CRISPR-Cas9 knockdown of ESR1 in preoptic GABA-kisspeptin neurons abolishes the preovulatory surge and estrous cycles in female mice

Reviewer #1 (Public Review):

Summary:
The current study examines the necessity of estrogen receptor alpha (ESR1) in GABA neurons located in the anteroventral and preoptic periventricular nuclei and the medial preoptic nucleus of the hypothalamus. This brain area is implicated in regulating the pre-ovulatory LH surge in females, but the identity of the estrogen-sensitive neurons that are required remains unknown. The data indicate that approximately 70% knockdown of ESR1 in GABA neurons resulted in variable reproductive phenotypes. However, when the ESR1 knockdown also results in a decrease in kisspeptin expression by these cells, the females had disrupted LH surges, but no alterations in pulsatile LH release. These data support the hypothesis that kisspeptin cells in this region are critical for the pre-ovulatory LH surge in females.

Strengths:
The current study examined the efficacy of two guide RNAs to knockdown ESR1 in GABA neurons, resulting in an approximate 70% reduction in ESR1 in GABA neurons. The efficacy of this knockdown was confirmed in the brain via immunohistochemistry and the reproductive outcomes were analyzed several ways to account for differences in guide RNAs or the precise brain region with the ESR1 knockdown. The analysis was taken one step further by grouping mice based on kisspeptin expression following ESR1 knockdown and examining the reproductive phenotypes. Overall, the aims of the study were achieved, the methods were appropriate, and the data were analyzed extensively. This data supports the hypothesis that kisspeptin neurons in the anterior hypothalamus are critical for the preovulatory LH surge.

Weaknesses: One minor weakness in this study is the conclusion that the guide RNAs didn't seem to have unique effects on GnRH cFos expression or the reproductive phenotypes. Though the data indicate a 60-70% knockdown for both gRNA2 and gRNA3, 3 of the 4 gRNA2 mice had no cFos expression in GnRH neurons during the time of the LH surge, whereas all mice receiving gRNA3 had at least some cFos/GnRH co-expression. In addition, when mice were re-categorized based on reduction (>75%) in kisspeptin expression, most of the mice in the unilateral or bilateral groups received gRNA2, whereas many of the mice that received gRNA3 were in the "normal" group with no disruption in kisspeptin expression. Thus, additional experiments with increased sample sizes are needed, even if the efficacy of the ESR1 knockdown was comparable before concluding these 2 gRNAs don't result in unique reproductive effects.

Reviewer #2 (Public Review):

Clarkson et al investigated the impact of in vivo ESR1 gene disruption selectively in preoptic area GABA neurons on the estrogen regulation of LH secretion. The hypothalamic pathways by which estradiol controls the secretion of gonadotrophins are incompletely understood and relevant to a better understanding of the mechanisms driving fertility and reproduction. Using CRISPR-Cas9 methodology, the authors were able to effectively reduce the expression of estrogen receptor (ER)-alpha in GABA neurons located in the preoptic area of adult female mice. The results obtained were rather variable except in the animals with concomitant suppression of kisspeptin in the rostral periventricular region of the third ventricle (RP3V), which displayed interruption of ovarian cyclicity and an altered estradiol-induced LH surge. The experimental approach used allowed for a cell-selective, temporally-controlled suppression of ER-alpha expression, providing further evidence of the critical role of RP3V kisspeptin neurons in the estrogen positive-feedback effect. Nevertheless, the assessment of the estradiol-induced LH surge was limited to only one terminal blood collection. The preovulatory LH surge is a variable phenomenon and would require serial blood sampling for a conclusive evaluation of the surge occurrence or alteration, such as in shape, amplitude, or timing. The animals were not assessed for ovulation either, which might be a functional readout for the effectiveness of the LH surge. Thus, the actual effect on the preovulatory LH surge was not fully characterized. Finally, the study leaves unanswered the role of GABA itself. As there was no evident phenotype for the ESR1 knockdown in GABA neurons that do not coexpress kisspeptin, this suggests that GABA neurotransmission in the preoptic area is not involved in the estrogen regulation of LH secretion.

Reviewer #3 (Public Review):

Summary:
Although the key role of estrogen receptor alpha (ERα ; encoded by ESR1 gene) in the control of reproduction has been known for more than two decades, the identity of the neuronal population(s) underlying the control of the negative and positive feedbacks exerted by estrogens on the hypothalamic pituitary ovary axis has been more complicated to pinpoint. Among the factors contributing to the difficulty in this endeavor are the cellular heterogeneity of the preoptic area and hypothalamus and the pulsatile activity of the axis. Several neuronal populations have been identified that control the activity of GnRH neurons, the hypothalamic headmasters of the axis. Among them, the kisspeptin neurons of the rostral periventricular aspect of the third ventricle (RP3V) have been considered the major candidates to convey preovulatory estrogen signals to GnRH neurons, which do not express ERα. Yet, the existence of other populations of kisspeptin neurons (notably in the arcuate nucleus) has made it difficult to selectively ablate ESR1 in one population. A first study (Wang et al., 2019) reported that knocking down ESR1 specifically in RP3V kisspeptin neurons led to decreased excitability of Kp neurons, blunted spontaneous as well estrogen-induced preovulatory LH surge, and reduced or absent corpora lutea indicative of impaired ovulation, but cyclicity was left unaltered.

As GABAergic afferences to GnRH neurons are also implicated in mediating the effects of estrogens on the HPG axis, the present study sought to investigate their role in the positive feedback using genetically driven Crispr-Cas9 mediated knockdown of ERα in VGAT expressing neurons in a specific subregion of the preoptic area. To this end, they stereotaxically delivered viral vectors expressing validated guide RNA into the preoptic area and evaluated their impact on estrus cyclicity and the ability of mice to mount an LH surge induced by estrogens and associated activation of GnRH neurons assessed by the co-expression of Fos. The results demonstrate that knocking down ESR1 in preoptic gabaergic neurons leads to an absence of LH surge and acyclicity when associated with severely reduced kisspeptin expression suggesting that a subpopulation of neurons co-expressing Kp and VGAT neurons are key for LH surge since total absence of Kp is associated with an absence of GnRH neuron activation and reduced LH surge (although this was not confirmed by the post-hoc). Although the implication of kisspeptin neurons was highly suspected already, the novelty of these results lies in the fact that estrogen signaling is necessary for only a selected fraction of them to maintain both regular cycles and LH surge capacity.

Strengths:
Remarkable aspects of this study are, its dataset which allowed them to segregate animals based on distinct neuronal phenotypes matching specific physiological outcomes, the transparency in reporting the results (e.g. all statistical values being reported, all grouping variables being clearly defined, clarity about animals that were excluded and why) and the clarity of the writing. This allows the reader to understand clearly what has been done and how the analyses have been carried out. The same applies to the discussion which describes clearly possible interpretations as well as the limitations of this study based on a single in vivo experiment.

Another remarkable feature of this work lies in the analysis of the dataset. As opposed to the cre-lox approach which theoretically allows for the complete ablation of specific neuronal populations, but may lack specificity regarding timing of action and location, genetically driven in vivo Crispr-Cas9 editing offers both temporal and neuroanatomic selectivity but cannot achieve a complete knockdown. This approach based on stereotaxic delivery of the AAV-encoded guide RNAs comes with inevitable variability in the location where gene knockdown is achieved. By adjusting their original grouping of the animals based on the evaluation of the extent of kisspeptin expression in the target region, the authors obtained a much clearer and interpretable picture. Although only a few animals (n=4) displayed absent kisspeptin expression, the convergence of observations suggesting a central impairment of the reproductive axis is convincing.

In particular, the lack of activation of GnRH neurons in these mice despite a non-significant effect on the reduced LH levels in the post-hoc following a significant ANOVA (which is likely due to the limited number of concerned animals [n=4]), is convincing. Did the authors test whether LH concentrations correlated with the percentage of GnRH+ESR1 positive neurons? This could reinforce the conclusion.

Moreover, the apparent complete absence of kisspeptin expression in these 4 animals is compelling as it provides indirect confirmation of the key role of Kisspeptin neurons in this phenomenon using a different LH surge induction paradigm than Wang et al., 2019. Yet, the quality of the kisspeptin immunostaining in control animals does seem suboptimal and casts doubts about this conclusion (see the section on weaknesses for more details).

It is also interesting that a few animals with unilateral reduction in Kp expression also showed deficits in GnRH neuron activation suggesting that the impact of Kp may not be limited to the side of the brain where it is produced or the existence of a dose effect of Kp activation of GnRH neurons.

Finally, the observation that the pulsatile secretion of LH is maintained in the absence of Kp expression in the RP3V lends support to the notion that LH surge and pulsatility are regulated independently by distinct neuronal populations, a model put forward by the corresponding author a few years ago.

Weaknesses:
One aspect for which I have ambiguous feelings is the minimal level of detail regarding the HPG axis and its regulation by estrogens. This limited amount of detail allows for an easy read with the well-articulated introduction quickly presenting the framework of the study. Although not presenting the axis itself nor mentioning the position of GnRH neurons in this axis or its lack of ERα expression is not detrimental to the understanding of the study, presenting at least the position of GnRH neurons in the axis and their critical role for fertility would likely broaden the impact of this work beyond a rather specialist audience.

The expression of kisspeptin constitutes a key element for the analysis and conclusion of the present work. However, the quality of the kisspeptin immunostaining seems suboptimal based on the representative images. The staining primarily consists of light punctuated structures and it is very difficult to delineate cytoplasmic immunoreactive material defining the shape of neurons in LacZ animals. For some of the cells marked by an arrow, it is also sometimes difficult to determine whether the staining for ESR1 and Kp are in the same focal plane and thus belong to the same neurons. Although this co-expression is not critical for the conclusions of the study, this begs the question of whether Kp expression was determined directly at the microscope (where the focal plan can be adjusted) or on the picture (without possible focal adjustment). Moreover, in the representative image of Kp loss, several nuclei stained for fos (black) show superimposed brown staining looking like a dense nucleus (but smaller than an actual nucleus). This suggests some sort of condensed accumulation of Kp immunoproduct in the nucleus which is not commented. Given the critical importance of this reported change in Kp expression for the interpretation of the present results, it is important to provide strong evidence of the quality/nature of this staining and its analysis which may help interpret the observed functional phenotype.

As acknowledged in the introduction, this study is not the first to use in vivo Crisp-Cas editing to demonstrate the role of kisspeptin neurons in the control of positive feedback. Although the present work achieved this indirectly by targeting VGAT neurons, I was surprised that the paper did not include more comparison of their results with those of Wang et al., 2019. In particular, why was the present approach more successful in achieving both lack of surge and complete acyclicity? Moreover, why is it that targeting ESR1 in a selected fraction of GABAergic neurons can lead to a near-complete absence of Kp expression in this region? This is briefly discussed in the penultimate paragraph but mostly focuses on the non-kisspeptinergic GABAneurons rather than those co-expressing the two markers.