A Shot in the Dark Exposes More Trees in the Forest: Adding a Histone Demethylase (Jmjd3) and RIP-cre Neurons to the Coordination of Female Reproductive Function

Rat insulin promoter (RIP)-cre animals were originally generated to evaluate gene activity in pancreatic β-cell development and function (1, 2). Ectopic cre-recombinase activity was later discovered in the brain (3), with particular concentration in the hypothalamus. Based on that discovery, subsequent work has leveraged this model to evaluate neuronal control of metabolism (4–7). The work reported by Song et al. (8) in this month’s issue of Endocrinology characterizes reproductive physiology of RIP-cre animals.

Song et al. (8) chose to disrupt Jmjd3, a gene that generates a histone-demethylating enzyme. Specifically, Jmjd3 removes methylation from the trimethylated (me3) Histone 3 (H3) lysine 27 (K27). The authors rationalized this choice by citing a recent, solitary finding; that is, H3K27me3 abundance is higher in juvenile female rats than in late-pubertal female rats, near the kisspeptin (kiss1) gene in the arcuate nucleus (ARC) (9). Given their own scant evidence that RIP-cre–positive cells coexpress kisspeptin in the ARC and minimally coexpress kisspeptin near the anteroventral periventricular nucleus (AVPV) (8), it would seem that expectations of manipulated reproductive function in RIP-cre–Jmjd3 conditional knockout (Rip-jmjd3cKO) animals are more or less shots in the dark. To that notion, we may someday invoke the adage that the most prominent discoveries come from the least likely places. Until that day, the most pragmatic evaluation of their findings parses the paper into three distinct discoveries: (1) the influence of Jmjd3 activity in RIP-cre neurons on reproductive physiology; (2) the interaction of Jmjd3 with the kiss1 gene, and the sensitivity of this interaction to estradiol (E2) in the AVPV; and (3) the influence of E2 and nonclassical estrogen receptor (ERα) signaling on Jmjd3 expression. Each of these findings represents intriguing and potentially very important pieces of information (i.e., the “trees”); however, caution is warranted in considering these data as a unified mechanism (i.e., “the forest”), as suggested by the article title.

To begin, Song et al. (8) describe the reproductive physiology phenotype of Rip-Jmjd3cKO female mice. Most notably, these animals exhibit slightly delayed vaginal opening, abnormal estrus cycles (with significantly less time spent in proestrus), subfertility, low plasma estradiol levels, and atypical ovarian histology (i.e., few corpora lutea). In addition, female Rip-Jmjd3cKO animals exhibit late-onset obesity and reduced energy expenditure. However, this was shown to be secondary to estrogen deficiency, suggesting a primary effect of atypical reproductive function/ovarian cycling.

Based on the abnormal estrus cycles, few corpora lutea, and low estradiol, the authors further hypothesize that Jmjd3 participates in the induction of E2-induced preovulatory luteinizing hormone (LH) surges (8). This mechanism is already known to depend on classical, ligand-dependent signaling of ERα (10) in kisspeptin neurons of the AVPV (11). Song et al. (8) show that kisspeptin is, indeed, coexpressed with a small proportion of RIP-cre neurons. However, the coexpressing neurons stray from the characteristic pattern of E2-sensitive kisspeptin neurons tightly lining the ventricle in the AVPV. This raises the question as to whether the reproductive physiology noted in Rip-Jmjd3cKO animals is directly related to kiss1 gene activity, as suggested by the article title.

Importantly, although the authors show that E2 exposure mitigates the impact of conditional Jmjd3 disruption on body weight and predictably influences kiss1 expression in the AVPV and ARC of ovariectomized wild-type animals (8), notably missing is the influence of E2 on kiss1 expression in Rip-Jmjd3cKO animals. Without indication that E2-induced kiss1 expression is compromised in Rip-Jmjd3cKO animals, we cannot conclude that any of the other molecular data presented by Song et al. (8) have any relevance to the physiology observed in Rip-Jmjd3cKO animals. In addition, and perhaps more fundamentally necessary, there is no evidence that Jmjd3cKO influences E2-induced LH surges. This can be easily accomplished using a positive-feedback scheme and, again, is necessary to connect the noted physiology to atypical E2 signaling.

Even given these missing pieces, connecting conditional gene deletion in RIP-cre neurons to atypical reproductive function/ovarian cyclicity hints at participation of yet-to-be-identified hypothalamic neurons that influence reproductive function. Given the widespread hypothalamic expression of cre recombinase in RIP-cre animals, it is possible, or perhaps probable, that the affected neurons are already characterized for their participation in reproductive function (e.g., POMC, AgRP, NPY, oxytocin). However, as we enter an era with exceptional new tools [e.g., DROP-sEquation (12)] for discovery of hypothalamic cell phenotypes, we may find that the RIP-cre animal has, indeed, unveiled an important mediator of female reproduction.

Outside of the Rip-Jmjd3cKO animal phenotype, Song et al. (8) report intriguing and clear evidence that Jmjd3 participates in E2-mediated kiss1 expression in the AVPV—an apparently critical step toward generation of LH surges. Most importantly, they show that E2 influences the interaction of Jmjd3 with the kiss1 5′ region, and the abundance of H3K27me3 (a repressive histone modification) at the kiss1 gene in the AVPV of wild-type animals. Together, these data suggest Jmjd3 participates in the E2-mediated creation of a more permissive chromatin environment at the kiss1 gene in the AVPV. Intriguingly, however, based on their in vitro work, the influence of Jmjd3 on kiss1 expression may not depend on the H3K27me3 demethylating activity. Specifically, they show that combining Jmjd3 overexpression with kiss1 promoter-driven luciferase reporter constructs, which do not interact with histones like native DNA, results in increased luciferase activity in 293T cells (8). This suggests a direct DNA-based, not histone-based, influence of Jmjd3 on kiss1 expression.

Importantly, Song et al. (8) point out that Jmjd3 directly interacts with ERα and potentiates ERα activity in MCF-7 cells (8, 13). The importance of this potentiation is covered later, but first, the clear influence of E2 on Jmjd3 association with the kiss1 gene and consequential reduction in H3K27me3 suggest Jmjd3, indeed, participates in an E2-mediated epigenetic regulation of AVPV kiss1 expression. To this end, it is possible that demethylation of H3K27me3 is necessary for displacement of the PRC2 complex, which would enable the AVPV-specific chromatin looping previously, and eloquently, described by Tomikawa et al. (14). If this is the case, Song et al. (8) appear to be the first to identify an enzyme participating in the well-characterized (14, 15) epigenetic and conformational changes related to AVPV-specific E2 stimulation of kiss1 expression.

Last, Song et al. (8) show that AVPV Jmjd3 expression is itself responsive to E2 exposure. Based on this, the authors suggest an E2-mediated rise in Jmjd3 is critical to E2 stimulation of AVPV kisspeptin. Again, this conclusion is not exactly warranted, but the rapid E2-driven rise in Jmjd3 expression does hint at an intriguing possible point of Jmjd3 participation in AVPV-driven LH surges. Interestingly, E2 drives Jmjd3 expression through nonclassical ERα signaling through PI3k, which, at first glance, might seem to contradict what we already know; that is, E2-mediated AVPV kiss1 expression depends on classical ERα signaling (10). However, given that Jmjd3 expression rises within 2 hours of E2 exposure in ovariectomized animals (8), it is possible nonclassical ERα signaling drives expression of the ERα potentiator (Jmjd3), which may be necessary for maximal, rapid expression of kiss1 before an LH surge.

Importantly, the previously described Jmjd3 potentiation of ERα activity in MCF-7 cells is related to demethylation of H3K27me3. In fact, if the opposing enzyme (Ezh2, a H3K27 methylator) was eliminated, H3K27 remained nonmethylated and led to constitutive expression of the associated gene (13). This raises an important point in terms of the LH surge. If, in fact, Jmjd3-mediated demethylation is related to E2-induced upregulation of kiss1 and the eventual LH surge, then is Ezh2 also necessary to shut down the surge generator? Presumably this must be so. In this case, we will find ourselves with another ying-yang (i.e., inhibition-activation) relationship in the control of female fertility—this time at the molecular level. Critical questions then emerge. What factors control Ezh2 and Jmjd3 before, during, and after the LH surge? How might environmental conditions impinge on these mechanisms? How might we leverage this information to generate novel approaches to infertility treatments?

Clearly, the work by Song et al. (8) raises many questions and inspires speculation; however, seeing the forest (physiology) rather than the trees (Jmjd3 and RIP-cre neurons) still depends on a substantial degree of conjecture. Nonetheless, the findings are different and intriguing, and filling in some of the blanks noted here will help the field move forward.