Minireview: Metabolic Control of the Reproductive Physiology: insights from genetic mouse models

Abstract

Over the past two decades, and in particular over the past 5-7 years, there has been a tremendous advancement in the understanding of the metabolic control of reproductive physiology. This has been in large part due to the advancement and refinement of gene targeting tools and techniques for molecular mapping. Yet despite the emergence of exciting and often times thought-provoking data through the use of new mouse models, the heavy reliance on gene targeting strategies has become fundamental in this process and thus caution must be exercised when interpreting results. This minireview article will explore the generation of new mouse models using genetic manipulation, such as viral vector delivery and the use of the Cre/loxP system, to investigate the role of circulating metabolic hormones in the coordination of reproductive physiology. In addition, we will also highlight some of the pitfalls in the use of genetic manipulation in the current paradigms. However, it has become clear that metabolic cues employ integrated and plastic neural circuits in order to modulate the neuroendocrine reproductive axis, and despite recent advances much remains to be elucidated about this circuitry.

Introduction

Sufficient energy stores and nutrient availability are critical for reproductive competency in mammals, and is therefore necessary for the survival of the species. As such, the central reproductive axis must be in constant dialogue with circulating hormones, including leptin, insulin, and ghrelin, signaling energy stores. Many have since described how peripheral hormones interact with the reproductive axis (Ahima et al., 2000; Elias, 2012; Elias and Purohit, 2013; Hill et al., 2013; Tena-Sempere, 2008). In the last twenty years rodent models utilizing genetic knockout or knockin strategies have provided key insights into how peripheral metabolic cues alter reproductive physiology. The manipulation of genetically modified mice has, at often times, provided exciting and sometimes surprising data. This minireview will focus on the use of genetic tools to elucidate mechanism by which metabolic hormones such as leptin, insulin, and ghrelin influence reproductive functioning with a particular focus on recent data stemming from the last five years (Table 1).

Table 1.

Overview of key findings of genetically engineered mouse models.

Mouse model	Metabolic phenotype	Reproductive phenotype	Reference
Leptin
Leptin deficiency (ob/ob) & Leptin receptor deficiency (db/db)	Obese Diabetic	Infertile	(for a comprehensive review (Elias and Purohit, 2013)
Leptin Circuitry
Neuron-specific deletion of LepR	Obese Diabetic	?	(McMinn et al., 2004)
Neuron-specific recovery of LepRb	Leaner	Improved fertility	(Kowalski et al., 2001)
GnRH specific deletion of LepRb	None	None	(Quennell et al., 2009)
Deletion of LepRb in Kiss-1 cells	None	None	(Donato et al., 2011)
Re-expression of LepRb in Kiss-1 cells	Obese	Infertile	(Cravo et al., 2013)
Re-expression of LepRb specifically in the PMv	Obese	Improved fertility	(Donato et al., 2011)
Ablation of AgRP neurons in ob/ob	Lean	Improved fertility	(Wu et al., 2012)
Deletion of LepRb in vGluT2 cells	None	None	(Kong et al., 2012; Zuure et al., 2013)
Deletion of LepRb in Vgat cells	Obese	Delay in puberty Prolonged estrous cycles	(Zuure et al., 2013)
Leptin Signaling
Deletion of STAT5 in LepRb cells	None	None	(Singireddy et al., 2013)
Deletion of STAT3 in LepRb cells	Obese	Fertile	(Singireddy et al., 2013)
Point mutation in LepRb STAT3 signaling Tyr1138→ S1138	Obese	Fertile	(Bates et al., 2003)
Point mutation LepRb ERK signaling Tyr985→ L985	Lean	Fertile	(Bjornholm et al., 2007)
Point mutation LepRb STAT5 signaling Tyr1077 → F1077	Mild increase in adiposity	Slightly lengthened estrous cycle	(Patterson et al., 2012)
Global deletion of Crtc1	Increased body weight	Infertile or None	(Altarejos et al., 2008; Breuillaud et al., 2009)
Insulin
Neuron specific deletion of IR	Increased body weight	Impaired fertility	(Bruning et al., 2000)
Global IRS2 deletion	Increased circulating leptin	Low gonadotropin secretion	(Burks et al., 2000)
Global deletion of p110α	Heterozygotes leaner	Heterozygotes fertile	(Foukas et al., 2006)
GnRH specific deletion of IR	None	None	(Divall et al., 2010)
Deletion of IR in Kiss-1 cells	None	Slight delay in pubertal onset	(Qiu et al., 2013)
Deletion of IR in POMC cells	None	None	(Hill et al., 2010)
Double deletion of IR and LepRb in POMC cells	Insulin resistance	Impairments in fertility Hyperandrogenemia	(Hill et al., 2010)
Deletion of IRS2 in LepRb cells	Increased body weight	None	(Sadagurski et al., 2012)
Ghrelin
Global GHSR KO	Resistant to high fat diet	?	(Zigman et al., 2005)

Leptin

Leptin, a 16-KD adipocyte derived hormone that circulates in proportion to body fat mass, signals long-term energy stores to the central nervous system (Considine et al., 1996; Maffei et al., 1995; Zhang et al., 1994). In states of negative energy balance, the drop in leptin levels signals energy insufficiency eliciting food seeking appetitive behaviors while concurrently restricting reproductive physiology (Ahima et al., 1996; Considine et al., 1996). Leptin is generally accepted as a permissive factor for sexual maturation and for signaling adequate energy stores for reproduction (Ahima et al., 1997; Cheung et al., 1997; Elias, 2012; Elias and Purohit, 2013). There are six isoforms of the leptin receptor (LepRa, LepRb, LepRc, LepRd, LepRe, LepRf), of which only the long form LepRb has signaling capacity through autophosphorylation of the janus kinase 2 (JAK2) protein. It is the long form that primarily mediates the physiological effects of leptin (as reviewed by (Ahima and Osei, 2004; Ahima et al., 2000; Bates and Myers, 2003)). Leptin deficiency (ob/ob) or leptin receptor deficiency (db/db) results in severe obesity, diabetes, and infertility (Farooqi et al., 1999; Ingalls et al., 1950). These individuals display abnormal gonadal physiology, low circulating gonadotropins, and low sex steroid secretion. Exogenous leptin administration to ob/ob individuals can reverse body weight, insulin sensitivity, and fertility (Chehab et al., 1996; Farooqi et al., 1999).

While LepRb is expressed in various tissues, including the ovaries and the pituitary, its primary action on energy balance and reproduction is mediated by the brain. The use of the Cre recombinase/loxP system in genetically engineered mice, in which the cre recombinase protein recognizes genomic loxP sites to mediate excision of the genome, has been crucial at determining the role for LepR in the brain. Targeted deletion of the LepRb in neurons by cre-mediated excision of loxP sites flanking exon 17 of the LEPR gene, encoding the JAK2 binding site (LepR^loxp/loxp), recapitulated the obese and diabetic phenotypes seen in ob/ob and db/db mice (McMinn et al., 2004; McMinn et al., 2005). Conversely, selective recovery of LepRb within neurons using neuron-specific markers (neuron specific enolase and synapsin) restores both body weight and fertility (Kowalski et al., 2001). Although these papers were crucial at highlighting the importance of the brain, it remained unclear as to whether the rescue of fertility was secondary to weight loss or the correction of other endocrine processes associated with the restoration of LepR signaling in the brain. In this sense, genetically engineered mouse models have been an invaluable tool at parsing neuroanatomical and biological functions in discrete regions of the CNS.

The use of genetically modified mouse models in the identification of circuitry and function of leptin action on reproduction

As the central controller and final common pathway of the reproductive axis, neurons synthesizing gonadotropin releasing hormone (GnRH) were originally postulated to directly integrate leptin signaling. Evidence from studies using in vitro cultures of immortalized GnRH cell lines suggested that GnRH neurons contained LepRb and that leptin could stimulate GnRH release through direct transmission on GnRH neurons (Magni et al., 1999; Zamorano et al., 1997). However, a series of in vivo experiments provided convincing evidence that leptin acts indirectly on GnRH neuron as the use of Cre/loxP mediated excision of LepRb in GnRH neurons resulted in no change in fertility (Quennell et al., 2009). Since then a number of research articles have focused on leptin action in various cell types and hypothalamic nuclei in mediating leptin action upstream of GnRH neurons. A great deal of interest formed surrounding the kisspeptin neural network, as mutations in kisspeptin or its receptor, Kiss1R, leads to hypogonadotropic hypogonadism resembling that of leptin or leptin receptor deficiency (d'Anglemont de Tassigny et al., 2007; de Roux et al., 2003; Seminara et al., 2003; Topaloglu et al., 2012). The location of the kisspeptin neuronal population in the arcuate nucleus of the hypothalamus (ARH) appeared serendipitous in that it also houses an abundance of leptin receptors (Caron et al., 2010; Clarkson et al., 2009; Gottsch et al., 2004; Scott et al., 2009; Smith et al., 2005). Following gonadectomy, nearly 40% of kisspeptin neurons in the ARH express the leptin receptor, suggesting leptin's stimulatory effect on the reproductive axis could be mediated by kisspeptin (Smith et al., 2006). However, studies using mice expressing the fluorescent reporter eGFP under the endogenous promoter of Tac2 (a protein expressed in kisspeptin cells of the arcuate nucleus) revealed that exogenous leptin administration induces P-STAT3 in only 5%-10% of cells in intact female mice (Louis et al., 2011). Moreover, cre mediated excision of LepRb in kisspeptin expressing cells lead to no deficits in reproductive capacity (Donato et al., 2009), suggesting that leptin action in kisspeptin is not required for fertility. Consistent with that hypothesis, leptin also has the ability to stimulate LH release in mice null for Kiss1R (Bellefontaine et al., 2014)

More recently, a new reactivable LepR mouse model has been used in our laboratory. A transcriptional blocker flanked by loxP sites was inserted into the Lepr gene between exon 16 and exon 17 resulting in a truncated form of the LepR lacking the Jak2 binding site and rendering a global knockout of the LepRb (LepR^loxTB/loxTB ). Like the natural mutation in the long form of the leptin receptor, db/db, the LepR^loxTB/loxTB mice are both obese and infertile. These mice, however, allow for the reactivation of LepRb using the efficient and widespread use of cre-mediated excision. When crossed to our Kiss1-cre line, the reactivation of the leptin receptor only in kisspeptin expressing cells did not drive the progression through puberty nor did it consequently improve fertility in these mice (Cravo et al., 2013). Direct leptin action on kisspeptin neurons, it appears, is neither necessary nor sufficient for leptin's effect on reproduction. It would, however, be prudent to suggest that kisspeptin does not play a role. In both mouse models (Kiss1-cre; LepR^loxp/loxp and Kiss1-cre; LepR^{loxpTB/loxpTB}) the recombination would have occurred during development with the first expression of cre recombinase and thus it remains a possibility that kisspeptin plays a role after the completion of the pubertal process during adulthood.

The ventral premammilary nucleus (PMv) is part of the sexually dimorphic circuitry and expresses the nuclear estrogen receptor and is responsive to sexual odorant molecules (Donato and Elias, 2011; Leshan et al., 2009; Merchenthaler et al., 2004; Simerly et al., 1990). It also houses a dense population of leptin receptors, and was recently found to be important for leptin's stimulatory action on the neuroendocrine reproductive axis (Donato et al., 2011; Donato et al., 2009; Leshan et al., 2009). Indeed, food deprivation over a prolonged period reduces luteinizing hormone (LH) levels and reduces estrogen and progesterone secretion, effectively suppressing fertility. Exogenous administration of leptin during the fast partially rescues LH levels and prevents a prolonged cycle during fasting conditions, indicating that the reduction in leptin levels signals neuroendocrine changes and that administration of leptin may increase LH levels and sustain fertility (Ahima et al., 1996). Interestingly, excitotoxic lesions to the PMv prevented leptin's ability to rescue LH levels following a fast, suggesting that the PMv relays, in part, leptin signaling to GnRH neurons (Donato et al., 2009). Importantly, PMv neurons are directly connected to GnRH neurons which was visualized through the expression of the transsynaptic tracer barley lectin under the promoter of GnRH. The PMv showed the presence of the barley lectin, as detected by immunoreactivity for wheat germ agglutinin (WGA), suggesting synaptic contact between GnRH and neurons in the PMv (Boehm et al., 2005). Because GnRH fibers apparently do not reach as caudal as the PMv, the PMv → GnRH neuron connections are likely to be afferent to GnRH neurons. Moreover, some of the WGA positive PMv neurons were responsive to leptin (visualized by the induction of the phosphorylation of STAT3 by leptin), demonstrating that leptin sensitive neurons may interact directly with GnRH cells (Leshan et al., 2009; Louis et al., 2011).

To examine further the role of the PMv in the cross talk between metabolism and reproduction, our laboratory utilized the LepR^neo/neo mouse model, in which a PGK-neo cassette flanked by FRT sites was inserted between exon 16 and 17 of the Lepr gene, rendering the mouse unable to transcribe the JAK2 binding site on exon 17, critical for intracellular signaling on the long form of the leptin receptor (Coppari et al., 2005). These mice were intended to mimic the phenotype of the db/db mice with the advantage that the neo cassette could be excised through viral injection of Flpe (AAV-flpe-GFP) (Coppari et al., 2005). The LepR^neo/neo mouse model therefore allowed for the selective reactivation of the leptin receptor in site specific tissue at any time point during development and circumvents recombination early in life. Strikingly, the selective reactivation of the LepRb in LepR^neo/neo female mice rescued the infertile phenotype while having no effect on body weight, demonstrating a dichotomy of leptin action in the neuroendocrine brain (Donato et al., 2011). Mice with the reactivation of LepRb showed increase uterine weight, a marker of increased circulating estrogen, and some females were able to become pregnant (Donato et al., 2011). However, many of the LepR^neo/neo mice were unable to carry pups to term or to nurse pups (Donato et al., 2011). This may be due to their severe metabolic phenotype, altered development of the mammary glands, and separate circuits involved in the maintenance of gestation. Alternatively, variability in injection sites and sizes may also factor in the heterogeneity in reproductive success between AAV-flpe-GFP LepR^neo/neo females. Surprisingly, reactivation of LepRb in male LepR^neo/neo mice had no effect on reproductive capacity, perhaps pointing to an alternative site within the brain involved in the dialogue between reproduction and metabolism in males (Donato et al., 2011).

Although kisspeptin within the ARH is neither necessary nor sufficient for leptin's effect on fertility, the ARH houses many leptin receptor expressing neurons that have been previously implicated in the metabolic control of reproduction. Recently, Palmiter and colleagues demonstrated that AgRP neurons could be one integral link between leptin and reproduction. To test their model, the authors produced a mouse model in which the diphtheria toxin receptor (DTR) was placed under the promoter for Agrp (AgRP^DTR), such that when mice are exposed to the diphtheria toxin (DT), neurons expressing the receptor will be rapidly ablated (Luquet et al., 2005). Ablation of AgRP neurons using this method in adult mice is lethal, although ablation of these neurons in early neonatal life results in an unaltered metabolic phenotype. Indeed, during early postnatal life metabolic circuits are being established (Bouret et al., 2004), likely allowing for compensatory changes to occur in AgRP^DTR (Luquet et al., 2005). Here the authors took advantage of selectively targeting AgRP neurons with DT in leptin deficient ob/ob mice to see an improvement on metabolism. Strikingly, the ablation of AgRP neurons in ob/ob did not result in death, but rather these mice showed body weight reduction and an improvement of fertility (Wu et al., 2012). There are several caveats to note, one is that this phenotype was only seen in mice with a moderate obese phenotype (35-40g). Secondly, following the ablation of AgRP neurons, the female ob/ob mice showed a drastic reduction in body weight prior to mating and thus the improvement in fertility could be secondary to body weight loss. Finally, AgRP neurons also co-express neuropeptide Y (NPY) and GABA, thus the effects on reproduction following the ablation of these neurons may be due to a loss of one or more of the neurotransmitters. Recent data have shown that the AgRP and melanocortin system may play a role in the effect of leptin on reproduction, as the leptin receptor deficient db/db mice when crossed with a MC4 global knockout or an AgRP global knockout showed improved fertility (Israel et al., 2012; Sheffer-Babila et al., 2013). Because the deletion of NPY in leptin deficient ob/ob mice also renders mice with reproductive capacity (Erickson et al., 1996), the specific roles for AgRP, NPY, and/or GABA in this population need to be carefully examined.

GABAergic neurons appear to mediate much of the anorectic effects of leptin, while glutamatatergic transmission has little influence upon leptin's antiobesity effects (Kong et al., 2012; Vong et al., 2011; Zuure et al., 2013). Using the Cre/loxP system, the vesicular GABA transporter (Vgat) or the vesicular glutamate transporter (Vglut) were used to target the deletion of LepRb in either GABA neurons or glutamate neurons. Similar to the metabolic phenotype, GABA transmission rather than glutamate transmission seems necessary for adequate reproductive functioning (Zuure et al., 2013). Indeed, the ablation of LepRb in GABA neurons resulted in delayed pubertal onset and first estrous, as well as a prolonged period in estrous once sexually mature (Zuure et al., 2013). The lack of effect on reproduction in mice with the targeted deletion of LepRb in vGluT2 expressing cells is somewhat curious. The PMv, which has been shown to relay leptin's signal to the neuroendocrine reproductive axis, is dense with glutamatergic neurons (Donato et al., 2011). Thus, it may be likely that the early activation of cre recombinase in these mice allows for compensatory mechanisms to take place. Alternatively, LepRb activation in the PMv is sufficient for leptin's stimulatory effect on reproduction but not a requirement. The latter of which may be less plausible as lesions to this region during adulthood impair leptin's action on the reproductive axis (Donato et al., 2009), indicative of developmental plasticity of the metabolic reproductive axis. The reproductive axis is a complex system and thus requires multiple pathways regulating the metabolic control of reproductive physiology. The advancement in gene targeting, injectable viral vectors, and accessibility of optogenetic tools will help in the future to define site specificity of the effects and development versus late postnatal necessity of neural circuits.

Neurons synthesizing the gaseous neurotransmitter nitric oxide (NO) may also represent another key pathway in leptin signaling to the neuroendocrine reproductive axis. Exogenous leptin administration activates neuronal NO synthase (nNOS), the enzyme required to produce NO, in several nuclei of the hypothalamus, including the preoptic area (POA), ARH, and PMv (Bellefontaine et al., 2014; Donato et al., 2010). The selective deletion of LepRb in nNOS neurons results in extreme obesity and a delay in sexual maturation (Leshan et al., 2012). Further, global genetic deletion or pharmacological blockade of nNOS during adulthood renders leptin unable to stimulate the reproductive axis (Bellefontaine et al., 2014). Moreover, nNOS null mice when crossed to the ob/ob mouse line did not progress through puberty following chronic leptin administration. This effect may be mediated by nNOS neurons of the POA as pharmacological inhibition of nNOS, via stereotaxic infusions of a nNOS inhibitor into the POA, in ob/ob mice prevents the rescue of puberty and cyclicity by leptin.

Importantly, the administration of leptin continued to decrease body weight in the ob/ob mice infused concurrently with L-NAME, providing further evidence for the division of leptin signaling in metabolic homeostasis and reproductive functioning (Bellefontaine et al., 2014). Thus, NO synthesizing neurons within the POA may represent another circuit integrating leptin signaling to the central reproductive axis.

Leptin receptor signaling: Genetic tools used to define intracellular signaling pathways involved in the metabolic control of reproduction

An in depth look at leptin signaling through LepRb is beyond the scope of this review, however many alternative reviews detailing LepRb signaling may provide a clearer picture (Ahima and Osei, 2004; Bates and Myers, 2003; Myers and Olson, 2012). In brief, leptin upon binding to its receptor promotes the autophosphorylation of the JAK2 protein associated to LepRb, which in turn promotes the phosphorylation of specific tyrosine residues: Tyr985, Tyr1077, Tyr1138. Utilizing the ability to create point mutations in the tyrosine residues has been paramount in our understanding of the specific role for each intracellular signaling pathway in LepRb signaling. Indeed, point mutations in Tyr985 and Tyr1138, which recruit SHP2 and STAT3 respectively, mediate several aspects of energy balance and suppression of the LepRb signaling (Bates et al., 2003; Bjornholm et al., 2007). However, despite the extreme metabolic and endocrine dysfunction in these mice, reproductive functioning remained relatively undisturbed demonstrating a clear separation in roles for each signaling pathway. Recently, the generation of a knockin point mutation of the Tyr1077, responsible for the recruitment of the latent transcription factor STAT5, resulted in both a mild metabolic and reproductive phenotype. Females demonstrated a mild increase in adiposity, which was slightly exacerbated when exposed to a high fat diet, although the males appear relatively undisturbed. While the age of onset of puberty (both vaginal opening and first estrous) were similar in wildtype littermates, estrous cyclicity was perturbed in that the point mutation lengthened the time between ovulatory periods (Patterson et al., 2012). However discrepancies exist as the deletion of STAT5 in LepRb expressing neurons results in no reproductive deficits and is assumed not required for fertility (Singireddy et al., 2013). Regardless, the little to no phenotypes seen with regards to the STAT5 pathways does not recapitulate the infertility seen in db/db mice and so it remains relatively unclear as to the intracellular signaling pathway largely responsible for leptin's effect on the reproductive axis.

Alternative nontraditional regulators have also been suggested to mediate aspects of leptin signaling on reproduction. One such regulator known as the cyclic AMP responsive element-binding protein-1 regulated transcription coactivator-1 (Crtc1) has been implicated in both energy balance and reproduction. Using a gene-trap method to delete endogenous Crtc1, Altaregos et al. (2008) found that in mice null for Crtc1 both male and female mice displayed obesity and no offspring were obtained from mating Crtc1 null mice with wildtype animals. LH levels in the Crtc1 deficient mice were low, which correlated with the low levels of kisspeptin in the hypothalamus (Altarejos et al., 2008). These results however remain controversial as another group, Breuillaud et al (2009) using the same gene trap method to create a global deletion of Crtc1, were unable to replicate the results on fertility and serum gonadotropin levels, as well as hypothalamic kisspeptin levels appeared comparable to their wildtype littermates (Breuillaud et al., 2009).

Insulin

Insulin is a hormone secreted by the β islets cells of the pancreas notable for its prominent role in glucose homeostasis. Upon binding to its receptor, the insulin receptor undergo tyrosine phosphorylation and rapidly recruits the insulin receptor substrate (IRS), of which there are 4 members: IRS1 and IRS2 which are wildly expressed, IRS3 expressed in adipose tissue, and IRS4 expressed in the thymus, brain, and kidney (Keller et al., 1993; Lavan et al., 1997; Lavan and Lienhard, 1993). The IRS proteins in turn recruit and activate several downstream effectors to induce a cellular and molecular response. The central deletion of the insulin receptor (IR) by selective deletion of IR in nestin expressing cells resulted in increased body weight, mild insulin resistance, and impairments in fertility (Bruning et al., 2000). Indeed, females and males expressed a reproductive phenotype and showed a corresponding decrease in gonadotropin secretion. The gonads were also affected as few spermatogenic cells were seen in males, while female mice showed a decreased in the number of corpora lutea (Bruning et al., 2000). Moreover, both sexes had a reduction in the number of successful offspring produced, although this was more prominently seen in females (Bruning et al., 2000). The selective reactivation of the IR under the transthyretin promoter, in which the re-expression of IR is restricted to the brain, hepatocytes, and pancreatic β cells, in mice otherwise null for IR improves fertility in both sexes (Okamoto et al., 2004). No deficits or very mild phenotypes were recorded in the length of the estrous cycle, ovarian histology, or reproductive fecundity (Nandi et al., 2010; Okamoto et al., 2004), highlighting the importance of IR expression in the brain and its control on fertility. Global deletion of IRS2 revealed a drastic reproductive phenotype. Indeed, low gonadotropin and gonadal steroid levels were seen, while abnormal morphology was observed in the ovary itself (Burks et al., 2000). Development of the testis was also altered in IRS2 null mice resulting in reduced testis weight, although when normalized for size no difference was seen in the number or the maturation of spermatozoa (Griffeth et al., 2013). Due to the widespread expression of IRS2, it is probable that the reproductive deficits occur at more than one level of the hypothalamic-pituitary-gonadal axis. Insulin signaling at the level of the pituitary has indeed been shown to be an important site for fertility in mice. The IR gene was designed with loxp sites flanking exon 4 that, when exposed to cre recombinase, resulted in a premature stop codon truncating the IR (Brothers et al., 2010; Divall et al., 2010). The deletion of IR within the pituitary was generated by cre-mediated excision of IR from cells expressing the α- subunit of gonadotropins (PitIRKO). Although fertility in PitIRKO females was undisturbed in normal conditions, when exposed to a high fat diet, LH levels remained similar to wildtypes on a chow diet, cyclicity remained regular, and copora lutea were present. The PitIRKO high fat diet females had fertility rate that was six times higher than in the wildtype mice exposed to a high fat diet (Brothers et al., 2010). Interesting, the PitIRKO females still displayed many symptoms of the metabolic syndrome, including increased body weight, and increased serum insulin and leptin levels, demonstrating a separation of insulin signaling in metabolism and reproduction (Brothers et al., 2010).

As IR signaling through the PI3K pathway plays an integral role in food intake and glucose homeostasis (Foukas et al., 2006; Hill et al., 2009; Niswender et al., 2003), it is an obvious candidate for insulin mediated control of reproductive physiology. Yet, few targeting studies have examined the role for PI3K in reproduction. Mice heterozygous for p110a loss-of-function mutation or those with POMC-specific deletion of one of the catalytic subunits of the PI3K, p110α, resulted in dysregulation of the energy homeostasis (Foukas et al., 2006; Hill et al., 2009). The global deletion of p110α in homozygosity is embryonically lethal, however heterozygotes survive. Despite the metabolic dysfunctions and growth retardation seen in the heterozygous mice, they were viable and fertile (Foukas et al., 2006). Unfortunately, the authors did not detail the reproductive phenotype and therefore it is not known if these mice displayed reproductive deficits. Regardless, to circumvent the use of heterozygous mice and to discern the role for PI3K pathway in insulin signaling to the reproductive axis further studies targeting specific cell populations are required. However, additional challenges will remain, including the differentiation between the role for PI3K in metabolism and the role for PI3K in reproduction. Identification of specific cell types and pathways will be crucial in this exercise, as well as the use viral vectors and/or optogenetic and designer receptors exclusively activated by designer drugs (DREADD) technology will be important to determine the role for insulin signaling in reproductive physiology.

Neural pathways involved in the metabolic control of fertility by insulin: A long way to go

Although insulin signaling in the brain is critical for insulin's effect on reproduction, few studies have attempted to elucidate the neural pathways involved in the metabolic control of the reproductive axis by insulin. Generation of mice with cre recombinase under the promoter of GnRH was used for cre-mediated deletion of the insulin receptor. The selective ablation of IR from GnRH neurons does not delay sexual maturation nor does it perturb fertility and so, like leptin, insulin acts on the neuroendocrine reproductive axis afferent to GnRH neurons themselves (Divall et al., 2010). The kisspeptinergic network was again a viable system in which insulin might influence GnRH neurons as IR is highly expressed in the ARH. Use of the Kiss1-cre eGFP mouse model allowed for the identification of approximately 20% of kisspeptin neurons that co-expressed IR mRNA (Qiu et al., 2013). While the specific deletion of IR in kisspeptin neurons resulted in a slight delay in pubertal onset in both males and females, the phenotype was mild and fertility during adulthood was unaltered and no abnormalities were observed in the gonads (Qiu et al., 2013). It is therefore conceivable that one or more cell populations distinct from kisspeptin are involved in the relay of insulin's effect on GnRH neurons.

Insulin and Leptin: Compensation of phenotypes through shared molecular pathways?

Insulin and leptin share a presumed intracellular signaling pathway in that PI3K is activated in the presence of insulin or leptin alone, and as such PI3K represents a common pathway through which metabolism may influence reproduction (Acosta-Martinez, 2011; Carvalheira et al., 2005; Niswender et al., 2003; Niswender et al., 2001). Neurons in the ARH such as POMC neurons express both the IR and LepRb. Although POMC-LepRb expressing neurons and POMC-IR expressing neurons are thought to represent separate and distinct populations, these cells may have overlapping projections and functions (Hill et al., 2010; Williams et al., 2010). The single deletion of IR or LepRb from POMC neurons does not produce a profound effect on either body weight or reproduction, yet the ablation of both IR and LepRb from POMC neurons results in insulin resistance and impaired fertility (Hill et al., 2010). Females with double IR and LepRb deletion in POMC neuron show high gonadotropin release, follicular arrest, hyperandrogenemia, and lengthened estrous cycles, symptoms strikingly similar to polycystic ovarian syndrome (Hill et al., 2010). It seems therefore that lacking one receptor in one subset of POMC neurons may allow for functional redundancies in another POMC population expressing the other receptor to compensate for the loss of function.

Although insulin and leptin have overlapping signaling it appears that it is rather distinct from one another. IRS2 is important for both energy balance and fertility, as shown by studies using global IRS2 knockouts. In a recent study Sadagurski et al hypothesized that LepRb expressing neurons that express IRS2 were a crucial target controlling energy balance (Sadagurski et al., 2012). To test their model, the authors created a mouse model in which the coding portion of the Irs2 gene is flanked by loxp sites and was crossed to a mouse expressing cre under the endogenous LepR promoter, thereby ablating IRS2 from LepRb neurons (Lepr^Irs2). While metabolic parameters were altered in these mice, as the males had higher body weight, were hyperphagic, and had decreased energy expenditure, the effects were independent of leptin itself (Sadagurski et al., 2012). Despite the alterations in energy balance, both females and males were fertile (Sadagurski et al., 2012), indicating that another subset(s) of neurons expressing IRS2 distinct from LepRb neurons are responsible for the reproductive phenotype of the IRS2 global knockouts. Candidate populations would likely include insulin and/or insulin- like growth factor responsive cells which act on related receptors converging on the downstream IRS2 substrate.

Leptin and insulin also activate the downstream IRS4 substrate that is predominately expressed in the ventral hypothalamus (Fantin et al., 1999; Numan and Russell, 1999; Sadagurski et al., 2013) and overlaps with LepRb expressing regions. The IRS4 knockout exhibits very mild metabolic and reproductive phenotypes (Fantin et al., 2000), and it was therefore hypothesized that IRS2 could compensate for the loss of IRS4 through complimentary functions. To test this hypothesis Sadagurski et al. (2013) generated a novel mouse line with the deletion of IRS2 in nestin expressing neurons using the cre/loxP system, in which loxP sites flanked Irs2 alleles were excised in neurons, and further crossed to the IRS4 null mouse (bIrs2^-/-; Irs4^-/-) (Sadagurski et al., 2013). Although an obese metabolic phenotype was noted in the bIrs2^-/-; Irs4^-/- mice, no deficits in reproductive capacity were observed nor were any further metabolic dysfunctions noted with the authors crossed bIrs2^-/-; Irs4^-/- to the Lepr^Irs2 mouse (Sadagurski et al., 2013). Thus it remains to be seen which pathways relay insulin signaling to the neuroendocrine reproductive axis and how leptin may functionally overlap with insulin signaling.

Ghrelin

Ghrelin is a 28 amino acid peptide secreted by the stomach and is largely associated with the control of appetitive behaviors and energy homeostasis. Signaling through its receptor growth-hormone secretagogue receptor (GHSR) promotes food intake and decreases energy expenditure (Castaneda et al., 2010; Kamegai et al., 2001; Kojima et al., 1999; Tschop et al., 2000; Zigman and Elmquist, 2003). As with most metabolic hormones, ghrelin is a pleiotropic hormone that has a variety of functions, including a putative control on the reproductive axis. Generally thought to be inhibitory signal to the reproductive axis, the exogenous administration of ghrelin potentially suppresses LH secretion in a number of species, including the rodent and the sheep (Fernandez-Fernandez et al., 2005; Furuta et al., 2001; Iqbal et al., 2006) (for an indepth review on sheep see in this issue (Clarke, 2014)). Given the potent effect of exogenous ghrelin on food intake, it was surprising that the global deletion of GHSR results in mild metabolic phenotype unless exposed to high fat diet (Zigman et al., 2005). Although no reproductive phenotype was described, the female GHSR mice also demonstrated a resistance to body weight gain on a high fat diet (Zigman et al., 2005). It is tempting then to speculate that given the resistance to diet induced obesity that the GHSR null females may be protected against high fat diet-induced infertility.

Similar to LepRb, the expression of GHSR has not been detected in GnRH neurons suggesting that, should ghrelin have a definitive role on the neuroendocrine reproductive axis, the signals must come afferent to GnRH neurons (Smith et al., 2013). However, the expression of GHSR has been described in many areas of the hypothalamus, including those implicated in reproductive functioning and regions expressing estrogen receptor α (ERα) (Frazao et al., 2014; Zigman et al., 2006). Interestingly, the orexigenic effects of gherlin is altered depending on the hormonal milieu in female rodents (Clegg et al., 2007). Accordingly, GHSR mRNA colocalizes with immunoreactive ERα cells within the AVPV, MPO, VMH, and ARH (Frazao et al., 2014).

Furthermore in the presence of estrogen, ARH GHSR hybridization shows a striking increase from 20% to 80% in ERα cells (Frazao et al., 2014). Because ERα is expressed in the majority of kisspeptin neurons of the AVPV and ARH, our laboratory utilized the Kiss-cre GFP reporter mice to determine the morphological and electrophysiological relationship between kisspeptin cells and ghrelin responsive cells. Approximately 10% of AVPV kisspeptin neurons express GHSR mRNA regardless of circulating estrogens, while the presence of estrogen increases GHSR mRNA in ARH kisspeptin neurons from 25% to 80% (Frazao et al., 2014). Although it was also shown that ghrelin depolarizes Kiss-cre GFP expressing cells in the ARH (Frazao et al., 2014), the physiological relevance of GHSR expressing kisspeptin neurons in modulating reproductive physiology remains to be evaluated.

Conclusion

In recent years the advancement and relative availability of genetically modified mouse models and viral vectors have allowed great strides to be taken with regards to our understanding of how peripheral metabolic hormones mediate aspects of the neuroendocrine reproductive axis. With a particular interest to leptin, molecular mapping tools and selective targeting using the cre/loxP or flp/FRT system has been imperative in our identification and understanding of how specific nuclei transmit leptin action. Yet caution must be exercised should a lack of leptin action in one population preclude pubertal onset and fertility. Functional redundancies and plasticity must exist in the metabolic control of reproduction since it ensures survival of the species. Nevertheless, with new technologies emerging (eg. optogenetics and DREADD) in conjunction with the cre/loxP system well established, it proves to be an exciting time to examine the intricacies in the fundamental features of how metabolic cues influence reproductive physiology.

Highlights.

• Reviews current state of research in the metabolic control of reproduction

• Analyses the use of gene targeting strategies and novel mouse models

• Highlights neural circuitry and potential signaling pathways through which metabolic hormones influence the reproductive axis