ERα in Tac2 Neurons Regulates Puberty Onset in Female Mice

Abstract

A variety of data suggest that estrogen action on kisspeptin (Kiss1)-containing arcuate nucleus neurons (which coexpress Kiss1, neurokinin B (the product of Tac2) and dynorphin (KNDy) neurons restrains reproductive onset and function, but roles for estrogen action in these Kiss1 neurons relative to a distinct population of rostral hypothalamic Kiss1 neurons (which does not express Tac2 or dynorphin) have not been directly tested. To test the role for estrogen receptor (ER)α in KNDy cells, we thus generated Tac2^Cre and Kiss1^Cre knock-in mice and bred them onto the Esr1^flox background to ablate ERα specifically in Tac2-expressing cells (ERα^Tac2KO mice) or all Kiss1 cells (ERα^Kiss1KO mice), respectively. Most ERα-expressing Tac2 neurons represent KNDy cells. Arcuate nucleus Kiss1 expression was elevated in ERα^Tac2KO and ERα^Kiss1KO females independent of gonadal hormones, whereas rostral hypothalamic Kiss1 expression was normal in ERα^Tac2KO but decreased in ERα^Kiss1KO females; this suggests that ERα in rostral Kiss1 cells is crucial for control of Kiss1 expression in these cells. Both ERα^Kiss1KO and ERα^Tac2KO females displayed early vaginal opening, early and persistent vaginal cornification, increased gonadotropins, uterine hypertrophy, and other evidence of estrogen excess. Thus, deletion of ERα in Tac2 neurons suffices to drive precocious gonadal hyperstimulation, demonstrating that ERα in Tac2 neurons typically restrains pubertal onset and hypothalamic reproductive drive.

Hypothalamic-pituitary-end organ feedback systems are crucial mediators of endocrine homeostasis. In the case of the female reproductive system, gonadal estradiol acts via estrogen receptor (ER)α to mediate negative feedback during most of the reproductive cycle, as well as appropriately timed positive feedback to promote the midcycle preovulatory surge in gonadotropins (1). Because the final output for the hypothalamic control of reproductive function, GnRH neurons do not appear to express ERα, other cells that express ERα likely mediate these responses to estradiol (1).

Hypothalamic neurons that produce the neuropeptide kisspeptin (Kiss1) are one likely mediator of central estradiol feedback. Humans with mutations in KISS1 or its receptor (KISS1R) exhibit hypothalamic infertility (2, 3), and rodent studies reveal crucial roles for Kiss1 or Kiss1r in the control of puberty onset and fertility (3–7). In many species, the hypothalamus contains 2 distinct populations of Kiss1-synthesizing neurons (8–11). One set, in the arcuate nucleus (ARC) in rodents, coexpresses Kiss1, neurokinin B (NKB) (the product of Tac2), and dynorphin (KNDy) cells (12–15). The second population of hypothalamic Kiss1-containing neurons does not contain NKB or dynorphin and lies in the rostral hypothalamus (largely in the anteroventral periventricular nucleus [AVPV] region and extending into the periventricular nucleus [PeN]) in rodents. Most cells in both hypothalamic populations of Kiss1 neurons contain ERα (7, 16, 17).

Kiss1 stimulates GnRH neurons (18–20), but although both populations of Kiss1 neurons appear to synapse on GnRH neurons, estradiol regulation of the 2 Kiss1 cell types differs; estrogen suppresses Kiss1 expression in the ARC while enhancing Kiss1 expression in the AVPV (8, 21–23). Many have hypothesized that the AVPV Kiss1 neurons mediate estrogenic positive feedback, whereas KNDy neurons mediate negative feedback (17, 21, 24–27).

Deletion of ERα from all Kiss1 neurons dramatically advances puberty onset and promotes a phenotype of early-onset gonadal hyperstimulation in mice (25, 26), suggesting that estrogen action via ERα in Kiss1 neurons restrains gonadal function and pubertal onset. Selective ablation of ARC KNDy neurons in adult females blunts the increase in circulating gonadotropins that accompanies ovariectomy, but roles for these cells in development and pubertal onset have not been directly studied (28). The failure to detect normal rostral hypothalamic Kiss1-immunoreactivity (IR) in animals lacking ERα in all Kiss1 cells is consistent with a role of ERα in KNDy cells for pubertal restraint (25), but this has not been tested directly. Here, we have ablated ERα in all Kiss1 cells or specifically in Tac2 cells (leaving ERα intact in rostral hypothalamic Kiss1 cells) to distinguish roles for ERα in KNDy neurons.

Materials and Methods

Animals

All animals were bred in our colony in the Unit for Laboratory Animal Medicine at the University of Michigan in accordance with the guidelines and approval of the University Committee on the Care and Use of Animals. The Esr1^flox mouse line has been previously described (29). Genotyping was by PCR to identify a loxP insertion (for primers, see Table 1). Occasional early embryonic recombination (Δ) around the loxP sites of this allele was detected using a separate forward primer together with the reverse primer used above; such animals were excluded from study.

Table 1.

Primers Used

Genotyping
	Cre Forward	WT Forward	Common Reverse
Kiss1cre	acctggcctggtctggacac	ccaaggcagggagcttcta	gagggtgtgggcatatgagt
Tac2cre	acctggcctggtctggacac	cccccagctttggcatcctc	gcgaatgacagaccatctctcc
	Forward	Reverse	Δ Forward
Esr1flox	aaggctgcaaggctttcttt	gaaattcttagccacagcttc	gaatccaccagctgctgtag
Riboprobe Generation
	Forward	Reverse
Kiss1	aattaaccctcactaaagggagactgtgtcgccacctatgg	taatagcactcactatagggagatctagaagctccctgccttg
Tac2	aattaaccctcactaaagggagagctgtcctcagct	taatacgactcactatagggagatttgaggatgccaaagctccccctg

Generation of Kiss1^cre and Tac2^cre knock-in mice

5.6 kb of genomic DNA containing portions of exon 2 and the 3′-untranslated region (UTR) of the murine Kiss1 gene was amplified by PCR from R1 ES cells and cloned into a plasmid for insertion of a FRT-NEO-FRT-IRES-CRE cassette between the STOP codon and the polyadenylation site. Similarly, 5.2 kb of gDNA containing portions of exon 6 and the 3′-UTR of the murine Tac2 gene was amplified and cloned for insertion of the same FRT-NEO-FRT-IRES-CRE cassette. The targeting constructs were linearized using NotI and electroporated into R1 mouse embryonic stem cells at the University of Michigan Transgenic Animal Model Core. Neomycin-resistant clones were analyzed by quantitative real-time PCR (30) and further confirmed by Southern blotting using HindIII and an external 580-bp probe (Kiss1) or EcoRV and an external 560-bp probe (Tac2). Correctly targeted ES cells were injected into C57Bl/6J blastocysts to generate chimeras. Male chimeras were then bred to C57Bl/6J females, and pups were genotyped to confirm insertion of internal ribosome entry site (IRES)-cre into the appropriate locus. To remove the FRT-flanked NEO cassette, mice were bred to a line expressing Flp recombinase in the germline mouse line (Jax 012930), and excision was confirmed by PCR.

Mouse model validation

Kiss1^cre/+ and Tac2^cre/+ males were bred to homozygous cre-inducible eYFP reporter females (Jax 006148) to generate Kiss1^eYFP and Tac2^eYFP animals, respectively. Dual-label immunohistochemistry/in situ hybridization (IHC/ISH) was performed as previously described (31) on brains from adult female Kiss1^eYFP and Tac2^eYFP mice. Enhanced yellow fluorescent protein (eYFP) was detected using a chicken anti-green fluorescent protein primary antibody (1:5000; Abcam).

Perfusion and IHC

Blood was drawn from the heart of female mice (age 21 d or 9–10 wk) after administration of an overdose of sodium pentobarbital. They were then perfused transcardially with PBS (pH 7.4) followed by 10% formalin. Blood was allowed to clot for 30 minutes at room temperature and then centrifuged for 15 minutes at 2000g, and the serum was removed and stored at −20°C. The brain was removed and postfixed in 10% formalin for 2–4 hours and then dehydrated in 30% sucrose in PBS until the time of sectioning. Brains were cut in 30-μm coronal sections on a sliding microtome, collected in 4 representative series, and stored at −20°C in cryoprotectant (32).

For Kiss1 immunostaining, free-floating brain sections were thoroughly washed with PBS to remove the cryoprotectant, incubated in 1% H₂O₂ in PBS, blocked in 3% normal donkey serum/3% Triton X-100 in PBS and then incubated with primary antibody (rabbit anti-Kiss1, 1:2000; Millipore) overnight. On the following day, sections were washed again with PBS and then incubated in biotinylated donkey-antirabbit (1:200; Jackson ImmunoResearch) for 2 hours, followed by avidin-biotin-complex labeling (Vectastain Elite kit; Vector Laboratories). Signals were developed with diaminobenzidine resulting in a brown precipitate.

For eYFP and ERα immunostaining, sections were blocked as described above and then incubated in primary antibodies (chicken anti-green fluorescent protein, 1:5000 [Abcam]; rabbit anti-ERα, 1:1000 [Santa Cruz Biotechnology, Inc]) overnight. Brain sections were washed and then incubated with Alexa Fluor-conjugated secondary antibodies (goat antichicken 488 and donkey antirabbit 568, 1:250; Invitrogen). Sections were mounted onto gelatin-coated slides and coverslipped with ProLong Antifade mounting medium (Invitrogen).

Longitudinal study

To ablate ERα from Kiss1- and Tac2-neurons, we crossed each Cre/+ line (Tac2^cre/+ and Kiss1^cre/+) to Esr1^flox/flox mice, and then crossed the resultant Cre/+;Esr1^flox/+ males to Esr1^flox/+ females. The F2 litters contained Cre/+;Esr1^flox/flox (ERα^Kiss1KO and ERα^Tac2KO, respectively) mice, along with Esr1^flox/flox and Cre/+ control mice for study. When appropriate, Cre/+ and Esr1^flox/flox mice were combined into one larger control group for comparison with the appropriate Cre/+;Esr1^flox/flox animals.

Females were checked daily for vaginal opening beginning at postnatal day 10. Mice were individually housed beginning at postnatal day 21; body weight was measured weekly. After vaginal opening, estrous cycle phase was determined daily based on vaginal cytology. At 8 weeks of age, body composition was determined by a Minispec LF90II (Bruker Optics) analyzer. Tissue was harvested from 9-week-old mice (some controls in diestrus and some in estrus, based on vaginal cytology). Using a mouse brain matrix, the ARC and a triangle of tissue containing the AVPV/PeN plus preoptic area were dissected and individually snap frozen on dry ice. Brain tissue was stored at −80° for later RNA extraction. Ovaries and uterus were removed, weighed, and fixed overnight in Z-fix (Anatech) and then transferred to 70% ethanol for storage. Ovaries were sectioned at 4 μm on a cryostat and stained with hematoxylin and eosin by the University of Michigan Comprehensive Cancer Center Tissue Core. The number of follicles or corpora lutea was counted by an individual blind to the genotype.

RNA was extracted from frozen microdissected brain tissue using TRIzol (Invitrogen) and then converted to cDNA using the iScript cDNA Synthesis kit (Bio-Rad). Gene expression was analyzed in triplicate by quantitative RT-PCR using Gapdh (endogenous control; VIC) and Kiss1 (FAM) TaqMan assays (Applied Biosystems). Relative mRNA expression values were calculated by the 2^−ΔΔCt method.

Ovariectomy

Nine-week-old female mice (ERα^Kiss1KO, ERα^Tac2KO and their littermate controls) were ovariectomized (OVX) under isoflurane anesthesia. Blood was collected in a heparinized capillary tube from the tail vein before ovariectomy. Seven days after ovariectomy, all mice were killed by anesthetic overdose, and hypothalamic regions were microdissected as described above.

Microscopy and image analysis

Microscopic images were obtained using an Olympus BX-51 microscope with a DP30BW camera (Olympus) or a Nikon Eclipse 90i microscope with a DS-Fi1 color camera (Nikon) (for color images [ovaries histology, dual IHC/ISH]). Images from fluorescent labeling experiments were pseudocolored and merged using Adobe Photoshop.

Hormone analysis

Serum or plasma was analyzed by the University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core (estradiol) or the University of Michigan Chemistry Core (LH and FSH). Estradiol concentrations were determined by ELISA (Calbiotech) (intraassay coefficient of variation (CV) 6.3%; interassay CV 8.1%), whereas LH and FSH concentrations were assessed using the Millipore MILLIPLEX MAP rat pituitary panel; all were run in a single assay (intraassay CV 9.4% [FSH] and 14.1% [LH]).

Statistics

Data were analyzed and graphs were generated using either OriginPro 8 or GraphPad Prism software. Data were normally distributed. Student's t test was used when only 2 groups were compared. ANOVA with Bonferroni post hoc analysis was used when comparing 3 or more groups. Differences were deemed significant if P < .05. Data are presented as mean ± SEM.

Results

Generation and validation of Tac2 and Kiss1 cell-specific mouse lines

Reasoning that neurons coexpressing NKB and ERα likely specify KNDy cells, we set out to generate tools to ablate the gene encoding ERα (Esr1) in Tac2- or all Kiss1-expressing cells. We thus generated knock-in mouse lines in which we inserted an IRES plus the coding sequence of cre recombinase in the 3′-UTR of the terminal exons of the Tac2 and Kiss1 loci (Tac2^cre and Kiss1^cre mice, respectively) to mediate the expression of cre from the mRNA encoding the relevant neuropeptide (Figure 1, A and B).

Figure 1.

Generation of mice to study Kiss1- and Tac2-expressing neurons and coexpression of ERα and eYFP in Tac2^eYFP and Kiss1^eYFP female mice. Tac2^cre mice carry an IRES-cre cassette in the 3′-UTR of the endogenous murine Tac2 gene (A). Kiss1^cre mice carry the same IRES-cre cassette in the 3′-UTR of the murine Kiss1 gene (B). Both cre lines were crossed to ROSA26-eYFP and Esr1-flox animals to generate fluorescent reporter and neuron-specific ERα deletion strains (shown for Tac2^cre). eYFP- and ERα-IR (pseudocolored green and purple, respectively) in the ARC (C) and the AVPV (D) of a Kiss1^eYFP mouse and in the ARC (E) and AVPV (F) of a Tac2^eYFP mouse. eYFP- and ERα-IR (pseudocolored green and purple, respectively) in the ARC (G) and the AVPV (H) of a Kiss1^eYFP/ERα^Kiss1KO mouse and in the ARC (I) and AVPV (J) of a Tac2^eYFP/ERα^Tac2KO mouse. Neurons expressing both eYFP and ERα appear white. 3V, third ventricle; ME, median eminence. K, Counts of Tac2 and Kiss1 neurons in Tac2^eYFP and Kiss1^eYFP female mice in the indicated brain regions (black bars) and numbers of Tac2 and Kiss1 neurons colocalized with ERα (gray bars). L, Data from K plotted to show % Tac2 and Kiss1 neurons colocalizing with ERα in each region. Mean ± SEM is plotted (from a 1:4 series through the brains of n = 3 animals per genotype).

We crossed Tac2^cre and Kiss1^cre mice onto the cre-activated ROSA26-eYFP reporter mouse line, generating Tac2^eYFP and Kiss1^eYFP animals to reveal sites of cre expression (Figure 1, A and B). To examine the overlap of YFP expression with Kiss1- and Tac2-expression within the brain in these 2 novel mouse lines, we performed dual-label IHC/ISH on hypothalamic sections from Kiss1^eYFP and Tac2^eYFP female mice (Supplemental Figure 1). In the ARC of Tac2^eYFP females, most (96%) eYFP-IR colocalized with Tac2 mRNA and Kiss1 mRNA (97%). Although we observed Kiss1 hybridization adjacent to the third ventricle in the AVPV of Tac2^eYFP females, we found few eYFP cells in that area (and no colocalization), consistent with the absence of Tac2 in AVPV/PeN Kiss1 neurons. Similarly, Kiss1 probe and eYFP-IR colocalized in Kiss1^eYFP sections, confirming the coexpression of cre and Kiss1 in these mice.

We also examined the colocalization of ERα with eYPF in Tac2^eYFP and Kiss1^eYFP mice. In Kiss1^eYFP mice, nearly all (93 ± 11%; n = 3) of ARC eYFP cells express ERα and approximately 2/3 (66 ± 7%; n = 3) of AVPV eYFP cells express ERα (Figure 1, C and D). In Tac2^eYFP female mice, nearly all (96 ± 1%; n = 3) ARC eYFP cells express ERα, but we detected no eYFP/ERα cells in the AVPV/PeN (Figure 1, E, F, and K). Although we found occasional eYFP/ERα cells elsewhere in the brain of Tac2^eYFP mice, they did not cluster in any particular region and were few in number (Figure 1K). Thus, ARC Tac2/ERα cells, which approximate KNDy cells, represent the vast majority (∼90%) of central nervous system Tac2/eYFP+ERα cells.

Cre-mediated excision of Esr1^flox by Kiss^cre ablated ERα from Kiss1 cells in both the ARC and AVPV/PeN, as revealed by the lack of ERα-IR in eYFP neurons in Kiss1^eYFP/ERα^flox/flox females (0% of ARC and 6% of AVPV/PeN eYFP cells contained ERα-IR; n = 2) (Figure 1, F–I). Excision of Esr1^flox by Tac2^cre similarly ablated ERα from Tac2 cells in the ARC of a Tac2^eYFP/ERα^flox/flox mouse (4% of ARC eYFP cells contained ERα-IR; n = 2) (Figure 1, K and L).

Consistent with previous reports, we detected eYFP cells in the ovary of Kiss1^eYFP (Supplemental Figure 2). We also detected evidence of Tac2 expression in the pituitary and ovary of Tac2 reporter mice. Although our IHC analysis detected no ERα in these cells in WT animals (data not shown), it remains possible that Kiss1^cre and Tac2^cre ablate functional ERα from some or all of the cells that express Kiss1 or Tac2 outside of the brain (others report Kiss1 expression in pancreatic islets, adipose, and other tissues, as well) (33, 34), in addition to deleting ERα from cells in the ARC and/or AVPV/PeN. Although our analysis (Supplemental Figure 1) and other reports suggest that virtually all ARC Tac2 cells coexpress Kiss1 (12, 13), it is also theoretically possible that some non-KNDy cells are affected in the brain of Tac2^eYFP/ERα^flox/flox animals, although AVPV/PeN Kiss1 neurons should be unaffected.

Analysis of Kiss1 expression in ERα^Kiss1KO and ERα^Tac2KO females

In wild-type animals, estrogen promotes AVPV Kiss1 expression while inhibiting ARC Kiss1 expression (Smith et al [21] and Frazao et al [26]). To examine the potential role for ERα in specific cell types for the control of Kiss1 expression in these areas in our mouse models, we extracted RNA from microdissected tissue (ARC or the AVPV/PeN-containing rostral hypothalamus) from ERα^Kiss1KO and ERα^Tac2KO females to determine relative Kiss1 mRNA levels (Figure 2). Deletion of ERα from Kiss1 and Tac2 neurons increased ARC Kiss1 expression in ERα^Kiss1KO and ERα^Tac2KO females, consistent with a cell-autonomous role for ERα in the suppression of Kiss1 expression in KNDy cells (Figure 2, A and B). Deletion of ERα from all Kiss1 cells in ERα^Kiss1KO females decreased Kiss1 in the rostral hypothalamus, whereas deletion of ERα solely from KNDy neurons in ERα^Tac2KO females did not alter Kiss1 mRNA in the rostral hypothalamus (Figure 2, C and D). Similarly, AVPV Kiss1-IR was diminished in adult ERα^Kiss1KO females (Figure 2, E and F), but not ERα^Tac2KO females (Figure 2, G and H), consistent with the observed changes in rostral hypothalamic gene expression. This could reflect a decrease in the total number of AVPV Kiss1 cells and/or decreased intensity of staining and soma detection due to a decrease in Kiss1 content in each AVPV Kiss1 neuron. Thus, ERα in Tac2 neurons suppresses ARC Kiss1 expression, whereas ERα in non-Tac2 cells independently controls rostral hypothalamic Kiss1 mRNA expression.

Figure 2.

Kiss1 expression in intact ERα^Kiss1KO and ERα^Tac2KO females. Relative expression of Kiss1 in the ARC (A and B) and rostral hypothalamus (C and D) of intact female ERα^Kiss1KO (left panels) and ERα^Tac2KO (right panels) mice. Controls were collected in diestrus/metestrus (control) or estrus (estrous control). One-way ANOVA, Bonferroni post hoc: *, P < .05; ***, P < .001. Numbers in parentheses above each bar denote the n for each measurement. Representative Kiss1-IR (from ≥3 animals of each genotype) in the AVPV/PeN of intact adult females of the indicated genotype (E–H).

Estradiol feedback regulation

To assess gonadal feedback on Kiss1 expression in ERα^Kiss1KO and ERα^Tac2KO mice, we also examined ARC and rostral hypothalamic Kiss1 mRNA after ovariectomy (Figure 3). ARC Kiss1 expression in control animals rose dramatically as a result of ovariectomy, to levels that were indistinguishable from those of intact or OVX ERα^Kiss1KO and ERα^Tac2KO females. In contrast, ARC Kiss1 expression was not changed by ovariectomy in the ERα^Kiss1KO or ERα^Tac2KO females (Figure 3, A and B). Thus, ERα in KNDy neurons suppresses Kiss1 expression by these cells in response to gonadal signals.

Figure 3.

Kiss1 expression in intact and OVX ERα^Kiss1KO and ERα^Tac2KO females. Relative expression of Kiss1 in the ARC (A and B) or rostral hypothalamus (C and D) of intact and OVX control and ERα^Kiss1KO (left panels) or ERα^Tac2KO (right panels) adult females. Numbers in parentheses above each bar denote the n for each measurement. Two-way ANOVA, Bonferroni post hoc: *, P < .05; **, P < .01; ***, P < .001. NS, not significant; ND, not detected.

Although ovariectomy tended to decrease rostral hypothalamic Kiss1 expression in control animals, as reported previously (21), this difference was only significant in the control littermates of ERα^Tac2KO females (Figure 3, C and D). As above, rostral hypothalamic Kiss1 expression was diminished in intact ERα^Kiss1KO but not ERα^Tac2KO females. Ovariectomy decreased rostral hypothalamic Kiss1 expression not only in ERα^Tac2KO females but also further reduced the already low rostral hypothalamic Kiss1 expression in ERα^Kiss1KO mice (Figure 3C).

Body weight and composition in ERα^Kiss1KO and ERα^Tac2KO females

The body mass of ERα^Kiss1KO females was not significantly altered compared with littermate controls at early ages, but at 8 weeks of age, ERα^Kiss1KO females were slightly heavier than control littermates (Supplemental Figure 3). A similar effect was observed in ERα^Tac2KO females, which were slightly heavier than their littermate controls starting at 5 weeks of age (Supplemental Figure 3). Body composition analysis of 8-week-old females revealed that both ERα^Kiss1KO and ERα^flox/flox mice had lower body fat percentage than Kiss1^IRES−cre/+ mice and also that ERα^Kiss1KO females have increased fluid percentage compared with both littermate control genotypes (Supplemental Figure 3). The same effect was observed in ERα^Tac2KO mice and their littermate controls; Tac2^IRES−cre/+ mice have significantly higher fat percentage compared with both ERα^flox/flox and ERα^Tac2KO animals and we found increased fluid percentage in ERα^Tac2KO females compared with both littermate control genotypes (Supplemental Figure 3).

Reproductive function in ERα^Kiss1KO and ERα^Tac2KO females

Vaginal opening is an external signal of puberty onset in rodents; administration of Kiss1 accelerates this and other markers of pubertal onset (20). Vaginal opening occurred in ERα^Kiss1KO and ERα^Tac2KO mice significantly earlier than their littermate controls (Figure 4, A and B). Vaginal cytology revealed early cornification of the vaginal epithelium in ERα^Kiss1KO and ERα^Tac2KO females, most of which displayed estrogenized (cornified) cytology when first checked at 21 days of age; in contrast, controls exhibited first estrus between 35 and 40 days of age (data not shown). Furthermore, ERα^Kiss1KO and ERα^Tac2KO mice spent significantly less time in diestrus or metestrus and significantly more time in estrus than their littermate controls (Figure 4, C and D).

Figure 4.

Advanced vaginal opening and persistent cornification in ERα^Kiss1KO and ERα^Tac2KO females. Average age at vaginal opening of littermate controls and ERα^Kiss1KO females (A) and ERα^Tac2KO females (B). Numbers in parentheses above each bar denote the n for each measurement. Representative vaginal cytology over time for control and ERα^Kiss1KO females (C) and control and ERα^Tac2KO females (D). Percentage of time spent in diestrus/metestrus and estrus (based on vaginal cytology) from the time of vaginal opening until 8 weeks of age for mice of the indicated genotypes. Numbers in parentheses above each bar denote the n for each measurement. One-way ANOVA, Bonferroni post hoc: ***, P < .001. VO, vaginal opening; C, cornified epithelial cells; N, nucleated epithelial cells; L, leukocytes; D/M, diestrus/metestrus; E, estrus.

In addition to their precocious and persistent estrus, both ERα^Kiss1KO and ERα^Tac2KO females exhibit additional abnormalities in their reproductive organs, including enlarged, fluid-filled uteri (Figure 5, A and B). Furthermore, we observed an increase in the total number of follicles, including hemorrhagic cysts, within the ovaries of ERα^Kiss1KO and ERα^Tac2KO females, coupled with a reduced number of corpora lutea (Figure 5, C–H). Thus, ablation of ERα from either Kiss1- or Tac2-expressing neurons results in early vaginal opening and estrogenization, along with ovarian and uterine histology consistent with hyperstimulation of the hypothalamic-pituitary-gonadal axis.

Figure 5.

Ovarian and uterine phenotypes of adult ERα^Kiss1KO and ERα^Tac2KO females. Uterine mass (normalized to body mass [g/g]) of 9-week-old control (C) (Kiss1^IRES−Cre/+ and ERα^flox/flox females dissected in diestrus/metestrus), estrous control (EC) (littermate controls dissected in estrus), and ERα^Kiss1KO females (A) (n = 7–18). Uterine mass of control, estrous control, and ERα^Tac2KO females (B) (n = 8–16). Number of follicles and corpora lutea per 2 ovary sections per animal from control and ERα^Kiss1KO (C) or ERα^Tac2KO (D) females. Numbers in parentheses above each bar denote the n (number of animals) for each measurement. One-way ANOVA (A and B), Bonferroni post hoc: ***, P < .001. Student's t test (C and D): *, P < .05; **, P < .01; ***, P < .001. Representative hematoxylin and eosin (H&E)-stained ovary sections from littermate control (E) and ERα^Kiss1KO females (F) and littermate control (G) and ERα^Tac2KO females (H). Blue stars denote corpora lutea, whereas yellow arrows indicate hemorrhagic cysts. All images were taken at the same magnification (×4) and were scaled identically.

Gonadotropins and gonadal hormones

To assess whether deletion of ERα from Kiss1 or Tac2 cells altered gonadotropin and/or estradiol levels consistent with the phenotype of the ERα^Kiss1KO and ERα^Tac2KO females, we analyzed serum from juvenile (3 wk old) (Figure 6) and adult (9 wk old) (Figure 7) animals. Serum LH concentrations were elevated in juvenile ERα^Kiss1KO females compared with littermate controls (Figure 6A), although FSH was not affected (Figure 6C). Not surprisingly, given their precocious vaginal opening, juvenile ERα^Kiss1KO females also had significantly higher estradiol levels than their littermate controls (Figure 6E). We observed a similar effect on serum LH and estradiol in the ERα^Tac2KO females (Figure 6B). The ERα^Tac2KO females displayed elevated FSH concentrations (Figure 6D), however.

Figure 6.

Gonadotropin and estradiol levels of intact juvenile ERα^Kiss1KO and ERα^Tac2KO females. Serum LH (A and B), FSH (C and D), and estradiol (E and F) in control (C) and ERα^Kiss1KO (left panels) or ERα^Tac2KO females (right panels). Numbers in parentheses above each bar denote the n for each measurement. Student's t test: *, P < .05; **, P < .01; ***, P < .001.

Figure 7.

Gonadotropin and estradiol levels in intact adult ERα^Kiss1KO and ERα^Tac2KO females. Serum LH (A and B), FSH (C and D), and estradiol (E and F) levels of 9-week-old diestrous control (C), estrous control (EC), and ERα^Kiss1KO (left panels) or ERα^Tac2KO females (right panels). Numbers in parentheses above each bar denote the n for each measurement. One-way ANOVA, Bonferroni post hoc: *, P < .05; ***, P < .001.

Because ERα^Kiss1KO and ERα^Tac2KO females showed persistent vaginal cornification, we assayed serum from a group of littermate controls in estrus, as well as diestrus, alongside the adult ERα^Kiss1KO and ERα^Tac2KO animals. Adult female ERα^Kiss1KO animals displayed elevated LH concentrations compared with estrous, but not diestrous, controls, whereas LH concentrations in adult ERα^Tac2KO females were elevated compared with both cycle stages in control mice (Figure 7, A and B). FSH levels were similar among all genotypes in the adult females (Figure 7, C and D). Both ERα^Kiss1KO and ERα^Tac2KO adult females displayed estradiol levels that were approximately 3 times higher than controls (Figure 7, E and F).

Discussion

Important aspects of estradiol feedback upon reproductive neuroendocrine activity has been postulated to be mediated by ERα-expressing Kiss1 neurons of the AVPV and ARC, but the roles of the Kiss1 neurons in these individual regions has not been empirically distinguished. To test directly roles for ERα in Tac2 (including KNDy) neurons, while maintaining endogenous ERα in AVPV/PeN Kiss1 neurons, we genetically ablated ERα in all Kiss1-expressing cells or in Tac2 cells. The essentially complete colocalization of cre reporter expression and mRNA from the relevant neuropeptide is as expected based upon the expression of cre from the Tac2 and Kiss1 mRNAs in these knock-in mice, and based upon the previous findings of others with similar cre lines and for the colocalization of Tac2 and Kiss1 in the ARC (12, 13, 25, 26, 35). Because KNDy neurons coexpress the product of the Tac2 gene (NKB), as well as Kiss1, we used our novel Tac2^cre mouse line to ablate ERα in KNDy neurons, but not AVPV/PeN Kiss1 cells. Because ARC Tac2/ERα cells, which approximate KNDy cells, represent the vast majority (∼90%) of central nervous system Tac2/eYFP+ERα cells; our results therefore likely reflect effects of ERα deletion in ARC Tac2 cells.

Although our IHC analysis detected no ERα in the pituitary and ovarian cells that exhibit Kiss1^cre- and Tac2^cre-mediated recombination in animals containing wild-type Esr1 (data not shown), it remains possible that Kiss1^cre and Tac2^cre ablate functional ERα from some or all of the cells that express Kiss1 or Tac2 outside of the brain (others report Kiss1 expression in pancreatic islets, adipose, and other tissues, as well) (33, 34), in addition to deleting ERα from cells in the ARC and/or AVPV/PeN. Although our analysis (Supplemental Figure 1) and other reports suggest that virtually all ARC Tac2 cells coexpress Kiss1 (12, 13), it is also theoretically possible that some non-KNDy cells are affected in the brain of Tac2^eYFP/ERα^flox/flox animals, although AVPV/PeN Kiss1 neurons should be unaffected.

The ablation of ERα from all Kiss1 cells in our hands resulted in early vaginal opening and cornification of vaginal epithelial cells, and a phenotype of ovarian hyperstimulation characterized by elevated circulating gonadotropins and estradiol, fluid-filled uteri, increased follicle development with hemorrhagic cysts, and few corpora lutea, consistent with previous findings (25, 26).

Our finding of a similar phenotype upon the deletion of ERα from Tac2 neurons thus refines our understanding of the locus of central estradiol negative feedback on puberty onset and gonadal restraint. The similarities between the phenotypes presented by ERα^Kiss1KO and ERα^Tac2KO females suggest that the precocious estrogenization and ovarian hyperstimulation results from the loss of ERα from the ARC KNDy neurons, because this is common to both knockout lines.

Body weight was increased in both ERα^Kiss1KO and ERα^Tac2KO females due to increased fluid content; this likely reflects the increase in fluid in their uteri, as noted above. In contrast, there was no effect of ablation of ERα from Kiss1 or Tac2 cells on adiposity (although the presence of the Esr1^Flox allele decreased adiposity compared with cre only controls).

Others have shown that the ablation of ARC KNDy cells in adult animals blunts the increase in LH secretion that follows ovariectomy (28). The present data suggest that the deletion of ERα from Tac2-expressing neurons sets in motion a series of changes that likely begins with an increase in Kiss1 expression and culminates in increased gonadotropin drive. That is, during normal prepubertal development, estradiol restrains a positive signal from KNDy neurons, rather than promoting an inhibitory KNDy-derived signal. The negligible to modest elevations in LH, FSH, and estradiol concentrations in adult ERα^Kiss1KO and ERα^Tac2KO females relative to the elevations observed in younger animals also suggests the potential for compensatory circuit reorganization with age in these animals.

Circulating estradiol controls Kiss1 mRNA expression differently in the 2 hypothalamic Kiss1 cell types, suppressing Kiss1 in ARC cells and supporting Kiss1 in rostral hypothalamic neurons (21, 26). Ablation of ERα from all Kiss1 neurons or specifically from KNDy neurons promotes ARC Kiss1 gene expression, mimicking the levels found in OVX animals. These findings are consistent with a crucial role for estrogen/ERα in KNDy cells in the repression of Kiss1 in these cells. Although deletion of ERα from KNDy neurons in ERα^Tac2KO mice did not alter AVPV Kiss1 expression, deletion of ERα from all Kiss1 cells decreased Kiss1 expression in this area, demonstrating that estrogen acts via ERα in AVPV/PeN Kiss1 cells to support Kiss1 expression in these cells. Similarly, AVPV Kiss1-IR is diminished in adult ERα^Kiss1KO females, but not ERα^Tac2KO females, consistent with the observed changes in rostral hypothalamic gene expression. Thus, ERα in Tac2 neurons suppresses ARC Kiss1 expression, whereas ERα in non-Tac2 cells independently controls rostral hypothalamic Kiss1 mRNA expression. This could reflect a decrease in the total number of AVPV Kiss1 cells and/or decreased intensity of staining and soma detection due to a decrease in Kiss1 content in each AVPV Kiss1 neuron. This is consistent with previous findings suggesting that early deletion of ERα in Kiss1 neurons interferes with the normal development (and Kiss1 expression) of these cells (25).

Because the deletion of ERα alters Kiss1 expression oppositely in ARC vs AVPV Kiss1 cells, ERα may recruit different cofactors to the Kiss1 promoter region in each cell type, or ERα may act via distinct second order transcriptional regulators in each population of Kiss1 cells. Indeed, estrogen may increase Kiss1 expression by a unique epigenetically regulated enhancer in AVPV/PeN Kiss1 cells (36).

Because ovariectomy further suppressed already low Kiss1 expression in the rostral hypothalamus of ERα^Kiss1KO females, gonadal hormones contribute to AVPV Kiss1 expression not only via ERα in AVPV Kiss1 neurons, but also via mechanisms independent of ERα in these cells. Thus, in contrast to Tac2 neurons in the ARC, where the control of Kiss1 expression absolutely requires ERα, a second gonadal signal (distinct from estrogen/ERα) independently controls the expression of Kiss1 by AVPV neurons. Additionally, the finding that estrogen decreases LH concentrations in ERα^Kiss1KO females demonstrates the existence of other (non-Kiss1) loci of estrogenic negative feedback in adults (37). Indeed, we observed similar effects in ERα^Kiss1KO and ERα^Tac2KO females (Supplemental Figure 4). Although estrogenic feedback on the pituitary is likely to participate, other sites in the ARC or elsewhere may also play a role (38).

Our study was not designed to specifically test the postulate that the non-KNDy Kiss1 neurons of the rostral hypothalamus mediate estrogen positive feedback to the reproductive axis. Furthermore, there may be developmental changes in the circuitry of Kiss1 cells and/or other neurons that control reproductive function in our knockout animals; indeed, others have suggested that early deletion of ERα from AVPV Kiss1 cells impairs their development and expression of Kiss1 (25).

Overall, our findings demonstrate that ERα in Tac2 neurons normally restrains pubertal onset and hypothalamic reproductive drive, because deletion of ERα from these cells does not alter AVPV/PeN ERα or Kiss1 expression, but promotes precocious onset of puberty and subsequent gonadal hyperfunction. The ARC KNDy neurons thus constitute a critical element of the sensitivity to gonadal steroids that regulate puberty onset and sexual maturation.