Implication of novel BMP15 and GDF9 variants in unexpected poor ovarian response

Unexpected poor ovarian response (UPOR) occurs when nine or fewer oocytes are retrieved from a young patient with normal ovarian reserve. Bone morphogenetic protein15 (BMP15) and Growth differentiation factor 9 (GDF9) are two oocyte-specic factors with pivotal role in folliculogenesis. The aim of this study was to assess the relation between BMP15 and GDF9 variants with UPOR. All participants were aged 39 and younger with AMH ≥ 1.27 IU/ml who were divided into UPOR cases and normal ovarian responders (NOR), based on their oocyte number. After genomic DNA extraction, the entire exonic regions of BMP15 and GDF9 were ampli�ed and examined by direct sequencing. Western blotting was performed to determine the expression levels of BMP15 and GDF9 in follicular �uid. Additionally, in-silico analysis was applied to predict the effect of discovered mutations. From four novel variants, silent mutations (c.744T>C) and (c.99G>A) occurred in both groups, whereas missense variants: c.967-968insA and c.296A>G were found exclusively in UPORs. The latter variants caused reduction in protein expression. Moreover, the mutant allele (T) in a GDF9 polymorphism (C447T) found to be more in NOR individuals (58% NOR vs. 37% UPOR (OR=2.3, CI 1.32-4.11, p=0.004). The novel missense mutations which were predicted as damaging, along with other mutations that happened in UPORs might result in ovarian resistance to stimulation. The mutant allele (T) in C447T polymorphism, has a protective effect. Our study proves that BMP15 and GDF9 variants play crucial roles in follicular development and ovarian response, however further investigation is needed for related mechanisms.


Introduction
Poor ovarian response (POR) is de ned as the development of fewer than four follicles following a controlled ovarian hyperstimulation (COH), and it is considered to be aggravated by advanced female age or diminished ovarian reserve [1].As indicated by Sunkara et al, the number of retrieved eggs might be used as an independent predictor of live birth rate; accordingly, adequate number of oocytes is a crucial prerequisite in a successful infertility treatment [2]; as a result, poor responders represent one of the most challenging groups of patients in fertility clinics, and the termination of assisted reproduction is accepted by both physicians and patients as a consequence of unavailable embryos due to insu cient number of eggs [3].
Moreover, the evident diversity in the de nition of POR makes it more di cult to assess the effectiveness of the proposed interventions [3,4].The European Society of Human Reproduction and Embryology (ESHRE) has made an effort to introduce a uniform de nition for POR (the so-called Bologna criteria) [6] which later were critiqued for certain heterogeneity regarding clinical and biological features in the population described [7]; consequently, a shift was proposed to more detailed Patient-Oriented Strategy Encompassing IndividualizeD Oocyte Number (POSEIDON) criteria that "low prognosis" patients are strati ed into four groups not only based on the number of retrieved oocyte but also their age and ovarian reserve biomarkers (Antral Follicle Count (AFC) and Anti-Müllerian Hormone (AMH) or both) [8][9][10].
As mentioned by Humaidan et al., 2016, 'adequate ovarian reserve' parameters refer to AFC ≥ 5 and AMH ≥ 1.2 (ng/ml), and retrieval of more than 9 oocytes (10 to 15 ones) corresponds with the 'optimal' or normal ovarian response (NOR) [9] which seems to elevate the chance of live birth rate in fresh embryo transfer (ET) cycles [2].
Cumulative live birth rates are decreased as a result of the retrieval of nine and less oocytes [11].
Having adequate ovarian test parameters, POSEIDON group1 (< 35 years old) and group 2 (> 35 years old) patients end up yielding an unexpected poor (< 4) or suboptimal (4-9 retrieved oocytes) ovarian response to previous COH cycles, and they are referred to as unexpected poor ovarian responders (UPOR), whereas POSEIDON's group 3 and 4 (the expected low ovarian response) encompass women with diminished ovarian reserve [8] in whom increment of gonadotropin dosage during a COH appears to be of limited success [12].
The exact mechanisms explaining the unexpected hyporesponse (resistance) to gonadotropins stimulation are not entirely clear.While some researchers suggest speci c genotypic factors affecting the gonadotropins and their receptors might be the possible explanations, other investigations have focused on paracrine signals of oocyte growth factors and their pivotal role in pathogenesis of ovarian and follicular conditions [13] .
Growth differentiation factor 9 (GDF9) and its ortholog, bone morphogenetic protein 15 (BMP15, also known as GDF9B) are two members of transforming growth factor β (TGF β) which is a major oocyte-secreted factor (OSF).
GDF9 and BMP15 have been known for their critical roles in the regulation of folliculogenesis via autocrine/paracrine mechanisms [16][17][18][19].BMP15 is an X-linked gene, and its encoded protein is secreted from oocyte from primordial/primary stages.GDF9 gene is located on chromosome 5 (the long arm) and its mRNA is synthesized from early primary up until after ovulation [18].Both GDF9 and BMP15 are synthesized as proprotein, which are further processed into an active mature proteins by furin-like proteases [19].There are high similarities between GDF9 and BMP15 in terms of amino acid, and protein structure.Furthermore, regarding the expression pattern and function in the ovary, the two factors are closely associated [20].Increasing evidence on the synergistic interactions of BMP15 and GDF9 have shown that BMP15/GDF9 heterodimers, known as cumulin, operate to regulate the granulosa cell (GC) and cumulus cell (CC) function, and result in oocyte quality improvement [21], [22] and nally provide support for future healthy embryo and fetal growth [23], [24].Pathogenic mutations regarding these genes can have a disruptive effect on their transcription and translation and lessen their interaction [25].Also, nucleotide sequence alterations are correlated with abnormalities in ovulation rate [26].It has shown that an aberrant function of BMP15 and GDF9 can result in premature ovarian failure (POF) [27], [28].Two separate studies described biallelic mutations in GDF9 in affected with primary amenorrhea [29], [30].
Wu et al found that the worst IVF results among poor responders belong to those patients who are older than 35 years old with low FF BMP15, while younger responders with higher FF BMP15 levels appear to have better results [33].
BMP15 levels in FF are potential predicting factor for oocyte quality and subsequent development of embryo and higher BMP15 levels are associated with the best embryo grading [34].
Considering the importance of BMP15 and GDF9 in folliculogenesis and subsequent embryo quality, this prospective study was aimed to investigate the mutations and subsequent protein expression of these two genes in the Iranian population suffering from unexpected inadequate response to ovarian stimulation.Further bioinformatics analysis was performed to evaluate alterations in both genes.

Subjects
A prospective case-controlled study was designed and approved by the medical ethics committee of the Iran University of Medical Sciences, Tehran, Iran (IR.IUMS.FMD.REC.1400.265).Since women of 40 years of age and older experience a signi cant reduction in the chance of achieving pregnancy and delivery in an IVF cycle, our study population included 39 years old and younger participants.From the 322 patients undergoing ovarian stimulation from January 2019 to July 2021 at IVF center of Afzalipour hospital (Kerman, Iran), 50 patients were assigned to the case group.Women who are regarded as POSEIDON group 1 (34 years of age and younger) and POSEIDON group 2 (aging 35-39 years old) with AMH ≥ 1.27 (ng/ml) who showed unexpected poor and suboptimal ovarian response to COH and were retrieved less than 9 oocytes on the day of egg collection (UPOR).Age-matched participants with similar AMH levels and 10-15 follicles, were recruited in control group (NOR).
A written consent was taken from all the study population.All the participants in this study underwent antagonist cycle and received a daily dose of 150-225 IU recombinant FSH (Cinal F, Cinagen, Iran) from day 3 of a menstrual period.GnRH antagonist (Orgalutran, MSD, America) was given subcutaneously daily (0.25 mg) from the day that the diameter of the leading follicle reached 12 mm to the day of HCG administration.When at least three follicles were measured ≥ 17 mm in diameter, patients received their last GnRH antagonist injection in the morning and nal follicular maturation was induced the same evening by 10000 IU HCG (Pooyesh Darou, Iran).Oocyte retrieval took place 34-36 h after trigger by transvaginal ultrasound-guided double lumen needle aspiration.
The exclusion criteria comprised any of the following conditions: endocrine disorders, polycystic ovarian syndrome, hyperprolactinemia, prior chemotherapy, abnormal karyotype, and other factors affecting ovarian function, such as ovarian surgery and endometrioma.

Evaluation of Oocyte Maturation
Isolated oocytes from follicular uid (FF) samples were assessed by removing the cumulus oophorus and corona radiate from oocytes using both mechanical and chemical methods.Determination of nuclear maturation was established by observing the rst polar body.

Assessment of fertilization and embryo
Fertilization was assessed 16-18 hours post ICSI procedure when two distinct pronuclei (2PN) and two polar bodies were visible.Embryo grading is as follows: A top quality embryo was described as 4-5 blastomeres on day 2 and 7 or more blastomeres on day 3 with less than 20% of fragmentation and complete absence of multinucleated blastomers throughout early cleavage [35].Fertilization rate was calculated as 2PN divided by metaphase II (MII) oocytes.
DNA extraction and PCR protocol 5 ml venous blood was drawn from each participant and stored in ethylene diamine tetra acetic acid (EDTA) containing tubes.Extraction of Genomic DNA from whole blood sample was out using the salting out method.The concentration of the yielded DNA samples were quanti ed using spectrophotometry.Polymerase chain reaction (PCR) was done to amplify the exonic regions of BMP15 and GDF9 genes.The speci c primers for BMP15 and GDF9 were as follows (Table 1):  The annealing temperatures of 60°C and 58°C were used for exon 1 and 2 of BMP15 and 62°C and 62.5°C for two exons of GDF9 were applied.The PCR products were resolved on 2% agarose electrophoresis and the bands were stained and observed using Gel Documentation equipment (Bio Rad).Sanger's dideoxy chain termination method was performed on all samples across two exons of BMP15 and GDF9.Sequence alignment reports were interpreted using online BLAST sequence analyzer and Chromas software.
In-silico analysis By applying straightforward physical and evolutionary comparative considerations, PolyPhen-2 software (Polymorphism Phenotyping v2) predicts the possible effect of amino acid changes on both structure and function aspects of human proteins (http://genetics.bwh.harvard.edu/pph2/)[39].

Analysis of expression levels of BMP15 and GDF9 in follicular uid
Western blot analysis was performed to determine BMP15 and GDF9 expression levels according to the protocol described by Duffy et al. [40] Brie y, protein of FF was loaded and run for electrophoresis on 10% sodium dodecyl sulfate (SDS) gel under reduced conditions.The separated protein bands were transferred to polyvinyl di uoride for 3 hours.The blots were then blocked in Tris-buffered saline solution with 0.1% Tween 20 (TBST) and non-fat powdered milk for one hour.The blots were incubated for 16-18 hours with the diluted rst antibody against β-actin (Santa Cruz Biotechnology, β-Actin (C4): sc-47778, 1: 300), BMP15 (Santa Crus Biotechnology, GDF-9B (F-7): sc-271824) and GDF9 (Santa Crus Biotechnology, GDF-9 (C-6): sc-514933).After being washed with TBST for 3 times, blots were incubated with mouse anti-rabbit IgG-HRP: sc-2357 secondary antibody and m-IgGκBP-HRP: sc-516102.After washing the blots with TBST (0.5%) the bands were detected with ECL-Dura (Thermo Scienti c).

Statistical analysis
The collected data were analyzed statistically using Graphpad Prism 9. Continuous variables were presented as mean ± standard deviation.Differences between UPOR and NOR groups evaluated with the t-test or one-way analysis of variance (ANOVA).Analysis of qualitative variables were done with the Chi-square test.Statistical signi cance was set at < 0.05.

Results
In this study, GDF9 and BMP15 genes were examined to investigate the variants concerning the unexpected poor ovarian response in 50 POSEIDON's group 1 and 2 patients (UPOR) and 50 normal ovarian responder (NOR) control females.PCR and subsequent DNA sequencing were carried out for two exon parts of both genes for all participants.
The clinical characteristics of NOR and UPOR groups are presented in ( The Hardy-Weinberg equilibrium test for genotype frequency of rs3810682 and rs254286, was done via the Chisquared test, and results were as follows: p = 0.51 and Chi-square = 1.32 for BMP15 rs3810682 (-9C > G) and p = 0.44 and Chi-square = 1.63 for GDF9 rs254286 (447C > T).
c.308 A > G and 788-789ins TCT mutations occurred once and c.538 G > A happened twice in UPOR participants and not in normal responders.c.744 T > C occurred once in a UPOR patient in homozygous (CC) form and twice in NOR individuals, though in heterozygous (TC) type.Table 4 outlines fertility treatment details of patients with mutations.
GDF9 SNPs.A sum of four SNPs was found across the GDF9 exonic locus, including two novel mutations (Table 3).Both of the novel mutations happened in the rst exon of GDF9 gene.Change from G to A at position 99 (c.99G > A), made another codon for Glutamine in GDF9 protein (p.Q33Q; Fig. 1G) and had a (0) score by PolyPhen-2.The second novel c.296A > G, made a replacement of Lysine by Arginine (p.K99R; Fig. 1H) and has probably damaging (0.981) impact on ovarian response.Moreover the two previously found variants: rs764784870 (c.309T > A; Fig. 1I) and rs254286 (c.447 C > T; Fig. 1J) were located in the rst and second exon of GDF9, respectively.
The frequency of c.99 G > A genotypes in NOR and UPOR women were not signi cantly different, (22% vs 32% GG, 78% vs 68% GA), however, no individual with AA genotype was found in either of the two aforementioned classi cations.c.296A > G and rs764784870 (c.309T > A) variants were both happened just once and simultaneously in one UPOR patient who had c.538 G > A variation too.The genotype frequency of 447 C > T (CC, CT and TT) in the NOR patients (20%, 44% and 36%) were signi cantly different from UPOR individuals (46%, 34% and 20%, p = 0.01) (Table 5).
Western blot analysis: Two bands at 35 kDa and 16 kDa represent propeptide and mature forms of BMP15, respectively, and a single band at 57 kDa indicated GDF9 (Fig. 2).Reduction in thickness (low intensity) of propeptide and mature bands of BMP15 was detected in number 2, 3, 5 and 8. Band 2 corresponded with a UPOR patient with 308 A > G mutation, and 3 belonged to the same classi cation but a patient with 967-968insA variation.Bands 5 was associated with 538 G > A mutation and number 8 is related to 782-783insTCT variation.Furthermore, similar faint (low) bands were seen in GDF9 expression in number 3, 5 and 8.A description of the bands is given in Table 6.

Discussion
The exact pathogenesis of unexpected insu cient ovarian response to COH remains to be elucidated, yet genotypic variants of BMP15 and GDF9 as two oocyte-speci c factors which belong to TGF-β superfamily with substantial role in ovarian physiology and regulating folliculogenesis could be a possible explanation for that [13].
We have found four novel mutations in BMP15 and GDF9 genes in poor responders with normal ovarian reserve tests.Three female patients (2 NOR and 1 UPOR) presented a novel polymorphism (c.744T > C).Normal responders were heterozygotes (TC), while the UPOR patient was homozygote (CC).Nucleotide substitution of (T→C) resulted in another Phenylalanine codon and was denoted a (0) score by PolyPhen-2 software.Interestingly BMP15 showed thick and sharp bands in those patients follicular uid.It seems that the protein maintain its function in (TC) form, while homozygous (CC) version could end in a defect in the ovulation process which result in 3 oocyte after COH and no subsequent embryo (Table 4).The novel variant c.967-968insA could hamper the reading frame of BMP15 exon 2, also the missense mutation in sequence position 296 in GDF9 both exert damaging effects on ovarian response as predicted by in-silico analysis.The predicted effects were further con rmed by low and reduced bands of associated protein in Western blot analysis (Fig. 1).
It has been proven through this investigation that a polymorphism (C447T) in GDF9 gene is associated with unexpected poor ovarian response to controlled ovarian stimulation for IVF.Since normal homozygote genotype (CC) was associated with the unexpected poor ovarian response (46% UPOR vs. 20% NOR), it can be concluded that the mutant genotype has a protective effect.The negative effects of GDF9 polymorphisms on ovarian reserve have described in previous studies [41], [42].It is shown that this polymorphism can impair follicular growth as well [43].
The genotype frequencies for the − 9C > G in the 5'UTR of BMP15 gene showed no signi cant difference in different groups of the study.In a previous study by Havenik and colleagues [44], a statistically signi cant association was reported between − 9G allele and hyper response to COH, while no signi cant difference was found between this SNP and poor responders.
For a greater success rate in fertility treatment of younger women, more attention should be paid to unexpected poor ovarian responders with normal ovarian reserve (POSEIDON group 1 and 2 women), however the exact reason underlying the mechanisms of UPOR remain to be elucidated.Zhuang et al., discovered an interesting similar and unexpected incidence of POR in younger than 35 years of age women to older ones [45].
As speci ed by [46], increase in FSH concentration is associated with increasing age and nally lead to ovarian dysfunction and inability to conceive.Table 2 showed that women in both groups were showing normal range of FSH according to standard reference value, although UPORs presented signi cantly higher FSH when compared to NOR patients (6.81 ± 1.3 vs 6.26 ± 1.5, P = 0.048), respectively.Moreover in the present study strati cation of unexpected poor responders was solely based on their AMH.Results of Gnoth and coworkers [47] concluded that AMH ≤ 1.26 (ng/ml) is signi cantly associated with poor ovarian response.Based on their ndings, our selection criteria was AMH ≥ 1.27 (ng/ml).However, we found a signi cant difference between AMH from normal and unexpected poor responders (3.14 ± 0.97 NOR versus 2.32 ± 1.00 UPOR, p < 0.0001).
Three previously known variants in the exonic part of BMP15 were found in cases which included two missense mutations, in the rst exonic part of BMP15, Ledig et al., introduced c.538G > A as a polymorphism instead of a mutation, since it was occurred in both POF and control groups.Same result was con rmed by Vanyan et al., [49] study in which they found heterozygotes in their both study groups, revealing no association of G allele with POR, however Di Pasquale and colleagues [50] published evidence which demonstrated association of c.538G > A (p.Ala180Thr) with premature ovarian failure.In our setting, c.538G/A was found only in our case group and associated with unexpected poor response.A study demonstrated (788-789insTCT) polymorphism to be commonly found not only in POF patients, but also the control participants [28].
As opposed to the previous study, this variant discovered to be associated with POR in Chinese Hui patients.Similarly we found (788-789insTCT) in just one of our UPOR individual and as a result it is associated with unexpected poor ovarian response.
BMP15 and GDF9 have fundamental role in primordial follicle activation and take part throughout the folliculogenesis and nally regulate germ cell formation and ovulation [20].Since BMP15 and GDF9 cooperate for oocyte growth, regulation of ovulation process and follicle production, any alterations in BMP15 gene codon can result in a less bioactive and mature protein with aberrant connection with GDF9 protein, resulting in a detrimental effect on ovarian response to induction of ovulation [47].
In conclusion we report a total of four novel polymorphic sites in BMP15 and GDF9 genes sequences, out of which, one insertion in BMP15 gene and a missense mutation in GDF9 gene is plausibly associated with UPOR phenotypes.
A polymorphism of GDF9 with protective effect is associated with unexpected poor ovarian response.Investigations in a larger cohort study is necessary to clarify the underlying mechanisms and their connections with unexpected poor response to ovarian stimulation.

BMP15 Exon 1
rs41308602 (c.308A > G) substitution cause an Asparagine-to-Serine change in 103 protein position.Vanyan and colleagues reported association between this variant with POR in Russian population where in total of 150 patients (65 POR and 85 NOR) only 3 affected with POR had c.308A > G SNP. Furthermore, based on Dixit et al., substituted G allele can be found in both healthy and amenorrhea patients, though compared to normal individuals, G is more frequently associated with G (c.-9C > G) or T (c.852C > T) in amenorrhea patients [48].

Figure 1 Sequence
Figure 1

Table 1
Primer sequences for exon 1 and 2 of BMP15 and GDF9

Table 5
Genotype and allele frequencies of polymorphisms in normal ovarian responders (NOR) and unexpected poor ovarian respodners (UPOR)

Table 6
Description of each band on the gel: band 1(no variation in BMP15, but GDF9 holding two variations (99 GA and 447 TT) and 6 and 7 are from control individuals, band 2,3,4,5 and 8 are affected UPORs (POSEIDON1and 2)