|
INSERM Subrepository The human cumulus--oocyte complex gene-expression profile * Correspondence should be adressed to: John De Vos Email: devos/at/montp.inserm.fr * Correspondence should be adressed to: Samir Hamamah Email: s-hamamah/at/chu-montpellier.fr ![]() ![]() | |||||||||
Abstract BACKGROUND The understanding of the mechanisms regulating human oocyte maturation is still rudimentary. We have identified transcripts differentially expressed between immature and mature oocytes, and cumulus cells.METHODS Using oligonucleotides microarrays, genome wide gene expression was studied in pooled immature and mature oocytes or cumulus cells from patients who underwent IVF.RESULTS In addition to known genes such as DAZL, BMP15 or GDF9, oocytes upregulated 1514 genes. We show that PTTG3 and AURKC are respectively the securin and the Aurora kinase preferentially expressed during oocyte meiosis. Strikingly, oocytes overexpressed previously unreported growth factors such as TNFSF13/APRIL, FGF9, FGF14, and IL4, and transcription factors including OTX2, SOX15 and SOX30. Conversely, cumulus cells, in addition to known genes such as LHCGR or BMPR2, overexpressed cell-tocell signaling genes including TNFSF11/RANKL, numerous complement components, semaphorins (SEMA3A, SEMA6A, SEMA6D) and CD genes such as CD200. We also identified 52 genes progressively increasing during oocyte maturation, comprising CDC25A and SOCS7.CONCLUSION The identification of genes up and down regulated during oocyte maturation greatly improves our understanding of oocyte biology and will provide new markers that signal viable and competent oocytes. Furthermore, genes found expressed in cumulus cells are potential markers of granulosa cell tumors.Keywords: Carrier Proteins, biosynthesis, Down-Regulation, Female, Fertilization in Vitro, Gene Expression Regulation, Developmental, Granulosa Cells, physiology, Growth Substances, biosynthesis, Humans, Membrane Glycoproteins, biosynthesis, Membrane Proteins, biosynthesis, Neoplasm Proteins, biosynthesis, Oligonucleotide Array Sequence Analysis, Oocytes, physiology, Protein-Serine-Threonine Kinases, biosynthesis, RANK Ligand, Receptor Activator of Nuclear Factor-kappa B, Semaphorins, biosynthesis, Transcription Factors, biosynthesis, Tumor Necrosis Factor Ligand Superfamily Member 13, Tumor Necrosis Factor-alpha, biosynthesis, Up-Regulation Keywords: oocytes, germinal cells, microarray, cumulus | |||||||||
The quality of oocytes obtained under controlled ovarian stimulation (COS) for assisted reproductive technology (ART) varies considerably. While most oocytes are amenable to fertilization, only half of those fertilized complete preimplantation development and even fewer implant. During follicle growth, the oocyte obtains the complement of cytoplasmic organelles and accumulates mRNAs and proteins that will enable it to be fertilized and to progress through the first cleavage divisions until embryonic genes start to be expressed. Transcriptional activity decreases as the oocyte reaches maximal size (Fair et al., 1995) and later on the oocyte depends on stored RNAs for normal function during maturation, fertilization and early embryonic development (Moor et al., 1998). After oocyte retrieval, the mature oocyte (MII) and some still immature oocytes (GV and MI) are surrounded by the cumulus oophorus. Several layers of cumulus cells surround the oocyte in antral follicle and play an important support and regulation role in oocyte maturation (Dekel and Beers, 1980; Larsen et al., 1986). Analysis of the oocyte maturation using microarray analysis techniques could detail the genes involved in this process and the specific checkpoints regulating acquisition of full competence for ovulation and fertilization. The understanding of the molecular processes involved in the development of a competent oocyte under COS conditions could guide the choice of ovarian hyperstimulation protocols and lead to improvements in oocyte quality, oocyte culture and manipulation. Some studies demonstrate that changes in gene expression during COS, such as GDF9 or Bone Morphogenic Protein-15 (BMP15) in oocyte, or Pentraxin 3 (PTX3) in cumulus cell, can be monitored for selecting oocytes for fertilization and embryos for replacement (Elvin et al., 1999; Yan et al., 2001; Zhang et al., 2005). Therefore, transcriptome studies in human oocytes and cumulus cells could contribute not only to elucidate the mechanisms of oocyte maturation, but could also provide valuables molecular markers of abnormal gene expression in oocytes with reduced competence. The aims of the present study were to establish: (1) whole genome transcriptome of human immature and matures oocytes and cumulus cells, (2) specific gene expression signatures of immature and mature oocytes and cumulus cells and (3) genes whose expression progressively increase during oocyte maturation. | |||||||||
Oocytes and cumulus cells Oocytes and cumulus cells were collected from patients consulting in our center for conventional in vitro fertilization (cIVF) or for intracytoplasmic sperm injection (ICSI). This study has received institutional review board approval. Patients were stimulated with a combination of gonadotropin-releasing hormone agonist (GnRH-a) (Decapeptyl PL 3; Ipsen, Paris, France) and recombinant FSH (Puregon and Gonal F; Organon and Serono respectively) or Menopur (Ferring). Ovarian response was evaluated by serum estradiol level and daily ultrasound examination to observe follicle development. Retrieval of oocytes occurred 36 hours after hCG administration and was performed under ultrasound guidance. Cumulus cells were removed from a mature oocyte (MII) 21 hours post insemination. Immature oocytes (GV and MI) and unfertilized MII oocytes were collected 21 hours or 44 hours post insemination or post microinjection by ICSI. Cumulus cells and oocytes were frozen at −80°C in RLT buffer (RNeasy kit, Qiagen, Valencia, CA, USA) before RNA extraction. Pools of 20 GV (7 patients, age 30 years ±4.6), 20 MI (6 patients, age 30.1 years ±6.7) and 16 MII oocytes (6 patients, age 34 years ±4.5) were analyzed by DNA microarrays. All these oocytes were from couples referred to our center for cIVF (tubal infertility) or for ICSI (male infertility).Complementary RNA (cRNA) preparation and microarray hybridization RNA was extracted using the micro RNeasy Kit (Qiagen) and the RNA integrity was assessed by using an Agilent 2100 Bioanalyzer (Agilent, Palo Alto, CA). RNA quantity was also assessed for some samples using the Nanodrop ND-1000 spectrophotometer (Nanodrop Technologies Inc., DE, USA). cRNA was prepared according to the manufacturer’s protocol “small sample protocol II” starting from total RNA (ranging from ~4 ng pooled oocytes to 100 ng cumulus cells), and hybridized to HG-U133 plus 2.0 GeneChip pangenomic oligonucleotide arrays (Affymetrix, Santa Clara, CA, USA). HG-U133 plus 2.0 arrays contain 54 675 sets of oligonucleotide probes (“probeset”) which correspond to ≈ 39 000 unique human genes or predicted genes. The GeneChip system is a robust microarray system with more than 3000 publications using this technology (http://www.affymetrix.com/community/publications/index.affx), little lab-to-lab variability and a good accuracy and precision (Irizarry et al., 2005). Primary image analysis of the arrays was performed by using GeneChip Operating Software 1.2 (GCOS) (Affymetrix), resulting in a single value for each probe set (“signal”). Data from each different array experiment were scaled to a target value of 100 by GCOS using the “global scaling” method. The dataset was floored to 2, i.e. each signal value under 2 was given the value 2.Statistical analysis Samples were analyzed using a pair wise comparison using the GCOS 1.2 software (Affymetrix). Of interest, this algorithm provides the information of 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, “absent” when both perfect match and mismatch probes display a similar fluorescent signal, or “marginal” when the probeset does neither comply to the “present” nor to the “absent” call criteria. A gene was denoted as exclusively expressed in one category when this gene displayed a detection call “Present” in this given category and “Absent” or “Marginal” in all other three categories. A gene was considered as over- or underexpressed in a category when all three possible pair wise comparisons showed a significant change p-value (P ≤ 0.01) according to the GCOS 1.2 software and a ratio ≥ 3 or ≤ 0.333 for the genes increased and decreased respectively. We also determined a list of genes whose expression progressively increased during oocyte maturation by selecting the probesets with a significant increase according to the GCOS 1.2 algorithm and matching the following ratio constraints: cumulus < GV (with GV/Cumulus ≥3), GV < MI (MI/GV ≥ 1.73) and MI < MII (MII/MI ≥ 1.73), where cumulus, GV, MI and MII stands for signals values in these samples. Note that 1.73 × 1.73 = 3. Gene annotation was based on Unigene Build 176.For hierarchical clustering, data were filtered (15 000 genes with a significant expression (“present” detection call) in at least one sample and with the highest variation coefficient), log transformed, median centered, and processed with the CLUSTER and TREEVIEW software packages with the average linkage method and an uncentered correlation (Eisen et al., 1998). Gene Ontology (GO) annotations (http://www.geneontology.org/) were obtained and analyzed via the FATIGO web site tool (http://www.fatigo.org/) using level three annotations. In some cases we used the GO annotations downloaded from the Affymetrix NetAffx database. Genes with a role in cell-to-cell communication function were obtained by filtering the genes on the following criteria: cellular component comprising the terms “membrane” or “extracellular”. Bibliographical search was carried out in Pubmed using Boolean logic. For each gene G present in table 2 and 3, using its Hugo approved abbreviation or any of its aliases, we looked for publication matching the query « gene G AND (gamete or “germ cell” or “germ cells” or egg or eggs or oocytes or oocyte or meiosis) » for genes found preferentially expressed in oocytes, and the query « gene G AND (gamete or “germ cell” or “germ cells” or egg or eggs or oocytes or oocyte or cumulus or granulosa) » for genes overexpressed in cumulus cells. The expression, including signal values, of all genes cited in Tables 2 and 3 can be examined on our web site as online supplemental data : http://amazonia.montp.inserm.fr/the_human_oocyte_transcriptome.html. Expression of these genes in various normal tissues transcriptome datasets, including ovarian and testis samples, is provided through the Amazonia! database web page (Manuscript in preparation).
| |||||||||
Identification of genes expressed in human oocytes and cumulus cells Total cRNA was synthetized from pools of GV, MI or MII stage oocytes, or cumulus cells, then labeled and hybridized to pan-genomic oligonucleotide microarrays. We analyzed the detection call (GCOS 1.2 software) of all 54 675 probes in oocytes and cumulus samples. Oocytes express in average 8728 genes. The lowest number of genes expressed was found in MII oocytes (n = 5 633) and highest in GV oocytes (n = 10 892) (Table 1). We found that expression variations between MI, MII or GV samples was low as illustrated by tight scatter plots and high correlation coefficients (0.63 – 0.92), as opposed to a marked difference of expression between the cumulus sample and the oocytes samples as illustrated by dispersed scatter plots and low correlation coefficients (0.39 – 0.50) (Figure 1A).
We visualized the respective gene expression across all samples using hierarchical clustering. Average linkage hierarchical clustering on 15 000 genes showed that oocytes cluster together, demonstrating a common gene expression, but are only distantly related to cumulus cells (Figure 1B). These results highlight that feminine germ cells and their nourishing neighbor cumulus cells display very different expression profile, in agreement with a very different but complementary biological function and with cell lineage disparity. Specific transcription program in each sample type We next examined which genes were specific to each sample category, using two different approaches. First we determined the genes that were only detected in one sample and not in the three other samples. These genes were called “exclusively expressed” (Table 1). As expected, cumulus cells have the largest number of exclusively expressed genes (n = 1829), likely because they display a very different transcriptome as compared to oocytes (n = 234 – 739). Second, we considered the probes that were overexpressed or underexpressed in one sample compared to all three other samples, with a fold ratio of at least three. Again, cumulus cells show the largest lists of genes, overexpressing 2600 and underexpressing 1514 genes as compared to oocytes. Using this rather stringent criteria (fold change of at least 3 between a given sample and the three other samples), we found very few genes over or under expressed in GV and MI oocytes. This shows that very few genes modify their expression between GV and MI oocytes, as opposed to MII oocytes that overexpress more than 400 genes and underexpress more than 800.We compared functional Gene Ontology annotations of overexpressed genes versus under expressed genes in oocytes and cumulus cells. We observed that certain functional annotations were more represented in either oocytes or cumulus cells (Figure 2). There were significantly more genes involved in “Response to stimulus”, “Secretion”, “Extracellular matrix” in cumulus cells, suggesting that cumulus cells are more active in cell-to-cell communication processes. Conversely, genes annotated “Reproduction”, “Ubiquitin ligase complex”, “Microtubule associated complex”, “Microtubule motor activity”, “Nucleic acid binding”, “Ligase activity” were significantly more frequently associated with genes overexpressed in oocytes, in agreement with the major processes involved in meiosis and implying microtubules attachement to chromosomes and the ubiquitin ligase complex APC/C regulation.
Whole genome transcriptome of oocytes We observed that 1514 genes were expressed with at least a 3-fold increase in oocytes, i.e. underexpressed in cumulus cells when compared to oocytes. Selected genes are highlighted in Table 2, which is also available as web supplemental data including the expression histogram for each gene (http://amazonia.montp.inserm.fr/the_human_oocyte_transcriptome.html). This list includes genes already recognized as specifically expressed in male and female germinal cells in mammals such as DAZL, the RNA helicase DDX4/VASA or DPPA3/STELLA (full names are listed in Table 2). Numerous well recognized actors of meiosis were highly expressed in oocytes: the components of the maturation-promoting factor (MPF) (CDC2/CDK1, CCNB1, CCNB2), CDC25 phosphatases (CDC25A, CDC25B and CDC25C), components of the spindle checkpoint (BUB1, BUBR1, MAD2L1/MAD2, CENP-A, CENP-E), CDC20 which is a components of the anaphase promoting complex (APC/C), and a downstream target, the meiosis specific sister chromatid arm cohesin STAG3 (Figure 3A). As expected, we observed the overexpression of genes known to be specific of oocytes such as the Zona Pellucida genes (ZP 1, 2, 3 and 4), members of the transforming growth factor-beta superfamily such as Growth differentiation factor 9 (GDF9), Bone morphogenetic protein 6 and 15 (BMP6 and BMP15), FGFR2, the chromatin remodeling molecules histone deacetylase HDAC9 and the oocyte-specific H1 histone H1FOO (Figure 3B). Thus, the data are in complete agreement with published studies. Interestingly, we show here that many genes, previously found expressed in oocytes in various animal models, are indeed highly expressed in human oocytes. Hence, our microarray data are of sufficient scope and accuracy to pave the way to a systematic gene expression exploration of oocyte and cumulus transcriptome.
We observed that several genes previously reported to be expressed in male germ cells are also highly expressed in human oocytes, in all maturation stages, such as Aurora Kinase C (AURKC), SOX30, or Sperm Associated Antigen 16 (SPAG16/PF20). Still, the majority of the genes we found overexpressed in oocytes were not yet reported to be associated with gamete biology. Some of these previously unrecognized “oocytes genes” are listed in Table 2 and comprise several functional categories. After fertilization, the spindle checkpoint inhibition is released and the APC/C complex degrades the securins, resulting in entry into anaphase. We found that genes of the centromere protein CENPH that interacts with the spindle checkpoint, the anaphase promoting complex subunits ANAPC1 and ANPC10 are highly expressed in oocytes. Moreover, the securing genes PTTG1 and 3 are 58 and 50 times more expressed in oocytes than in cumulus cells, respectively. We found several growth factors and growth factor receptors significantly overexpressed in oocytes (IL-4, FGF9, FGF14, TNFSF13/APRIL), transcription factors (SOX15, OTX2, FOXR1), three anti-apoptosis molecules (BCL2L10, BNIP1, BIRC5/Survivin), and the glucose transporter SLC5A11. Whole genome transcriptome of cumulus cells Conversely, we observed that 2600 genes are overexpressed in cumulus cells compared to all three oocytes samples. The cumulus sample we studied was obtained from a MII oocyte during ovulation. First, we observed a marked expression of the LH receptor LHCGR in cumulus cells, which primes these cells to respond to the LH surge. Second, we observed that genes overexpressed in MII cumulus cells comprise the main genes that are induced by the LH surge during ovulation (Table 3). We observed a very high expression of the progesterone receptors PGRMC1 and 2, and the steroidogenic acute regulatory (STAR) that are induced by LH. Similarly we found that eicosanoids biosynthesis enzymes such as the two Prostaglandin Endoperoxyde Synthetase PTGS1 and PTGS2/COX2 and the Prostaglandin I2 (Prostacyclin) Synthase PTGIS, the Prostaglandin Receptor PTGER2, and two downstream effector of this signalling pathway, Interleukin IL1beta and Pentaxin-Related 3 (PTX3), are also overexpressed in cumulus cells. These genes were mostly described in animal models, and we show here for the first time that the RNA expression of these genes is also highly induced in human cumulus cells obtained after ovulation. Two chemokines are highly produced by cumulus cells, CXCL1/GRO-alpha and IL8, in agreement with the invasion of the granulosa by leucocytes during ovulation. Interestingly, the metalloprotease ADAMTS1, as well as its target Versican whose cleavage has been shown to contribute to the proteolytic disintegration of the cumulus matrix, were also highly induced. The transcription factor CEBPB, induced after the gonadotrophin surge and mediating the upregulation of inhibin alpha (INHA), is found overexpressed in our post stimulation cumulus cells. Accordingly, INHA, as well as INHBA/Activin A, are 5 and 34 times more expressed in cumulus cells than in oocytes respectively. Another transcription factor characteristic of granulosa cells, GATA6, is also highly overexpressed in comparison to oocytes. We observed the upregulation of peroxiredoxins (PRDX2, 4, 5 and 6) that are part of a family of peroxidases involved in antioxidant protection and cell signaling and recently reported in bovine ovaries (Leyens et al., 2004), as well as a lysosomal cysteine proteinase, cathepsin K (CTSK). Genes coding for protein found in follicular fluid such as PAPPA are also found overexpressed in cumulus cells. Thus, genes found overexpressed in cumulus cells by our whole genome transcriptome analysis recapitulates previous expression studies on post-LH surge granulosa cells carried out in various species.Considering that cell-to-cell communication genes are a functional category that plays an essential role in the maturation of the cumulus-oocyte complex, we focused on genes filtered on the Gene Ontology cellular localization annotations “membrane” or “extracellular” (see material and method): 615 genes passed this filter. The most noticeable genes from this list were ligands (BMP1, BMP8B) or receptors (BAMBI, BMPR2) from the TGF superfamily, ligands (TNFSF11/OPGL/RANKL) or receptors from the TNFR superfamily (TNFRSF1A/TNF-R, TNFRSF10B/DR5, TNFRSF12A), components of the complement (CFHL1, C7, IF, CFH, C1S, C1R) and one inhibitor of the complement system (CLU), semaphorins (SEMA3A, SEMA6A, SEMA6D), tetraspanins (TM4SF1, TM4SF6, TM4SF8, TM4SF10) and various CD members (CD24, CD44, CD47, CD58, CD59, CD63, CD74, CD81, CDW92, CD99, CD151, CD200). Table 3 lists these genes, references key publications relevant for feminine reproduction biology. Furthermore, components of the cumulus-oocyte complex signalling pathways were retrieved such as connexin 43. We found that this connexin was expressed at a high signal in both cumulus cells and in all oocytes categories, in line with its extracellular domains that provide strong and specific homophilic adhesion properties. Most interestingly, many of these genes were never before highlighted as expressed in granulosa cells. Differences in genes expression variation during oocyte maturation An important feature of our work is that we established a transcriptome for each of the three stages of oocyte maturation: GV, MI and MII. We were thus able to identify genes whose expression gradually increased during oocyte maturation (see material and method). Fifty two probesets were retrieved, including the phosphatase CDC25A, PCNA and SOCS7. However, most of the resulting genes are poorly characterized or only predicted coding sequences. All these genes are candidate marker for oocyte cytoplasmic and/or nuclear maturation. | |||||||||
We undertook to establish the molecular transcriptome phenotype of the human oocyte and its surrounding cumulus cells by using oligonucleotide microarrays covering most of the genes identified in human. Relying on a recently developed technique of double in vitro transcription, that amplifies more than 100 000 times the initial RNA input, we were able to establish the expression profile of pooled oocytes from distinct maturation stages, and from cumulus cells of MII oocytes. Thus, for the first time, we report in human samples, the variation of gene expression during oocyte nuclear maturation, and that of their neighboring cumulus cells, at whole genome scale. A global analysis of the number of genes detected in each sample category showed a progressive decrease of the number of genes expressed during oocyte nuclear maturation, with the lowest number of genes expressed found in MII oocytes compared with GV or MI oocytes. This is in agreement with the significant decrease, both in quantity and in diversity, of maternal RNAs observed in mouse oocytes (Bachvarova et al., 1982)(Wang et al., 2004). Indeed, GV and MI oocytes over or under expressed few genes compared with the other samples (Table 1), reflecting a very similar expression profile. By contrast, MII oocytes differed markedly, underexpressing specifically many genes (n = 803), which may be explained by the RNA content decrease. In addition, MII oocytes overexpress 444 genes, which may be due to a specific expression pattern related to the near completion of meiosis, or to the longer in vitro incubation time secondary to the IVF procedure (21 or 44 hours post-insemination). Hierarchical clustering demonstrated that oocytes expression profiles where markedly different from cumulus cells (Figure 1). We compared the oocytes samples to the cumulus cells and we found that 1514 genes were upregulated in oocytes whereas 2600 genes were upregulated in cumulus cells. Analyzing these lists of genes, we observed that oocyte markedly overexpressed genes involved in meiosis process such as MPF, APC/C and spindle checkpoint complexes. Full completion of meiosis is only accomplished after fecundation because metaphase exit is prevented by the activity of cytostatic factor (CSF) that will only be relieved by gamete fusion. As expected, EMI1, which was recently found to be part of CSF, is highly expressed in all oocytes samples, as well as MOS. We also found that the two major cyclin-dependent kinase inhibitors CDKN1A/p21 and CDKN1B/p27, acting at the G1-S transition, were found markedly downregulated in oocytes as compared to cumulus cells (see Table 3). The separation of sister chromatids at the metaphase-to-anaphase transition is activated by proteases called separases that are activated by the destruction of the inhibitory chaperone securins. Interestingly, we found two securins highly expressed in all oocytes pools: PTTG1 and 3. These securins are expressed at least 15 times more in oocytes than in cumulus cells, CD34+ sorted bone marrow cells, B lymphocytes or mesenchymal stem cells (data not shown). PTTG1 expression was reported in mice oocytes, but not human oocytes, whereas PTTG3 marked expression in oocytes was not previously noted. Considering that post-ovulation oocytes are germinal cells that have just escaped the very long meiosis I arrest and are due to the second meiosis arrest, securins, that are crucial to these processes, must be expressed at a high level. We propose that PTTG1 and 3 play this role in oocytes (Figure 3). The metaphase-to-anaphase transition is associated with a rapid drop of securin protein level mediated by the proteases of the separase family. Degradation of securins leads to the destruction of cohesins, a ring structure formed by a multisubunit complex that holds sister chromatids together. We confirm the specific upregulation of the meiosis-specific cohesin subunit STAG3 in human oocytes, whereas the mitotic cohesin STAG2 is markedly downregulated in oocytes compared to cumulus cells or other somatic cells (data not shown). Thus, as for the securins, two homologs of an essential component of the cell division machinery are differentially expressed between human oocytes and somatic cells, implying that one homolog (the cohesin STAG2) is operating during mitosis, whereas the other homolog (the cohesin STAG3) is replacing the first one during the very specialized cell division process of meiosis. The high conservation of many of the molecular determinants of gametogenesis in the animal kingdom, sometimes from yeast to mammals, suggests that genes found in mammals oocytes should be expressed in human oocytes. We provide here the unambiguous demonstration for many genes that they are indeed strongly overexpressed in the three pools of oocytes (Table 2). These genes include CENPA, CENPE, PTTG1, FBXO5/EMI1 or BMP6. These results underscore the consistency of our approach. Furthermore, the inventory of human genes essential for nuclear and cytoplasmic oocyte maturation is an important step toward the comprehensive understanding of oocyte biology. Although female and male gametes differ in many aspects, they share a common meiosis machinery. Indeed, we see here that genes reported to be expressed specifically in spermatozoa are also highly overexpressed in oocytes in comparison with somatic cumulus cells. This is the case for Aurora Kinase C (AURKC), Sperm Associated Antigen 16 (SPAG16/PF20) and SOX30 (Osaki et al., 1999; Horowitz et al., 2005; Yan et al., 2005). Three aurora kinases have been identified (AURA/STK6, AURKB and AURKC) that share a conserved catalytic domain and play a role in centrosome separation and maturation, spindle assembly and segregation, and cytokinesis (Giet et al., 2005). Whereas AURA and AURKB are involved in mitosis in somatic cells, AURKC was only found highly expressed in testis, suggesting a tissue specific role in meiosis. It is therefore of special interest to observe that AURKC is also 49 times more expressed in pure oocytes samples than in somatic cells. Since AURKB and AURKC have a similar cellular localization and a similar biological activity such as SURVIVIN/BIRC5 binding, we propose that AURKC is replacing AURKB during meiosis in both male and female gametes. In line with this proposition, our data shows that in oocytes samples, AURKB expression is close to background whereas survivin/BIRC5, a known partner of the AURKC complex (Yan et al., 2005) is also strongly overexpressed. We found the specific upregulation in oocytes of two methyltransferase enzymes (DNMT1 and 3B), one histone deacetylase (HDAC9) and an oocyte specific histone (H1FOO). Interestingly, the Chromosome Condensation Protein G (HCAP-G) which is a components of the condensin complex that mediates genome-wide chromosome condensation at the onset of mitosis and directly interacts with DNMT3B (Geiman et al., 2004) is also found preferentially expressed in oocytes, suggesting that this condensin is essential to the nuclear maturation of oocytes. Keeping in line with epigenetic modifications of the genome, we screened our list of oocytes genes for imprinted genes. Of note, one paternally imprinted gene, MEST, was highly overexpressed in all three oocytes samples as compared to cumulus cells, while other paternally imprinted gene such as IGF2 or NNAT were not. We noted the overexpression of two pro-apopototic genes in oocytes (BNIP1 and BCL2L10). These findings strongly argue in favor of a model where the survival of oocytes is mediated by external signals provided by surrounding cumulus cells rather than by intrinsic cues such as overexpression of anti-apoptotic factors. Accordingly, we found many receptors for growth factors overexpressed on oocytes, including a BMP receptor (BMPR2), the receptor for the stem cell factor (KIT), a member of the EGF receptor familiy (ERBB4), and a frizzled receptors (FZD3) member of the WNT pathway. In addition we observed 6 poorly characterized G protein-coupled receptors in oocytes (GPR37, GPR39, GPR51, GPR126, GPR143, GPR160). The fact that oocytes overexpress these growth factors receptors strongly suggests that the ligands of these receptors are involved in conveying surviving and maturation cues from the oophorus cumulus to the oocytes. Conversely, oocytes express many growth factors. Among the genes, we noted the remarkable overexpression of a ligand from the TNF superfamily, TNFSF13/APRIL that we found 131 times more expressed in oocytes than in cumulus cells. We did not see a significant expression of the two TNF receptors for APRIL, TNFRSF13B/TACI and TNFRSF17/BCMA (data not shown). But it was recently described that APRIL’s binding to proteoglycan was necessary for the survival signal conveyed by this cytokine to targets cells (Ingold et al., 2005). Since cumulus cells overexpress several proteoglycan such as CSPG2/VERSICAN (Table 3) and SYNDECAN4 (data not shown), APRIL could mediate a comparable trophic signal from the oocyte to the surrounding cumulus cells. We also focused our analysis on genes which expression increased progressively during oocyte meiosis. We postulate that they could be interesting candidate genes for oocyte maturation. Indeed, if these genes fail to be upregulated in MII-stage oocytes, it is likely that the maturation process was defective. Genes increasing progressively during oocyte maturation comprise SOCS7. This gene is part of a family of proteins negatively regulating intracellular signal transduction cascades (Krebs and Hilton, 2000). Its overexpression in MIIstage oocytes may indicate the shutting down of specific cytokine signalling. For this category, it must also be noted that many genes are still not characterized and remain without any hint about their function (20 out of 48 genes, = 42%). It is not a surprise if so many genes from this list have escaped bioinformatics or biological functional investigations to date, because (i) MII-stage oocytes are a very rare cell type, (ii) it is a very specialized cell type expressing numerous genes that may not be found in any other tissue type, including genes devoided of any molecular motif found in other tissues, and (iii) we used here pangenomic microarray to study here for the first time gene expression of this cell type without any selection bias. It will be essential to describe in detail the function of these genes to obtain further insights in oocyte biology. In order to decipher the tight relationship weaved between the oocyte and its surrounding follicle cells, we also analyzed the transcriptome profile of cumulus cells. Indeed, 24% of the 2600 genes overexpressed in cumulus cells are annotated either “membrane” or “extracellular”, demonstrating a strong bias towards genes involved in cell-to-cell communication processes. The signalling pathways involved comprise the progesterone and its receptors, eicosanoids and several enzymes involved in their biosynthesis and chemokines. We showed in this study that cumulus cells up regulated hormonal receptors and hormones such as LHCGR, Inhibin alpha, Inhibin beta A, GNRH1 and progesterone receptor membrane component1 and 2. Interestingly, cumulus cells overexpress BMPR2 which is the receptor for GDF9 which is overexpressed by oocytes, demonstrating a typical intercellular communication process. In addition to the inhibins INHA and INHBA, we also observed the overexpression of BMP1 and BMP8B, as well as the pseudoreceptor BAMBI, lacking an intracellular serine/threonine kinase domain and thus negatively regulating TGF-beta signalling. Another important growth factor superfamily found to be overexpressed in cumulus is the TNF superfamily. The marked overexpression of TNFSF11/OPGL/RANKL (80 times more expressed in cumulus cells than in oocytes) is intriguing and awaits further investigations. Magier et al. suggested a positive effect of cumulus cells on fertilization, a protective effect and a possible beneficial effect on further embryo development (Magier et al., 1990). In addition, Platteau et al. (Platteau et al., 2004) suggested that the exogenous luteinizing hormone activity may influence treatment outcome in IVF but not in ICSI. We provide here molecular evidence for cumulus cells expression by of hormones and growth factors that could mediate these functions. Another puzzling observation is the increased expression of seven complement factors or closely related genes. Whether this overexpression is involved in the cellular destruction process taking place in the antrum during ovulation needs to be considered. Finally, cumulus cells express several other cell surface gene families such as semaphorins, first identified for their role in neuron guidance, tetraspanins, with one member, CD9, directly involved in fertilization (Le Naour et al., 2000), and many other CD molecules with various function (Table 3). Very interestingly, some genes overexpressed in granulosa cells are also found expressed in ovarian tumors. We found for example a high expression in cumulus cells of CD24 and CD99 which are expressed in ovarian tumors and have been proposed as either diagnostic tools (Choi et al., 2000) or as prognostic tools (Kristiansen et al., 2002). These findings suggest that many of the genes overexpressed in cumulus samples, including the cell surface markers of cumulus cells listed in Table 3, could provide ovarian cancer markers. We pooled oocytes according to their maturation stage for this first, exploratory, whole genome transcriptome analysis. This strategy leveled down differences that would be associated with different IVF settings such as maternal age, sperm exposure or in vitro incubation time length. In order to describe the expression modifications that may be related to specific conditions, we are currently analyzing the transcriptome of oocytes pooled according to the hormonal profile at day 3, maternal age or ovarian hyperstimulation protocol. Nevertheless, to appreciate variations in gene expression according to each patients idiosyncrasy, we will need to achieve reliable transcriptome analysis from single oocytes. In conclusion, DNA microarray provided us with the opportunity to analyze human oocytes and cumulus cells expression profiles on a genome scale and permitted a significant progress to understand the molecular events involved in the process governing oocyte maturation. Many of the genes described here may well provide markers to monitor health, viability and competence of oocytes. In addition, underpinning oocyte growth factors receptors should help to design optimal in vitro culture conditions for oocyte and early embryo development. | |||||||||
| |||||||||
Acknowledgments We are grateful to Stephan Gasca, Irène Fries, Benoit Richard, Benoit Latucca and Benoît Crassou for helpful discussions. We wish to thank all members of our ART team for their assistance during this study. This study was supported by grants from Ferring and Organon Pharmaceuticals France. | |||||||||
References
| |||||||||