Comparative Gene Expression Profiling in Human Cumulus Cells according to Ovarian Gonadotropin Treatments

In in vitro fertilization cycles, both HP-hMG and rFSH gonadotropin treatments are widely used to control human follicle development. The objectives of this study are (i) to characterize and compare gene expression profiles in cumulus cells (CCs) of periovulatory follicles obtained from patients stimulated with HP-hMG or rFSH in a GnRH antagonist cycle and (ii) to examine their relationship with in vitro embryo development, using Human Genome U133 Plus 2.0 microarrays. Genes that were upregulated in HP-hMG-treated CCs are involved in lipid metabolism (GM2A) and cell-to-cell interactions (GJA5). Conversely, genes upregulated in rFSH-treated CCs are implicated in cell assembly and organization (COL1A1 and COL3A1). Interestingly, some genes specific to each gonadotropin treatment (NPY1R and GM2A for HP-hMG; GREM1 and OSBPL6 for rFSH) were associated with day 3 embryo quality and blastocyst grade at day 5, while others (STC2 and PTX3) were related to in vitro embryo quality in both gonadotropin treatments. These genes may prove valuable as biomarkers of in vitro embryo quality.


Introduction
The gonadotropin-releasing hormone (GnRH) antagonist and agonist protocols with either highly purified human menopausal gonadotropin (HP-hMG) or recombinant FSH (rFSH) preparations are the most widely used protocols for controlled ovarian stimulation (COS) for both intracytoplasmic sperm injection (ICSI) and in vitro fertilization (IVF) [1][2][3]. At present, most of the mature oocytes retrieved after COS are capable of fertilization; however, only half of them develop into good embryos and only a few implants. There is increasing evidence that cumulus cells (CCs), which are somatic cells that surround the oocyte, play a crucial role in folliculogenesis and oocyte developmental competence acquisition [4,5]. Several authors propose the use of CC gene expression as a noninvasive approach to predict oocyte aneuploidy, and oocyte competence, as well as embryo and pregnancy outcomes during assisted reproductive technology (ART) procedures [6][7][8][9][10][11][12][13][14][15][16][17]. Despite the recent molecular advances in the knowledge of human CCs, our understanding is far from complete. We believe that the characterization of the biology of these cells following COS might explain observed changes in in vitro embryo development. Several studies have compared the effects of HP-hMG and rFSH on oocyte and embryo quality, follicular fluid biochemical profile, and pregnancy rate [18][19][20][21][22][23]. However, their specific effects on the gene expression profile of individual CC samples have not been investigated. To date, only two such studies have been reported. They compared the gene expression profiles of pooled human granulosa cells (GCs) from periovulatory follicles of six patients in one study and eight patients in the other study. In both studies, the patients were treated with HP-hMG or rFSH in a GnRH agonist long protocol. Significant differences have been observed [24,25]. The aims of the present study were (i) to compare the gene expression profiles of large cohorts of individual CCs isolated from periovulatory follicles of patients stimulated with HP-hMG or rFSH in a GnRH antagonist protocol and (ii) to determine the relationship between in vitro embryo development and expression profiles of CCs isolated from mature oocytes after COS.

Study
Oversight. This research was approved by our Institutional Review Board. All patients provided their written informed consent for the use of CC samples for research.

Sample Collection and Treatment
Cycle. This study is a retrospective analysis of data from of a subgroup of eleven randomly selected patients, who participated in an openlabel, assessor-blind, parallel groups, multicenter trial (Clin-icalTrials.gov Identifier: NCT00884221) that was previously described [26]. CCs ( = 146) were collected from all oocytes retrieved from four patients treated with HP-hMG (Menopur, Ferring Pharmaceuticals) and seven patients treated with rFSH (Follitropin beta, Puregon; MSD) following a GnRH antagonist protocol (Ganirelix Acetate, Orgalutran; MSD), respectively. Stimulation with HP-hMG or rFSH was started at a dose of 150 IU/day (first 5 days of the COS protocol), and the patients' follicular response during stimulation was monitored by transvaginal ultrasound. The GnRH antagonist (daily dose of 0.25 mg) was initiated at day 6 and continued throughout the stimulation period. Transvaginal ultrasound echo guidance, FSH, LH, and estradiol levels were used to monitor the ovarian response. A single injection of 250 g human chorionic gonadotropin (hCG) (choriogonadotropin alfa, Ovitrelle; Merck Serono) was administered to induce the final follicular maturation when three or more follicles ≥17 mm in diameter were observed. Cumulus-oocytecomplexes were collected 36 h after hCG administration (day 0). Supplemental Table SI (see Supplementary Materials available online at http://dx.doi.org/10.1155/2013/354582) shows a summary of the patients' clinical features, end-of-stimulation data, and the number of retrieved oocytes/patients. All CCs were mechanically removed shortly after oocyte retrieval, washed in culture medium, and frozen immediately prior to total RNA extraction. MII oocytes were used for ICSI. All embryos and blastocysts were assessed daily by the embryologists until 5 days after oocyte retrieval. Embryo quality was assessed at 26 ± 2 and 92 ± 2 hours after insemination. On day 5, the quality evaluations of blastocysts consisted of expansion and hatching status, inner cell mass grading (grade A-C), and trophectoderm grading (grade A-C) [26][27][28]. Each CC sample included only CCs from a single oocyte. The number of CCs isolated from oocytes at GV, MI, and MII stages and the in vitro embryo outcome for the two patients' groups (HP-hMG or rFSH) are reported in (Figure 1).

Cumulus Cells RNA Extraction.
The RNeasy Micro kit (ref. 74004, Qiagen) was used to extract total RNA from each CCs sample ( = 146) according to the manufacturers' recommended protocols. The quantity and purity of the total RNAs were determined by using a NanoDrop ND-1000 spectrophotometer (NanoDrop ND-Thermo Fisher Scientific, Wilmington, DE, USA) and their integrity by using the Agilent 2100 Bioanalyzer (Agilent Technologies, Palo Alto, CA, http://www.agilent.com/). All RNA samples were stored at −80 ∘ C until the microarray experiments.

Preparation of cRNA and Microarray Hybridization.
Total RNA (50 ng) was used to prepare cRNA (one cycle of amplification) using the Affymetrix 3 IVT express protocol. An oligo-dT primer with a T7 promoter sequence was used to synthesize the first-strand cDNA. After generating the second strand, the complete cDNA was amplified by in vitro transcription (linear amplification) with a T7 RNA polymerase. The amplified RNA (aRNA) was generated and quantified by using a NanoDrop ND-1000 spectrophotometer (NanoDrop ND-Thermo Fisher Scientific, Wilmington, DE, USA), and biotinylated nucleotide analog was incorporated during in vitro transcription step. RNA from the GeneChip Eukaryotic Poly-A RNA Control Kit (Affymetrix, Santa Clara, CA), which contains mRNAs from Bacillus subtilis genes (lys, phe, thr, and dap), was amplified and labeled under the same conditions as positive controls. After fragmentation, the labeled antisense aRNA (15 g) was hybridized to HG-U133 Plus 2.0 GeneChip pan-genomic oligonucleotide arrays (Affymetrix) containing 54,675 sets of oligonucleotide probes (probeset) which correspond to ≈25,000 unique human genes or predicted genes. Each cumulus cell sample was put individually on a microarray chip. Microarray experiments were performed in DNA microarray platform of our Institute of Research in Biotherapy at the Montpellier University Hospital.

Data Processing and Gene Expression Profile Analysis.
After image processing with the Affymetrix GeneChip Operating 1.4 software (GCOS), the CEL files were analyzed using the Affymetrix Expression Console Software v1.3.1 and normalized with the MAS5.0 algorithm by scaling each array to a target value (TGT) of 100 using the global scaling method to obtain an intensity value signal for each probe set. This algorithm also determines whether a gene is expressed with a defined confidence level or not ("detection call"). This "call" can either be "present" (when the perfect match probes are significantly more hybridized than the mismatch probes, < 0.04), "marginal" (for values of >0.04 and <0.06) or "absent" ( > 0.06). Gene annotation was performed using NetAffx (http://www.affymetrix.com/, March 2009). A first selection of microarray data was based on the detection call (present in at least 50% of the CC samples of each group). Then, the Significant Analysis of Microarrays (SAM) (http://www-stat.stanford.edu/∼tibs/SAM/) with the Wilcoxon test and sample label permutation ( = 300) was used to identify genes of which expression varied significantly between the HP-hMG and rFSH CC samples. The lists of significant genes (fold change, FC ≥1.5 and false discovery rate, FDR ≤5%) as well as common genes were analyzed using the Ingenuity Pathway Analysis (IPA) software (http://www.ingenuity.com/) to identify the biological functions that were specific of each CC group and in common between the two treatments, respectively. Only annotations with significant value ( < 0.05) were considered. Then, the SAM analysis (FC ≥1.5, FDR ≤5%) was used to link gonadotropin-specific genes in CCs or those that are irrespective of gonadotropin treatment to subsequent embryo outcome at day 3 (top, good embryo versus poor) or day 5 (good blastocyst versus bad). Hierarchical clustering analyses based on the expression levels of the differentially expressed genes were performed by using the Cluster and Treeview software packages [29]. Box-and-whisker plots depicted the comparisons of the expression levels of candidate genes carried out using SPSS 12.0 (SPSS, Chicago, IL, USA) software.

Microarray Data Validation by Quantitative RT-PCR.
Quantitative RT-PCR was performed to validate the expression of selected genes identified as differentially expressed between the two CC groups by using mRNAs from HP-hMG ( = 4) and rFSH ( = 4) CC samples as described in [30]. The primer sequences are shown in (Supplementary data,

Common Transcriptional Gene Profile in HP-hMG/rFSH
CCs. In view of few differences between the two gonadotropin treatments, we examined the list of genes in common to HP-hMG and rFSH groups (list of 9,805 genes; see Supplementary data, Table SIII). We used IPA software to explore the specific functional properties of this common molecular signature. Estrogen receptor signaling (83 genes) ( value = 8.17 − 08) was one of the top canonical pathways related to this molecular signature. On the other hand, the HP-hMG signature rFSH signature HP-hMG CCs rFSH CCs top network involving 35 genes was articulated around the "cell death and survival, DNA replication, recombination, and repair" functions. The detailed list of genes involved in this network can be found in (Supplementary data, Table  SIV). Interestingly, the most common HP-hMG/rFSH genes were associated with multiple signaling pathways including FGF signaling (FGFR and GRB2), IGF signaling (IGF1R and IGFBP3), EGF signaling (EGFR and MAPK1), and PDGF signaling (PDGFRA and PDGFD). It is important to note that no difference was observed in the mRNA CC level between treatments for receptors (LHCGR and BMPR2), aromatase (CYP19A1), cytochrome P450 (CYP11A1), or steroidogenic genes (StAR, HSD3B2, ACVR1, ACVR1B, INHBC, and INHBB).

Relationship between the HP-hMG or rFSH CC Expression Profiles and In Vitro Embryo Development.
Of the 146 CC samples, 101 were isolated from MII mature oocytes which underwent ICSI. In the HP-hMG group, 77% of injected oocytes were fertilized and 61% achieved blastocyst stage at day 5. In the rFSH group, these values were, respectively, 86% and 52%. Fertilized MII oocytes ( = 23 in the HP-hMG and = 61 in the rFSH group) were divided into oocytes that developed into (i) top/good quality (52% in the HP-hMG and 70% in the rFSH group, no significant difference ( = 1.65)) or poor quality embryos at day 3; and then into (ii) good (AA and AB) (43% for the HP-hMG and 29% for the rFSH group, no significant difference ( = 1.28)) or bad grade (AC, BC, CC, and CB) blastocysts at day 5 ( Figure 1). Then, the transcription profile of the cumulus cell samples isolated from these 101 MII oocytes was evaluated relative to day 3 embryo quality and blastocyst    (Figure 3(a)). After rFSH treatment, upregulation of GREM1 (x1.59, FDR = 0) and PSPH (x1.6, FDR = 0) was significantly associated with top/good quality day 3 embryos; OSBPL6 (x1.59, FDR = 0) upregulation was found in CCs from oocytes that developed into good blastocyst at day 5 ( Figure 3(b)). In the two gonadotropin groups, PTX3 (x-1.81, FDR = 0) downregulation and STC2 (x1.76, FDR = 0) upregulation were observed in CCs isolated from MII oocytes that developed into top/good day 3 embryos, whereas TRIM65 (x-1.62, FDR = 0) and GSTM2 (x-1.67, FDR = 0) expressions were downregulated in CCs associated with good blastocyst grading (Figure 3(c)).

CC mRNA Content and In Vitro Blastocyst Outcome at Day 5.
Independently of the type of gonadotropin treatment used, the relation between amplified mRNA content of CC samples and in vitro blastocyst development at day 5 was also investigated. Seventeen CC samples, isolated from MII oocytes that developed into top quality 8-cell embryos at day 3, were selected and divided in three groups: (i) CCs from MII oocytes that developed into good quality (grade AA-AB, = 7), (ii) intermediary (grade BB, = 6), and (iii) bad (grade CC and others, = 4) blastocysts. The amount (mean ± SEM) of amplified mRNA from CCs from MII oocytes leading to good quality blastocysts was 1044.28 ± 159.18 ng/ L. This value decreased to 796.66 ± 150 ng/ L in the intermediary group and to 627.50 ± 76.25 ng/ L in the bad blastocyst grade group (Figure 4).

Discussion
Following global genomic assessment of 146 human CCs transcriptome under HP-hMG and rFSH treatments, the present study revealed a small but significant distinct molecular signature of 94 genes between the two treatments, suggesting that these treatments impact differentially the CC gene expression profile. This may be accounted for by the differences in the origin of the two pharmaceutical preparations. More precisely, overexpression of genes involved in the metabolism of lipids such as GM2A, AKR1C1 and AKR1C2, as well as genes related to the intercellular signaling (GJA5 and FOS) was observed in the CCs treated with HP-hMG, while genes involved in "cellular assembly and organization" (COL1A1, COL3A1, MT3, TNFAIP6, and TNFAIP8) were overexpressed in the rFSH CCs. Each of these functions plays a central role in oocyte maturation and/or oocyte competence [31][32][33]. Indeed, the metabolism of lipids represents the main energy source for protein synthesis during oocyte nuclear maturation and early embryo development [34,35]. Simultaneously, adequate communication between oocyte and CCs and appropriate assembly and organization of the CC matrix are required for both oocyte maturation and competence [36][37][38]. Most of the genes, identified in the present investigation as differentially expressed in CCs treated with HP-hMG and rFSH, were reported for the first time, except for TNFAIP6 and GJA5 (connexin 40) which have been previously identified as potential markers of oocyte competence in CCs from bovine preovulatory follicles [39] and biomarker of oocyte maturation in canine cumulusoocyte complexes matured in vitro, respectively [38]. Furthermore, the comparison of our data with the two other transcriptomic studies comparing the same gonadotropin treatment in granulosa cells (GCs) using the GnRH agonist long protocols [24,25] indicates that GM2 ganglioside activator is upregulated in HP-hMG CCs (this study) and rFSH GCs [24]. GM2A is known to play an important role in the hydrolysis of phospholipids or small glycolipids [40]. In addition, among the 9 common genes of our study and the one by Brannian et al. [25], six genes (ATP7A, BTRC, LRRN3, STRN3, PTER, and SUPT3) are upregulated in both CCs and GCs after rFSH treatment; one (H19) was upregulated in both rFSH CCs and HP-hMG GCs and the two others (SERPINI1 and SSFA2) in HP-hMG CCs and rFSH. The use of different GnRH analogs might explain these discrepancies, but we cannot exclude the possibility that gonadotropin stimulation might have different effects on CCs and GCs. More investigations are required to address this issue.
On the other hand, we reported an important common CC molecular signature revealing the preservation of numerous growth factor signaling between the two types of treatments including the IGF, PDGF, FGF, and EGF pathways (See Figure SIII). These signaling pathways have been previously reported to play a central role in the control of the intrafollicular androgen/estrogen ratio for the IGF members [41], in angiogenesis and embryo development for the FGF and PDGF members [42] and in oocyte maturation for the members of the EGF family [43][44][45]. The interactions between these signaling pathways in CCs under COS will be a precious itinerary to explore in future works in order to complete the oocyte competence puzzle.
Another important finding of this study is that the mRNA level for key genes involved in ovulation process including hormonal receptors (LHCGR and BMPR2) and regulators of steroidogenesis (StAR, HSD3B2, Activins, and Inhibins) was comparable in the HP-hMG and rFSH CC groups. This suggests a similar potency of the two protocols to induce hormonal receptors and similar estrogenic capacity of the CC samples stimulated by HP-hMG and rFSH. This is in line with several studies reporting that CCs in vitro were able to secrete estradiol during COCs culture from patients undergoing stimulated cycles, probably as a consequence of the action of gonadotropins [46].
We also identified a significant relationship between some CC genes that were specifically upregulated following stimulation with HP-hMG or rFSH and in vitro embryo development. In the HP-hMG group, upregulation of NPY1R and NPY2R in CCs was associated with top/good embryo quality at day 3. NPY modulates steroid production through NPY receptors [47] and plays a role in human ovarian steroidogenesis directly at the level of the granulosa cells of the follicles in the early stage of luteinization [48,49]. Additionally, the association of ubiquitin specific protease 45 (USP45) with good blastocyst quality suggests the requirement of proteasomal activity in HP-hMG-treated CCs. Proteasomal activity has been reported to have multiple functions in CCs expansion, in oocyte meiosis, and in the modification of cumulus-oocyte communication [50].
In the rFSH group, upregulation of gremlin 1 (GREM1) in CCs was associated with top/good embryo quality at day 3 and OSBPL6 upregulation with good blastocyst grading at day 5. Only CC expression of GREM1, a member of the bone morphogenic protein (BMP) antagonist family, has been reported as positively correlated with embryo quality [7,12,51]. The regulation of BMP through GREM1 is thought to contribute to CCs expansion and therefore to the final maturation of oocytes [52]. The gene OSBPL6 codes for the oxysterol binding protein-like-6 receptor. Oxysterols, which bind to this receptor, are potent modulators of expression of cholesterol synthesis in human granulosa cells [53]. Recently, Watanabe et al. [54] reported that variation in cholesterol contents in cumulus-oocyte complexes during in vitro maturation of porcine oocytes affected their ability to be fertilized, suggesting that, under rFSH regime, cholesterogenesis at a nearby site of oocyte growth and maturation might also be involved in in vitro blastocyst outcome. On the other hand, we also identified CC genes associated with day 3 embryo quality and blastocyst grading at day 5, independently of the type of gonadotropins. Among these genes, we report for the first time the expression of STC2, GSTM2, and TRIM65, as well as PTX3 which has been shown in previous studies to either be associated with fertilization rate [55] or to have no relationship with high-quality embryo on day 3 [51]. A possible reason for higher stanniocalcin 2 (STC2) expression in the CCs isolated from MII oocytes that developed into top/good day 3 is the modulation of the angiogenic [56] or steroidogentic pathways [57] or principal processes in ovarian function [58][59][60]. Conversely, we observed an increased expression of GSTM2 and TRIM65 in CCs from oocytes that developed into bad blastocyst grading. GSTM2 and TRIM65 play a role in the protection against lipid peroxidation [61] and in DNA repair [62] respectively, suggesting an increase in cellular resistance against oxidative stress and damaged DNA. The implications of these genes, at the CC level, deserve to be addressed in future studies in order to understand their function in follicular growth.
Furthermore, independently of the type of gonadotropin treatment, we found an association between blastocyst grading at day 5 and the amount of amplified mRNA in CC samples from MII mature oocytes with comparable top/good embryo quality at day 3. Lower mRNA values were detected in CCs from MII oocytes that developed into bad blastocysts as compared to CC samples from oocytes that developed into intermediary or good quality blastocysts at day 5. This suggests that CCs surrounding an incompetent oocyte are less transcriptionally active.
These results are in line with our previously published data showing a general reduction in transcriptomic activity of CCs associated with poor oocyte competence and negative clinical outcome [6].

Conclusion
Analysis of the microarray data of CCs from patients, who underwent GnRH-antagonist COS, highlights a significant difference in the gene expression profile of CCs following treatment with HP-hMG or rFSH. Components of signaling pathways (the EGF, IGF, FGF, and PDGF cascades) were conserved in CCs under the two gonadotropin stimulation regimens. Some genes specific to each gonadotropin treatment or commonly expressed in both groups were associated with in vitro embryo development. Moreover, independently of the gonadotropin preparation used, the amount of amplified mRNA in each CC was associated with blastocyst grading at day 5. These genes may prove valuable as biomarkers of in vitro embryo quality and can be useful for understanding the biology of stimulation.