ADAMTS1 and HSPG2 mRNA levels in cumulus cells are related to human oocyte quality and controlled ovarian hyperstimulation outcomes

C57BL/6 mice were obtained from the Zhejiang Academy of Medical Science, China. Animal care and experimental procedures were conducted in accordance with the Animal Research Committee guidelines of Zhejiang University.

Female wild-type C57BL/6 mice at postnatal day (PD) 23 were primed with 5.0 IU pregnant mare serum gonadotrophin (PMSG). And 48 h later, mice were injected with human chorionic gonadotropin (hCG) for different treatment times (0, 4, 8, 12, and 16 h). At these timepoints, ovaries were dissected, and follicles were punctured using needles. Then, mural GCs were harvested directly and CCs were obtained by removing the oocytes mechanically from COCs.

Enclosed COCs were collected by puncturing the ovaries from mice with 5 IU PMSG treatment for 48 h. IVM medium (Easy Check, China, M2115) with 2.5 μM milrinone (MCE, USA, HY-14252) was used to inhibit CCs expansion during collection and culture with siRNA. Prior to COCs collection, specific siRNAs (5′-GGAAGTACTGTGAAGGCAA-3′) (RIBOBIO, China, S181130134515) for the mouse ADAMTS1 gene and a negative control at a final concentration of 50 pmol/ml were mixed with Lipofectamine™ 3000 Transfection Reagent (Thermo Fisher Scientific, USA, L3000015) for 30 min and then added into the cultured IVM medium containing 2.5 μM milrinone. The COCs were transferred to the RNAi medium and cultured for 24 h in a 5% CO₂ incubator at 37 °C. After transfection with siRNA or negative control for 24 h, COCs were transferred and allowed to resume meiosis in new fresh IVM medium (Easy Check, China, M2115) without milrinone and then cultured in a 5% CO₂ incubator at 37 °C for 12 h. The CCs were harvested from 10 matured COCs by a mechanical method for RNA isolation.

This study included 148 women undergoing IVF/ICSI because of PCOS (n = 45; ICSI, n = 5; IVF, n = 40) or tubal factor or male factor (n = 103, ICSI, n = 23; IVF, n = 80) at the Reproductive Center of Sir Run Run Shaw Hospital affiliated to Zhejiang University from July 2017 to June 2018. The experiment was certified by the ethics committee of Sir Run Run Shaw Hospital affiliated to Zhejiang University (20190215–2). All patients signed informed consent regarding the collection of CCs.

Of all patients undergoing IVF/ICSI, we selected patients who were between 20 and 35 years old. Our study did not affect clinical treatment options, but we only chose patients who were treated with a long protocol. Briefly, mid-luteal gonadotropin-releasing hormone (GnRH) agonist (Ferring) and ovarian stimulation with a daily subcutaneous dose of hMG (Menogon, Ferring) were started on the third day of the menstrual cycle. When the leading follicles reached 18 mm in diameter, women received hCG (Merck, Serono). Patients with a minimum of seven large follicles ≥ 14 mm at the final ultrasound before oocyte retrieval were invited to attend.

Based on the Rotterdam criteria [39], patients with PCOS and two of the following criteria were included: (1) rare ovulation or anovulation; (2) hyperandrogenism or clinical manifestations of hyperandrogenism (such as hairy, acne); and (3) polycystic ovarian changes. Patients with other causes of hyperandrogenism (such as hyperprolactinemia and thyroid diseases, congenital adrenal cortical hyperplasia, Cushing’s syndrome, androgen-secreting tumors, 21-hydroxylase deficiency atypical adrenal cortical hyperplasia, and exogenous androgen use) were excluded.

Aspiration of the oocytes was performed transvaginally by ultrasound guidance 36 h after hCG administration. Immediately after the isolation of COCs, professional staff cut part of the cumulus mass off. CCs from all 18–20 mm follicles of a patient were pooled. Erythrocytes were removed by adding erythrocyte lysis buffer (EL-buffer, Qiagen, Germany). Then, CCs were washed with 1 ml phosphate-buffered saline (PBS) twice and centrifuged for 10 min at 800 x g to form a pellet. The pellet was collected and stored at − 80 °C until RNA extraction.

Human mural GCs were isolated from the follicular fluid by a pipette, and erythrocyte lysis buffer was added to remove erythrocytes. Similar to CCs, mural GCs were washed with PBS twice and centrifuged for 10 min at 800 x g to form a pellet. The pellet was collected and stored at − 80 °C until RNA extraction. Mural GCs from all 18–20 mm follicles of a patient were pooled.

Total RNA was isolated using the RNeasy Plus Micro Kit (Qiagen, Germany, 74,034) according to the manufacturer’s protocol. Then, RNA concentration was determined by NanoDrop 2000 (Thermo Fisher) and cDNA was obtained from total RNA (50 ng) of each sample by reverse transcription using oligo (dT)₁₅ and reverse transcriptase (Promega, USA, A3500), according to the manufacturer’s instructions, at 42 °C for 15 min, followed by 95 °C for 5 min in 20 μl total volume.

Quantitative real-time PCR was performed using SYBR Green Master Mix (ABI, Germany, DBI-2044) in an Applied Biosystems 7500 Real-Time PCR System. An aliquot (10%) of cDNA was subjected to 40 amplification cycles of PCR with the primers listed in Supplementary Table 1. Cycling parameters were one cycle for 2 min at 50 °C, one cycle for 10 min at 95 °C, 40 cycles at 95 °C for 15 s, and 60 °C for 20 s. For each experiment, three replicates were included in each qPCR reaction. We performed melting curve analysis at the end of each run to ensure a single amplicon.

The relative levels of endogenous β-actin mRNA (ACTB) mRNA were used as an internal control [40], and data were analyzed using the 2^-(△△Ct) method. Only samples whose cycle threshold (Ct) values of ACTB were between 21 and 22 were included in the analysis.

Regarding clinical outcomes, the development conditions of oocytes and embryos were observed and recorded by professional staff. MII oocytes with their corresponding polar bodies and fertilized oocytes with pronuclei were observed at 4–6 h and 16–18 h after insemination, respectively. Good quality embryos and transferable embryos were judged according to a cleavage embryo scoring system [41] at day 3. Professional staff observed the number of cleavage spheres, their symmetry, cytoplasmic morphology, and the number of fragments produced during division. According to these parameters, the embryos were divided into four grades. Grade I–II embryos have equally sized blastomeres with 0%–20% fragmentation. Grade III embryos have unequal sizes and 20% fragmentation. Grade IV embryos have low developmental potential with an abnormal appearance. The grade I–II embryos were recorded as good quality embryos, and grade I–III embryos were recorded as transferable embryos. Only transferable embryos were frozen on day 3, after the assessment of embryo quality. Concurrently, grade IV embryos were discarded. When the patient had reached the condition of embryo transfer, two embryos were thawed and transferred, including endometrial thickness between 8 and 12 mm, E2 > 100 IU/L, and LH <10 IU/L. High-level embryos were transferred as a priority. For implantation, we included only the first transfer from each patient, and all embryos transferred were vitrified-warmed.

The fertilization rate of IVF = the number of zygotes / the total number of retrieved oocytes × 100%.

The fertilization rate of ICSI = the number of zygotes / the number of MII stage oocytes × 100%.

The transplantable embryo rate = the number of transferable embryos (grade I–III) / the number of total embryos × 100%.

The good quality embryo rate = the number of good quality embryos (grade I–II) / the number of total embryos × 100%.

The research team was blinded to experimental data until all clinical data had been collected from all patients. The clinical staff obtained the clinical data, including the number of oocytes retrieved, MII-stage oocytes, good quality embryos, transferable embryos, good quality embryos rate, transferable rate, and implantation outcome.

All analyses were performed using SPSS software version 23.0 (SPSS Inc. Chicago, IL, USA). Tests of statistical significance were two sided and considered significant when P < 0.05. For the relationship between ADAMTS1 or HSPG2 mRNA level and COH outcomes, the Kolmogorov–Smirnov test was used to judge whether parameters showed continuous distribution. If so, unpaired Student’s two-tailed t-tests were applied. If not, non-parametric tests were applied. Analyses using non-parametric tests were marked with # in Tables 1, 2, 3. The mean ± standard error (SE) was used for the descriptive statistics of data. To determine whether ADAMTS1 and HSPG2 mRNA levels were associated with successful implantation, receiver operating characteristic (ROC) analyses were used to determine the predictive potential of ADAMTS1 and HSPG2 mRNA level on implantation. Based on the ROC curve, positive and negative predict values were calculated on the Youden index cut-off point.

We wished to evaluate the expression timing of ADAMTS1 and HSPG2 during ovulation and therefore conducted experiments in the mouse evaluating the expression of these genes following hCG injection. Human oocytes were collected at 36 h after the injection of hCG in the COH cycle, and this timepoint was equivalent to approximately 12 h in mice. The expression levels of ADAMTS1 or HSPG2 were comparable between mural GCs and CCs at hCG 12 h in mice or 36 h in humans (Fig. 1A–D). To evaluate the timing of the expression of ADAMTS1 and HSPG2 during ovulation, mouse ovarian CCs at different timepoints were collected after hCG treatment for RT-PCR. As shown in Fig. 1E, ADAMTS1 mRNA levels were low before 8 h, sharply increased at 12 h, and then decreased to the basal level at 16 h, at which time the level at hCG 12 h was over tenfold higher compared with 0 h. In contrast to ADAMTS1, the mRNA level of HSPG2 in mouse CCs decreased first and then increased, with the lowest expression at 12 h after hCG injection (Fig. 1F). This result indicates that the dynamic expression profiles of ADAMTS1 and HSPG2 mRNA have an inverse relationship.

We wished to determine the effect of ADAMTS1 expression on HSPG2 expression and therefore conduced RNAi experiments knocking down ADAMTS1 expression in the mouse during in vitro maturation, and CCs were collected for RT-PCR. When ADAMTS1 mRNA was depleted in mouse CCs, the relative HSPG2 mRNA level was significantly increased (Fig. 1G). This result indicates the essential role of ADAMTS1 to regulate HSPG2 mRNA levels during ovulation.

To explore the relationship between the expression of ADAMTS1 and HSPG2 and COH outcomes, we compared the mRNA levels of ADAMTS1 and HSPG2 in the control group with tubal or male factor patients (n = 103) and the PCOS group (n = 45). Basal information of the paired groups is shown in Table 1. Patients were between 20 and 35 years old (mean ± SE: 28.796 ± 0.073 years in the PCOS group vs. 29.544 ± 0.034 years in the control group, NS). The body mass index (BMI) of subjects was between 18 and 26 kg/m² in both groups (mean ± SE: 21.379 ± 0.056 kg/m² in the PCOS group vs. 21.776 ± 0.025 kg/m² in the control group, NS). Basal FSH was within the normal range (mean ± SE: 6.120 ± 0.038 IU/L in the PCOS group vs. 6.270 ± 0.194 IU/L in the control group, NS), whereas basal LH level in the PCOS group was nearly 1.6-fold higher than that in the control group (mean ± SE: 8.227 ± 0.096 IU/L vs. 5.015 ± 0.056 IU/L, P = 0.0104). Human anti-Müllerian hormone (AMH) in the PCOS group was over twofold higher compared with the control group (mean ± SE: 10.226 ± 0.105 IU/L vs. 4.818 ± 0.022 IU/L, P < 0.0001). Regarding the hCG dose for triggering ovulation, PCOS patients required a significantly lower dose than control patients (mean ± SE: 6130.95 ± 29.977 IU vs. 6868.69 ± 12.752 IU, P = 0.0009).

To confirm the CCs cDNA samples, we analyzed the mRNA level of ADAMTS1, which was reported to be expressed at a lower level in CCs derived from PCOS patients than control patients [4, 18]. In our study, the PCOS group had lower ADAMTS1 expression in CCs than control patient CCs (0.0081 ± 0.00011 vs. 0.0104 ± 0.00006, P < 0.05) (Fig. 2A). In line with the inverse expression pattern of ADAMTS1 in mice, higher HSPG2 expression was observed in the PCOS group (0.0285 ± 0.00034 vs. 0.0172 ± 0.00012, P < 0.05) (Fig. 2B). Compared with the control group, we obtained more oocytes from the PCOS group (17.739 ± 0.183 vs. 12.223 ± 0.046, P < 0.0001) (Fig. 2C), who displayed significantly lower fertilization rates (64.245 ± 0.359% vs. 85.502 ± 0.147%, P < 0.0001) (Fig. 2D). These results not only further validated the inverse expression pattern in human CCs and implied that CCs derived from PCOS patients had lower ADAMTS1 and higher HSPG2 mRNA levels.

Because a significantly different hCG dose was used between PCOS and control patients, we verified whether hCG dose affected ADAMTS1 and HSPG2 expression. A subanalysis for patients with similar hCG was performed in each group. PCOS patients were divided into two groups according to the hCG dose lower or higher than the median value (hCG ≤ 5000 IU or hCG > 5000 IU). The normal ovulatory group was also divided into two groups according to the hCG dose lower or higher than the median value (hCG ≤ 7000 IU or hCG > 7000 IU). We did not find any difference in ADAMTS1 and HSPG2 expression, retrieved oocytes number, and fertilization rate between PCOS and control patients (Supplementary Tables 2, 3). This suggested the dose of hCG did not affect the expression of ADAMTS1 and HSPG2.

Next, we investigated whether the ADAMTS1 and HSPG2 expression levels were associated with oocyte developmental competence. Thus, normal ovulatory patients were recruited and divided into two groups according to whether the ADAMTS1 or HSPG2 expression levels in their CCs were higher or lower than the median value of all patients (higher, n = 52; lower, n = 51). Basal characteristics (including age, BMI, basal FSH, LH, serum AMH, and hCG dose) of the paired groups divided by the ADAMTS1 or HSPG2 expression levels were comparable, as shown in Tables 2 and 3, which reduced the likelihood of confounding bias.

The retrieved oocyte number and matured (MII) oocyte number in patients with high ADAMTS1 levels were significantly higher than that in patients with a low ADAMTS1 mRNA level (13.490 ± 0.102 vs. 10.981 ± 0.074, P = 0.0072; 13.000 ± 0.101 vs. 10.692 ± 0.073, P = 0.0335) (Table 2). Regarding day 3 embryos, more transferable embryos (7.137 ± 0.077 vs. 5.385 ± 0.059, P = 0.0137) (Table 2) and more good quality embryos (3.863 ± 0.053 vs. 2.462 ± 0.037, P = 0.0097) (Table 2) were obtained from the high ADAMTS1 group compared with the low ADAMTS1 group (Table 2). Although a higher transferable rate and good quality embryo rate were observed in the high ADAMTS1 group compared with the low ADAMTS1 group, this did not reach statistical significance (81.188 ± 0.405 vs. 72.872 ± 0.510, P = 0.0824; 42.469 ± 0.448 vs. 33.360 ± 0.468, P = 0.055) (Table 2). These results suggested that ADAMTS1 levels were linked with oocyte and embryo number and quality.

Regarding HSPG2, although similar oocyte numbers were retrieved from the low and high HSPG2 groups (12.423 ± 0.092 vs. 12.020 ± 0.092, NS) (Table 3), more mature MII oocytes (12.192 ± 0.091 vs. 11.471 ± 0.089, P = 0.0474) (Table 3), more transferable embryos (6.962 ± 0.067 vs. 5.529 ± 2.471, P = 0.045) (Table 3), and more good quality embryos (3.827 ± 0.048 vs. 2.471 ± 0.044, P = 0.0047) (Table 3) were obtained from the low HSPG2 group than from the high HSPG2 group. Moreover, the transferable embryo rate and good quality embryo rate are valuable characteristics to evaluate oocyte developmental competence. The transferable rate and good quality embryo rate were significantly higher (85.217 ± 0.311% vs. 68.600 ± 0.545%, P = 0.004; 45.850 ± 0.444% vs. 29.734 ± 0.436%, P = 0.0005) in patients with lower HSPG2 mRNA levels (Table 3). These results indicated HSPG2 was closely associated with oocyte quality and developmental competence. Overall, the high ADAMTS1 and low HSPG2 expression levels were associated with better oocyte developmental competence.

ROC analysis was performed and the area under the curve (AUC) was used to estimate the accuracy of potential biomarkers. To investigate the predictive effect of the relative mRNA expression of ADAMTS1 and HSPG2 on COH outcome, implantation was evaluated after the IVF/ICSI procedure and embryo transfer.

Implantation was determined by serum hCG levels measured at 14 days after embryo transfer. When we analyzed the correlation of ADAMTS1 or HSPG2 mRNA level with implantation separately, there was no statistical significance (ADAMTS1: AUC = 0.512, P = 0.837; HSPG2: AUC = 0.488, P = 0.837) (Fig. 3A–B). Interestingly, the AUC of ADAMTS1 and HSPG2 to predict pregnancy was 0.738. Further calculation showed that the positive predictive value was 79.3%, and the negative predictive value was 55.6% (Fig. 3C). Therefore, we concluded that the co-measurement of ADAMTS1 and HSPG2 assists in the prediction of pregnancy outcomes.