Mutation analysis of the FOXL2 and BMP15 genes in patients with premature ovarian insufficiency

Premature ovarian insufficiency (POI) is defined as irregular menstrual cycles or amenorrhea for at least four months occurring before the age of 40. This diagnosis is established when two or more follicle-stimulating hormone (FSH) measurements obtained at least four weeks apart demonstrate levels ≥25 mIU/mL⁽¹⁾. POI affects approximately 1% of women and commonly presents with infertility and hypoestrogenic symptoms^{(1, 2)}. In addition to infertility, women with POI are at an increased risk of developing osteoporosis, cardio-cerebrovascular, neurodegenerative, and metabolic diseases⁽³⁾. The pathogenesis is heterogeneous and may involve a reduced primordial follicle pool, accelerated follicular atresia, or impaired follicular growth, with genetic, autoimmune, metabolic, infectious, iatrogenic (e.g., gonadotoxic cancer treatments), and environmental contributors^{(4, 5)}. Genetic factors account for 25-30% of cases of POI and up to ~50% in familial POI, and more than 100 genes have been implicated^{(6, 7)}. In this context, we focused on FOXL2 and BMP15, key regulators of oocyte-granulosa cell communication. FOXL2 is critical for granulosa-cell differentiation and identity, and loss-of-function variants cause blepharophimosis-ptosis-epicanthus inversus syndrome, which may be associated with POI^{(8, 9)}. BMP15 is an oocyte-derived transforming growth factor β superfamily ligand that modulates granulosa-cell proliferation, steroidogenesis, and FSH responsiveness; pathogenic BMP15 variants have been linked to ovarian dysgenesis/POI through a dosage-sensitive mechanism^(10-12). This study aimed to investigate FOXL2 and BMP15 variants in Turkish women with POI and to assess the association of the BMP15 c.-9C>G promoter variant with POI in a case-control setting.

A total of 75 unrelated patients were included in the study: 50 who presented to the Ankara University Faculty of Medicine, Department of Medical Genetics clinic with a preliminary diagnosis of POI and 25 who presented to the Zekai Tahir Burak Women’s Health Education and Research Hospital, Clinic of Medical Genetics. The Ankara University Clinical Research Ethics Committee approved the study (approval number: 05-187-13, date: 25.03.2013), and 75 patients were included after obtaining their informed consent. The inclusion criteria of the study were being under 40 years of age, serum FSH ≥25 mIU/mL on two occasions at least four weeks apart, presence of primary or secondary amenorrhea, normal karyotype: 46,XX, and negative FMR1 CGG repeat testing. The exclusion criteria were a history of ovarian surgery, pelvic chemotherapy, pelvic radiation exposure, or endocrine or autoimmune disease. Eighty women who presented to our clinic and were assessed as having normal ovarian function were recruited as the control group. Controls were ethnically matched women who had regular menstrual cycles, no history of primary or secondary amenorrhea, and no prior diagnosis of POI. Women with a personal history suggestive of POI or amenorrhea were excluded. Genotype distributions of BMP15 c.-9C>G were compared in POI cases and controls using Fisher’s exact test. In addition, exploratory inheritance-model analyses were performed, and effect sizes were reported as odds ratios (ORs) with 95% confidence intervals (CIs). Departure from Hardy-Weinberg equilibrium for BMP15 c.-9C>G was assessed in the control group using an exact test. All statistical analyses were performed using SPSS Statistics, version 31.0.1.0.

Lymphocyte Cell Culture from Peripheral Blood

Peripheral blood lymphocytes were cultured using standard cytogenetic procedures. Heparinized peripheral blood was incubated in culture medium at 37 °C for 72 hours. Colchicine treatment, hypotonic incubation, and fixation were subsequently performed, and metaphase chromosome analysis was carried out on the prepared samples.

Fragment Analysis

Genomic deoxyribonucleic acid was isolated from peripheral blood collected in ethylenediamine tetraacetic acid tubes, using the Magna Pure LC Instrument (Roche Applied Science). Fragment analysis was performed using fluorescently labeled polymerase chain reaction (PCR) primers and capillary electrophoresis with the ROX 1000 size standard. Fragment sizes were analyzed using GeneMapper software (Applied Biosystems).

Sanger Sequencing

FOXL2 and BMP15 were analyzed by bidirectional Sanger sequencing. The single exon of FOXL2 and the two exons of BMP15, including exon-intron junctions, were amplified by PCR using gene-specific primers. Amplification was verified by agarose gel electrophoresis, purified, and then sequenced on a 3130 Genetic Analyzer (Applied Biosystems). Sequence data were analyzed using SeqScape version 2.7 and Sequencing Analysis version 5.1 software. NM_023067.4 and NM_005448.2 were used as the reference transcripts for FOXL2 and BMP15, respectively. Variants were described and classified according to American College of Medical Genetics and Genomics (ACMG) guidelines⁽¹³⁾. Further details of the laboratory protocols are available from the corresponding author upon reasonable request.

The ages at diagnosis ranged from 11 to 39 years, with a mean of 29.93±6.86 years. Of the 75 patients included in the study, 11 (14.67%) were diagnosed with primary amenorrhea and 64 (85.33%) with secondary amenorrhea. Hormonal analysis revealed a mean FSH level of 70.69±29.58 mIU/mL and a mean luteinizing hormone level of 29.76±15.54 mIU/mL. In the patients, chromosomal analysis and FMR1 CGG repeat testing were normal. The clinical and laboratory characteristics of the patients are summarized in Table 1.

FOXL2 sequencing was performed in three fragments in 67 of the 75 patients; in the remaining eight patients, only the first two fragments could be analyzed because of technical difficulties. Variants in the FOXL2 gene were detected in four of the 75 patients. Three patients carried both the c.501C>T (rs61750361) and c.536C>G (rs7432551) variants, and one patient carried the c.672A>T variant. The FOXL2 variants c.501C>T and c.536C>G were classified as benign, whereas the c.672A>T variant was classified as likely benign. BMP15 variants were detected in 36 of the 75 patients. The c.-9C>G variant was homozygous in two patients (cases 6 and 20) and heterozygous in 34 patients. In this heterozygous group, the c.-9C>G (rs3810682) promoter variant was present alone in 21 patients, in combination with c.308A>G (rs41308602) in 12 patients, and in combination with c.352G>A (rs142156356) in one patient. The BMP15 variants c.-9C>G and c.308A>G were classified as benign, whereas the c.352G>A variant was classified as likely benign. Detailed information on FOXL2 and BMP15 variants is presented in Table 2.

A control group of 80 women (ethnically matched women with normal ovarian function) was analyzed to determine the frequency of the BMP15 c.-9C>G variant in the Turkish population. In the control group, 15 women were heterozygous, and 9 were homozygous for this variant. The distribution of BMP15 c.-9C>G genotypes in the patient and control groups is shown in Table 3.

In the control group, the BMP15 c.-9C>G genotype distribution deviated from the Hardy-Weinberg equilibrium (exact p<0.01). Formal haplotype analysis was not performed in our study. In the patient group, the c.308A>G variant was observed only in individuals carrying the c.-9G allele (CG or GG genotypes), which may suggest linkage between these two variants. However, the same pattern was observed in the control group, and the distribution of c.308A>G among c.-9C>G carriers was similar in patients and controls. Therefore, our current data do not support a clear disease-specific effect attributable to a c.-9C>G/c.308A>G haplotype. The observed association appears more likely attributable to c.-9C>G heterozygosity, although this interpretation remains preliminary in the absence of formal haplotype-based analysis.

In the case-control comparison, the genotype distribution for BMP15 c.-9C>G differed significantly between POI cases and controls (2×3 Fisher’s exact test, p=0.0006). The association signal was mainly driven by enrichment of the heterozygous CG genotype in POI cases (34/75, 45.3%) compared with controls (15/80, 18.8%). In exploratory inheritance-model analyses, the overdominant model (CG vs. CC + GG) provided the best fit, and showed that heterozygous carriage was associated with increased risk of POI (OR=3.59, 95% CI 1.74-7.40; Fisher’s exact test, p=0.0005). By contrast, the GG genotype was not significantly associated with case status. The distribution of the c.-9C>G and c.308A>G variants in the patient and control groups is shown in Table 4.

The etiology of POI is the result of the interplay of several genes, and FOXL2 and BMP15 represent functionally important genes that have a role in this process⁽³⁾. In the present study, we analyzed FOXL2 and BMP15 in patients with POI to determine the spectrum and frequency of variants and to assess their pathogenicity.

We did not identify any FOXL2 or BMP15 pathogenic variants in the analyzed regions of our cohort. Because FOXL2 sequencing was incomplete in eight patients, rare variants in the unanalyzed FOXL2 region cannot be fully excluded. Although the BMP15 promoter variant c.-9C>G is classified as benign according to ACMG criteria, its relatively high frequency in our cohort and the case-control comparison suggested a possible association between the CG genotype at position -9 and POI susceptibility. This finding should not be interpreted as evidence of pathogenicity, but rather as a preliminary association signal that requires cautious interpretation and confirmation in larger independent cohorts.

Functional in vitro studies have demonstrated that the BMP15 c.-9C>G promoter variant results in a significant increase in BMP15 expression and has been proposed to contribute to the pathogenesis of POI⁽¹⁴⁾. In the study by Fonseca et al.⁽¹⁴⁾, the BMP15 c.-9G allele was associated with POI. It has been established that BMP15 is co-expressed with the transcription factor PITX1 (pituitary homeobox 1), which binds to the BMP15 promoter between positions -14 and -8. PITX1 can transactivate the BMP15 promoter carrying either the C or the G allele at position -9; however, the c.-9C>G substitution modifies the PITX1-binding site and results in enhanced BMP15 transcription, which has been implicated in POI⁽¹⁴⁾. It has been reported that BMP15 expression increases significantly in the presence of the c.-9G allele, leading to altered granulosa cell proliferation. Moreover, high ovarian BMP15 levels have been shown to further reduce FSHR messenger RNA expression. Consistent with these findings, a transgenic mouse model with high BMP15 expression exhibited an increased number of primary follicles but a reduced number of secondary follicles, and increased granulosa-cell mitosis was associated with accumulation of primary follicles and atresia of secondary follicles⁽¹⁵⁾. Collectively, these data provide biological plausibility for a possible contributory role of the BMP15 c.-9G allele in ovarian dysfunction; however, they do not establish this common variant as a pathogenic cause of POI.

Several studies have shown that the BMP15 c.-9C>G variant is associated with POI^(16-18). In a study of 202 Indian patients with POI, Dixit et al.⁽¹⁶⁾ demonstrated that the BMP15 c.-9G allele forms part of a haplotype associated with the disease. They performed haplotype analysis of three variants that are commonly observed in POI patients (c.-9C>G, c.308A>G, and c.852C>T) and found a significant association between the G-G-C haplotype (c.-9G, c.308G, c.852C) and POI⁽¹⁶⁾. Although c.308A>G was observed only in carriers of the c.-9G allele in our cohort, a similar pattern was present in controls, and formal haplotype analysis was not performed. Therefore, our data do not allow firm conclusions regarding a disease-specific BMP15 haplotype. Morón et al.⁽¹⁷⁾ investigated the association between BMP15 alleles and ovarian hyperstimulation syndrome. The study included 307 women undergoing in vitro fertilization treatment, of whom 35 had a high ovarian response to stimulation. A significant association was found between carriage of the BMP15 c.-9G allele and a high ovarian response to ovarian stimulation in this cohort⁽¹⁷⁾. In another study, Hanevik et al.⁽¹⁸⁾ examined the association between BMP15 variants and the ovarian response to stimulation. They found a significant association between a high ovarian response and carriage of the BMP15 c.-9G allele. Taken together, these studies suggest that the BMP15 c.-9G allele is associated with an increased ovarian response to stimulation, perhaps by increasing the number of primary follicles at the early stages of folliculogenesis⁽¹⁸⁾.

Although functional and association studies support a potential link between the BMP15 promoter variant c.-9C>G and reproductive phenotypes, the published evidence remains inconsistent across populations. Peluso et al.⁽¹⁹⁾ analyzed 186 infertile Brazilian women undergoing their first assisted reproduction cycle and reported higher serum AMH levels among women homozygous for c.-9C>G; however, they found no significant association with ovarian stimulation parameters or assisted reproduction outcomes. Mehdizadeh et al.⁽²⁰⁾ screened BMP15 exon 1 in 70 Iranian women with polycystic ovary syndrome and detected c.-9C>G in 31.4% of patients (28.6% heterozygous and 2.9% homozygous), suggesting a possible contributory role in disease susceptibility rather than a primary causal effect. In contrast, several case-control studies in POI reported no significant association between c.-9C>G and POI risk, including cohorts from Brazil (74 women with POI and 88 controls)⁽²¹⁾ and the Chinese Hui population (63 women with POI and 58 controls)⁽²²⁾; similarly, no significant difference for BMP15 rs3810682 was observed in a Korean recurrent implantation failure cohort (133 patients and 317 controls)⁽²³⁾. In a single-case report of POI coexisting with blepharophimosis-ptosis-epicanthus inversus syndrome, Settas et al.⁽²⁴⁾ identified a de novo FOXL2 mutation alongside BMP15 variants and did not consider c.-9C>G to be causally related to the POI phenotype in that patient.

Study Limitations

An important limitation of this study is incomplete sequencing of FOXL2 in eight patients. Although FOXL2 sequencing was successfully completed for all three fragments in 67 of the 75 patients, only the first two fragments could be analyzed in the remaining eight patients because of technical difficulties. Therefore, rare FOXL2 variants located in the unanalyzed region cannot be excluded from this subset. This limitation is important for the interpretation of our negative FOXL2 findings, which should therefore be interpreted with caution, particularly in these partially analyzed cases.

A significant limitation of this study is the deviation from the Hardy-Weinberg equilibrium observed for the BMP15 c.-9C>G variant in the control group. This pattern, characterized by a deficit of heterozygotes and an excess of homozygotes, may be attributable to sampling variation related to the modest sample size, subtle population stratification, or, less likely, technical/genotyping issues. Although genotyping was performed by bidirectional Sanger sequencing, which reduces the likelihood of systematic misclassification, systematic misclassification, this possibility cannot be completely excluded. Because deviation from Hardy-Weinberg equilibrium may bias estimates of genotype-specific associations, particularly for the CG genotype, the observed association between BMP15 c.-9C>G and POI should be interpreted with caution. Therefore, our findings should be considered preliminary and require validation in larger, independent, well-matched cohorts. Because BMP15 c.-9C>G is classified as a benign variant under ACMG criteria, the observed case-control association should be interpreted not as evidence of pathogenicity but as a hypothesis-generating finding.

Formal haplotype analysis of BMP15 variants was not performed. Therefore, the possible combined contribution of c.-9C>G and c.308A>G to POI susceptibility could not be fully assessed in this cohort.

In our study, no pathogenic variant associated with POI was detected in BMP15 or in the analyzed regions of FOXL2. Because FOXL2 sequencing was incomplete in eight patients, rare variants in the unanalyzed region cannot be entirely excluded. Although the BMP15 c.-9C>G variant was relatively frequent in our Turkish POI cohort and showed a genotype-based association in the case-control comparison, it is classified as benign according to ACMG criteria and should not be interpreted as a pathogenic cause of POI. Rather, it may represent a common variant associated with disease susceptibility in this cohort. These findings should be interpreted cautiously and confirmed in larger, unrelated, well-matched populations, ideally supported by functional and haplotype-based analyses.