According to the inclusion criteria, PCOS patients admitted to the First Affiliated Hospital of Xiamen University in Fujian Province from June 2018 to June 2019, and healthy volunteers were recruited. Meanwhile, healthy volunteers were recruited for the non-PCOS group. All participants signed an informed consent form before the experiment. All participants underwent a general physical examination and an ultrasonic-B scan before enrollment ( Supplementary Table 1 ).

All patients with PCOS fulfilled the Rotterdam 2003 diagnostic criteria:

According to BMI and the PCOS diagnosis results, the 41 volunteers were divided into four groups, 11 individuals in the PCOS with obesity (PO) group, eight individuals in the non-PCOS with obesity (NO) group, 8 individuals in the PCOS and non-obese (PS) group, and 14 individuals in the non-PCOS and non-obese (NS) control group. Microbiota samples were collected from the posterior fornix of the vagina were collected using a clean and sterile cotton swab during the non-menstrual period (Greenbaum et al., 2019). Swabs were placed into collection tubes containing fixative (YXA, Xiamen, China) and then delivered to the laboratory in dry ice boxes and stored in a −80 °C refrigerator until nucleic acid extraction. Total DNA from the swaps was then extracted using the DNeasy PowerSoil Kit (Qiagen, Hilden, Germany), as recommended by the Human Microbiome Project. The DNA libraries for NGS sequencing were prepared using the Tn5 tagmentation-based protocol (Hennig et al., 2018). DNA segmentation and addition adaptors simultaneously with the Tn5 enzyme were performed according to Hennig’s protocol (Hennig et al., 2018). The purified DNA libraries were then sequenced on the Illumina HiSeq X sequencing platform using a pair-end (PE150) strategy.

The raw shotgun sequencing data were used for taxonomic analysis using MetaPhlAn3 (Metagenomic Phylogenetic Analysis, Version 3.1.0) (Beghini et al., 2021) and Kraken2 (Wood et al., 2019). The retrieved microbiota analyzed included bacteria, viruses, and fungi. The function of the microbiota was analyzed using Humann3 (Beghini et al., 2021), followed by megahit assembly (Dinghua et al., 2015).

The one-way ANOVA test was used to test the significance of the difference in abundance of specific species between groups. The correlation between the bacteria and phage/virus species was analyzed using the R package Hmisc and plotted using the R package pheatmap. The differentially enriched microbial species in the PCOS and obesity groups were analyzed using LEfSe tool (Segata et al., 2011). The significance of the difference in relative bacterial abundance between PCOS and non-PCOS was analyzed with a t-test. The PCA analysis of the vaginal microbiome in samples from different groups was performed using the R package DESeq2 and plotted using ggplot2.

A total of 41 PE150 sequencing data sets were obtained, and each sample received at least 20 million reads. The data have been uploaded to Sequence Read Archive (SRA) database (Accession No. SRR24561304 to SRR24561344).

The rarefaction curve of gene families retrieved from Humann and OTUs output from Kraken indicate that the sequencing depth is adequate for metagenomic analysis ( Supplementary Figure 3 ).

The functional analysis using HUMAnN output revealed that a total of 23,438 UniRef90 genes were identified, with 1,278 classified under Lactobacillus crispatus, 607 under L. iners, and 21,553 remaining unclassified.

The pathway abundance results encompassed 27 pathways, with the majority of the samples falling within group NS and group NO. Notably, the PWY-3781 pathway, associated with aerobic respiration I (cytochrome c), was exclusively observed in the obese group. A more specific omics study should be undertaken to shed light on the functionality of the PCOS and obesity microbiome.

The results of this study revealed distinct differences in the composition of vaginal microbiomes between the PCOS (P) and non-PCOS (N) groups, as well as between the obesity (O) and non-obese (S) groups ( Table 3 ). The non-PCOS group exhibited a relatively normal vaginal microbiome with low diversity. In contrast, the PCOS and obesity groups showed signs of dysbiosis characterized by increased microbial diversity.

Among the various bacterial genera identified in the vaginal microbiomes, several genera were significantly different in abundance between the groups. Notably, Lactobacillus, Pseudomonas, Bacillus, and other genera were enriched in the non-PCOS and non-obese (NS) group, while Staphylococcus, Clostridium, Leptospira, and others were relatively more abundant in the PCOS and obesity groups. It is worth noting that several pathogenic and opportunistic pathogens, including Streptococcus pyogenes, Leptospira santarosai, Citrobacter amalonaticus, Listeria ivanovii, and Clostridium perfringens, were detected in higher abundance in the PCOS and obesity groups. This finding raises concerns about potential health risks associated with the alteration of the vaginal microbiome in these conditions.

The higher abundance of benign commensals bacteria could inhibit the growth of pathogens. The fact that the pathogenic bacteria were lower in the NS group could be explained by the former finding (Kang et al., 2017); the lactic acid and other chemicals produced by Lactobacillus could inhibit the growth of Staphylococcus aureus and prevent the formation of biofilm (Gulzar, 2018). Generally, the Lactobacillus species is the dominant bacteria in the vaginal microbiome of healthy reproductive-age women (Spurbeck and Arvidson, 2011). Lactobacillus species are the common bacteria that sustain health. The abundance of Lactobacillus species is significantly higher in the NS group. This result is concordant with the previous study (Spurbeck and Arvidson, 2011; Greenbaum et al., 2019). Lactobacillus species sustain the acid and anaerobic conditions for the normal vaginal microenvironment and form the front line to defend against vaginally acquired infections (Spurbeck and Arvidson, 2011). But in the PO, PS, and NO groups, Lactobacillus species are not so dominant, and this suggests dysbiosis of the vaginal microbiome is linked with obesity or PCOS. The disturbance of normal vaginal bacteria has been approved to be linked with different types of diseases, such as bacterial vaginosis and virus infection (Spurbeck and Arvidson, 2011).

In this study, Pseudomonas species are also much higher in the NS group ( Figure 2 ), suggesting they interact with each other and keep a balance. Pseudomonas species are not generally considered pathogenic, but studies have identified that they are present at a low level in the indigenous microbiota of various body sites, including the uterus, and can lead to pelvic inflammatory disease (Scales et al., 2014). The presence of Pseudomonas and Pseudomonas phages in the NS group but not in the PCOS or obesity groups suggests that, while Pseudomonas is a common bacteria in vaginal microbiomes.

Bacillus cereus is a facultative anaerobe has mechanisms for both aerobic and anaerobic respiration. Some of the major products produced from carbon sources such as sucrose or glucose during anaerobic respiration include L-lactate, acetate, formate, succinate, ethanol, and carbon dioxide (Mols et al., 2007). These metabolites produced by B. cereus help to sustain the acid conditions in the vaginal microenvironment and contribute to the health.

On the other hand, most of the pathogenic bacteria lower in the NS group are positively correlated with each other, such as Listeria ivanovii, Clostridium perfringens, Citrobacter amalonaticus, Staphylococcus aureus, and Streptococcus pyogenes; and L. ivanovii negatively correlated with the benign commensals bacteria Bacillus cereus. Dysbiosis can lead to the overgrowth of pathogens, which might be linked to an increased risk of infections, inflammation, and other health issues. Understanding these associations could have implications for diagnostic and therapeutic strategies in managing PCOS and obesity. This result suggested that in the PCOS or obesity groups, some groups of pathogens linked with the diseases. Furthermore, the pathogens linked with PCOS and obesity are quite different, suggesting the vaginal microbiome’s pathogenic mechanisms are different between PCOS and obesity.

In summary, the vaginal bacteria in healthy are similar, but PCOS and obesity are associated with different bacterial shifts from the normal one which are benign commensals and produce lactic acids and help to sustain the acid conditions.

This study annotates 14 genera (24 species) of viruses and phages. This suggests that the distribution of viruses or phages in the vaginal microbiome is very common. Bacteriophages contribute to the maintenance of ecological balance and the evolution of bacterial species (Martín et al., 2009; Wright, 2021). Regarding the function of the vaginal bacteria phage, a previous study has illustrated that the Lactobacillus phage can eliminate the Lactobacillus culture in a petri dish (Kilic et al., 2001; Miller-Ensminger et al., 2020). But in this study, Lactobacillus phage Lv1 was only found in two specimens of the NS group, and in these two specimens, several Lactobacillus bacteria were detected. This implies that the Lactobacillus phage contributes to the ecological balance of vaginal microecology while not eliminating the Lactobacillus. Furthermore, the presence of the Lactobacillus phage is an indicator of the presence of relatively abundant Lactobacillus (Miller-Ensminger et al., 2020).

Lactobacillus phage Lv1 is a dsDNA virus isolated from a human vaginal Lactobacillus jensenii strain (Martín et al., 2010), which coexisted and positively correlated in this study. It has been illustrated that the stable coexistence of phages and bacteria is very common in the urogenital microbiota (Miller-Ensminger et al., 2020; Wright, 2021). The coexistence of lactobacilli and phages in the vaginal microbiome is widely spread, and many of the lysogenic lactobacilli could release phages at a high frequency (Kilic et al., 2001; Martín et al., 2009; Wright, 2021). In summary, lysogenic phages have a complex relationship with their host bacteria. They can both provide advantages and pose threats to the host, depending on the specific genes carried by the prophage and the environmental conditions. This interplay between lysogenic phages and bacteria is an important aspect of microbial ecology and evolution. lysogeny in vaginal lactobacilli is widely spread. Some lysogenic lactobacilli spontaneously release phages with a broad host range. Even though the phages were all temperate, they were able to cause lytic infection in various strains (Kiliç et al., 2001). The detailed relationship between Lactobacillus phage Lv1 and Lactobacilli in vaginal microbiome needs further study.

In this study, Pseudomonas phages/viruses, including Pseudomonas virus EL, F10, phiCTX, B3, PA16, and LKA1, have the common host Pseudomonas aeruginosa (Kwan et al., 2006; de Smet et al., 2017; Miller-Ensminger et al., 2020). Pseudomonas aeruginosa is an opportunistic pathogen that grows on the skin in moist parts, including the genital area, and it can cause severe infection when the immune system is weakened. A recent study suggested that the colonization of Pseudomonas aeruginosa in the genital tract can lead to infertility by inhibiting sperm mobility (Vander and Prabha, 2019). This could explain why some dysbiosis of the vaginal microbiome may lead to infertility. On the other hand, phage therapy has been proven to be an effective way to treat infection with Pseudomonas aeruginosa (Pires et al., 2015). The relationship between Pseudomonas phages/viruses and the presence of Pseudomonas in the vaginal microbiome, and their impacts on health needs further investigate.

The abundance of human endogenous retrovirus K (HERV-K) is significantly higher in the NS group when compared with the PS, PO, and NO groups. The most recent study suggests that HERV-K has multiple copies in the human genome and that some of its open reading frames are transcribed and translated in early embryogenesis (Garcia-Montojo et al., 2018). Though HML-2 of HERV-K has been reported to be associated with certain types of cancer and neurodegenerative diseases in adult tissues, the functional associations between HERV-K (HML-2) and diseases have not been completely clarified, and further work is still needed to confirm the specific HERV loci associated with specific diseases (Xue et al., 2020). In this study, we only sequenced DNA, not RNA, and the sequencing depth was relatively low, so it cannot reflect HERV-K transcriptional levels in the vaginal or specific HERV-K loci in the genome. The relationship between HERV-K and PCOS needs further study.

Among these viruses, Cyprinid herpesvirus 3 showed the highest frequency and widest distribution. It is detected in 33 out of the 38 samples, and the average frequency is 50% of the viruses/phages. Cyprinid herpesvirus 3, also known as CyHV-3, koi herpes virus or KHV, is a species of dsDNA virus that causes a viral disease that is very contagious to the common carp Cyprinus carpio. CyHV-3 has been proposed as a biological control agent for the invasive carp in Australia, and the safety to non-target species has been tested, which suggests that CyHV-3 is safe for the non-target species tested, including 14 species of fish, two species each of amphibians, and reptiles, and single species of bird, mammal, and invertebrates (McColl et al., 2016). CyHV-3 is a large double-stranded DNA herpesvirus classified in the Alloherpesviridae family in the Herpesvirales order. This order encompasses herpesviruses that infect different hosts, including mammals, avians, and reptiles. CyHV-3 is divergent from the other mammalian, avian, and reptilian herpesviruses, and bears homolog genes similar to CyHV-1, CyHV-2, AngHv-1, IcHV-1, and RaHV-1 (Ilouze et al., 2012). In this study, the annotation of Cyprinid herpesvirus 3 from the human vaginal microbiota shotgun metagenomes may be because there is a CyHV-3-like virus in the specimens. The high frequency of this CyHV-3-like virus may contribute to diseases such as PCOS. But more information about this CyHV-3-like virus in the vaginal microbiome and its function needs further study.

The analysis of correlations between bacterial genera and phages/viruses provided insights into the complex interactions within the vaginal microbiome. Notably, there were strong positive correlations among various Lactobacillus species and among Pseudomonas species, suggesting the presence of microbial communities with shared ecological niches. The positive correlation between Lactobacillus phages and Lactobacillus, as well as Pseudomonas phages and Pseudomonas suggest lysogenic phages are also involved in the microbiome regulation and may play an role. The third correlation is the negative one between benign commensal bacteria and pathogenic bacteria, which suggests that benign commensal bacteria play a central role in the maintenance of vaginal health. It also indicates that the disruption of the benign commensal ecology will cause a dysbiosis of the vaginal flora, which lead to invasion of pathogenic bacteria and cause PCOS in women of reproductive age by inducing inflammation, causing insulin resistance, and overproduction of androgens (Azziz et al., 2009).

The results of this study highlight the potential clinical significance of vaginal microbiome dysbiosis in PCOS and obesity. Dysbiosis can lead to the overgrowth of pathogens, which might be linked to an increased risk of infections, inflammation, and other health issues. Understanding these associations could have implications for diagnostic and therapeutic strategies in managing PCOS and obesity. Because the interaction between the microbiota and polycystic ovary syndrome is a bidirectional one, it is likely that restoration of the vaginal flora may contribute to the relief of PCOS symptoms by reducing inflammation and restoring normal microbiome metabolites and their signaling.

It’s important to acknowledge the limitations of this study, such as the relatively small sample size and the observational nature of the findings. Further research with larger and more diverse cohorts is needed to validate these results. Additionally, future studies may explore the mechanisms underlying the observed microbiome alterations and their potential impacts on gynecological and metabolic health.