SIRT1 mediates obesity- and nutrient-dependent perturbation of pubertal timing by epigenetically controlling Kiss1 expression

Puberty is regulated by epigenetic mechanisms and is highly sensitive to metabolic and nutritional cues. However, the epigenetic pathways mediating the effects of nutrition and obesity on pubertal timing are unknown. Here, we identify Sirtuin 1 (SIRT1), a fuel-sensing deacetylase, as a molecule that restrains female puberty via epigenetic repression of the puberty-activating gene, Kiss1. SIRT1 is expressed in hypothalamic Kiss1 neurons and suppresses Kiss1 expression. SIRT1 interacts with the Polycomb silencing complex to decrease Kiss1 promoter activity. As puberty approaches, SIRT1 is evicted from the Kiss1 promoter facilitating a repressive-to-permissive switch in chromatin landscape. Early-onset overnutrition accelerates these changes, enhances Kiss1 expression and advances puberty. In contrast, undernutrition raises SIRT1 levels, protracts Kiss1 repression and delays puberty. This delay is mimicked by central pharmacological activation of SIRT1 or SIRT1 overexpression, achieved via transgenesis or virogenetic targeting to the ARC. Our results identify SIRT1-mediated inhibition of Kiss1 as key epigenetic mechanism by which nutritional cues and obesity influence mammalian puberty.

S exual maturity and reproductive capacity are attained at puberty 1 . Understanding the intimate mechanisms controlling this process has gained new urgency in view of the observations that a secular trend towards an earlier puberty appear to be re-initiated 2,3 . This is worrisome because early puberty in humans is associated with a number of undesirable outcomes, such as cardiovascular disease, obesity, insulin resistance, hypertension, diabetes type 2, increased incidence of breast cancer, increased susceptibility to mental illness and behavioral disorders, and lower adult height [4][5][6] .
Although puberty is under strong genetic determination 7 , its timing is modulated by endogenous and exogenous cues 1 . Among those, nutritional and metabolic signals are known to play very prominent regulatory roles 8,9 . Consequently, conditions of metabolic stress, ranging from malnutrition to early-onset obesity, are linked to perturbed pubertal timing 10 ; child obesity appears to be particularly relevant because of its association with early puberty worldwide 2,11 . Yet, despite the importance of these findings, the mechanisms conveying nutritional status information to the cellular networks controlling puberty remain ill defined.
The pubertal process depends on changes in the secretory activity of hypothalamic neurons producing gonadotropinreleasing hormone (GnRH). Because GnRH neurons themselves appear to be devoid of the appropriate sensing mechanisms 10 , nutrition-dependent regulation must instead operate within cellular networks functionally connected to GnRH neurons. One such network is a group of neurons known as KNDy neurons because they produce Kisspeptin, NKB (neurokinin B) and Dynorphin [12][13][14] . They are located in the arcuate nucleus (ARC) of the hypothalamus 13,14 , and drive the changes in GnRH secretion that set in motion the endocrine manifestations of puberty 15 . Because both Kiss1 (encoding kisspeptin) and Tac3 (encoding NKB) are essential for puberty to occur 16,17 , KNDy neurons appear to be uniquely posed to serve as nodal portals for nutritional cues to influence reproductive development. In rodents, another population of neurons expressing Kiss1, but not Tac3, is located in the rostral hypothalamus, mainly at the anteroventral periventricular nucleus (AVPV) 18 . However, its role in the nutritional modulation of puberty has not been established.
Sirtuins are the homologs of the ageing-related factor, Silent Information Regulator 2 (Sir2), discovered in Saccharomyces cerevisiae [19][20][21] . Among sirtuins, SIRT1, with abundant expression in the brain and different metabolic peripheral tissues, is the most extensively studied 19 . SIRT1 operates as NAD + -dependent deacetylase 22 , acting on histones and other cellular targets to conduct a wide array of biological functions, including epigenetic control and modulation of life/health span [19][20][21] . Sirtuins act as bona fide cell energy sensors: activation of sirtuins is tightly coupled to changes in the availability of the metabolic cofactor, NAD + , and the related intermediates, NADH and nicotinamide, so that conditions as caloric restriction and nutrient deprivation, which increase the NAD + /NADH or NAD + /nicotinamide ratios, result in increased SIRT1 content and activation in various tissues 23,24 . CNS mapping studies in adult mice revealed abundant expression of Sirt1 in key hypothalamic nuclei controlling neuro-vegetative functions, including the ARC; central SIRT1 activity seemingly plays a key role in controlling longevity 25 , and might mediate the effects of caloric restriction on life/health span in mammals 20,24 .
The nutritional status of an individual has strong environmental promptings, suggesting that nutrition may influence reproductive development via epigenetic mechanisms. Indeed, KNDy neurons are subjected to a repressive epigenetic control, imposed by the Polycomb (PcG) silencing complex that prevents the premature unleashing of the pubertal process 26 . This repressive influence is counter-balanced by the Trithorax group (TrxG), which operate as a central element of the activating epigenetic machinery that offsets PcG actions 27 . Yet, little is known about putative epigenetic regulatory mechanisms conveying nutritional information to KNDy neurons. Because SIRT1 -as energy-sensing molecule that allows the cell to respond to both reduction and increases in nutrient availability 19,24 -is activated in the adult hypothalamus by decreased nutrient availability 28 and is abundant in the ARC [28][29][30] , we explored the notion that SIRT1 is a key component of the epigenetic machinery that regulates the timing of puberty by conveying nutritional information to KNDy neurons. We also evaluated the potential pathophysiological role that SIRT1 may play in eliciting pubertal perturbations associated with early-onset obesity and undernutrition.
Our present results support an essential role of SIRT1 as a central hub of an epigenetic mechanism mediating the effect of nutritional cues on female puberty, through regulation of Kiss1 transcription. Our findings show that SIRT1 is expressed in hypothalamic Kiss1 neurons and suppresses Kiss1 expression, via interaction with the Polycomb silencing complex. As puberty approaches, SIRT1 is evicted from the Kiss1 promoter facilitating a switch in local chromatin configuration from repressive to permissive. As evidence for their translational relevance, these changes are accelerated by early-onset obesity, which induces precocious puberty, and postponed by undernutrition, which delays puberty. The pivotal role of SIRT1 on the latter phenomenon is evidenced by the ability of central pharmacological activation of SIRT1 or SIRT1 overexpression, either globally or targeted to the ARC, to mimic the pubertal delay evoked by undernutrition.

Results
Developmental and nutritional changes in hypothalamic SIRT1 during puberty. SIRT1 content decreased in the preoptic area (POA)-medial basal hypothalamus (MBH) unit analyzed in toto before and during the initiation of female puberty. This change, already apparent by PND10 ( Supplementary Figure 1), became firmly established by the end of juvenile development (PND28), and reached prominence between the end of juvenile development (PND28) and the peripubertal period (PND 36; Fig. 1a). The decrease was also apparent when the MBH and POA were analyzed separately, although it was proportionally greater in the MBH (>65%) than in the POA (Fig. 1b). The drop in hypothalamic SIRT1 content that occurs between PND10 and PND36 coincided with a significant increase in Kiss1 expression levels, both in the POA and MBH (Fig. 1c). Tac3 mRNA content in the MBH also increased during this period (Supplementary Figure 2a).
The relationship that exists between nutritional status and hypothalamic SIRT1 was evidenced by a reduction in SIRT1 abundance observed at PND29 in overnourished (ON) animals (Fig. 1d). This change was associated with a significant elevation in both Kiss1 and Tac3 mRNA levels ( Fig. 1e and Supplementary  Figure 2b). ON rats also displayed increased body weight (Fig. 1f) and early puberty, as assessed by the age at vaginal opening (28.1 ± 0.4 d vs. normally fed (NN) controls, 32 ± 0.3 d, p < 0.01; Student's t test) and at first ovulation, two well-established signs of puberty. Thus, while >80% of obese animals exhibited vaginal opening (Fig. 1g) and corpora lutea (CL) (Fig. 1h) at PND29 (p < 0.001, X 2 test), none of the normally fed controls had signs of puberty at this age (Fig. 1g, h). Both the uterus (index of estrogen stimulation; Fig. 1i) and serum LH levels ( Fig. 1j; reflecting hypothalamic-pituitary activity) were significantly greater in obese than in normally fed rats. These results show that overnutrition instituted during early postnatal life reduces hypothalamic SIRT1 content and causes precocious puberty.
In contrast, undernourishment (UN) imposed at the beginning of the juvenile period resulted in increased hypothalamic SIRT1 content and decreased Kiss1 expression, assessed at PND36 (Fig. 2a, b); a non-ignificant decrease in Tac3 levels was also observed (Supplementary Figure 2c). UN also caused a reduction in body weight (Fig. 2c), delayed vaginal opening (Fig. 2d), and ovulatory failure (Fig. 2e). Both uterine weight (Fig. 2f) and LH levels ( Fig. 2g) were reduced in underfed animals. Thus, prepubertal undernutrition increases SIRT1 content and lowers Kiss1 expression in the hypothalamus, and delays female puberty.
Functional manipulation of hypothalamic SIRT1 alters puberty. To determine if the inverse correlation that exists between hypothalamic SIRT1 and the time of female puberty involves a causal relationship, we employed three different approaches. First, we studied transgenic mice overexpressing SIRT1 31,32 (SIRT1-Tg). Second, we used the allosteric SIRT1-activator, SA3, to stimulate centrally endogenous SIRT1 activity 33 . Third, we used a virogenetic approach to selectively increase SIRT1 expression in the ARC of the hypothalamus. SIRT1-Tg mice had normal body weight ( Fig. 3a) and displayed elevated SIRT1 content in the hypothalamus (Fig. 3b), coupled to a decrease in the abundance of histone 3 acetylated at lysines 9/14 (H3K9/14Ac) (Fig. 3c), likely reflecting an increase in SIRT1-dependent deacetylase activity. In addition, SIRT1-Tg mice showed a significant reduction in hypothalamic Kiss1 mRNA levels (Fig. 3d), but not Tac3 mRNA (Supplementary Figure 2d). Importantly, SIRT1-Tg animals had a strikingly delayed vaginal opening (32.7 ± 1.3 d vs. WT, 28.3 ± 0.7 d, p = 0.008; Student's t test). By PND33, all WT mice had vaginal opening, in contrast to only 35% of the SIRT-Tg mice (p < 0.0001; X 2 test) (Fig. 3e); none of the Tg mice had ovulated at this time (Fig. 3f). Similar to undernourished rats, SIRT1-Tg animals had reduced LH levels at puberty (Fig. 3g). Thus, as initially suggested by others 34 , our data conclusively document that transgenic overexpression of SIRT1 delays reproductive maturation.
We next injected SA3 into the lateral ventricle of immature female rats daily between PND26 and PND35. Like SIRT1-Tg mice, SA3-treated rats had a normal body weight (Fig. 4a), but delayed vaginal opening. At PND33, only 20% of SA3-injected animals had vaginal opening in contrast to 75% of vehicleinjected rats (p = 0.06; X 2 test; Fig. 4b). At the time of tissue collection (PND35), 40% of SA3-treated rats had not ovulated  In panels b-f, i and j, *p < 0.05; **p < 0.01; ***p < 0.001 (two-sided Student's t test). For protein analyses in panels a, b and d, three representative bands per group, run in the same original western blots, are presented. The scale bar in panel h corresponds to 600 μm. Total group sizes were: NN = 11 and ON = 12; while phenotypic and hormonal parameters were assayed in the whole groups, hypothalamic protein/RNA (d, e) and ovarian histological (h) analyses were conducted in a representative subset of randomly assigned samples from each group, with the following distribution: d, e NN = 5; ON = 6−8; h n = 6 in each group (Fig. 4c). Mimicking the phenotype of SIRT1-Tg mice, SA3injected rats had lighter uteri (Fig. 4d) and reduced serum LH levels (Fig. 4e). In addition, hypothalamic SIRT1 content was increased (Fig. 4f), and the abundance of H3K9/14Ac was reduced ( Fig. 4g), likely due to increased SIRT1-dependent deacetylase activity. Noteworthy, SA3-treated rats had reduced Kiss1 expression in the MBH (Fig. 4h), but unaltered Tac3 levels (Supplementary Figure 2e). Given the location of KNDy neurons in the ARC, we used a virogenetic approach to target SIRT1 overexpression to this hypothalamic region. Stereotaxic-guided delivery of an AAV vector expressing SIRT1 resulted in increased SIRT1 content (Fig. 5e), and decreased ratios of acetylated K9/14 H3 vs. total H3 (AcH3/H3) in the MBH (Fig. 5e). Congruent with the results observed using SIRT1-Tg mice and rats centrally injected with SA3, increasing SIRT1 abundance in the ARC resulted in delayed vaginal opening; while >90% of rats injected with control AAV had vaginal opening by PND33, <40% of AAV-SIRT1 injected rats displayed VO at this age (p < 0.01; X 2 test; Fig. 5f). Likewise, AAV-SIRT1 rats had a markedly delayed first estrus (Fig. 5g) and stunted follicular maturation, with 60% of the animals failing to ovulate at PND36 (Fig. 5h). Pubertal female rats injected with AAV-SIRT1 displayed also a significant decrease in ovarian weight (Fig. 5i). Thus, by selectively enhancing SIRT1 abundance in the ARC (and hence, preventing the drop of expression that occurs before puberty), female reproductive maturation was delayed.
SIRT1 is expressed in Kiss1 neurons where it is modulated by nutritional cues. Double fluorescence in situ hybridization (FISH) and single cell (sc)-PCR demonstrated that KNDy neurons in the MBH/ARC do express Sirt1 mRNA (Fig. 6a, b). Detailed quantification demonstrated that~50% of all Kiss1expressing neurons in the ARC express Sirt1 (Fig. 6b). We confirmed this colocalization using a double immunohistofluorescence approach. In fact, in agreement with our previous data (see Figs. 1 and 2), SIRT1 content in Kiss1 neurons was decreased in pubertal overfed ON rats (Fig. 6c), while it was increased in undernourished UN rats (Fig. 6d). Altogether, these data support the concept that prepubertal nutrient manipulation can affect KNDy neuron activity (as evidenced by changes in Kiss1 expression) via changes in SIRT1 that occur within KNDy neurons themselves.
SIRT1 evokes a repressive histone configuration at Kiss1 promoter that abates at puberty. Undernourishment and obesity/ overnutrition affected hypothalamic Kiss1 expression at puberty in opposite directions: the former decreasing Kiss1 mRNA levels, the latter increasing gene expression. To determine if these effects are directly exerted on Kiss1 neurons, we interrogated the Kiss1 promoter, using ChIP-qPCR assays, targeting a promoter region involved in the epigenetic control of Kiss1 expression 26 . We found that SIRT1 is evicted from the Kiss1 promoter at the completion of puberty (PND36) (Fig. 7a). This loss takes place earlier (at PND29) in overfed animals, but fails to occur in undernourished rats (Fig. 7a). Two activating histone modifications (H3K9ac and H4K16ac), normally removed by SIRT1mediated deacetylation, showed a pattern of abundance at the Kiss1 promoter opposite to that of SIRT1. While the content of both marks increases during the pubertal transition, this increase were: NN = 20 and ON = 10; while phenotypic and hormonal parameters were assayed in the whole groups (in the case of LH levels, for all serum samples that were available), hypothalamic protein (a) and RNA (b), as well as ovarian histological (e) analyses were conducted in a representative subset of randomly assigned samples from each group, with the following distribution: a n = 5; b n = 6−8; e n = 5−6 determinations occurs prematurely in overfed animals and is averted in undernourished rats (Fig. 7a). In contrast, the content of H3K27me3, a repressive histone modification catalyzed by the PcG complex 35 , paralleled that of SIRT1 at the Kiss1 promoter, decreasing at the completion of normal puberty, diminishing earlier in obese rats, and increasing at the promoter of underfed rats (Fig. 7a). Because the Trithorax group (TrxG) of proteins is the main epigenetic force counterbalancing the repressive actions of the PcG complex 27,36 , we determined if the abundance of H3K4me3, an activating histone mark catalyzed by TrxG 27,36 , changes at Kiss1 promoter in response to nutritional manipulations. While H3K4me3 content, measured at the end of juvenile development (PND29), was greatly increased in obese rats, undernutrition did not significantly alter H3K4me3 levels at either PND29 or PND36 (Fig. 7a). This suggests that TrxG influence might be more prominent in conditions of energy surplus. The genomic specificity of the changes described above was documented by ChIP assays targeting intron 2 of Kiss1. As shown in Fig. 7b, none of the changes in histone mark abundance observed at the Kiss1 promoter occurred at intron 2 of this gene.
Although pubertal expression of Tac3 was only mildly influenced by the nutritional status, we explored SIRT1-related interactions with the Tac3 promoter, a genomic region recently shown to be subjected to epigenetic control at puberty 27 . SIRT1 association to the Tac3 promoter decreased between late juvenile development and the completion of puberty, but neither obesity nor undernourishment altered this association (Supplementary Figure 3a). Similarly, nutritional manipulations did not change the H3K9ac and H4K16ac content at the Tac3 promoter on PND29, but underfeeding reduced the content of H3K9ac mark on PND36 (Supplementary Figure 3a). The abundance of the repressive histone mark, H3K27me3, at the Tac3 promoter increased mildly at PND36 in underfed rats, without any changes in the content of H3K4me3 (Supplementary Figure 3a). Finally, neither normal puberty nor nutritional manipulations altered the association of SIRT1 to, or the histone configuration at, the Pdyn promoter (Supplementary Figure 3b). Thus, it appears that SIRT1 prevents the premature activation of KNDy neurons in immature animals by mostly repressing Kiss1 activity, and that earlier or late puberty in over-and underfed animals, respectively, correlate with changes in SIRT1 binding and activity mainly at the Kiss1 promoter. It is also clear that these modifications do not occur in the POA, which contains the AVPV; the only changes in this region were a tendency for a decrease of SIRT1 association to the Kiss1 promoter and an increase in the content of the activating histone mark, H4K16ac, observed in overfed ON rats (Supplementary Figure 4). SIRT1 cooperates with EED to repress Kiss1 promoter activity. The parallel loss of SIRT1 and H3K27me3 association to the Kiss1 promoter elicited by pubertal maturation and overnutrition suggests that SIRT1 and the PcG complex act in a coordinated fashion to repress Kiss1 activity in KNDy neurons of prepubertal animals. Promoter assays revealed that Kiss1 promoter activity is modestly reduced by either SIRT1 or EED, a protein required for PcG silencing activity 37 , but is more clearly inhibited by the simultaneous presence of both proteins (Fig. 8a). Coimmunoprecipitation studies demonstrated that SIRT1 and EED interact physically (Fig. 8b). When SIRT1 was overexpressed in a rat (R22) immortalized hypothalamic cell line (Fig. 8c), there was increased interaction of SIRT1 protein to the Kiss1 promoter WT Tg ). This was accompanied by decreased content of the activating mark H3K9ac and the TrxG-dependent mark H3K4me3, and increased abundance of the PcG-catalyzed mark H3K27me3 (Fig. 8f). Concomitantly, EED recruitment to the promoter rose (Fig. 8g). These results indicate that SIRT1 and EED act together at the Kiss1 promoter in KNDy neurons to turn a permissive histone landscape into a repressive histone configuration, resulting in decreased Kiss1 transcription.

Discussion
Our understanding of the molecular mechanisms responsible for the precise timing of puberty has expanded recently, due in part to the recognition of the role of epigenetic mechanisms in the control of normal pubertal maturation 26,38 . However, characterization of the repertoire of epigenetic regulatory molecules, especially as it pertains to the nutritional gating of puberty and its deregulation in conditions of metabolic stress, remains incomplete. In principle, molecules that use epigenetic mechanisms to link metabolic activity with neuroendocrine reproductive control should have four basic attributes that enable them to perform these functions: (a) to be able to engage directly with the basic machinery of epigenetic regulation; (b) to respond to alterations in cellular metabolism with meaningful changes in expression and/or activity; (c) to sense nutrient-dependent changes in the activity of nodal components of relevant metabolic pathways; and (d) to modify the activity of genes directly involved in the control of GnRH release. Our results conclusively identify SIRT1, the mammalian ortholog of Sir2 19,24 , as a regulatory molecule that meets all of these criteria, and unveil a central regulatory hub, involving SIRT1-mediated transcriptional repression of Kiss1 in KNDy neurons of the ARC, as an essential mechanism regulating the timing and metabolic gating of puberty. The inhibitory influence of central SIRT1 signaling on pubertal timing was disclosed by a combination of physiological and molecular approaches applied to various preclinical models. Not only the hypothalamic content of SIRT1 declines during postnatal/pubertal maturation, but translational models of precocious and delayed puberty, caused by early over-or undernutrition of female rats, display coherent changes in hypothalamic SIRT1 levels: advanced puberty is accompanied by decreased hypothalamic SIRT1 content, whereas delayed puberty associates with increased SIRT1 levels. The causal nature of these associations is demonstrated by our studies involving functional manipulation of SIRT1 levels/activity. Thus, SIRT1 overexpression by either transgenesis, using a validated mouse model 31,32 , or directed to the ARC via an AAV-mediated, stereotaxic-guided targeting approach, caused a delay in pubertal progression, reminiscent of that caused by undernutrition. A similar effect occurred following the pharmacological enhancement of SIRT1 activity in the brain. The changes in hypothalamic SIRT1 abundance caused by these experimental manipulations were related to inverse changes in Kiss1 expression in vivo, suggesting a key role for Kiss1 as putative conduit for the regulatory actions of SIRT1 on pubertal timing.
KNDy neurons in the ARC are a well-conserved, indispensable component for the control of pulsatile secretion of GnRH and gonadotropins 39 , and therefore the timing of puberty in mammals. In line with this key role, our expression and functional   For protein analyses in panels f and g, three representative bands per group, run in the same original western blots are presented. The scale bar in panel c corresponds to 600 μm. Total group sizes were: C = 10 and SA3 = 10; while phenotypic and hormonal parameters were assayed in the whole groups (in the case of LH levels, for all serum samples that were available), hypothalamic protein (f, g) and RNA (h) analyses were conducted in a representative subset of randomly assigned samples from each group, with the following distribution: C: n = 6−8 (protein and RNA, respectively); SA3: n = 6 analyses demonstrate that SIRT1 targets this neuronal population to metabolically gate pubertal timing, mainly by repressing Kiss1 transcription. This contention is supported by the demonstration that (a) SIRT1 (mRNA and protein) is expressed in KNDy neurons, with a steady-state average of >50% of KNDy neurons expressing Sirt1; (b) Kiss1 expression is suppressed in the MBH of pubertal female rats after transgenic overexpression and pharmacological activation of SIRT1 in vivo; (c) SIRT1 reduces Kiss1 expression in vitro; and (d) SIRT1 content in ARC KNDy neurons and SIRT1 recruitment to the Kiss1 promoter in vivo change according to the nutritional status of the animal; while overnutrition decreases neuronal content and promoter association, undernutrition increases both parameters. In addition, (e) virogenetic overexpression of SIRT1 in the ARC delays puberty onset. Although the potential contribution of SIRT1 to regulating the activity of other hypothalamic cell types remains a distinct possibility, our findings strongly suggest that SIRT1 is a bona fide energy sensor that operates in KNDy neurons to transduce metabolic information into changes of Kiss1 transcriptional activity. Importantly, proper progression through normal puberty appears to also require the timed eviction of SIRT1 from the Kiss1 promoter, as a permissive requisite for the upsurge of hypothalamic Kiss1 expression that occurs during the pubertal transition 18 .
The molecular underpinnings of this phenomenon have been also disclosed by our work. Eviction of SIRT1 from the Kiss1 promoter during puberty is associated with increased abundance of activating histone modifications (H3K9ac, H4K16ac, H3K4me3), and a decrease in the repressive histone mark H3K27me3 catalyzed by the PcG complex 35 . These changes appear to occur selectively in ARC KNDy neurons, since similar modifications were not detected in the AVPV region, which   18 . Moreover, our in vitro analyses demonstrate that SIRT1 recruits the PcG member, EED, to the Kiss1 promoter, where it enhances the repressive action of SIRT1 on Kiss1 transcription. Therefore, SIRT1 appears to operate as central epigenetic link between energy status and Kiss1 expression in KNDy neurons, by changing the histone landscape of the Kiss1 promoter. This molecular switch from a repressive to a permissive histone configuration is seemingly dictated by the nutritional influence on SIRT1 activity at the Kiss1 promoter: conditions of energy excess, which induce precocious puberty, accelerate the removal of the epigenetic brake imposed by SIRT1/ PcG on Kiss1 expression. In contrast, persistent energy deficit, which causes pubertal delay, prolongs this repressive influence (Fig. 9). Our results pave the way for the identification of upstream regulators of hypothalamic SIRT1 that may, thereby, modulate pubertal timing. Besides nutritional cues, these might include metabolic hormones with known roles in puberty control, such as ghrelin, which has been shown to increase hypothalamic SIRT1 activity to regulate energy homeostasis in adulthood 40 , and to delay puberty onset 41 . Regulation of SIRT1 in KNDy neurons by ghrelin and/or other metabolic transmitters (e.g., adipokines), which are likely deregulated in some of our preclinical models of pubertal perturbation (e.g., early over-and undernutrition), as well as the almost certain contribution of epigenetic factors, warrants future investigation.
In the last decades, SIRT1 has been recognized as master molecular link (and potential druggable target) for the integration of metabolic sensing and key developmental and neuroendocrine phenomena [19][20][21] . Interestingly, genetic loss-of-function studies in mice had previously suggested a developmental role of SIRT1 in early stages of GnRH neuron maturation; mice with congenital Sirt1 deficiency display hypogonadotropic hypogonadism (HH) due to defective neuronal migration 42,43 . Our work unveils another dimension of SIRT1 biology in the hypothalamic control of neuroendocrine function. Hitherto known to be an essential factor for the control of longevity and healthy ageing 21,44 , SIRT1 appears to also function as fundamental epigenetic conduit that links obesity and nutritional status with changes in pubertal timing. This might not be the only puberty-related function of SIRT1 in the developing brain. Because the SIRT1 locus in humans is strongly associated with susceptibility to major  45 , and an increase in SIRT1 abundance in certain brain areas, such as the nucleus Accumbens, induces depressivelike behaviors in mice 46 , failure of brain SIRT1 expression to decrease in a timely fashion during the adolescence transition to adulthood might contribute to the well-known exacerbation of depressive disorders at puberty 47 , some of which occur in the context of obesity 48 .

Methods
Animals. We used Wistar female rats bred in the vivarium of the University of Cordoba and housed under constant conditions of light (14/10 h light/dark cycle) and temperature (22 ± 1°C), with free access to tap water and standard laboratory chow, unless otherwise stated (see models of nutritional stress). For experiments involving functional manipulations, either nutritional, pharmacologic or virogenetic, the animals were randomly assigned to the different experimental groups. We also used transgenic female mice overexpressing Sirt1 (Sirt1-Tg; kindly provided by Dr. Manuel Serrano, Institute for Research in Biomedicine (IRB), Barcelona, Spain). These animals were generated as previously described 31 New Jersey) to induce early-onset obesity 50 . NL rats were fed after weaning with a control diet (Diet D12450B, 10% fat content). A model of peripubertal undernutrition (UN) was also generated; female rats reared as NL and fed the control diet were subjected to a 25% reduction in daily food intake from weaning (PND21) to PND36, as described earlier 50 . NL rats fed with the above-described control diet served as controls given an NN.
Phenotypic evaluation of pubertal maturation. Somatic and reproductive indices of pubertal development were evaluated as previously described 50 , including (a) body weight (BW) gain; (b) age of vaginal opening (VO), a consensus external marker of puberty in female rodents; (c) uterine and ovarian weight; and (d) serum LH and FSH levels. Based on previous data on the normal timing of puberty in female rodents [50][51][52] , VO was monitored daily from PND25 until termination of the experiment, time at which uterine and ovarian weight were recorded and serum hormone levels were assayed. Further assessment of the precise age of completion of puberty was achieved by histological analysis of the ovary, as described below.
Ovarian histological analysis for precise assessment of the completion of puberty. Ovaries (including the oviduct and the tip of the uterine horn) were fixed for at least 24 h in Bouin solution, before dehydration and embedding in paraffin. Serial (7 μm) sections were stained with hematoxylin and eosin, and evaluated under a microscope. Pubertal progression was estimated using a scoring method (Pub-Score), recently validated by one of our groups, based on histometric analyses of follicular development and development of corporal lutea 52 . This method dates pubertal maturation based on the combined analysis of follicular development and corpus luteum dynamics (the latter, for animals that has completed first ovulation). In nonovulating animals, the most advanced healthy antral follicle class, from small follicles measuring less than 275 µm in diameter to antral follicles (from F1 to F4) was determined, allowing to date prepubertal maturation from stage −4 to −1 (representing the interval expected between the age at analysis and the first ovulation). In addition, for animals that had undergone ovulation, dating of CL, as a morphological sign of ovulation, was also implemented, based on major histological features, allowing staging of pubertal timing at one-day intervals, from +1 (equivalent to CL1) to +4 (equivalent to CL4). Negative scores denote expected days until first ovulation, whereas positive scores indicate days after the first ovulation, therefore providing an integral assessment of the stage of pubertal maturation, even in animals that have not completed puberty. Surgical procedures. For intracerebral injections, standard procedures for cannulation of the lateral cerebral ventricle, followed by chronic intracerebroventricular (icv) administration, were carried out using previously published protocols 53,54 . In brief, to access the lateral cerebral ventricle, the cannulae were lowered to a depth of 3 mm beneath the surface of the skull, with an insert point that was 1 mm posterior and 1.2 mm lateral to bregma. Immature female rats received icv injections of the allosteric SIRT1 activator, Sirt1-Activator 3 (SA3; Cayman Chemical Co.) 28,33 , at a dose of 5 nmol, twice a day between PND26 and PND35. Pair-aged rats injected with vehicle (VH; 40% dimethyl sulfoxide in 0.9% sodium chloride) served as controls.
Tissue collection. For expression analyses, hypothalamic tissue containing the POA and the MBH was dissected immediately upon decapitation of the animals, by a horizontal cut ∼ 2 mm in depth with the following boundaries: one cut made 1 mm anterior of the optic chiasm, a posterior cut made at the border of the mammillary bodies, and two lateral cuts made at the hypothalamic sulci 53,55 . In addition, expression analyses were conducted on the POA, containing the AVPV region, and the MBH, which includes the ARC; dissection of these regions was carried out as previously recommended 56 , with modifications. In brief, from the above rostral limit (1 mm anterior to the rostral chiasm), the POA region was dissected by a second cut immediately caudal to the optic chiasm, and two additional lateral cuts along the borders of the optic chiasm and the border of the anterior commissure. The distal hypothalamic fragment corresponded to the MBH, delimited by lateral cuts along the hypothalamic sulci and the mammillary bodies caudally. The tissue fragments were frozen in liquid nitrogen and stored at −80°C until used for further analysis.
Hormone assays. Serum LH and FSH levels were measured using RIA kits supplied by the National Institutes of Health (Dr. A.F. Parlow, National Hormone and Peptide Program, Torrance, CA). Rat LH-I-10 and FSH-I-9 were labeled with 125 I by the chloramine-T method, and hormone concentrations were expressed using reference preparations LH-RP-3 and FSH-RP-2 as standards. Intra-and inter-assay coefficients of variation were less than 8 and 10% for LH and 6 and 9% for FSH, respectively. The sensitivity of the assay was 5 pg/tube for LH and 20 pg/tube for FSH. Accuracy of hormone determinations was confirmed by assessment of rat serum samples of known concentrations, used as external controls.
Virogenetic overexpression of SIRT1 in the ARC. Using a stereotaxic approach 58,59 , immature female rats (PND20) were injected bilaterally into the ARC with an adeno-associated virus (AAV) vector expressing green fluorescent protein (GFP; AAV-C) or SIRT1 (AAV-SIRT1  Fig. 9 Mode of SIRT1 action in the control of puberty and its modulation by metabolic cues. a Major events occurring during normal female pubertal maturation. Transition from late juvenile (PND29; left panel) to peripubertal (PND36; right panel) stages is defined by eviction of SIRT1 and EED, a key member of the PcG silencing complex, from the Kiss1 promoter in KNDy neurons, which changes the chromatin landscape from a predominantly repressive to a permissive histone configuration. According to this model, these changes would result in enhanced Kiss1 transcription, mandatory for puberty onset. b Predicted changes in Kiss1 transcriptional activity occurring under opposite nutritional conditions. In the left panel, early-onset obesity (caused by overnutrition) induces the premature eviction of SIRT1/EED from the Kiss1 promoter, and the rearrangement of histone configuration from repressive to permissive. This change, already evident by PND29, would allow enhancement of Kiss1 transcription, leading to precocious puberty. In contrast, prepubertal under-nutrition (right panel) prevents the eviction of SIRT1/EED from the Kiss1 promoter, which maintains a repressive histone configuration still at PND36. The model predicts that this protracted repression would lead to decreased Kiss1 transcription and delayed puberty stereotaxic frame (David Kopf Instruments) under ketamine/ xylazine anesthesia. The injections were targeted to the ARC (stereotaxic coordinates from Bregma AP: −2 mm; L: ±0.3 mm; DV: −9.4 mm) using a 33-gauge Nano-Fil needle (WPI) connected to a Nano-Fil 10-μl syringe (WPI). AAV vectors expressing GFP (1 × 1010 pfu/ml; Tebu-Bio) or SIRT1 (1×10 11 pfu/ml; Tebu-Bio) were delivered at a rate of 200 nl/min for 1 min (1 μl/ injection site) and the entire injector system was left in place for an additional 5 min after the injections were completed. Phenotypic monitoring of puberty included assessment of VO, which was monitored daily from PND28 to 36, and occurrence of the first estrus (after VO). On PND36, animals were euthanized, body, ovarian and uterus weights were recorded, and brain samples were collected to visualize the location of the injections by fluorescein isothiocyanate (Sigma Aldrich) or fluorescence latex microspheres (Lumafluor Inc), coadministered with AAV injections, both in AAV-C and AAV-SIRT1 animals, while assessment of GFP was used as marker of effective infection. Immunohistofluorescence localization of GFP was conducted according to previously validated protocols, using a polyclonal antibody against GFP (Abcam; 1/ 2000 dilution) 58 . In addition, in a subset of AAV-C and AAV-SIRT1 animals, MBH tissue samples were collected upon euthanasia to measure SIRT1 and histone levels by western blots (as described in the section Western blots).
Chromatin immunoprecipitation (ChIP) assays. To assess the recruitment of SIRT1 to specific gene promoters, and the association of different histone modifications to these promoters in vivo, we performed ChIP assays using chromatin extracted from the MBH and POA of prepubertal NN, UN (PND29 and PND36), and ON (PND29) female rats. Because initial analyses revealed only modest changes at Kiss1 promoter in POA samples, ChIP assays at this site were restricted to the assessment of interactions with SIRT1, H3K9Ac, and H4K16Ac. To assess the changes in histone modifications resulting from Sirt1 overexpression, ChIP assays were performed using chromatin extracted from embryonic rat hypothalamic cells, R22 (Cedarlane Laboratories, Canada), infected with either LV-Sirt1-HA or LV-GFP constructs. ChIP procedures were described previously by one of our groups 26,60,61 , and were carried out with minimal modifications, as follows. Cells were harvested for chromatin immunoprecipitation 48 to 72 h after infection. The cells and tissue fragments were washed once in ice-cold PBS containing a protease inhibitor cocktail (PI, 1 mM phenylmethyl-sulfonylfluoride, 7 μg/ml aprotinin, 0.7 μg/ml pepstatin A, 0.5 μg/ml leupeptin), a phosphatase inhibitor cocktail (PhI, 1 mM β-glycerophosphate, 1 mM sodium pyrophosphate and 1 mM sodium fluoride), and an HDAC inhibitor (20 mM sodium butyrate). Thereafter, cells and tissue fragments were cross-linked by exposing them to 1% formaldehyde for 10 min at room temperature. After two additional washing steps in PBS the samples were lysed with 200 µl SDS buffer (0.5% SDS, 50 mM Tris-HCl, 10 mM EDTA) containing protease, phosphatase, and HDAC inhibitors and sonicated for 45 s to yield chromatin fragments of approximately 500 base pairs (bp) using the microtip of a Fisher Scientific FB 705 sonicator. Size fragmentation was confirmed by agarose gel electrophoresis. The sonicated chromatin was clarified by centrifugation at 14,000 rpm for 10 min at 4°C, brought up to 1 ml in Chip Dilution Buffer (16.7 mM Tris-HCl, pH 8.1, 150 mM NaCl, 1.2 mM EDTA, 1.1% Triton X-100, and 0.01% SDS) containing the PI and PhI cocktails, and the HDAC inhibitor described above. The samples were then stored at −80°C for subsequent immunoprecipitation. For this step, chromatin was precleared with Protein A/G beads qPCR detection of chromatin immunoprecipitated DNA. Genomic regions of interest were amplified by qPCR. Accession numbers of the genes analyzed as well as the chromosomal position of the 5′-flanking region amplified, using the position of the TSS as the reference point, were as follows: rKiss1 (accession number NM_181692.1, 91 bp product), rTAC3 (accession number NM_019162.2, 123 bp product) and rProDyn (accession number NM_019374.3, 109 bp product). Primer sequences used were: rKiss1 (forward3 5′-TCGGGCAGCCAGATAGAGGAAGC -3′; reverse3 5′-TTGAGGGCCGAGGGAGAAGAG-3′), rTac3 (forward2 5′-ACGT GCGTGTCTGGGTATGTGA-3′; reverse2 5′-GGAGGGTTTGGGGGAGTCG-3′) and rProDyn (forward 5′-CTGCCTTTCTCCTACTTTTGT CTCTGTTTT-3′; reverse: 5′-CGGGGGTGGATTCTCGGTGTAG-3′). A summary of primers used in ChIP assays is provided in Supplementary Table 2. PCR reactions were performed using 1 µl of each immunoprecipitate (IP) or input samples (see below), primer mix (1 µM each primer), and SYBR Green Power Up Master Mix™ (Thermo Fisher, Waltham, MA) in a final volume of 10 µl. Input samples consisted of 10% of the chromatin volume used for immunoprecipitation. The thermo-cycling conditions used were: 95°C for 5 min, followed by 40 cycles of 15 s at 95°C and 60 s at 60°C. Data are expressed as % of IP signal/Input signal.
Detection of Sirt1 expression in GFP-expressing kisspeptin neurons by singlecell-RT-PCR. To better visualize kisspeptin neurons of the ARC, we used 1-week ovariectomized (OVX) female Kiss1 Cre:GFP mice (n = 4 animals). Single cells were dispersed and harvested according to established procedures 49 , with minor modifications, as follows. Briefly, the ARC was micro-dissected from coronal slices obtained from Kiss1 Cre:GFP mice. Following extensive washing in oxygenated artificial CSF, the dispersed cells were visualized using a Leitz inverted fluorescent microscope, patched, and then harvested with gentle suction to the pipette using a Xenoworks manipulator system (Sutter Instrument; Novato, CA). The cells were expelled into a siliconized 0.65-ml microcentrifuge tube containing a solution of 1× Invitrogen Superscript III Buffer, 15 U of RNasin (Promega), 10 mM dithiothreitol and diethylpyrocarbonate-treated water in a total of 5 μl for a single cell. cDNA synthesis was performed on single cells as previously described 62 , and stored at −20°C. Controls were prepared to detect contamination in both the harvesting procedure and the PCR. Artificial CSF (aCSF) was collected from the plate in the vicinity of the dispersed cells as a negative control. Single cells were harvested and taken through the reverse transcription but without RT enzyme to confirm that there was no amplification of genomic DNA. In the PCR procedure, we included tissue controls consisting of 25 ng of basal hypothalamic RNA that were reverse transcribed with and without RT enzyme as a positive and negative control. A water blank was also included in the PCR that contained the master mix for the PCR but no cDNA as a negative control.
Primers for Kiss1 and Sirt1 amplification were designed using Clone Manager software (Sci Ed Software), to cross at least one intron−exon boundary and optimized as previously described 62 . The primers used were as follows: Sirt1 (accession number NM_019812; 198-bp product; forward primer 694−711 nt; reverse primer 872−891 nt), Kiss1 (accession number NM_178260; 120-bp product; forward primer 64−80 nt; reverse primer 167−183 nt). Details of the primers used for single-cell RT-PCR assays (including sequences) are provided in Supplementary Table 2. The PCR was performed using 2−3 μl of cDNA template from each RT reaction in a 30 μl PCR mix as follows: initial denaturation for 2 min at 94°C; 20 s at 94°C, 30 s at 61°C (Sirt1) or 57°C (Kiss1), 30 s at 72°C for 50 cycles of amplification, with a final extension of 5 min at 72°C. PCR products were sequenced to confirm their identity.
Quantitation of SIRT1 in kisspeptin neurons by immunhistofluorescence. Animals were intracardially perfused with 4% paraformaldehyde in PBS. After removing the brains, they were submerged in the same solution overnight at 4°C followed by two days in 30% sucrose in PBS. Thirty micron slices were made on a sliding cryostat, mounted into microscope slides and vacuum-dried overnight at room temperature. The sections were incubated overnight at 4°C with a rabbit polyclonal antibody against kisspeptin (Ab9754 from Millipore, diluted 1:2000), followed by biotyramine enhancement method, before developing the reaction to a red color with Stretpavidin Alexa 568 (Invitrogen; dilution 1:500). Next, sections were incubated with anti-SIRT1 antibodies (sc-74504, Santa Cruz Biotechnologies, Dallas, TX; 1:1000 dilution), and the reaction was developed the next day to a green color using Alexa 488 donkey anti-goat IgG (Invitrogen, 1:500). Fluorescent images were acquired with a Zeiss Axiovert 200 M microscope with a ×20 C-apochromat NA1.2 objective. Kisspeptin cells were identified as immunoreactive neuron-like cells that stood out above background, and that were clearly discernible from immunoreactive fibers. To quantify changes in cellular SIRT1 immunoreactivity, each kisspeptin cell was manually outlined and the intensity of the SIRT1 signal was measured using ImageJ software. Each outline was then moved over neighboring nonlabeled areas to obtain a measure of the mean intensity of the background. Following subtraction of the background values from each cell, the mean intensity values per group were calculated. Fluorescent in situ hybridization (FISH). Brains from PND28 female rats (n = 3) were fixed by intracardiac perfusion of 4% paraformaldehyde borate buffer, and were processed for hybridization histochemistry, as previously described 63 . In detail, we used the double FISH procedure employing nonradioactive complementary (c)RNA probes, as optimized by us 26 . A cRNA complementary to Sirt1 mRNA was labeled with fluorescein-12-UTP. A Kiss1 cRNA probe was labeled with digoxigenin-11-UTP. The labeling reactions were performed in a 10 µl volume, as reported 64 . Control sections were incubated with sense probes transcribed from the same plasmid, but linearized on the 3′ end to transcribe the coding strand of the cDNA template.
Statistics. Statistical analyses were performed using Prism software (Graphpad Prism version 7.00 for Windows, GraphPad Software, La Jolla, California, USA, www.graphpad.com). The data were first subjected to normality tests. For data passing these tests, the differences between several groups were analyzed by oneway ANOVA followed by the Student−Newman−Keuls multiple comparison test for unequal replications. The Student's t test was used to compare two groups. Analyses were two-sided. Quantitative, continuous data are expressed as the mean ± SEM for each group. When comparing percentages, groups were subjected to arc-sine transformation before statistical analysis to convert them from a binomial to a normal distribution; statistical differences of frequency distributions were calculated using X 2 tests. A P value of <0.05 was considered statistically significant. The sample size was selected based on previous experience with studies addressing molecular and neuroendocrine regulation of puberty, assisted by power analyses performed using values of standard deviation that we usually obtain when measuring parameters analogous to those examined in this study. Based on those calculations, a minimal group size of n = 6 animal per group was established as a general rule, as analyses using these sample size should provide at least 80% power to detect effect sizes using the tests indicated above, with a significance level of 0.05. Note that for physiological experiments, groups sizes largely exceeded this threshold; yet, based on standard procedures, while phenotypic and hormonal analyses were applied to all individuals, more complex molecular and histological analyses in these experiments were implemented in a representative subset of randomly assigned samples from each group. Further details are provided in the corresponding figure legends. For molecular (ChiP) assays, tissues from four individuals per experimental group were studied; assays were done in duplicate. As a general rule, the investigators directly performing the experimentation involving physiological/molecular determinations were not blinded to the group allocation, but primary data analyses conducted by senior authors were conducted independently to avoid any potential bias.