Genetic Insights into Intestinal Microbiota and Risk of Infertility: A Mendelian Randomization Study.

The intestinal microbiota has been regarded as the second human genome, mediating the development of various conditions [ 1 ]. The intestinal microbiome has many important roles in digestion, drug metabolism, immunity training, endocrine regulation, and so on, which may affect the host’s health through different pathways [ 2 ]. Currently, abnormal composition of intestinal microbiota has become a significant feature shared by many different pathogenesis [ 3 ]. It is suggested that gut microbiota might be a risk or preventive factor for many diseases, including metabolic conditions and cancer [ 3 , 4 ]. Meanwhile, gut microbiota is reportedly associated with multiple gynecologic conditions and female infertility [ 5 ]. Moreover, the metabolites of the gut microbiota may affect testicular function and male fertility [ 6 ]. However, the role of intestinal microbiota in infertility is less studied. Infertility is a prevalent and intricate reproductive disorder that affects approximately 8–12% of couples worldwide [ 7 ]. This disease can have profound emotional, physical, and social impacts on couples, leading to a heavy psychological burden [ 8 ]. Most infertile couples have identifiable causes, such as ovulatory dysfunction, male factor infertility, and tubal disease; however, approximately 15% of infertile couples have unexplained causes [ 9 ]. Thus, understanding the causes of infertility and exploring the causal link between risk factors and infertility are crucial for affected couples and healthcare professionals. In all, it has been documented that microbial dysbiosis of the genital tract incurred by various factors represents a risk factor for infertility [ 10 ]. However, the causal association of gut microbiota with infertility remains unclear. Theoretically, a randomized controlled trial (RCT) is the best method to establish a causal relationship between exposure and disease. However, it is not realistic to explore the causal association of intestinal microbiota with infertility using an RCT because of too many objective limitations and confounding factors. As a novel epidemiological method, Mendelian randomization (MR) uses single nucleotide polymorphisms (SNPs) as instrumental variables (IVs) to replace exposures and outcomes. This method has been widely used for causal inferences, avoiding confounding bias and reverse causality from a limited sample size and cross-sectional design [ 11 , 12 ]. Thus, we performed this two-sample MR analysis to explore the causal link between intestinal microbiota and infertility utilizing the genome-wide association study (GWAS) summary statistics.

The flowchart of this study is shown in Figure 1 . Summary statistics on human intestinal microbiota were obtained from the MiBioGen consortium [ 13 ]. As a retrospective design, this study included 18,340 individuals from 24 cohorts. The microbial composition and relative abundances of the intestinal microbiota were detected by using 16S ribosomal RNA gene sequencing. A microbiota quantitative trait loci mapping analysis was conducted to identify genetic variants in the host and evaluate the association between human genetic variants and intestinal microbiota. In this study, the lowest taxonomic level was genus. Furthermore, GWAS summary statistics of infertility (male and female) were retrieved from FinnGen ( https://r9.finngen.fi/ ). The summary statistics for male infertility included 1271 cases and 119,297 controls, and those for female infertility included 13,142 cases and 107,564 controls ( Figure S1 ). The diagnosis of infertility was based on the International Classification of Diseases (ICD) codes. The diagnostic codes were 606 (ICD-8 and ICD-9) and N46 (ICD-10) for male infertility and N97 (ICD-10) for female infertility ( Figures S2 and S3 ). All the participants were of European descent. To obtain adequate IVs and increase the statistical power, IVs were filtered from the identified SNPs at a genome-wide statistical significance of p < 1 × 10 −5 , as previous studies did [ 14 ]. The left SNPs were further pruned if the linkage disequilibrium r 2 was ≥0.01 at a window size of 10,000 kb. SNPs with a minor allele frequency (MAF) < 0.01 are generally accepted as rare SNPs, which have limited impact on the traits. Therefore, only SNPs with MAF ≥ 0.01 were reserved. Six methods were used to explore the causal effects of intestinal microbiota on infertility, including inverse variance weighted (IVW), MR-Egger, weighted median, maximum likelihood (ML), MR robust adjusted profile score (MR.RAPS), and MR pleiotropy residual sum and outlier (MR-PRESSO). In IVW models, all IVs are assumed to be valid and then combined using a meta-technique. This method is used as the main analysis due to its high statistical power. When heterogeneity existed, the random effects IVW model was used; otherwise, the fixed effects IVW estimator was adopted as the main analysis. To generate unbiased results even when pleiotropy existed, MR-Egger and weighted median methods relaxed this assumption and were used as sensitivity analyses. In addition, the ML method was also used for its minimal bias in limited sample sizes. The MR.RAPS method could produce consistent results when weak and pleiotropic SNPs exist. After excluding pleiotropic outliers, the MR-PRESSO estimator combined the effects from IVs into the IVW model. This estimator could be used to detect the presence of pleiotropy. The recalculated genetic association was used as the outcome data in the sensitivity analysis. The Cochran Q test was applied to evaluate the heterogeneity of IVs. Q statistics with a p -value < 0.05 indicated the presence of heterogeneity, and the random-effects IVW method was used to generate more conservative but robust estimates. To assess the horizontal pleiotropy, the MR-Egger intercept term and the global test from the MR-PRESSO estimator were used. The strength of SNPs was quantified by calculating the F-statistics of each bacterial taxon, as previously reported [ 15 ]. A F-statistic greater than 10 indicated a lower likelihood of weak instrumental bias. All statistical analyses were conducted using R 4.0.3 (R Foundation for Statistical Computing, Vienna, Austria). “TwoSampleMR”, “mr.raps”, and “MR-PRESSO” packages were used for data analyses. The “forest” package was used to draw the forest plot.

Two germs showed protective effects on female fertility ( Figure 2 ). A total of 13 SNPs were used as IVs in the Family XIII AD3011 group , with an average F statistic of 21.47 ( Table 1 ). A standard deviation (SD) increment in the abundance of the genetically proxied Family XIII AD3011 group led to a reduced risk of female infertility (OR: 0.87, 95% CI: 0.77–0.99, p for IVW = 0.041) ( Figure 2 ). The results remained significant when ML and MR-PRESSO were used (all p < 0.05). Meanwhile, the direction and effect size of other estimators were similar to IVW ( Figure 2 ). The scatter plot displayed a decreasing risk of female infertility with the increase of the SNP effect on the Family XIII AD3011 group ( Figure 3 D). The Ruminococcaceae NK4A214 group is a genus of Ruminococcaceae , and a total of 13 SNPs were used as IVs of this germ with an average F statistic of 21.66 ( Table 1 ). The IVW estimator detected a protective effect of the Ruminococcaceae NK4A214 group against female infertility (OR: 0.85, 95% CI: 0.74–0.98, p for IVW = 0.021) ( Figure 2 ). This result remained significant in ML, RAPS, and MR-PRESSO (all p < 0.05). The scatter plots showed decreased risks of female infertility with the increase of the SNP effect on the Ruminococcaceae NK4A214 group ( Figure 3 F). Moreover, no pleiotropy or heterogeneity were detected for these two germs (all p > 0.05) ( Figure 4 , Table 1 and Table 2 ). Four germs showed adverse effects on female fertility ( Figure 2 ). There were 10 SNPs used as IVs in Betaproteobacteria with an average F statistic of 21.78 ( Table 1 ). The IVW estimator revealed an adverse effect of Betaproteobacteria on female fertility (OR: 1.18, 95% CI: 1.01–1.38, p for IVW = 0.037) ( Figure 2 ). The results were still significant in ML, RAPS, and MR-PRESSO (all p < 0.05), and the direction of other estimators was consistent with IVW. With the increase in SNP effects on Betaproteobacteria , the scatter plot showed an increasing risk of female infertility ( Figure 3 A). At the order level, a total of 10 SNPs were used as IVs of Burkholderiales , with an average F statistic of 22.27 ( Table 1 ). The IVW estimator showed that increased abundance of genetically proxied Burkholderiales led to an increased risk of female infertility (OR: 1.18, 95% CI: 1.01–1.38, p for IVW = 0.034) ( Figure 2 ). The results remained significant when ML, RAPS, and MR-PRESSO were used (all p < 0.05), and the findings of other estimators did not reveal discordance. Genetically proxied Burkholderiales abundance was positively related to the risk of female infertility ( Figure 3 B). No heterogeneity or pleiotropy in MR analyses were observed (all p > 0.05) ( Figure 4 , Table 1 and Table 2 ). A total of 9 SNPs were used as IVs in Candidatus Soleaferrea , with an average F statistic of 20.78 ( Table 1 ). The IVW estimator revealed a potentially causal association of Candidatus Soleaferrea with female infertility (OR: 1.12, 95% CI: 1.01–1.25, p for IVW = 0.048) ( Figure 2 ). This finding remained consistent in ML, RAPS, and MR-PRESSO. With the increase in SNP effect on Candidatus Soleaferrea , the effect of SNP on infertility ascended ( Figure 3 C). At phylum level, there were 9 SNPs used as IVs in Lentisphaerae with an average F statistic of 22.05 ( Table 1 ). At phylum level, the causal relationship between Lentisphaerae and female infertility was identified (OR: 1.11, 95% CI: 1.01–1.20, p for IVW = 0.044) ( Figure 2 ). The directions of other estimators were consistent with IVW. The scatter plots displayed increased risks of female infertility with the increase in SNP effects on Lentisphaerae ( Figure 3 E). Additionally, no pleiotropy or heterogeneity were detected for these germs (all p > 0.05) ( Figure 4 , Table 1 and Table 2 ). Three germs showed protective effects on male fertility. At family level, there were 8 SNPs used as IVs in Bacteroidaceae with an average F statistic of 22.29 ( Table 1 ). The IVW estimator revealed a protectively causal effect of Bacteroidaceae against male infertility (OR: 0.57, 95% CI: 0.33–0.96, p for IVW = 0.036) ( Figure 5 ). This finding was supported by ML, RAPS, and MR-PRESSO (all p < 0.05). The scatter plots showed that with the increment of SNP effects on the abundances of Bacteroidaceae , SNP effects on male infertility decreased ( Figure 6 B). No pleiotropy was identified by the global test and MR-Egger method (all p > 0.05) ( Table 1 ), and no heterogeneity was detected by the Cochran Q test according to the funnel plot (all p > 0.05) ( Figure 7 , Table 2 ). At genus level, a total of 8 SNPs were used as IVs in Bacteroides with an average F statistic of 22.29 ( Table 1 ). The protective effect of genetically proxied Bacteroides on male fertility was identified by the IVW estimator (OR: 0.57, 95% CI: 0.33–0.96, p for IVW = 0.036) ( Figure 5 ). This causality was further supported by ML, RAPS, and MR-PRESSO (all p < 0.05). Scatter plots visualizing the increased SNP effect on Bacteroides and the decreased SNP effect on male infertility were displayed ( Figure 6 C). No pleiotropy or heterogeneity was detected by the global test, the MR-Egger method, or the Cochran Q test (all p > 0.05) ( Figure 7 , Table 1 and Table 2 ). There were 13 SNPs used as IVs in the Ruminococcaceae NK4A214 group with an average F statistic of 21.66 ( Table 1 ). The IVW estimator disclosed the protective effect of the Ruminococcaceae NK4A214 group on male fertility (OR: 0.61, 95% CI: 0.39–0.97, p for IVW = 0.037). This finding was consistent with ML, RAPS, and MR-PRESSO (all p < 0.05). Of note, the protective role of the Ruminococcaceae NK4A214 group was also detected in female infertility, suggesting that Ruminococcaceae NK4A214 may be a core genus affecting fertility. All directions and effect sizes of the MR-Egger and weighted median for the three above-mentioned germs were in accordance with the IVW estimator. The scatter plots showed decreased risks of male infertility with the increase of the SNP effect on the Ruminococcaceae NK4A214 group ( Figure 6 D). No pleiotropy was detected by the global test and MR-Egger method (all p > 0.05) ( Table 1 ), and no heterogeneity was identified by the Cochran Q test according to the funnel plot (all p > 0.05) ( Figure 7 , Table 2 ). There was only one germ that had an adverse causal effect on male fertility. A total of 13 SNPs were used as IVs of Anaerotruncus , with an average F statistic of 20.84 ( Table 1 ). With an increase in SD in the abundance of genetically proxied Anaerotruncus , an increased risk of male infertility was detected (OR: 1.81, 95% CI: 1.14–0.99, p for IVW = 0.011) ( Figure 5 ). This causality was also detected by ML, RAPS, and MR-PRESSO (all p < 0.05). The directions of other estimators were consistent with IVW. The scatter plots displayed increased risks of male infertility with the increase in SNP effects on Anaerotruncus ( Figure 6 A). No pleiotropy was identified by the global test and MR-Egger method, and the Cochran Q test detected no heterogeneity according to the funnel plot ( p > 0.05) ( Figure 7 , Table 1 and Table 2 ).

Infertility is a major health issue in the world, which will lead to a severe decline in child births and considerable population declines [ 16 ]. Meanwhile, infertility is a multi-factorial pathological condition with high complex and heterogeneous etiologies [ 17 ]. Although many risk factors have been documented, the interaction between gut microbiota and infertility requires further investigation. As the second genome of humans, the gut microbiota plays an important role in the pathogenesis of many diseases [ 18 ]. The gut microbiota may participate in the development of many conditions by producing microbial metabolites and subsequently activating downstream signaling pathways, which eventually change cellular function and initiate the abnormality [ 19 ]. Currently, the relationship between gut microbiota and infertility remains unclear. Here, we used genetic data to explore the causal link between intestinal microbiota and infertility in order to avoid bias from confounding factors. In our MR analysis, protective effects of the Family XIII AD3011 group and the Ruminococcaceae NK4A214 group on female fertility were identified, while Betaproteobacteria , Burkholderiales , Candidatus Soleaferrea , and Lentisphaerae showed adverse effects on female fertility. At the same time, Bacteroidaceae , Bacteroides , and Ruminococcaceae NK4A214 group revealed protective effects on male fertility, while an adverse causal association between Anaerotruncus and male fertility was detected. This study identified causal links between intestinal microbiota and female or male infertility, offering novel insights into the assessment of potential causes of infertility. At the genus level, previous evidence supporting the protective role of the Family XIII AD3011 group in female fertility is scarce. In a case-control study, the abundance of the Family XIII AD3011 group was found to be related to several markers of polycystic ovary syndrome (PCOS) [ 20 ]. Meanwhile, it is reported that the Family XIII AD3011 group was negatively correlated with adipic acid, affecting the metabolic phenotype and host inflammation [ 21 ]. Thus, the protectively causal effect of the Family XIII AD3011 group on female fertility may be involved in the inhibition of PCOS-related molecular expression and inflammatory signaling, which may impair the function of reproductive organs. The Ruminococcaceae NK4A214 group is a genus of Ruminococcaceae . It should be noted that the Ruminococcaceae NK4A214 group in our study showed protective influences on both female and male fertility, suggesting a potentially core role for the Ruminococcaceae NK4A214 group in the pathogenesis of infertility. Reduced abundance of the Ruminococcaceae NK4A214 group may result in abnormal spermatogenesis through reduced bile acid levels and vitamin A absorption [ 22 ]. For female infertility, significantly decreased Ruminococcaceae were seen in the PCOS group compared with healthy controls [ 23 ]. Meanwhile, increasing commensal Ruminococcaceae via fecal microbiota transplantation (FMT) may improve the ovarian function of aged mice by inhibiting pro-inflammatory interferon (TNF)-γ signaling and promoting anti-inflammatory interleukin (IL)-4 signaling [ 24 ]. Taken together, the Ruminococcaceae NK4A214 group might have a crucially protective effect against infertility through increasing vitamin A absorption, inhibiting inflammatory signaling, and PCOS-related genetic expression. Betaproteobacteria are a class of Proteobacteria . At the phyla level, a higher abundance of Proteobacteria in peritoneal fluid could be seen among infertile patients with endometriosis, in which several inflammatory factors, including IL-6, IL-10, IL-13, and TNF-α, may be involved [ 25 ]. Moreover, it is suggested that Betaproteobacteria and Proteobacteria might mediate the effects of estrogen on peripheral organs [ 26 ]. Thus, Betaproteobacteria may adversely affect female fertility through inflammatory and estrogen signaling pathways, impairing the function of reproductive organs and interfering with processes such as ovulation, implantation, and embryonic development. Burkholderiales are a class of bacteria known to have both beneficial and pathogenic effects on human health [ 27 ]. A molecular analysis showed that Burkholderiales was detectable in the endometrial fluid of infertile women and was negative in fertile controls [ 28 ]. Meanwhile, some genera of Burkholderiales might be associated with cystic fibrosis and could cause a broad range of infections in hosts [ 29 ]. Thus, we hypothesize that inflammatory signaling and fibrotic mechanisms might underlie the causal association of Burkholderiales with female infertility. At the genus level, Candidatus Soleaferrea was observed to be related to several autoimmune diseases via the effects of metabolites and the immune-gut axis [ 30 ]. However, the evidence related to the link between Candidatus Soleaferrea and female infertility is scarce. Whether the causal effect of Candidatus Soleaferrea on infertility depends on the immune-gut axis needs further investigation. It is reported that patients with lung cancer had higher levels of Lentisphaerae than the healthy controls, and certain specific bacteria were correlated with serum inflammatory indicators in those patients, indicating the potential links between Lentisphaerae and systemic immunity and inflammation in hosts [ 31 ]. Meanwhile, a decreased abundance of Lentisphaerae was detected in patients with autoimmune hepatitis compared to healthy controls [ 32 ]. These results suggested that Lentisphaerae and its metabolites might be associated with abnormal immune responses or suppression, which may potentially impact fertility-related processes. Bacteroides is a type genus of the family Bacteroidaceae . In our MR analysis, both Bacteroides and Bacteroidaceae showed protective effects on male fertility, indicating the potential role of Bacteroides in the management of infertile males. It has been found that Bacteroides might affect the iron metabolism in infertile females, highlighting the links between Bacteroides and reproductive function [ 33 ]. Several species of Bacteroides were detected to be correlated with non-obstructive azoospermia in male infertility [ 34 ]. Meanwhile, fecal microbiota transplantation may improve the semen quality of infertile males with a high-fat diet by increasing gut Bacteroides , improving liver function, and ameliorating the testicular microenvironment [ 35 ]. All in all, the influence of Bacteroides on male fertility might be positive, in which the gut microbiota-testis axis and spermatogenesis signaling may be involved. Anaerotruncus is a genus assigned to the family Ruminococcaceae . As a conditional pathogenic bacterium, Anaerotruncus has an important role in the maintenance of microbial diversity, which is crucial to host energy homeostasis affected by a high-fat/sucrose diet [ 36 ]. Meanwhile, Anaerotruncus was reportedly associated with the formation of hepatocellular carcinoma related to non-alcoholic fatty liver disease [ 37 ]. Similarly, decreased abundance of Anaerotruncus was considered to mediate the therapeutic effect of liraglutide on fatty liver in db/db mice [ 38 ]. Thus, the adverse effect of Anaerotruncus on male fertility may be due to disrupted energy homeostasis and glucolipid metabolism, which might interfere with male fertility-related processes. Several limitations in this study should be noted when explaining the results. Firstly, the trans-ethnic population (mainly Europeans) was included in the dataset to obtain the genetic links between intestinal microbiota and SNPs. The mixed ethnic background may produce bias to the genetic association. Further identification with a simpler ethnic background and a larger sample size is necessary. Secondly, the abundance of gut microbiota is easily affected by many environmental factors, such as diet and drugs. Meanwhile, the dataset originates from 24 cohorts in different countries. Therefore, different environmental factors among different cohorts may produce significant heterogeneity and bias the genetic association. Thirdly, due to the limitations of summary-level statistics, the combined role of several risky germs cannot be determined, which may need further investigation. Additionally, the etiology and risk factors of infertility are complex and often mixed, and most links between gut microbiota and different causes remain unknown. In this study, the causes of infertility were diagnosed by ICD codes as in other Mendelian randomization studies, but the results were not adjusted for documented causes or etiologies, such as hormonal profiles, karyotype, and spemogram data, which may affect the conclusion and should be investigated in the future.

In this study, we found that the genetically proxied Family XIII AD3011 group , Ruminococcaceae NK4A214 group , Betaproteobacteria , Burkholderiales , Candidatus Soleaferrea , and Lentisphaerae had potentially causal effects on female infertility. Meanwhile, genetically proxied Bacteroidaceae , Bacteroides , Ruminococcaceae NK4A214 group , and Anaerotruncus had potentially causal effects on male infertility. Among them, the Ruminococcaceae NK4A214 group was found to be the core germ, which has a protective effect on both female and male infertility. Further investigations are needed to identify the biological mechanisms linking gut microbiota and infertility.