Transcriptomic integrity of human oocytes used in ARTs: technical and intrinsic factor effects

Abstract

Background

Millions of children have been born throughout the world thanks to ARTs, the harmlessness of which has not yet been fully demonstrated. For years, efforts to evaluate the specific effects of ART have focused on the embryo; however, it is the oocyte quality that mainly dictates first and foremost the developmental potential of the future embryo. Ovarian stimulation, cryopreservation, and IVM are sometimes necessary steps to obtain a mature oocyte, but they could alter the appropriate expression of the oocyte genome. Additionally, it is likely that female infertility, environmental factors, and lifestyle have a significant influence on oocyte transcriptomic quality, which may interfere with the outcome of an ART attempt.

Objective and rationale

The objective of this review is to identify transcriptomic changes in the human oocyte caused by interventions specific to ART but also intrinsic factors such as age, reproductive health issues, and lifestyle. We also provide recommendations for future good practices to be conducted when attempting ART.

Search methods

An in-depth literature search was performed on PubMed to identify studies assessing the human oocyte transcriptome following ART interventions, or in the context of maternal aging, suboptimal lifestyle, or reproductive health issues.

Outcomes

ART success is susceptible to external factors, maternal aging, lifestyle factors (smoking, BMI), and infertility due to endometriosis or polycystic ovary syndrome. Indeed, all of these are likely to increase oxidative stress and alter mitochondrial processes in the foreground. Concerning ART techniques themselves, there is evidence that different ovarian stimulation regimens shape the oocyte transcriptome. The perturbation of processes related to the mitochondrion, oxidative phosphorylation, and metabolism is observed with IVM. Cryopreservation might dysregulate genes belonging to transcriptional regulation, ubiquitination, cell cycle, and oocyte growth pathways. For other ART laboratory factors such as temperature, oxygen tension, air pollution, and light, the evidence remains scarce. Focusing on genes involved in chromatin-based processes such as DNA methylation, heterochromatin modulation, histone modification, and chromatin remodeling complexes, but also genomic imprinting, we observed systematic dysregulation of such genes either after ART intervention or lifestyle exposure, as well as due to internal factors such as maternal aging and reproductive diseases. Alteration in the expression of such epigenetic regulators may be a common mechanism linked to adverse oocyte environments, explaining global transcriptomic modifications.

Wider implications

Many IVF factors and additional external factors have the potential to impair oocyte transcriptomic integrity, which might not be innocuous for the developing embryo. Fortunately, it is likely that such dysregulations can be minimized by adapting ART protocols or reducing adverse exposure.

Graphical Abstract

The multiplicity of assisted reproductive interventions has the potential to modify the oocyte or follicle transcriptome, which has possibly been previously modulated by patient characteristics and their environmental exposures.

assisted reproduductive technologies, human oocyte, IVF, RNA-seq, transcriptome

Introduction

ARTs have allowed the birth of millions of children worldwide and the proportion of couples undergoing infertility treatments is in continuing expansion (de Geyter et al., 2020; Wyns et al., 2020, 2021). ARTs have been recognized as safe, but >40 years after the first baby born via IVF (Steptoe and Edwards, 1978) numerous studies have reported a higher incidence of birth defects and congenital disorders including imprinting disorders in the population conceived ‘in vitro’ (Davies et al., 2012; Vermeiden and Bernardus, 2013). Over recent years, efforts have been made to improve the safety of such techniques after identification of several factors that may be detrimental for gametes and embryos.

There is no doubt that during an IVF cycle, oocytes are particularly vulnerable to exogenous exposure, notably at the MII (metaphase II) stage when there is transcriptional quiescence. Transcriptional activity of the oocyte dictates the developmental potential of the future embryo, and small perturbations of the oocyte microenvironment have the capacity to impair oocyte quality and developmental competence, with potential long-lasting effects (Mermillod et al., 2008; Krisher, 2013). Assessing the integrity of the oocyte transcriptome has been proposed as a tool to evaluate oocyte quality following several ART exogenous exposures including ovarian stimulation, IVM and culture conditions, cryopreservation, oocyte denudation, pipetting, and other in vitro external factors such as thermal stress and plastic pollution. Additionally, it is likely that environmental factors and lifestyle have a significant influence on oocyte transcriptomic quality that may interfere with the success of an IVF/ICSI protocol (Krisher, 2013; Carré et al., 2017).

For years, efforts have focused mainly on the optimization of in vitro culture conditions of the embryos, known to be crucial for the outcomes (Wale and Gardner, 2016). However, the impact of environment for oocyte quality should not be underestimated; the quality of the oocyte quality at fertilization is the fundamental checkpoint to ensure correct early embryo development. Exposing the oocyte to stressful environment forces the oocyte to adaptations, and the acquisition of necessary transcriptional material to support embryo development may be compromised. Intrinsically, the embryo expresses pro-death pathways, which are tightly repressed by autocrine trophic signals (such as growth factors) which activate cAMP response element-binding protein (CREB) and AKT signaling, playing a pro-survival role (O’Neill et al., 2012). An inherited defective maternal-effect machinery, as hypothesized by Bertoldo et al. (2015) based on evidence from early mouse embryos, would deprive the embryo of those trophic signals, and switch the embryonic transcriptome from pro-survival to pro-apoptotic settings via accumulation of TRP53 protein (Bertoldo et al., 2015). The purpose of this review is to identify and interpret changes in the transcriptome or expression of targeted genes in the human oocyte caused by interventions specific to ART but also external factors such as maternal age, reproductive diseases and lifestyle. The three main objectives of our study were: (i) to determine whether there is concordance between the findings of the studies carried out in humans according to each disruptive factor; (ii) to identify whether common transcriptional dysregulations of molecular pathways exist between the different disruptive factors; and (iii) to identify whether some regions/genes are particularly sensitive to the different disruptive factors.

Methods

The focus of this review is human oocyte gene expression in the context of ART technical and intrinsic factors; thus, an in-depth literature search was performed on PubMed to identify studies relevant to this subject, without publication date limit. The factors considered included: endometriosis, polycystic ovary syndrome (PCOS), diminished ovarian reserve (DOR), maternal age, smoking, BMI, pollutants, ovarian stimulation, IVM, and cryopreservation. Studies evaluating both the full (via microarrays, RNA-seq or scRNA-seq) or partial transcriptome (using RT-qPCR) or only a small number of genes were considered in this review. Due to the narrative nature of this review, no formal quality of evidence assessment of studies was conducted. For all studies, we recorded information on study design, the number, and the list of differentially expressed genes (DEGs), and information of biological pathways altered in the context considered. In the absence of studies directly on human oocytes for some factors, studies on somatic cells surrounding the oocytes (i.e. mural granulosa cells (MGCs) or cumulus cells (CCs)) were also considered. In total, 50 studies were recorded evaluating specific ART interventions or intrinsic factor effects on the human oocyte transcriptome or that of its surrounding cells. In some cases, studies on animal oocytes were also discussed in the absence of studies directly on human oocytes.

To facilitate and justify the comparison of studies, we gathered lists of DEGs retrieved in each individual study. No custom significance threshold was applied for the selection of DEGs; all DEGs have been selected based on the criteria applied in each study. Then, we separated DEGs into up- and down-regulated genes according to the two groups compared (reference versus other condition tested) and subjected them to gene symbol update with the UpdateSymbolList function from Seurat package for harmonization (Hao et al., 2021). This function finds current gene symbol of old or alias gene symbols from the HGNC database. All lists of DEGs were intersected with each other to find overlaps in affected genes between studies, and Venn diagrams were used for visualization. We also sought to cross-reference these genes with a list of genes involved in chromatin-based processes such as DNA methylation, heterochromatin modulators, histone modifiers, and remodeling complexes.

Patients’ characteristics

The interaction between patients’ characteristics and ART remains a grey area, but if clarified, this could lead to improve patient management in the future. The fact that oocytes have lower competence with specific patients’ conditions might be seen at the transcriptomic level and that could help scientists to develop new personalized ART protocols.

Infertility

Endometriosis

Endometriosis is a common disorder in the general population (∼1 in 10 women) (Eisenberg et al., 2018) and it is estimated that 35–50% of infertile women are affected by this disease (Meuleman et al., 2009; Bulletti et al., 2010). Several studies have shown that endometriosis does not reduce endometrial receptivity and have suggested that decreased oocyte quality could be the reason why patients have reduced implantation rates (Sung et al., 1997; Garcia-Velasco et al., 2015; Miravet-Valenciano et al., 2017; Tan et al., 2021). The transcriptomic profile of MII oocytes from patients with endometriosis has recently been compared with healthy controls to understand the molecular mechanisms altering oocyte quality (Ferrero et al., 2019) (Supplementary Table S1). Using scRNA-seq on 16 MII oocytes from 5 healthy oocyte donors and 16 MII oocytes from 7 patients with ovarian endometrioma, the authors identified 520 DEGs (the majority up-regulated due to endometriosis), which were enriched for processes such as steroid metabolism, response to oxidative stress, cell growth regulation, mitochondrial function, methylation, and angiogenesis functions. Interestingly, robust sub-group analysis when oocytes came from an affected or unaffected ovary from the same patient showed no difference between the groups, suggesting that endometriosis decreases globally oocyte quality independently of the presence of an endometrioma. A study on RNA-seq in pools of cumulus cells from patients with endometrioma reached the same conclusions (da Luz et al., 2021), while another study with a similar protocol found different pathways and defective functions such as MAPK and Wnt signaling, apoptosis, and cell junctions (Shi et al., 2022). The small number of samples (9 and 5, respectively) could explain the heterogeneity of the results. It remains to be determined to what extent these transcriptomic modifications could compromise oocyte fertility and have a negative influence on ART outcomes for patients with endometriosis (Senapati et al., 2016). Elevated levels of reactive oxygen species (ROS) in follicular and peritoneal fluids may explain a substantial number of these transcriptomic alterations, and cumulus-oophorus rinsing rapidly after follicular puncture may prevent them. Additionally, half of the causes of endometriosis can be attributed to genetic factors (Saha et al., 2015), and polymorphisms may impact the expression of some oocyte transcripts, but this is unclear to date.

Polycystic ovary syndrome

A second reproductive disorder that has been comprehensively studied at the molecular level is PCOS. The frequency of PCOS is also high in the infertile population with great variability depending on ethnicity and world regions (5–20%) (Lizneva et al., 2016; Wolf et al., 2018). It is an endocrine and metabolic disorder characterized by hyperandrogenism, ovulatory dysfunction, and polycystic ovaries. Of concern, an altered endocrine microenvironment could alter the quality of the oocyte and compromise its ability to develop.

Single-cell RNA-seq and microarray studies in small cohorts of germinal vesicle (GV) and MII oocytes (n = 6, 11, and 3 PCOS versus n = 6, 9, and 6 controls) demonstrated that oocytes from patients with PCOS show transcriptomic abnormalities in genes related to meiosis (such as genes encoding microtubules) (Wood et al., 2007; Liu et al., 2016; Li et al., 2021). In addition, a recurrent observation in PCOS oocyte studies is the perturbation of mitochondrial processes, which would cause impaired energy metabolism and oxidative phosphorylation (Qi et al., 2020; Li et al., 2021). The suggested mechanism in PCOS is that mitochondrial functions are prematurely activated at the GV stage instead of the MII stage, leading to the production of ROS with detrimental effects for oocyte competence (Qi et al., 2020). Abnormal expression of 826 transposable elements was also reported in oocytes from women with PCOS (Li et al., 2021). As the authors noticed, some of these elements may be involved in complex transcriptional regulation processes, negatively affecting the oocyte transcriptome, or could be biomarkers of aberrant expression of important genes for oocyte developmental competence, such as tubulin genes (Li et al., 2021).

An adaptation of stimulation and maturation protocols (using mild controlled ovarian stimulation or medical IVM) may assist PCOS patients in obtaining oocytes with higher competence while also decreasing the risk of OHSS.

Women’s age

Over the last few decades, maternal age at birth has significantly increased worldwide (Attali and Yogev, 2021) and now it is common for women to rely on ART after the age of 35. Advanced maternal age is associated with a higher proportion of poor-competence oocytes and difficulty in achieving a successful IVF outcome (Cimadomo et al., 2018). Studies agree that the main biological processes involved in reproductive aging are cell cycle, mitosis and meiosis, oxidative stress, mitochondrial functions, and RNA metabolism (Steuerwald et al., 2007; Grøndahl et al., 2010; Reyes et al., 2017; Zhang et al., 2020b; Llonch et al., 2021; Ntostis et al., 2021; Yuan et al., 2021) (Supplementary Table S1). Specifically, aging is associated with an increased production of ROS, a decrease in the expression of oxidation-protective genes, and a gradual alteration of mitochondrial activity (Llonch et al., 2021). The completion of oocyte maturation is energy intensive, thus relying on a minimum level of mitochondrial activity to ensure basal energy production. Reduced ATP production and subsequent reduced protein metabolism could negatively affect oocyte maturation if the basal activity is not ensured (Schatten et al., 2014).

Interestingly, GV and MII oocytes do not seem to be equally affected by maternal age revealing that maturation differentially mediates the transcriptomic effects of maternal aging. Llonch et al. (studying continuous changes from 18 to 43 years old), Ntostis et al. and Reyes et al. (studying <30 versus >40 years old) showed with scRNA-seq that MII oocytes display more age-related transcriptional abnormalities than oocytes at the GV stage (Reyes et al., 2017; Llonch et al., 2021; Ntostis et al., 2021). There was not even a partial overlap between age-related DEGs at both stages, signifying that age-related modifications at the GV stage were not carried over to MII oocytes despite transcription being less active during the transition between stages. The deviation in gene expression occurring during maturation may be caused by elevated cellular stress as maternal age increases, and the lack of compensatory mechanisms in response to stress, with the transcriptional quiescence occurring at MII. A limitation to those conclusions is that the IVM used in the Llonch et al. and Reyes et al. studies, which could have engendered a confusion between IVM and age-related effects between GV and MII oocytes. Nevertheless, aging effects may preferentially occur within the time frame of oocyte maturation and protecting oocytes from oxidative stress during this period is crucial in women with advanced maternal age. Adapting ART protocols toward the use of antioxidant molecules during IVM might be beneficial for women aged over 35 (Liang et al., 2017; Liu et al., 2018; Cao et al., 2020), and oxidative stress could also be limited by minimizing the exposure to in vitro environments, with short and soft processes. Counteracting the degradation of mitochondrial activity with age is also one of the reasons given for using mitochondrial transfer. While there may be research opportunities to improve IVM for these women and increase pregnancy rates, at present, oocyte vitrification might be a safer alternative for women who plan to delay childbearing.

Diminished ovarian reserve

DOR naturally occurs with maternal aging but is also found in a small proportion of women under 40 years old in cases of premature ovarian insufficiency and other infertility factors (Kaur and Arora, 2013; Chon et al., 2021). A microarray study determined whether the quantity of antral follicles in the ovary could directly be related to the transcriptomic quality of oocytes (Barragán et al., 2017) (Supplementary Table S1). For that purpose, four groups of healthy fertile women were recruited, including women of young and advanced maternal age, with low (5–10) or high (20–35) antral follicle count, and their MII oocytes were collected from natural cycles. The authors determined that ovarian reserve levels are associated with particular non-coding RNA profiles (mainly long ncRNAs), which have key roles in transcription and regulation of development (Bouckenheimer et al., 2016, 2018). In addition, the few mRNAs found dysregulated in relation to DOR were not related to any pathways. It is interesting that long ncRNAs are supposed to be involved in the pathological mechanisms leading to DOR because they directly regulate follicle development via the activation of different signaling cascades (Tu et al., 2020; Dong et al., 2022; Lv et al., 2023). Unfortunately, the absence of annotation for a very large proportion of those ncRNAs restrict the analysis to a superficial interpretation and the functional importance of those specific ncRNA for the oocyte quality remains unknown.

However, there was limitation to this study lying in the absence of a description of the potential causes of DOR, notably for the young group. Definitively, more studies are needed to explore the oocyte transcriptome from DOR patients and to determine whether DOR due to maternal aging impacts the same biological pathways as DOR caused by other infertility factors.

Lifestyle and environmental exposure

Oocytes that have been exposed to multiple detrimental exposures over the lifecourse are even more susceptible to stress exposure induced in vitro. A current threat to oocyte quality is pollution and exposure to environmental contaminants such as drug residues in food and water, toxicants or chlorinated organic compounds (Foster et al., 2008; Carré et al., 2017). Unfortunately, the literature is sparse for human oocytes due to the difficulty in establishing comparative groups. Mouse exposed to particulate matter <2.5 μm during 12 weeks (8 h per day, 6 days per week) yielded a lower number of oocytes and with increased ROS compared with their counterparts exposed to filtered air (Guo et al., 2021). This correlated with data at the transcriptomic level as the DEGs were mainly related to mitochondrial protein complexes. Differences were even assessed until the blastocyst stage, in which metabolism and metabolic regulation genes were dysregulated by particulate exposure, suggesting that increased oxidative stress may have potential long-term health consequences (Guo et al., 2021). Other harmful substances such as persistent endocrine disruptors, nonylphenols, showed alterations in mitochondrial and apoptosis genes expression that may also be mediated by accumulation of ROS in mice (Xu et al., 2020). Phthalates are also suspected to disrupt oocyte gene expression, according to a study in bovines (Kalo and Roth, 2017).

Along the same lines, some lifestyle habits are generators of oxidative stress. One study on mouse oocytes hypothesized that cytotoxic smoke constituents may be metabolized by the cytochrome P450 2E1, which in turn increases ROS production (Sobinoff et al., 2013). This may consecutively trigger complex ovotoxicity mechanisms negatively influencing pathways involved in cancer, cellular development, cell death, inflammatory responses, and detoxification pathways (Sobinoff et al., 2013; Mai et al., 2014). In human, up-regulation of oxidative stress genes in GC and CC has been reported with smoking, confirming the oxidant–antioxidant imbalance generated by smoking observed in animals (Budani et al., 2017; Konstantinidou et al., 2021). Research should go further and evaluate daily cigarette consumption how many cycles after stopping smoking are required for significant improvements in oocyte competencies.

Additionally, increased BMI (>25 kg/m²) as compared with normal BMI (18.5–24.9 kg/m²) in GV, MI, and MII human oocytes is associated with an up-regulation of metabolic, inflammation and oxidative stress-related genes (Ruebel et al., 2017). Modifications in the follicular fluid composition (insulin, leptin, triglyceride, C-reactive protein, cytokine, polyunsaturated fatty acids concentrations) might create a pro-inflammatory, oxidant, and altered metabolic environment which has repercussions on the follicle. However, the cohort in this study comprised patients undergoing IVF treatment for a multitude of infertility factors including endometriosis and ovulatory disorders, which may weaken some results (Ruebel et al., 2017).

It is likely that many other lifestyle factors and environmental exposures have detrimental effects (Varghese et al., 2011). Alcohol consumption is associated with reduced quantities of oocytes collected in IVF cycles, reduced fertilization rates, reduced live birth rates, and increased pregnancy loss (Firns et al., 2015), and also with oocyte dimorphisms (dark cytoplasm, fragmented polar body, zona pellucida hardening) (Setti et al., 2022). Those dimorphisms may also be negatively influenced by refined sugar consumption but ameliorated by consumption of vegetables, highlighting nutritional habits as an important parameter in oocyte quality (Setti et al., 2022).

Remaining to be studied are the effects at the transcriptomic level, and the constitution of human cohorts to study the effects of disruptive environmental factors, such as particulate matter, is important. Moreover, the studies have often been performed on young individuals, and age is a cumulative risk factor for all these lifestyle factors and exposures, potentially augmenting their individual effects. It is thus important to adopt a healthy lifestyle in the context of an IVF attempt.

As the proceeding sections outlined, before attempting an IVF cycle, the oocyte may have been exposed to multiple exposures which are potentially harmful to its transcriptomic integrity. Furthermore, specific ART interventions involving gene expression modifications can exacerbate or superimpose on these pre-existing alterations. The following sections discuss the interactions between the transcriptome of the oocyte or its surrounding cells with current ART practices.

Ovarian stimulation

Ovarian stimulation is a conventional step in IVF protocols, with the aim of artificially triggering maturation and ovulation to obtain a high number of oocytes and increase the chances of a successful live birth per attempt. This process may be damaging to oocyte quality because of the forced growth and maturation of some antral follicles.

Due to obvious ethical considerations, the transcriptomic effects of ovarian stimulation directly on mature human oocytes have never been investigated. Nevertheless, studies on cumulus cells and MGCs provide a non-invasive indication of the influence of ovarian stimulation on oocyte transcriptomic patterns. Both cell types originate from undifferentiated granulosa cells that divides into two cell lineages. CC constitute the layers of cells in contact with the oocyte while MGC delineate the follicle (Uyar et al., 2013). Bidirectional communication is established between an oocyte and its surrounding cells and this has led investigators to consider MGC and CC as indirect markers of oocyte quality (Uyar et al., 2013; Dumesic et al., 2015).

Ovarian stimulation protocols

Effects of hormonal protocols used to inhibit a premature rise in luteinizing hormone

Ovarian stimulation is practiced with gonadotropins in combination with gonadotropin releasing hormone agonists (GnRH-a) or antagonists (GnRH-anta). Their role is to put the pituitary gland at rest and thus block the risk of spontaneous ovulation by preventing the risk of an LH surge. From a gene expression perspective, Devjak et al. (2012) demonstrated with microarrays that the use of GnRH-a or GnRH-anta does not result in transcriptomic divergence in CC of MI (metaphase I) and MII oocytes. This supports the observation that both treatments are clinically and molecularly equally effective (Al-Inany et al., 2016).

Effects of gonadotropin stimulation protocols

Each fertility clinic relies on their own gonadotropin stimulation protocols, which are adapted to ensure for heterogeneous groups of patients the best chance of retrieving a relatively good number of mature oocytes, as expected for their age and ovarian reserve. Studies on CC and MGC suggest that different hormone treatments used for stimulation do not result in similar gene expression profiles in patients with similar characteristics (ART prescribed for male factor or tubal disease or unexplained infertility, <35 years old, normal BMI) (Supplementary Table S2). We identified nine studies assessing gene expression profiles at both candidate genes (with RT-PCR) and with the whole transcriptome level (using different versions of microarrays). Six of the studies were interested in comparing recombinant FSH (rFSH) stimulation versus highly purified hMG (HP-hMG, human menopausal gonadotropin) and all reported differentially expressed genes in both MGC and CC (Grøndahl et al., 2009; Adriaenssens et al., 2010; Brannian et al., 2010; Assou et al., 2013; Gatta et al., 2013; Cruz et al., 2017).

rFSH versus HP-hMGH

Concerning CC analyses, investigations showed little consensus in either the number of differentially expressed genes between rFSH and HP-hMG regimens and the biological significance of the pathways retrieved disturbed in four different studies (Adriaenssens et al., 2010; Assou et al., 2013; Gatta et al., 2013; Cruz et al., 2017) (Supplementary Table S2). This might stem from heterogeneities in methodology and the populations analyzed including either oocyte donors or patients with unexplained infertility (Supplementary Table S2). However, Assou et al. (2013) and Gatta et al. (2013) found DEGs in identical molecular pathways, related to cell-to-cell interactions, lipid metabolism and cellular assembly and organization, while Gatta et al. (2013) and Cruz et al. (2017) found DEGs related to system development. Lipid metabolism represents the principal energy source for resuming oocyte maturation (Khan et al., 2021) and defective cell-cell interactions of an oocyte and its surrounding CC might break a cascade of essential hormonal and metabolic signals (Russell and Robker, 2007; Li and Albertini, 2013). Our cross-over of DEGs between studies revealed common genes up- or down-regulated following HP-hMG stimulation compared with rFSH when studies were compared one by one (Fig. 1A and B): COL1A1, DKFZp451A211, ENDOD1, GAL, HS3ST1, MT3, NFIB, NPY1R, OSBPL6, WNT3A. Notably, up-regulation of NPY1R in CC was associated with good embryo quality at day-3; NPY receptors are implicated in the modulation of steroid production (Assou et al., 2013). In the same study, OSBPL6 up-regulation in CC was associated with higher blastocyst grading and is a modulator of cholesterol synthesis (Assou et al., 2013). The nuclear factor I B (NFIB) expression in CC was also associated with the oocyte capacity to generate embryos capable of implanting and may play a role in embryogenesis and organ development (Steele-Perkins et al., 2005; Assou et al., 2008). In bovine CC, COL1A1, a sub-unit of type I collagen and component of the extracellular matrix may regulate several important biological pathways such as ovarian cell cycle, proliferation or apoptosis (Fu et al., 2019). Taking the comparisons together, there is evidence that rFSH and HP-hMG stimulations do not mobilize similar CC metabolic pathways.

Figure 1.

Differentially expressed genes between studies comparing different types of gonadotropin stimulation protocols. (A) Genes up-regulated with HP-hMG in CC as compared to rFSH. (B) Genes down-regulated with HP-hMG in CC as compared to rFSH. (C) Genes up-regulated with HP-hMG in MGC as compared to rFSH. (D) Genes down-regulated with HP-hMG in MGC as compared to rFSH. (E) Overlap of genes differentially expressed between stimulation protocols in Cruz et al. HP-hMG, urinary FSH and rFSH compared. (F) Overlap of genes differentially expressed between stimulation protocols in Gatta et al. (FSH, rFSH, and FSH + rFSH compared). (G) Overlap of genes differentially expressed between stimulation protocols in Gurgan et al. (HP-hMG, rFSH, and rLH + rFSH compared). CC: cumulus cells; HP-hMG: highly purified human menopausal gonadotrophin; MGC: mural granulosa cells; rFSH: recombinant follicle stimulating hormone.

The main difference between rFSH and HP-hMG stimulation was the absence of exogenous LH activity when rFSH was administered alone and surprisingly, there were no differences in the expression of LH/hCG receptor in the three microarray studies in CC. This suggests that despite activating different molecular mechanisms that mobilize distinct intracellular pathways, both treatments have nonetheless a similar potency to induce hormonal receptors. However, the results still must be interpreted with caution owing to the limited sample size in the two studies (Supplementary Table S2) (Gatta et al., 2013; Cruz et al., 2017).

In parallel, in MGC, although Grøndahl et al. (2009) found only 85 DEGs and Brannian et al. (2010) 20 times more but with a small sample size (n = 4 per group), the main observations pointed to the perturbation of metabolic processes (lipid metabolism, protein phosphorylation, cholesterol and steroid synthesis). Genes found to be down-regulated with HP-hMG common to both studies included ATP2C1, EPHA4, FGG, HMGCS1, RNGTT, and TFRC (Fig. 1C and D). HMGCS1 is notably required for cholesterol biosynthesis but contrary to the results of CC studies, the gene LHCGR encoding the LH/hCG receptor was down-regulated with HP-hMG in Grøndahl et al. (2009), which could explain the shutdown of the cholesterol de novo synthesis pathway.

Recombinant LH supplementation

In a small study on three markers of developmental competence (HAS2, GREM1, and PTGS2), recombinant LH (rLH) supplementation during the rFSH stimulation protocol suggested a positive effect on oocyte maturation manifested by the up-regulation of HAS2 and PTGS2 with rLH in CC, as assessed via RT-PCR (Barberi et al., 2012). In a larger study using microarrays, rLH supplementation was associated with a differential expression of 496 genes (Gatta et al., 2013), but the results were not discussed and the expression of HAS2 and PTGS2 was similar between the rLH + rFSH and rFSH groups.

Origin of FSH

Two studies have investigated different origins of FSH stimulation, but the results have not yet been replicated to date (Gurgan et al., 2014; Cruz et al., 2017). In CC, stimulation with either rFSH or human-derived FSH alone versus rFSH + FSH together showed variable responses on gene expression, notably on genes related to cholesterol metabolism, cell growth and proliferation, cell morphology and death, cell adhesion and differentiation, as well as protein synthesis (Gurgan et al., 2014). A limitation in this study is the absence of the description of the population analyzed and underlying infertility causes that is important regarding the response to treatment (Supplementary Table S2). Urinary FSH was also compared with rFSH and HP-hMG in CC from healthy oocyte donors (Cruz et al., 2017). Unsurprisingly, urinary FSH and rFSH stimulation had similar effects on the transcriptome with only 44 DEGs and without enrichment in any biological process. Three times more DEGs were found between urinary FSH and HP-hMG, related to metabolic processes, which corroborates previous observations on the various effects of HP-hMG and FSH isoforms on metabolic activity.

Differences in induced regulation from the different FSH isoforms may stem from their different glycosylation profiles, which play an essential role in FSH pharmacokinetics, metabolic stability, and receptor signaling (Ulloa-Aguirre et al., 1999; Gurgan et al., 2014). For example, human urinary FSH is more sialylated than rFSH, thus more acidic (Andersen et al., 2004). It has been shown that less acidic isoforms have a higher half-life and potency in estrogen stimulation, which may result in different oocyte biological activity (Timossi et al., 2000). Recently, a new FSH analog with an increased half-life (corifollitropin alfa) has entered the clinic (Anderson et al., 2018). Now, transcriptomic comparisons have to be performed with other FSH isoforms.

In the end, studies comparing more than three different regimens support the observation that proximal treatments of different origins have variable effects on the transcriptome, which is manifested by the low number of overlaps between DEGs (Gatta et al., 2013; Gurgan et al., 2014; Cruz et al., 2017) (Fig. 1E–G).

Dose effects

A recent study investigated the benefits of a mild ovarian stimulation protocol compared with the conventional ovarian stimulation in CC from poor ovarian responders with advanced maternal age (Liu et al., 2022). The main difference between the two protocols was the daily gonadotropin dose administered to patients, which was closer to the physiological environment in mild treatments. The RNA-seq analysis revealed 425 DEGs (192 up-regulated and 233 down-regulated with mild stimulation) which were mainly involved in cytokine activities, regulation activities, immune responses, oocyte development, cytokine–cytokine receptor interactions, the TGF-β signaling pathway, the cGMP-PKG signaling pathway, and metabolic pathways. The authors concluded that the mild protocol is beneficial for poor responders because they further demonstrated, via western blot and hormone assays, the importance of the up-regulation of the TGF-β signaling pathway in this group for the cross-talk between MGC and oocyte and embryo quality.

Ovulation triggering protocol

Following ovarian stimulation, ovulation induction is artificially triggered. This ovulatory surge is the crucial step in oocyte meiosis and maturation, as well as in maturation of its supporting cells. Various effective triggering methods are commonly used, including hCG (classically recombinant) or GnRH-a administration or the combination of both. As assessed with microarrays, genes related to steroidogenesis were up-regulated with GnRH-a compared with hCG in CC and notably the expression of the LHCGR gene was up-regulated, which may promote progesterone production and facilitate implantation (Borgbo et al., 2013). In contrast, Haas et al. (2014), still studying CC but by using RT-qPCR, reported lower expression of genes related to steroidogenesis (CYP19A1, CYP11A1, HSD3B1) with a GnRH-a trigger and no difference in LHR expression (Fig. 2A and B, Supplementary Table S2). Again, discrepancies in the results might stem from variability in the populations analyzed, notably the inclusion or exclusion of endometriosis patients who may have a particular local intrafollicular environment (presence of ROS, high estrogen, and progesterone levels) (Sanchez et al., 2017). In MGC, the principal findings suggest that GnRH-a might reduce the risk of developing OHSS by down-regulating the angiogenesis pathway (Borgbo et al., 2013). Interestingly, these observations were partially confirmed by a study in MGC from OHSS patients with RT-qPCR on VEGF (a central vasoactive factor) whose expression was lowered with the use of GnRH-a compared with hCG (Haas et al., 2014). In addition, high or low doses of GnRH showed similar effects on three regulator genes of oocyte maturation, HSD3B1, LHCGR, and IHNBA expression in both CC and MGC of healthy oocyte donors (Vuong et al., 2017).

Figure 2.

Differentially expressed genes between studies comparing different types of ovulation triggering protocols (GnRH-agonist trigger, hCG, and dual trigger). (A) Genes up-regulated with GnRH-a as compared to hCG. (B) Genes down-regulated with GnRH-a as compared to hCG. (C) Genes up-regulated with hCG + GnRH-a as compared to hCG. (D) Genes down-regulated with hCG + GnRH-a as compared to hCG. GNRH-a: gonadotrophin releasing hormone agonist; hCG: human chorionic gonadotrophin.

By using RNA-seq to compare hCG single versus double (hCG + GnRH-a) regimens in CC from poor and normal responders to stimulation, dual triggering showed up-regulation of pathways such as oocyte maturation, cell cycle, and apoptosis (Fuchs Weizman et al., 2019). However, the few similarities in the genes disturbed by the triggering method in both responder groups (Fig. 2C and D, Supplementary Table S2) revealed that patients’ characteristics, such as advanced maternal age (>40 years old) or low levels of anti-Müllerian hormone resulting in poor ovarian response to stimulation, are associated with variable effects of the triggering method compared with a cohort of normal responders. In MGC, comparison of the hCG single trigger versus the double trigger showed AREG, EREG, and SERPINF1 up-regulation and GJA1 down-regulation with the double trigger, which was not observed in the Fuchs Weizman et al. study on CC (Haas et al., 2016; Fuchs Weizman et al., 2019) (Fig. 2C and D). Higher levels of amphiregulin (AREG) and epiregulin (ERG) may be associated with improved oocyte maturation, while higher levels of PEDF (SERPINF1), which has anti-angiogenic actions, may reduce the risk of OHSS.

Collectively, despite the inter-patient and inter-cell type heterogeneity of pathway responses to the triggers, these results still highlight the benefit of using a dual trigger to promote oocyte developmental competence while preserving patients from the detrimental effects of a single hCG trigger (Haas et al., 2016). Nonetheless, GnRH-a still has apoptotic effects, notably on GC, which could induce premature demise of the corpus luteum (Gonen et al., 2021).

Ovarian stimulation effects compared with natural cycles

Different ovarian stimulation regimens shape the oocyte transcriptome but with differences that may remain limited when one is compared with another. However, recourse to ovarian stimulation itself may substantially affect oocyte gene expression compared with natural cycles and could give hints as to why oocytes have lower developmental competence after ovarian stimulation (Blondin et al., 1996; Lee et al., 2017). A study on CC from oocyte donors showed a mild effect of ovarian stimulation on the transcriptome using microarray (18 DEGs related to immune processes, meiosis and ovulation) but even in the natural cycles, recombinant hCG was administered to trigger ovulation which could have hidden ovulation induction effects (de los Santos et al., 2012). In contrast, Papler et al. (2014) reported that in CC from natural cycles and ovarian stimulation, 66 DEGs were revealed by microarray, mainly related to RNA degradation and transport, ribosomes, metabolism and oxidation-reduction processes. Finally, Lu et al. (2019), this time analyzing MGC by using RNA-seq, identified 1002 DEGs, enriched for genes related to immune processes for those up-regulated, and genes related to cell cycle, meiosis, steroidogenesis, and oocyte maturation for those down-regulated with ovarian stimulation. Cross-over between up- and down-regulated genes in those three studies showed limited overlap, as only GSTA1, GSTA2, and CTSV were found to have similar effects associated with ovarian stimulation in at least two studies (Fig. 3A and B). GSTA1 and GSTA2 down-regulation might be a manifestation of lower steroidogenesis with ovarian stimulation, as demonstrated in bovine MGC (Rabahi et al., 1999). The low sample size in all these studies and the lack of replication remain a serious limit in assessing the transcriptomic effects of ovarian stimulation and results should be taken with caution (Supplementary Table S2).

Figure 3.

Differentially expressed genes between studies comparing ovarian stimulation and natural cycles. Natural cycles were set as the reference group. (A) Genes up-regulated with COH as compared to natural cycles. (B) Genes down-regulated with COH as compared to natural cycles. COH: (controlled) ovarian stimulation.

Overall, it remains unknown whether modifications in CC and MGC may reflect oocyte integrity, even if there is a close interaction between these cell types. Evidence of alterations in gene expression in ovarian stimulation protocols in the oocyte is only available from mouse and farm animals. The expression of some maternal effect genes, which could be considered molecular markers of oocyte quality, has been investigated in mouse oocytes following superovulation by RT-qPCR (Jahanbakhsh-Asl et al., 2018). The expression of BMP15, HDGF, DNMT1, DPPA3, and ZFP53 was significantly reduced in superovulated oocytes compared with naturally ovulated oocytes. The three latter genes are involved in epigenetic reprogramming, which could explain the presence of methylation errors in superovulated oocytes (Market-Velker et al., 2010). BMP15, along with two others preferentially expressed growth factors in oocytes (GDF9, FGF8), showed no difference in FSH stimulated mouse oocytes compared with those produced in vivo (Sánchez et al., 2010). However, the increase in FSH dose resulted in significant up-regulation of BMP15, GDF9, and FGF8. A recent study compared, at the same time, the transcriptome and proteome of superovulated mouse oocytes with their counterparts in vivo (Taher et al., 2021). The transcriptomes were very similar in contrast to the proteomes, and the results indicated that superovulation does not cause direct harmful effects on oocytes but rather it produces oocytes that have not reached complete cytoplasmic maturity.

In conclusion, the lack of studies directly on oocytes in humans is a notable barrier in our understanding of how ovarian stimulation could affect oocyte quality at the molecular level. A striking aspect of the studies is the almost complete absence of genes commonly dysregulated under the same conditions. This may stem from the various technologies used (different arrays versus RNA-seq), the low sample size which may over-estimate the magnitude of an association between gene expression and a specific condition, the heterogeneity between the clinical characteristics of patients, and the CC and MGC heterogeneity in response to treatment in the same patient, which is unknown. Evidence from CC and MGC and studies in small and large animal models suggest there are no drastic transcriptomic damages, but oocytes produced via ovarian stimulation appear not to have achieved complete cytoplasmic maturation compared with their naturally produced counterparts. A fundamental step forward in the field would be to assess if there is a heterogeneous response of sibling CC/MGC and sibling oocytes (i.e. from the same follicle and patient) following an identical ovarian stimulation treatment. Some ovarian stimulation protocols might be a risk for patients by increasing the risk of developing OHSS and therefore moving toward mild ovarian stimulation and dual trigger is beneficial. For most patients (male factor or tubal disease infertility, <35 years old, normal BMI), the standard protocols used for decades are well adapted, but approaches to reduce unpleasant treatment experiences with the use of dual trigger and doses closer to physiological environment could become more widely used. A main limit to the scientific evidence regarding transcriptomic effects of ovarian stimulation is the low number of studies on specific populations, including those with PCOS, poor ovarian responders, or aged patients, who display a hormonal profile suboptimal for the effectiveness of standard protocols. In the era of personalized medicine, individualized ovarian stimulation will soon be preferred above standard protocols (la Marca and Sunkara, 2014).

In vitro conditions

IVM of oocytes

Medical IVM is defined as the maturation of oocytes (most often as cumulus–oocyte complexes) collected from small and intermediate antral follicles (generally diameter <13 mm) to MII in vitro using an appropriate IVM medium (Dahan et al., 2016). It must be differentiated from rescue IVM (rIVM), which is performed on GV-stage oocytes denuded of their CC and collected from ovarian stimulation cycles and thus not a common procedure (de Vos et al., 2016). Medical IVM is an effective method for women with resistant ovary syndrome and for reducing the risk of developing OHSS; it can also be used for oncological patients to preserve fertility over time before a cryopreservation step (Shalom-Paz et al., 2012; Virant-Klun et al. 2022).

Problematically, there is a high heterogeneity among fertility clinics in the techniques used to perform IVM/rIVM (e.g. standard or biphasic IVM, follicular priming methods, choice of the culture medium and supplementations, time of culture (Yang et al., 2021)) that may contribute to their high variability in clinical outcomes (Dahan et al., 2016). Additionally, IVM/rIVM is affected by reports depicting poor oocyte developmental competence and multiple other clinical outcomes despite decades of research (Hatırnaz et al., 2018; Krisher, 2022). Oocyte transcriptomic analysis appears to be crucial to unravel the molecular cause behind the altered oocyte competencies.

The first study on the transcriptomic effects of rIVM compared with in vivo maturation on pools of human oocytes revealed, on the one hand, up-regulation of genes related to signal transduction and metabolism (nucleic acids, tRNA, lipids, steroids) and, on the other hand, down-regulation of genes involved in protein biosynthesis, translational regulation, cell cycle, and homeostasis (Wells and Patrizio, 2008). Although the differences remained mild and MII oocytes produced via rIVM had an expression pattern highly similar to the in vivo matured oocytes, this study suggested that their expression pattern displayed vestiges of the GV stage, which could explain the greater difficulties in producing successful pregnancies following rIVM. Additionally, other studies on pooled oocytes have established that rIVM alters the regulation of transcription, cell cycle, transport, cellular metabolism, respiratory process, epigenetics, and embryogenesis (Jones et al., 2008; Virant-Klun et al., 2013) (Supplementary Table S3). The high number of DEGs identified in those studies could be indicative of failure in the correct post-transcriptional regulatory events during nuclear maturation, such as poly- or de-adenylation, and thus explains the low developmental competence of rIVM oocytes that still resemble GV oocytes (Jones et al., 2008).

More recently, next-generation sequencing has brought a finer resolution of transcriptomic changes related to IVM and rIVM in oocytes. The high number of DEGs retrieved in four studies between in vitro and in vivo maturation confirmed previous observations (Zhao et al., 2019; Ye et al., 2020; Lee et al., 2021; Yang et al., 2022) (Supplementary Table S3). In rIVM oocytes, Zhao et al. (2019) identified key genes related to metabolism that turned out to be kinases and enzymes related to CoA production. All are key regulators of oocyte maturation as kinases are necessary for completing the meiosis cycle and CoA is critical in the regulation of substrate production in the Krebs cycle, and down-regulation of CoA-related enzymes observed with IVM would cause a decline in energy metabolism leading to deficient maturation (Palmer et al., 1998; Downs et al., 2009). DEGs associated with rIVM in the Lee et al. study (Lee et al., 2021) were over-represented for genes participating in metabolic processes, biosynthesis and oxidative phosphorylation. The GATA-1/CREB1/WNT signaling pathway was particularly down-regulated with rIVM, promoting apoptosis and decreasing the chance of correct cytoplasmic maturation. Corroborating the observations from rIVM, DEGs identified by scRNA-seq in Ye et al. (2020) with IVM in a population of healthy volunteers undergoing caesarean delivery were related to cell cycle, mRNA metabolism, mitochondrial respiration biogenesis and ATP metabolism notably. In Yang et al., using oocytes collected from pregnant women undergoing caesarean delivery (Yang et al., 2022), among the 2281 DEGs identified from IVM oocytes, 160 were mitochondria related genes and were involved in oxidative phosphorylation, amino acid and carbon metabolism, and fatty acid degradation. DEGs were also enriched in mitochondrial respiration processes, cellular structure, metabolic process, NADH dehydrogenase activity and ribosomal activity. This might highlight compensatory mitochondrial mechanisms activated by the oocyte to adjust for the energy needed to complete cytoplasmic maturation. These results could not be extrapolated to patients needing medical IVM (e.g. PCOS patients).

Transcriptomic effects of rIVM have also been assessed in patients with PCOS, who are most likely to benefit from this treatment, but only in CC. Corroborating previous observations, >5000 DEGs have been identified in CC from patients with PCOS and involve the down-regulation of oocyte maturation and cumulus expansion-related genes with rescue IVM (Ouandaogo et al., 2012).

Overall, due to the inability to access complete data for most studies on in vitro oocyte maturation, analysis of DEGs common to multiple studies was not pertinent. From a methodological standpoint, most studies did not provide sufficient clinical information on the patients included, and the quality of the evidence for the results presented could be questioned, particularly for medical IVM (Supplementary Table S3). To gain insight on how IVM/rIVM has different effects regarding the population studied, information on maternal age and causes of infertility must be clearly presented. However, a clear message emanating from all studies is the perturbation of processes related to the mitochondrion, oxidative phosphorylation and metabolism. Incomplete cytoplasmic maturation might be explained by these deficient energy processes. Improvement of energy availability for the oocyte could be a lever of development for the future of IVM culture conditions (Katz-Jaffe et al., 2009; Gao et al., 2017; Zhang et al., 2021). Refinement of the culture medium composition is also a promising field of research. Notably, co-culture of immature oocytes with CC of mature oocytes from the same patient led to a transcriptomic profile with reduced divergence compared with the differences observed between classic IVM and in vivo matured oocytes (Virant-Klun et al., 2018). Addition of growth hormone in the IVM media was associated with the up-regulation of AURKA, CENPE, PDIA6, LINGO2, and CENPJ, which probably contribute to accelerating meiotic progression, balancing redox homeostasis, and thus promoting oocyte maturation (Li et al., 2019). Protection of oocytes against oxidative stress during IVM with antioxidant supplementation offers a new perspective in improving IVM efficiency (Cao et al., 2020; Sallem et al., 2022).

Oxygen tension

A key component in in vitro steps that is largely overlooked in human is oxygen tension. Depending on the species in mammals, it is estimated that the physiological intrauterine and intratubal oxygen tension in vivo varies between 2% and 8%, with levels that decrease as a follicle develops. Current practices in IVF laboratories are either performed at 5% O₂ or 20% O₂ but, transiently, follicles and oocytes extracted from the genital tract could be exposed to variable oxygen concentrations (Christianson et al., 2014). Clinical outcomes comparing low (5%) and high (20%) oxygen tension effects indicated no positive effects in a high O₂ environment (Eppig and Wigglesworth, 1995; Kasterstein et al., 2013; Lim et al., 2021). It is known that O₂ increases the production of ROS and could have cytotoxic effects and activate stress pathways. Oocytes would adapt in response to different oxygen environments, notably through genes regulated by the hypoxia-inducible factor (HIF) (Lim et al., 2021).

Whether the transcriptional content of oocytes is altered depending on physiological or ambient oxygen tension is currently unknown in human. In mice, using RNA-seq on pools of primary oocytes cultured to the GV phase for 9 days under 5% O₂ or 14 days under 20% O₂, ontologies corresponding to female gamete genes and chromatin organization tended to be up-regulated with 20% O₂, whereas stress response genes were up-regulated with 5% O₂ (Naillat et al., 2021). In another mouse study, this time using scRNA-seq, oocytes from secondary follicles cultured to GV for 12 days under 20% O₂ versus 7% O₂ were compared (Takashima et al., 2021). Oocytes grown in 20% O₂ showed up-regulation of genes related to apoptosis and the sphingolipid signaling pathway, including ceramide metabolism. The authors concluded that mitochondrial dysfunction and impaired developmental abilities were associated with ambient oxygen tension as compared to in vivo conditions, whereas the in vivo and 7% O₂ environments resulted in oocytes highly similar in their transcriptomic profiles. Studies in a large animal model, yak oocytes matured under 5% or 20% O₂ revealed, for the hypoxia group, up-regulation of genes involved in the cell cycle (ERK1 and ERK2 cascade, PI3K-Akt signaling) and maturation with a decrease in the expression of genes involved in oxidative phosphorylation; the authors concluded that there was an improved quality of oocytes grown under hypoxic conditions (Li et al., 2020).

In conclusion, three studies with different animal models showed contradictory results on whether hypoxia or normoxia is beneficial at the transcriptomic scale. The studies suggest the need to further assess to what extent non-physiological gas conditions disturb key oocyte biological processes in humans.

Temperature

IVM is rigorously performed at 37°C but the temperature could transiently vary during oocyte transfer from collection to the IVF laboratory and then during the further oocyte manipulations (e.g. during oocyte denudation and fertilization). After oocyte collection, it is important to minimize the time for CC removal and manipulations because dishes are also prone to temperature decreases, with the risk of slightly modifying pH conditions of the buffered media and generating ROS (Agarwal et al., 2022). IVM performed under large temperature differences (38.5°C versus 41°C) does not alter the expression of major heat shock proteins and oocyte quality markers in bovine (BMP15, GDF9) but the overall effect of temperature on the whole transcriptome remains to be evaluated (Payton et al., 2011; Roth, 2018).

Pollutants

While maternal exposure to pollutants can affect the oocyte, additional pollution specific to the laboratory environment is directly in close contact with the oocyte. It involves plastic consumables such as tips for oocyte holding and denudation, tubes or plastic dishes for collection, maturation, fertilization and cryopreservation, as the worrying presence of bisphenol A, S, and AF was measured in those consumables from different brands, even if they do not leach into ART media (Togola et al., 2021). Contrary to in vivo conditions, there is no glucuronidation in IVM and bisphenols cannot be turned into their inactive form (Togola et al., 2021). Gene expression changes with exposure to bisphenol A in double-strand break signaling and repair related genes has been observed with RT-qPCR in human fetal oocytes (Brieño-Enríquez et al., 2012). In mouse oocytes, IVM under supraphysiological concentrations of bisphenol A showed alterations in gene expression of major genes involved in cell cycle, hormone signaling, and translational regulation (Ferris et al., 2016). In addition, there could be a synergetic or additive effect of different pollutants on oocyte transcriptome, and the careful assessment of all sources of pollution and their uncontrolled effects on the transcriptome of gametes is necessary.

Cryopreservation and storage time

Cryopreservation was first introduced in the 1980s with the slow freezing method but has been gradually replaced by vitrification among IVF clinics, an ultrarapid freezing and cost-effective technique. For ethical and clinical reasons, there is a growing use of cryopreservation with respect to delayed childbearing, maternal aging, donation programs, and cryopreservation before gonadotoxic therapy (Cobo et al., 2021).

A recurrent observation is the overall decrease in total mRNA content following slow freezing (Chamayou et al., 2011; Monzo et al., 2011; Stigliani et al., 2015) and vitrification (Huo et al., 2021) as compared with fresh human oocytes. Maternal mRNAs drive early embryo development until the maternal-to-zygotic transition and sufficient maternal mRNAs should be stored in the oocyte beforehand (Flach et al., 1982; Dworkin and Dworkin‐Rastl, 1990; Sirard et al., 2006). Additionally, it is likely that major pathways including DNA structural organization, chromosomal structure maintenance and organization, cell cycle processes, response to DNA damage stimuli, and cellular response to stress and DNA repair are altered by the slow freezing method according to two microarray and one RT-PCR studies on pools of oocytes (Chamayou et al., 2011; Monzo et al., 2011; Stigliani et al., 2015) (Supplementary Table S4). Vitrification and slow freezing do not have the same influence compared with fresh oocyte, and genes modified by both techniques are partially different. The study by Monzo et al., using microarrays (Monzo et al., 2011), showed that only 88 DEGs were common to both techniques, among respectively 389 and 608 DEGs identified for slow-frozen and vitrified MII unfertilized oocytes. Slowly frozen oocytes showed down-regulation of genes associated with DNA repair, transcriptional regulation in response to DNA damage, cell cycle regulation and maintenance of chromosome stability, which were absent when comparing vitrified to fresh oocytes (Monzo et al., 2011). Differences between vitrified and fresh oocytes appear less deleterious than those after slow freezing, as confirmed by the mRNA content of a panel of genes involved in important oocyte functions (cell cycle, chromosome structure, mitochondrial pathways) which are more conserved in supernumerary vitrified MII oocytes than with slow freezing (Chamayou et al., 2011). Vitrification also showed no differential expression in genes related to cytokinesis (PLK1, DCTN1,2,3 and 6), known to be involved in zygotic arrest, and no difference for genes involved in basic oocyte functions such as GAPDH, BMP15, GDF9, and OCT4 in supernumerary MII oocytes (Di Pietro et al., 2010; D’Aurora et al., 2019). Although it has less of an impact than slow freezing, pathways altered by vitrification are related to transcriptional regulation, ubiquitination, cell cycle and oocyte growth (Monzo et al., 2011; Huo et al., 2021), which is not insignificant regarding oocyte and embryo developmental competence. Recent work by Barberet et al., using both scRNA-seq and a strong study design consisting of sibling oocytes from the same patients, was reassuring regarding the use of vitrification. The number of DEGs remained low (108 versus 1987 DEGs in Huo et al.) and the difference between the fresh and vitrified groups was largely limited (fold change <1) (Huo et al., 2021; Barberet et al., 2022). In addition, the use of either manual of semi-automated vitrification is also without consequence for the oocyte (Barberet et al., 2022). By comparing the DEGs of the studies, we did not find any overlapping genes from one study to another, either for vitrification or slow freezing, but complete data were not available for one study (Monzo et al., 2011). Evidence from animal studies globally corroborates the observations in human oocytes, notably, alterations in cell cycle, ubiquitination, and transcriptional regulation linked to cryopreservation (Wang et al., 2017; Huang et al., 2018; Barberet et al., 2020; Ma et al., 2022).

Finally, it does not seem that storage time exacerbates differences after cryopreservation for a few months compared with >5 years periods (Stigliani et al., 2015; Huo et al., 2021).

The nature and the concentration of cryoprotective agents might be a source of variation for the oocyte transcriptome and oocyte quality due to vitrification, as indicated by studies in bovines (Zhang et al., 2020a). Indeed, decreased concentrations of CPAs reduced the number of aberrant transcriptional modifications in immature oocytes vitrified in liquid helium, but not liquid nitrogen (Zhang et al., 2020a). Optimization of the vitrification technique is still possible to reduce the few molecular side effects reported to date in human oocytes and to ameliorate clinical outcomes.

ART factors not studied to date

The Cairo consensus recently noticed that ‘there is only one thing that is truly important in an IVF laboratory: everything’. It would be interesting to investigate to what extent overlooked factors such as light exposure could affect gene expression since early mouse embryos at different stages showed light-induced changes in genes related to regulation of transcription, apoptosis, cell cycle and cellular responses to stimulus (Lv et al., 2019). Variations in oocyte handling media formulation also exist among fertility clinics, but there are no studies comparing their transcriptomic effects to date. Whether micromanipulation or vigorous pipetting could trigger severe mechanical stress is unknown as well. The exposure time to hyaluronidase treatment for oocyte denudation might also influence the oocyte transcriptome. We still do not know whether the incubation time of oocytes before insemination or sperm microinjection, which can delay the fertilization step, could induce in vitro aging and directly impact fertilization and developmental abilities. While all these factors may have a limited impact on oocyte competence independently, the combination of all these factors cannot be neglected.

Implications of different ART and external factors for common transcriptional regulation mechanisms

Examining common genes dysregulated due to different exposures is challenging owing to the different methodologies and significance thresholds employed in each study. However, we grouped all the DEGs found per ART interventions (ovarian stimulation versus natural cycle, cryopreservation versus fresh, IVM versus in vivo maturation) independently of the study and we observed that 8 DEGs were common to the different exposures evaluated in this review: CDC45, CENPA, GINS1, IQGAP3, MED17, TNFRSF1B, TNNI3, and UHRF1. For most of them, their precise role in oocytes is unknown, but they may be linked to centromere position (CENPA), initiation of DNA replication (CDC45, GINS1), mediation of RNA polymerase II transcription (MED17), and apoptosis (TNFRSF1B). Interestingly, UHRF1 plays a major role in DNA methylation by recruiting DNMT1 (Maenohara et al., 2017). Mouse Uhrf1-null oocytes notably showed increased aneuploidy rates, DNA double-strand breaks, and spindle aberrations (Cao et al., 2019). However, these results need to be interpreted with caution because it remains a qualitative assessment of genes deregulated in different studies (i.e. common genes of three ART processes yet not reported by all studies).

We can still hypothesize that dysregulation in genes involved in chromatin-based processes such as DNA methylation, heterochromatin modulation, histone modification, and complex remodeling, but also otherwise genomic imprinting, may be a common mechanism linked to an adverse oocyte environment and explaining global transcriptomic modifications. Indeed, when we focused on chromatin modifier genes, we systematically found dysregulations of such genes either after ART intervention or lifestyle exposure, as well as due to internal factors such as maternal aging and reproductive diseases (Table 1). Among the most frequently found dysregulated genes with such functions were CHD8, DNMT1, PBRM1, PHF10, and SMARCA4 (Table 1). The role of DNMT1 is particularly crucial for global DNA methylation patterns and genomic imprints, as DNMT1 is primarily required for maintaining methylation patterns established in gametogenesis. DNMT1 can also participate in the de novo methylation of a small but essential part of the genome (Edwards et al., 2017). Careful assessment of genes involved in chromatin-based regulation appears warranted in future research regarding ART and external factors influencing the oocyte transcriptome because these genomic regions seem to be particularly at risk for oocyte quality.

Table 1.

List of chromatin modifier genes, imprinted genes, and transposable elements shown to be differentially expressed in studies investigating ART, maternal aging, lifestyle, and endometrial diseases’ effects.

CC: cumulus cells; GnRH-a: gonadotrophin releasing hormone agonist; hCG: human chorionic gonadotrophin; HP-hMG: highly purified human menopausal gonadotrophin; MGC: mural granulosa cells; PCOS: polycystic ovary syndrome; rFSH: recombinant follicle-stimulating hormone; rLH: recombinant luteinizing hormone.

*

Genes in red, blue, and green are found in, respectively, two, three, and four different studies.

Transposable elements have barely been studied to date but would also deserve more attention as they are key regulators of genomic stability.

Conclusion and perspective

We have seen that many IVF factors and additional external factors have the potential to impair oocyte transcriptomic integrity, which might not be innocuous for the developing embryo. In line with our research questions, our findings include three aspects. (i) We found similarities in dysregulated biological pathways, but not genes, between studies for each factor considered (summarized in Figs 4 and 5). (ii) We provide an overview of oocyte transcriptomic effects integrating all the types of intervention and we identified common oocyte alterations in cell cycle and division, energy (metabolism, mitochondria activity), and transcriptional and translational regulation pathways, which are essential for governing the development of the embryo until the maternal-to-zygotic transition. Additionally, each factor may mobilize other distinct biological pathways. (iii) For the first time, we also revealed a repetitive hypersensitivity of genes related to epigenetic mechanisms, which deserves more attention regarding the influence of ARTs, as they are considered risk factors for epigenetic defects, notably, at the DNA methylation level.

Figure 4.

Summary of the current evidence regarding factors affecting the transcriptomic integrity of human oocyte during ART protocols. CC: cumulus cells; COH: (controlled) ovarian stimulation; MGC: mural granulosa cells.

Figure 5.

Summary of the current evidence regarding environmental factors, infertility and aging effects on the transcriptomic integrity of human oocyte. CC: cumulus cells; MGC: mural granulosa cells.

However, it is unclear whether all these alterations are detrimental for the future embryo and prone to be pathogenic later in life. Maternal transcripts are recruited in a temporally complex manner to fit in with the biological needs of the early embryo. Some of them, designated as maternal effect genes, have already proven associations with structural birth defects in the absence of expression in mouse models (Mitchell, 2022). Further identification of all the maternal transcripts crucial before the embryo takes control of its own genome will help in our understanding of how the oocyte environment determines the success or failure of the first mitotic divisions. One limit to the scientific evidence of all of the studies assessed in this review lies in the heterogeneity of ART factors transcriptomic effects that is rarely studied in populations of sibling oocytes. Indeed, the studies considered here mainly pooled oocytes or did not report data per patient. The fact that ART effects might not be similar from one oocyte to another in the same patient under the same conditions is an important parameter for the success of an ART attempt. It also remains undetermined whether the oocyte transcriptomic modifications associated with ART have a short- or long-term effect in humans, although two mice studies have considered this question for cryoinjuries and ovarian stimulation (Eroglu et al., 2020; Taher et al., 2021). Overall, it is completely unknown for human oocytes whether the transcriptomic and individual gene alterations described in this review correlate with proteomic modifications. Simultaneous assessment of the human oocyte proteome and transcriptome could give better insights on the consequences of transcriptomic alterations found with ART and intrinsic factors, but there are still technical challenges to overcome.

Although we lack considerable evidence for some aspects of ART, the majority of children born through those techniques remain healthy (Hart and Norman, 2013). Even so, this review could identify the weakest points in ART to ameliorate in the near future in order to improve ART outcomes. From this perspective, we have elaborated on some recommendations for ART practices which include adaptations in hormonal treatment and culture conditions, and a reduction in exposure factors in the in vitro environment. These adaptations must be set against the small number of studies, the absence of human oocyte evidence and their sometimes conflicting results for certain factors (Supplementary Fig. S1).

ART should not force the oocyte to adapt to a modified environment but should put the oocyte in the best possible conditions so as not to have a detrimental impact on embryonic development. While it is not possible to act on certain parameters such as maternal age, it is necessary to avoid any cumulative stress during gamete manipulation and embryo development, and efforts to provide safer ART protocols are still required in this direction.

Supplementary data

Supplementary data are available at Human Reproduction Update online.

Data availability

No new data were generated or analyzed in support of this research.

Authors’ roles

B.D. performed the literature search and data extraction. B.D. and P.F. drafted the manuscript. C.P. and J.T. provided critical revision and edited the article. All the authors approved the final version of the article.

Funding

This work was supported by funding from the Agence Nationale pour la Recherche (‘CARE’-ANR JCJC 2017).

Conflict of interest

The authors declare no conflict of interest.