Novel bone morphogenetic protein 15 (BMP15) gene variants implicated in premature ovarian insufficiency

Background Bone morphogenetic protein 15 (BMP15) is expressed in oocytes and plays a crucial role in the reproduction of mono-ovulating species. In humans, BMP15 gene mutations lead to imperfect protein function and premature ovarian insufficiency. Here we investigated the BMP15 gene variants in a population of Iranian women with premature ovarian insufficiency. We conducted predictive bioinformatics analysis to further study the outcomes of BMP15 gene alterations. Methods Twenty-four well-diagnosed premature ovarian insufficiency cases with normal karyotype participated in this study. The entire coding sequence and exon-intron junctions of the BMP15 gene were analyzed by direct sequencing. In-silico analysis was applied using various pipelines integrated into the Ensembl Variant Effect Predictor online tool. The clinical interpretation was performed based on the approved guidelines. Results By gene screening of BMP15, we discovered p.N103K, p.A180T, and p.M184T heterozygous variants in 3 unrelated patients. The p.N103K and p.M184T were not annotated on gnomAD, 1000 Genome and/or dbSNP. These mutations were not identified in 800 Iranians whole-exome sequencing that is recorded on Iranom database. We identified the p.N103K variant in a patient with secondary amenorrhea at the age of 17, elevated FSH and atrophic ovaries. The p.M184T was detected in a sporadic case with atrophic ovaries and very high FSH who developed secondary amenorrhea at the age of 31. Conclusions Here we newly identified p.N103K and p.M184T mutation in the BMP15 gene associated with idiopathic premature ovarian insufficiency. Both mutations have occurred in the prodomain region of protein. Despite prodomain cleavage through dimerization, it is actively involved in the mature protein function. Further studies elucidating the roles of prodomain would lead to a better understanding of the disease pathogenesis.


Background
Premature Ovarian Insufficiency (POI) also called premature ovarian failure (POF) is considered by loss of ovarian activity before the age of 40. It can be distinguished as primary amenorrhea with delayed menarche, secondary amenorrhea and oligomenorrhea over 4 months. POI is diagnosed by an increase in follicle-stimulating hormone (FSH) of higher than 25mIU/ml, repeated twice over 4 weeks [1]. Antimullerian hormone (AMH) is produced by the granulosa cells of growing follicles and is of considerable value in the timely detection of POI. Ovarian reserve can be detected by transvaginal ovarian ultrasound and estimated by AMH levels in POI cases [2]. The prevalence of POI was mainly considered 1 -1.5%. However, a recent meta-analysis reported a global prevalence of 3.7% with a higher rate in medium and low developed countries [3].
In the aetiology of POI, a wide range of explanations have been considered including involvement of infectious, autoimmune, iatrogenic or genetic factors. Among them, genetic aberrations account for approximately 20-25% of cases. FMR1 permutation is the most common genetic abnormality in POI, followed by bone morphogenetic protein 15 (BMP15) gene defects [4]. BMP15 gene is located on Xp11.2 and encodes a 392-amino acid protein of transforming growth factor-β superfamily (TGF-β). The protein is composed of the signal peptide, prodomain and transforming growth factor β-like (TGFβ-like) domain [5]. BMP15 is expressed in oocytes and acts synergically with growth differentiation factor 9 (GDF9) towards the regulation of folliculogenesis [6]. Using a mouse model, it has been shown that BMP15 promote follicular development along with FSH [7]. In vitro study revealed that the simultaneous addition of GDF9 and BMP15 to the culture leads to improvement of the human primordial follicles activation [8].
BMP15 peptides either form homodimers or bind to GDF9 and make heterodimers called cumulin. Dimerization occurs via the TGFβ-like domain. A furin-like protease cleaves prodomain upon protein maturation, however, prodomain remains associated with secreted mature dimer. BMP15 and cumulin control the migration and proliferation of primordial germ cells through separate signal pathways [9,10].
Pathogenic mutations can disrupt the transcription/ translation of BMP15 protein or weaken the interaction with GDF9 [11]. Preliminary studies on knockout models have shown that GDF-9 deficiency led to impaired folliculogenesis and infertility, while a null mutation in the BMP-15 gene caused only reduced fertility in female mice [12]. Subsequent studies suggested differences in BMP15 function between mice and other mammals. A study aiming to identify factors regulating ovulation rate in sheep found that heterozygous carriers of BMP15 mutation had increased ovulation while homozygous were infertile [13]. Additional evidence from the porcine knockdown model confirmed that the role of BMP15 in single-ovulatory species is more pronounced than in multi-ovulatory species [14]. This difference is explained in the balance of GDF9 and BMP15. The dominance of GDF9 signaling leads to multiple ovulations and high fertility, whereas the production of BMP15 modulate primordial germ cell sensitivity to gonadotropins and leads to single ovulation, low fertility and appropriate ovarian reserve [15].
Considering the importance of BMP15 in the pathogenesis of POI, we aimed to investigate the related mutations in a population of affected Iranian women. We further conducted predictive bioinformatics analysis in a comprehensive manner, to evaluate BMP15 gene alterations linked to POI.

Patients
This study was approved by the Ethics Committee of Tarbiat Modares University, Tehran, Iran (IR.MODARES. REC.1399.006). The clinical interpretation was performed based on the European society of human reproduction and embryology (ESHRE) guidelines [2]. All participants delivered written informed consent before participating in the study. Physical examination, blood testing and sonography were performed as part of routine clinical workup by the experienced gynecologists. Twenty four normal-karyotype women with idiopathic secondary amenorrhea before the age of 40 were entered into our study. Based on the genetic counselling, patients were further allocated into sporadic and familial groups. Having at least two serum FSH levels higher than 25mIU/ ml was the main inclusion criteria. The exclusion criteria comprised any of the following: chromosomal aberrations including FMR1 premutation, personal or family history of autoimmune diseases, ovarian surgery, radiotherapy or chemotherapy. To rule out autoimmune disease, related serological tests containing thyroid peroxidase antibody, anti-tissue transglutaminase IgG and thyroglobulin antibody were checked.

DNA extraction and PCR-Sequencing
Genomic DNA was extracted from whole blood samples using the salting-out method. The purity and concentration of the extracted DNA were measured by spectrophotometry. Specific primers were designed for two exons of BMP15 using Oligo explorer and Primer Express software as follows: exon1F;GTT GTG GAG CCA GGA TGC AG, exon1R;AAG CCT GAC AGT AAA CCC ACC, exon2F;AAT TTT AGG GCT GAT TAT AGC, exon2R;TTG GTA CAG GAT TAC TTG CAG.
To perform the PCR reaction, 2X PCR master mix (Amplicon, Pishgam, Iran) containing an ultimate concentration of 1.5 mM MgCl2 was used in a final volume of 50 µl. The PCR program included 5 min initial priming at 95 °C and then 30 cycles with 30 s at 95 °C, 30 s at 59 °C, 30 s at 72 °C with a final extension of 72 °C for 10 min. The PCR products were electrophoresed on 1.5% agarose gel along with the 100 base pairs DNA marker. Bi-directional Sanger sequencing was performed on all samples across two coding exons and flanking intronic sequences. Sanger sequencing results were interpreted using the Chromas software version 2.01.

In silico analysis
The probable pathogenic effect of the amino acid changes was detected by SIFT, PolyPhen, etc., which were integrated into the Ensembl Variant Effect Predictor (VEP) https:// asia. ensem bl. org/ Homo_ sapie ns/ Tools/ VEP? db= core. The prevalence of the variants was checked on the Iranome website http:// www. irano me. ir/. The Iranome browser represents whole-exome sequencing data of 800 healthy individuals from eight Iranian major ethnic groups with the approximate 1:1 Female/Male ratio. All samples were selected from individuals who were > 30 years old to reduce the bias of late-onset Mendelian disorders. The mean age at blood draw was 50.61 ± 9.33, and the age range 30-84 years old in the whole project [16]. Structural modeling of BMP15 wild-type and missense variants were computed by the SWISS-MODEL servers. The protein models were assessed using the Ramachandran plot from the MolProbity program.

Statistical analyses
The collected clinical data were statistically analyzed using SPSS version 24 with a significance level chosen at 5%. Correlation studies of the quantitative and qualitative data were performed by ANOVA and Chi-square tests, respectively.

Clinical status of patients
All of the patients had spontaneously entered menarche and none of the patients was exhibited primary amenorrhea. The mean age at menarche was calculated at 13.12 ± 1.56 years old. The mean age at POI diagnosis was 26.84 ± 9.07 years old ranged from 12 to 39. Median and mean levels of serum FSH were 40.8 and 75.92 mIU/ml, respectively. The analyses of AMH serum levels were available for 19 out of 24 patients and the majority of them were below 0.02ng/ml. The sonographic evaluation revealed that 79.16% of the patients had atrophic ovaries and 20.83% had normal ovarian pattern. Patients' interviews and genetic counselling disclosed that 6 out of 24 cases had a family history of menstrual irregularities, infertility and/or secondary amenorrhea.

BMP15 genotyping
As shown in Fig. 1, by the PCR-sequencing analysis of BMP15, we discovered p.N103K, p.A180T, and p.M184T heterozygous variants in 3 unrelated patients. This is the first report of the link between p.N103K and p.M184T to the POI pathogenesis.
The patient carrier of p.N103K variant entered menarche at the age of 16 and secondary amenorrhea happened when she was at the age of 17. As specified in Table 1, elevated FSH levels and a low normal range of AMH were detected in this patient. The sonographic evaluation exhibited an atrophic ovaries pattern. She was born from a consanguineous background without any indication of POI or related conditions in her family members.
p.M184T was detected in a sporadic POI patient who was prematurely menopaused at the age of 31. High FSH levels of 100mIU/ml and very low levels of AMH were detected by hormonal profiling. Ultrasound examination showed an atrophic pattern in both ovaries. Detailed family history revealed that the parents were not relatives and there was no history of POI and/or related disorders among her relatives.   We identified p.A180T in a 42-years-old patient who was a descendant of consanguineous marriage with a familial history of POI. She had been diagnosed at age of 39 following a history of oligomenorrhea for more than 18 months. The hormonal assessment showed high FSH and very low AMH levels. The sonographic evaluation indicated normal ovarian status (Table 1).
We also detected two known variants of BMP15; c.-9 C > G and p.N103S [17]. Previous studies from Syria and India reported the p.N103S, in both cases and controls [18,19]. This variant is mentioned in the ClinVar database as a benign variant causing ovarian dysgenesis. The annotation of its corresponding allele, c.308 A > G showed an allele frequency of 0.057 and 0.028 in gno-mAD and 1000 Genome, respectively. Iranome browser specified a high c.308 A > G allele frequency of 0.075 in the Iranian population. In our study 1 out of 24 patients (1/48 alleles) was the carrier of this SNP variant.

rs3810682, c.-9 C > G
The 5`-UTR variant, c.-9 C > G, was detected in 29.16% of the patients, all in heterozygous status including 2 compound heterozygous with p.A180T and p.N103S. Statistical analysis was performed to calculate the correlation between c.-9 C > G SNP and clinicopathological parameters of the patients. Table 2 shows a statistically significant association between c.-9 C > G SNP and patients FSH levels (p value = 0.02).

In-silico analysis
Further in-silico analysis was conducted to search the effects of the newly identified p.N103K and p.M184T on protein biogenesis and function. Table 3 lists the insilico predicted values of these mutations in comparison to previously reported nonsynonymous missense alterations of BMP15. The indicated data were extracted from the Ensembl VEP online tool.
The newly identified p.N103K was not annotated on gnomAD, 1000 Genome or dbSNP. Investigation on the Iranome database revealed that p.N103K did not occur in 800 healthy Iranians. The SIFT and PolyPhen results showed scores of 0.35 and 0.006, respectively. In the same way, the p.M184T new mutation was not annotated in databases. Bioinformatics analysis showed scores of 0.13 and 0.003 by the SIFT and PolyPhen prediction tools, respectively. Figure 2 shows the conserved score of the M184 and N103 residues among different species indicating a higher score for M184. Both mutations are located in the secondary structure elements of the prodomain (Fig. 3B), which are essential for regulating mature protein production, secretion, and activity. M184 and N103 are present in β6 and β1` sheet of the peptide structure, respectively [26]. As shown in Fig. 3 A, M184 and A180 are situated in the same β-sheet of the peptide structure. Figure 4 A shows molecular modeling of wild-type M184 and N103 as compared with mutant T184 and K103, respectively. The results confirmed the localization of amino acids on the protein β-sheet structures. Figure 4B provides Ramachandran plots exhibiting Φ/Ψ backbone dihedral angles of the wild-type and mutant proteins. The wild-type showed 88.4% and 4.03% residues in the favored and outliers regions of the plot. The generated model for M184T was found to have fewer residues in the favored region (87.6%). The Rotamer Outliers was also increased for the M184T (0.79%) vs. wild-type (0.16%) which indicates AF Frequency of existing variants in 1000 genomes combined population, gnomAD AF Frequency of existing variants in gnomAD exomes combined population, Iranome AF Allele frequency in Iranome database a decrease in protein stability. The mutant N103K displayed the same configuration as that of the wild-type structure.

Segregation analysis of the p.A180T
As mentioned earlier, p.A180T was found in a patient with a family history of POI. The c.538G˃A missense variant located in exon 2 of BMP15 is presented in databases as the rs104894767. As illustrated in the family pedigree, two older sisters of the proband were also reached menopause before the age of 40, however, their mother entered physiological menopause at the age of 55. By Sanger sequencing, the c.538G˃A variant was analyzed in the female family members. The obtained results indicated a heterozygote alteration in the 64-years-old mother and one of the affected sisters who was menopaused at 35. The other POI sister with menopause at 40 and the 43-years-old healthy sister with regular menses did not carry the p.A180T variant (Fig. 5).

Discussion
In this study, we provided the molecular analysis of the BMP15 gene in a series of unrelated Iranian women with idiopathic POI. Our results revealed two novel variants; p.N103K and p.M184T as well as one known substitution; p.A180T.
With the development of next-generation sequencing, several new genes have been linked to the disease pathogenesis which supports the concept of POI heterogeneity. Investigation of copy number variations led to the identification of microdeletions/microduplications on multiple chromosomal regions. Using high-resolution SNP arrays a 4 MB deletion in Xp11.23-p11.22 was discovered in a patient with secondary amenorrhea at the age of 34. This deletion consists of 65 genes, including the entire BMP15 [28]. A more recent study reported the deletion of both paternal and maternal copies of BMP15 in a 16-yearold patient with primary amenorrhea and FSH levels of 141mIU/ml. The deletion expanded 36 KB and contained the entire BMP15 in the Xp11.22 region. In addition, the study showed intragenic duplication of the tumor protein P63 (TP63) gene in two sisters who were also the carrier of BMP15 p.Y235C [29]. Remarkably, p.Y235C is the first discovered BMP15 mutation that was reported in 2004 in the heterozygous Italian sisters [25]. The same research group further identified p.R68W and p.A180T alterations in non-familial POI patients in 2006 [21]. In 2009 they repeated the BMP15 gene screening in 300 patients and found p.R138H and p.L148P in the prodomain region. They also reported p.S5R mutation which is localized in the signal peptide of the nascent protein. By conducting a functional in vitro assay in the human granulosa cell line, Fig. 4 A Modeled BMP15 amino acid replacement on the three-dimensional structure. B General (no proline or glycine) Ramachandran plots of the BMP15 wild-type, M184T and N103K proteins which were provided by SWISS-MODEL online server. The different colored areas correspond to favored, allowed, generously allowed and sterically disallowed conformations