All animal procedures were performed with the approval from the University of California (Riverside, CA) Animal Care and Use Committee and in accordance with the National Institutes of Health Animal care and Use Guidelines. Breeding pairs of FVB.129P2-Fmr1tm1Cgr/J (Fmr1 KO) and their congenic controls (WT) mice were obtained from Jackson Laboratories and bred in-house. Mice were maintained under a 12-h light, 12-h dark cycle and received food and water ad libitum. Since we were interested in determining the mechanisms of premature ovarian failure in women with mutations in the FMR1 gene, only female mice were used for our studies. Estrous cycle stage was determined with vaginal smears and females were collected in a specific estrous cycle stage, as indicated for each analysis. For fertility studies, Fmr1 KO females and WT controls were housed with WT males and their litters and numbers of pups per litter were recorded. Ovariectomy was performed, as described before, using 8-week-old mice (43). Animals were allowed to recover and seven days later blood was collected for hormone analyses as described below.

Previous studies using FXS mouse models demonstrated that these mice have heightened response to stress and altered levels of glucocorticoids (44), which can lead to a decrease in luteinizing hormone (LH) levels (45, 46). To reduce stress, animals were acclimated by daily handling and tail massage for two weeks prior to hormone measurements. For LH measurements, blood was collected from the tail and analyzed by an in house ultra-sensitive ELISA. The capture monoclonal antibody (anti-bovine LH beta subunit, 518B7) was provided by Janet Roser, University of California. The detection polyclonal antibody (rabbit LH antiserum, AFP240580Rb) was provided by the National Hormone and Peptide Program (NHPP). HRP-conjugated polyclonal antibody (goat anti-rabbit) was purchased from DakoCytomation (Glostrup, Denmark; D048701-2). Mouse LH reference prep (AFP5306A; NHPP) was used as the assay standard. Assay sensitivity is 0.016 ng/ml, while intra-assay coefficient of variation is 2.2% and inter-assay coefficient of variation was 7.3% at the low end of the curve. Other hormone assays were performed by the University of Virginia, Ligand Core using serum that was obtained from the inferior vena cava and serum prepared per their instructions. The University of Virginia Center for Research in Reproduction Ligand Assay and Analysis Core is supported by the Eunice Kennedy Shriver NICHD/NIH Grant R24HD102061. FSH was assayed by RIA using reagents provided by Dr. A.F. Parlow and the National Hormone and Peptide Program, as previously described (47). Mouse FSH reference prep AFP5308D was used for assay standards. Inhibin and steroid hormone levels were analyzed using validated commercially available assays, information for which can be found on the core’s website: http://www.medicine.virginia.edu/research/institutes-and-programs/crr/lab-facilities/assay-methods-page and reported in (48). Limits of detection were 2.4 ng/ml for FSH, 300 pg/ml for progesterone, and 100 pg/ml for testosterone. Intra- and inter-assay coefficients of variation were 6.9%/7.5%, 6.0%/11.4% and 4.4%/6.4% for FSH, progesterone (P) and testosterone (T), respectively. Each animal used for each hormone analysis is represented as a dot in the figure.

Pulsatile secretion of LH strictly corresponds to GnRH secretion (49, 50). LH is used as an indicator of GnRH secretion, since GnRH secretion into median eminence cannot be measured in mice. To ascertain if LH or GnRH neuron secretion is affected, we measured LH pulses and used an ultrasensitive ELISA assay for LH that allows for LH measurement in 5 µL of whole blood (51). Mice were acclimated for 2 weeks by daily tail massage. 10µL of blood was collected every 8 min for 3 hours from the tail vain (45, 52). LH levels were analyzed using ELISA described above. LH amplitude was determined by subtracting the LH value at the peak from the basal value prior to the onset of the pulse and averaged for each mouse. Mean LH concentration was calculated by averaging LH values, while pulse frequency was determined using freeware DynPeak algorithm (53).

8-week old female mice in diestrus were perfused with ice cold PBS, brains rapidly removed and flash frozen in isopentane on dry ice. Coronal brain sections of 500 μm were obtained using vibratome, hypothalamus dissected, RNA isolated using the RNAqueous^®-Micro Kit (Ambion) and quantified using Nanodrop. Gene expression in 50 ng RNA per sample was analyzed using the Nanostring instrument as described before (54), according to manufacturer’s instruction, with the nCounter Mouse Neuroinflammation Panel (770 genes, gene list available at the manufacturer’s website). The panel was customized with an addition of 30 custom probes for: Gnrh1, Kiss1, Kiss1r, Pdyn, Tac2, Tac3r, Agrp, Npy, Pomc, Cart, Mc3r, Mc4r, Hcrt, Ghrh, Crh, Trh, Oxt, Avp, Prl, Prlr, Adcyap1, Slc6a3, Slc32a1, Slc17a6, Th, Lif, Lifr, Gabra5, Gabra1, Gabrg2. Only samples with an RNA integrity number RIN over 7 were used after passing QC, with no imaging, binding, positive control, or CodeSet content normalization flags. Data analysis was performed using nSolver Analysis Software 4.0, including nCounter Advanced Analysis (version 2.0.115). Genes with the expression lower than the limit of detection after background subtraction, and compared to negative controls included in the panel, were excluded. Seven housekeeping control genes that are included in the panel, were used for normalization. A heatmap of differentially expressed genes (DEG) was created using Heatmapper software from University of Alberta (Edmonton, Canada (55);). Results are plotted in the Volcano plot as log fold change vs. log p-value, and genes with changes higher than 20% were indicated with colors in the figures: red indicates genes higher in KO compared to WT, while green indicates genes that are higher in the WT compared to KO. Genes with significant changes in expression are indicated above the dashed line. Gene ontology (GO) enrichment analysis of the DEG genes was performed using the ShinyGo 0.76.3 platform (South Dakota State University (56)). False discovery rate (FDR) cutoff was 0.05, with the pathway minimum set to 10. Data is deposited in GEO repository (accession number GSE222723).

Total RNA was extracted from ovaries using Trizol (Invitrogen, CA), and from hypothalamus and pituitary using the RNAqueous^®-Micro Kit (Ambion), quantified using Nanodrop and the same amount per sample reverse transcribed using Superscript IV (Invitrogen, CA). To dissect the hypothalamus, coronal brain sections of 300 μm were obtained using vibratome. Sections containing anterior hypothalamus and posterior hypothalamus were used to dissect the 1 mm x 1 mm mediobasal portion for RNA isolation. Conditions and primers were reported before (57–61).

WT controls and Fmr1 KO mice were anesthetized, perfused with 20 ml PBS and 20 ml 4% paraformaldehyde; and tissues were collected. Ovaries were fixed in 4% paraformaldehyde, embedded in paraffin, and cut to 20 μm sections. Slides were deparaffinized in xylene, rehydrated and H&E stain was performed to count ovarian follicles. For ovarian vasculature and innervation studies, frozen floating sections were stained with antibody to CD31 (1:2000 dilution, 553370, BD Biosciences) or with antibody to tyrosine hydroxylase (TH, 1:5000, ab112, Abcam) for 48 hours at 4°C, followed by overnight incubation with goat anti-rat IgG-Alexa 488 (1:2000, A11006, Vector Laboratories, Burlingame, CA) or goat anti-rabbit IgG-Alexa 488 (1:1000, A11034, Vector Laboratories, Burlingame, CA), respectively. Vascularization of corpora lutea and antral follicles was quantified by the mean fluorescent intensity (MFI) using Fiji ImageJ. Ovarian innervation was quantified by counting the number of neuronal projections in direct contact with follicles or corpora lutea.

Hypothalami were sectioned to 100 μm sections. Sections containing organum vasculosum laminae terminalis (OVLT) where GnRH neurons are located, were blocked and stained for GnRH using rabbit anti-GnRH antibodies (1:10,000 dilution) kindly provided by Greg Anderson (University of Otago; Dunedin, New Zealand (62)), GABAγ2 receptor subunit (1:10,000 dilution, guinea pig anti-GABAγ2, Synaptic systems 224 004), VGAT (1:5,000, mouse anti-VGAT, Synaptic systems 131 011) for 72 hours at 4°C. After PBST washes, slides were incubated overnight at 4°C with secondary antibodies goat anti-rabbit IgG-Alexa 488 (1:1000, A11034, Vector Laboratories, Burlingame, CA); anti-mouse IgG-Alexa 594 (1:1000, A11032, Vector Laboratories, Burlingame, CA); anti-guinea pig–biotin (1:1000, BA-7000) followed by streptavidin-Cy5 (1:1000, 434316, Vector Laboratories, Burlingame, CA). Secondary antibody-only controls were performed to determine antibody specificity. To determine puncta density, we followed our established protocol as previously published (60, 63–66). Puncta were counted in the individual neurons, by an investigator blinded to the group, where at least 45 μm of the axon proximal to soma can be observed using z-stack acquired by confocal Leica SP2 microscope. At least 15-20 individual neurons from 4-5 different sets of mice were counted. 3–D reconstruction was performed by Imaris software (Bitplane, Inc; Concord, MA).

Immunostaining for FMRP was performed using antigen retrieval methods, as previously described (67). Slices were stained overnight with mouse anti-FMRP (1:1000; Developmental Studies Hybridoma Bank, catalog #2F5-1-s, RRID: AB_10805421). Secondary antibody was donkey anti-mouse Alexa 594 (1:300; Molecular Probes, A-21202). Slices were mounted on slides with Vectashield mounting medium containing DAPI (Vector Laboratories, H-1200).

Whole cell lysates were obtained from the dissected hypothalami from WT controls and Fmr1 knockout mice. The same amount of protein from each sample, determined by Bradford assay, was resolved on SDS-PAGE, transferred on nitrocellulose membrane and probed for: GABAγ2 receptor subunit (1:1000, 14104-1-AP, Proteintech), NMDAR1 (1:1000, 32-0500 Invitrogen), NMDAR2B (1:1000, 07-632 EMD Millipore), postsynaptic density protein 95 (PSD-95; 1:1000, 3409, Cell Signaling), microtubule-associated protein 2 (MAP2; 1:5000, ab5392, Abcam) or β-tubulin (1:1000, sc-9104, Santa Cruz Biotechnology). Bands were quantified using ChemiDoc imaging system (Bio-Rad, Hercules, CA).

Statistical differences between WT control and Fmr1 KO mice (p < 0.05) were determined by t-test, or ANOVA when appropriate, followed by Tukey’s post-hoc test for multiple comparisons using Prism software (GraphPad, CA).

Since women with FMR1 mutation experience increased risk of early menopause, to begin investigating the role of FMRP in reproductive function, we first determined if the lack of FMRP in female Fmr1 KO mice can mimic reproductive dysfunctions observed in women with reduced levels of FMRP due to FMR1 mutation. In affected people, full expansion of the unstable CGG trinucleotide repeats (>200) leads to the loss of FMRP and FXS. CGG expansion in a range of 50-200 repeats, or FMR1 premutation, exhibits reduced FMRP levels and FXTAS. Due to differential methylation between human and mouse genes, the Fmr1 KO is a widely used mouse model to study Fragile X Syndrome and is considered a better model than putative mimics of the CGG repeat expansion (68, 69). Our study showed that Fmr1 KO female mice experienced early vaginal opening, an external sign of puberty in mice ( Figure 1A , FMR1, Fmr1 knock-out (KO); WT, wild-type control; each point represents a mouse, while bars represent group average). Fmr1 KO mice demonstrated vaginal opening at postnatal day 29 (p29), compared to p31 in WT controls. At 8 weeks of age, we paired Fmr1 KO females and control WT females, with control males and recorded birth dates and number of litters, and the number of pups in each litter, until they stopped reproducing. There was no difference in the length of time between litters ( Figure 1B ) or in the estrous cycle duration ( Figure 1C ). We determined that Fmr1 KO female mice exhibited early cessation of reproductive function, which is similar to early menopause in women with FMR1 mutation or premutation ( Figure 1D , each point represents a mouse) (70, 71). Fmr1 KO mice stopped having litters at 5.5 months of age and an average age of the last litter was p163 (FMR1, black squares), compared to p263 for WT control females (WT, open circles). Similar to the penetrance in women with mutations in the FMR1 gene (32), we observed different degrees of premature cessation of reproductive function. Fmr1 KO females split into two groups based on the age at last litter: females more severely affected with the mutation that stopped reproducing at p97, and less severely affected females that had the last litter at p221. Nonetheless, the difference between controls (p263) and less affected females (p221) was also statistically significant. We counted the number of pups in the litters and determined that Fmr1 KO females had larger first three litters ( Figure 1E ). The average litter size of the first litter was 10.6 pups for Fmr1 KO and 7.5 pups for controls; the average size of the second litter was 11.4 pups for Fmr1 KO and 8.8 pups for controls, and the average size of the third litter was 10.8 pups for Fmr1 KO and 7.1 pups for controls. However, while control females continued to produce litters for at least 10 litters, the number of Fmr1 KO females that continued to produce litters decreased after each litter, and only 5 out of 11 Fmr1 KO females had a fourth litter, while none had eighth litter. There was no difference in weight at any age between KO and control females and thus, change in reproductive function does not stem from a difference in weight. Therefore, similar to women with a mutation in the FMR1 gene, Fmr1 KO female mice experience early cessation of reproductive function.

To examine if early cessation of reproductive function is due to initially diminished ovarian reserve or to an accelerated loss of follicles, we counted the number of primordial follicles in pre–pubertal females at 3 weeks of age (p21, Figure 2A count, and Figure 2D representative images). Three equal size areas of the ovarian cortex were selected per mouse, primordial follicles counted, and the numbers were averaged for each mouse. There was no difference in the number of primordial follicles between Fmr1 KO and WT females, indicating that development of the initial pool is not affected by the Fmr1 loss. The analysis of the number of corpora lutea (CL) in the ovaries at 6 weeks of age (p42) showed that Fmr1 KO females had over 4 times more corpora lutea than WT controls. Fmr1 KO had 10.2 average number of corpora lutea per ovary compared to 2.2 corpora lutea per ovary in controls ( Figure 2B count, Figure 2E representative images). Given that Fmr1 KO females exhibited early vaginal opening, to confirm that the increase in corpora lutea in KO females does not stem from early puberty, we counted the number of corpora lutea at 9 weeks of age (at p63; Figure 2C ). At p63, an average number of corpora lutea per ovary was significantly higher in Fmr1 KO females (8.6 corpora lutea) compared to control WT mice (5.4 corpora lutea). Together, these results demonstrate that Fmr1 KO female mice stop reproducing early, have the same primordial follicle pool, but increased number of corpora lutea corresponding to the larger litter size in young animals, indicating potentially higher recruitment to the growing pool in young animals.

Given that ovarian function is regulated by gonadotropin hormones from the pituitary, we analyzed the hormone levels in female mice in diestrus at 9 weeks of age ( Figure 3 ). LH and FSH levels were significantly higher in diestrus Fmr1 KO. LH doubled in KO to 0.84 ng/ml from 0.42 ng/ml in controls. Serum FSH was also higher with 4.2 ng/ml in KO, compared to 2.3 ng/ml in diestrus controls ( Figure 3A ). These results may demonstrate that high FSH leads to higher recruitment of follicles in the growing pool, which together with high LH results in more corpora lutea. Since LH and FSH β-subunit transcription, that is unique for each hormone, precedes changes in hormone concentration in the circulation, and fluctuations in mRNA levels in the gonadotrope correlate with concentration of the hormones (18), we analyzed pituitary mRNA levels ( Supplemental Figure 1 ). Both Lhb (LHβ) and Fshb (FSHβ) expression was increased in Fmr1 KO mice, while expression of the common Cga (αGSU, Glycoprotein hormones common subunit alpha), Gnrhr (GnRH receptor) or other pituitary hormones was unchanged. These results indicate that concentrations of LH and FSH in the circulation correlate with β-subunit mRNA levels.

Previous studies postulated that ovarian impairment contributed to diminished negative feedback, which in turn caused increased FSH observed in affected women (40, 41). To address this possibility, we analyzed ovarian hormones that provide feedback to the hypothalamus and pituitary in 8-week-old diestrus females before cessation of reproductive function. Steroid hormones primarily provide feedback to the hypothalamus, while inhibin regulates FSH levels (72–77). Testosterone was significantly increased in Fmr1 KO female mice, 279 pg/ml in KO compared to 200 pg/ml in controls ( Figure 3B , T). Progesterone was elevated as well to 3 ng/ml in Fmr1 KO from 1.7 ng/ml in controls ( Figure 3B , P), demonstrating that negative feedback is present and increased LH cannot be explained by reduced negative feedback. We also analyzed inhibin B levels in diestrus females in the circulation, and determined that inhibin B was higher in KO mice, 1.9 ng/ml compared to 1.5 ng/ml in controls ( Figure 3B , Inh B). Our results demonstrate that inhibin B is higher in young animals, which may be a result of larger number of follicles, or alternatively that is stimulated by higher FSH levels. This means that inhibin feedback is also present, and cannot explain elevated FSH. This implicates central mechanisms, rather than ovarian insufficiency in the reproductive phenotype of Fmr1 KO mice. Together, our results demonstrate elevated gonadotropin hormone levels, higher testosterone and progesterone, more corpora lutea, larger litters in young animals, and early cessation of reproductive function.

To discern ovarian contribution from the hypothalamic origin of the disorders, we ovariectomized (OVX) the mice and a week later analyzed LH and FSH levels. Over 10-fold higher LH and over 20-fold higher FSH confirmed the successful OVX (compare levels in Figures 3 , 4 ). Interestingly, LH remained significantly higher in OVX KO mice (8 ng/ml) compared to OVX WT mice (6 ng/ml), while there was no difference in FSH levels between WT and Fmr1 KO females after OVX. Therefore, increased LH in unmodified animals likely stems from central dysregulation, while increased FSH involves ovaries.

To address seemingly discordant results, that Fmr1 KO females exhibit higher levels of ovarian hormones than WT controls, which can provide negative feedback, and also increased FSH, which as revealed by ovariectomy is due to ovarian dysregulation, we analyzed ovarian vascularization and innervation. Using an endothelial cell marker, CD31 we stained ovarian sections and analyzed vascularization around follicles and corpora lutea. Follicles from WT and Fmr1 KO had the same degree of vascularization ( Figure 5A , representative image; Figure 5B , quantification). However, corpora lutea (CL) were more highly vascularized in Fmr1 KO than in WT mice ( Figure 5C low magnification, top). Higher magnification revealed more abundant and thicker vasculature in the Fmr1 KO CL ( Figure 5C , bottom, Figure 5D , quantification).

Although FMRP is expressed at high level in neurons, ovarian innervation was not previously examined to possibly explain ovarian phenotype observed in Fmr1 KO mice. Using antibodies to tyrosine hydroxylase, the rate-limiting enzyme in catecholamine synthesis, we counted numbers of neuronal fibers that reach the theca layer of growing follicles. Secondary follicles in Fmr1 KO ovaries had significantly more neuronal fibers than WT follicles; 4.8 average fibers per secondary follicle in KO compared to 2.3 average fibers in WT ( Figure 6A , representative images; Figure 6B quantification). Innervation of CLs was very variable within each animal and between animals, and we did not identify significant differences between WT and KO mice ( Figure 6C bottom left quarter of CL with innervation presented, Figure 6D , quantification). Together, ovarian histology demonstrated increased vascularization of corpora lutea and increased innervation of secondary follicles in Fmr1 KO mice.

Our results demonstrate elevated LH before and after ovariectomy, which is regulated by GnRH secretion. To address the mechanisms of changed LH, we examined changes in the hypothalamus, first focusing on gene expression changes caused by the loss of Fmr1 gene. Although FMRP is an RNA-binding protein that regulates protein levels of its targets, genome-wide changes in the RNA expression signatures can identify pathogenic pathways that may be indirectly regulated by FMRP. Especially since FMRP regulates levels of several transcription factors and other transcriptional regulators, which may exhibit extensive changes on the transcriptome. Previous analyses of Fmr1 KO transcriptomes focused on embryonic hippocampus and cortex, and identified overexpression of immune-related genes and downregulation of genes implicated in behavioral phenotype (78). For that reason, we used Nanostring neuroinflammation panel, which contains 770 genes implicated in neurological disorders, neuronal injury, neurotransmission, neuron-glia interactions, neuroplasticity, cell integrity, neuroinflammation, and metabolism; and added 30 custom probes for hypothalamic neuropeptides and their receptors. The complete list of genes that changed in KO compared to WT is presented by the heatmap ( Figure 7A ). There were 59 genes that were upregulated >120% from WT levels, and 39 genes that were downregulated <80% of WT levels, delineated with a dashed line. Significant changes in expression of neuropeptides and other genes of interest were highlighted in the volcano plot ( Figure 7B , log fold change vs. log p-value, statistically significant change in expression indicated with a dashed line; red, upregulated genes in KO compared to WT; green, downregulated gene in KO compared to WT). Immediate early gene, transcription factors Egr1, Fos and Jun, that are used as markers of neuronal activation, were upregulated in Fmr1 KO mice. Genes encoding GABA_A receptor γ2 subunit, (GABARγ2, Gabrg2) which is the obligatory subunit of the pentameric GABA_A receptor (79); and PSD-95 (Dlg4), a postsynaptic scaffolding protein anchoring glutamate receptors, were upregulated in KO mice. Ppfia4, involved in neurotransmitter release, and Opalin, important for oligodendrocyte differentiation, were also upregulated. On the other hand, genes correlated with DNA repair, Ercc2; neurodegenerative disorders, Serpina3n; hypoxia, Hif1a; and apoptosis, Hcar2 and Bag4, were downregulated. Importantly, genes encoding GLAST, Slc1a3, and VGLUT2, Slc17a6, were also downregulated. Of interest, neuropeptide gene encoding GnRH, Gnrh1, was upregulated, while genes for kisspeptin, Kiss1, neurokinin B, Nkb, Tac3; and cocaine and amphetamine regulated transcript, Cart, were downregulated. We confirmed changes in Gnrh1 and Kiss1 expression by qPCR of hypothalamic lysates ( Figure 7C ). GO pathway analysis indicated that pathways such as AP1 complex, comprised of Fos and Jun, myelin adaxonal regulation, spine and dendrite development were upregulated, while neuropeptide binding, ubiquitin ligase binding, and receptor signaling pathways were downregulated ( Figure 7D ).

Considering changes in excitation/inhibition synaptic balance in the cortical neurons of Fmr1 KO mice (80), we next investigated the levels of several synaptic proteins in hypothalamus at the protein level. Although previous studies showed changes in neurotransmitter receptor levels, such as GABA_A receptor for GABA, and NMDA receptor for glutamate, in several brain regions of male Fmr1 KO mice, no studies were done in females or in the hypothalamus. Hypothalami were dissected from diestrus female brains, and levels of synaptic molecules were measured by western blotting. We analyzed GABARγ2 (Gabrg2) levels in the hypothalami of KO and control female mice in diestrus, since GABA transmission activates GnRH neurons, via activation of GABA_A receptor (26, 81), and determined that Fmr1 KO females had significantly higher levels of GABARγ2 GABA_A receptor than controls ( Figure 8A , representative western blots; Figure 8B quantification). Protein levels correlate with gene expression analysis, indicating that GABARγ2 may be indirectly regulated by Fmrp. We then analyzed levels of the glutamatergic NMDA receptors (NMDARs, or NRs) (82, 83), since 30-50% of GnRH neurons respond to NMDA (84, 85). NR1 is an obligatory subunit that forms a heterotetramer with either 2A or 2B subunits (other isoforms are less frequent). We determined that NR1 (Nmdar1, GluN1) levels were increased in the hypothalami of KO mice compared to controls ( Figure 8C ). Depending on the NR2 isoform, NMDAR is localized at synapses or extrasynaptically with different effects on long-term potentiation or negative feedback, respectively (64, 86). NR2B is localized extrasynaptically, and we determined that the levels of NR2B are lower in the KO mice compared to controls ( Figure 8D ). Genes encoding these proteins did not change at the transcriptional level, which indicates that they may be regulated by Fmrp at the protein level. We also analyzed levels of PSD-95, since PSD–95 anchors glutamate receptors (87), and determined that PSD-95 protein levels were the same in Fmr1 KO and controls, which is contrary to gene expression studies and again points to direct regulation by Fmrp ( Figure 8E ). Taken together, an increase in the synaptic GABA_A and NR1 receptors, and a decrease in extrasynaptic NR2B, may contribute to altered activity of hypothalamic neurons in Fmr1 KO female mice.

To determine if Fmrp loss alters neurotransmitter receptor levels specifically in GnRH neurons, we first confirmed that 82% of GnRH neurons express Fmrp protein ( Figure 9A , left image, confocal microscopy, right image, 3D reconstruction to demonstrate Fmrp inside GnRH neurons). Antibody specificity was determined by staining the hypothalami from Fmr1 KO mice ( Supplemental Figure 2 ). We also analyzed GnRH neuron number in Fmr1 KO mice to compare to WT to ascertain if increased in GnRH neuron number contributes to higher LH. There was no difference in the number of GnRH neurons in WT and KO mice ( Supplemental Figure 2 ).

Since GABA can regulate GnRH neuron activity (26, 81), and we determined elevated GABAγ2 subunit of the GABA_A receptor in the hypothalami of KO mice compared to controls by western blot, we analyzed if GABA_A receptor immunoreactivity is increased specifically in GnRH neurons. To determine GABA_A receptor distribution in GnRH neurons, we immunostained 100 μm coronal sections of the preoptic area of the hypothalamus for GABAγ2 and GnRH. After staining, sections were imaged with high-resolution confocal microscopy. GABAγ2 receptors showed puncta-like distribution in GnRH neurons. GABAγ2 receptor puncta colocalized with GnRH immunoreactivity were identified by closely apposed puncta when no black pixels were visible between two signals in optical slices, and counted blind to condition by scrolling through the series of captured z-stack images for each GnRH soma and along the process, at 15-μm intervals, for each GnRH neuron. At least 15 neurons were counted from each mouse, and the average for each mouse was calculated (represented by a dot in the Figure 9C with bars representing group average). We determined that GABAγ2 puncta numbers increased significantly in the GnRH neuron soma and in the first 15 μm of the process proximal to the soma ( Figure 9C quantification, Figure 9B , representative images, top WT, bottom Fmr1 KO). The increase in GABAergic inputs in this area is significant, since this region of the neuron is the region where action potentials are initiated (88), and it exhibits synaptic plasticity during development and in different hormonal milieu (89–91). Since GABA is excitatory for GnRH neurons (20, 26–28), the alteration of the receptor levels may enhance GnRH neuron responsiveness and neuropeptide secretion, which in turn would cause changes in gonadotropin levels.

To determine GnRH innervation, we also analyzed appositions of GABAγ2 subunit of the GABA_A receptor with vesicular GABA transporter (VGAT), presynaptic marker of GABAergic terminals. We performed a triple stain for GnRH, GABAγ2 and VGAT, and as above, counted number of puncta where VGAT was in a close opposition to GABAγ2 in GnRH neurons ( Figure 10A ). We counted at least 15-20 neurons from each mouse, 5 pairs of mice; and used Imaris software to perform 3-D modeling of VGAT-GABA_A receptor appositions ( Figure 10B ). Fmr1 KO mice had a higher number of GABAergic appositions in GnRH neuron soma and proximal process, in the segment 1–15 μm and segment 16–30 μm from the soma, than WT controls ( Figure 10C ). The increase in synaptic GABA_A receptors in the proximal process indicates higher innervation of GnRH neurons in the area that is plastic and receives synaptic input.

Pulsatile secretion of LH strictly corresponds to GnRH secretion (49, 50). LH is used as an indicator of GnRH secretion, since GnRH secretion into median eminence cannot be measured in mice. To ascertain whether GnRH neuron secretion was affected, we measured LH pulses and used an ultrasensitive ELISA assay for LH (51) that allows for LH measurement in 5 µl of whole blood. Mice were acclimated for 2 weeks by daily tail massage. Serial sample collection every 8 min for 3 hours from the tail vein was performed (45, 52). Representative LH pulse profile from unmodified WT and KO mice are presented in Figure 11A , right side (WT control, top; Fmr1 KO, bottom). Number of LH pulses per 2.5 hours of measurement were determined using DynPeak algorithm (53) and compared between genotypes. LH, and therefore GnRH, pulse frequency was significantly higher in Fmr1 KO mice compared to WT controls ( Figure 11A , left side). Amplitude was determined by subtracting the highest LH value from the basal value prior to the onset of the pulse and averaged for each mouse ( Figure 11A , middle).

Pulsatile LH analyses were performed using ovariectomized animals as well. Frequency of LH secretion was faster in OVX Fmr1 KO animals compared to OVX WT mice ( Figure 11B , left; representative profiles right), while pulse amplitude was the same. Representative pulse profiles are shown on the right side. These experiments determined that the lack of Fmrp increased LH pulse frequency, indicating higher GnRH neuron activity corresponding to higher GABAergic innervation of GnRH neuron. It is possible the lack of Fmrp alters GnRH neuropeptide secretion leading to the faster GnRH pulse frequency and elevated LH, which contributes to ovarian dysregulation in young animals.