of age). As observed in leptin-signaling–deficient mice (14–15, 53, 54), LepRneo/neo mice displayed significantly lower uterine weight compared with WT (0.0259 ± 0.003 g vs. 0.074 ± 0.006 g, P < 0.0001). Subsequently, in order to assess whether PMV-hits that displayed vaginal opening but did not become pregnant showed any progress in sexual maturation, we compared their uterine weight with that of intact LepRneo/neo and PMV-misses. We found no difference in uterine weight between intact LepRneo/neo and PMV-misses. However, we observed that PMV-hits exhibited a significant increase in uterine weight compared with intact LepRneo/neo and PMV-misses (P = 0.0252; Figure Figure6B).6B). The uteri of PMV-hits also showed increased cross-sectional area (227500 ± 14250 vs. 94880 ± 15230 μm2, P = 0.0017), parenchyma size (206.0 ± 35.46 vs. 82.86 ± 6.237 μm, P = 0.010), and epithelial length (1334 ± 155.8 vs. 720.6 ± 119.4 μm, P = 0.0136; Figure Figure6C)6C) compared with PMV-misses. The ovaries of PMV-hits and PMV-misses showed a similar number of primary, secondary, and antral follicles. However, we found that only the ovaries of PMV-hits contained Graafian follicles and corpora lutea (Figure (Figure6D),6D), suggesting follicular maturation and the occurrence of ovulation. As mentioned, in order to identify the reactivated neurons, all mice were fasted for 24 hours before perfusion. Moreover, of 9 PMV-hits, 3 were perfused during miscarriage (to confirm implantations), 1 was perfused 2 days after giving birth, 3 were perfused on diestrus, and 2 on estrus (endpoint defined by time after surgery). Consequently their hormonal profiles are not comparable. However, the increased uterus size and the presence of corpora lutea in the ovaries of PMV-hits are both reliable evidence for the increase in circulating levels of estrogen and LH, respectively (49, 55, 56).
The LepRneo/neo male mice were also monitored for fertility. After 4 weeks of injections, males were housed with 2 sexually competent WT FVB females and were perfused 6 weeks later. We found that 1 in 11 PMV-hits (9%) impregnated 1 female (after 4 weeks of mating) and none of the PMV-misses were fertile. However, we found that 1 in 2 control mice with AAV-GFP injected into the PMV was also fertile (impregnated 1 female after 4 weeks of mating). In order to assess differences in the maturation of the reproductive tract, we additionally collected the seminal vesicles and testes of intact LepRneo/neo (n = 9, 18 to 20 weeks of age) and of WT C57BL/6 mice (n = 5, 18–20 weeks of age). We found no difference in seminal vesicle weight (P = 0.4957; Supplemental Figure 1A; supplemental material available online with this article; doi: 10.1172/JCI45106DS1). Additionally, compared with WT, LepRneo/neo showed only a tendency for a decrease in testis weight (P = 0.0613; Supplemental Figure 1B). We also compared the seminal gland and testis weight of the intact LepRneo/neo, the PMV-hits, and the PMV-misses. We found no differences comparing all groups (P = 0.3878 for seminal vesicle weight and P = 0.7045 for testis weight; Supplemental Figure 1, C and D). We then assessed testis histology and found no identifiable morphological changes comparing PMV-misses, PMV-hits, and WT (Supplemental Figure 1, E–J).
As mentioned, in order to identify the sites of reactivation, the PMV-hits, PMV-misses, and WT littermates were perfused after fasting, a condition known to decrease gonadotropins and testosterone levels (57, 58). In agreement, mice displayed comparable low levels of testosterone (WT: 29.48 ± 9.9 ng/dl; PMV-misses: 38.47 ± 5.9 ng/dl; PMV-hits: 34.52 ± 5.1 ng/dl, P = 0.7060), except for 1 fertile PMV-hit which showed high testosterone level (279.87 ng/dl). It is well reported that a small percentage of leptin-deficient male mice are fertile (59). Therefore, our findings indicate that no clear or significant improvement of the reproductive physiology of the LepR-null male mice was attained following reexpression of LepR in the PMV.
An increase in excitatory inputs and/or in sensitivity of glutamate receptors of GnRH neurons is a key event at the onset of puberty and at the time of the preovulatory LH surge (60–63). Thus, in order to determine whether PMV neurons coexpress LepR and glutamate, we performed colocalization studies in a mouse model carrying a LepR-dependent reporter gene. Mice expressing Cre under the control of the endogenous Lepr promoter (LepRb-IRES-Cre mice) (64) were crossed to mice carrying a Cre-inducible LacZ allele. In our reporter mice (LepR-LacZ), virtually all β-gal immunoreactive neurons in the PMV expressed pSTAT3 following leptin administration (38). We found that 93.9% ± 0.5% of PMV neurons that express β-gal immunoreactivity coexpress vesicular glutamate transporters (vGluT2) mRNA (Figure (Figure7,7, A and B).
To assess whether PMV neurons innervate GnRH terminals and/or Kiss1 neurons, we injected male and female C57BL/6 mice with the anterograde tracer biotin dextran amine (BDA) (Figure (Figure7C).7C). Using a high-resolution microscope with a Z-stack module allowing 3D reconstruction of 1-μm optical sections (ApoTome Microscope, AxioVision software; Zeiss), we observed PMV fibers in close apposition with GnRH terminals in the median eminence/mediobasal hypothalamus and with Kiss1 neurons in the AVPV (Figure (Figure7,7, D–F).
Because the PMV densely innervates the AVPV and apparently Kiss1 neurons, we assessed whether the rescue of the reproductive phenotype of LepR-null mice by reexpression of LepR in the PMV is consequent to changes in Kiss1 gene expression. We initially assessed whether Kiss1 mRNA expression is different, comparing C57BL/6 WT (n = 4 females, 18 to 20 weeks of age) and intact LepRneo/neo female mice (n = 4 females, 18 to 20 weeks of age). We found no difference in Kiss1 mRNA (total area and individual cells) or in the number of Kiss1 neurons in the female AVPV (Supplemental Figure 2, A–C). In the female Arc, we did not find differences in total Kiss1 mRNA or the number of Kiss1 neurons (Supplemental Figure 2, E and G). However, as shown in leptin-deficient ob/ob male mice (28), we did observe a decrease in Kiss1 mRNA/cell in the Arc of LepRneo/neo female mice (P = 0.0065; Supplemental Figure 3F). We then assessed whether reactivation of LepR in the PMV caused changes in Kiss1 mRNA/cell. We found no difference in Kiss1 mRNA /cell in the AVPV or the Arc comparing PMV-misses and PMV-hits (Supplemental Figure 3). In males, only the Arc was assessed due to the sexually dimorphic nature of Kiss1 in the AVPV (n = 4, WT and n = 5, LepRneo/neo) (39, 40, 65). No difference in Kiss1 gene expression (total area or individual cells) or number of cells was detected in LepRneo/neo male mice compared with WT male mice (Supplemental Figure 2, H–J) or between PMV-misses and PMV-hits. Our findings indicate that leptin action on PMV neurons to rescue the reproductive deficits of LepR-null mice is independent of changes in Kiss1 gene expression.
To further test our model, we assessed changes in the GnRH neuronal system in LepR-null and reactivated mice. We initially assessed possible differences in localization and number of GnRH neurons as well as GnRH gene expression comparing C57BL/6 WT (n = 4 males and n = 4 females, 18 to 20 weeks of age) and intact LepRneo/neo male and female mice (n = 5 males and n = 4 females, 18 to 20 weeks of age). In LepRneo/neo male and female mice, as in WT, GnRH neurons were distributed in the medial septal nucleus, in the preoptic area, surrounding the vascular organ of lamina terminalis, and scattered in the ventral aspect of the anterior hypothalamic area. No apparent deficit in GnRH neuron migration was noticed in LepRneo/neo mice. We also found no changes in GnRH mRNA expression (Figure (Figure8A8A and Supplemental Figure 4A) or number of GnRH neurons (males: 740.0 ± 15.00 neurons in WT vs. 778.8 ± 25.36 neurons in LepRneo/neo, P = 0.3782; females: 775.0 ± 52.95 neurons in WT vs. 669.0 ± 49.31 neurons in LepRneo/neo, P = 0.1833, in 1 series of sections). However, we found increased GnRH content (observed via immunoreactivity) in the median eminence/mediobasal hypothalamus of LepRneo/neo female mice compared with WT on diestrus (Figure (Figure8,8, B–D). No difference in GnRH content between male LepRneo/neo and WT mice was noticed (Supplemental Figure 4, B–D).
We then assessed whether reactivation of LepR in PMV neurons induced changes in GnRH expression or peptide content. Because states of negative energy balance may increase GnRH immunoreactivity (66), we compared PMV-misses, PMV-hits, and WT littermates. As mentioned, these mice were fasted for 24 hours before perfusion. We found no changes in the number or in the amount of GnRH mRNA/cell of females comparing all groups (Figure (Figure8,8, E and F). However, PMV-misses displayed increased GnRH immunoreactivity at the median eminence/mediobasal hypothalamus compared with WT (P = 0.0166; Figure Figure8,8, G–I). Remarkably, PMV-hits showed a comparable amount of GnRH immunoreactivity to WT (Figure (Figure8,8, G–M), suggesting that the reactivation of LepR selectively in PMV neurons stimulates GnRH release and, therefore, normalizes the GnRH content in the median eminence/mediobasal hypothalamus of female LepR-null mice.
It has been known for decades that nutritional state affects puberty and the reproductive status in many species, including humans (1, 2). Sexual maturation is usually initiated when energy stores meet the demands of the reproductive function including the production of hormones and gametes, copulation, pregnancy, and lactation. If excessive leanness occurs in young women, puberty is often delayed (2). Moreover, the existence of a link between the advance of childhood obesity and the increasing rates of precocious puberty has also been postulated. A previous study published in 1997 reported that, in the USA, 6.7% of girls had clinical evidence of puberty at age 7 years and 14.7% at age 8 years. Authors also described the youngest population age at puberty onset as 9.9 ± 1.8 years (8). In the last decade, an increment of 5%–8% in the number of girls with clinical evidence of puberty at age 7 and 8 years has been reported (5). The cause for this alarming phenomenon is not quite understood. One possibility is the increased levels of circulating leptin in obese children. Increased adiposity is often correlated to high leptin levels (67, 68), a condition known to accelerate pubertal development (69, 70). However, the sites of leptin action on puberty initiation and therefore the mechanisms by which this physiological event is regulated have been difficult to determine.
The role played by the Kiss1/Kiss1r system in puberty and fertility has been well described. Loss-of-function mutations of Kiss1 or Kiss1r genes in rodents and humans result in hypogonadism, abnormal sexual maturation, and decreased circulating levels of sex steroids and gonadotropins (71–74). Administration of Kisspeptin to juvenile rodents induces vaginal opening, increases LH secretion, and induces ovulation (75, 76). Hypothalamic levels of Kiss1 and Kiss1r mRNAs and electrophysiological responses of GnRH neurons to Kisspeptin increase across puberty (76, 77). Collectively, these findings indicate that Kiss1 neurons play an essential role in pubertal development, but what triggers the Kiss1/Kiss1r system is still unsettled. Leptin has been a plausible candidate, and therefore, leptin-responsive Kiss1 neurons would be well positioned to relay leptin’s effect on puberty initiation. To test this model, we generated mice with selective deletion of LepR from Kiss1 neurons and assessed their sexual maturation. We found that male and female mice progress through puberty normally and are fertile, suggesting that leptin signaling in Kiss1 neurons is not required for pubertal development and fertility. Studies from different laboratories have shown that kisspeptin induces LH secretion via direct stimulation of GnRH neurons (75, 78–81). But the estrogen-induced LH surge still occurs in Kiss1r-knockout mice, suggesting the existence of an additional relevant pathway in the regulation of GnRH-induced LH secretion (82). Our findings are in agreement with this concept and indicate that leptin’s effect in the onset of puberty is relayed by an alternative route outside Kiss1 neurons. Nevertheless, the physiologic relevance of direct leptin signaling in Kiss1 neurons in different paradigms is yet to be determined, as developmental adaptations may have occurred in our mouse model.
The role played by PMV neurons in mediating leptin’s effects has been largely ignored. The PMV neurons express sex steroid receptors (83, 84) and are activated following copulation or exposure to sexually relevant odors (85–91), but its role in pubertal development had not been anticipated. In fact, the lack of mouse models in which to selectively manipulate LepR in the PMV represents an additional difficulty in assessing the physiological role played by PMV neurons. In this study, we employed stereotaxic techniques to lesion PMV neurons of ob/ob mice or to reactivate endogenous expression of LepR in PMV neurons of LepR-null mice. These techniques inherently produce heterogeneous experimental subjects due to variable injection sizes and injection sites. Moreover, in order to precisely determine the functional reactivation of LepR or lesion of neurons that express LepR, mice were fasted to potentiate leptin-induced pSTAT3-ir. This precluded a systematic analysis of their hormonal profile and evaluation of differential reproductive parameters. However, our results are unambiguous. We identified the PMV as a key site of leptin action in the onset of puberty. Reexpression of LepR in adjacent sites outside the PMV did not trigger pubertal development in LepR-null mice. Likewise, partial lesion of the PMV or of neurons outside the PMV did not blunt leptin’s effect of inducing sexual maturity in ob/ob female mice.
Previous studies have suggested that leptin plays a small or virtually no role in the progress of pregnancy (92). Therefore, we established the copulatory plugs as the endpoint for the lesion experiment. In this experimental design, we also assessed leptin’s effect on LH secretion. Leptin treatment prevents the fall in LH levels during states of negative energy balance, when leptin levels are low (34, 58, 93–96). As previously demonstrated (14, 54), all intact ob/ob female mice exhibited low levels of LH, which were increased following leptin administration. However, leptin was ineffective at increasing LH levels in PMV-lesioned ob/ob mice. These findings confirm and extend previous data (34) indicating that PMV is also required for leptin’s stimulatory effect on LH secretion in leptin-deficient ob/ob mice. During proestrus, the increase in LH levels also stimulates progesterone production, which in turn is essential for copulatory behavior (97, 98). Following copulation, female rodents display a sustained increase in progesterone as a result of neuroendocrine reflex generated by vaginal-cervical stimulation (99). Therefore, the low levels of progesterone observed in PMV-lesioned mice may be a consequence of the decreased LH secretion and/or may indicate a deficient copulatory behavior in these mice.
Selective reexpression of LepR in PMV neurons induced puberty and sexual maturation in a high percentage of LepR-null female mice. Remarkably, this effect was independent of changes in body weight or food intake. Thus, endogenous reexpression of LepR in the PMV did not correct the obese phenotype and likely did not rescue most of the endocrine dysfunctions of LepR-null mice. This is worth emphasizing, as most of the pregnant females underwent miscarriages. It is now well documented that obesity as well as diabetes may increase the probability of miscarriage and malformations in rodents and humans (100–102). Therefore, the miscarriages and the absence of pregnancy in some of the reactivated mice may be caused by the remaining metabolic or endocrine abnormalities following reactivation of LepR in PMV neurons of LepR-null mice.
Deficits in gonadal development of leptin-deficient ob/ob male mice are well described (14, 16, 54). In contrast, and in agreement with our findings, the absence of a clear morphological abnormality in the reproductive tract of LepR-deficient male mice has also been reported by others (53, 103). But intriguingly, reexpression of LepR in the PMV of LepR-null mice did not ameliorate male fertility. This may indicate that the sites of leptin action in the reproductive physiology of rodents are sexually dimorphic. The generation of mouse models to specifically manipulate genes of interest in PMV neurons will allow further evaluation of the contribution of leptin signaling in PMV neurons to male sexual maturation and also to female pregnancy.
In the present study, we demonstrate that leptin’s effect in pubertal development is relayed by PMV neurons expressing the excitatory neurotransmitter glutamate. Leptin directly depolarizes PMV neurons (89), which may, in turn, “activate” their downstream neuronal targets, including GnRH cell bodies in the preoptic area (89, 104, 105) and/or GnRH terminals in the median eminence/mediobasal hypothalamus. This model is in agreement with a recent study that reported that glutamate stimulates GnRH secretion in the absence of the Kiss1/Kiss1r system (106). Thus, our findings indicate that PMV neurons may serve as an alternative pathway in this circuitry, conveying inputs on changing levels of leptin directly to GnRH neurons. As observed in db/db and ob/ob mice (53, 54), LepR-null mice displayed no deficits in GnRH neuronal migration, GnRH gene expression, or peptide production. We also noticed that LepR-null mice show modest or undetectable changes in Kiss1 gene expression compared with control littermates. Together, these data suggest that the infertility phenotype of the LepR-null mice is not caused by deficient expression of Gnrh or Kiss1 genes. Rather, the leptin-signaling deficient mice display increased GnRH content in the median eminence/mediobasal hypothalamus, strengthening the model that GnRH secretion is inhibited in females with leptin-signaling deficiency (53, 54). This elevated GnRH content is not observed in the infertile Kiss1 gene knockout mice (73, 74), suggesting that the disruption in the normal pubertal development observed in Kiss1 and in leptin-signaling deficiencies is caused by dissociated mechanisms. Importantly, reexpression of LepR in PMV neurons of LepR-null mice normalizes their GnRH content in the median eminence/mediobasal hypothalamus. This is in concert with the well-established concept acquired from studies conducted in vivo and in vitro, in different species and sexes, that leptin acts in the brain stimulating GnRH secretion (107–111). Although the bulk of evidence collected in the present study favors a direct action of PMV neurons in GnRH terminals, we cannot rule out the existence of additional and/or alternative relays. In many physiological systems and, in particular, in the reproductive function, redundant pathways are essential for species survival. Here, we demonstrate that a direct action of leptin on PMV neurons is required and sufficient for leptin’s effect in the onset of puberty. But it is not clear whether glutamatergic PMV neurons exert their role by directly stimulating GnRH secretion or by acting in an additional relay (e.g., kisspeptin neurons) that impinges on GnRH neurons. In any case, the hypothalamic circuitry (i.e., GnRH and Kiss1 neurons) must be in place to provide the underlying substrate for leptin’s effect as a “permissive” factor in the onset of puberty.
The role played by leptin as a permissive factor for the onset of puberty is well determined. But the efforts to dissect the effects of leptin and the mechanisms by which puberty is initiated have been obstructed by the lack of information on where in the brain this effect takes place. Our study establishes the PMV as a key site for leptin action in the onset of puberty and demonstrates that direct leptin signaling in Kiss1 neurons is not required for pubertal development and fertility. Our findings also add to the growing literature suggesting that leptin’s multiple actions to control metabolism and reproduction are anatomically dissociated.
Adult male and female Kiss1-Cre, LepRflox/flox, LepRneo/neo, LepR-IRES-Cre, β-gal reporter/B6.129S4-Gt(ROSA)26–Sortm1Sor/J (Jackson Labs), and C57BL/6 mice and female ob/ob mice were housed in the University of Texas Southwestern Medical Center Facility. All animal studies were approved by the University of Texas Southwestern Medical Center Institutional Animal Care and Use Committee, Dallas, Texas, USA.
We performed stereotaxic injection of BDA unilaterally in the PMV of WT mice, of NMDA (0.15 M, Sigma-Aldrich) bilaterally into the PMV of ob/ob female mice, and of adeno-associated virus vector (AAV-Flp or AAV-GFP) unilaterally into the PMV of male and female LepR-null mice. Mice were housed individually after surgery. Before sacrifice, AAV-Flp–injected mice, AAV-GFP–injected mice, and NMDA-injected mice were fasted for 24 hours and received i.p. leptin 40 minutes before perfusion (provided by A.F. Parlow, Harbor-UCLA Medical Center, Los Angeles, California, USA; through the National Hormone and Peptide Program). Brain sections were submitted to immunohistochemistry to detect pSTAT3 as a marker for leptin signaling.
Mice were deeply anesthetized with ketamine (5 mg/100 g) and xylazine (1 mg/100 g). Before perfusion or tissue collection, blood samples were collected directly from the heart. Hormone assays were performed by The University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core. Brain and reproductive organs were collected and post-fixed in buffered formalin. Brains were cut (25 μm sections, 5 series) in a freezing microtome and stored at –20°C. Uteri, ovaries, and testes were processed for standard paraffin-embedded sectioning and H&E staining. Brain sections were processed for the detection of the reporter gene (β-gal; Aves Lab), pSTAT3 (Cell Signaling Technology), Kisspeptin (Millipore), and GnRH (LHRH; Immunostar) immunoreactivity. Standard immunoperoxidase using DAB as chromogen as well as standard immunofluorescence using Alexa Fluor fluorophores–conjugated secondary antisera (Invitrogen) was used.
Brain sections were mounted onto SuperFrost plus slides (Fisher Scientific) and pretreated in 0.1 M citric acid, pH 6.0, and microwave, as described (38). The riboprobes (GnRH, Kiss1, LepR, and vGluT2) were generated by in vitro transcription with 35S-UTP. The 35S-labeled probes were diluted (106 dpm/ml) in a hybridization solution that was applied to each slide, and sections were incubated overnight at 56°C. Sections were submitted to stringency washes and dehydrated. Hybridization signal was detected following standard autoradiographic protocol. The specificity of GnRH, Kiss1, LepR, and vGluT2 probes was described before (34, 38, 112).
Hypothalamic sections containing the PMV of LepR-IRES-Cre LacZ mice (n = 3 males and n = 3 females) were submitted to in situ hybridization (vGluT2 riboprobes) followed by a standard immunoperoxidase for the detection of β-gal immunoreactivity (Aves Labs) using DAB as chromogen. The procedure is the same described in previous publications (38, 90). Slides were then processed for standard autoradiographic methods.
The hybridization signal (Kiss1 and GnRH) and the GnRH immunoreactivity were estimated by the analysis of integrated optical density (IOD) using ImageJ software ( http://rsb.info.nih.gov/ij). The IOD values for Kiss1 mRNA and GnRH immunoreactivity were calculated as the total IOD of a constant area subtracting the background. The IOD values for GnRH mRNA/cell and Kiss1 mRNA/cell were calculated as the total IOD of each GnRH neuron surrounding the vascular organ of the lamina terminalis and each Kiss1 neuron on one side of one representative section of the AVPV and Arc. We considered a neuron GnRH positive only if the cell had IOD values at least 3× higher than that of background. Because Kiss1 expression is lower compared with GnRH, we considered a neuron Kiss1 positive only if the cell had IOD values at least 2× higher than that of background. The number of GnRH neurons was obtained from 1 series of sections extending from the medial septal nucleus to the preoptic area. The number of Kiss1 neurons was obtained from 1 representative section of the AVPV and of the Arc.
Mice were decapitated, brains were removed, and the hypothalamus was divided in 2 blocks: 1 containing the preoptic area and medial septum and 1 containing the Arc. After RNA extraction using the TRIzol reagent (Invitrogen), 2 μg of total RNA was incubated in DNase I (Roche Diagnostics) for 30 minutes at 37°C. Subsequently, the cDNA was generated using random hexamers (Roche Diagnostics) and SuperScript II Reverse Transcriptase (Invitrogen) during incubation at 42°C for 50 minutes. We used the SYBR Green PCR master mix (Applied Biosystems) for quantitative real-time PCR analysis, and the assays were performed using an Applied Biosystems Prism 7900HT sequence detection system. The mRNA contents were normalized to β-actin content (forward: 5′-CATCGTGGGCCGCTCTA; reverse: 5′-CACCCACATAGGAGTCCTTCTG), and the resulting values were expressed as fold change above control levels. The primers for GnRH mRNA quantification were 5′-CCACTGGCCCCGTTCAC (forward) and 5′-GCTTCCTCTTCAATCAGACTTTCC (reverse). The primers for Kiss1 mRNA quantification were 5′-GGCAAAAGTGAAGCCTGGAT (forward) and 5′-GATTCCTTTTCCCAGGCATT (reverse).
Sections of brain, uterus, ovary, and testis were analyzed in a Zeiss Axioplan or ApoTome microscope (Zeiss). Photomicrographs were produced by capturing images with a digital camera (Axiocam; Zeiss) mounted directly on the microscope. Image-editing software (Adobe Photoshop CS3) was used to combine photomicrographs into plates. Only the sharpness, contrast, and brightness were adjusted. Cytoarchitectonic details were added by using a camera lucida.
Data are expressed as mean ± SEM. Comparison between 2 groups was carried out using the unpaired 2-tailed Student’s t test. One-way ANOVA followed by the pairwise Tukey test were used to compare 3 or more groups simultaneously. Statistical analysis was performed using GraphPad Prism software, and an α value of 0.05 was considered in all analyses. The net effect of the lesions of PMV neurons and adjacent sites (pSTAT3 and Kiss1 neurons in the Arc) on the copulatory plugs was evaluated by General Regression Models, using the software Statistica v.9 (Statsoft). Results were expressed as Pareto chart of t values for coefficients.
We thank N. Dunning, M. Choi, S. Osborne-Lawrence, and S. Rovinsky for technical support. This work was supported by the NIH (R01HD061539 to C.F. Elias; R01MH61583 and RL1DK081185 to J.K. Elmquist; K08DK068069 and R01DA024680 to J.M. Zigman), Capes (to L.O. Margatho), and Fapesp (05/58997-4 to J. Donato Jr.; 05/59286-4 to C.F. Elias). C.F. Elias is a Distinguished Scholar in Medical Research (University of Texas Southwestern Medical Center [UTSW]). Histology of the reproductive organs was provided by the UTSW Molecular Pathology Core. Hormone assays were performed by The University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core, supported by the Eunice Shriver NICHD/NIH (SCCPIR) (U54-HD28934). We thank Aktar Ali and Laura Brule and the Mouse Metabolic Phenotyping Core at UTSW (supported by PL1 DK081182-01 and UL1RR024923).
Conflict of interest: The authors have declared that no conflict of interest exists.
Citation for this article: J Clin Invest. 2011;121(1):355–368. doi:10.1172/JCI45106.
See the related Commentary beginning on page 34.