Prevalence of Genotypes and Subtypes of Gardnerella vaginalis in South African Pregnant Women

Background Gardnerella vaginalis, a microorganism highly linked to bacterial vaginosis (BV), is understudied in terms of genotypic heterogeneity in South African populations. This study investigated the prevalence of G. vaginalis genotypes in BV-positive, BV-intermediate, and BV-negative South African pregnant women. Methods The study population included n = 354 pregnant women recruited from a public hospital in Durban, South Africa. The women provided self-collected vaginal swabs for BV diagnosis by Nugent scoring. For the genotyping assays, the 16S rRNA and sialidase A genes from BV-negative, BV-intermediate, and BV-positive samples were amplified with G. vaginalis-specific primers. The16S rRNA amplicon was digested with TaqI to generate genotyping profiles, and subtypes were determined by correlating BamHI and HindIII digestion profiles. Phylogenetic analysis was performed on the 16S rRNA and sialidase A sequences. The data analysis was performed with R Statistical Computing software, version 3.6.2. Results Two different genotypes, GT1 and GT2, were detected. The most prevalent genotype was GT1. Four subtypes (1, 2B, 2AB, and 2C) were shown to be present. The most prevalent subtype was 2B, followed by subtypes 1, 2C, and 2AB. The phylogenetic analysis of the 16S rRNA showed the presence of 5 clusters. The tree displayed clusters which contained sequences from the same BV group with different genotypes and subtypes. Clusters with sequences from across the BV groups carrying the same genotype and subtype were present. Diversity of the sialidase A across BV groups and genotypes was observed. Finally, the study did not find a significant association (p > 0.05) between reported symptoms of abnormal vaginal discharge and genotype harboured. Conclusion This study provided the first report on the diversity of G. vaginalis in South African pregnant women. Diversity assessments of G. vaginalis with respect to genotypes and virulence factors may aid in a greater understanding of the pathogenesis of this microorganism.


Background
Bacterial vaginosis (BV) is an imbalance of the vaginal microenvironment [1]. The condition is characterized by a lower abundance of "healthy" lactobacilli and overgrowth of diverse anaerobic bacteria such as Gardnerella, Atopobium, Mobiluncus, Prevotella, Bacteroides, Anaerococcus, Peptostreptococcus, Sneathia, and Leptotrichia and members of the class Clostridia [2]. Bacterial vaginosis has been associated with preterm birth and poor perinatal outcomes [3]. A strong association between BV and sexually transmitted infections (STIs) has also been reported [4].
Gardnerella vaginalis is found in most women with vaginosis and has been reported to be the main cause of clinical signs and symptoms used to diagnose BV [2,5]. G. vaginalis was originally discovered by Leopold [6] who described this microorganism as a "Haemophilus-like" species associated with prostatitis and cervicitis. G. vaginalis has the necessary virulence factors including production of sialidase which aids in the adherence to the host vaginal epithelium in order to compete with normal vaginal flora for dominance [7]. Sialidase produced by G. vaginalis degrades mucosal sialoglycans, believed to be important in BV [8]. Additionally, sialidase produced by some strains of G. vaginalis has been shown to interfere with host immune modulation resulting in adverse pregnancy outcomes [9].
For over three decades, researchers have been conducting extensive bacterial typing assays, in order to identify different virulence traits among Gardnerella spp. [10]. Phenotypic assays have been used to assess the diversity of Gardnerella spp. based on their biochemical properties such as production of b-galactosidase, lipase, and hippurate hydrolysis. However, the early typing assays had failed to reveal the diversity of G. vaginalis 5 [8],. The genetic heterogeneity of G. vaginalis species has been determined using molecular approaches, such as Amplified Ribosomal DNA restriction analysis (ARDRA) [11]. ARDRA is a simple, fast, and reproducible method for microbial molecular epidemiology and taxonomy [12]. The ARDRA genotyping approach developed by Balashov and coworkers was shown to be less error-prone [13]. In the study by Ingianni et al. [14], the ARDRA method allowed for G. vaginalis to be separated into at least 4 genotypes.
Despite the availability of useful genotyping techniques for G. vaginalis, it has been documented that there is limited data on the prevalence of G. vaginalis genotypes from across the globe [15]. This study investigated the diversity of G. vaginalis from noncultured vaginal swabs obtained from pregnant women by ARDRA.
Past studies have described a link with sialidase production and particular G. vaginalis ARDRA genotypes [9,15]. A recent study by our research group had found high copy numbers of the sialidase A gene across BV-intermediate and BVpositive women and in women with and without abnormal vaginal discharge (unpublished). However, the association between sialidase and G. vaginalis ARDRA genotypes was not performed. This current study will attempt to fill this gap in evidence. In addition, no clear association between BV and any of the ARDRA genotypes has been reported [11]. Through this study, the distribution of G. vaginalis ARDRA genotypes linked to BV status and clinical symptoms of BV such as abnormal vaginal discharge will be determined. The 16S rRNA gene specific to G. vaginalis was amplified using primers: forward: 5′-TTCGATTCTGGCTCAGG and reverse: 5 ′ -CCATCCC AAAAGGGTTAGGC. The primers were synthesized based on their published sequences described by Pleckaityte et al. [15]. The PCR was performed in a 50 μL final volume and comprised 0.2 μM of each primer, 30 ng of genomic DNA, and 1.5 U of High-Fidelity PCR enzyme mix (ThermoFisher Scientific, Massachusetts, United States). The reaction mixture was subjected to 28 cycles of denaturation at 94°C for 30 seconds, primer annealing at 52°C for 45 seconds, and extension at 72°C for 1 minute 25 seconds. PCR conditions were as per Pleckaityte et al. [15]. All PCR reactions were performed using a T100 thermocycler (BioRad, California, United States). The PCR products were separated on a 1% agarose gel and viewed under a UV transilluminator (Gene Genius, SYNGENE).

Sequence
Analysis of the 16S Ribosomal RNA. To confirm the identity of the PCR amplicons prior to genotyping, the amplicons were sequenced using the Sanger method [17] at Inqaba Biotechnological Industries in Pretoria, South Africa. The amplicons were sequenced using an ABI3500XL genetic analyser, and the raw sequence data was edited using Chromas software V2.6.5 (Technelysium, Queensland, Australia). The edited forward and reverse sequences were aligned using the DNAMAN software (Lynnon Biosoft, California, United States), and the identity of the edited sequences was confirmed using the National Center for Biotechnology Information (NCBI) Basic Local Alignment Search Tool (BLAST). 2.3.7. Detection of the Sialidase A Gene from G. vaginalis. The presence of the sialidase A gene was detected using the following specific primers: forward: 5 ′ -GACGACGGCGA ATGGCACGA-3′ and reverse: 5′-TACAAGCGGCTTTA CTCTTG-3′. The primers were synthesized based on their published sequences described by Pleckaityte et al. [15]. The PCR conditions were as follows: an initial denaturation at 95°C for 10 minutes was followed by 40 cycles of denaturation at 95°C for 30 seconds, annealing was performed at 58°C for 1 minute and extension at 72°C for 2 minutes, and this was followed by final extension at 72°C for 7 minutes. All PCR reactions were performed using a T100 thermocycler (BioRad, California, United States). PCR products were separated on a 1% agarose gel and viewed under a UV transilluminator (Gene Genius, SYNGENE).

Sequence Analysis of Sialidase A.
Amplicons generated using the sialidase A-specific primers were sequenced using the Sanger method as previously described. The sequence data generated for the sialidase A gene was compared across the genotypes.
2.3.9. Data Analysis. The data analysis was performed in R Statistical Computing software, version 3.6.2. To assess the association between the symptoms and the BV status for each genotype, the Chi-squared goodness of fit test for one sample was used. The results were also presented as component bar charts.

BV Diagnosis.
Of the 354 samples analysed, 124 were BV-positive, 37 were BV-intermediate, and 113 were BVnegative. The remaining slides (100) were unreadable due to poor quality of the slide (inadequate sample material on slide). We randomly selected 50 BV-negative, 37 BV-intermediate, and 50 BV-positive specimens for the genotypic analysis. A total of 137 samples were analysed. Attempts to generate amplicons for the unsuccessful samples were investigated such as increasing the concentration of template DNA and adjusting primer and amplification conditions. All attempts were unsuccessful. The possibility of sample inhibitors affecting the PCR reactions or failed DNA extractions was ruled out since the same DNA samples were amplifiable for other genes not included in this study. A set of 37 samples were used for further analysis. The DNA sequencing hits of the 16S rRNA showed identity (97%) to G. vaginalis strain GS10234 (MH898659.1) and G. vaginalis strain N153 (98%) (JQ354973.1).
3.3. Genotyping Analysis. The distribution of the genotypes based on TaqI digestion for the 37 specimens analysed is shown in Table 1. The subtypes of the genotypes which were determined by combining the banding profiles of BamHI and HindIII digestions are also presented in Table 1.
3.3.1. Genotypes Based on TaqI Digestion. Restriction digestion with TaqI revealed the presence of two different genotypes, i.e., GT1 and GT2. GT1 was carried by 20/37 specimens (54%), followed by GT2 which was present in 9/37 specimens (24%). Of the 37 specimens analysed, 7 specimens were not ascribed genotypes. Two of the specimens from the BV-positive sample group produced a banding profile (i.e., a single band at 500 bp) that was not described in previously published studies. One specimen from the BV-intermediate group produced a very faint profile which was difficult to interpret. The remaining 3 specimens did not produce any bands; the gel lanes appeared blank for those samples. These specimens were across both BV status groups. Within the BV-positive sample group, 13/23 specimens carried GT1 (57%) and 6 of the 23 specimens (26%) carried GT2. Two specimens produced a differing banding profile (9%), and 2 specimens did not produce any bands (9%) (Figure 1).

Phylogenetic Analysis of 16S rRNA Genotypes and
Subtypes. The phylogenetic tree revealed the presence of 5 sequence clusters ( Figure 5). The tree displayed clusters which contained groups of specimens from a particular BV group (clusters 1, 3, and 5). Within these same BV groups, there were however differences noted for either the genotypes assigned and/or subtypes present. Additionally, there were clusters which contained specimens from across both BV groups such as clusters 2 and 4. Despite the heterogeneity with respect to the BV group, cluster 2 contained specimens of the same genotype (GT1) with the majority carrying the same subtype (S2B). Cluster 4 on the other hand contained  in women who reported abnormal vaginal discharge when compared to women who did not report the discharge (p > 0:05) (Figure 6(a)). Similarly, for the women harbouring GT2, there was no significant difference in the BV-positive women who reported abnormal vaginal discharge when compared to women who did not report discharge (p > 0:05). All BVintermediate women with GT2 reported no symptoms of abnormal vaginal discharge (Figure 6(b)).

Discussion
Gardnerella vaginalis is one of the most frequently isolated microorganisms from women who present with symptoms of BV [19]. High microbial loads of G. vaginalis in the vaginal tract have been linked to reproductive health issues such as infertility and preterm labour [20]. The pathogenesis of G. vaginalis in the vaginal tract is not completely understood since this microorganism has been shown to be present across the BV groups (BV-negative, BV-intermediate, and BV-positive). Differentiation of G. vaginalis strains and subgroups according to sequence variations in 16S rRNA and the cpn60 genes has been made possible using molecular biology approaches [13].
In this study, the diversity of the G. vaginalis 16S rRNA was analysed across BV-intermediate and BV-positive pregnant women who were diagnosed by the Nugent method. A BV-negative group was not included in the diversity analysis since none of the BV-negative samples produced a positive PCR amplicon for the 16S rRNA specific for G. vaginalis. However, the presence of Lactobacillus crispatus was shown to be present in the negative specimens eliminating the possibility of a failed DNA extraction or PCR amplification for the negative specimens (data not shown). Our failure to amplify the G. vaginalis 16S rRNA in the negative samples differs from previously published works which had shown the presence of G. vaginalis in BV-negative specimens based on PCR detection of the 16S rRNA gene [13,21]. However, a fairly recent study conducted by our research group was able to detect G. vaginalis in our BV-negative pregnant cohort using the highly sensitive Droplet Digital PCR (ddPCR) System (submitted for publication). This leads to the assumption that ARDRA may not be a very sensitive method for detecting G. vaginalis directly from clinical samples in women classified as BV-negative. Our assumption is validated by an earlier study conducted by Verhelst et al. [22] which used ARDRA in order to assess the diversity of the vaginal microbiome. In that study, ARDRA failed to identify G. vaginalis in women who were classified as BVnegative; however, G. vaginalis was detected in women who were BV-positive. Despite the suggested limitation, ARDRAhas been useful in identifying different G. vaginalis genotypes [9,14,23].
Based on the ARDRA technique used in this study, restriction digestion with TaqI revealed the presence of two different genotypes, i.e., GT1 and GT2. Similarly, a study by Pleckaityte et al. [15] reported on the presence of GT1 and GT2 in a population of women with BV in Lithuania. However, the Lithuanian study was unable to detect specific subtypes associated with GT1. All the GT1 sequences in their population of women were identical. However, the present study observed different subtypes associated with GT1. This suggests that a level of diversity does exist between G. vaginalis present in different geographical locations as well as across different population groups (pregnant versus nonpregnant). Additionally, the present study observed a difference in the prevalence of the different subtypes across BV-intermediate and BV-positive women. Among the BV-positive women, the most prevalent subtype was 2B whereas in the BV-   The study further investigated the link between genotypes and clinical symptoms of abnormal vaginal discharge. Among the BV-intermediate and BV-positive groups, a higher percentage of the women did not present with symptoms of abnormal vaginal discharge (i.e., were asymptomatic). This study found no significant association between genotypes harboured and symptoms of abnormal vaginal discharge. However, a study conducted by Santiago et al. [9] showed GT2 to be the most prevalent genotype associated with symptomatic BV.
This study also investigated the diversity of the sialidase A gene, the virulence factor of G. vaginalis in relation to genotypes. In this study, there was a correlation between the DNA sequences of the sialidase A gene and the respective genotypes. Two of the four sequence clusters contained samples of the same genotype. A clear link between genotype and sialidase production was previously reported by Santiago et al. [9]. However, the present study cannot be directly compared to that of Santiago et al. [9] since the present study detected the sialidase A gene directly from noncultured vaginal swabs whereas Santiago et al. [9] evaluated pure cultures for sialidase activity and diversity. The present study also showed that the diversity of the sialidase A gene was based on the BV group. BV-intermediate and BV-positive sequences formed distinct clusters on the phylogenetic tree indicating a level of diversity of sialidase A gene across the two BV groups. Previous studies on the diversity of sialidase A gene were not investigated in women with intermediate BV. The current study now provides additional data on the diversity of the sialidase A gene for this BV group.
The limitations of the study are as follows: the sample size used for the analysis was small. However, despite the small However, from the noncultured clinical specimens, the diversity assessments performed still provided substantial evidence. Lastly, due to the cross-sectional nature of the study, we did not associate the genotypes with pregnancy outcomes and acquisition of other infections such as HIV and genital infections. All the limitations described here will be addressed in a study that is planned for commencement in 2021. V03 -GT2, S2B V10 -GT1, S1 V56 -GT1, S1 V21 -GT1, S2B V27 -GT1, S1 V85 -S1 V102 -S1 V44 -GT1, S2C 100 Figure 5: Phylogenetic analysis according to distribution of genotypes. The tree was constructed using the Neighbour Joining method. The optimal tree with the sum of branch length = 15:35302543 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (100 replicates) is shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using the Maximum Composite Likelihood method and are in the units of the number of base substitutions per site. Evolutionary analyses were conducted in MEGA X [18].

Conclusion
This study provides the first report on the most prevalent genotypes and subtypes of G. vaginalis across BVintermediate and BV-positive South African pregnant women. Restriction analysis revealed the presence of two different genotypes, i.e., GT1 and GT2, as well as four subtypes (1, 2B, 2AB, and 2C) circulating in our population. In addition, diversity across the BV groups, genotypes, and subtypes for sialidase A was evident in this study. The observed diversity can be used as a foundation for future studies which are aimed at understanding the pathogenesis of G. vaginalis across BV groups in women from different populations.   Figure 7: Phylogenetic analysis of sialidase A in relation to genotype. The tree was constructed using the Neighbour Joining method. The evolutionary history was inferred using the Neighbour Joining method. The optimal tree with the sum of branch length = 0:18320599 is shown. The percentage of replicate trees in which the associated taxa clustered together in the bootstrap test (100 replicates) is shown next to the branches. The tree is drawn to scale, with branch lengths in the same units as those of the evolutionary distances used to infer the phylogenetic tree. The evolutionary distances were computed using evolutionary analyses which were conducted in MEGA X [18].