FSHB  Transcription is Regulated by a Novel 5′ Distal Enhancer With a Fertility-Associated Single Nucleotide Polymorphism

Abstract

The pituitary gonadotropins, follicle-stimulating hormone (FSH) and luteinizing hormone, signal the gonads to regulate male and female fertility. FSH is critical for female fertility as it regulates oocyte maturation, ovulation, and hormone synthesis. Multiple genome-wide association studies (GWAS) link a 130 Kb locus at 11p14.1, which encompasses the FSH beta-subunit (FSHB) gene, with fertility-related traits that include polycystic ovary syndrome, age of natural menopause, and dizygotic twinning. The most statistically significant single nucleotide polymorphism from several GWAS studies (rs11031006) resides within a highly conserved 450 bp region 26 Kb upstream of the human FSHB gene. Given that sequence conservation suggests an important biological function, we hypothesized that the region could regulate FSHB transcription. In luciferase assays, the conserved region enhanced FSHB transcription and gel shifts identified a binding site for Steroidogenic factor 1 (SF1) contributing to its function. Analysis of mouse pituitary single-cell ATAC-seq demonstrated open chromatin at the conserved region exclusive to a gonadotrope cell-type cluster. Additionally, enhancer-associated histone markers were identified by immunoprecipitation of chromatin from mouse whole pituitary and an immortalized mouse gonadotrope-derived LβT2 cell line at the conserved region. Furthermore, we found that the rs11031006 minor allele upregulated FSHB transcription via increased SF1 binding to the enhancer. All together, these results identify a novel upstream regulator of FSHB transcription and indicate that rs11031006 can modulate FSH levels.

Follicle-stimulating hormone (FSH) is a key regulator of fertility. In females, FSH is critical for folliculogenesis and steroidogenesis. FSH drives oocyte maturation by promoting estrogen synthesis and preparing granulosa cell responsiveness to the ovulatory luteinizing hormone (LH) surge by inducing the expression of LH receptors (1, 2). FSH accelerates the conversion of testosterone to estrogen through its action on granulosa cells, where it regulates the expression of aromatase, the enzyme responsible for this conversion (3). In males, FSH stimulates Sertoli cell differentiation and proliferation during development and promotes spermatogenesis (4, 5). FSH is a glycoprotein hormone consisting of a unique beta-subunit (FSHβ), encoded by the FSHB gene, and a common alpha subunit (αGSU), encoded by the CGA gene, which also is a component of the other glycoprotein hormones including LH, thyroid-stimulating hormone, and human chorionic gonadotropin. The gonadotropins, FSH and LH, are both synthesized and secreted by gonadotrope cells within the anterior pituitary.

Constitutive FSH secretion is maintained throughout the cycle. Studies using immortalized mouse gonadotrope-derived LβT2 cells (6) have shown that transcription factors including Paired-like homeodomain 1 (PTX1), LIM homeobox gene 3 (LHX3), and Steroidogenic factor 1 (SF1, encoded by NR5A1) in cooperation with Nuclear transcription factor Y (NFY) bind to the Fshb proximal promoter to regulate basal transcription (7-9). FSH serum levels are primarily limited by the rate of Fshb transcription (10, 11).

In the human menstrual cycle, FSHB mRNA synthesis and FSH serum levels peak midcycle at the time of the LH surge and rise again during the late luteal/early follicular phase to initiate follicular maturation (12). Similarly, in rodents, there is a primary rise that occurs during LH-induced ovulation and a secondary rise that occurs during estrus (13). These increases in FSH have been shown to be mediated, in part, by gonadotropin-releasing hormone (GnRH) and activin. While hypothalamic GnRH and steroid hormones regulate FSH and LH transcription and secretion, FSH is also regulated by activins, inhibins, and follistatin (14-16). Activin signaling phosphorylates SMAD proteins, which bind to the proximal promoter of FSHB and activate its transcription in conjunction with the FOXL2 transcription factor (17, 18).

In humans, inactivating (loss-of-function) mutations in FSHB result in infertility in both sexes (19). Females fail to go through puberty (20-23), while both normal and absent puberty have been reported in males (22, 24, 25). Female, but not male, Fshb knockout mutant mice are infertile, and the phenotype of the female recapitulates failure to ovulate and reduced uterine and ovarian weight (26). Conversely, high FSH levels in females are associated with premature ovarian failure (27). Therefore, FSH serum levels must be tightly controlled to maintain female fertility.

Recent genome-wide association studies (GWAS) from both Caucasian and Han Chinese populations identified a 130 Kb locus, mapping to the chromosome 11p14.1 region containing FSHB, that is associated with polycystic ovary syndrome (PCOS) (28-31), age of natural menopause (32), and dizygotic twinning (33). Notably, these fertility-associated traits and diseases are all linked with FSH levels (28-33). PCOS is the most common cause of anovulatory infertility, affecting up to 10% to 15% of reproductive-age women (34). PCOS diagnosis with the Rotterdam criteria requires 2 of 3 diagnostic features: elevated androgens, anovulation, or the presence of ovarian cystic follicles (35). The 11p14.1 locus has also been associated with decreased FSH, increased LH, and an elevated LH/FSH ratio, characteristic of some PCOS patients (28-31, 36, 37). Strikingly, LH levels are also elevated in FSH-deficient female mouse models and in women with inactivating mutations of FSHB (19-21, 23, 26, 38), suggesting that FSH insufficiency may contribute to elevated LH in women with PCOS. Additionally, GWAS studies for PCOS identified risk loci encompassing the FSH and LH receptors (39, 40), suggesting that disturbances in gonadotropin production and action are potential mechanisms that contribute to PCOS etiology and pathophysiology.

The lead single nucleotide polymorphism (SNP) identified in age of natural menopause, dizygotic twinning, and a subset of PCOS GWAS studies is rs11031006 (G/A) (28, 29), which is located approximately 26 Kb upstream of the human FSHB transcriptional start site (TSS). The worldwide minor (A) allele frequency is approximately 8% (41). As of yet, the only polymorphism associated with FSHB to be functionally interrogated is the rs10835638 (G/T) SNP, located 211 base pairs upstream of the FSHB TSS in the proximal promoter. In humans, the rs10835638 minor (T) allele has been associated with lower FSH levels in males and infertility in both sexes (42-46). The minor (T) allele was found to decrease the binding of the LHX3 homeodomain transcription factor to the proximal promoter and reduce basal FSHB transcription in cell culture (47).

The rs11031006 SNP resides within a noncoding region that is highly conserved among vertebrates (Fig. 1A). We hypothesized that this conserved intergenic region is an enhancer of FSHB transcription, and that FSHB expression would be decreased by the disease-associated variant. Consistent with our hypothesis, we determined that the conserved enhancer region increased transcription driven by the FSHB proximal promoter via binding of the SF1 transcription factor to the enhancer region. This region also possessed markers of an enhancer including open chromatin exclusively in gonadotrope cells within the pituitary and enrichment of the histone markers H3K4me1 and H3K27Ac, which are associated with active enhancers (48, 49). Contrary to our expectations, conversion to the rs11031006 minor (A) allele increased SF1 binding to the enhancer and increased, rather than decreased, FSHB expression, indicating that the mechanism linking this SNP to fertility-related traits still remains to be elucidated.

Materials and Methods

Plasmids

The human −1028/+7 FSHB luciferase reporter plasmid (−1028/+7 FSHB-luc) in a pGL3 backbone (Promega) was provided by Daniel Bernard (50). A conserved intergenic region 26 Kb upstream of FSHB, spanning 450 base pairs (chr11:30,204,683-30,205,132, hg38 assembly), was amplified by polymerase chain reaction (PCR) from a human fosmid template (GenBank accession number: AL358944.12) containing the major allele of the rs11031006 SNP (G) using primers “FW human enh-KpnI” and “RV human enh-SacI” to add each restriction site (Table 1). It was ligated into the human −1028/+7 FSHB-luc reporter plasmid between the KpnI/SacI sites, directly upstream of the proximal promoter, to create the hEnh/G:1028 FSHB-luc plasmid. Truncations of the upstream region were constructed in the same way, with primers designed for intermediate sites (Table 1). To restore restriction sites within the pGL3 backbone, the mouse −1000/−1 Fshb-luc plasmid was constructed by amplifying the promoter region of the previously described mouse −1000FSHβluc plasmid (51) with PCR using primers “FW mouse pro-MluI” and “RV mouse pro-XhoI” (Table 1) to add each restriction site. The amplicon was ligated into the MluI/XhoI sites of the pGL3-Basic plasmid. The mouse equivalent, conserved 450-base-pair element 17 Kb upstream of Fshb (chr2:107,076,909-107,077,358, mm10) was amplified from C57BL/6 genomic mouse DNA with primers “FW mouse enh-KpnI” and “RV mouse enh-SacI” (Table 1) and subcloned into the −1000/−1 Fshb-luc plasmid at the KpnI/SacI sites to create the mEnh/G:1000 Fshb-luc plasmid. Plasmids with the human and mouse enhancer in reverse orientation (RV-hEnh/G:1028 FSHB-luc and RV-mEnh/G:1000 Fshb-luc) were constructed exactly as the forward enhancer plasmids except using primers with the KpnI and SacI sites switched. Plasmids containing the herpes simplex thymidine kinase (−81/+52 TK) promoter were constructed by subcloning the enhancer variants referenced above into the KpnI/SacI sites upstream of the TK promoter in a pGL3 backbone (52). Plasmids containing the mouse −523/+6 Lhb and −493/+9 Gnrhr promoters were constructed by amplifying each region with the “FW mouse Lhb-MluI” and “RV mouse Lhb-XhoI” or “FW mouse Gnrhr-MluI” and “RV mouse Gnrhr-XhoI” primer sets from C57BL/6 mouse genomic DNA and cloning into the MluI/XhoI sites in the human enhancer construct in place of the FSHB promoter to create hEnh/G:Lhb-luc and hEnh/G:Gnrhr-luc. The SF1 expression vector in pcDNA3 (Invitrogen) was cloned from a pCMV expression vector originally provided by Bon-Chu Chung (53). All plasmids were confirmed by Sanger sequencing (Eton Bioscience).

Mutagenesis

Construction of the SF1 site mutation plasmids, the rs11031006 (G>A) minor allele plasmids, and 10-base-pair deletions within the human conserved region was performed using the Quikchange II Site-Directed Mutagenesis Kit (Agilent, Cat # 200523) according to the manufacturer’s protocol. The hEnh/G:1028 FSHB-luc and the RV-mEnh/G:1000 Fshb-luc plasmids were used as templates. Primers were designed using the “Quikchange Primer Design Program” from the manufacturer’s website (54). Forward primer sequences are listed in Table 1; reverse primer sequences are the reverse complement.

Cell culture and transient transfection

LβT2 and NIH 3T3 cells were maintained in Dulbecco’s Modified Eagle Medium (DMEM) (Corning, cat # 10-013-CV) supplemented with 10% fetal bovine serum (Omega Scientific, cat # FB-01) and 1% penicillin-streptomycin (GE Life Sciences, cat # SV30010), and incubated at 37 °C, 5% CO₂. They were passaged at approximately 80% confluency by dissociating cells with 0.25% trypsin-EDTA (Gibco, cat # 25200056). Prior to transfection, 4.25 × 10⁵ LβT2 cells or 5 × 10⁴ NIH 3T3 cells were plated into 12 well plates (Thermo Scientific, cat # 150628) and cultured overnight. The following morning, cells were transfected using Polyjet (Sygnagen Laboratories, cat # SL100688) according to the manufacturer’s protocol. Each well was co-transfected with 500 ng of the luciferase reporter construct and 200 ng of a reporter plasmid encoding β-galactosidase driven by the herpes virus thymidine kinase promoter to control for transfection efficiency. NIH 3T3 cells were also transfected with 20 ng of SF1 expression vector or the same amount of the pcDNA3 backbone. After 5 hours, the transfection agent was removed and replaced with supplemented DMEM. Three technical replicates were included for each reporter.

Luciferase assay

Approximately 48 hours after transfection, cells were washed with 1× phosphate-buffered saline (PBS) and incubated for 5 minutes at room temperature in lysis buffer (0.1 M potassium phosphate [pH 7.8] and 0.2% Triton-X-100). Lysate was transferred into two 96-well plates. Luciferase and β-galactosidase activity were measured using a Veritas Microplate Luminometer (Turner BioSystems). For luciferase activity, 25 μL of lysate was treated with 100 μL of luciferase buffer (25 mM Tris-HCl [pH 7.8], 15 mM MgSO4, 10 mM ATP, and 65 μM luciferin) with a 1-second delay before measuring luminescence over 1 second. β-Galactosidase activity was assayed using Galacto-Light Plus reagents (Tropix, cat# T1009) according to the manufacturer’s protocol.

Single-cell assay for transposase-accessible chromatin using sequencing (scATAC-seq)

Pituitaries were collected from 2 adult male mice. Nuclei isolation for scATAC-seq was performed following the instruction provided by 10x Genomics (scATAC-V1) with some modifications. Briefly, 100 000 cells were added to a 2 mL microcentrifuge tube and centrifuged (300g for 5 minutes at 4 °C). The supernatant was removed without disrupting the cell pellet. Lysis Buffer [10 mM Tris-HCl (pH 7.4), 10 mM NaCl, 3 mM MgCl₂, 0.1% Tween-20, 0.1% Nonidet P40 Substitute, 0.01% Digitonin, and 1% BSA] was diluted by Lysis Dilution Buffer (10 mM Tris-HCl [pH 7.4], 10 mM NaCl, 3 mM MgCl₂, 1% BSA) to make 0.1× Lysis Buffer. Then 100 μL chilled 0.1× Lysis Buffer was added to the cell pellet, and pipette-mixed 10 times. The microcentrifuge tube was then incubated on ice for 5 minutes. Next, 1 mL of chilled Washing Buffer (10 mM Tris-HCl [pH 7.4], 10 mM NaCl, 3 mM MgCl₂, 0.1% Tween-20, and 1% BSA) was added and gently pipette-mixed 5 times. Nuclei were centrifuged (500g for 5 minutes at 4 °C) to remove the supernatant, and the wash was repeated. An appropriate volume of chilled Diluted Nuclei Buffer (10x Genomics) was added to resuspend nuclei at 2500 nuclei/μL to retrieve/target 6000 nuclei per library following the standard protocol provided by Chromium Single Cell ATAC Library and Gel Bead kit (10x Genomics, v1). Each sample library was uniquely barcoded. Libraries were then pooled and loaded on a NovaSeq Illumina sequencer and sequenced to ~50% saturation on average.

scATAC-seq analysis

Raw sequencing data were converted to FASTQ format using cell ranger atac mkfastq (10x Genomics, v.1.0.0). Read counts of a single library were aligned to mm10 reference genome for quantification and analyzed by cellranger-atac count (10x Genomics, v.1.0.0). To ensure identical detection sensitivity across libraries, we performed cellranger-atac aggr on integrated libraries and normalized all the data to identical sequencing depth. Pituitary cell-type clusters were obtained using t-distributed stochastic neighbor embedding (t-SNE) (55) to cluster together cells with similar open chromatin profiles. Processed and raw data have been deposited and can be downloaded from NCBI GEO, accession number GSE158070 (56).

Chromatin immunoprecipitation

Chromatin immunoprecipitation (ChIP) was performed using the Active Motif ChIP-IT High Sensitivity kit (cat # 53040). LβT2 cells were cultured in DMEM supplemented with 10% fetal bovine serum and 1% penicillin-streptomycin in 15-cm tissue culture dishes. Cells were harvested at approximately 80% confluency and fixed with “Complete Cell Fixation Solution” containing 1.3% formaldehyde for 10 minutes. Chromatin was extracted according to the manufacturer’s protocol, with the following modifications: After addition of the “Chromatin Prep Buffer” supplemented with 100 nM PMSF and 1 μL/mL of “Protease Inhibitor Cocktail,” the cells were lysed by passing through a 25-gauge needle 4 times. Prior to sonication, nuclei were diluted with “ChIP Buffer” to a density of approximately 3 million per 100 μL. Cells were sonicated in 300 μL aliquots in 1.5 mL Bioruptor Pico Microtubes (Diagenode, cat# C30010016) using the Bioruptor Pico (Diagenode, cat# B01060010), 30 seconds on/30 seconds off for 20 minutes total. Following centrifugation at 16 000g for 2 minutes to remove cellular debris, 25 μL of chromatin was collected as input. Immunoprecipitation was performed according to manufacturer’s protocol, using approximately 10 μg of chromatin per reaction. Five μg of each antibody targeting histone H3 lysine 27 acetylation (H3K27Ac) (Active Motif, cat # 93133) (57), and histone 3 lysine 27 trimethylation (H3K27me3) (Active Motif, cat # 39155) (58) were compared with a normal rabbit IgG control (Cell Signaling, cat # 2729) (59). Ten μL of antibody targeting histone H3 lysine 4 monomethylation (H3K4me1) (Active Motif, cat # 39297) (60) was compared with a normal rabbit serum control (Molecular Probes, cat # PLN5001). ChIP DNA was eluted in a total volume of 200 μL. Input chromatin was treated with RNAse A and Proteinase K. DNA was purified using the Active Motif Chromatin IP DNA Purification Kit (cat # 58002) and eluted in a total volume of 200 μL.

For the whole pituitary ChIP, 4 to 12 pituitaries from random-cycling adult C57BL/6 female mice were pooled in a glass Dounce and homogenized in “Complete Tissue Fixation Solution” containing 1% formaldehyde in PBS for 10 minutes. Chromatin was extracted as above. Prior to sonication, nuclei were diluted with “ChIP Buffer” to a density of approximately 2 million per 100 μL. Cells were sonicated in 150 to 300 μL aliquots in 1.5 mL Bioruptor Pico Microtubes (Diagenode, cat# C30010016) using the Bioruptor Pico (Diagenode, cat# B01060010), 30 seconds on/30 seconds off for 10 minutes total. Immunoprecipitation was performed according to manufacturer’s protocol, using approximately 25 μg of chromatin per reaction. Five μg of each antibody targeting H3K27Ac (Active Motif, cat # 93133) (57) or H3K4me1 (Active Motif, cat # 61633) (61) were compared with a normal rabbit IgG control (Cell Signaling, cat # 2729) (59). Input was collected from the flow-through of the IgG immunoprecipitation. Input and samples were treated with Proteinase K, purified using the QIAquick PCR Purification kit (Qiagen, cat# 28104), and eluted in a total volume of 100 μL.

Quantitative PCR analysis

Primers targeting the mouse conserved intergenic region (putative enhancer), Fshb proximal promoter, and control regions (Table 2) were designed using IDT PrimerQuest (62) and NCBI Primer-BLAST (63). DNA from input and immunoprecipitated samples were measured using iQ SYBR Green Supermix (Bio-Rad Laboratories, cat # 1708880) in a CFX Connect Detection System (Bio-Rad Laboratories). A standard curve of serial input dilutions was constructed for each plate and used to compute the concentration of each sample as % input. A dissociation curve was performed following PCR to ensure the presence of a single product. Three technical replicates were included per experiment.

Electrophoretic mobility shift assay

To prepare nuclear extracts, LβT2 cells were washed once with PBS and lysed in a buffer containing 20 mM Tris pH 7.4, 10 mM NaCl, 1 mM MgCl₂, 1 mM PMSF, 10 mM NaF, and 10 μL/mL of protease inhibitor cocktail (Sigma, cat # P8340). Cells swelled on ice for 15 minutes before they were passed 4 times through a 25G needle. Nuclei were collected by centrifugation at 3750g for 4 minutes at 4 °C. Nuclei were lysed in a hypertonic buffer containing 20 mM HEPES pH 7.9, 20% glycerol, 420 mM KCl, 1.5 mM MgCl₂, 0.5 mM EDTA, 0.1 mM EGTA, 1 mM PMSF, 10 mM NaF, and 10 μL/mL of protease inhibitor cocktail. Complementary oligonucleotides (Table 3) were annealed and ATP γ-32P end-labeled with T4 Polynucleotide Kinase (New England Biolabs, cat# M0201). Two μg of nuclear extract was incubated with labeled probes for 30 minutes in a buffer containing 10 mM HEPES pH 7.8, 50 mM KCl, 5 mM MgCl₂, 0.1% NP-40, 10% glycerol, 1 mM DTT, 0.1 mg/mL poly dI-dC, and 2 μg anti-SF1 (Millipore, cat # 07-618) (64), 2 μg rabbit IgG (Santa Cruz Biotechnology Cat# sc-2027) (65), or excess of cold competitor oligonucleotides where specified. Complexes were separated on a 5% acrylamide gel with 2.5% glycerol in 0.25X TBE. Gels were dried and exposed at −80°C or room temperature as indicated. Images are representative of at least 3 experiments yielding the same results.

Statistical analysis

Differences were analyzed via Student t test or ANOVA. One-way and two-way ANOVAs were followed by post hoc Tukey-Kramer honestly significant difference test when comparing all samples or Dunnett test when comparing to the mean of a control sample. Residuals were checked for normality using the Shapiro-Wilk test with P > 0.05 as the threshold for normality. As needed, data were subsequently log or square transformed and reanalyzed where indicated. For transient transfection, luciferase values were normalized to β-galactosidase values from the same tissue culture well. Triplicates were averaged, and results were expressed relative to pGL3 backbone levels from the same experimental replicate except where indicated. Prism 8 (GraphPad) was used for all analysis, and p ˂0.05 was the threshold for statistical significance.

Results

Conserved putative enhancer region increases transcription from the FSHB promoter

The highly conserved, putative enhancer region containing the rs11031006 SNP spans approximately 450 base pairs (Fig. 1A). In the human genome, it is located ~26 Kb upstream of the FSHB TSS, with the equivalent region ~17 Kb upstream of the Fshb TSS in the mouse genome. To determine whether the human conserved intergenic region containing the major rs11031006 allele (G) can function as an enhancer of FSHB transcription, we transfected gonadotrope-derived LβT2 cells (6) with hEnh/G:1028 FSHB-luc. The conserved element subcloned in the forward direction (hEnh/G) increased luciferase activity 2.4-fold compared with the FSHB promoter alone (Fig. 1B). Additionally, the conserved element subcloned in the reverse direction (RV-hEnh/G) increased luciferase activity 2.2-fold compared with the FSHB promoter alone (Fig. 1B). Reversibility is a classical property of enhancers (66). The results with the analogous mouse construct mEnh/G:1000 Fshb-luc were similar to the human construct, with a 1.7-fold (forward orientation) and 2.5-fold (reverse orientation) increase in luciferase activity compared with the Fshb promoter alone (Fig. 1C). Overall, these results demonstrated that the conserved intragenic element from both the human and mouse genomes enhanced expression driven from the FSHB promoter. From here onward, we will refer to the conserved region as the human FSHB enhancer or mouse Fshb enhancer.

To test whether the mouse and human enhancers are promoter-specific, we transfected LβT2 cells with hEnh/G:TK and mEnh/G:TK, which contain the herpes simplex virus thymidine kinase (TK) promoter driving luciferase expression. The human FSHB enhancer increased luciferase activity in the forward orientation (4.5-fold) (Fig. 1D), but the mouse Fshb enhancer did not (Fig. 1E). Because the mouse enhancer was stronger in the reverse orientation upstream of the Fshb promoter, we assessed whether it could act as an enhancer in this orientation on TK and found that reversing the enhancer restored its activity (1.6-fold).

To determine if the FSHB enhancer can also function upstream of other mammalian-derived promoters, we transfected LβT2 cells with constructs containing hEnh/G upstream of the Lhb (Fig. 1F) and Gnrhr promoters (Fig. 1G), both representing genes actively transcribed in gonadotropes. As with the TK promoter, the human FSHB enhancer increased expression driven by the Lhb and Gnrhr promoters. Overall, the enhancer is not promoter-specific, and its reversibility may be context-dependent.

Fshb enhancer is characterized by open chromatin exclusively in gonadotropes

Open chromatin structure facilitates gene expression by allowing transcriptional regulators and RNA polymerases to access DNA, while closed chromatin typically inhibits these interactions (67). To evaluate chromatin status in gonadotropes, scATAC-seq data from whole pituitary was analyzed by t-SNE (55) to cluster cells based on similar regions of accessible chromatin (Fig. 2). A gonadotrope cell-type cluster (Fig. 2A) was identified by open chromatin at genes expressed solely in pituitary gonadotropes such as Fshb, Lhb (encoding the LH beta-subunit), and Gnrhr (encoding the GnRH receptor). The coding regions of all 3 genes were open exclusively in this cluster which included 3.3% of cells, consistent with previous reports of the gonadotrope cell population size relative to total number of pituitary cells (68) (Fig. 2B and 2C). Because the closed chromatin status at Fshb, Lhb, and Gnrhr was consistent among all other clusters identified by t-SNE (shown by individual cell type in Fig. S1) (69), we combined them into one group (“other”) for simplicity (Fig. 2). To evaluate whether the Fshb enhancer was specifically open in gonadotropes, we compared the chromatin status in this region between the gonadotrope cell-type cluster and all other pituitary cell-type clusters (Fig. 2C). The enhancer was only accessible in gonadotrope cells. All findings were replicated in a second experiment (Fig. S2) (69).

H3K4me1 is enriched near the Fshb enhancer

Specific histone modifications in a region of chromatin correlate with its function. As enhancers are marked by H3K4me1, we used a ChIP assay to assess the presence of this marker at the Fshb enhancer in LβT2 cells (Fig. 3A). As compared with the negative (gene desert) control, H3K4me1 was enriched at the enhancer and the positive control region (β-actin intron 1). At the proximal Fshb promoter, H3K4me1 was also slightly elevated as compared with the negative control region, although it did not reach statistical significance. H3K4me1 has been reported to be highest at enhancers, moderately elevated immediately downstream of the TSS, and lowest at proximal promoters and nonregulatory intergenic regions (70), consistent with our results.

Next, we asked whether the enhancer is active in this cell line, especially given that endogenous basal Fshb expression is low in LβT2 cells (71). Active enhancers and promoters are characterized by enrichment of H3K27Ac and an absence of the repressive marker H3K27me3, whereas poised or inactivated enhancers show the opposite pattern (48, 49). As predicted from low Fshb expression levels in LβT2 cells, H3K27Ac was not enriched at the Fshb enhancer, Fshb proximal promoter, or negative (gene desert) control in LβT2 cells as compared with the positive control region (β-actin intron 1) (Fig. 3B). Interestingly, H3K27me3 was also absent at the Fshb enhancer, Fshb promoter, and negative control region (β-actin intron 1) compared with the positive control MyoD, which is not expressed in LβT2 cells (Fig. 3C). Enhancers negative for both H3K27me3 and H3K27Ac have been classified as “intermediate” between active and poised and are correlated with lower expression levels of their target genes than observed from active enhancers, but are not correlated with Polycomb repression proteins typical of “poised” enhancers, possibly indicative of a transitionary stage (72).

Because we did not observe H3K27Ac enrichment in LβT2 cells, we performed ChIP in female whole pituitary to assess H3K27Ac recruitment and also to confirm the presence of H3K4me1 at the Fshb enhancer. The repressive marker H3K27me3 is expected to be present in non-Fshb cell linages, so it could not be measured in this context. As compared with the negative (gene desert) control region, H3K4me1 was elevated at the Fshb enhancer as well as the Fshb proximal promoter in whole pituitary (Fig. 3D). In contrast to LβT2 cells, H3K27Ac was enriched at the Fshb enhancer and Fshb proximal promoter compared with the negative (gene desert) control (Fig. 3E). A well-characterized enhancer 10 Kb upstream of the POU1F1 TSS (expressing PIT1) was selected as a positive control for H3K4me1 and H3K27Ac immunoprecipitation in whole pituitary, but levels were not directly compared with the Fshb enhancer and promoter because POU1F1 is expressed in most pituitary cells (Fig. 3F) (73). Together, these results confirm that the Fshb enhancer is positive for the active enhancer markers H3K4me1 and H3K27Ac in adult female pituitary.

The 216-341 region within the enhancer is sufficient to increase transcription

To narrow the search for regulatory sites within the conserved enhancer region, we cloned subregions of the human FSHB enhancer upstream of the −1028/+7 FSHB-luc promoter and assessed luciferase expression (Fig. 4A). Subregions were selected to divide the enhancer into halves and quarters to identify the minimal region sufficient for enhancer activity. A reporter including the subregion spanning 216-341 (chr11:30,204,898-30,205,023, hg38) significantly enhanced luciferase expression 2.2-fold over the promoter alone. This result was comparable to the 2.4-fold increase by the full 450 bp enhancer and 1.9-fold increase by 216-450. The 216-341 subregion maintained enhancer function in the reverse orientation, although its activity was reduced as compared with the forward orientation (Fig. 4B). Together, these results indicate that 216-341 is sufficient for basal enhancer function.

Putative ZEB1, PTX1, SMAD, and SF1 binding sites are important for enhancer function

We next used site-directed mutagenesis to create a series of plasmids containing the 450 bp human FSHB enhancer with sequential 10-base-pair deletions within the identified minimal region sufficient for basal enhancer activity, 216-345. Compared with the intact human FSHB enhancer, 1 deletion increased and 3 decreased enhancer activity (Fig. 4C), demonstrating that several regulatory elements within the 216-345 region of enhancer contribute to its activity. Using Match 1.0 software (GeneXplain) to search the TRANSFAC database, we identified putative transcription factor binding motifs corresponding to each region including ZEB1 (Δ226-235), PTX1 (Δ246-255), SMAD (Δ256-265), and SF1 (Δ296-305) (Fig. 4C). Interestingly, the deletion from 296-305 contains the rs11031006 SNP, suggesting that the SNP resides within, and could possibly alter, a functional element. Given its potential relevance for FSHB regulation in health and in disease, we further investigated the role of SF1 binding at this element.

The rs11031006 G>A SNP increases SF1 binding to the enhancer

The rs11031006 SNP falls within an element with striking similarity to the SF1 consensus sequence (74) (Fig. 5A). SF1 is a transcriptional activator that is critical for fertility in both sexes and regulates multiple genes at all levels of the hypothalamic-pituitary-gonadal (HPG) axis (75). SF1 has been shown to bind to the proximal promoters of Lhb, Fshb, and Cga to regulate basal expression of the gonadotropin genes (76). The human minor (A) allele of rs11031006 corrects a mismatch between the SF1 consensus sequence and the sequence with the major (G) allele, creating a full SF1 consensus sequence (Fig. 5A). To test whether SF1 can bind the putative SF1 binding element at 296-305 and to compare binding between the major and minor allele, we performed an electrophoretic mobility shift assay (EMSA) using 30 bp radiolabeled double-stranded DNA oligonucleotides (Table 3) from mouse and human containing the rs11031006 SNP. Incubation of the probe containing the rs11031006 major (G) allele with nuclear extracts isolated from LβT2 cells resulted in a DNA-protein complex that was supershifted when it was incubated with a SF1 antibody, indicating that SF1 binds to this region of the enhancer (Fig. 5B). Notably, binding of SF1 increased when the labeled probe contained the rs11031006 minor (A) allele. A similar pattern was seen for probes corresponding to both human and mouse enhancer sequences encompassing the SF1 binding element.

To further confirm that the DNA-protein complex contained SF1, competition with various probes in excess was utilized. A labeled probe containing an SF1 consensus sequence (also called gonadotrope-specific element, GSE) (77) from the CGA promoter was competed by cold human major (G) allele or minor (A) allele probe or the corresponding probes from mouse (Fig. 5C). All human and mouse probes were able to outcompete GSE. In agreement with SF1 binding in Fig. 4B, the minor (A) allele outcompeted the labeled probe at a lower concentration than was needed for competition by the major (G) allele for both human and mouse. Furthermore, SF1 binding to the probe was eliminated by a 2-base-pair mutation (CAGGGTCA >CAGTTTCA) within the SF1 motif (Fig. 5D). This mutation does not target the rs11031006 base pair. These findings confirm that SF1 binds to this region of this enhancer and suggests that rs11031006 could affect FSHB expression by altering SF1 binding.

The rs11031006 site increases enhancer activity

Because the rs11031006 minor allele G>A increases SF1 binding to the enhancer in the gel-shift assay (Fig. 5), it is important to determine whether this SNP affects enhancer activity. We used site-directed mutagenesis to create hEnh/A:1028 FSHB-luc, containing the minor (A) allele at the rs11031006 site. Activity from the rs11031006 minor allele construct (hEnh/A) was increased 1.5-fold compared with the major allele construct (hEnh/G) and 4.3-fold compared with promoter alone (Fig. 6A). An equivalent mutation was created in the reversed mouse Fshb enhancer since it had higher activity than the forward direction, and it produced a similar increase in transcription compared with the wild-type (WT) enhancer (2.4-fold) or promoter alone (4.2-fold) (Fig. 6B). These results indicate that the rs11031006 minor (A) allele has a distinct effect on gene expression driven from the FSHB promoter compared with the major (G) allele.

SF1 binding site contributes to enhancer function

We next sought to further assess how the SF1 site affects function of the human FSHB enhancer. To expand upon the effect of the deletion from 296-305 (Fig. 4B), the SF1 binding site was more narrowly targeted by creating the same 2-base-pair mutation in the human FSHB enhancer as in Fig. 5D. The SF1 mutation significantly decreased luciferase activity to 0.8-fold WT levels (Fig. 6C). This result was consistent with the deletion from 296–305, although luciferase activity driven by the SF1 mutant human FSHB enhancer was still greater than from the promoter alone (1.8-fold). An equivalent mutation in the mouse enhancer similarly decreased luciferase activity to 0.8-fold WT levels but was still elevated 1.5-fold relative to promoter alone (Fig. 6D). To determine if SF1 overexpression would increase transcription from the TK promoter linked to the FSHB enhancer, NIH 3T3 cells were transfected with an SF1 expression vector (Fig. 6E). NIH 3T3 cells do not express endogenous SF1. As compared with the empty pcDNA3 expression vector, SF1 produced a small but significant increase in transcription (1.2-fold), suggesting that it is sufficient for activation of transcription through the enhancer. These results indicate that the SF1 binding site contributes to both the human and mouse enhancer activity in vitro.

Discussion

Precise regulation of FSH is critical for optimal fertility. For this reason, identifying FSHB transcriptional regulatory elements is necessary to understand how genetic variants within these elements might contribute to the pathophysiology of disease. In this study, we identified a novel upstream enhancer of FSHB, confirming that it is reversible, accessible in gonadotropes, and marked by histone modifications associated with active enhancers.

As of yet, there are no ENCODE ChIP-seq databases for gonadotropes. It was previously reported that a potential enhancer, marked by open chromatin and H3K27Ac and H3K1me1 peaks in ENCODE data, resides about 2 Kb downstream of the rs11031006 variant (30). However, these data were from a lymphoblastoid cell line, which is not expected to express FSHB, and our scATAC-seq analysis did not show that this region is accessible to transposase in gonadotropes. In contrast, our results clearly demonstrated that the chromatin was accessible at the enhancer region 17 Kb upstream of the Fshb promoter and that H3K4me1 and H3K27Ac were enriched at this element. While enrichment of the enhancer-specific histone marker H3K27me1 and an absence of the repressive marker H3K27me3 were detected at the 450 bp upstream region in LβT2 cells, the active regulatory marker H3K27Ac was not enriched, although levels were comparable to the proximal promoter. This result contrasted with the results in whole pituitary in which both H3K4me1 and H3K27Ac were enriched. Enhancers positive for H3K4me1 and negative for H3K27 acetylation or methylation have been termed “intermediate” and can be activated by stimulatory cues (72). Bulk ATAC-seq of GnRH-treated LβT2 cells showed that the Fshb enhancer was not accessible in LβT2 cells (78). Consistent with that finding, Fshb mRNA expression in LβT2 cells was relatively low in the absence of activin stimulation (71). Thus, stimulatory cues, such as activin and/or steroid hormones, potentially signaling through the putative SMAD binding site, may be necessary for H3K27Ac recruitment and chromatin accessibility in gonadotropes. How this affects enhancer accessibility with regards to sex, developmental stage, and estrous cycle stage remains an intriguing question for future studies.

To mechanistically interrogate the enhancer, we used systematic deletion and sequence analysis to identify an SF1 binding site that overlaps with a fertility-associated variant and contributes to the function of the enhancer. We also predicted binding sites for the transcription factors PTX1 and SMADs. Deletion of each individual element reduced enhancer-mediated transcription to promoter-only levels, potentially indicating that physical interactions between SF1, PTX1, and SMADs are necessary for FSHB upregulation by the enhancer. In gonadotropes, PTX1/SMAD have been demonstrated to interact on the Fshb and Lhb proximal promoters and PTX1/SF1 interact on the Lhb proximal promoter (7, 79-81). Together, SF1 and PTX1 may confine enhancer activity to gonadotropes while the SMAD site could confer activin responsiveness, although this remains to be tested experimentally.

It was previously shown that a randomly integrated, 10 Kb human transgene encoding FSHB flanked by 4 Kb of upstream and 2 Kb of downstream sequence was sufficient to restore gonadotrope-specific expression and fertility in Fshb knockout mice (82). While this implies that regulatory elements within the transgene were sufficient to localize FSHB expression to gonadotropes and restore fertility, this study was not designed to evaluate FSHB expression levels driven by the transgene, as copy number of the FSHB transgene and the effect of any regulatory elements near the site of random transgene integration were not able to be controlled. Therefore, it is difficult to infer whether additional intragenic sequences not incorporated in the human transgene can also modulate FSHB gene expression levels. Our findings widen the scope of FSHB regulation beyond the boundaries of the transgene, raising the potential for this element or additional novel regulatory regions to explain unanswered questions related to FSHB gene regulation. In addition to the enhancer element described in this study, scATAC-seq revealed an additional region of open chromatin located 4 Kb more proximal to the Fshb TSS and possibly indicative of an additional enhancer to be considered in future studies.

GWAS is a powerful technique for identifying genetic correlants of specific diseases or physiological phenotypes, but it is limited because it does not provide a direct assessment of causality. Variants in high linkage disequilibrium segregate together, confounding whether a disease-correlated SNP has a functional effect or merely segregates with a variant that does. For this reason, there is a growing need for mechanistic studies to evaluate specific genetic variants (83). Variants within protein-coding regions are obvious candidates for further evaluation but comprise only a small fraction of disease-associated SNPs (84). Therefore, it is essential to also evaluate genetic variants in noncoding regions for regulatory function. Our results show that the rs11031006 minor allele has a demonstrable effect on FSHB gene expression, suggesting that genetic variation in the FSHB locus may result in changes in FSHB gene expression and FSH serum levels that impact the reproductive axis. Furthermore, gel-shift and transcriptional assays support SF1 binding as a mechanism to explain how the rs11031006 minor allele may alter FSHB expression.

Evolutionary conservation is a well-established tool for finding novel regulatory regions (85-87). In this study, we demonstrate that evolutionary conservation can also be used as a tool to prioritize genomic variants for functional analysis. Although in this case, the hypothesized causal variant rs11031006 was also the most statistically significant SNP in several PCOS GWAS studies, this is not always true. While fine-mapping of variants within a disease-associated region of linkage disequilibrium can further refine candidate causal variants, these experiments are costly and require a large sample size (88). Prioritizing variants that occur in putative regulatory elements, marked by evolutionary conservation, histone markers, and/or open chromatin, provides a straightforward method to narrow down functional candidates, and can be evaluated using publicly available databases such as phyloP for conservation and ENCODE for indicative histone modifications, when available.

The chromosome 11p14.1 region containing rs11031006 and FSHB is of particular interest for PCOS research as it is also associated with levels of FSH, LH, and a female-specific effect on testosterone levels (28-31, 37, 89). LH levels, also elevated in FSH-deficiency female mouse models and women with FSHB loss-of-function mutations, may be increased due to reduced negative feedback on the HPG axis (19-21, 23, 26, 38). The observed effect of rs11031006 on FSHB transcription could either directly contribute to PCOS etiology through FSH dysregulation, possibly contributing to anovulation, or indirectly through altering LH levels. Increased LH is associated with an array of problems, including elevated testosterone levels (90).

Our results strongly support the conclusion that the enhancer targets FSHB given that the open chromatin status at the enhancer is exclusive to gonadotropes within the pituitary and that FSH is an important regulator of fertility. Additionally, LH levels and the LH/FSH ratio in women show a copy number correlation, with higher serum LH levels and LH/FSH ratio when both copies of the rs11031006 minor allele are present compared with one or no copies (36, 91). Therefore, any contribution of rs11031006 to gonadotropin levels most likely results from changes to FSHB regulation mediated by the enhancer. While FSHB is the most likely target of the enhancer, located within the chromosome 11p14.1 locus 122 Kb downstream of the enhancer is ARL14EP, an effector protein that regulates MHC Class II export and is co-expressed with SF1 in the gonads. Our studies do not rule out that the enhancer could target ARL14EP or another gene outside the 11p14.1 locus in FSHβ-positive cell populations in the pituitary or in non-pituitary tissues.

Given that some women with PCOS exhibit low FSH and that even a single copy of the rs11031006 minor allele was associated with an increase in the LH/FSH ratio (28, 29), we were surprised that the minor allele increased FSHB transcription in vitro in our study. One possibility is that the direction of response was altered by the DNA context; in our study, the enhancer was subcloned directly upstream of the FSHB proximal promoter and studied in a gonadotrope cell culture model. In the in vivo context, it is possible that increased binding of SF1 to the enhancer with the rs11031006 minor allele results in a differential transcriptional response especially since other regulatory elements and factors could interact with the enhancer to alter its effect on the promoter. An alternative possibility is that the rs11031006 minor allele increases FSHB transcription during development, resulting in organizational changes to the HPG axis that contribute to PCOS pathogenesis and potentially, other fertility-related traits. It is intriguing that infants small for their gestational age have been shown to have elevated FSH levels during the “mini-puberty” that occurs in infancy (92) and an increased risk for PCOS in adulthood (93) that may correlate with increased follicle development and anti-Müllerian hormone (AMH) levels, as has been observed in premature infants who also have elevated peak FSH levels (94). In women, elevated AMH has been hypothesized to contribute to hyperandrogenism by inhibiting FSH-stimulated aromatase expression or by directly increasing GnRH neuron excitability (95-98), which favors the synthesis of LH over FSH (99-102). Whether increased FSHB synthesis driven by the minor allele could elevate AMH and promote later development of PCOS is unknown and would need to be tested experimentally.

Finally, further appreciation of the role of the upstream enhancer in the regulation of FSHB gene expression will require the use of mouse models to delineate whether the putative enhancer region is necessary for appropriate regulation of FSH levels and fecundity. Deletion of the enhancer may be sufficient to reduce Fshb levels and mirror the phenotype of other FSH-deficiency models, such as reduced fecundity, elevated LH, and/or reduced number of corpora lutea. In addition, creation of a mouse model with a point mutation to represent the rs11031006 minor allele will help determine whether genetic variation at the Fshb locus is an important factor in reproductive function. Studies in infancy and adulthood may help clarify the discrepancy between the observed increase in Fshb expression due to the minor allele versus the decrease in Fshb expression that would be predicted to be associated with PCOS. As the function of the conserved region is maintained in mouse, the rs11031006 SNP model has the potential to emerge as an innovative tool providing a genetic model to facilitate future fertility research.

Abbreviations

Abbreviations
 
• AMH

  anti-Müllerian hormone

 
• ChIP

  chromatin immunoprecipitation

 
• DMEM

  Dulbecco’s Modified Eagle Medium

 
• FSH

  follicle-stimulating hormone

 
• FSHB

  FSH beta-subunit gene

 
• GnRH

  gonadotrophin-releasing hormone

 
• GSE

  gonadotrope-specific element

 
• GWAS

  genome-wide association study

 
• HPG

  hypothalamic-pituitary-gonadal

 
• LH

  luteinizing hormone

 
• LHX3

  LIM homeobox gene 3

 
• NFY

  Nuclear transcription factor Y

 
• PBS

  phosphate-buffered saline

 
• PCOS

  polycystic ovary syndrome

 
• PCR

  polymerase chain reaction

 
• PTX1

  Paired-like homeodomain 1

 
• SF1

  Steroidogenic factor 1

 
• SNP

  single nucleotide polymorphism

 
• TK

  thymidine kinase

 
• t-SNE

  t-distributed stochastic neighbor embedding

 
• TSS

  transcriptional start site

 
• WT

  wild-type

Acknowledgments

We would like to thank Daniel Bernard and Bon-Chu Chung for kindly providing the human FSHB luciferase plasmid and mouse SF1 expression plasmid, respectively. Members of the Mellon laboratory, especially Shanna N. Lavalle and Jessica B. W. Cassin, kindly provided critical review of this manuscript and valuable discussion. We would also like to thank Mary Jean Sunshine, Ichiko Saotome, Chengxian Victoria Shi, and Qianlan Xu for technical assistance, and Michal Krawczyk for assistance in creating figures.

Financial Support: This work was supported by National Institutes of Health (NIH) Eunice Kennedy Shriver National Institute of Child Health and Human Development (NICHD) Grants R01 HD082567, HD100580, and HD072754 to P.L.M., as well as P50 HD012303 (to P.L.M. and V.G.T.) as part of the National Centers for Translational Research in Reproduction and Infertility. P.L.M. was also partially supported by NIH P30 DK063491, P30 CA023100, and P42 ES010337. S.C.B. was partially supported by NIH F31 HD096838 and NIH T32 NS061847. V.G.T. was partially supported by NIH R01 HD095412. D.S.K. was partially supported by NIH R01 EY027011 and RPB Special Scholar Award. The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health.

Additional Information

Disclosure Summary: Authors have no conflict of interest.

Data Availability

All data generated or analyzed during this study are included in this published article or in the data repositories listed in References.

References