A critical view of the use of genetic tools to unveil neural circuits: the case of leptin action in reproduction

Abstract

The remarkable development and refinement of the Cre-loxP system coupled with the nonstop production of new mouse models and virus vectors have impelled the growth of various fields of investigation. In this article, I will discuss the data collected using these genetic tools in our area of interest, giving specific emphasis to the identification of the neuronal populations that relay leptin action in reproductive physiology. A series of mouse models that allow manipulation of the leptin receptor gene have been generated. Of those, I will discuss the use of two models of leptin receptor gene reexpression (LepR^neo/neo and LepR^loxTB/loxTB) and one model of leptin signaling blockade (LepR^flox/flox). I will also highlight the differences of using stereotaxic delivery of virus vectors expressing DNA-recombinases (Flp and Cre) and mouse models expressing Cre-recombinase. Our findings indicate that leptin action in the ventral premammillary nucleus is sufficient, but not required, for leptin action in reproduction and that leptin action in Kiss1 neurons arises after pubertal maturation; therefore, direct leptin signaling in Kiss1 neurons is neither required nor sufficient for the permissive action of leptin in pubertal development. It also became evident that the full action of leptin in the reproductive neuroendocrine axis requires the engagement of an integrated circuitry, yet to be fully unveiled.

over the past 10 years, the scientific literature has been inundated with studies using genetic manipulation to define brain circuitries related to specific functions. In particular, the remarkable development and refinement of the Cre-loxP system coupled with the nonstop production of new mouse models and virus vectors have impelled the growth of various fields of investigation. In this article, I will discuss the data collected using these genetic tools in our area of interest, giving specific emphasis to the identification of the neuronal population(s) that relay leptin action in reproductive physiology. It is my intention to present a critical view of the experiments performed in our laboratory and discuss our choices, the pros and cons of the techniques employed, and the problems we faced using these tools.

Leptin Action in Reproductive Neuroendocrine Axis

The role of the adipocyte-derived hormone leptin in the control of the reproductive physiology has been extensively studied by different laboratories worldwide, and many reviews exploring different aspects of leptin actions have been published in recent years (12, 53, 66, 84, 120). Briefly, it is now well accepted that leptin has two major roles in reproductive function: 1) it acts as a permissive factor for the pubertal development (2, 5, 25, 52) and 2) it signals energy sufficiency for the neuroendocrine reproductive axis in adult life (4, 45, 64, 99, 127). This concept originated from data in mice and humans with mutations in the leptin (Lep/LEP) or leptin receptor (Lepr/LEPR) genes and from experimental data showing recovery of the reproductive system in conditions of negative energy balance upon leptin replacement. Mice that are homozygous for loss-of-function mutations in Lep or Lepr genes (ob/ob, for obese and db/db for diabetic, respectively) are morbidly obese and display multiple neuroendocrine abnormalities (30, 69, 136). Males and females are infertile and do not undergo puberty. The gonads display morphological and biochemical abnormalities, and circulating levels of sex steroids, luteinizing hormone (LH), and follicle-stimulating hormone (FSH) are decreased in both sexes (7, 8, 51, 74, 97, 118).

In humans, leptin signaling deficiency due to genetic mutations in the LEP or LEPR genes is rarely found, but individuals with either mutation are obese and infertile (27, 57–59, 63, 81, 90, 91, 104). Lack of a pubertal growth spurt and secondary sexual characteristics are apparent. Recombinant leptin therapy in leptin-deficient subjects induces puberty and restores fertility (55, 56).

Likely because of the high energetic costs of the reproductive function, such as pregnancy and lactation, states of negative energy balance rapidly inhibit reproduction (39, 52, 61, 62, 72, 107, 121). Rodents and primates subjected to fasting exhibit decreases in sex steroids, pulsatile LH secretion, and fertility (15–17, 87, 88, 103, 128). Leptin administration prevents the fasting-induced suppression of LH secretion, restores female cyclicity, and improves fertility (4, 45, 64, 99, 127) (but see also Ref. 123). In women with hypothalamic amenorrhea caused by extreme weight loss, leptin treatment increased LH and sex steroid levels and improved ovarian physiology (22, 82, 94, 126, 129).

Leptin has biological effects in every node of the hypothalamus-pituitary-gonads axis (HPG) (5, 12, 53, 120), but in vitro and in vivo studies have determined that leptin acts primarily in the brain by indirectly stimulating gonadotropin-releasing hormone (GnRH) neurons and, consequently, activating the HPG axis (77, 78, 102, 131, 134). In our laboratory, we have been interested in deciphering the mechanisms by which leptin modulates the reproductive neuroendocrine axis and acts as a key metabolic cue for the reproductive system. The initial step was to define the anatomical site(s) engaged in this function. To do that, we used a series of genetically modified mouse models and virus vectors, discussed in detail below.

Generation of Mice for Manipulating the Leptin Receptor Gene

LepR is a member of the class 1 cytokine receptors and has six isoforms (3, 26, 60, 96, 98, 119, 135). All isoforms have a common extracellular ligand-binding domain and, except for one of them (soluble LepR-LepRe), an identical transmembrane domain. The long form of LepR (LepRb) contains the full intracellular domain that is required for the activation of STAT signaling, while the short forms contain intracellular domains of variable sizes with weak signaling capacity (11).

The diabetic db/db mouse has a single nucleotide mutation in the leptin receptor (Lepr) gene, resulting in a 106-nucleotide insertion at the junction where the long and the short forms of the receptor diverge (24). This inserted sequence leads to premature termination of the long intracellular domain, generating a LepR reminiscent of the short isoforms. In the db/db mutation, a truncated LepR with intact extracellular and transmembrane domains is produced, and, therefore, leptin can still bind to its receptor but virtually no or very weak intracellular signaling is triggered. As a consequence, the obese, diabetic and infertile phenotype of these mice is believed to be due to lack of the signaling, intracellular, domain characteristic of the LepRb (long) isoform. This concept is in line with seminal data suggesting that the obesity syndrome of db/db and ob/ob mice is indistinguishable when they are bred to the same genetic background (29).

The use of genetic strategies to delete or express the Lepr gene in various organs and tissues gave rise to the idea that most of leptin's effect is mediated by the brain. For instance, db/db (null for LepRb) and db^3j/db^3j (null for all LepR isoforms) mice with a transgene expressing the long form of LepR under the control of the neuron-specific enolase (NSE) or synapsin I promoters were rescued from most of their obese, diabetic, and infertility phenotypes (41, 73). In addition, selective neuronal deletion of LepR was achieved by crossing synapsin I-Cre transgenic mice (137) with ObR^flox/flox mice, which express loxP sites flanking exon 1 of the leptin receptor gene. The resulting mouse model had low levels of all LepR isoforms in neurons and displayed increased adiposity, hyperinsulinemia, and diabetes (28).

Once the brain was defined as a prime target of leptin, the next challenge was to identify the brain sites or neuronal cell groups that mediate specific physiological functions, e.g., food intake, energy expenditure, and reproduction. In the attempt to solve this issue, several laboratories have developed mouse models to selectively delete or reexpress the endogenous LepR in defined neuronal populations. While the ObR^flox/flox mouse is an excellent model to investigate the role of all LepR isoforms, the generation of the LepR^flox/flox mouse model allowed a more detailed assessment of the role of the LepRb in specific functions. The LepR^flox/flox mouse has loxP sites flanking exon 17 of the Lepr gene, a region that encodes the JAK docking site required for STAT signaling (92). Cre-mediated deletion of the flanked region results in blockade of the leptin-induced JAK-STAT signaling pathway, while LepR short forms remain intact. The LepR^flox/flox mouse model has been used by many laboratories and has been extremely valuable for unraveling the brain circuitry related to leptin signaling (6, 44, 47, 65, 67, 79, 106, 113, 124, 125).

The LepR^flox/flox mouse was generated using a strategy defined by the authors as the “allelogenic” approach (93), i.e., in a single gene targeting event, the Lepr allele was designed to contain the recognition sequences of two different DNA recombinases: frt and loxP. The LepR allelogenic mouse contains an insertion of the phosphoglycerate kinase promoter-driven neomycin phosphotransferase cassette (PGK-neo) sequence flanked by frt sites into intron 16 and loxP sites flanking the frt-PGK-neo-frt cassette and exon 17 of the Lepr gene. The presence of the PGK-neo cassette blocks the full expression of the Lepr gene, producing a LepR with virtually no JAK-STAT signaling capacity, i.e., a mouse null for the LepRb (LepR^neo/neo) with a phenotype similar to that of the db/db mouse. Flp recombinase-mediated deletion of the frt-flanked neo cassette produces a normally functioning Lepr allele with exon 17′ (JAK docking site) flanked by loxP sites (LepR^flox/flox mouse model).

More recently, another LepR-null reactivable mouse model was generated to allow the use of the Cre-loxP system to endogenously reexpress LepR (10). In this mouse model (LepR^loxTB/loxTB), a loxP-flanked transcription-blocking cassette (loxTB) was inserted between exons 16 and 17 of the Lepr gene, resulting in a mouse that globally lacks the LepR intracellular domain and is obese, diabetic, and infertile. Cre-mediated excision of loxP sites generates LepR with signaling capacity. Because of the widespread use and remarkable development of the Cre-loxP system, the LepR^loxTB/loxTB mouse model is a crucial tool to the investigation of leptin physiology. The LepR^neo/neo mouse, however, has a more limited use due to the lack of resources or mouse models expressing Flp-recombinase driven by specific genes and, consequently, the need to use stereotaxic delivery of Flp-recombinase for genetic manipulation in the brain. However, this mouse has also been explored by different laboratories and was the starting point of our study on leptin action in reproductive function using genetically modified mouse models.

Endogenous Reexpression of LepR: Use of LepR^neo/neo and Virus Vectors

Following the description of the distribution of LepR in the rodent brain, it became clear that leptin potentially engages multiple brain circuits to exert its function. The role of specific cell groups was initially predicted on the basis of the outstanding hypothalamic mapping performed by different laboratories (18, 19, 21, 109, 110, 116). Regarding reproductive control, one hypothalamic site immediately attracted our attention: the ventral premammillary nucleus (PMV) (48). The PMV projects to brain areas related to reproductive function, and LepR-expressing cells in this site directly innervate GnRH neurons (13, 19, 80, 83, 108). Previous studies have shown that sexually relevant odors stimulate PMV neurons and that excitotoxic lesions targeting this area preclude odor-induced LH secretion and disrupt female cyclicity in rodents (9, 20, 45, 46, 50, 80, 133). Because only a subset of PMV neurons expresses LepR (80), studies using excitotoxic lesions of the entire neuronal population are only suggestive of a certain function. One could argue that non-LepR neurons are the crucial players. With this in mind, we opted to employ genetic tools and mouse models to assess the role of specific neuronal populations (i.e., those responding to leptin) and the contribution of LepR to the phenotype previously observed using neuronal lesions. As a first step, we assessed whether leptin signaling only in PMV neurons is sufficient to restore the infertility phenotype of the LepR-null mice. To do so, we used stereotaxic delivery of an adeno-associated virus vector (AAV) expressing Flp-recombinase and eGFP (AAV-Flp-eGFP) into the PMV of the LepR-null (LepR^neo/neo) mouse (34, 47) (Fig. 1). The efficacy of the method was determined by assessing the ability of leptin to induce JAK-STAT signaling in cells targeted by the virus vector (AAV-Flp-eGFP); in other words, by mapping the distribution of leptin-induced phosphorylation of STAT3 immunoreactivity in the hypothalamus. Reactivation of LepR in PMV neurons of LepR-null mice induced puberty and improved the fertility of females (47). No amelioration of their metabolic phenotype was observed; the mice were still very obese and diabetic. Therefore, this finding indicated that dissociation between brain sites controlling metabolism and reproduction exists.

Fig. 1.

Reexpression of leptin receptor (Lepr) in the ventral premmammilary nucleus (PMV) neurons using the LepR^neo/neo mouse model and stereotaxic delivery of adeno-associated virus vector expressing Flp-recombinase (AAV-Flp-eGFP). A: LepR^neo/neo mouse has a transcriptional blocker (PGK-neo) cassette flanked by frt sites inserted between exons 16 and 17, precluding the expression of full intracellular domain of LepR. B: AAV-Flp-eGFP virus was delivered unilaterally into the PMV. C: reexpression of functional LepR after step B.

We observed that 90% of females with successful reactivation of LepR in PMV neurons entered puberty and showed some degree of sexual maturation, as defined by increased uterine weight and the presence of ovarian corpora lutea. These mice were housed with sexually experienced males for 6 wk, and 50% were fertile. Unexpectedly, the males did not respond to the genetic manipulation, i.e., reactivation of LepR in the PMV neurons of adult LepR-null male mice did not restore their infertility. As reported for ob/ob mice (75), we found that 10% of the LepR-null mice with or without LepR reactivation in PMV neurons were fertile. The findings of this experiment were remarkable, and, as usual, several questions emerged, and others remained unanswered. Why were only 50% of the females fertile while 90% showed sexual maturation? Is it a matter of time? Is it because of the inherent variability of the technique? As expected, for stereotaxic injections, the number of PMV neurons showing LepR reactivation was variable among the animals. Could the massive obesity and diabetes interfere with the progression of ovulation (anovulatory cycles), implantation, or embryo development? What about the males? Is there a sexually dimorphic site involved in the metabolic control of reproduction? Could the metabolic dysfunction also change the outcome of LepR reactivation in males?

Several experimental details also deserve a closer look. These mice were adults at the time of the LepR reactivation; therefore, we were able to bypass the developmental adaptations frequently observed in studies using genetic manipulation early in life (86, 89, 132). However, leptin also has an effect on brain circuitry development (1, 14), which may have produced definitive deficits and the resultant lack of response (partial in female and complete in male) due to deficient innervation of related nuclei. The use of adult animals was also a challenge because the LepR-null mice were fairly obese, which may have concealed potential positive data due to the secondary effects of the obesity and comorbidities (e.g., diabetes and hypercorticosteronemia). The injections were performed unilaterally, a choice based on the difficulty of specifically targeting a nucleus with the size of the PMV (∼300 μm wide). Unilateral injections increased our chances of hitting only PMV neurons. Because it is not possible to know a priori the number of neurons necessary to attain a physiological response, the data could have been more robust if bilateral injections had been performed. This leads to another question: if stereotaxic injections are so challenging and bilateral reactivation of PMV neurons could have produced stronger data, why not use a mouse model to drive Flp recombinase expression selectively in PMV neurons instead of AAV injections? To answer this question, a few variables must be considered. Depending on the gene driving Flp expression, this alternative would generate a different model: a mouse with LepR expression only in PMV neurons during specific developmental stages. While this procedure would answer some important and different questions, mouse models with these characteristics are not available. New technology or resources must be generated to allow this approach.

Requirement of Leptin Signaling in PMV Neurons for Reproduction

Endogenous reexpression of LepR selectively in the PMV neurons of adult mice yielded the exciting finding that leptin action only in the PMV is sufficient to induce puberty and improve fertility in female mice (47). However, the “reexpression approach” is not appropriate to determine whether this specific pathway is required for the physiological role of leptin in the reproductive axis. It is plausible that other pathways mediate these effects in intact mice without the need for PMV neuron engagement. To assess whether leptin action in PMV neurons is required for the female reproductive function, we initially used the LepR^flox/flox mouse and stereotaxic delivery of AAV-expressing Cre recombinase and eGFP (AAV-CRE-eGFP; Vector Laboratories) to eliminate LepR from PMV neurons. As described in previous sections, leptin exerts two well-defined functions in reproductive physiology: it acts as a permissive factor for pubertal initiation, and it signals energy sufficiency to allow fertility in adult life (53). To assess the requirement of LepR in PMV neurons for the permissive effect of leptin on puberty onset, we designed an experimental protocol to delete LepR only from the PMV neurons of prepubertal female mice using bilateral injections of AAV-Cre-eGFP (Fig. 2). Because vaginal opening, an external sign for initiation of puberty onset in female mice, is observed between postnatal days (PND) 23–28 in C57BL/6 mice, we injected the virus vector into the PMV of 19-day-old mice (PND19) to allow enough time for infection of the cells and expression of Cre-recombinase and DNA recombination. This experiment was extremely laborious as PND19 female mice are very small for this type of approach (average weight = 9 g), and, again, it is not possible to know a priori the number of neurons engaged in a certain function; bilateral and complete deletion of LepR from PMV neurons would likely be necessary to test our model. Following several attempts, we concluded that the experimental design would not work as planned. In our hands, stereotaxic injections in prepubertal mice delayed puberty onset independent of the injection sites (Elias CF and Donato J, Jr, personal observation), likely due to the stress caused by the surgical procedure. This approach was discarded until a mouse expressing Cre-recombinase selectively in PMV neurons can be used to test our model.

Fig. 2.

Strategy to delete leptin signaling from the PMV using the LepR^flox/flox mice and stereotaxic delivery of adeno-associated virus vector expressing Cre recombinase (AAV-Cre-eGFP). A: LepR^flox/flox mice has loxP sites flanking the exon 17 of Lepr gene, which encodes the JAK-docking site. B: AAV-Cre-eGFP was delivery bilaterally into the PMV of prepubertal mice (postnatal day/PND19). This strategy did not work as planned. C: the expected result: deletion of exon 17 and blockade of leptin signaling in PMV neurons.

Alternatively, we obtained provocative data using the ob/ob mouse and bilateral excitotoxic lesions of PMV neurons (47). The obese ob/ob mouse remains in a prepubertal condition unless treated with leptin (7, 23), a procedure that induces pubertal development and rescues fertility. With this in mind, we were able to use adult female ob/ob mice and produce bilateral lesions of the PMV before leptin treatment and after full recovery from the stress of the surgery. We asked whether ablation of PMV neurons would abrogate the effects of leptin replacement on the reproductive axis of ob/ob mice. Although no difference in time for vaginal opening was noticed, ob/ob females with successful bilateral PMV lesions had a significant delay in sexual maturation. They also failed to show an increase in LH secretion after acute leptin administration (47). Interestingly, a recent study reported a similar effect using nNOS-Cre crossed with the LepR^flox/flox mouse model (79), i.e., female mice with deletion of LepR from the neurons expressing neuronal nitric oxide synthase (nNOS) showed the regular time for vaginal opening but delayed sexual maturation. Importantly, the PMV shows the highest rate of colocalization between nNOS and LepR of the entire brain (49, 79); therefore, one can postulate that this effect is the result of the lack of leptin signaling in PMV neurons. However, several aspects should be noted: 1) in our experiments using excitotoxic lesions, virtually all neurons, not only those expressing LepR, are ablated, and, therefore, a lack of specificity is apparent, as discussed in a previous section; 2) deletion of LepR in nNOS neurons caused severe obesity; it is not known whether the delayed sexual maturation observed in this mouse model is secondary to the obesity phenotype and comorbidities; and 3) nNOS and LepR are coexpressed in other hypothalamic nuclei and, therefore, leptin action in an alternative nNOS neuronal population may also play a role.

Collectively, these findings indicate that leptin action in PMV neurons is required for the normal progression of sexual maturation, but puberty still occurs, suggesting that the PMV is not the only site relaying leptin action in reproduction. This is, in fact, highly expected because of the predicted redundancy of brain pathways with a role in reproductive physiology, which increases the probability of species survival.

Searching for Alternative Neuronal Population(s) that Relay Leptin Action in Reproduction

In search of neuronal population(s) outside of the PMV that relays leptin action in reproduction, we turned our attention to kisspeptin neurons (31, 101, 105). In mice and humans, loss-of-function mutations in kisspeptin (Kiss1/KISS1) or kisspeptin receptor (Gpr54/GPR54, also known as Kiss1R/KISS1R) genes result in infertility due to a lack of pubertal maturation and to hypogonadotropic hypogonadism (40, 42, 76, 114, 122). The role of kisspeptins as a key mediator of leptin action in reproduction was proposed on the basis of findings from several laboratories showing that Kiss1 gene expression is usually decreased in conditions of low leptin levels and that Kiss1 and LepR mRNAs are coexpressed in a subset of neurons of the arcuate nucleus (Arc) (71, 85, 117). The Arc is a hypothalamic site that houses several groups of neurons well known to mediate the action of leptin in metabolism (32, 33, 54, 111, 130). These groups include neurons expressing proopiomelanocortin (POMC) and agouti-related protein (AgRP), which are both part of the melanocortin system. POMC and AgRP neurons comprise distinct populations of LepR-expressing cells with complementary functions in leptin physiology. POMC neurons are stimulated by leptin, whereas AgRP neurons are inhibited by leptin (35, 36, 68, 95); α-melanocyte-stimulating hormone (αMSH, a POMC product) inhibits food intake, whereas AgRP increases food intake (33, 95). Because of this intermingled distribution of neurons that express LepR (POMC, AgRP, and Kiss1) in the Arc, the use of stereotaxic injections of a virus vector to deliver Flp- or Cre-recombinase is compromised since the outcome would be difficult to interpret. To circumvent this problem, we generated a mouse model with Cre-recombinase expressed in Kiss1 neurons. This mouse was produced using BAC transgenics and was validated using the distribution of two reporter genes (eGFP and LacZ) and the colocalization with Kiss1 mRNA in adult female mice under low and high levels of sex steroids (38). This approach was used due to the remarkable changes in Kiss1 mRNA expression observed in different steroid milieus in distinct brain nuclei expressing the Kiss1 gene. Using this mouse, we initially assessed whether leptin action in Kiss1 neurons is required for pubertal development. We deleted LepR from Kiss1 neurons by crossing our Kiss1-Cre mouse model with the LepR^flox/flox mouse (Fig. 3A). Male and female Kiss1-Cre LepR^flox/flox mice showed no deficit in pubertal development, fertility, and fecundity (47). These mice also had a normal body weight, food intake, and body mass index until the end of the experiment (35 wk of age), demonstrating that the direct action of leptin in Kiss1 neurons is not required for sexual maturation and metabolic regulation.

Fig. 3.

Deletion and reexpression of leptin signaling in Kiss1 neurons. A: Kiss1-Cre mouse was crossed with LepR^flox/flox mouse to generate mice with selective deletion of leptin signaling in Kiss1 neurons. B: Kiss1-Cre mouse was crossed with LepR^loxTB/loxTB mouse to generate mice with selective reexpression of leptin signaling in Kiss1 neurons. A and B: images show an example of Cre expression in the arcuate nucleus (Arc) of Kiss1-Cre reporter (eGFP) mouse.

This surprising observation led us to speculate that developmental adaptations or circuit redundancy might have overcome the lack of a key physiological pathway (e.g., kisspeptin) to allow reproductive development and species survival. Following this reasoning, we asked the next question: what if we avoid the redundant or parallel circuits? Or, in other words, is leptin action only in kisspeptin neurons sufficient to allow puberty onset and improve fertility in LepR-null mice? To determine the answer, we crossed the Kiss1-Cre mice with LepR^loxTB/loxTB mice to generate a mouse model with LepR expression only in Kiss1 neurons (37) (Fig. 3B). Kiss1-Cre LepR^loxTB/loxTB male and female mice showed no sexual maturation and no improvement of any aspect of the reproductive physiology until 6 mo of age. They also remained very obese and diabetic, showing no improvement in their metabolic dysfunction. These negative data were, at first sight, somewhat disappointing, but a deeper look revealed an unexpected and remarkable finding. The presence of Cre activity was confirmed by the recombination assay, and both mouse models had been validated and used before in previous publications (10, 38, 47); however, we were unable to demonstrate convincing reexpression of LepR in Kiss1 neurons in LepR-null (Kiss1-Cre LepR^loxTB/loxTB) mice. This was assessed by both the colocalization of leptin-induced pSTAT3 immunoreactivity and Kiss1 mRNA expression using in situ hybridization and by the expression of the Kiss1-Cre reporter gene (GFP) and LepR mRNA using in situ hybridization. However, these mice remained in a prepubertal state, and the Kiss1 mRNA expression was very low, making it difficult to be fully detected using in situ hybridization. The same challenge was evident for the LepR mRNA: low expression, a very common characteristic of membrane receptors.

To enhance the visualization of Kiss1 neurons and to assess Cre activity in these cells, we added a reporter gene (tdTomato) to our Kiss1-Cre LepR^loxTB/loxTB mouse model. With this approach, we were able to demonstrate that, as previously indicated by others using in situ hybridization, leptin signaling-deficient mice display decreased numbers of Kiss1-expressing neurons in the Arc and in the anteroventral periventricular nucleus (37, 117). This finding is also in agreement with the literature, demonstrating that the level of Kiss1 mRNA is lower in prepubertal mice (100, 115).

We next asked whether the decreased number of neurons expressing Kiss1 in the Arc could be the cause of the decreased coexpression of Kiss1 and LepR at that site. An alternative approach to answer this question would be the identification of possible differences in coexpression levels during postnatal development. Because of the lack of good procedures to detect kisspeptin and LepR immunoreactivity or gene expression in wild-type mice, we generated a mouse model that expresses human renilla GFP (hrGFP) driven by Kiss1 regulatory elements and crossed this mouse with a line showing tdTomato fluorescence in LepR-expressing cells (LepR-Cre tdTomato or LepR reporter mice). The LepR-Cre mouse is a previously described and validated knock-in model (43, 112), and the Kiss1-hrGFP was generated using the same BAC that was used for the Kiss1-Cre mouse, being validated also in female mice in different steroid conditions. Using this approach, we found no colocalization of the two reporter genes (hrGFP and tdTomato) in prepubertal mice, revealing that leptin signaling in Kiss1 neurons arises after completion of sexual maturation, i.e., in adult life (37). We were not able to reactivate LepR in Kiss1 neurons of LepR-null mice likely because the lack of global leptin signaling in the LepR-null mice precludes sexual maturation. These findings indicate that direct leptin signaling in Kiss1 neurons is neither required nor sufficient to induce puberty and improve the fertility of LepR-null mice.

If not Kisspeptin, Then What?

Using genetically modified mouse models and conditional deletion and reexpression of the Lepr gene, we demonstrated that kisspeptin neurons are not the direct players that link leptin action and reproductive function. An alternative neuronal population remain unidentified. Recently, two studies from different laboratories using distinct approaches have suggested a role for the melanocortin system (70, 132). In one study, the authors used the cytotoxic property of diphtheria toxin to ablate AgRP neurons and evaluated the effects of the lack of AgRP neurons in ob/ob mice. Lack of leptin signaling, as observed in ob/ob and db/db mice, generates a model of sustained high AgRP expression in the Arc (32, 33). By ablating AgRP neurons from ob/ob mice, the authors investigated the role of elevated Agrp gene expression in the obese and infertile phenotype of these mice. The researchers generated a mouse model with human diphtheria toxin receptor (DTR) expressed selectively in AgRP neurons (DTR cDNA targeted to the Agrp locus). Exposure to diphtheria toxin (DT) would cause toxicity only in AgRP neurons, ultimately leading to cell death. The authors found that ablation of AgRP at the young adult age (body weight = 35–40 g) leads to improvement in the obesity and infertility of ob/ob mice (132). Curiously, a similar approach using younger (leaner) or older (fatter) mice resulted in death from starvation. Following the same line, db/db mice with global knockout of AgRP (db/db Agrp^−/−) or heterozygous for melanocortin receptor 4 knockout (db/db Mc4r^+/) showed improvement of the infertility phenotype (70). Interestingly, selective deletion of LepR from AgRP neurons (AgRP-Cre LepR^lox/lox) has little effect in metabolism and causes no reported reproductive deficit (124). These findings suggest that although AgRP neurons are not required for fertility, excess AgRP may suppress the reproductive axis.

Perspectives and Significance

The use of genetic tools has generated tremendous advances in scientific knowledge and, with it, great excitement in the scientific community. The potential is enormous. Uncountable manipulations and experimental designs have become possible, and unexpected findings have taught us unanticipated lessons. Far from being definitive, the techniques of in vivo genetic manipulations have called our attention to biological events that are usually disregarded in investigations using adult animals, such as developmental adaptations, plasticity, and redundancy. In this review, I have attempted to share our experience with these approaches and what we have learned from using them. Our findings indicate that leptin action in PMV neurons is sufficient but not required for leptin action in reproduction and that leptin action in Kiss1 neurons arises after pubertal maturation; therefore, leptin signaling directly in Kiss1 neurons is neither required nor sufficient for the permissive action of leptin in pubertal development. Importantly, it became evident that the full action of leptin in the reproductive neuroendocrine axis requires the engagement of an integrated circuitry, yet to be fully unveiled.

GRANTS

The research in my laboratory has been funded by grants from the National Institutes of Health (R01HD-061539 and R01HD-069702), the Foundation for Prader-Willi Research, Regent's Research Award and Young Researcher Award (from University of Texas Southwestern Medical Center, Dallas-TX) and startup funds from University of Michigan, Ann Arbor, MI.

DISCLOSURES

No conflicts of interest, financial or otherwise, are declared by the authors.

AUTHOR CONTRIBUTIONS

Author contributions: C.F.E. conception and design of research; C.F.E. analyzed data; C.F.E. interpreted results of experiments; C.F.E. prepared figures; C.F.E. drafted manuscript; C.F.E. edited and revised manuscript; C.F.E. approved final version of manuscript.