Age-Related Decrease of Expression of GDF9 and BMP15 Genes in Follicle Fluid and Granulosa Cells from Poor Ovarian Responders

Background: Growth differentiation factor 9 ( GDF9 ) and bone morphogenetic protein 15 ( BMP15 ) genes play important roles in folliculogenesis. Altered expression of the GDF9 and BMP15 genes have been found among patients with poor ovarian response (POR). In this prospective cohort study, we determined the expression of the GDF9 and BMP15 genes in follicle fluid (FF) and granulosa cells (GCs) from poor ovarian responders grouped by age, and explored its correlation with the outcome of in vitro fertilization and embryo transfer (IVF-ET) treatment. Methods: A total of 196 patients with POR were enrolled from a tertiary teaching hospital. The patients were diagnosed based on the Bologna criteria and sub-divided into different age groups: group A (< 35 year old), group B (35 ~ 40 year old) and group C (> 40 year old). A GnRH antagonist protocol was conducted for all patients, and FF and GCs were collected after oocyte retrieval. Expression of GDF9 and BMP15 in the FF and GCs was determined with enzyme linked immunosorbent assay (ELISA), quantitative real-time polymerase chain reaction (qRT-PCR), and Western blotting. Results: Compared with group C, groups A and B had significantly more transplantable embryos and higher rates of implantation and clinical pregnancy ( P < 0.05). Group A also had significantly more two pronuclei (2PN) oocytes ( P < 0.05). The expression level of GDF9 and BMP15 in the FF and GCs differed significantly among the three groups ( P < 0.05), showing a trend of decline with age. Conclusion: For poor ovarian responders, the expression of GDF9 and BMP15 is declined with increased age in accompany with poorer oocyte quality and IVF outcome, particularly for those over 40.


Introduction
Poor ovarian response (POR) poses a major challenge for in vitro fertilization and embryo transfer (IVF-ET) treatment. Patients with POR have fewer oocytes retrieval, lower numbers of transferable embryos, lower pregnancy rates, and greater odds for cycle cancellation and miscarriage [1]. The prevalence of POR ranged from 9% to 24% in various IVF clinics [2] and affects approximately 11.9% of Chinese women undergoing IVF [3]. Although various stimulation protocols and strategies have been proposed to improve the outcome of IVF, patients with POR still benefited little from such treatment [4].
According to the Bologna criteria, women with POR are diagnosed with at least two of the three criteria: advanced maternal age (> 40 year old) or any other risk factors for POR; previous history of POR (≤ 3 oocytes with a conventional ovarian stimulation protocol); and an abnormal ovarian reserve test: antral follicle count (AFC) < 5 ~ 7 or anti-Müllerian hormone (AMH) < 0.5 ~ 1.1 ng/mL [5]. Although the Bologna criteria is useful for predicting ovarian response and counseling purpose, it grouped together women with various pathologies and characteristics [6]. Some have reported that, compared with elder patients, young poor ovarian responders demonstrated higher embryo quality and pregnancy rate but lower miscarriage rate [7][8][9]. Therefore, age has been proposed as an independent predictor for POR [7,8]. Studies have shown that elder poor responders, in particular those over 40, achieved a significantly lower cumulative live birth rate (CLBR) compared with younger poor responders (< 35) [10][11][12]. Therefore, 35 and 40 year old are considered as the thresholds by the Poseidon Criteria and Bologna Criteria, respectively [5,6]. Age-related infertility and decreased ovarian reserve (DOR) are the main obstacles for elder poor responders, whereas in younger poor responders, ovarian ageing is independent of age. The heterogeneous prognosis may be connected with various pathogenesis in such patients.
Risk factors for POR include fecundity decline with age, ovarian cystectomy, chronic smoking, genetic factors, previous chemotherapy and/or radiotherapy [13]. The pathophysiology of POR is complex, including the declined endocrine feedback of ovarian factors [14][15][16]. Oocyte-secreted factors (OSFs) can regulate follicle development [17]. And transforming growth factor β (TGFβ) superfamily is one of the most important OSFs which controls the function of granulosa cells (GCs), which in turn can nourish the oocyte [18]. Growth differentiation factor 9 (GDF9) and bone morphogenetic protein 15 (BMP15, also known as GDF9B) are two essential members of the TGF-β superfamily. Both play important roles in follicle recruitment, cumulus expansion, oocyte maturation and ovulation [15,17]. Previous studies have shown that genetic variants and abnormal expression of GDF9 and BMP15 may induce follicle atresia and early exhaustion of ovarian reserve [15,16]. Altered expression of such genes have been found in patients with premature ovarian insufficiency (POI), DOR and infertility [14][15][16]19]. Moreover, for patients undergoing IVF-ET treatment, GDF9 and BMP15 in follicular fluid (FF) and GCs have been associated with poorer oocyte quality and outcome [20]. Hence, they were proposed as biomarkers for the potential of oocyte development [20,21].
Considering the important roles of GDF9 and BMP15 in folliculogenesis, this prospective study was designed to analyze the expression of GDF9 and BMP15 in the FF and GCs from patients with POR from more homogeneous age groups. Identification of the underlying mechanism may facilitate design of individualized treatment protocols for subgroups of patients with POR.

Population
This study was registered with the Chinese Clinical Trial Registry Center (Registration No. ChiCTR1800016107) and approved by the Medical Ethics Committee of Sichuan Provincial Hospital for Woman and Children. All patients had given written informed consent. The study also conformed to the Declaration of Helsinki for Medical Research involving Human Subjects (2013 revision).
Ethnic Han Chinese patients diagnosed with POR according to the Bologna criteria [1] were enrolled from May 2018 to October 2019 and divided into three groups: group A (< 35 year old), group B (35 ~ 40 year old) and group C (> 40 year old). All patients underwent IVF-ET treatment. Patients were excluded from the study should they meet any of the following criteria: (1) Congenital uterine malformation, endometriosis, polycystic ovarian syndrome, intrauterine adhesion, single ovary; (2) Systemic lupus erythematosus and/or sicca syndrome; (3) Uncontrolled endocrinopathy such as diabetes, hyperthyroidism, hypothyroidism, and hyperprolactinemia; (4) Abnormal karyotype; (5) Controlled ovarian stimulation (COS) in the past three months; and (6) intracytoplasmic sperm injection (ICSI) cycle due to male factor infertility. Medical history was taken for all participants, including regularity of menstrual cycle, duration of infertility and pre-treatment protocols. The height and body weight (with shoes and heavy clothing taken off) were measured. Body weight index (BMI) was calculated as weight divided by height squared (kg/m 2 ). When the leading follicle reached 18 mm in diameter, 10 000 IU of urinary human chorionic gonadotrophin (uHCG) (Livzon, China) was injected to trigger the ovulation. Oocytes were retrieved by trans-vaginal ultrasound guidance within approximately 36 h after the trigger, and follicle flushing was not used.

Controlled ovarian stimulation (COS) and IVF procedures
Oocytes were fertilized by conventional IVF for 4 ~ 6 h, mature oocyte was defined as being at the metaphase Ⅱ (MⅡ) stage with the first polar body visible in the cytoplasm. 17 ~ 18 h after the IVF, normal fertilized oocyte was confirmed if it contained two pronuclei (2PN). Cultured embryos were evaluated on day 3 based on the number of blastomeres and degree of fragmentation. Embryos of grade A ~ C on day 3 were defined as transplantable embryos [22]. One or two transplantable embryos were transferred. Luteal phase support was started on the oocyte retrieval day by the injection of 60 mg/d progesterone oil (Zhejiang Xianju Pharmaceutical Co., Ltd. Taizhou, China) or vaginal progesterone (Crinone 8% gel, Merck, Germany). The reasons for canceled cycles included follicular growth failure (10 days after COS, leading follicle diameter < 10 mm), absence of oocyte retrieval at the time of follicle aspiration, no transplantable embryos (no mature oocyte, abnormal fertilization or cleavage), and accumulation of embryos and progesterone > 2.5 ng/mL on the trigger day. Clinical pregnancy was defined as detection of embryonic heartbeat. The rates of implantation, clinical pregnancy, multiple pregnancy, and miscarriage were calculated [8].

Measurement of basal endocrine parameters in serum
Endocrine parameters including estradiol (E 2 ), progesterone (P), total testosterone (TT), prolactin (PRL), FSH and LH were measured with an electrochemiluminescence immunoassay platform (Roche Diagnostics GmbH, Mannheim, Germany). AMH was measured with an enzyme linked immunosorbent assay kit (Guangzhou Kangrun Biotech, Co., Ltd, Guangdong, China). Intra-and inter-assay coefficients for the above variables were set as < 5% and 10%, respectively.

Collection of follicular fluid (FF) and granulosa cells (GCs)
FF samples without blood contamination were carefully collected from follicles with a diameter ≥ 18 mm, and centrifuged immediately at 700× g for 5 min. The supernatant was stored at -80℃. GCs were obtained by follicular aspiration and isolated from blood cells and cellular debris with a lymphocyte separation medium (Beijing Solarbio Science and Technology Corporation, Beijing, China) by centrifugation at 700× g for 10 min. Residual red blood cells were removed with a red blood cell lysis buffer (Solarbio Science and Technology Corporation, Beijing, China). The GCs were stored at -80℃ until the time of use. For each patient, the FF and GCs were collected from all follicles and pooled as one sample.

Determination of GDF9 and BMP15 in FF by ELISA
The concentrations of GDF9 and BMP15 in FF were measured with a commercial enzyme-linked immunosorbent assay kit (Elabscience Biotechnology Co., Ltd., Wuhan, China) by following the manufacturer's instructions. The absorbance value was measured at 450 nm with a Perlong DNM-9602G microplate spectrophotometer (Perlong New Technology Co., Ltd., Beijing, China). The sensitivity of the assay was set as 100 pg/mL.

Determination of mRNA expression in GCs by quantitative real-time polymerase chain reaction (qRT-PCR)
Total RNA was isolated from GCs using a RNAprep Pure Micro Kit (Tiangen Biotech Co., Ltd., Beijing, China). The purity and concentration of RNA were determined with Nanodrop-2000 (Thermo Fisher Scientific, Waltham, Massachusetts, USA) at the absorbance of 260 nm/280 nm. The RNA was reversely transcribed into cDNA using a PrimeScript TM RT Reagent Kit with gDNA Eraser (TaKaRa, Tokyo, Japan). The cDNA was then amplified using TB Green TM Premix Ex Taq TM Ⅱ (TaKaRa, Tokyo, Japan) by qRT-PCR in triplicate. PCR conditions were 95℃ for 30 s, followed by 40 cycles of 95℃ for 10 s, 60℃ for 30 s, and 65℃ for 5 s. Formation of a single product was verified with a melting curve method. GAPDH was used as the internal control. The mRNA levels of the target genes were calculated using the 2 -ΔCT method and expressed as fold change relative to the controls. All PCR reactions were conducted in triplicates. Primers used in the qRT-PCR are shown in Table 1.  Table 1 Sequences of primers used in the qRT-PCR

Statistical analysis
Data were analyzed using SPSS 17.0 software (SPSS Inc., Chicago IL, USA). Continuous variables were expressed as mean ± standard deviation (SD). The Kolmogorov-Smirnov test was used to assess the normality of data distribution. Continuous variables with normal distribution were compared using one-way ANOVA with post hoc Bonferroni test. Categorical data were compared using Chi-squared test. Significance level was set as < 0.05, and two-tailed test was used for all hypothesis tests.

Baseline characteristics of the patients
Baseline characteristics of the patient are shown in Table 2. The average age was significantly different among the three groups (P < 0.05). Compared with group C, the duration of infertility was significantly longer in group A (P < 0.05). The abnormal menstrual cycle rate, BMI, AFC, AMH and basal hormone levels did not significantly differ among the three groups (P > 0.05).  Table 2 Baseline characteristics of the patients

Decline of oocyte quality and IVF outcome with increased age
Thirty-one cycles were canceled. The main reasons have included follicular growth failure (10 cycles), absence of oocyte retrieval at the time of follicle aspiration (6 cycles), no transplantable embryos (5 cycles), accumulation of embryos (5 cycles), and progesterone > 2.5 ng/mL on the trigger day (5 cycles). Compared with group C, groups A and B had significantly more transplantable embryos and higher rates of implantation and clinical pregnancy (P < 0.05). Group A also had significantly more two pronuclei (2PN) oocytes (P < 0.05). The expression level of GDF9 and BMP15 in FF and GCs also differed significantly among the three groups (P < 0.05).
Both showed a trend of decline with increased age. The dosage of rFSH, duration of COS, endometrial thickness and E 2 levels on the trigger day, the number of retrieved oocytes, MⅡ oocytes and embryos per ET, and the rates of miscarriage and multiple pregnancy did not significantly differ among the three groups (P > 0.05) ( Table 3). Note: Data are presented as mean±SD or percentage (number). COS: controlled ovarian stimulation; E2: estradiol; ET: embryos transferred; FF: follicle fluid; GDF9: growth differentiation factor 9; BMP15: bone morphogenetic protein-15. Chi-squared test was u to compare the rates of maturation, fertilization, cleaved embryo, higher quality embryo, cancel cycle, implantation, clinical pregnancy, miscarriage and multiple pregnancy between the two groups. P < 0.05 was considered as statistically significant. a P < 0.05 group A versus group C. b P < 0.05 group B versus group C. c P < 0.05 group A versus group B. Table 3 Controlled ovarian stimulation, IVF outcomes and biochemical markers in FF

Decreased expression of GDF9 and BMP15 in FF with increased age
As shown in Table 3

Decreased expression of GDF9 and BMP15 in GCs with increased age
As shown in Figures 1A and 1B, the expression of GDF9 and BMP15 in GCs, at the levels of both mRNA and protein, have differed significantly between the three groups (P < 0.05). Both showed a trend of decline with increased age. As shown in Figure 1C, the single band at 51 kDa represented GDF9, while the single band at 45 kDa represented BMP15.

Discussion
Through this study, we proved that the oocyte quality and IVF outcome will decrease with increased age among patients with POR, in particular those over 40, which is also correlated with decline of GDF9 and BMP15 expression in both FF and GCs.
Although the oocyte quantity was low in all groups in this study, the numbers of normal fertilized oocytes and transplantable embryos have decreased sharply to 1.37 and 1.22 in poor responders over 40. And the oocyte quality is one of the most critical factors for pregnancy. Similar to previous reports, the rates of implantation and clinical pregnancy in elder patients (> 40) have decreased to 10.14% and 11.54%, respectively [7,8]. This seems to have reflected the poor quantity and quality of the oocytes from elder poor responders whom may benefit less from IVF treatment. For poor responders between 35 and 40, the rates of implantation and clinical pregnancy were 24.39% and 29.82%, respectively, which were both lower than those of under 35 albeit with no statistical significance. Previous studies proposed that 35 ~ 37 year of age to be the threshold for decreased rates of euploidy embryos and live births [23]. Such patients should therefore consider IVF treatment timely. Meanwhile, younger poor responders (< 35) still have reasonable number of transplantable embryos (2.02 ± 0.57) and pregnancy rate (37.5%), and should be treatment actively. Of note, the miscarriage rate may also rise with increased age [8]. The underlying mechanisms for age-related decline of oocyte quality and IVF outcome are variable, including chromosomal aberration in oocytes and embryos due to advanced maternal age [24][25][26]. The autocrine function of oocytes in POR is also important but not yet fully understood.
GDF9 and BMP15 secreted by oocyte play an important role in the cross-talk between cumulus GCs and the oocytes [20]. Together they can activate the transmembrane bone morphogenetic protein receptors (BMPR), Sma-and Mad-related proteins (SMAD), and several signaling pathways in GCs which are involved in folliculogenesis [17,27]. The functions of GDF9 and BMP15 on folliclar development include to suppress GCs apoptosis and promote cell proliferation [28], enhance the effect of FSH and insulin-like growth factor-I (IGF-1) on GCs, provide more E 2 and glycolysis for oocyte [29,30], prevent premature luteinization and promote normal expansion of cumulus cells till the LH surge [31,32]. In turn, the GCs are necessary to nourish the oocytes and promote their maturation [17,33]. Therefore, altered expression and/or function of GDF9 and BMP15 may result in abnormal folliculogenesis and poor oocyte quality.
Reduced expression of the GDF9 and BMP15 genes has previously been discovered among patients with DOR [19,34]. In this study, we further proved that such decrease is associated with increased age, particularly among those over 40. FF and GCs compose the microenvironment of the oocytes, and certain components of the FF and GCs may reflect the metabolism and endocrine status of the oocyte [33,35]. In this study, the decline of GDF9 and BMP15 expression were accompanied with poor oocyte quality and IVF outcomes in elder patients. Compared with group A (< 35), group B (35 ~ 40) had significantly lower level of expression. The oocyte quality and IVF outcome were also poorer albeit with no statistical significance. Hence, GDF9 and BMP15 may provide more sensitive and early biomarkers for the fertility of poor ovarian responders. Declined expression of the two may, at least in part, account for the age-related poor oocyte quality in poor responders. Physiologically, GDF9 and BMP15 can promote the normal expansion of cumulus GCs till the LH surge [31,32]. For elder poor responders, the steep decrease in GDF9 and BMP15 expression is accompanied with relative deficiency of LH. For such patients, co-injection of exogenous LH may be a strategy to improve the outcome of IVF-ET treatment [36].
The limitation of this study lies in its relatively small sample size and lack the live birth rate data due to the limited study period. Moreover, as the main purpose of this study was to evaluate GDF9 and BMP15 expression in female age groups, we have excluded ICSI cycles due to male factors. The latters, in particular advanced paternal age, may also affect the embryo quality and IVF outcomes.

Conclusions
In summary, we have discovered an age-related decline of GDF9 and BMP15 expression among patients with POR, which may in part account for the age-related poor oocyte quality and IVF outcome. Identification of the heterogeneity may facilitate design of individualized protocols to achieve better treatment outcome and avoid repeated cycles. Study of larger cohorts are required to validate the results of this study. Researchers have recently suggested that exogenous recombinant GDF9 and/or BMP15 were able to increase blastocyst formation during IVF treatment [37,38]. Should such strategy be adapted for poor ovarian responders, dosage dependent on age and basal levels of GDF9 and BMP15 expression should be considered.

Availability of data and materials
The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethics approval and consent to participate
Ethics approval was obtained from the Medical Ethics Committee of Sichuan Provincial Hospital for Woman and Children.

Consent for publication
Not applicable.

Figure 1
The gene and protein levels of GDF9 and BMP15 in GCs were decreased with increasing age. A The relatively mRNA levels of GDF9 and BMP15 were significantly different between the three groups (P < 0.05). They were decreased with increasing age. GAPDH was used as internal control. B1 The relatively protein levels of GDF9 and BMP15 were significantly different between the three groups (P < 0.05). They were decreased with increasing age. GAPDH was used as internal control. B2 A single band at 51 kDa represented GDF9, and 45 kDa represented BMP15 by western blot, respectively.

Supplementary Files
This is a list of supplementary files associated with this preprint. Click to download.