Transcriptomic Analysis and Meta-Analysis of Human Granulosa and Cumulus Cells

Specific gene expression in oocytes and its surrounding cumulus (CC) and granulosa (GC) cells is needed for successful folliculogenesis and oocyte maturation. The aim of the present study was to compare genome-wide gene expression and biological functions of human GC and CC. Individual GC and CC were derived from 37 women undergoing IVF procedures. Gene expression analysis was performed using microarrays, followed by a meta-analysis. Results were validated using quantitative real-time PCR. There were 6029 differentially expressed genes (q < 10−4); of which 650 genes had a log2 FC ≥ 2. After the meta-analysis there were 3156 genes differentially expressed. Among these there were genes that have previously not been reported in human somatic follicular cells, like prokineticin 2 (PROK2), higher expressed in GC, and pregnancy up-regulated nonubiquitous CaM kinase (PNCK), higher expressed in CC. Pathways like inflammatory response and angiogenesis were enriched in GC, whereas in CC, cell differentiation and multicellular organismal development were among enriched pathways. In conclusion, transcriptomes of GC and CC as well as biological functions, are distinctive for each cell subpopulation. By describing novel genes like PROK2 and PNCK, expressed in GC and CC, we upgraded the existing data on human follicular biology.


Introduction
Precise and timely coordination between oocyte maturation and the process of somatic cell proliferation and differentiation is crucial for the proper development of the ovarian follicle [1]. Granulosa cells (GC) differentiate into two functionally distinct groups during ovarian follicle growth: cumulus cells (CC) that are in physical contact with the oocyte, and mural GC, which line the wall of the antral follicle [2]. Bidirectional communication between the oocyte and its surrounding somatic cells through paracrine signalling and gap junctions is required during folliculogenesis for the development of follicular compartments, oocyte maturation and competence acquisition [3].
In order to secure its own optimal growth and maturation [4], the oocyte regulates the differentiation, proliferation, apoptosis and luteinisation of GC and CC by secretion of paracrine factors such as growth differentiation factor 9 (GDF-9) and bone morphogenetic protein 15 (BMP-15) [5]. GDF-9 and BMP-15 activate signalling pathways and induce GC gene expression which leads to a functional change of these cells necessary for oocyte development [6].
On the other hand, somatic follicular cells provide support for oocyte growth, resumption of meiosis, nuclear and cytoplasmic maturation and control oocyte's transcriptional activity through the activation of different signalling pathways [7]. For example, in mice, GC derived Kit ligand (KitL) has been shown to stimulate oocyte growth via activation of phosphatidylinositol 3-kinase (PI3K) pathway [8]. Furthermore, binding of KitL to its receptor Kit on the oocyte activates Ras, Raf, mitogen activated protein kinase (MAPK) and protein kinase B (PKB/Akt)-apoptosis related signalling pathways [9].
Identification of key molecules and signalling pathways within GC and CC is necessary to gain insight of how the oocyte acquires developmental competence during its growth within the follicle and for uncovering molecular regulators of this process. This could help optimize in vitro techniques for oocyte maturation. There have been some studies conducted on human GC and CC in order to provide information on cell-type specific transcriptome [10,11,12]. These studies identified specific enriched biological processes in GC (cell cycle, immune response, cell proliferation, steroidogenesis) and CC (cell signalling, protein degradation, angiogenesis). However, our knowledge of the key genes and pathways of human folliculogenesis and oocyte competence acquisition is still incomplete. Identification of specific genes and functions of human GC and CC would help in understanding the functional specifics of each follicular compartment and its role in the process of human folliculogenesis.
The aim of the present study was to identify differentially expressed genes and analyse biological processes in human GC and CC obtained during IVF procedures.

Patient population and IVF procedure
Thirty-seven (37) IVF patients were included in this study. All patients were less than 35 years old, had a body mass index (BMI) between 17 and 26 kg/m 2 , were enrolled in the IVF procedure due to tubal cause of infertility and had their first or second IVF procedure. Their partner's spermiogram was normal according to the WHO criteria [13]. The study was approved by the National Medical Ethics Committee of the Republic of Slovenia and written informed consent was obtained from all women prior to participation.
All patients were treated with recombinant FSH (Puregon; Schering Plough, New Jersey, USA) and GnRH antagonist cetrorelix acetate (Cetrotide; Asta Medica AG, Frankfurt, Germany). Vaginal ultrasound examination was performed to monitor follicular development. Final follicular maturation was induced by administering 10,000 IU of human chorionic gonadotropin (hCG) (Pregnyl; N.V. Organon, Oss, the Netherlands) when at least three follicles were 17 mm. Ultrasound-guided transvaginal puncture was performed 34-36 h later. Each follicle was aspirated separately. After aspiration, cumulus-oocyte complexes (COC) were removed from the follicular fluid. Cumulus cells of each oocyte were removed immediately. Classical IVF was used for oocyte fertilization. Embryo transfer was performed on day 5 after oocyte retrieval. The GC and CC samples in this study were all derived from follicles containing mature, MII, oocytes. real-time PCR (qPCR) validation performed on 19 individual GC and 19 individual CC samples derived from 16 women that were newly included in the study. Data from two previous studies, where comparison of human GC and CC gene expression profiles was performed, were used for the meta-analysis for the identification of consistent differentially expressed genes.

Samples collection and preparation
Immediately after oocyte retrieval, CC of each oocyte were mechanically removed by a needle and a glass denudation pipette (Swemed, Göteborg, Sweden), washed in phosphate-buffered saline (PBS), snap frozen in liquid nitrogen and stored at -80°C until RNA isolation.
For GC purification, Dynabeads CD45 Magnetic Beads (Invitrogen Dynal AS, Oslo, Norway) were used for the depletion of CD45+ cells according to the manufacturer's protocol with some adjustments. The beads were used to separate leukocyte CD45+ cells from GC to exclude the risk of contamination [14]. Follicular fluids were centrifuged at 800x g at 4°C for 10 minutes. Follicular fluids were then snap frozen in liquid nitrogen and stored at -80°C. The sediment was resuspended in 2 ml of lysis solution (0.15 M NH4Cl) to remove red blood cells, incubated for 10 minutes and centrifuged at 600x g at 4°C for 10 minutes. The supernatant was discarded and the sediment resuspended in 1 ml PBS with 0.1% bovine serum albumin (BSA) and 0.002 M ethylenediaminetetraacetic acid (EDTA). The suspension was added to the tubes containing 50 μl of Dynabeads CD45 and incubated at 4°C for 30 minutes with gentle tilting and rotation of the mixer. The tubes were pre-coated with 2 ml of Roswell Park Memorial Institute medium (RPMI) with 1% fetal calf serum (FCS) for 5 minutes. After incubation, the samples were diluted with PBS with 0.1% BSA and 0.002 M EDTA to 4 ml and placed in a magnet for 10 minutes. The supernatant was transferred to a new pre-coated tube and centrifuged at 800x g at 4°C for 15 minutes. The supernatant was discarded and sediment resuspended in 1 ml of PBS. The samples were then centrifuged at 2000 x g for 5 minutes, supernatant discharged and sediment snap frozen in liquid nitrogen and stored at -80°C. The purity of the samples was tested with flow cytometry. The samples were prepared from a GC suspension before isolation and compared to cell suspension after CD45+ depletion. The contamination with leucocytes was investigated using antibodies CD45 conjugated with Fluorescin Isothiocyanate (FITC) and CD14 Phycoerythrin (PE) (both from BD Bioscience, San Jose, USA). Samples were washed and resuspended in PBS with 1% BSA and incubated with antibodies for 20 minutes. The purity of the samples was characterized by Forward Scatter (FSC) versus Side Scatter (SSC) followed by gating on CD45 and CD14. Data were collected on FACSCanto flow cytometer (BD Bioscience, San Jose, USA) and expression of various markers was assessed using FlowJo analysis software (TreeStarInc, Ashland, USA).
Before depletion of CD45+ cells there were, in average, 14.8% of granulocytes, 9.7% of lymphocytes, 9.7% of monocytes and 66.9% of GC present. After the depletion, the proportions of leukocytes were significantly lower. In average, there were 4.7% of granulocytes, 1.8% of monocytes, 1.7% of lymphocytes and 91.8% of GC. Thus, we have achieved a sufficient purity of GC samples, as it has been established that a gene expression profile of a more than 75% pure sample was indistinguishable from a 100% pure sample [15].

RNA extraction and cDNA synthesis
TRI reagent (Sigma-Aldrich, St.Louis, USA) was used for RNA extraction according to a slightly modified manufacturer's protocol. Due to a small sample volume, glycogen was used as a carrier to increase RNA yield. Individual GC and CC samples were homogenized in 500 μL TRI reagent supplemented with 125 μg of glycogen (Ambion, Austin, USA). After 2 min incubation at room temperature, 100 μL of chloroform was added and the sample was vortexed vigorously. RNA was precipitated from the aqueous phase with isopropanol and collected after 15 min of centrifugation at 12,000x g and 4°C. RNA pellet was washed three times with 75% ethanol, dried and dissolved in 15 μL of RNAse free water. The integrity of the RNA samples was assessed with an Agilent 2100 Bioanalyzer (Agilent, Palo Alto, USA) and the total RNA quantity was measured with a Nanodrop ND-1000 spectrophotometer (Nanodrop Technologies, Wilmington, USA). SuperScript Vilo reverse transcriptase (Invitrogen, Carlsbad, USA) was used for cDNA synthesis from 150 ng RNA according to the manufacturer's instructions.

Microarray analysis
Microarray profiling of global gene expression on RNA samples from GC and CC was performed using Agilent SurePrint G3 Human gene expression 8x60K two-color microarrays (Agilent eArray design identifier: 028004), with 42.405 unique probes targeting for most of the RefSeq mRNA sequences. The experimental design comprised of co-hybridization of each tested sample with common universal reference RNA sample (Universal Human Reference RNA from Agilent). The labeling procedure was performed using Agilent's two-color Low Input Quick Amp Labeling Kit (two-color) according to manufacturers' instructions, labeling test samples with Cy3 and reference samples with Cy5 dye.
Microarray slides were scanned using Agilent High Resolution Microarray Scanner System after hybridization, using the manufacturer's recommended scanning settings. Subsequently, microarray features were extracted using Agilent Feature Extraction (FE) software v10.7.3.1. Background was subtracted using background de-trending algorithm implemented in FE software, features with non-uniform fluorescence profiles were removed from further analyses and linear lowess intra-array normalization was performed to correct for the presence of dye bias. The consistency of log ratio values was inspected using manufacturers provided spike-in probes and reproducibility was evaluated by investigating the extent of variance in the population of replicated probes on the array. The microarray data are available at the Gene Expression Omnibus (GEO) public repository (http://www.ncbi.nlm.nih.gov/geo) under the accession number GSE55654.

Statistical analysis of microarray data
All the steps described in this section, including post-processing, statistical comparisons and classification performance estimations were performed using packages from Bioconductor v2.8 project in R statistical environment version 2.13.1.
Processed fluorescence values obtained by feature extraction were imported and inspected for presence of missing values and other irregularities. Probe annotations were obtained from the Agilent's eArray service (earray.chem.agilent.com). Prior to the calculation of Cy3/Cy5 log2 ratios, fluorescence values were offset by 100 units to prevent an undesirable increase in log2 ratio variance in the population of features with low fluorescence values. Quality assessment of microarray hybridizations was performed by inspecting the signal distribution in box and MA plots. Principal component analysis was done to identify the presence of possible batch and other confounding effects.
Statistical comparisons of expression values were evaluated using moderated t-test approach implemented in limma [16]. Significance values and log2 fold-change were calculated and afterwards, p-values were controlled for multiple testing effect by the method described by Benjamini and Hochberg [17] and implemented into the limma workflow.

Meta-analysis
Two datasets that analysed global gene expression profiles in human GC and CC samples with accession numbers GSE18559 [10] and E-MEXP-3641 [11], together with our own data. were included in the meta-analysis. Raw data from microarray experiments were obtained from Gene Expression Omnibus (GEO) repository (http://www.ncbi.nlm.nih.gov/geo/) [18] and from ArrayExpress repository (www.ebi.ac.uk/arrayexpress/). Raw datasets were put through quality control steps implemented in arrayQualityMetrics package, followed by intra-array quantile and inter-array loess normalization, where necessary. Afterwards, intensities of probe replicates targeting specific transcripts were median averaged and only values for transcripts interrogated in all three studies were retained through further meta-analysis steps.
Differential expression of genes across all three studies was calculated using meta-analysis algorithms implemented in the RankProd package [19]. RankProd uses a non-parametric approach that selects genes displaying consistently highly ranked across different microarray studies and is therefore a feasible meta-analytic tool across a variety of microarray studies originating from different laboratories and performed under differing conditions [19]. We utilized RPadvance function, with origin parameter set to account for data originating from three different sources. P values were calculated using permutation approach (1000 permutations) and predicted false positive rates estimated across permutation cycles [20].

Gene ontology and IPA analysis
Gene ontology (GO) analysis was performed to define enriched, cell specific GO profiles using Gene Codis [21]. We uploaded our two sets of genes: one set of 524 genes that were up-regulated in GC and one set of 126 that were up-regulated in CC. All of the selected genes had a log 2 fold difference > ± 2 and q-value < 10 −4 so they were considered significantly differentially expressed. Up-regulation of genes in GC population, as mentioned in this manuscript, refers to down-regulation of genes in CC and vice versa.
Our list of significantly differentially expressed genes (log 2 FC > ±2, q-value < 10 −4 ) was subjected to Ingenuity Pathway Analysis (IPA) Software (Ingenuity Systems Inc, Redwood City, USA) for network enrichment analysis in each cell subpopulation.
Pre-designed TaqMan Gene Expression assays (Applied Biosystems, Foster City, CA, USA) were used for mRNA quantification. All reactions were performed in triplicates in 96-well formats in a 20 μl final volume. The measurements were performed on ABI Prism 7000 Sequence detection system (Applied Biosystems, Foster City, USA). Thermal cycling conditions were as follows: an initial step at 50°C for 2 min, denaturation at 95°C for 10 min, amplification for 40 cycles at 95°C for 15s and at 60°C for 1 min. The threshold cycle (Ct) values were then determined for each assay and normalized to internal control glyceraldehyde-3-phosphate dehydrogenase (GAPDH) that was co-ran with each sample. Differences in gene expression were calculated using the delta-delta Ct method, as previously described by Livak and Schmittgen [22]. The significance of the mean relative gene expression level differences was calculated using two-sample t-distribution test, differences were considered significant at α < 0.05.

Global differential gene expression
The comparison of global gene expression profiles of GC and CC showed cell type specific profiles. Complete transcriptional profiles of the two cell populations were subjected to principal component analysis (PCA); first two principal variance components clearly separated expression profiles of GC and CC populations (Fig 1).
There were 6029 genes significantly (q < 10 −4 ) differentially expressed between GC and CC in our study, with 650 having a log 2 fold change 2, with 524 genes higher expressed in GC and 126 genes higher expressed in CC. The lists of top 100 differentially expressed genes are represented in Tables 1 and 2 for GC and CC, respectively. The full list of differentially expressed genes is represented in S1 Table. Meta-analysis showed there were 3156 genes consistently significantly (q < 10 −4 ) differentially expressed between GC and CC, with 1596 showing higher expression in GC and 1560 in CC. A high concordance was found between the results of our and previous two studies in terms of the direction of expression of differentially expressed genes, as well as the selection of the highest significantly differentially expressed genes. The list of 500 most differentially expressed genes across the three studies is presented in S2 and S3 Tables for GC and CC, respectively.

Gene ontology and IPA analysis
To determine significantly (p < 0.01) overrepresented biological functions in each cell population the list of differentially expressed genes in our study (524 up-regulated in GC, 126 up-regulated in CC) was subjected to GO analysis. Functions such as immune response, inflammatory response and chemotaxis were among the overrepresented in the GC compartment (Table 3). In the CC compartment protein binding, cell adhesion and multicellular organismal development were among overrepresented (Table 4).
IPA Software was used for network enrichment analysis of differentially expressed genes in each cell subpopulation. Top network enriched in the GC compartment was connected with hematological system development and function, humoral immune response and cellular movement (Enrichment score: 32) (Fig 2). Thirty genes of this network were highly expressed in GC in our study; among them being tumor necrosis factor (TNF) gene, which is at the central position of the network.
In the CC compartment, the top enriched network was connected with cellular development, skeletal and muscular system development and function and tissue development (Enrichment score: 41) (Fig 3). There were 21 genes of this network highly expressed in CC in our study. The network includes insulin-like growth factor binding protein 5 (IGFBP5) and tenascin C (TNC) genes with functions in extracellular matrix.

Validation of differential expression by quantitative real-time PCR
Microarray data from our study were validated using qPCR. Expression levels of 8 genes were checked on 19 individual GC and 19 individual CC samples derived from 16 newly included women. According to the microarray data, genes DUSP6, FGG, IL1B and PROK2 showed higher expression in GC, whereas ACE2, HTRA1, PNCK and RYR2 showed higher expression in CC. Results of the qPCR analysis were in agreement with microarray data for 7 out of 8 selected genes (Fig 4). According to qPCR data, the expression of these 7 genes significantly differed between GC and CC samples (Fig 5). For ACE2 gene, the qPCR analysis failed to confirm the differential expression from the microarray data.

Discussion
In the present study differential gene expression between human GC and CC surrounding mature oocytes was analysed. Transcriptome profiling of individual human GC and CC samples revealed two distinct cell subpopulations with 7033 genes significantly (q < 10 −4 ) differentially expressed. Out of these, 650 genes had at least a 2-fold log 2 difference of expression.
As the number of study samples included in microarray studies is generally much lower than the number of tested variables (genes), testing large number of variables on a small number of study samples can result in false-positive and false-negative detections. This effect can contribute to low reproducibility of results between the studies performed by different research groups [23]. Combining datasets from several experiments where same condition has been investigated can increase the statistical power for detection of biologically relevant genes [24]. For this reason, we decided to perform a meta-analysis in order to maximize the potential for the discovery of consistently differentially expressed genes between human GC and CC. There were 3156 genes that showed consistent significant differential expression between GC and CC after meta-analysis. Among these, there were genes that have not yet been reported to be expressed in human somatic follicular cells like PROK2 and PNCK.

Differentially expressed genes
The list of significantly differentially expressed genes derived from our study, as well as from the meta-analysis, contained genes that have not yet been described in human GC and CC, like PROK2 and PNCK. qPCR analysis confirmed differential expression of these two genes.  The PROK2 gene was over-expressed in GC. Prokineticin 1 and 2 (PROK1 and 2) are multifunctional proteins that were first described in the gastrointestinal tract, where they induce contractions of smooth muscle fibres [25]. They exert their effects through G-protein coupled receptors, prokineticin receptor 1 and 2 (PROKR1 and PROKR2) [26]. PROK2 is mainly expressed in the central nervous system and nonsteroidogenic cells of the testes [27]. It has been established that PROK2 system is involved in angiogenesis [27], control of circadian rhythms [28] and sexual maturation [29]. PROK2 was shown to be expressed in the olfactory bulb of mice, where it is involved in neurogenesis and morphogenesis. Homozygous PROK2 and PROKR2 knockout mice show hypoplasia of the olfactory bulb [30] and atrophy of the reproductive system (testis, ovary, uterus, vagina, mammary gland) [31]. Defects of the PROK2 pathway in humans affect the neuroendocrine control of reproduction by causing hypogonadotropic hypogonadism due to the GnRH deficiency, which is caused by a reduction in the number of GnRH neurons in the preoptic region of the hypothalamus [32]. Furthermore, inactivating mutations of PROK2 and PROKR2 have been discovered in 2-7% of patients with Kallmann syndrome, who suffer from anosmia and hypogonadotropic hypogonadism [33]. According to the IPA analysis in our study, PROK2 gene was a part of the top enriched network connected to hematological system development and function, cancer and cellular movement. The development of perifollicular vascularization during folliculogenesis is of fundamental importance for the nourishment of the growing follicle and thus, development of a mature and competent oocyte [34]. Vascular endothelial growth factor (VEGF) is an important regulator of this process [35]. In the bovine corpus luteum (CL), PROK1 stimulates the expression of VEGF and this finding implies that it has an indirect role in the process of angiogenesis in the CL [36]. Expression of PROK2 in GC might imply that this gene also has a role in angiogenesis   Diversity of Granulosa and Cumulus Cells during folliculogenesis. Furthermore, PROK2 in the enriched IPA network was indirectly related to the tumor necrosis factor (TNF) gene at the central position. During mammalian ovulation, TNF causes apoptosis of ovarian surface epithelium and breakdown of the extracellular matrix in the follicle wall [37]. By this the oocyte can be released from the follicle. The connection of PROK2 with TNF might thus indicate that PROK2 is involved in the process of releasing the cumulus-oocyte complex from the follicle at the time of ovulation. Another gene that has not been previously described in somatic cells of ovarian follicles and showed higher expression on our set of CC samples and after meta-analysis was PNCK. PNCK is a calmodulin kinase I isoform [38]. PNCK has a role in the murine central nervous system development, as its mRNA and protein are present there during embryonic development [39]. In adult rats, it is also expressed in the breast, uterus, heart and stomach [40]. It has been shown that PNCK mRNA over-expression causes epidermal growth factor receptor (EGFR) protein degradation leading to inhibition of EGF-induced mitogen-activated protein kinase (MAPK) activation [41]. After the LH surge, EGF-like factors (amphiregulin, epiregulin, β-cellulin) act on EGFR to activate Ras-MAPK3/1 pathway in GC and CC; this activation is crucial for the CC expansion, final oocyte maturation and GC luteinization [42]. In our study, only CC samples that surrounded mature oocytes were used for analyses; therefore, we presume that PNCK expression in CC after LH/hCG triggering oocyte maturation perhaps maintains inactivation of Ras-MAPK3/1 pathway, as the effects of this pathway are no longer needed with CC expansion and oocyte maturation being complete.
Among consistently differentially expressed genes that have already been described in human GC and CC there were also FGG and RYR2. Gene FGG that codes fibrinogen was expressed higher in GC. This finding is in accordance with a study of Parrot et al. [43], where Diversity of Granulosa and Cumulus Cells higher expression of FGG was found in equine GC and the presence of FSH increased its expression. Fibrinogen is crucial for blood coagulation and tissue reparation, and its presence in the ovary is probably necessary for restoration of tissue integrity after follicle rupture at ovulation. Gene RYR2, a regulator of intracellular Ca++ concentrations, showed higher expression in CC. Positive correlation of RYR2 expression with expression of genes involved in cell to cell communication in CC has been described [11]. Our findings further support the assumption   that intercellular signaling in CC is dependent upon gap junction proteins as well as Ca+ + signaling. Interestingly, we also identified a number of long non-coding RNA (lncRNA) transcripts as differentially expressed, when comparing GC and CC expression profiles. LncRNAs are transcripts of >200 nucleotides without known protein coding function [44]. They play a role in regulating gene expression at the level of chromatin modifications, transcription and post-transcriptional processing [45]. In human CC, differential expression of several lncRNAs was found between mature and germinal vesicle oocytes [46] and between high and poor-quality embryos [47]. These results together with the results of our study show that lncRNAs play a role in the development of oocytes. It is possible that lncRNAs expression participates in driving the differentiation of GC and CC and further studies are required to characterize their biological functions during the process of oocyte maturation.

Biological functions enriched in GC
The 524 up-regulated genes in GC represent biological functions such as inflammatory and immune response, and these results are consistent with previous studies [10,11]. Ovulation is considered to be a process similar to inflammation due to the rupture of surface epithelium followed by a healing process [48]. After the release of the oocyte from the ovary the coagulation cascade is activated through invasion of leucocytes and cytokine release, that are needed for tissue healing [49]. Among the genes involved in the inflammatory process that were significantly up-regulated in GC in our study was IL1B. IL1 system is composed of ligands IL1A and IL1B [50], receptors IL1R1 and IL1R2 [51] and a receptor antagonist IL1RA [52]. In human cultured GC IL1B was shown to regulate plasminogen activator activity and steroidogenesis [53]. In our study IL1B, IL1RA and IL1R2 were up-regulated in GC, IL1A expression was undetectable and IL1R1 was up-regulated in CC. These findings are in accordance with the study of Martoriati and Gérard [54], where equine GC showed the same expression of IL1 system genes. They confirm the proposed involvement of the IL1 system in the events regulating oocyte maturation and ovulation [55] (Gérard et al., 2004). The higher expression of IL1R1 in CC probably indicates that IL1B exerts its effects on CC through this receptor. In the present study, IL6 and IL8 were also up-regulated in GC. This finding is in accordance with previous studies, where it has been established that IL1 induces expression of pro-inflammatory genes (IL6, IL8) at the site of the follicle rupture [56]. Cytokines IL6 and IL8 induce cellular proliferation which is needed for the healing of the ovarian surface epithelium [57]. Furthermore, IL6 has an important role in ovulation process as it is involved in regulation of meitoic maturation, CC expansion and CL formation [58].
The group of 'toll-like' receptors (TLR1,2, 4, 6, 7, 8), that has a role in the innate immune response in somatic cells of ovarian follicles [59], was found to be up-regulated in GC in our study. Our results are in line with a previous study that compared transcriptomes of GC and CC [10]. In mouse CC, TLR genes activate the expression of several signalling molecules (IL6, PTGS2, PTX3) that have a known role in CC expansion [60], cumulus-oocyte complex oviductal migration [61] and fertilization [62]. In hen GC, treatment with LH led to an increase in Diversity of Granulosa and Cumulus Cells TLR's mRNA in pre-ovulatory follicles [59]. In the study of Woods et al. [63] it was shown that TLR signaling pathway influences hen GC steroidogenesis and apoptosis in relation to follicle size. Additionally, it has been suggested that TLR's function as a detection system to initiate tissue regeneration after injury [64]. According to these findings, we may presume that TLR signaling pathway also modulates human folliculogenesis. Furthermore, the high expression of TLR's in GC of pre-ovulatory follicles might support the suggestion that they are important in the healing process at the sight of the follicle rupture after ovulation. Additional studies are needed to determine the effects of TLR mediated signaling in GC during folliculogenesis.

Biological functions enriched in CC
The 126 significantly up-regulated genes in CC represent biological functions as multicellular organismal development and cell differentiation. Genes IGFBP5 and TNC were among the genes involved in signal transduction. IGFBP5 is a binding protein for insulin-like growth factors 1 and 2 (IGF1, IGF2), that are known to stimulate extracellular matrix production [65] and regulate steroidogenesis [66] during folliculogenesis. IGFBP proteins control the bioactivity of IGFs and controlled access to IGF1 is crucial for the proper coordination of early oocyte and follicular development [67]. In the present study, IGFBP5 expression was increased in CC and IGF1 expression was increased in GC. These results are in line with the study of Ingman et al. [68] where treatment of CC culture media with IGF1 increased extracellular matrix IGFBP5 by 2.5-fold. Our results probably indicate that in pre-ovulatory follicles, bioavailability of IGF1 protein derived from GC is regulated by IGFBP5 expression in CC.
TNC is an important glycoprotein of the extracellular matrix [69]. It is involved in cellular functions as adhesion, migration, embryonic development, wound healing and tumour metastasis [70]. It was shown to be overexpressed in the stroma of malignant ovarian tumours and it was proposed to be involved in the process of tumour invasion [71]. In ovarian cancer cell line expressions of transforming growth factor beta 1 (TGFB1) and TNC are significantly related [72]. Furthermore, TGFB1 induces secretion of TNC from ovarian fibroblasts [68]. During folliculogenesis TGFB1 enables CC hyaluronic acid production and expansion, regulates CC steroidogenesis and promotes GC proliferation [73,74,75]. TNC was significantly higher expressed in CC after our microarray analysis as well as after the meta-analysis. Over-expression of TNC in CC might thus indicate that TNC is one of the genes through which TGFB1 exerts its effects during the process of oocyte maturation. Evidently, if this process is pathological, it is related to ovarian tumour invasion. The precise role of TNC in ovarian folliculogenesis and the possible relation with TGFB1 will need to be determined in future studies.
In conclusion, with this study we upgraded the existing data on differentially expressed genes between GC and CC by description of PROK2 and PNCK genes whose expression has, to the best of our knowledge, not yet been reported in human GC and CC. These genes could be involved in any number of the processes regulating folliculogenesis. Further research is needed to elucidate their potential physiological roles. Pathway enrichment and IPA analyses suggested that GC are involved in angiogenesis and ovarian healing process after follicle rupture, whereas pathways enriched in CC are connected with intercellular communication and extracellular matrix formation. Identification of genes and pathways that contribute to the development of mature and competent oocytes has the potential to improve oocyte in vitro maturation procedures.

Author Contributions
Conceived and designed the experiments: TBP EVB ANK LL. Performed the experiments: TBP AM ANK LL. Analyzed the data: TBP AM. Contributed reagents/materials/analysis tools: TBP AM ANK LL. Wrote the paper: TBP EVB AM ANK LL.