Comparison of the vaginal microbiota diversity of women with and without human papillomavirus infection: a cross-sectional study

Background The female genital tract is an important bacterial habitat of the human body, and vaginal microbiota plays a crucial role in vaginal health. The alteration of vaginal microbiota affects millions of women annually, and is associated with numerous adverse health outcomes, including human papillomavirus (HPV) infection. However, previous studies have primarily focused on the association between bacterial vaginosis and HPV infection. Little is known about the composition of vaginal microbial communities involved in HPV acquisition. The present study was performed to investigate whether HPV infection was associated with the diversity and composition of vaginal microbiota. Methods A total of 70 healthy women (32 HPV-negative and 38 HPV-positive) with normal cervical cytology were enrolled in this study. Culture-independent polymerase chain reaction-denaturing gradient gel electrophoresis was used to measure the diversity and composition of vaginal microbiota of all subjects. Results We found significantly greater biological diversity in the vaginal microbiota of HPV-positive women (p < 0.001). Lactobacillus, including L. gallinarum, L. iners and L. gasseri, was the predominant genus and was detected in all women. No significant difference between HPV-positive and HPV-negative women was found for the frequency of detection of L. gallinarum (p = 0.775) or L. iners (p = 0.717), but L. gasseri was found at a significantly higher frequency in HPV-positive women (p = 0.005). Gardnerella vaginalis was also found at a significantly higher frequency in HPV-positive women (p = 0.031). Dendrograms revealed that vaginal microbiota from the two groups had different profiles. Conclusions Our study is the first systematic evaluation of an association between vaginal microbiota and HPV infection, and we have demonstrated that compared with HPV-negative women, the bacterial diversity of HPV-positive women is more complex and the composition of vaginal microbiota is different.


Background
Microbes inhabit virtually all sites of the human body and play an important role in human health, yet we know relatively little about them. The Human Microbiome Project (HMP) [1], funded by the National Institutes of Health Roadmap for Biomedical Research, was implemented in 2007. To date, the HMP has released over 100 million 16S rRNA gene reads and more than 8 trillion bytes of shotgun metagenomic sequences [2]. Studies of the gastrointestinal microbiota showed that bacteria maintain homeostasis with the host in a healthy gut [3]. However, when bacterial dysbiosis (microbial imbalance) occurs in the gut, the host may experience inflammation, a loss of barrier function, and possibly serious disease such as ulcerative colitis, Crohn's disease and colorectal cancer [4][5][6].
The female genital tract is an important habitat for human microbiota. Investigation of "normal" vaginal microbiota typically reveals Lactobacillus species as the predominance genus in the vagina, which helps to promote a healthy vaginal milieu [7]. Bacterial vaginosis (BV), characterized by a loss of indigenous Lactobacillus species and a concurrent overgrowth of anaerobic bacteria, has been associated with vaginal discharge syndrome, poor pregnancy outcomes, pelvic inflammatory disease, post-operative wound infections, and endometritis after elective abortions [8][9][10][11]. Additionally, BV predisposes women to infection by human papillomavirus (HPV) [12]. Guo et al. [13] reported that compared with women without BV, those with BV had a lower clearance of HPV. And Dols et al. [14] resisted that in women with HPV infection, the prevalence of L. crispatus was significantly reduced and there was a shift in the composition of the Lactobacillus microbiota in HPV infection.
Persistent HPV infection is the central factor in the development of cervical cancer, and is a prerequisite for progression to high-grade cervical lesions [15]. However, few HPV infections progress to cervical cancer, and most HPV infections are eventually cleared [16]. The reason(s) why high risk HPV infection is cancerous in some women but not others is unknown. Some studies imply that BV is associated with HPV acquisition; a meta-analysis of twelve studies with a total of 6,372 women indicated a positive association between BV and HPV infection, with an overall estimated odds ratio of 1.43 (95% confidence interval, 1.11-1.84) [12]. However, these studies focused on the relationship between BV (diagnosed using clinical Amsel criteria or Nugent's score) and HPV infection [17,18], and did not examine the composition of vaginal microbial communities involved in HPV acquisition.
It is difficult to assess the microbial community in an environment where more than 80% of microbiota is nonculturable (such as the vagina) [19,20]. However, the advances of molecular biotechnology, such as cultureindependent polymerase chain reaction-denaturing gradient gel electrophoresis (PCR-DGGE), enable better characterization of complex microbial communities [21,22]. The combination of PCR and DGGE allows for the rapid and reliable examination of the vaginal microbiota, and these techniques have been used widely in bacterial microbiota studies [23][24][25]. The method allows numerous samples to be screened, because the microbial nucleic can be derived directly from human specimens, without the need for culture enrichment. In addition, PCR-generated DNA fragments of the same length but different base-pair sequences can be separated by DGGE. Although PCR-DGGE has some shortcomings, including the fact that it needs costly equipment, the "GC clamp" is very expensive, and the formamide is poisonous, it is a powerful tool for microbiota studies. But the next generation sequencing [26] might will replace the current molecular biotechnology in the near future.
To examine the relationship between the diversity and composition of vaginal microbiota and HPV positivity, we used PCR-DGGE to examine the vaginal microbiota of 38 HPV-negative and 32 HPV-positive healthy women with no BV (using Amsel criteria). To our knowledge, this is the first study of the association between HPV infection and vaginal microbiota to use a molecular biological technique to examine the microbiota.

Subject selection
A total of 100 healthy women (50 HPV-negative and 50 HPV-positive) with normal cervical cytology, who accepted a routine gynecological examination and Thinprep Cytology Test (TCT) in Beijing Cancer Hospital from January 2012 to June 2012, were initially recruited for this study. Subsequently, TCTs were reexamined by two cytologists and HPV infection was reexamined by PCR. If there was a difference between the first and second TCT reports, or between the results from the two cytologists performing the second TCT, the woman was excluded from the study. In addition, if a recruited woman was infected with more than one HPV type, or there was a difference between the initial HPV report using Hybrid Capture II and the second HPV report using PCR, the woman was excluded from the study. A total of 30 women were excluded from the study based on these four criteria. Inclusion criteria were age <50, no BV by the Amsel method, no use of antibiotics or vaginal antimicrobials (orally or by topical application in vulvar/vaginal area) in the previous month, and no vaginal intercourse or vaginal lavage within the last 3 days. All subjects were free of systemic diseases such as diabetes, autoimmune disease, and malignant tumors. Informed written consent was obtained from all participants prior to enrollment. This study was approved by the ethical committee of Peking University Cancer Hospital and Institute, Beijing, China.

Sample collection and preparation
When women underwent genital examination, a sterile swab sample was taken from near the vaginal fornix and cervix from each participant. The swab was placed into 1 ml sterile saline, placed on ice packs immediately, and transferred to the laboratory within 30 min. The sample was pelleted by centrifugation at ≥10,000 × g (25°C) for 10 min and stored at −80°C until further analysis, as previously published [27].

Total bacterial genomic DNA extraction
Bacterial DNA was extracted according to the procedures described by Signoretto [28] and Zijnge [29]. Briefly, the sample was incubated for 1 h at 58°C with 1 ml of lysis buffer (10% SDS and 0.2 mg/ml proteinase K in 25 mM Tris-HCl, pH 8). The sample was then incubated at 80°C for 10 min to inactivate the proteinase K. DNA was purified from the lysate by repeated phenol-chloroformisoamyl alcohol extraction, precipitated with sodium acetate and ethanol, dissolved in 100 μl sterile Milli-Q water and stored at −20°C in aliquots. The concentration of extracted DNA was determined by a Nano Photometer™ Pearl ultramicro ultraviolet spectrophotometer (Implen, Munich, Germany). The quality of the DNA was checked by agarose gel electrophoresis. All DNA was stored at − 20°C before further analysis.

Nested PCR
The hypervariable region chosen for amplification can influence the PCR-DGGE profiles. The V2-V3 region of the 16S rRNA gene is reported to be the most reliable [30,31]. The consensus primers used in this study were as follows: S-D-Bact −0008-a-S-20/ S-*-Univ-1492-b-A-21: AGA GTT TGA TCC TGG CTC AG/ ACG GCT ACC TTG TTA CGA CTT; and HDA1-GC/ HDA2: CGC CCG GGG CGC GCC CCG GGC GGG GCG GGG GCA CGG GGG GAC TCC TAC GGG AGG CAG CAG/ GTA TTA CCG CGG CTG CTG GCA. The primers used in this study have been published previously [32,33].
The first PCR mixture contained 100 ng of DNA template, 5  The temperature was held at 4°C following the final cycle [35].

DGGE
The denaturing gradient gel was formed with 8% polyacrylamide stock solution containing either low (40%) or high (70%) concentrations of urea and formamide that increased in the direction of electrophoresis.
PCR products were mixed with a loading buffer containing bromophenol blue, xylene cyanol, and 70% glycerol in TAE buffer (0.02 M Tris base, 0.01 M acetic acid, 1 mM EDTA, pH 7.5) and loaded into the gels. About 20 μl of the reference mixture was combined with loading buffer, and the mixture was loaded so that it flanked the vaginal samples. Electrophoresis was performed for 16 h at 60 V and 58°C. The gels were stained with SYBR Green I (1:10,000) for 30 min.
DGGE images were digitally captured and recorded by the BIO-RAD Gel Doc™ XR + Gel Imaging System (Bio-Rad, Hercules, USA) and analyzed by Gel Compare® (Applied Maths, Kortrijk, Belgium). Each gel was normalized according to a DGGE standard marker (10 markers according to the GC base-pair percentage).

Cloning of excised DGGE bands
The main DGGE bands were excised, and the DNA fragments were amplified with primers HDA1 (without the CG clamp) and HDA2. Amplified material was cloned into the pGM®-T vector using the pGM®-T cloning Kit (Tiangen, Beijing, China), according to the manufacturer's instructions. The vector was then cloned into Escherichia coli Top10, and the extracted plasmid was sequenced by the Beijing Genomics Institute. Sequence identification of the plasmid was performed by searching the NCBI BLAST database.

Detection of HPV
HPV DNA was extracted from the remnant TCT sample by TIANamp Virus DNA/RNA Kit (Tiangen) according to the manufacturer's instructions. HPV DNA was first amplified using the HPV L1 consensus MY09/MY11 primer pair, followed by nested PCR with GP5+/GP6+ primers. The PCR amplification was performed as described previously [36,37].
PCR products were subjected to direct DNA sequencing by the Beijing Genomics Institute. The obtained sequences were compared with documented virus sequences available in the GenBank database using the BLAST program. The subjects whose results were different to the previous results from Hybrid Capture or who were infected with multiple HPV genotypes were eliminated from this study. HPV-negative women and women infected with one HPV (single high-risk) subtype were included in this study. The HPV types are not listed in this paper.

Statistical analysis
The similarities of PCR-DGGE DNA profiles were analyzed with Gel Compare® software (Applied Maths) using Dice's similarity coefficient. The clustering algorithm, unweighted pair-group method with arithmetic means (UPGMA), was used to calculate the dendrogram.
Vaginal microbiota diversity was expressed by the Shannon-Weiner diversity index [38][39][40] and calculated using the following formula: H' = −Σp i lnp i . For morphological analysis, p i is the proportion of individuals in the ith taxon. For PCR-DGGE analysis, p i is the importance probability of the bands in a gel lane and measured as p i = n i /N, where n i is the intensity of a band and N is the sum of all band intensities in the densitometry profile. The Mann-Whitney U test and Kruskal-Wallis tests were performed to compare the diversity indices, and p < 0.05 was interpreted to be statistically significant.

DGGE profiles in HPV-positive and HPV-negative women
A total of 38 HPV-negative women (mean age 37.0) and 32 HPV-positive women (mean age 37.8) were enrolled in this study. There was no significant difference in age between the two groups (p = 0.57).
The DGGE profiles of vaginal samples were obtained from all 70 women. Figure 1 shows representative DGGE profiles of vaginal microbiota from the two groups. The left panel depicts DGGE profiles from five HPV-negative women (N1-N5) and the right panel shows profiles from five HPV-positive women (H1-H5). Lane M comprises 10 markers (a-j), and the bands of each lane in each DGGE profile are classified by each marker's GC base-pair percentage. The bilateral part of the figure shows the marker files from different DGGE gels. Each gel was normalized according to a DGGE standard marker, and each peak in the marker files stands for one marker. Pairwise comparisons between each gel marker using Gel Compare® software demonstrated that the marginal discrepancies of marker bands between each gel are negligible. The number of bands in the DGGE profile from each woman is shown in Table 1. Previous studies have reported that healthy vaginal microbiota only contains one or two predominant species [41,42]. We compared the number of bands in samples from HPV-positive women versus HPV-negative women. Half of the vaginal samples from HPV-negative women contained more than two bands in their DGGE profiles, whereas 87.5% of HPV-positive women had more than two bands. The mean band number from HPV-negative women was 3.45, and the mean number from HPV-positive women was 6.47 (Chi-squared test, p = 0.001).
Vaginal microbiota diversity was expressed using the Shannon-Weiner diversity index and compared using the Mann-Whitney U test. The diversity index for subjects with and without HPV infection is displayed in Figure 2. We found significantly greater biological diversity in HPV-positive women (mean = 1.64; range, 0 to 3.09) than in HPV-negative women (mean = 0.93; range, 0 to 2.62) (p < 0.001).

Cluster analysis based on DGGE profiles
The UPGMA clustering algorithm was used to construct a dendrogram of the DGGE profiles of vaginal microbiota from all 70 women enrolled in this study. Hierarchical cluster analysis of the dendrogram indicated that six clusters were formed, and revealed differences in the composition of vaginal microbiota between HPV-negative women and HPV-positive women (Figure 3). Most samples from HPV-positive women fell within clusters 1-4. Conversely, clusters 5 and 6 primarily comprised samples from HPVnegative women. There were significantly more bands in cluster 1-4 than in clusters 5 and 6, indicating that the vaginal microbiota of HPV-positive women had a greater biological diversity than that of HPV-negative women. Of note, in cluster 6 the DGGE profiles consisted of only one or two bands.
The discriminative character analysis of DGGE profiles showed that the bands in box A are present in both HPVpositive and HPV-negative women ( Figure 3). However, the bands in box B mostly appeared in cluster 2 (which primarily consists of HPV-positive women). The bands in box C were detected in cluster 2, cluster 3, and other   clusters, but the bands in box C are mostly from HPVpositive women. We compared the number of bands, Shannon-Weiner diversity index and the proportion of samples from HPVpositive and HPV-negative women in each of the six clusters (Table 2). There were large differences in all three characteristics (all p values <0.001). The Kruskal-Wallis test was used to compare the Shannon-Weiner diversity indices from each of the six clusters, and we found that the diversity of at least one of the clusters was significantly different from the other clusters (p < 0.001).

Identification of vaginal microorganisms
Excised DGGE gel bands were sequenced and BLAST database searches were used to identify 18 bacterial species present in vaginal samples from the 70 women studied. The name of each microorganism, sequence identities and NCBI Sequence ID are shown in Table 3. The sequence identities of most microorganism were above 97%.
L. gallinarum and L. iners are the two bands in box A of Figure 3. Gardnerella vaginalis is the predominant genus of the bands in box B, and other bands in box B include Escherichia fergusonii, Atopobium vaginae, Streptococcus australis, Streptococcus intermedius and Alloscardovia omnicolens. L. gasseri is the predominant genus of the bands in box C, and another band in box C is ureaplasma parvum serovar 3 str.

Discussion
Biologic susceptibility to HPV acquisition and immune competence for clearance of an HPV infection can be affected by vaginal bacterial infection, which disrupts the balance of vaginal microbiota [12,43]. Previous studies have not investigated the difference in microbiota communities between HPV-negative and HPV-positive women. To our knowledge, the current study is the first systematic evaluation of an association between vaginal microbiota and HPV infection. Our study analyzes the diversity and abundance of the vaginal microbiota by PCR-DGGE, and compares the microbiota by HPV infection status. This study may be helpful to reveal the role of the vaginal microbiota in the natural history of HPV infection.
If the target 16S rRNA gene copy number in the sample is low, it is likely that it cannot be amplified sufficiently to be visualized as a band. Therefore, the bands generated in DGGE reflect the most abundant genus from each vaginal sample [44]. In our study, the number of DGGE bands in HPV-positive women was significantly higher than that in HPV-negative women. Therefore, we conclude that the vaginal microbiota of HPV-infected women is more complex than that of HPV-negative women in our study.
The Shannon-Weiner diversity index uses the total number and relative intensities of DGGE bands, which makes it more suitable to estimate bacterial diversity than the number of bands alone [45]. A Mann-Whitney U test comparing the Shannon-Weiner diversity index between HPV-positive and HPV-negative women demonstrated that there was significantly greater biological diversity in HPV-positive women than HPV-negative women (p <0.001). Other studies have also demonstrated a difference in the abundance of bacterial species present in the vagina when comparing healthy women to those with BV or cervicitis [25,46]. This suggests that an increase in bacterial diversity may be associated with the shift from health to disease. Of note, the opposite is true for oral health, where a decrease in bacterial diversity is associated with disease [47]. The UPGMA clustering algorithm has been used previously [48], and in our study it was used to identify samples that generate similar DGGE profiles. The UPGMA dendrogram showed that six distinct clusters are formed, and that samples from HPV-positive and HPV-negative women tend to be in distinct clusters. The discriminative characters analysis of DGGE profiles indicate that the bands of box B and box C, which are different in clusters 1-4 versus cluster 5-6, might generate the distinction of clusters.
Over 120 species of Lactobacillus have been identified, and more than 20 species have been detected in the vagina. However, vaginal microbiota is not reported to contain very many different species of Lactobacillus [49]. Typically, one or two Lactobacilli species are predominant. For example, L. crispatus and L. jensenii were the most common genera for white women [50], while L. crispatus and L. gasseri were more common in Japanese women [51]. Recently, a study from China reported that L. crispatus, L. iners and L. gasseri were the most common genera in Chinese women [52].
In our study, sequence analysis identified three Lactobacillus members: L. gallinarum, L. iners and L. gasseri, all of which are obligately homofermentative species. L. gallinarum and L. iners are the two bands in box A (Figure 3), and there was no difference between HPVnegative women and HPV-positive women, there was no difference among each cluster. However, L. gasseri is the main species in box C, and was detected more frequently in the HPV-positive women (p = 0.005). Cherpes et al. reported that L. gasseri was vaginal colonization from the rectum [53], and a study of homosexual women found an association between L. gasseri and BV [54]. And Dols et al. [14] resisted that the prevalence of L. crispatus was significantly reduced and there was a shift in the composition of the Lactobacillus microbiota in women with HPV infection.
But we were not able to distinguish L. gallinarum from the other Lactobacillus spp., because the BLAST result indicated that the similarity score of L. gallinarum was the same as that of other Lactobacillus spp., such as L. crispatus, L. acidophilus, L. amylovorus, L. rhamnosus, L. fermentum, L. helveticus, L. kitasatonis and L. ultunensis. One possible explanation for this is that the genera of Lactobacillus spp. mentioned above have the same DNA sequence as L. gallinarum in the V2-V3 region of 16S rDNA. Future research will seek to resolve this issue.
The bands in box B are the primary bands that distinguish cluster 2 from other clusters. The main microorganisms in box B are Gardnerella vaginalis and Atopobium vaginae. Gardnerella vaginalis is a facultative anaerobic bacterium of the Bifidobacteriaceae family [55], while the genus Atopobium lies within the family Coriobacteriaceae and forms a distinct branch within the phylum Actinomycetes [56]. Because Gardnerella vaginalis and Atopobium vaginae were frequently detected in association with BV, Menard deemed the combination of the two bacterial genera as predictive criteria for the diagnosis of BV [57]. In   our study, Gardnerella vaginalis (10/32 vs. 4/38, p = 0.031) and Atopobium vaginae (2/38 vs. 5/32, p = 0.150) were more frequently detected in HPV-infected women than in HPV-negative women, though the difference of Atopobium vaginae in the two groups was not statistically significant.
Because this was a cross-sectional study, we were unable to determine whether a change in vaginal microbiota preceded HPV infection, or whether HPV infection preceded a change in vaginal microbiota. Furthermore, we initially recruited 100 women with normal cervical cytology and excluded 30 from the study, which led to a relatively small sample size. However, because this is the first study to use PCR-DGGE to examine the association between vaginal microbiota and HPV infection, even with the limitations mentioned above we believe it has substantial importance.

Conclusion
In conclusion, we found that the bacterial diversity and composition in HPV-positive women were more complex than in HPV-negative women. Gardnerella vaginalis and L. gasseri were detected significantly more frequently in HPV-positive women. Abnormal vaginal microbiota might act as a co-factor for the acquisition of HPV. Continued exploration of the interplay between vaginal bacterial communities and HPV may shed light on biologic susceptibility to HPV infection, as well as immune competence for clearance of HPV, which may provide new insights into the early steps in the development of cervical cancer.