Article Text
PDF
Abstract
Objectives To investigate if the participation of Atopobium vaginae, Megasphaera sp. and Leptotrichia sp. in the bacterial community of bacterial vaginosis (BV) is associated with distinct patterns of this condition.
Methods In this cross-sectional controlled study, 205 women with BV and 205 women with normal flora were included. Vaginal rinsing samples were obtained for measuring the levels of pro-inflammatory cytokines and bacterial sialidases. Real-time PCR was used to quantify the BV-associated bacteria and to estimate the total bacterial load using the 16S rRNA. Principal component analysis (PCA) using the measured parameters was performed to compare the BV samples with lower and higher loads of the species of interest.
Results Higher bacterial load (p<0.001), levels of interleukin 1-β (p<0.001) and sialidase activity (p<0.001) were associated with BV. Women with BV and higher relative loads of A vaginae, Megasphaera sp. and Leptotrichia sp. presented increased sialidase activity, but unchanged cytokine levels. PCA analysis did not indicate a different pattern of BV according to the loads of A vaginae, Megasphaera sp. and Leptotrichia sp.
Conclusions Greater participation of A vaginae, Megasphaera sp. and Leptotrichia sp. in vaginal bacterial community did not indicate a less severe form of BV; moreover, it was associated with increased sialidase activity.
- Bacterial Vaginosis
- Vaginal Microbiology
- Inflammation
- Microbiology
- Molecular Biology
Statistics from Altmetric.com
Introduction
Bacterial vaginosis (BV) is the most common type of abnormal vaginal flora in women of childbearing age.1 This condition is characterised by decreased Lactobacillus spp. and the overgrowth of anaerobic bacteria.1 ,2 These anaerobes produce several metabolites, such as sialidase, which are capable of degrading the local immunoglobulin A.3 Sialidase-producer strains of Gardnerella vaginalis, a well known BV-associated bacterium, present stronger adherence to epithelial cells and biofilm formation.4 Additionally, higher levels of sialidase in early pregnancy were already associated with an increased risk for preterm birth.5
Considering the heterogeneous feature of the bacterial composition of BV, it is expected that the inflammatory response may also vary among the cases. The innate immune response to BV is characterised by increased interleukin (IL)-1β and unchanged IL-6 and tumour necrosis factor (TNF)-α levels, while divergences remain regarding the IL-8 levels.6–9 In fact, cytokine production in response to BV is one of the proposed mechanisms for increased risk of acquisition of sexually transmitted infections.10 Thus, it is reasonable to propose that bacterial composition of BV can indicate BV cases that can be more deleterious to women's health.
Molecular-based investigations have provided new insights regarding the bacterial composition of BV by associating unculturable species to this condition, such as Atopobium vaginae, BV-associated bacteria (BVAB1-3), Megasphaera sp. and Leptotrichia sp.11 ,12 In a culture-independent study, Zhou et al13 demonstrated that asymptomatic women may have vaginal community dominated by A vaginae, Megasphaera sp. and Leptotrichia sp., and supported that, although these are BV-associated bacteria, they are lactic acid-producers. Based on this, it was proposed that the replacement of vaginal Lactobacillus spp. by A vaginae, Megasphaera sp. and Leptotrichia sp. should not be considered as abnormal, since these species could also provide a healthy environment by lowering the local pH.14–16
The observation of asymptomatic women with vaginal flora dominated by lactic acid-bacteria, other than lactobacilli, raises the hypothesis that a greater participation of these species in the local community would not be deleterious for the vaginal environment. Therefore, the present study aimed to evaluate the vaginal levels of pro-inflammatory cytokine and sialidase activity according to the relative loads of A vaginae, Megasphaera sp. and Leptotrichia sp. and to verify whether a greater participation of these species in the local bacterial community defines a less severe form of BV.
Material and methods
From May 2009 to October 2011, premenopausal and non-pregnant women attending an outpatient clinic of the University of Campinas and one unity of primary medical care in Botucatu-SP, were invited to participate in this controlled and cross-sectional study and were requested to sign a consent term. Women presenting vaginal bleeding, urinary loss, sexual intercourse (<72 h) or use of antibiotics (<30 days) were excluded. Demographic, behavioural and clinical data were obtained by self-reports.
During speculum examination, vaginal pH was assessed by pressing strips (4.0–7.0, Merck, Darmstadt, Germany) against the vaginal wall. Whiff test was performed by adding 10% KOH solution to the samples with results interpreted by the practitioner as positive, doubtful or negative. For microscopic evaluation of the vaginal flora, samples were taken from the mid-lateral vaginal wall. Wet mount microscopic analysis was performed to detect Trichomonas vaginalis and Candida sp. Vaginal flora was classified according to Nugent et al17 on Gram-stained smears. Endocervical samples were taken for Chlamydia trachomatis (CT) and Neisseria gonorrhoeae (NG) detection by PCR.18 ,19 Vaginal rinsings were performed by allowing the contact of 3 ml of sterile 0.9% NaCl on the lateral vaginal wall. Sterile pipettes were used to recover the samples, which were discarded in the presence of blood or if volumes were inferior to 3 ml. Rinsing samples were centrifuged at 800×g (10 min) and supernatants and pellets were stored separately at −80°C.
From 944 women seen at both services, a total of 313 (33.2%) with microscopy finding of BV (score ≥7) were included. Cases of BV that were PCR-positive for CT (n=74, 23.6%), NG (n=2, 0.6%) and for both (n=4, 1.3%) were excluded. From the remaining 233 BV women, those presenting Candida sp. morphotypes on the smears (n=28, 12.0%) were also excluded, independent of their symptoms. Consequently, the 205 BV samples included had a minimal chance of co-infection. A total of 205 women with normal flora (Score ≤ 3), not presenting any concurrent genital infection, were randomised to constitute the control group. Minimum sample size was estimated in at least 100 women for each group based on the estimated variability of IL-1β and IL-8, assuming α=0.05 and β=0.20.
Genomic DNA was extracted from the pellets of vaginal samples by DNEasy Blood & Tissue Kit (Qiagen, California, USA) with a final elution step in 100 μl of the buffer provided. Loads of G vaginalis, A vaginae, Megasphaera sp., Leptotrichia sp. and the conserved sequence of 16S rRNA gene were determined by quantitative PCR (qPCR). As template, 2 μl of extracted DNA was used in the qPCRs. Taxon-directed and 16S rRNA qPCRs were performed individually in final volumes of 13 μl using Maxima SYBR Green/ROX (Fermentas, St. Leon-Rot, Germany), with primers targeting sequences from the 16S rRNA gene (see online supplementary table 1) in 40 amplification cycles (LineGeneK, Bioer, China).20–22 Unknown numbers of copies were calculated from the standard curves constituted by 10-fold dilutions of plasmidial DNAs. To obtain the plasmids, taxon-specific sequences of A vaginae, Megasphaera sp. and Leptotrichia sp. were amplified from clinical samples and from G vaginalis (ATCC14018). The broad spectrum conserved region of 16S rRNA gene was amplified from Staphylococcus aureus (ATCC19095) to determine the total bacterial load. Products were sequenced (Applied Biosystems 3500 Genetic Analyzer, Inc., Foster City, CA, USA) and final sequences were submitted to BLAST to confirm they matched to the species of interest. Sequences were cloned into Escherichia coli DH5-α (Invitrogen, Carlsbad, California, USA) and the plasmidial DNA was used for determining the number of copies/μl using Avogadro's equation. All samples were tested in duplicate and retested if more than one cycle threshold was detected between them. Results from the qPCR, relative to 2 μl of template, were multiplied by 50 in order to reach the elution volume of 100 μl that corresponded to 1 ml of vaginal rinsing. Thus, final bacterial loads were expressed by the number of copies/ml of vaginal rinsing.
Relative loads of A vaginae, Megasphaera sp. and Leptotrichia sp. in BV samples were determined to evaluate their participation in the local bacterial community. This value was obtained by the sum of each individual qPCR of the three species, subsequently dividing it by the total bacterial load, estimated by the conserved 16S rRNA. The median of the relative load from all samples was determined and those with inferior or equal loads to the median were considered as the BV group with lower participation of A vaginae, Megasphaera sp. and Leptotrichia sp. in the local bacterial community, while relative loads above the median were considered as samples with higher participation of these species.
IL-1β, IL-6, IL-8 and TNF-α levels were measured by ELISA (Duo Set Kits, R&D Systems, Minnesota, USA) in the supernatants of vaginal rinsings. All samples were tested in duplicate and those with values set above the standard curve range were diluted (1 : 5 and 1 : 10) and retested. The minimum detectable levels for IL-1β, IL-6, IL-8 and TNF-α assays were, respectively, 0.2, 3.2, 20.0 and 1.1 pg/ml. Intra- and inter-assay variability remained <10.0% for all cytokines, except for TNF-α assay that presented an intra-assay variability of 23.4% and inter-assay of 25.5%.
Measurement of sialidase activity in the supernatants was performed using the fluorogenic substrate 2-(4-methylumbelliferyl)-α-D-N-acetylneuraminic acid (MUAN; Sigma-Aldrich, Missouri, USA). In a black 96-well plate, samples were added to 50 μl of 0.35% MUAN (wt/vol) and incubated for 30 min at 37°C. The reactions were read at 450 nm, 365 nm excitation and filter at 420 nm. Samples were tested in duplicate and compared with the standard curve (range 1000.0–1.0 ng/ml) of commercial Clostridium perfringens sialidase (Sigma-Aldrich). In every assay, a negative control consisting of a sample previously heated at 95°C (15 min) and a positive control of a known sialidase-positive sample with visible fluorescence on ultraviolet were run simultaneously.
Comparison of discrete and continuous variables of the two study groups were performed respectively by χ2 and Mann–Whitney tests. Bacterial load, cytokine and sialidase levels between the groups were compared using Mann–Whitney test. Correlation between the species was evaluated by the Spearman correlation test. Analyses were performed using GraphPad Prism 5.0 software (GraphPad, San Diego, California, USA) and p<0.05 was considered as significant. Principal component analysis (PCA) was performed to compare the samples of normal flora with those of BV, as well as to compare BV cases with lower and higher relative loads of A vaginae, Megasphaera sp. and Leptotrichia sp. based on the combination of their cytokine levels, sialidase activity and bacterial loads. Auto-scaled PCA analysis with log10 transformed data was performed using MVSP software (V.3.13a, Kovach Computing Services, Anglesey, Wales, UK).
Results
Demographic, behavioural and gynaecological characteristics of the women enrolled are shown in table 1. Population was homogeneous in the BV and normal flora groups, not differing for any of the demographic or behavioural variables evaluated, except for smoking habit that was more frequent in BV (p=0.01). Regarding the gynaecological data, women with BV were more likely to present a previous episode of BV (p=<0.0001), as well as multiparity (p=0.003), higher vaginal pH (p<0.0001), and positive or doubtful whiff test (p<0.0001).
Demographic, behavioural and gynaecological characteristics from 205 women with normal flora and 205 with BV included in the study
The results from taxon-directed qPCR showed that all BV-associated bacteria were detected in BV and normal flora (table 2). The species with the strongest association with BV was A vaginae, detected in 97.6% of the cases. Regarding the loads of each individual species, they were significantly higher in BV when compared with normal flora (p<0.0001). Results of the conserved sequence of 16S rRNA revealed that the total bacterial load in the vagina of women with normal flora can be highly variable, but still significantly lower than in BV.
Bacterial vaginosis (BV)-associated bacteria and total bacterial load in women with normal flora and BV
Cytokine and sialidase levels were also compared between BV and normal flora. As demonstrated in figure 1, the levels of IL-1β were significantly higher in women with BV than in normal flora (p<0.0001). No significant difference on IL-6 and IL-8 was observed between BV and normal flora (p>0.05). Only few samples presented detectable TNF-α and its levels did not differ between the groups (p>0.05). Sialidase activity was detected in 128 (62.4%) samples of BV (median 7.3 ng/ml; range 0.0–987.9 ng/ml) and only 4 (0.02%) with normal vaginal flora (p<0.001).
Vaginal levels of pro-inflammatory cytokines and sialidase in women with bacterial vaginosis (BV) (n=205) and normal flora (n=205). IL, interleukin.
As shown in table 3, important correlations exist between the loads of BV-associated bacteria and levels of cytokine and sialidase in the vaginal flora of women experiencing BV. All correlation tests between the microorganisms yielded significant positive coefficients, except for G vaginalis and Leptotrichia sp. The strongest correlation was found between A vaginae and Megasphaera sp. Spearman correlation tests were also performed between bacterial loads and cytokines revealing that IL-1β was significantly correlated with all bacteria, except Megasphaera sp. and Leptotrichia sp. IL-8 was only inversely correlated with Megasphaera sp. while no other significant correlation was observed between bacterial loads with IL-6 and TNF-α. Regarding sialidase activity, it was positively correlated with all species, as well as with the total bacterial load and levels of IL-1β. Strong correlations were also observed between each of the cytokines IL-1β, IL-6 and IL-8, while TNF-α was only significantly correlated with IL-6.
Spearman correlation rank between the parameters measured in women with bacterial vaginosis
As shown in table 4, no difference on cytokine levels was observed in BV samples according to the participation of A vaginae, Megasphaera sp. and Leptotrichia sp. in the local bacterial community. Regarding sialidase activity, cases of BV with greater participation of these species presented increased levels of these enzymes (p<0.0001).
Cytokine and sialidase levels according to the relative loads of Atopobium vaginae, Megasphaera sp. and Leptotrichia sp. (AML) in BV
PCA was first performed to evaluate the distribution of normal flora and with BV samples according to the parameters: cytokine, sialidase and bacterial load. As shown in figure 2A, a clear division is observed between the groups. A second PCA was performed to compare the cases of BV with lower and higher relative loads of A vaginae, Megasphaera sp. and e Leptotrichia sp. In this analysis, the loads of A vaginae, Megasphaera sp., e Leptotrichia sp. and total bacteria were excluded, as these parameters were used to constitute the two groups. As shown in figure 2B, samples were randomly distributed on the graph area, independent of the relative load of the three species.
Scatter plots of the principal component analysis showing variation among the measured parameters from samples of women presenting (A) normal vaginal flora and bacterial vaginosis (BV) (B) with low and high relative loads of the lactic acid-producers. Relative loads were determined by the sum of the loads of Atopobium vaginae, Megasphaera sp. and Leptotrichia sp. divided by the total bacterial load of each sample assuming the median value as the cut-off to constitute the groups. The bacterial loads were obtained by amplification of the 16S rRNA by quantitative PCR, using as primers sequences described in previous studies.20–22
Discussion
In this study, the levels of cytokines, sialidase activity, total bacterial load and loads of BV-associated bacteria, A vaginae, Megasphaera sp., Leptotrichia sp. and G Vaginalis, were determined in the vaginal samples from women with BV and normal flora. The inclusion criteria were very strict to assure that none of the samples had concurrent infection. In fact, 80 (25.6%) BV samples were excluded, as they tested positive for CT and/or NG. The high rate of CT and NG in BV is in agreement with the literature,23 which hinders the execution of studies aiming the analysis of cases of BV solely.
The studied population was homogeneous, not differing between the groups in most of the demographic variables. Although important studies showed that BV is associated with ethnicity, number of sex partners and education level, the current data failed to confirm their findings.24 ,25 However, this study was not designed to determine BV-associated factors and almost one of three of the women were excluded due to the presence of sexually transmitted infections, like CT and NG, which may explain why sexual history was no longer associated with BV. Women with BV were more likely to smoke regularly and had been pregnant at least once, agreeing with previous findings.25 ,26 As expected, BV was associated with higher vaginal pH and positive Whiff test, both well-recognised markers of BV.27
Regarding bacterial load, BV-associated bacteria were detected more frequently and they are found in significantly higher loads in BV. In fact, several reports already suggested the potential use of quantitative molecular methods as reliable tools for diagnosing BV.21 ,28 ,29 Although increases in the bacterial load in BV was already described,30 ,31 this study adds the information in terms of number of copies/ml of vaginal sample by the amplification of 16S rRNA. Although this method presents some limitations, as there are differences in the efficiency of the reaction among species, it is more reliable in determining the total bacterial load than the simple quantification of the extracted DNA, which carries variable amounts of host's DNA from epithelial and inflammatory cells.
The current analysis also adds to the previous information that the BV-associated bacteria are found simultaneously in vagina,13 ,20 ,32 and the fact that their growth seems to be positively correlated, supporting a synergistic relationship among them. In fact, the positive correlation between G vaginalis and A vaginae was previously reported by De Backer et al.33 Besides confirming their findings, the current data show important correlations among other BV-associated bacteria, which may encourage further researches focusing on the synergism among vaginal bacteria and their role in the local microbiota.
The cytokine levels and sialidase activity were assessed in the current study to detect possible disparities among the BV samples. As bacterial composition of BV is variable, differences in the innate immune response and bacterial products are also expected. The present findings on cytokine levels are in agreement with the literature that shows increased IL-1β and unchanged IL-6, IL-8 and TNF-α levels in BV when compared with normal flora.6–8 When analysing the current data on cytokines in BV samples, a positive correlation can be observed on the levels of IL-1β, IL-6 and IL-8. These findings are partially supported by Spear et al34 who showed a positive correlation on vaginal levels of IL-1β, IL-6 and TNF-α, but failed to demonstrate a correlation between IL-6 and IL-8, although their study did not evaluate women with BV exclusively.
Regarding sialidase activity, these data confirm it is almost only detected in BV.3 Moreover, sialidase levels in BV are particularly variable reflecting the high diversity on bacterial strains in this community.4 ,35 Correlation analysis also showed that sialidase is strongly correlated with all BV-associated bacteria and with the total bacterial load. Additionally, sialidase was positively correlated with vaginal IL-1β, but not with other pro-inflammatory cytokines, which is in agreement with previous reports.36 ,37 Although these former studies suggested that sialidase could be involved in blocking the downstream of the cytokine cascade after IL-1β release, this hypothesis has not yet been confirmed. In fact, although the factors involved in the impairment of the local innate immunity in BV are still unknown, the positive correlation between bacterial loads with IL-1β, but not with IL-6 and IL-8, indicates their presence in the vaginal milieu.
Slight changes in the microbial composition of the vaginal flora were already demonstrated to influence the local cytokine production.38 Nevertheless, no trend to increases on the levels of cytokine was found in relation to the participation of A vaginae, Megasphaera sp. and Leptotrichia sp. in the bacterial community of BV. When focusing on the local bacterial sialidases, the current data show that this enzyme is elevated in BV samples with higher relative loads of these species. These findings do not provide sufficient evidence of sialidase production by these species, but they clearly indicate an important synergistic relation between these microorganisms and sialidase-producing strains.
Considering the great diversity found on bacterial load, sialidase and cytokine levels in vaginal samples from women with BV, such parameters can be considered as useful to group samples based on their similarity. First, BV samples were compared with normal flora aiming to: (1) validate this method to group vaginal samples based on these parameters, which was confirmed by the separation of BV and normal flora samples and (2) check if this method of representation reflects the heterogeneous feature of BV, which could be observed by the diffuse distribution of BV samples on the graph, differently from the normal flora that tended to be more concentrated. However, in contrast with the clear division between BV and normal flora, no tendency of separation was observed between samples with lower and higher relative loads of A vaginae, Megasphaera sp. and Leptotrichia sp. in the second PCA. Therefore, BV samples with different participation of these species in the local bacterial community do not constitute different patterns of BV.
In conclusion, the current findings do not corroborate the assumption that the greater participation of A vaginae, Megasphaera sp. and Leptotrichia sp. on the bacterial community of BV is associated with a healthy vaginal environment.
Key messages
-
Bacterial vaginosis (BV) presents heterogeneous microbial features regarding the microbial composition, sialidase activity and cytokine release.
-
Higher relative loads of Atopobium vaginae, Megasphaera sp. and Leptotrichia sp. do not interfere in the local innate immunity, but are associated with increased sialidase activity.
-
Participation of A vaginae, Megasphaera sp. and Leptotrichia sp. in bacterial community of BV is variable, but not related to reduced severity of this condition.
References
Supplementary materials
Supplementary Data
This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.
Files in this Data Supplement:
- Data supplement 1 - Online appendix
Footnotes
-
Contributors CM: designed, acquired the data, analysed and interpreted the results, wrote the paper and approved the final version for publication. GGD and PCG: analysed and interpreted the data, revised the draft and approved the final version of the manuscript. CMGLP: acquired and interpreted the data, revised the draft and approved the final version of the manuscript. MGS: designed, analysed and interpreted the data; wrote the paper and approved the final version for publication.
-
Funding This work was supported by Fundação de Amparo a Pesquisa do Estado de São Paulo-FAPESP, Grant numbers 2008/55420-6 and 2009/50560-7, and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior-CAPES, Grant number 1179-09-8.
-
Competing interests None.
-
Patient consent Obtained.
-
Ethics approval Ethics approval provided by the Boards of the Botucatu Medical School and the University of Campinas.
-
Provenance and peer review Not commissioned; externally peer reviewed.