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

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.

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].

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-chloroform-isoamyl 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.

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 pmol of each primer, 25 μl of 2× PCR Master Mix (0.05 unit/μl Taq DNA Polymerase, 4 mM MgCl₂, 0.4 mM dATP, 0.4 mM dCTP, 0.4 mM dGTP, 0.4 mM dATP and 0.4 mM dTTP) (Fermentas, Vilnius, Lithuania) and RNase-free H₂O in a final volume of 50 μl. The cycling parameters were 95°C for 5 min; 25 cycles of 95°C for 2 min, 42°C for 30 s, 72°C for 4 min; and a final cycle of 72°C for 20 min. The temperature was held at 4°C following the final cycle [34].

The second PCR mixture contained 5 μl of PCR product from the above reaction, 10 pmol of each primer, 25 μl of 2× PCR Master Mix (0.05 unit/μl Taq DNA Polymerase, 4 mM MgCl₂, 0.4 mM dATP, 0.4 mM dCTP, 0.4 mM dGTP, 0.4 mM dATP and 0.4 mM dTTP) and RNase free H₂O in a final volume of 50 μl. The cycling parameters were 95°C for 5 min; 30 cycles of 95°C for 30 s, 56°C for 30 s, 72°C for 1 min; and a final cycle of 72°C for 8 min. The temperature was held at 4°C following the final cycle [35].

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).

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.

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 [38‐40] 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.

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).

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 HPV-negative 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 HPV-positive 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 HPV-positive women.

We compared the number of bands, Shannon–Weiner diversity index and the proportion of samples from HPV-positive 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).

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%.

Lactobacillus was the most predominant genus, and was detected in all women included in this study. The Lactobacillus members are grouped into three species: L. gallinarum was the most abundant (45/70, 64%), followed by L. iners (41/70, 59%) and L. gasseri (23/70, 33%) (Table 4). The next most predominant genus was Gardnerella vaginalis (14/70, 20%), followed by Atopobium vaginae (7/70, 10%). A Chi-squared analysis revealed that there is no difference in the frequency of identification of L. gallinarum between HPV-negative and HPV-positive women (p = 0.775). Likewise, no significant difference existed in the frequency of identification of L. iners (p = 0.717) or Atopobium vaginae (p = 0.150). However, L. gasseri and Gardnerella vaginalis were isolated more frequently in HPV-positive women than in HPV-negative women (p = 0.005 and p = 0.031, respectively).

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.