In vitro oocyte maturation (IVM) holds great promise as a tool for enhancing clinical treatment of infertility, enhancing availability of nonhuman primates for development of disease models, and facilitating endangered species preservation. However, IVM outcomes have remained significantly below the success rates obtained with in vivo matured (VVM) oocytes from humans and nonhuman primates. A cDNA array-based analysis is presented, comparing the transcriptomes of VVM oocytes with IVM oocytes. We observe a small set of just 59 mRNAs that are differentially expressed between the two cell types. These mRNAs are related to cellular homeostasis, cell-cell interactions including growth factor and hormone stimulation and cell adhesion, and other functions such as mRNA stability and translation. Additionally, we observe in IVM oocytes overexpression of PLAGL1 and MEST, two maternally imprinted genes, indicating a possible interruption or loss of correct epigenetic programming. These results indicate that, under certain IVM conditions, oocytes that are molecularly highly similar to VVM oocytes can be obtained; however, the interruption of normal oocyte-somatic cell interactions during the final hours of oocyte maturation may preclude the establishment of full developmental competence.
- assisted reproduction
- nonhuman primate
assisted reproduction technologies (ARTs) are achieving increasing prominence in reproductive medicine. The Centers for Disease Control and Prevention (CDC) reports that the number of ART cycles nationally for 2003 was 122,872, with 48,576 children born in this country for that year alone (∼200,000 babies worldwide/year), and by 2006 over 3 million children had been born worldwide after ARTs (47, 116). With this increasing application of ARTs comes increased interest in optimizing efficiency while minimizing potential risks to the offspring. Societal and ethical concerns about ARTs, as well as a desire to avoid multiple-gestation pregnancies, are pushing the ART field to adopt strategies in which small numbers of high-quality embryos are produced and transferred to establish pregnancy (21, 44, 68, 106). The emerging awareness of potential long-term health consequences of ARTs (3, 12, 13, 20, 23, 24, 26, 27, 35, 36, 39, 41, 55, 57, 61, 66, 67, 73, 82, 88, 96, 114) has also resulted in an increased desire to establish methods that minimize genetic or epigenetic abnormalities in the oocyte or early embryo that may contribute to diseases later in life.
One area of assisted reproduction in which improvements are being sought is in vitro oocyte maturation (IVM). Current practices for achieving in vivo maturation (VVM) entail a range of costly daily hormone injections, which can be accompanied by a range of side effects (42). The quality of oocytes thus obtained varies considerably both within and between patients. This can limit the overall success of the procedures and the ability to select the highest-quality oocytes for fertilization or the highest-quality embryos for embryo transfer. IVM could increase the supply of high-quality oocytes of more uniform characteristics and would provide new opportunities for restoring female fertility after chemotherapy. IVM has been applied when immature oocytes are recovered along with mature oocytes in patients undergoing ART, and a small number of pregnancies and deliveries have resulted (2, 4, 119). However, the overall efficiency of IVM in the human has been limited. Additionally, the emerging interest in the possibility of therapeutic cloning to produce patient-specific stem cells for clinical treatments may lead to an even greater demand for high-quality oocytes. Improved IVM would also enhance endangered species preservation and the production of valuable nonhuman primate disease models. There is thus considerable interest in improving IVM. Key objectives toward achieving more efficient IVM will be to establish the molecular determinants of oocyte quality, identify specific biological processes or mechanisms that may be disrupted by ARTs, and identify specific modifications to procedures to eliminate these deficiencies.
The rhesus monkey provides an ideal nonhuman primate model for the study of human reproduction and the development of in vitro methods that can be applied to the human. Studies of granulosa cell culture highlighted the importance of a nonhuman primate model for modeling human granulosa cell function (43, 51, 77, 113). Studies of gene expression in rhesus monkey embryos have revealed striking differences compared with rodent embryos (74, 75, 111). Moreover, with a suitable nonhuman primate model it is possible to undertake experimental and functional studies that are not ethically permissible in the human because they entail the use of high-quality oocytes and embryos for laboratory analyses.
Previous studies in the rhesus monkey have shown that high-quality oocytes matured in vivo can be subjected to in vitro fertilization (IVF), with the subsequent development of blastocysts at a rate as high as 61% and subsequent term development (9, 89, 93). IVM, however, remains much less efficient. In unprimed monkeys, oocytes subjected to IVM and IVF form blastocysts at an efficiency of <1% (90). Prior stimulation with follicle-stimulating hormone (FSH) elevates the blastocyst formation rate to 15–40% (92–94). Coculture of maturing oocytes from nonstimulated rhesus monkey females with granulosa cells from FSH-stimulated females enhances blastocyst formation by >10-fold (90).
These observations are intriguing, given the known importance of oocyte-follicle cell interactions in establishing oocyte developmental competence (22, 31, 34, 53). Intimate bidirectional interactions occur between the oocyte and granulosa cells, and between the oocyte, follicle, and endocrine system (5, 7, 15, 25, 29, 30, 40, 56). Disruptions in these interactions within the follicle yield poor-quality oocytes that do not support normal embryogenesis (16, 22, 32, 33, 121). Supplementation of IVM medium with androstenedione elevates blastocyst formation significantly, to as high as 51%, and enhances embryonic genome activation (93). These results indicate that oocyte-follicle cell interactions throughout the maturation period establish full developmental competence, that the absence of these interactions can be partly compensated by hormonal stimulation in vitro, but that developmental competence after IVM is nevertheless restricted compared with VVM.
The implication of these observations is that follicular aspiration of oocytes and IVM preclude essential developmental events that support later embryogenesis. The specific molecular and cellular pathways that are thus affected have not been identified. The elucidation of the affected pathways would enhance our understanding of the nature and timing of follicular events that are critical for primate oogenesis, and would also provide valuable insights that may be applicable to the refinements of IVM and other ART methods. In this study, we have undertaken a global comparison of mRNA expression profiles between in vitro and in vivo matured metaphase II stage rhesus monkey oocytes. These studies reveal a very selective set of differentially expressed mRNAs that are related to a narrow range of cellular functions.
MATERIALS AND METHODS
Adult female rhesus macaques were housed at the California National Primate Research Center. All procedures employed to obtain oocytes were conducted according to recommendations of the Guide for the Care and Use of Laboratory Animals and the Animal Welfare Act and its amendments. All procedures for maintenance and handling of the animals were reviewed and approved in advance by the Institutional Animal Use and Care Administrative Advisory Committee at the University of California at Davis. Details of oocyte isolation and culture are described in previous studies (107–109). Animals were caged individually with a 0600-1800 light cycle and maintained at a temperature of 25–27°C. Animals were allowed to socialize by being housed in pairs during the day from approximately 0800 to 1400. Animals were fed a diet of Purina Monkey Chow and water ad libitum. Seasonal produce, seeds, and cereal were offered as supplements for environmental enrichment. Only females with a history of normal menstrual cycles were selected for study.
Females were observed daily for signs of vaginal bleeding, and the first day of menses was denoted cycle day 1. Recombinant macaque FSH (r-mFSH; National Hormone and Peptide Program, A. F. Parlow, UCLA) was administered intramuscularly (37.5 IU) twice daily, starting on cycle days 1–4 for 7 days total (Fig. 1). To obtain oocytes for IVM, the follicles were aspirated on the morning following day 7 of FSH. Animals were immobilized with ketamine hydrochloride (10 mg/kg), and oocytes were aspirated with an 18-gauge needle coated in 10,000 IU/ml heparin (Elkins-Sinn, Cherry Hill, NJ) by ultrasound-guided oocyte collection as previously described (107). Oocytes were collected into Tyrode lactate (TL)-HEPES medium (37°C) containing 0.1 mg/ml polyvinyl alcohol and 5 ng/ml recombinant human FSH (r-hFSH; Organon). Aspirates were immediately placed in a heated isolette (37°C), where oocytes were retrieved from aspirates with an EM Con filter (Veterinary Concepts, Spring Valley, WI) and placed in fresh TL-HEPES medium. Immature oocytes were recovered at the germinal vesicle stage and cultured for 24 h in 70-μl drops under oil in CMRL medium (11) containing 10% bovine calf serum (Gem Cell, Woodland, CA), hFSH, hLH (0.03 IU/ml Pergonal, Ares-Serono), and 1 μg/ml androstenedione (Steraloids, Newport, RI) [also reported as CMRLb medium (93)]. For VVM oocytes, in addition to the hormonal treatment outlined above for immature oocyte collection, females were injected (im) with recombinant human chorionic gonadotropin (hCG; 1,000 IU Ovidrel; Serono, Rockland, MA) on treatment day 8 for induction of oocyte maturation. Oocytes were aspirated 27–32 h after hCG as described above for immature oocyte collection. The EM Con filters were rinsed with 10 mg/ml hyaluronidase (MP Biomedicals, Solon, OH) and rinsed twice with TL-HEPES before oocytes were rinsed from the filter. Oocytes were placed in 70-μl drops of protein-free hamster embryo culture medium 9 (HECM-9) (70) under oil and incubated at 37°C in a humidified atmosphere of 5% CO2 in air until the metaphase II stage of nuclear maturation was reached. Although many oocytes are not at the metaphase II stage when aspirated, if the time of aspiration is later than 32 h, some ovulations occur and the entire cohort of oocytes cannot be recovered; therefore follicle aspiration at this time is standard for the monkey model (78, 92, 93, 115). For both IVM and VVM oocytes, cumulus cells were carefully removed by using a pulled glass Pasteur pipette and gently moving oocytes through the smallest section of the pipette tip while in a 10 mg/ml hyaluronidase (MP Biomedicals, Solon, OH) solution. After cumulus removal oocytes were rinsed in HECM-9 medium, and only oocytes that were confirmed by microscopy to have one polar body and no remaining cumulus cells were used for analysis.
RNA purification, amplification, and array hybridization.
RNA purification was performed with the PicoPure kit (Arcturus, Mountain View, CA) as described previously (112). Oocytes were lysed two per tube in lysis buffer and stored at −70°C until processing. Oocytes of each kind (IVM or VVM) were combined into pools of eight, each pool representing oocytes from at least three different females. From each pool, half the RNA was reverse transcribed and the cDNA was subjected to two rounds of amplification, consisting of a first round of in vitro transcription followed by random priming and a second round of reverse transcription and in vitro transcription with biotinylated nucleotides to achieve a linear amplification (Affymetrix Small Sample Technical Bulletin, www.affymetrix.com) with minor modifications (initial 5- μl volume for annealing and reverse transcription for 30 min at 42°C followed by 30 min at 45°C). The amplified cRNA samples were fragmented, and 10 μg was hybridized to Affymetrix rhesus macaque genome arrays in the University of Pennsylvania Microarray Facility, which were then processed on fluidic stations and scanned according to the manufacturer's instructions.
Array data analysis.
Array signals data were processed initially with the Microarray Analysis Suite 5.0 (MAS 5.0, Affymetrix) program, with default analysis parameters and global scaling to target a mean equal to 150 signal units. The raw data from all arrays are available online at the Gene Expression Omnibus (GEO) repository (http://www.ncbi.nlm.nih.gov/geo) under series GSE11895. Quality control parameters for the samples were within ranges shown in Supplemental Table S1.1 To minimize false positive signals, only genes called “present” in at least three of four replicates were used for further analysis with all statistical packages. The MAS metric output was loaded into GeneSpring GX v7.3.1 (Agilent Technologies, Foster City, CA) with per chip normalization to the 50th percentile and per gene normalization to the median. The filtered MAS metrics output was loaded into TIGR-MEV v4.0 (87). The Significance Analysis of Microarray [SAM (104)] algorithm was applied to identify genes with significant differences among samples at the 5% false discovery rate (FDR). Genes were further selected as differentially expressed on the basis of t-test (P < 0.05). Fold changes of expression were calculated after SAM analysis. The lists of differentially expressed genes were further evaluated with Ingenuity Pathway Analysis (Ingenuity Systems, Redwood City, CA) and Expression Analysis Systematic Explorer (EASE, version 2.0) programs to analyze gene ontology for overrepresentation (48).
Quantitative mRNA expression analysis.
Independent evaluation and confirmation of gene expression differences were sought with two methods of analysis applied to samples of oocytes obtained independently of those used for array analysis. Where possible we employed a quantitative RT-PCR-based method that involves whole transcriptome amplification as cDNA, followed by dot blotting and quantitative analysis of cDNA probe hybridization (quantitative amplification and dot blotting, QADB). The QADB method is a reliable, sensitive, efficient, and quantitative approach, ideally suited to the task of quantitative gene expression studies in samples of preimplantation mammalian embryos, and offers the advantage that a small number or even a single oocyte per sample can be assayed and the same sample set can be assayed repeatedly for comparisons of gene expression patterns. For QADB analysis, 10 or 7 samples of 1 or 2 oocytes per sample were assayed each for IVM and VVM oocytes, respectively, representing oocytes from at least 3 females for each condition. It should be noted that because the RT-PCR method amplifies the entire mRNA population quantitatively, input in the range of 1 to 2 oocytes per sample does not affect results (14, 50). cDNA probes were obtained by PCR (Supplemental Table S2) from reverse-transcribed rhesus monkey ovary, testis, or liver RNA. Blot preparation, probe preparation, hybridization, and quantitative analyses were performed as described previously (121–124). For some mRNAs, however, array signals indicated that mRNA abundance was likely too low (<500 raw intensity signal units) for efficient detection and quantitation by the QADB method. For these mRNAs, quantitative real-time RT-PCR (qRT-PCR) was employed. Two pools of 10 oocytes each, representing at least 3 females, were obtained for IVM and VVM oocytes. Total cell RNA was isolated with the PicoPure kit (Arcturus) as described above and subjected to qRT-PCR with an ABI Prism 7000 instrument, as recommended by the manufacturer and described previously (112), using primer pairs available from Applied Biosystems (Foster City, CA) and based on the human cDNA sequences. The mRNA abundance of a target gene was normalized to an internal gene control (UBB) for sample-to-sample comparisons, and the relative expression ratio of IVM to VVM groups was obtained by the comparative threshold cycle (CT) method (62). Statistical analysis was performed between IVM and VVM groups with Relative Expression Software Tool (81) for the genes assayed by qRT-PCR method and Student's t-test for those assayed by the QADB method.
Transcription factor binding motif analysis.
To identify potentially shared transcription factor binding motifs among the genes encoding mRNAs elevated in IVM oocytes, the list of official gene symbols was analyzed with oPOSSUM analysis software (http://www.cisreg.ca/cgibin/oPOSSUM/opossum) (45). This software combines methods for detecting transcription factor binding sites (TFBS) documented in the JASPAR database, statistical methods for identifying overrepresented TFBS among a group of genes, and a database of DNA regions that are conserved among moderately divergent organisms (i.e., phylogenetic footprinting). The TFBS thus identified are then evaluated by two statistical measures (45). One is the Z score for overrepresentation, which examines the number of occurrences within the target sequence population, reflects increased prevalence of common sites, and is sensitive to the number of occurrences within the members of the gene set. The second is the one-tailed Fisher exact probability value to test for a nonrandom association between the gene set and a given TFBS. This test is insensitive to the number of occurrences within each gene, and a significant value indicates that a significant fraction of the members of the gene list contain the site. In the study by Ho Sui et al. (45), TFBS displaying a combination of Z > 10 and P < 0.01 yielded a false positive rate of 15% for a set of 15 genes, whereas using either score alone yielded much greater rates of false positives. For this analysis, we retrieved data for genes passing the threshold of Z ≥ 5.0 and Fisher P < 0.05 for further examination.
Overview of array results.
We obtained four high-quality array data sets for each kind of oocyte. Among the eight arrays, percent present call ranged from 33.61 to 40.35, corresponding to ∼1.75 × 104–2.09 × 104 probe sets, which is well within the acceptable range (120), and an overall presence call of 36.5% (1.9 × 104 of 5.2 × 104 probe sets) (Supplemental Table S1). The other quality control parameters for all the samples were within the following ranges: scale factor 0.873–1.59 (accepted range: 0.5–5.0) and background 43.6–65.28 (accepted range: 20–100) (Supplemental Table S1). The absence of contaminating cumulus cells is indicated by the absence of detected signals for follicle-stimulating hormone receptor (FSHR) mRNA.
We used K-means hierarchical clustering (HCL) to ascertain the overall degree of similarity between the transcriptomes of the IVM and VVM oocytes (Fig. 2). The replicates of IVM and VVM oocytes readily clustered appropriately, with no apparent outliers. This clustering pattern indicates a high degree of reproducibility and small biological variability between samples of each kind of oocyte, and also indicates sufficient difference between the two kinds of oocytes to allow them to cluster separately.
Differentially expressed gene sets.
Comparisons between the two kinds of oocytes revealed a strikingly small number of differentially expressed genes. We applied the criteria of FDR of <0.05 and significance value of P < 0.05 for t-test, with no threshold filtering for identifying these genes. This yielded two sets of expression differences, one set of 7 genes expressed more highly in VVM oocytes (Table 1) and another set of 52 genes expressed more highly in IVM oocytes (Table 2). This represents only a very small fraction, ∼0.31%, of the 1.9 × 104 total probe sets receiving a positive presence call and employed for the SAM analysis, or ∼0.39% after accounting for redundant probe sets for individual genes. Among the seven genes expressed more highly in VVM oocytes, the fold differences ranged from 1.91- to >60-fold, with average expression values in VVM oocytes ranging from a relatively low value of 56 units to a moderate value of 528 units. For the 52 genes expressed more highly in VVM oocytes, fold differences ranged from 1.70- to >17-fold. The average signals for IVM oocytes were relatively low (<100 units) for some genes but ranged to much higher values for many others.
Quantitative analysis by QADB and qRT-PCR.
We examined the expression of 19 of the 59 differentially expressed mRNAs identified above, using either of two alternate methods (Fig. 3). Nine were examined with the QADB method, with a large number of biological replicates, 7 VVM and 10 IVM, and employing mRNA-specific rhesus monkey cDNAs as probes. Of these nine mRNAs, eight displayed significant differences in expression in the same direction as indicated by the array data. This includes both of the annotated, known mRNAs judged to be higher in VVM oocytes in the array data. The SPINK2 mRNA also appeared to be differentially expressed; however, the difference did not reach statistical significance with the QADB method. An additional 10 mRNAs were judged on the basis of raw array signals to be of low abundance. These mRNAs were quantified with the real-time qRT-PCR assay. All 10 of these mRNAs were confirmed to be expressed significantly higher in IVM oocytes compared with VVM oocytes.
Biological functions of differentially expressed genes.
Among the seven genes with higher expression in VVM oocytes, three are associated with known functions. One is aldo-keto reductase family 1 member 1 (AKR1C1), an enzyme in the cytoplasm that catalyzes the conversion of aldehydes and ketones to alcohols with NADH and/or NADPH. Another (ATP6V0A4), a subunit of the H+-ATPase pump, responsible for proton extrusion and acidification of intracellular compartments and a variety of intracellular processes (e.g., protein sorting, zymogen activation, receptor-mediated endocytosis), the expression of which is most commonly reported for the kidney and mutations of which are associated with defects in renal function and deafness (99). The third was cholesterol 25-hydroxylase (CH25H), which converts cholesterol to oxysterol, a potent regulator of lipid metabolism (64).
Of the genes overexpressed in IVM oocytes 42 are annotated for function or expression pattern, and these are associated with a range of cellular functions. To explore biological functions represented by these genes, we applied EASE analysis, which identifies Gene Ontology categories that are overrepresented among a gene list relative to the background population (Table 3). This analysis revealed steroid biosynthesis and metabolism, cell-cell signaling, ion transport and homeostasis, cell growth, and stress response as the major functional categories that are altered in IVM compared with VVM oocytes. Smaller functional categories not identified as overrepresented but nevertheless encompassed among the affected genes included ion or macromolecular transport (KCNK3, LDLR, NPTX2, SLC12A5, SLC1A5, SLC6A6), cellular adhesion (CD83, PGAP1), RNA binding and control of translation (CUGBP2, RG9MTD3, RPS6KA5, TDRD1, SND1), cytoskeleton/cell architecture (MYLIP), transcription (SND1), and proteolysis (MDM4, USP54).
To gain further insight into the relevance of the genes that were more highly expressed in IVM oocytes to specific biological processes within cells, we employed the Ingenuity Pathway Analysis (IPA) program. The IPA analysis seeks to identify biological pathways or networks of interacting gene products. The network encompassing the largest number of affected genes, 21 of the 42 annotated, differentially expressed mRNAs (Fig. 4), displays interactions related to cell growth, maintenance, and cell-cell signaling. SOCS3 occupies a key location in the network, interfacing with components of JAK/STAT pathway signaling. INHBA likewise occupies a key point in the pathway, coupling cellular signals to a variety of other mediators such as STAR and CYP19A1 that are also differentially expressed. Additional key regulators included in the network are insulin, NFKB, p38/MAPK, CREB, and PDGFB, which in turn impact on other differentially expressed genes. Overall, this network indicates that numerous components of a large network affecting cell growth and signaling may be altered in IVM oocytes. The second network identified by IPA (Fig. 5) encompasses a large number of genes related to steroid metabolism and endocrine and paracrine control of cell growth and differentiation. Key points of regulation in this network include a number of growth factors (EGF, IGF1, BDNF) and hormones or their receptors (FSHR, progesterone) that may not be expressed in the oocyte or are not altered in expression in the oocyte, but which may signify altered interactions with somatic cells. The components of the network that are upregulated in IVM oocytes are predominantly distal targets of these factors, indicating that the phenotypic response of the oocyte to endocrine or paracrine factors may be altered. The two networks have several features in common, such as steroid metabolism and biosynthesis, and pathways related to growth factor and hormonal stimulation. Thus the predominant difference between IVM and VVM oocytes appears to be changes related to cell-cell interactions that potentially affect cell growth, differentiation, and maintenance.
Transcription factor binding motif analysis.
To determine whether the genes showing elevated mRNA expression in IVM oocytes might be coordinately regulated at the level of transcription, we analyzed those annotated genes with approved gene symbols (n = 44), using oPOSSUM software. The analysis retrieved and analyzed available data for 36 genes. This yielded a set of five transcription factors that passed the initial filter of Z ≥ 5.0 and Fisher P < 0.05 (Table 4). Four of these were quite prevalent, with binding sites being detected in 25 or more of the 36 genes analyzed, and two of these (NKX2-5 and NOBOX) passed the more stringent cutoff of Z ≥ 10.0. The genes containing the relevant binding sites are indicated in Table 4.
3′ Untranslated region analysis.
mRNA content and stability in the oocyte are highly subject to posttranscriptional regulation, with stability often being conferred upon mRNAs by the presence of a cytoplasmic polyadenylation element [CPE (69, 79, 80, 97)]. The CPE is responsible for binding maternal mRNAs into a messenger ribonucleoprotein complex with Maskin and CPE binding protein 1 (CPEB1) for storage, and it subsequently regulates translational recruitment via a process that includes Maskin and CPEB phosphorylation, mRNA polyadenylation, and binding of translational initiation factors to the 5′ mRNA cap (8, 17, 37, 46, 86). We therefore examined the 3′ untranslated regions (UTRs) of the mRNAs that are overexpressed in IVM oocytes. The 3′UTR exon sequences were retrieved for 44 of the mRNAs from the sequence database with the tools available at http://genome.ucsc.edu/cgi-bin/hgGateway?db=hg10. These sequences were then examined for the presence of any of 14 published sequences that can function as CPEs in either mouse or frog (Supplemental Table S3; Refs. 79, 97). This analysis (Table 5) revealed several interesting features. First, a number of the affected mRNAs possess quite long 3'UTRs, with an average length of 1,213 nt, a median value of 962 nt, and a range of 111–8,227 nt. Second, 36 (82%) of the mRNAs contained one or more putative CPEs. Of these, six (CUGBP2, MYLIP, PGAP1, PLAGL1, RPS6KA5, STC1, and TMEM26) contained five or more putative CPEs and five (LDLR, MDM4, MPPE1, SOCS3, USP54) contained three or four CPEs. Calculating the number of CPEs per nucleotide of 3′UTR reveals an overall average of 0.00246 and an overall median value of 0.00183. Eleven of the mRNAs (CUGBP2, IL6ST, MDM4, MT2A, MYLIP, OSMR, PGAP1, PLAGL1, RPS6KA5, TMTM26, and USP54) exceeded twice the median value, and four (IL6ST, MDM4, MT2A, and RPS6KA5) exceeded the average value. A regression analysis comparing the array fold change to the length of 3′UTR, the number of CPEs, or the number of CPEs/length of 3′UTR revealed no significant correlation.
Expression of imprinted genes.
The discovery that PLAGL1, a maternally imprinted gene, is overexpressed in IVM oocytes prompted us to investigate whether other imprinted genes might be overexpressed in IVM oocytes, indicative of a potential broader disruption in imprinted gene regulation. We extracted from the array data the intensity values for maternally and paternally imprinted genes (Supplemental Tables S4 and S5, assembled from http://www.geneimprint.com/site/genes-by-species.Homo ±sapiens). We found that among the maternally imprinted genes two genes known to be imprinted (PLAGL1, MEST) and two predicted as potentially imprinted (C20ORF82, C6ORF117) were overexpressed in IVM oocytes, while among the paternally imprinted genes one known imprinted gene (H19) and one suspected imprinted gene (ANKRD11) were overexpressed in IVM oocytes.
This global comparison of transcriptomes of VVM and IVM oocytes is the first such study to be undertaken in a nonhuman primate model. The study has yielded several novel insights pertaining to the molecular basis for differences in oocyte quality. First, the analysis has revealed that, despite significant differences in developmental potential, only a relatively small number of significant differences exist between the transcriptomes of VVM oocytes and IVM oocytes. Conceivably, a difference in oocyte quality could be the result of many differences in mRNA expression, accumulation, and stability. Alternatively, oocyte quality could be determined by more specific changes in oogenesis affecting a narrow array of transcripts. Our results clearly argue for the latter of these two scenarios. Second, the results indicate that the removal of the oocyte from the follicle for IVM may prematurely terminate important cell-cell interactions that foster the emergence of full oocyte developmental competence.
The close similarity between transcriptomes of VVM and IVM oocytes contrasts with gene expression differences previously reported between IVM and VVM rhesus monkey oocytes and IVM and VVM human oocytes (52, 121). This difference most likely reflects differences in the IVM conditions employed in different studies. In previous rhesus monkey studies (93, 121) IVM was achieved with CMRLa medium, which employs 20% instead of 10% serum and lacks androstenedione. The developmental competence of IVM oocytes obtained with CMRLb medium as employed here is much greater and much more similar to that of VVM oocytes than that observed with CMRLa medium (93). Hence, it is not surprising that the transcriptomes of IVM oocytes obtained with CMRLb medium would likewise be highly similar to those of VVM oocytes. The androstenedione in CMRLb medium may provide an important molecular precursor for steroidogenesis (71). It was suggested that the inferior development of embryos produced from IVM oocytes obtained with CMRLa medium could reflect a failure either to acquire or to maintain ooplasmic components (e.g., maternal mRNAs) that support later embryogenesis, and that the beneficial effects of androstenedione could be related to the preservation of such ooplasmic components. Our results are consistent with the explanation that a more faithful maintenance of maternal mRNAs that support later embryogenesis is indeed realized with the CMRLb IVM system. In the case of human array studies, IVM was applied to oocytes that failed to respond to in vivo hCG stimulation, and thus may have been of inherently inferior quality. These studies likewise employed simpler culture media that, like CMRLa, may not optimally support retention of key ooplasmic components needed for later embryogenesis, with the result that >2,000 transcripts differed between metaphase II stage IVM and VVM oocytes. Overall, our results indicate that IVM may result in the production of oocytes with dramatically altered transcriptomes that are incompatible with efficient embryogenesis, but that alterations and supplements in the IVM culture system can yield oocytes that are molecularly as well as developmentally highly similar to VVM oocytes.
IVM oocytes obtained even with the CMRLb system display significant reductions in developmental potential compared with VVM oocytes (93). Our analysis reveals alterations in several functional categories, the largest ones being steroid biosynthesis and metabolism, cell-cell interactions (signaling and adhesion), ion transport and homeostasis, cell growth, stress response, and RNA binding and translation. The IPA analysis provides additional insight into these results. The first network (Fig. 4) highlights key roles for INHBA and SOCS3 in regulating cell growth and signaling. These molecules mediate a range of effects via many other factors that are expressed in the oocyte but not upregulated and at least three exogenous factors (IL1, insulin, and LDL). A majority of the overexpressed genes lies at the periphery of the network, suggesting that these genes may comprise downstream mediators of cellular processes controlled by the signaling pathways encompassing SOCS3 and INHBA. The second network (Fig. 5) incorporates multiple hormones and growth factors as signals to regulate a range of processes, including stress and hormone responses via PLAGL1, NPTX2, KCNK3, STC1, AKR1C1, SLC1A5, and MDM4. As with the first network (Fig. 4), the majority of overexpressed mRNAs correspond to products that occupy peripheral locations within this network, surrounding a core of unaltered products, and other products that are not expressed in the oocyte, but many of which are expressed in cumulus cells. This indicates that many of the overexpressed mRNAs correspond to products that are regulated by interactions between the oocyte and somatic cells. Moreover, it appears that pathways involving signaling via progesterone, FSH, EGF, and aldosterone, as well as insulin and PDGF (Fig. 4), may be altered with IVM.
Of the 42 annotated genes, 11 (CD5L, CD83, EFNA1, GP43, IL26, IL6ST, INHBA, NPTX2, OSMR, PGAP1, STAR) participate in cell-cell interactions and signal transduction, suggesting that the reciprocal dialogue between oocyte and follicle cells may be disrupted. In particular, STAR regulates responses to steroid hormones, and two other genes (CD83 and PGAP1) regulate cell adhesion, both important components of oocyte and follicle development. STC1 expression by thecal cells negatively regulates granulosa cell differentiation (65), and its expression in the rhesus monkey oocyte indicates that oocyte-derived STC1 may affect differentiation within the somatic compartment (i.e., cumulus and granulosa cells). The increased STC1 expression in IVM oocytes compared with VVM oocytes may reflect a disruption in the cell-cell interactions and final development of both the oocyte and somatic cells. STC1 is regulated by glucorticoids in the male (60), creating a potential interaction with the steroid biogenesis pathway. SOCS3 negatively regulates cellular differentiation (18, 101, 102), and INHBA helps to provide for oocyte-directed control of cumulus cell expansion (28). Overexpression of these genes in IVM oocytes may be indicative of a temporal or quantitative disruption in key oocyte-cumulus cell interactions.
The overexpression of other mRNAs may reflect a premature stage of oocyte development. TDRD1 encodes a key component of ribonucleoprotein nuage particles for maternal mRNA storage (19), CUGBP2 encodes an RNA binding protein that suppresses translation and stabilizes mRNA (76, 85, 100), and SND1 (aka TDRD11) encodes a component of the RNA-induced silencing complex (RISC) involved in RNA degradation (103). The increased expression of these mRNAs could reflect a delay in final oocyte development or incomplete establishment of mechanisms for regulating maternal mRNA stability and content. We reported (93) that a number of maternal mRNAs appear to be regulated incorrectly in oocytes and embryos resulting from IVM in CMRLa medium. The premature nuclear maturation described for oocytes in IVM CMRLa medium (76) is also present in IVM oocytes from CMRLb medium (referred to as M1A in Ref. 78), despite the present finding of aberrant expression for many fewer genes. Thus the previous suggestion by Schramm and Bavister (91) that oocyte developmental competence is acquired progressively after the initial maturation stimulus seems to be supported at the molecular level by these data. However, the data indicate that the process may continue for longer than previously anticipated, with key events occurring throughout the period leading up to ovulation. Consequently, aspiration of the oocyte-cumulus cell complex a day before normal ovulation may interrupt these final events, compromising oocyte quality.
We observed overrepresentation and a high prevalence of sites for NKX2-5 and NOBOX among the affected genes. NOBOX is a key regulator of oogenesis (49, 84), and the increased expression of some NOBOX target genes with IVM is consistent with the hypothesis that the final development of these oocytes is incomplete, with prolonged retention of mRNAs indicative of earlier stages of oogenesis. Of further potential interest is that binding sites for multiple members of the Forkhead (FOX) or winged helix DNA binding protein family were retrieved, which regulate a range of homeostatic and developmental functions in many tissues.
Genes for steroidogenesis and steroid metabolism are also overexpressed in IVM oocytes. CYP19A1 is the cytochrome P-450 enzyme aromatase that catalyzes the formation of C18 estrogens from C19 androgens and has a critical role in steroidogenesis in the ovary. CYP3A5 is part of a larger family of genes (CYP1, 2, and 3) involved in steroid metabolism and is expressed in the human ovary (10, 58). The addition of androstenedione to the IVM culture medium may be indirectly responsible for the high levels of expression of both of these genes since aromatase converts it to estrone and the CYP3 family can metabolize both estrogens and androgens. However, the transcriptional regulation of aromatase is complex (72) and may involve several hormonal and paracrine factors that are present during IVM culture.
One mechanism by which oocyte mRNAs may be altered with IVM is via an effect on maternal mRNA translational regulation. Our analysis of the 3′UTRs of the mRNAs overexpressed in IVM oocytes revealed that >80% contained putative CPE motifs, a property that is shared with actively translating maternal mRNAs in mouse metaphase II stage oocytes (83). A number of these mRNAs possess quite long 3′UTRs, and many of them have multiple CPE elements. We failed to observe a direct relationship between fold difference in expression and either length of the 3′UTR or the number of CPEs present. Thus there does not appear to be a specific characteristic with respect to CPE elements to account for the increased expression of this particular group of mRNAs. Our previous studies of rhesus monkey oocytes obtained by IVM in CMRLa medium (121) revealed that the regulation of the maternal mRNA population may be aberrant in these oocytes, particularly without prior FSH stimulation. The small number of affected mRNAs observed here indicates that many fewer maternal mRNAs may be aberrantly regulated in this way with the CMRLb medium. Thus a defect in maternal mRNA regulation is apparently not an obligate consequence of IVM per se, but rather may accompany suboptimum IVM conditions.
The altered expression of many of the indicated mRNAs could affect embryogenesis. The overexpression of potassium ion channel mRNAs (KCNK3, SLC12A5) and amino acid transporter mRNAs (SLC1A5, SLC6A6) and reduced expression of the mRNA encoding the hydrogen ion channel ATP6V0A4 could severely disrupt intracellular pH regulation and thereby a myriad of cellular processes. The increased expression of mRNAs encoding growth factors and hormones (IL26, INHBA, NPTX2, STC1), growth factor receptors (OSMR), and modulators of cellular responses to such factors (HSD11B2, IL6ST, SOCS3, STAR) could alter the in vitro response of oocytes and embryos to serum components. STC1 overexpression could disrupt calcium and phosphate homeostasis and intracellular pH (118). Ionic adjustments to the culture medium may therefore enhance development of embryos produced with IVM oocytes. Interestingly, transgenic mice overexpressing human STC1 display reduced litter sizes, indicating possible effects on ovulation or early embryo development rates (38). The LDLR gene is a candidate for a locus controlling metabolic syndrome susceptibility (95, 117). The overexpression of other mRNAs related to various forms of stress (MDM4, CD83, CYPA43, MT2A, PLAGL1) could affect embryo response to the culture environment, particularly via p53-mediated pathways (123).
The discovery of PLAGL1 as an overexpressed mRNA in IVM oocytes is notable within the context of genomic imprinting. PLAGL1 was identified in a screen for imprinted human genes involving comparisons between androgenetic hydatidiform mole DNA and parthenogenetic DNA obtained from a chimeric patient (54). PLAGL1 was subsequently shown to be expressed solely from the paternal allele in humans (54). An imprinting control region exists that is methylated in the oocyte but not the sperm (6). PLAGL1 is a member of a coregulated group of imprinted genes, and its expression may affect the expression of other imprinted genes (110). Its expression in the oocyte is unexpected because of maternal imprinting. Indeed, its expression in VVM oocytes was near background level, while its expression in IVM oocytes was significantly elevated. This could result from aberrant activation of the biallelic promoter P2 (105), delayed degradation of maternal transcripts in IVM oocytes, or effects of IVM on the imprints themselves, including interruption of the DNA methylation process or loss of imprinting. A number of imprinted genes undergo specific methylation at imprinting control regions during the growth phase in both mice and humans and can occur quite late during oogenesis (63). This again suggests that removal of the oocyte from the in vivo environment prematurely could disrupt essential events such as genomic imprinting. It has also been suggested that assisted reproduction methods, including superovulation, could alter DNA methylation of imprinted genes in human oocytes (36, 61). We also found that MEST1 and two additional potentially maternally imprinted genes were overexpressed in IVM oocytes. Conversely, only one known and one potentially paternally imprinted gene were overexpressed. These observations raise concern about possible disruptions in epigenetic information and imprinted gene regulation accompanying IVM, particularly for maternally imprinted genes. Because PLAGL1 is a putative tumor suppressor, its overexpression in the early embryo could retard the cell cycle and enhance apoptotic events in the oocyte or early embryo (1, 98). The overexpression of MEST is less likely to be problematic for the early embryo, because MEST gene deficiency is associated with mild growth restriction and defects in maternal behavioral response to pups and placentophagia (59).
In summary, this study provides the first comparison of transcriptomes of IVM and VVM oocytes in a nonhuman primate species. The results show that a limited number of mRNAs are differentially expressed. The two kinds of oocytes differ in that the IVM oocytes lack 1 day of development in vivo as well as the in vivo response to hCG. This difference in cellular history, while seemingly small, is sufficient to generate transcriptomes that segregate from each other on K-means HCL analysis and that differ from one another in a very select array of transcripts. The small number of observed changes could arise through either effects on the pace of individual maternal mRNA translation and degradation or an effect at the level of gene transcription just before germinal vesicle breakdown. We propose that the identities of the affected genes combined with the statistical separation on HCL analysis reflect an interruption or truncation in the normal series of oocyte-somatic cell interactions that produce a developmentally competent oocyte, with essential interactions continuing throughout the period leading up to ovulation. Removal of the cumulus-oocyte complexes from the follicular environment before these interactions are complete may terminate key developmental processes in the oocyte, with additional changes that compromise early embryogenesis. Moreover, the removal of the cumulus-oocyte complexes from the follicle and exposure to an in vitro environment may disrupt genomic imprinting information, potentially causing overexpression of PLAGL1 and altering the regulation of other imprinted genes. The data presented here thus provide novel insight into the nature of oocyte-follicle cell interactions, the potential molecular and cellular consequences of altering these interactions, and the basis for compromised developmental competence following IVM procedures in a nonhuman primate model. These results also raise concerns about applying IVM clinically without addressing such developmental defects, but indicate that these deficiencies may be overcome by further improvement in IVM culture systems.
This work was supported by National Center for Research Resources Grants RR-15253 (K. E. Latham), RR-000169 (C. A. VandeVoort), and RR-13439 (C. A. VandeVoort).
↵1 The online version of this article contains supplemental material.
Address for reprint requests and other correspondence: K. E. Latham, Fels Institute for Cancer Research & Molecular Biology, Temple Univ. School of Medicine, 3307 North Broad St., Philadelphia, PA 19140 (e-mail:).
The costs of publication of this article were defrayed in part by the payment of page charges. The article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
- Copyright © 2008 the American Physiological Society