Characterisation and comparison of semen microbiota and bacterial load in men with infertility, recurrent miscarriage, or proven fertility

Abstract

Several studies have associated seminal microbiota abnormalities with male infertility but have yielded differing results owing to their limited sizes or depths of analyses. The semen microbiota during recurrent pregnancy loss (RPL) has not been investigated. Comprehensively assessing the seminal microbiota in men with reproductive disorders could elucidate its potential role in clinical management. We used semen analysis, terminal-deoxynucleotidyl-transferase-mediated-deoxyuridine-triphosphate-nick-end-labelling, Comet DNA fragmentation, luminol reactive oxidative species (ROS) chemiluminescence, and metataxonomic profiling of semen microbiota by 16S rRNA amplicon sequencing in this prospective, cross-sectional study to investigate composition and bacterial load of seminal bacterial genera and species, semen parameters, ROS, and sperm DNA fragmentation in men with reproductive disorders and proven fathers. 223 men were enrolled, including healthy men with proven paternity (n=63), the male partners in a couple encountering RPL (n=46), men with male factor infertility (n=58), and the male partners of couples with unexplained infertility (n=56). Rates of high sperm DNA fragmentation, elevated ROS, and oligospermia were more prevalent in the study group compared with control. In all groups, semen microbiota clustered into three major genera-dominant groups (1, Streptococcus; 2, Prevotella; 3, Lactobacillus and Gardnerella); no species clusters were identified. Group 2 had the highest microbial richness (p<0.001), alpha-diversity (p<0.001), and bacterial load (p<0.0001). Overall bacterial composition or load has not been found to associate with semen analysis, ROS, or DNA fragmentation. Whilst global perturbation of the seminal microbiota is not associated with male reproductive disorders, men with unidentified seminal Flavobacterium are more likely to have abnormal seminal analysis. Future studies may elucidate if Flavobacterium reduction has therapeutic potential.

Keywords: fertility; human; medicine; microbiota; semen.

Figure 1.. Characterisation of semen microbiota composition at genera level.

(A) Heatmap of Log10 transformed read counts of top 10 most abundant genera identified in semen samples. Samples clustered into three major microbiota groups based mainly on dominance by Streptococcus (Cluster 1), Prevotella (Cluster 2), or Lactobacillus and Gardnerella (Cluster 3) (n=223, Ward’s linkage). (B) Relative abundance of the top 6 most abundant genera within each cluster. (C) Silhouette scores of individual samples within each cluster. (D) Species richness (p<0.0001; Kruskal-Wallis test) and (E) alpha-diversity (p<0.0001; Kruskal-Wallis test) significantly differed across clusters. (F) Assessment of bacterial load using qPCR showed Clusters 2 and 3 have significantly higher bacterial loads compared to Cluster 1. Dunn’s multiple comparison test was used as a post hoc test for between-group comparisons (*p<0.05, ****p<0.0001).

Figure 1—figure supplement 1.. Genera-level categorisation of seminal microbiota identified three major clusters using average silhouette scores for number of clusters.

Figure 2.. Co-occurrence network estimated with SparCC from 16S sequencing counts at species level.

Network representing co-occurrence patterns (edges), between various taxonomic units, assigned at species level (nodes). Edges are coloured by their estimated SparCC correlation coefficient (ρ). Edges with a SparCC bootstrapped p-value<0.05, ρ<0.25, and singleton nodes are not shown. Node colour represents network community membership. Node sizes are proportional to the mean relative abundance of their respective taxon.

Figure 2—figure supplement 1.. Characterisation of semen microbiota composition at species level.

(A) Heatmap of Log10 transformed read counts of top 25 most abundant species identified in semen samples. Samples clustered into 15 microbiota groups. (B) Silhouette scores of individual samples in microbial groups. (C) Average silhouette scores for 15 clusters at species level.

Figure 3.. Relative abundance and prevalence matrices of Flavobacterium in relation to semen quality and morphology.

(A) Relative abundance of Flavobacterium was significantly higher in samples with abnormal semen (p=0.0002, q=0.02). (B) Detection of Flavobacterium was significantly more prevalent in abnormal semen quality samples (p=0.0003). (C) Flavobacterium relative abundance was significantly higher in samples with <4% morphologically normal forms (p=0.0002, q=0.01). (D) Flavobacterium was also significantly more prevalent in samples with low percentage of morphologically normal sperm (p=0.0009).

Appendix 2—figure 1.. Seminal quality and function parameters according to recruited cohorts.

Comparison of microscopic semen parameters (A) concentration (p<0.0001, Kruskal-Wallis rank-sum test), (B) progressive motility (p<0.0001, Pearson’s chi-squared test), and (C) morphology (p<0.0001, Pearson’s chi-squared test) suggested poor semen quality for male factor infertility (MFI) patients. Comparison of clinical semen qualities: (D) reactive oxidative species (ROS), (E) sperm DNA fragmentation index.

Appendix 2—figure 2.. Ecological parameters of seminal microbiota for the recruited study cohorts.

(A) Species richness (p=0.30) and (B) Simpson’s diversity index (p=0.49) were not significantly different based on recruited study cohorts. Kruskal-Wallis tests with Dunn’s multiple comparison p-values demonstrated on the plots. (C) Bacterial load of seminal microbiota in recruited study cohorts. There were no significant differences in bacterial load based on recruited study cohorts using the number of 16S rRNA genes per 1 ml of semen (p=0.22, Kruskal-Wallis test).