Genetic landscape of human oocyte/embryo defects.

Conceptualization, L. Wang and Q. Sang; data curation, B.C. and Z.Y.; formal analysis, Q.L., T.W., J.W., and L.H.; methodology, Wenjie Wang, J.M., and Z.Z.; software, B.C., Z.Y., and Weijie Wang; investigation, H.G., H.F., Y.L., S.X., X.D., X.Z., R.H., and H.Z.; validation, F.X., D.G., H.J., R.Y., and Z.P.; resources, Y.K., J.S., X.S., Y.G., G.H., J.Z., J.F., L. Wu, Wenjing Wang, C.L., F.D., Q. Shi, L.L., S.X., D.L., X.D., and P.X.; visualization, B.C. and Weijie Wang; writing – original draft, B.C., Weijie Wang, Q. Sang, and L. Wang; Writing – review & editing, B.C., Weijie Wang, Q. Sang, and L. Wang; funding acquisition, L. Wang, Q. Sang, X.S., and Weijie Wang. All authors approved the final draft for submission.

We initially recruited 3,796 unrelated women diagnosed with primary infertility who had experienced at least two failed IVF/ICSI cycles caused by oocyte/embryo defects. To eliminate interference from other factors responsible for infertility, we conducted a careful screening of these patients as detailed in STAR Methods . A total of 3,627 patients were ultimately included in the case cohort ( Figure S1 ), including 1,554 cases (42.9%, 1,554/3,627) with oocyte defects, 248 cases (6.8%, 248/3,627) with abnormal fertilization, and 1,825 cases (50.3%, 1,825/3,627) with embryonic arrest ( Figures 1 and S2 A). Oocyte defects primarily included infertility resulting from the production of immature oocytes, including empty follicle syndrome, zona pellucida defects, oocyte maturation defects, morphological defects, and so forth. Abnormal fertilization was defined as the absence of normal zygotes (2PN) after IVF/ICSI, mainly manifested by multiple pronuclei or no pronuclei formation. Embryonic arrest was defined as the inability of embryos to develop into normal eight-cell embryos or blastocysts as well as recurrent implantation failures of viable embryos. Figure 1 Flowchart of genetic assessment of the case cohort with oocyte/embryo defects A total of 3,627 patients meeting the inclusion criteria were initially subjected to TUBB8 targeted screening. Those cases with negative results underwent WES for genetic diagnostic assessment. Flowchart of genetic assessment of the case cohort with oocyte/embryo defects A total of 3,627 patients meeting the inclusion criteria were initially subjected to TUBB8 targeted screening. Those cases with negative results underwent WES for genetic diagnostic assessment. To assess the genetic load of infertile women with oocyte/embryo defects, we evaluated the diagnostic yield in our case cohort by analyzing the known gene set, which consisted of 37 genes related to oocyte/embryo defects ( Figure 1 , Table S1 , and STAR Methods ). Among these known genes, 31 followed an autosomal recessive inheritance pattern, four followed an autosomal-dominant inheritance pattern, and two followed both autosomal-recessive and autosomal-dominant inheritance patterns ( Figure S2 B and Table S1 ). We first performed TUBB8 -targeted sequencing in our 3,627 cases and identified a total of 225 positive cases ( Table S2 ). Subsequently, after screening the WES data from the remaining 3,402 patients, we identified an additional 254 positive cases carrying variants of other known genes ( Table S2 ), indicating that these known genes accounted for approximately 13.2% (479/3,627) of infertile women who experienced IVF/ICSI failures due to oocyte/embryo defects ( Figure 2 A). Importantly, among the 479 positive cases, TUBB8 alone accounted for 47% (225/479) of all positive cases followed by PATL2 , which accounted for 9.4% (45/479) of all positive cases, suggesting that TUBB8 and PATL2 are the most commonly mutated genes in patients with primary infertility and at least two failed IVF/ICSI trials due to oocyte/embryo defects ( Figure 2 B). In our data, we identified five recurrent variants in TUBB8 and two recurrent variants in PATL2 (cases >5). Regarding the gene TUBB8 , the variant c.A10C was carried by up to 12 infertile women (12/255, 4.7%). For the gene PATL2 , the variant c.223-14_223-2del was the most common, with a total of 20 carriers (20/45, 44.4%) ( Table S2 ). In addition, we generated an integrated matrix of pathogenic variants identified in known genes and found that there was diversity in the clinical phenotypes, inheritance patterns, and variant types of the mutations identified in our case cohort ( Figure S3 ), indicating the complexity of the genetic mechanisms underlying IVF/ICSI failures caused by oocyte/embryo defects. Figure 2 Overview of the case cohort identified with known causative genes associated with oocyte/embryo defects (A) Overall diagnostic yield of the known causative genes in the 3,627 patients. (B) Proportional contribution of the top 20 genes among the 479 positive cases identified with 37 known causative genes. (C) Total positive rate and the positive rate of the top ten genes in cases with oocyte defects, abnormal fertilization, and embryonic arrest. (D) Schematic representation of the pathways associated with oocyte/embryo defects. The 37 known causative genes were largely categorized into eight processes: spindle assembly, maternal mRNA regulation, zona pellucida and sperm-egg interaction, meiotic double-strand break (DSB) formation and homologous recombination (HR), ion channels, mitochondrial function, subcortical maternal complex formation, and protein transport. These causative genes might be implicated in multiple processes with varied clinical phenotypes. Overview of the case cohort identified with known causative genes associated with oocyte/embryo defects (A) Overall diagnostic yield of the known causative genes in the 3,627 patients. (B) Proportional contribution of the top 20 genes among the 479 positive cases identified with 37 known causative genes. (C) Total positive rate and the positive rate of the top ten genes in cases with oocyte defects, abnormal fertilization, and embryonic arrest. (D) Schematic representation of the pathways associated with oocyte/embryo defects. The 37 known causative genes were largely categorized into eight processes: spindle assembly, maternal mRNA regulation, zona pellucida and sperm-egg interaction, meiotic double-strand break (DSB) formation and homologous recombination (HR), ion channels, mitochondrial function, subcortical maternal complex formation, and protein transport. These causative genes might be implicated in multiple processes with varied clinical phenotypes. In the oocyte defect group (1,554 cases), 24 known genes were involved in 272 positive cases with a diagnostic yield of 17.5% (272/1,554) ( Figures 2 C and S2 C). Oocyte maturation defect is one of the most typical oocyte defects, and TUBB8 , which encodes a primate-specific β-tubulin involved in human spindle assembly, was the first reported pathogenic gene shown to be responsible for oocyte maturation defects. 4 The contribution rate of TUBB8 to oocyte defect cases was approximately 10.0% (155/1,554) ( Figure 2 C). The primary phenotype of infertile women with TUBB8 variants is oocyte metaphase I arrest following a dominant inheritance pattern. Apart from TUBB8 , variants in five additional spindle-assembly-related genes— TUBA4A , KIF11 , HAUS6 , TACC3 , and KIF18A 3 , 10 , 11 —also resulted in oocyte defects (0.9%, 14/1,554) ( Table S3 ). This result highlights the important role of spindle-assembly regulatory genes in oocyte maturation. As disease-causing genes implicated in maternal mRNA homeostasis, 38 of 1,554 cases (2.4%) carried PATL2 variants and followed a recessive inheritance pattern ( Table S3 ). Moreover, 14 of 1,554 cases (0.9%) with variants in LHX8 , TBPL2 , ZFP36L2 , and PABPC1L were characterized by oocyte maturation defects due to disruption of maternal mRNA regulation ( Table S3 ). In addition, variants in the cell-cycle and checkpoint-related genes CDC20 , TRIP13 , and CDC23 also caused oocyte maturation defects in 0.8% (13/1554) of the cases ( Table S3 ). Some genes have also been shown to be responsible for unique phenotypes observed in IVF/ICSI failures. For example, PANX1 variants result in the “oocyte death” phenotype due to abnormal activation of the PANX1 channel, 12 whereas COX15 variants trigger oocyte ferroptosis and oocyte degeneration. 13 These two genes accounted for 0.9% of oocyte defect cases (14/1,554) ( Table S3 ). The zona pellucida is a translucent glycol-proteinaceous matrix that surrounds the mammalian oocyte and plays a critical role in sperm-egg binding during fertilization. 14 A total of 16 cases (1.0%, 16/1,554) carried variants in zona pellucida genes ( ZP1 , ZP2 , and ZP3 ), and these manifested as empty-follicle syndrome and as oocytes with no zona pellucida or with an abnormally thin zona pellucida ( Table S3 ). In the group of patients experiencing abnormal fertilization, 38 positive cases across 13 genes were identified, accounting for 15.3% (38/248) of cases ( Figures 2 C and S2 D). Notably, more than half (8.0%, 20/248) of the positive cases carried variants in TUBB8 and WEE2 ( Table S3 ), suggesting that these two genes can serve as primary genetic markers for diagnosing fertilization failure. In addition, variants in the cell-division-related genes CDC20 and CDC23 also resulted in fertilization failure in a small number of patients, while deficiency of ASTL caused multiple pronuclei after IVF. In this group, a total of five cases (2.0%, 5/248) carried variants in these three genes and exhibited the phenotype of either no pronuclei or multiple pronuclei ( Table S3 ). In the embryonic arrest group, a total of 23 causal genes accounted for 9.3% (169/1,825) of cases ( Figures 2 C and S2 E). Of note, among these 169 positive cases, 57 carried TUBB8 variants, accounting for 3.1% of the cases with embryonic arrest ( Figure S2 E). This demonstrates that TUBB8 also has the highest prevalence in cases with embryonic arrest. The subcortical maternal complex is crucial for embryonic developmental competence. 15 , 16 Among the 1,825 cases, a total of 41 (2.2%) carried variants in subcortical maternal complex-related genes ( TLE6 , PADI6 , OOEP , NLRP2 , NLRP5 , NLRP7 , and KHDC3L ) and exhibited the phenotype of embryonic arrest ( Table S3 ). In addition, variants in meiotic homologous recombination or double-strand break formation-related genes such as REC114 , TOP6BL , MEI1 , and MEI4 accounted for 0.9% (17/1,825) of cases with embryonic arrest ( Table S3 ). Moreover, variants in genes related to nuclear protein transport and the cell cycle, such as KPNA7 , MOS , and FBXO43 , were identified in 18 cases (1.0%, 18/1,825) with embryonic arrest ( Table S3 ). Zygotic cleavage failure is a unique embryonic arrest phenotype, primarily manifesting as zygotes that cannot complete the first cleavage and ultimately arrest at the one-cell stage. 17 , 18 BTG4 and CHEK1 were two main genetic factors associated with zygotic cleavage failure, accounting for 0.6% (11/1,825) of total cases with embryonic arrest ( Table S3 ). Taken together, these known causative genes highlight the crucial pathways that influence the developmental potential of oocytes and embryos ( Figure 2 D), including spindle assembly ( TUBB8 , TUBA4A , KIF11 , HAUS6 , TACC3 , and KIF18A ), maternal mRNA regulation ( PATL2 , LHX8 , BTG4 , TBPL2 , ZFP36L2 , and PABPC1L ), cell cycle and checkpoints ( WEE2 , CDC20 , CDC23 , CHEK1 , MOS , and FBXO43 ), zona pellucida and sperm-egg interaction ( ZP1 , ZP2 , ZP3 , and ASTL ), meiotic double-strand break formation and homologous recombination ( TRIP13 , REC114 , TOP6BL , MEI1 , and MEI4 ), ion channels ( PANX1 ), mitochondrial function ( COX15 ), subcortical maternal complex formation ( TLE6 , PADI6 , OOEP , NLRP2 , NLRP5 , NLRP7 , and KHDC3L ), and protein transport ( KPNA7 ). These known genes make up the current foundation for the genetic diagnosis of clinically infertile women caused by oocyte/embryo defects. Known genes accounted for approximately 13.2% of cases in this study, suggesting that there are other potential novel genes involved in the remaining cases. To further explore potential genetic factors associated with infertile women characterized by oocyte/embryo defects, we performed a case-control association analysis combined with gene set enrichment analysis including both dominant and recessive inheritance patterns ( Figure S4 and STAR Methods ). For the recessive model, bi-allelic variants of 10,953 genes in both the case and control cohorts were enriched and subsequently mapped to 4,903 reproduction-associated genes following gene prioritization ( Figure S4 and Table S4 ). Combined with transcriptome data from mouse primordial germ cells 19 and human oogenesis and embryogenesis 20 , 21 with fragments per kilobase per million mapped reads (FPKM) >10, a total of 122 candidate susceptibility genes were prioritized for functional enrichment analysis ( Table S5 ). The results showed that 91 out of the 122 genes were strongly enriched in Gene Ontology (GO) clusters associated with embryo development ( p = 1.4 × 10 −11 ), cell-cycle regulation ( p = 2.0 × 10 −10 ), DNA repair ( p = 3.3 × 10 −6 ), protein modification ( p = 1.3 × 10 −5 ), microtubule cytoskeleton organization ( p = 9.9 × 10 −4 ), chromosome segregation ( p = 1.1 × 10 −3 ), nuclear division ( p = 7.6 × 10 −3 ), and gene expression ( p = 6.7 × 10 −3 ) ( Figures 3 A and 3B). These results indicate that variants in these 122 candidate genes are implicated in oocyte/embryo defects and female infertility. Figure 3 Identification of novel susceptibility genes through population-based analyses and gene set enrichment analysis (A) Gene Ontology (GO) analysis of 122 genes by DAVID ( https://davidbioinformatics.nih.gov/ ), according to the filtering criteria of the gene set enrichment analysis. Most of the genes (91 out of 122) were strongly enriched in GO clusters related to multiple biological processes that may be involved in oogenesis and embryogenesis, including cell cycle, embryo development, protein modification, DNA repair, microtubule cytoskeleton organization, nuclear division, and gene expression. (B) Chord diagram of the GO clusters. The p values were obtained from population-based burden analysis following a recessive inheritance pattern using one-sided Fisher’s exact test. The representative genes are indicated in red font. (C) Distributions of altered amino acid residues across the proteins encoded by six novel disease-causing genes. Mutants that are likely to be in critical domains or motifs are indicated. Variant types are indicated as solid circles with different colors. The x axis represents the length of each protein sequence. Protein domains: CPC, chromosome passenger complex protein INCENP N-terminal domain; APK, inner centromere protein ARK-binding region; CM, coiled-coil motif; HATPase, histidine kinase-, DNA gyrase B-, and HSP90-like ATPase domain; PMS, DNA mismatch repair protein C-terminal domain; MutL_C, MutL C-terminal dimerization domain; TM, transmembrane region. Identification of novel susceptibility genes through population-based analyses and gene set enrichment analysis (A) Gene Ontology (GO) analysis of 122 genes by DAVID ( https://davidbioinformatics.nih.gov/ ), according to the filtering criteria of the gene set enrichment analysis. Most of the genes (91 out of 122) were strongly enriched in GO clusters related to multiple biological processes that may be involved in oogenesis and embryogenesis, including cell cycle, embryo development, protein modification, DNA repair, microtubule cytoskeleton organization, nuclear division, and gene expression. (B) Chord diagram of the GO clusters. The p values were obtained from population-based burden analysis following a recessive inheritance pattern using one-sided Fisher’s exact test. The representative genes are indicated in red font. (C) Distributions of altered amino acid residues across the proteins encoded by six novel disease-causing genes. Mutants that are likely to be in critical domains or motifs are indicated. Variant types are indicated as solid circles with different colors. The x axis represents the length of each protein sequence. Protein domains: CPC, chromosome passenger complex protein INCENP N-terminal domain; APK, inner centromere protein ARK-binding region; CM, coiled-coil motif; HATPase, histidine kinase-, DNA gyrase B-, and HSP90-like ATPase domain; PMS, DNA mismatch repair protein C-terminal domain; MutL_C, MutL C-terminal dimerization domain; TM, transmembrane region. To further demonstrate the reliability of these candidate susceptibility genes, we randomly selected genes with high expression as representative among each aforementioned enriched pathway for functional studies. Here, we prioritized the functional analysis of five of these genes, namely CNTD2 (cell cycle), SPDYC (cell cycle), DDOST (protein modification process), INCENP (cell cycle, nuclear division, chromosome segregation, and microtubule cytoskeleton organization), and MLH3 (DNA repair, cell cycle, nuclear division, and chromosome segregation) ( Figure 3 B, Table S6 , and STAR Methods ). The physiological and pathological mechanisms of other genes need to be deeply explored by combining in vivo and in vitro studies. Functional domain analysis revealed that most of the variants were located in the critical functional domain of the protein, suggesting that these variants have destructive effects on protein functions ( Figure 3 C). Importantly, we validated the roles of these candidate genes in female reproduction and assessed the pathogenicity of these variants through in vitro functional experiments or in vivo mouse models. CNTD2 and SPDYC are involved in crucial processes involving cyclin-dependent kinases during the phase transition of the mitotic cell cycle, 22 , 23 and in our case cohort we identified seven cases carrying bi-allelic variants in these two genes ( Table S6 ). Our functional experiment showed that overexpression of mutant CNTD2 (p.L59Wfs∗11, p.Y231N, and p.Y121Tfs∗14A) in mouse zygotes resulted in embryonic arrest ( Figure 4 A). In addition, SPDYC variants (p.R52Vfs∗50 and p.R188C) disrupted the ability of the protein to overcome milrinone-mediated inhibition of meiotic resumption in mouse germinal vesicle (GV) oocytes ( Figure 4 B). Figure 4 Experimental validation of novel susceptibility genes (A) Overexpression of CNTD2 variants in mouse zygotes resulted in early embryonic arrest. Representative images of embryos in different stages (left) and quantitative analysis of embryonic development rates by microinjection of mutant CNTD2 mRNA compared to wild type (WT) (right). (B) Overexpression of mutant SPDYC mRNA in mouse oocytes caused disruption of the ability of the protein to overcome milrinone-mediated inhibition of meiotic resumption. Representative images of oocytes undergoing germinal vesicle breakdown (GVBD) (left, red arrows) and quantitative analysis of GVBD rates by microinjection of mutant SPDYC mRNA compared to WT (right). (C) DDOST was knocked down in mouse oocytes using the Trim-Away method. Representative images of oocytes after extrusion of the first polar body (PB1) (left) and quantitative analysis of PB1 extrusion rates after knockdown of DDOST in mouse oocytes (right). (D) Incenp was knocked down in mouse oocytes by small interfering RNA (siRNA). Representative images of oocytes after extrusion of the first polar body (PB1) (left) and quantitative analysis of PB1 extrusion rates after knockdown of INCENP in mouse oocytes (right). Data are shown as the mean with standard deviation. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001, and ∗∗∗∗ p < 0.0001 by unpaired two-tailed Student’s t test. Scale bars, 100 μm. Experimental validation of novel susceptibility genes (A) Overexpression of CNTD2 variants in mouse zygotes resulted in early embryonic arrest. Representative images of embryos in different stages (left) and quantitative analysis of embryonic development rates by microinjection of mutant CNTD2 mRNA compared to wild type (WT) (right). (B) Overexpression of mutant SPDYC mRNA in mouse oocytes caused disruption of the ability of the protein to overcome milrinone-mediated inhibition of meiotic resumption. Representative images of oocytes undergoing germinal vesicle breakdown (GVBD) (left, red arrows) and quantitative analysis of GVBD rates by microinjection of mutant SPDYC mRNA compared to WT (right). (C) DDOST was knocked down in mouse oocytes using the Trim-Away method. Representative images of oocytes after extrusion of the first polar body (PB1) (left) and quantitative analysis of PB1 extrusion rates after knockdown of DDOST in mouse oocytes (right). (D) Incenp was knocked down in mouse oocytes by small interfering RNA (siRNA). Representative images of oocytes after extrusion of the first polar body (PB1) (left) and quantitative analysis of PB1 extrusion rates after knockdown of INCENP in mouse oocytes (right). Data are shown as the mean with standard deviation. ∗ p < 0.05, ∗∗ p < 0.01, ∗∗∗ p < 0.001, and ∗∗∗∗ p < 0.0001 by unpaired two-tailed Student’s t test. Scale bars, 100 μm. DDOST is a subunit of the oligosaccharyltransferase complex, and its functional defects lead to decreased N -glycosylation in humans, 24 while INCENP is an inner centromere protein that is responsible for regulating microtubule dynamics and chromosome segregation. 25 Biallelic variants in DDOST or INCENP were identified in a total of six cases ( Table S6 ). To investigate the functions of these two genes in oocyte development, we downregulated these two proteins in mouse oocytes and found that the loss of function of both proteins affected oocyte maturation ( Figures 4 C and 4D). Furthermore, immunofluorescence analysis revealed that DDOST variants (p.F47Lfs∗10, p.67-71del, and p.P226S) altered the subcellular localization of the protein in HeLa cells ( Figure S5 ). These in vitro results revealed the essential roles of these candidate genes in oocyte development and indicated the destructive effects of these variants. In addition, the candidate gene MLH3 is crucial for DNA mismatch repair during meiotic chromosome recombination, and Mlh3 −/− female mice exhibited infertility characterized by oocyte meiotic defects. 26 In our case cohort, we identified different MLH3 bi-allelic variants in six unrelated patients ( Table S6 ). The phenotypes of patients carrying variants in these five genes were presented in Table S6 . These in vitro and in vivo functional evaluations suggest that these recessive mutant genes are novel genes responsible for oocyte/embryo defects. In the dominant inheritance model, a total of 2,030 genes in our case cohort met the inclusion criteria, and these genes were subjected to collapsing analyses using one-sided Fisher’s exact tests ( STAR Methods ). As a result, a total of 19 heterozygous variants in TUBA1C were strongly enriched with the highest contribution rate (0.63%, 19/3,627) in the case cohort ( p = 1.5 × 10 −5 ) ( Figure S6 A). TUBA1C encodes an α-tubulin that is a major constituent of the microtubule cytoskeleton 22 and is highly expressed during both oogenesis and embryogenesis. Functional enrichment analysis also indicated that microtubule proteins such as TUBB8, TUBA4A, and TUBA1C can constitute a complete microtubule network ( Figure S6 B). These results indicate that TUBA1C might be a potential susceptibility gene associated with oocyte/embryo defects. In summary, a total of 123 novel candidate genes that might be involved in quality control of oocyte and embryo development following both dominant and recessive inheritance patterns were identified through gene set enrichment analysis. These novel candidate genes might provide additional genetic markers that will improve the diagnostic yield of infertile women suffering from IVF/ICSI failure due to oocyte/embryo defects.

Further information and requests for resources and reagents should be directed to and will be fulfilled by the lead contact, Lei Wang ( [email protected] ). No new unique reagents were generated in this study. Summary statistics of variant data are included in the supplemental information . The raw genotype data have been deposited in the Genome Variation Map ( https://ngdc.cncb.ac.cn/bioproject/ ) at the National Genomics Data Center, China National Center for Bioinformation/Beijing Institute of Genomics, Chinese Academy of Sciences, under project numbers PRJCA011994 , PRJCA012541 , and PRJCA027248 . It will be publicly available under protected access, as individual genomic data are protected due to patient privacy and Regulations on the Management of Human Genetic Resources of China. This paper does not generate original code. Any additional information required to reanalyze the data reported in this paper is available from the lead contact upon request.

In clinics, a number of patients experience recurrent failed IVF/ICSI attempts due to oocyte/embryo defects, and recent studies have provided strong evidence for the genetic basis of oocyte/embryo defects. Although some mutant genes were identified, the genetic landscape of human oocyte/embryo defects remains largely unknown. In this study, we clearly present that the reported mutant genes account for 13.2% (479 positive cases) of our 3,627 cases with oocyte/embryo defects, which is significantly higher than the proportion of variants in these genes observed in the control population (1.2%, 33/2,868) ( Figure S7 A). This demonstrates the enrichment of reported mutant genes in patients and strongly suggests the necessity for pursuing genetic diagnoses for corresponding clinical patients ( Figure 5 ). Because 3,627 cases were recruited from several clinical centers over a few years, we do not know the exact percentage of all IVF patients represented by these patients. Thus, the prevalence of genetics variants in all IVF patients cannot be evaluated and is worth investigating in the future. In addition, we systematically performed mutational screening of potential pathogenic genes in the remaining 3,148 cases and identified 123 novel candidate genes, five of which were verified by in vitro and in vivo functional studies. The candidate genes will greatly deepen our physiological and pathological understanding of human oocyte and embryo development. Figure 5 Clinical management of genetic diagnoses for infertile women with IVF/ICSI failures caused by oocyte/embryo defects According to the genetic screening results, clinicians can now understand female infertility with unexplained etiologies and can optimize the treatment strategy. Clinical management of genetic diagnoses for infertile women with IVF/ICSI failures caused by oocyte/embryo defects According to the genetic screening results, clinicians can now understand female infertility with unexplained etiologies and can optimize the treatment strategy. Our results provide a rationale for pursuing genetic diagnostic testing in patients who experience multiple failed IVF/ICSI attempts. It is worth noting that TUBB8 variants in our cohort were prevalent among infertile women with IVF/ICSI failures characterized by varied phenotypes, including oocyte maturation defects, abnormal fertilization, and embryonic arrest ( Figure 2 C). In contrast, there were some mutant genes that were mainly seen in specific groups. For example, PATL2 variants were strongly enriched in the oocyte defect group ( Figure 2 C). WEE2 variants and PADI6 variants were enriched among those with fertilization failure in the abnormal fertilization group and the embryonic arrest group, respectively ( Figure 2 C). In addition, PANX1 deficiency resulted in the oocyte death before or shortly after fertilization, 12 while the mutations in ZP1 , ZP2 , and ZP3 only caused defects of zona pellucida. 27 Abnormalities in oocyte meiosis and maturation lead to defects in oocyte development, fertilization, and embryonic development. Thus, although patients with IVF/ICSI failures can be generally divided into three subgroups (oocyte defects, abnormal fertilization, and embryonic arrest), there is some phenotypic heterogeneity in the patients in these subgroups. For example, in the subgroup of patients with oocyte defects, the majority of oocytes from these patients showed developmental defects, while some oocytes had seemingly normal morphology but still had abnormal fertilization or underwent embryonic arrest. 28 , 29 , 30 , 31 In addition, phenotypic heterogeneity was also observed in patients with different variants in the same genes. For example, some variants in TUBB8 caused typical oocyte maturation defects, while patients with other variants in TUBB8 showed phenotypic variability, including abnormal fertilization and embryonic arrest ( Figure S7 B). There are some possible explanations for phenotypic variability. First, differences in the degree of functional disruption caused by various variants might lead to phenotypic variations. TUBB8 variants with severe functional disruption cause failure of oocyte spindle assembly, thereby resulting in oocyte maturation defect. In contrast, oocytes carrying variants with modest functional disruption can assemble spindles, but the spindle assembly is abnormal, leading to abnormal cleavage or aneuploidy and ultimately causing fertilization failure or early embryonic arrest. Second, phenotypic characteristics can also be influenced by external factors such as stimulation strategies, lifestyles, and exposure to different environments. In addition, genetic modifiers might play a crucial role in regulating the expression of mutant TUBB8 and phenotypic diversity. Among the 123 novel candidate genes we identified, most of them appear to play critical roles in several molecular processes related to oogenesis and embryogenesis according to GO analysis, including cell cycle, embryo development, protein modification, DNA repair, microtubule cytoskeleton organization, chromosome segregation, nuclear division, and gene expression ( Figure 3 B). We provided functional evidence for the roles of five genes in human reproduction, namely CNTD2 , SPDYC , DDOST , INCENP , and MLH3 . Heterozygous variants in TUBA1C , which encodes a component of the microtubule cytoskeleton, 22 may impair oocyte spindle assembly and therefore lead to various phenotypes including oocyte defects, abnormal fertilization, and embryonic arrest ( Table S6 ). The cell-cycle regulators SPDYC and CNTD2 are assumed to play vital roles during mitosis in human cells, 22 , 23 but the two genes have unknown functions in human oocyte meiosis. Bi-allelic variants in SPDYC and CNTD2 may disrupt the meiotic cell cycle, ultimately resulting in human oocyte/embryo defects ( Table S6 ). The centromeric protein INCENP is involved in spindle assembly during oocyte meiosis in Drosophila , 32 and this suggests that compound heterozygous variants in INCENP may disrupt microtubule stabilization and spindle formation and thereby cause oocyte defects and fertilization failure ( Table S6 ). It was reported that deletion of Mlh3 , which encodes a DNA mismatch repair protein, leads to oocyte meiotic arrest in mice, 26 and homozygous and compound heterozygous variants in MLH3 may cause abnormal DNA repair during meiosis and result in human oocyte/embryo defects ( Table S6 ). In addition, a previous study showed that compound heterozygous variants in DDOST , a subunit of the N -glycosylation oligosaccharyltransferase complex, are associated with disorders of glycosylation. 24 Patients with homozygous and compound heterozygous variants in DDOST showed abnormal fertilization and embryonic arrest ( Table S6 ), indicating that abnormal protein N -glycosylation may also contribute to female infertility. These findings further demonstrate the crucial role of genetic factors and provide additional genetic markers and novel mechanisms for understanding human oocyte/embryo defects. Although we did not determine the bi-allelic status of compound heterozygous variants due to unavailability of DNA samples from the patients’ parents, extensive investigations into corresponding pathological mechanisms should therefore be undertaken in the future. In autosomal-recessive disorders, the phenotypes of somatic diseases are contributed by both the parental genomes. In contrast, the genetic factors underlying oocyte/embryo defects mainly derive from the maternal genome. The reason for this is that normal oocyte maturation is the key for successful fertilization and ensuring embryonic development. In addition, mammalian oocyte and early embryonic development can only utilize maternal mRNAs and proteins, since oocytes maintain transcriptional silence until zygotic genome activation (eight-cell embryo in humans). Therefore, oocyte/embryo defects caused by variants in these maternal genes cannot be rescued by the paternal genome after fertilization, as the zygotic genome has not yet been activated. This further indicates that the contributions of maternal and paternal genome are non-equivalent in oocyte/embryo defects. The paternal genome is an important component of the early embryonic genome after fertilization. In recent years, an increasing number of studies have focused on the impact of the paternal genome on early embryonic development, including paternal genome integrity, epigenetics, and genetic variations. 33 Previous studies have found that DNA fragmentation in morphologically normal sperm has a negative impact on embryonic development 34 and that sperm DNA fragmentation is associated with a higher miscarriage rate. 35 Furthermore, abnormal methylation of sperm DNA is associated with recurrent miscarriage and poor-quality embryos. 36 , 37 Additionally, studies have found that homozygous variants in ACTL7A carried by two infertile brothers cause early embryonic developmental arrest. 38 These studies highlight the critical role of the paternal genome in early embryonic development. Understanding the exact mechanisms of paternal genomic contribution will help identify new causes of early embryonic arrest and recurrent miscarriage in IVF/ICSI attempts. With a diagnostic yield of approximately 13.2% in our case cohort, the infertile causes of a significant portion of cases remains unknown. The following are possible explanations for undiagnosed cases. (1) We are currently focused on variants in coding regions, but variants in the non-coding regions, which account for 98% of the human genome, may play an important role in the pathogenesis of oocyte/embryo defects. Whole-genome sequencing is a powerful strategy for identifying novel genetic factors. (2) Epigenetic factors might play a potential role in oocyte/embryo defects, which is worthy of further exploration. (3) Non-genetic factors may contribute to oocyte/embryo defects, such as aneuploidy formation, IVF/ICSI procedures, and other technical limitations. With the extensive application of diverse research strategies and the continuous advancement of technologies, it is believed that more genetic factors of oocyte/embryo defects will be uncovered and female reproductive health will be improved in the future. In conclusion, our study provides a detailed genetic landscape of human oocyte/embryo defects. The identification of novel candidate genes not only broadens our mechanistic understanding of oocyte/embryo defects but also improves the genetic diagnostic yield in corresponding patients. These findings highlight the importance of utilizing WES to obtain genetic diagnoses for these patients ( Figure 5 ) and will lay the foundation for transforming the traditional clinical practice for failed IVF/ICSI attempts into genetic-based precision and personalized treatment for these patients. There are some limitations in our study. First, verification of independent samples was restricted due to unavailability of a large-scale cohort, including diverse cohorts or other populations. Second, the novel susceptibility genes show strong involvement in crucial pathways during oocyte and embryo development and should be further explored for experimental validation in the future. Third, this study lacks parental haplotypes and segregation analysis for many patients, due to the inability to collect parental samples. Therefore, the clinical pathogenicity of these novel susceptibility genes requires further investigation. Finally, the embryonic genome or paternal genome is also crucial for preimplantation embryonic development, and their contributions to early embryonic development need to be further explored in the future. In summary, this study depicts a comprehensive genetic landscape of human oocyte/embryo defects. Our findings highlight the importance of utilization of exome sequencing to assist genetic counseling and clinical practice for infertile women confronted with failed IVF/ICSI attempts characterized by oocyte/embryo defects.

Normal human oocyte and embryo development are required for successful human reproduction. The oocyte and embryo defects will result in oocyte maturation arrest, fertilization failure, early embryonic arrest, and, finally, infertility. 1 , 2 In contrast to the comparatively convenient evaluation of sperm competence, human oocyte and embryo competence cannot be assessed until attempts are made at in vitro fertilization (IVF) or intracytoplasmic sperm injection (ICSI). In clinics, a large number of infertile women showed oocyte/embryo defects and thereby underwent recurrent failure of IVF/ICSI. 3 However, the reason for human oocyte/embryo defects remained poorly understood for many years, troubling clinicians and leading to considerable economic and psychological burdens on patients. With the application of whole-exome sequencing (WES), we previously identified the mutant genes TUBB8 and PADI6 as being responsible for human oocyte maturation and embryonic arrest, 4 , 5 respectively, indicating that oocyte/embryo defects are Mendelian disorders and that genetic factors play important roles. Although some mutant genes have been identified subsequently, 1 , 6 the underlying genetic basis is still largely unknown, and several patients still experienced oocyte/embryo defects with unknown genetic reasons. It is estimated that the diagnostic yield of WES can explain from 10% to 40% of cases among a variety of genetic diseases, 7 , 8 , 9 and this strongly supports WES as a first-line diagnostic strategy in clinical practice. However, despite the discovery of a few pathogenic genes responsible for oocyte/embryo defects, there are no data clearly showing how many patients can be accounted for by these mutant genes, and the rate of genetic contribution remains unknown. This information is important for providing rationales for pursing genetic testing in these clinical patients. Thus, the aim of this study was to establish a genetic landscape of human oocyte/embryo defects and address the molecular diagnostic yield of known genes in infertile women with oocyte/embryo defects. Our study will provide a comprehensive understanding of the genetic basis of human oocyte/embryo defects and facilitates the clinical management of targeted diagnostic strategies and therapeutic interventions for recurrent IVF/ICSI failures.

The authors declare no competing interests.

REAGENT or RESOURCES SOURCE IDENTIFIER Cell lines , Reagents and vectors HeLa Cell Bank of Shanghai Institute for Biological Sciences Cat# SCSP-504 anti-GFP proteintech Cat# 66002-1-Ig; RRID: AB_11182611 anti-DDOST/OST48 Santa Cruz Cat# sc-74408; RRID: AB_1125745 Hoechst 33342 MedChemExpress Cat# HY-15559 pCMV6-Entry vector Origene Cat# PS100001 Fetal bovine serum Gibco Cat# 10099-141C KOD-Plus-Mutagenesis Kit TOYOBO Cat# SMK-101 AgeI/ScaII restriction enzymes New England BioLabs Cat# R3552S HiScribe T7 ARCA mRNA Kit New England BioLabs Cat# E2065S RNeasy MinElute Cleanup Kit Qiagen Cat# 217084 DMEM Gibco Cat# MA0212 Penicillin/streptomyci Gibco Cat# MA0110 PolyJet In Vitro DNA Transfection Reagent Signagen Cat# SL0068 M2 medium Nanjing Luanchuang Biotechnology Co. Cat# M01-B K-modifiedsimplex-optimized medium Nanjing Aibei Biotechnology Co. Cat# M1435 Human tubal fluid medium Nanjing Aibei Biotechnology Co. Cat# M1135 Deposited data Raw variants and genotypes data This paper. Genome Variation Map: PRJCA011994, PRJCA012541, PRJCA027248 HuaBiao Hao et al. 39 https://www.biosino.org/wepd/index DevOmics Yan et al. 21 http://devomics.cn/#/ GametesOmics Zhang et al. 20 http://gametesomics.cn/#/ Softwares and algorithms BWA (v.0.7.17) Li et al. 40 https://github.com/lh3/bwa/releases GATK (v.4.1.9.0) McKenna et al. 41 https://github.com/broadinstitute/gatk/releases PLINK (v.1.9) Purcell et al. 42 www.cog-genomics.org/plink/1.9/ ANNOVAR (v.2020-06-08) Wang et al. 43 https://annovar.openbioinformatics.org/ InterVar (ACMG/AMP 2015) Richards et al. 44 https://wintervar.wglab.org/ STRING (v.12.0) Szklarczyk et al. 45 https://string-db.org/ Cytoscape (v.3.10.2) Shannon et al. 46 https://cytoscape.org/ In this study, a total of 3,796 women diagnosed with primary infertility and experiencing at least two failed IVF/ICSI cycles caused by oocyte/embryo defects were recruited from 14 collaborating hospitals and reproductive centers. The recruitment process spanned ten years, from 2014 to 2023. To eliminate interference from other pathogenic factors, we conducted a more detailed screening of these patients. Ultimately, 3,627 patients met the screening criteria and formed our case cohort ( Figure S1 ). The inclusion criteria were as follows: (1) age younger than 40 years old; (2) primary infertility with unexplained etiology; (3) normal menstrual cycles, hormone levels, and chromosomes; (4) exclusion of male factors due to spermatogenic failure, hormonal disturbances, etc.; (5) IVF/ICSI failure due to oocyte defects, abnormal fertilization, or embryonic arrest, etc.; and (6) exclusion of other diseases that affect fertility such as ovarian dysfunction, fallopian tube lesions, endometriosis, endocrine disorders, etc. For the association analysis, the control cohort was composed of datasets of 1,944 unrelated female individuals from the HuaBiao project 39 and 924 in-house controls. All participants signed an informed consent and agreed to the use of their genetic data and clinical information for research. The DNA of all patients was extracted from their blood samples after they had signed the informed consent. The study was approved by the Ethics Committee of Shanghai Medical College, Fudan University. After TUBB8 targeted sequencing, WES datasets of patients with negative results were obtained using the Illumina platform (Illumina, San Diego, CA). After removing adapters and PCR duplicates, the clean FASTQ files were mapped to the human reference genome (GRCh37/hg19) using bwa-men (v.0.7.17), 40 and single nucleotide variants and indel variants were called using GATK (v.4.1.9.0). 41 To avoid including related participants or repeated samples, variants with a genotype call rate >95% were subjected to identity-by-descent analysis using PLINK 1.9. 42 To be included in the analysis, the variants had to pass recalibrated base quality, have a genotyping phred quality >20, have an alternative allele depth ≥10 with a read frequency ≥25%, and be part of an exonic or splice region. All qualified variants were annotated using the ANNOVAR pipeline, 43 including in silico prediction, population frequencies, and clinical significance. The pathogenicity of the variants in this study was manually determined according to the American College of Medical Genetics and Genomics/Association for Molecular Pathology (ACMG/AMP) guidelines. 44 The known causative genes responsible for oocyte/embryo defects in this study included mutated genes that were previously reported to be responsible for female infertility characterized by oocyte defects, abnormal fertilization, and embryonic arrest. We screened known genes in the PubMed and OMIM databases and then carefully evaluated the relevant studies in order to ensure that the pathogenicity of these deleterious variants had been validated through mouse models or in vitro functional studies. Ultimately, a total of 37 mutated genes were included in the list of known causative genes, and the relevant phenotypes for each gene are shown in Table S1 . To screen candidate genes, we performed gene set enrichment analysis. The gene set was a manually selected set of genes that might be related to the reproductive process with the following criteria: (1) genes involved in female gamete formation and embryonic development based on a literature review in PubMed; (2) genes related to female infertility/subfertility and embryonic lethality in the Mouse Genome Informatics database (MGI: http://www.informatics.jax.org/ ); (3) gene sets associated with cell division and embryogenesis according to the Molecular Signatures Database (MSigDB: https://www.gsea-msigdb.org/ ), including cell cycle (KEGG: hsa04110, GO:0016538) and embryo development (GO:0009790) pathways; and (4) other gene sets collected from the Gene Ontology Resource (GO: https://geneontology.org/ ) and Kyoto Encyclopedia of Genes and Genomes database (KEGG: https://www.genome.jp/kegg/ ), including those associated with oocyte meiosis (KEGG: hsa04114, GO:1900193, GO:0090306), DNA replication (KEGG: hsa03030), DNA damage repair (GO:0000077, GO:0006281), oxidative phosphorylation (KEGG: hsa00190), kinetochore organization (GO:0000776, GO:0051382, GO:0051383), centrosome cycle (GO:0046605), and microtubule cytoskeleton (GO:0007021, GO:0015631, GO:0005874), etc. In addition, in order to improve gene prioritization we manually collated a total of 4,903 reproduction-associated genes ( Table S4 ) and integrated these with the transcriptome profiles of mouse primordial germ cells, 19 human oogenesis, 20 and human embryogenesis. 21 For the gene-level phasing analysis, we performed both dominant and recessive pattern analyses. The dominant pattern included missense variants in both the case and control cohorts that were predicted to be “pathogenic” or “likely pathogenic” (P/LP) using a combination of several algorithms 47 as well as protein-truncating variants (including frameshift, stop-gain, start-loss, and splicing site mutations). The minor allele frequency (MAF) of the deleterious variants was defined as a cutoff of 0.1% in both the East Asian population and the total population of the genome aggregation database (gnomAD v2.1.1: https://gnomad.broadinstitute.org/ ). Genetic burden analysis for rare P/LP variants was collapsed to establish the association with the phenotype of female infertility using one-sided Fisher’s exact tests. In addition, for the recessive pattern, homozygous or compound heterozygous variants in both the case and control cohorts were enriched with an MAF cutoff of 1% in both the East Asian population and the total population of the gnomAD database. One-sided Fisher’s exact tests were performed by tabulating the co-occurrence number between cases and controls, and the adjusted p -value of the burden analysis was then calculated for multiple testing using the Bonferroni correction method. The coding sequences of the human CNTD2 , SPDYC , and DDOST genes were cloned into the pCMV6-Entry vector containing the C-FLAG/C-GFP tag. Site-directed mutagenesis was performed to introduce the identified variants into the wild-type (WT) vector according to the instructions of the KOD-Plus-Mutagenesis Kit (Toyobo Life Science). WT and mutant plasmids were confirmed by Sanger sequencing. For the in vitro transcription of CNTD2 and SPDYC mRNAs, WT and mutant plasmids were linearized by digestion with the AgeI/ScaII restriction enzymes (New England BioLabs). Linearized WT and mutant plasmids were used as templates for in vitro mRNA transcription using the HiScribe T7 ARCA mRNA Kit (New England BioLabs). Finally, the mRNA product was purified and dissolved in nuclease-free water using the RNeasy MinElute Cleanup Kit (Qiagen). HeLa cells (Cell Bank of Shanghai Institute for Biological Sciences) were cultured in DMEM (Gibco) supplemented with 10% fetal bovine serum (Gibco) and 1% penicillin/streptomycin (Gibco) in a 5% CO 2 atmosphere at 37°C. The WT and mutant DDOST -GFP plasmids were transfected into HeLa cells using the PolyJet In Vitro DNA Transfection Reagent (Signagen) according to a standard protocol. For cell imaging, the transfected cells were washed three times in PBS after culturing for 36 h. The cells were then incubated with Hoechst 33342 (MedChemExpress) for 15 min at room temperature. Images were captured with an LSM 880 confocal microscope (Zeiss). GV oocytes were isolated from the ovaries of ICR female mice (7–8 weeks old) by puncturing the antral follicles with a fine needle. To obtain embryos, female mice were intraperitoneally injected with 10 IU of PMSG followed by 10 IU of hCG 46 h later. After an additional 13 h, oocyte/cumulus masses were isolated from the oviducts and were mixed with sperm in human tubal fluid medium (Nanjing Aibei Biotechnology Co.) at 37°C and 5% CO 2 for IVF. The zygotes formed about 6 h after fertilization. For microinjection, GV oocytes were injected with WT and mutant SPDYC mRNA (500 ng/μL) and then cultured in M2 medium with 2.5 μM milrinone at 37°C for 7 h to assess the GVBD rate in each group. In addition, zygotes were injected with WT and mutant CNTD2 mRNA (1000 ng/μL) and then cultured in K-modified simplex-optimized medium (Nanjing Aibei Biotechnology Co.) at 37°C and 5% CO 2 . Embryonic development was monitored in each group. In order to assess the impact of INCENP and DDOST deficiency on oocyte maturation, we downregulated INCENP and DDOST using siRNA and Trim-Away, respectively. The maturation rate of oocytes was then evaluated through in vitro maturation culture. All animal experiments were approved by the Shanghai Medical College of Fudan University.

We thank the HuaBiao project: Whole-Exome Database of Han Chinese for providing some of the data on the control cohort used in this study. We also thank all the patients and physicians from the different hospitals who participated in our research. This study was supported by the 10.13039/501100001809 National Natural Science Foundation of China ( 82325021 to Q. Sang, 82288102 to L. Wang, 32130029 to L. Wang, 82171643 to Q. Sang. 82471885 to Weijie Wang, and 82171644 , to X.S.), the Fund of Fudan University and Cao’ejiang Basic Research ( 24FCB01 to L. Wang), and the New Cornerstone Science Foundation through the XPLORER PRIZE (to L. Wang). L. Wang is a SANS Exploration Scholar.