Atopobium vaginae and Prevotella bivia Are Able to Incorporate and Influence Gene Expression in a Pre-Formed Gardnerella vaginalis Biofilm

Bacterial vaginosis (BV) is associated with a highly structured polymicrobial biofilm on the vaginal epithelium where Gardnerella species presumably play a pivotal role. Gardnerella vaginalis, Atopobium vaginae, and Prevotella bivia are vaginal pathogens detected during the early stages of incident BV. Herein, we aimed to analyze the impact of A. vaginae and P. bivia on a pre-established G. vaginalis biofilm using a novel in vitro triple-species biofilm model. Total biofilm biomass was determined by the crystal violet method. We also discriminated the bacterial populations in the biofilm and in its planktonic fraction by using PNA FISH. We further analyzed the influence of A. vaginae and P. bivia on the expression of key virulence genes of G. vaginalis by quantitative PCR. In our tested conditions, A. vaginae and P. bivia were able to incorporate into pre-established G. vaginalis biofilms but did not induce an increase in total biofilm biomass, when compared with 48-h G. vaginalis biofilms. However, they were able to significantly influence the expression of HMPREF0424_0821, a gene suggested to be associated with biofilm maintenance in G. vaginalis. This study suggests that microbial relationships between co-infecting bacteria can deeply affect the G. vaginalis biofilm, a crucial marker of BV.


Introduction
Bacterial vaginosis (BV) is the most-common vaginal infection affecting fertile, premenopausal, and pregnant women [1]. It is associated with important adverse outcomes related to pregnancy [2] and infertility [3]. Additionally, it is associated with an increased risk of acquisition of HIV and other sexually transmitted infections (STIs) [4][5][6][7][8]. Microbiologically, BV is a complex polymicrobial infection where beneficial vaginal bacteria, mainly hydrogen peroxide and lactic acid-producing Lactobacillus species, which are usually dominant in vaginal microbiota of healthy women, are replaced by high concentrations of facultative and strict anaerobic bacteria [1,9]. The most prominent of these are Gardnerella spp., facultative anaerobes usually found embedded in a polymicrobial BV biofilm [10][11][12]. However, Gardnerella spp. are also commonly found in asymptomatic or BV-negative women [13]. This has aroused interest in whether genetic differences among G. vaginalis isolates might differentiate pathogenic from commensal organisms [14]. G. vaginalis was the only recognized species in its genus for over four decades, but very recently the Gardnerella taxonomic description was amended based on comparisons of whole-genome sequencing and matrix-assisted laser desorption/ionization time-of-flight (MALDI-TOF) mass spectrometry analysis, resulting in four species (G. vaginalis, G. leopoldii,

Quantification of the Biomass of Mono-, Dual-, and Triple-Species BV-Associated Biofilms
In the BV-associated vaginal ecosystem, resident microorganisms interact with each other in both synergistic and antagonistic manners, which might affect their ability to form biofilms in this polymicrobial community [35]. Initially, we compared the in vitro biofilm formation ability of each tested species for 24 h and 48 h using optimized [38] in vitro conditions ( Figure 1A). In both timepoints tested, G. vaginalis formed a biofilm with the highest total biomass, while P. bivia formed a biofilm with the lowest total biomass. Following the hypothesis that Gardnerella is the early colonizer in BV [9], we then assessed how A. vaginae and P. bivia could incorporate into a 24-h pre-formed G. vaginalis biofilm. Under the tested conditions, all the consortia reached the same level of total biofilm biomass ( Figure 1B). Curiously, when compared to the mono-species G. vaginalis 48-h biofilm, none of the consortia provided an added advantage in terms of an increase in total biomass. Under the tested conditions, all the consortia reached the same level of total biofilm biomass ( Figure 1B). Curiously, when compared to the mono-species G. vaginalis 48-h biofilm, none of the consortia provided an added advantage in terms of an increase in total biomass. While it was tempting to compare the 48-h consortia with A. vaginae or P. bivia 48-h mono-species biofilms, it is important to highlight that both species were only allowed to grow for 24 h, after the initial 24-h G. vaginalis biofilm was performed, as described in the methods section. Dual-and triple-species biofilms were initiated by inoculating a bacterial suspension of G. vaginalis into 24-well tissue culture plates in New York City III (NYC III) medium and by incubating the plates for 24 h, at 37 °C under anaerobic conditions. After 24 h, planktonic cells were removed, and each bacterial species, Atopobium vaginae or Prevotella bivia (for dual-species biofilms) and A. vaginae and P. bivia (for triple-species biofilms), were inoculated in the pre-formed G. vaginalis biofilms and incubated for another 24 h. Each data point represents the mean ± s.d. of three independent assays, with four technical replicates assessed each time. * Values were significantly different between 24-h and 48-h mono-species biofilms (paired t-test, p < 0.05). Abbreviations: A. vaginae (Av), G. vaginalis (Gv), and P. bivia (Pb).

In Vitro PNA Gard162 and PNA AtoITM1 Probes Specificity and Efficiency
Although the PNA Gard162 [39] and PNA AtoITM1 [40] probes' specificity had been previously tested for several BV-associated bacteria, we also analyzed these probes' specificity for the bacterial strains used in this study. Based on our results (Table 1), PNA Gard162 and PNA AtoITM1 probes hybridized with G. vaginalis and A. vaginae, respectively, whereas no hybridization was observed for the other species tested, showing a specificity of 100% as previously reported. Additional details of the specificity of the Gard162 and AtoITM1 probes are shown in Supplementary Figures S1 and S2. Table 1. Bacterial species used in peptide nucleic acid fluorescence in situ hybridization (PNA FISH) assays and their specificity with PNA probes a .

In Vitro PNA Gard162 and PNA AtoITM1 Probes Specificity and Efficiency
Although the PNA Gard162 [39] and PNA AtoITM1 [40] probes' specificity had been previously tested for several BV-associated bacteria, we also analyzed these probes' specificity for the bacterial strains used in this study. Based on our results (Table 1), PNA Gard162 and PNA AtoITM1 probes hybridized with G. vaginalis and A. vaginae, respectively, whereas no hybridization was observed for the other species tested, showing a specificity of 100% as previously reported. Additional details of the specificity of the Gard162 and AtoITM1 probes are shown in Supplementary Figures S1 and S2. Table 1. Bacterial species used in peptide nucleic acid fluorescence in situ hybridization (PNA FISH) assays and their specificity with PNA probes a .

Strains
Gard162 Probe Specificity AtoITM1 Probe Specificity G. vaginalis strain ATCC 14018 T ++++ − A. vaginae strain ATCC BAA-55 T − ++++ P. bivia strain ATCC 29303 T − − a PNA probes' (Gard162 and AtoITM1) specificity was tested for each species, with the following hybridization PNA FISH qualitative evaluation: (−) absence of hybridization; (++++); and optimal hybridization. As no P. bivia PNA FISH probe currently exists, the estimation of P. bivia counts could only be assessed indirectly by 4 -6-diamidino-2-phenylindole (DAPI) counterstaining, assuming that all cells with unlabeled PNA probes were P. bivia; however, this needs to be experimentally determined [41]. As such, we compared the data obtained from PNA FISH and DAPI counts for both G. vaginalis and A. vaginae pure-culture biofilms and the Pathogens 2021, 10, 247 4 of 15 planktonic fractions of the biofilm. Not surprisingly, each probe failed to detect 100% of the respective total cells. By performing serial dilutions of each sample, calibration curves were obtained for G. vaginalis biofilm (Figure 2A) or planktonic cells ( Figure 2B) and for A. vaginae biofilm ( Figure 2C) or planktonic cells ( Figure 2D).
As no P. bivia PNA FISH probe currently exists, the estimation of P. bivia counts co only be assessed indirectly by 4′-6-diamidino-2-phenylindole (DAPI) counterstaining suming that all cells with unlabeled PNA probes were P. bivia; however, this needs t experimentally determined [41]. As such, we compared the data obtained from PNA F and DAPI counts for both G. vaginalis and A. vaginae pure-culture biofilms and the pl tonic fractions of the biofilm. Not surprisingly, each probe failed to detect 100% of respective total cells. By performing serial dilutions of each sample, calibration cu were obtained for G. vaginalis biofilm (Figure 2A) or planktonic cells ( Figure 2B) and A. vaginae biofilm ( Figure 2C) or planktonic cells ( Figure 2D). Taking into consideration this data, it was possible to calculate the efficiency of e probe and to develop an equation that would correct PNA counts, to prevent overesti tion of DAPI counts as P. bivia counts, as listed in Table 2. Taking into consideration this data, it was possible to calculate the efficiency of each probe and to develop an equation that would correct PNA counts, to prevent overestimation of DAPI counts as P. bivia counts, as listed in Table 2.

Quantification and Distribution of Bacterial Populations in Dual-and Triple-Species BV-Associated Biofilms by PNA FISH
Taking advantage of the robustness of the PNA FISH/DAPI approach for the differentiation between G. vaginalis, A. vaginae, and P. bivia, we discriminated the bacterial populations into dual-and triple-species BV-associated biofilms and their planktonic fractions. Similar to what was described before [42], we showed that A. vaginae and P. bivia were able to incorporate in the dual-species biofilms, accounting for up to ∼23% and ∼38% of the total number of cells, respectively ( Figure 3). Curiously, in the triple-species biofilms, the relative load of A. vaginae was reduced to~8.3%, maintaining G. vaginalis as the main species. The lower ability of A. vaginae to integrate the triple-species biofilms is evidence that somewhat specific interactions are established when these three bacterial species act as a consortium, which in our tested conditions promoted the enhanced integration of P. bivia, in depreciation of A. vaginae. When looking at the bacterial populations found on the planktonic fraction of the biofilm, it is noteworthy that significant differences were found both for P. bivia and the triple-species consortia; in contrast the relative composition of both biofilms and planktonic fractions were similar for A. vaginae.

Quantification and Distribution of Bacterial Populations in Dual-and Triple-Species BV-Associated Biofilms by PNA FISH
Taking advantage of the robustness of the PNA FISH/DAPI approach for the differentiation between G. vaginalis, A. vaginae, and P. bivia, we discriminated the bacterial populations into dual-and triple-species BV-associated biofilms and their planktonic fractions. Similar to what was described before [42], we showed that A. vaginae and P. bivia were able to incorporate in the dual-species biofilms, accounting for up to ∼23% and ∼38% of the total number of cells, respectively ( Figure 3). Curiously, in the triple-species biofilms, the relative load of A. vaginae was reduced to ~8.3%, maintaining G. vaginalis as the main species. The lower ability of A. vaginae to integrate the triple-species biofilms is evidence that somewhat specific interactions are established when these three bacterial species act as a consortium, which in our tested conditions promoted the enhanced integration of P. bivia, in depreciation of A. vaginae. When looking at the bacterial populations found on the planktonic fraction of the biofilm, it is noteworthy that significant differences were found both for P. bivia and the triple-species consortia; in contrast the relative composition of both biofilms and planktonic fractions were similar for A. vaginae. Figure 3. Bacterial populations in dual-and triple-species BV-associated biofilms and in their respective planktonic fraction. Biofilms were disrupted and resuspended before quantification was performed, as described in methods section. Percentage of cells detected by PNA FISH for 48-h biofilms and in their planktonic fraction. Each data point represents the mean ± s.d. of three independent assays. For each assay, 20 fields were randomly acquired in each sample and the number of bacteria per image was counted using ImageJ Software. Values were significantly different between the percentage of each bacterial species that integrates the dual-or triple-species biofilm or planktonic fraction, namely, * Gv BIOF vs. Gv PLANK, γ Av BIOF vs. Av PLANK, τ Pb BIOF vs. Pb PLANK (two-way ANOVA test, p < 0.05). Abbreviations: A. vaginae (Av), G. vaginalis (Gv), and P. bivia (Pb); biofilm BIOF; planktonic (PLANK). Bacterial populations in dual-and triple-species BV-associated biofilms and in their respective planktonic fraction. Biofilms were disrupted and resuspended before quantification was performed, as described in methods section. Percentage of cells detected by PNA FISH for 48-h biofilms and in their planktonic fraction. Each data point represents the mean ± s.d. of three independent assays. For each assay, 20 fields were randomly acquired in each sample and the number of bacteria per image was counted using ImageJ Software. Values were significantly different between the percentage of each bacterial species that integrates the dual-or triple-species biofilm or planktonic fraction, namely, * Gv BIOF vs. Gv PLANK, γ Av BIOF vs. Av PLANK, τ Pb BIOF vs. Pb PLANK (two-way ANOVA test, p < 0.05). Abbreviations: A. vaginae (Av), G. vaginalis (Gv), and P. bivia (Pb); biofilm BIOF; planktonic (PLANK).
We also analyzed bacterial distribution in the intact structure of the dual and triplespecies biofilms by confocal laser scanning microscopy (CLSM). As shown in Figure 4, in both dual-and triple-species biofilms, A. vaginae and P. bivia were found well-distributed across the G. vaginalis biofilm, in small clusters of cells. Details in the orthogonal views of mono-, dual-, and triple-species biofilms are shown in Supplementary Figure S3.
We also analyzed bacterial distribution in the intact structure of the dual and triplespecies biofilms by confocal laser scanning microscopy (CLSM). As shown in Figure 4, in both dual-and triple-species biofilms, A. vaginae and P. bivia were found well-distributed across the G. vaginalis biofilm, in small clusters of cells. Details in the orthogonal views of mono-, dual-, and triple-species biofilms are shown in Supplementary Figure 3.

Impact of A. vaginae and P. bivia on G. vaginalis Virulence
Considering the central role often attributed to Gardnerella in BV etiology [1,43], shifts in the G. vaginalis transcriptome during the establishment of polymicrobial BV-associated biofilms could be a key for unveiling interspecies interactions that enhance the virulence of G. vaginalis. Thus, we focused on deciphering the impact of A. vaginae and/or P. bivia on G. vaginalis virulence. As such, we analyzed the expression of genes related to cytotoxicity, biofilm maintenance, antimicrobial resistance, and evasion of the immune system in cells from mono-, dual-, and triple-species biofilms. Regarding cytotoxicity, G. vaginalis produces the toxin vaginolysin (vly), which might induce lysis in vaginal cells membranes [44,45]. However, as shown in Figure 5A, according to our in vitro conditions, no significant differences were found in the expression of this gene by G. vaginalis between the different biofilm models. We also addressed sialidase (sld) expression, since sialidase appears to contribute to G. vaginalis cytotoxicity by the destruction of the protective mucus layer on the vaginal epithelium [46]. Although we detected a slight decrease in sld transcription, it was not statistically significant in dual-and triple-species biofilms, as compared with mono-species G. vaginalis biofilms ( Figure 5B).
Regarding the glycosyltransferases, it has been proposed that they are involved in the transfer of a sugar moiety to a substrate and are thus essential in the biosynthesis of glycoconjugates like exopolysaccharides and glycoproteins, which are important for bio-

Impact of A. vaginae and P. bivia on G. vaginalis Virulence
Considering the central role often attributed to Gardnerella in BV etiology [1,43], shifts in the G. vaginalis transcriptome during the establishment of polymicrobial BV-associated biofilms could be a key for unveiling interspecies interactions that enhance the virulence of G. vaginalis. Thus, we focused on deciphering the impact of A. vaginae and/or P. bivia on G. vaginalis virulence. As such, we analyzed the expression of genes related to cytotoxicity, biofilm maintenance, antimicrobial resistance, and evasion of the immune system in cells from mono-, dual-, and triple-species biofilms. Regarding cytotoxicity, G. vaginalis produces the toxin vaginolysin (vly), which might induce lysis in vaginal cells membranes [44,45]. However, as shown in Figure 5A, according to our in vitro conditions, no significant differences were found in the expression of this gene by G. vaginalis between the different biofilm models. We also addressed sialidase (sld) expression, since sialidase appears to contribute to G. vaginalis cytotoxicity by the destruction of the protective mucus layer on the vaginal epithelium [46]. Although we detected a slight decrease in sld transcription, it was not statistically significant in dual-and triple-species biofilms, as compared with mono-species G. vaginalis biofilms ( Figure 5B).
Regarding the glycosyltransferases, it has been proposed that they are involved in the transfer of a sugar moiety to a substrate and are thus essential in the biosynthesis of glycoconjugates like exopolysaccharides and glycoproteins, which are important for biofilm maintenance to maximize the full virulence of G. vaginalis [47,48]. Of note, the expression of the HMPREF0424_0821 transcript, which encodes glycosyltransferases type II, was up-regulated in all the tested conditions, being approximately 2.8-fold higher in triplespecies than in mono-species biofilms (p < 0.05; Figure 5C). We also analyzed the expression of transcripts encoding antimicrobial-specific resistance proteins belonging to efflux pump families (HMPREF0424_1122 and HMPREF0424_0156). Despite detecting slight changes in the transcription of these genes, these were not statistically significant ( Figure 5D,E). Nevertheless, a tendency was observed in the presence of P. bivia, with a downregulation of the tested G. vaginalis genes, similar to what we have previously found [37]. Finally, we analyzed the expression of HMPREF0424_1196 transcript, which encodes a Rib-protein that belongs to the α-like protein (Alf)-family of highly repetitive surface antigens [49], which elicit protective immunity through their inter-strain size variability [48]. Here the most striking difference was in the triple-species biofilm, as compared to the G. vaginalis mono-species biofilm (p < 0.05; Figure 5F). film maintenance to maximize the full virulence of G. vaginalis [47,48]. Of note, the expression of the HMPREF0424_0821 transcript, which encodes glycosyltransferases type II, was up-regulated in all the tested conditions, being approximately 2.8-fold higher in triplespecies than in mono-species biofilms (p < 0.05; Figure 5C). We also analyzed the expression of transcripts encoding antimicrobial-specific resistance proteins belonging to efflux pump families (HMPREF0424_1122 and HMPREF0424_0156). Despite detecting slight changes in the transcription of these genes, these were not statistically significant ( Figure  5D,E). Nevertheless, a tendency was observed in the presence of P. bivia, with a downregulation of the tested G. vaginalis genes, similar to what we have previously found [37]. Finally, we analyzed the expression of HMPREF0424_1196 transcript, which encodes a Rib-protein that belongs to the α-like protein (Alf)-family of highly repetitive surface antigens [49], which elicit protective immunity through their inter-strain size variability [48].
Here the most striking difference was in the triple-species biofilm, as compared to the G. vaginalis mono-species biofilm (p < 0.05; Figure 5F).

Discussion
The dynamic and complex nature of the vaginal microbiota and the likely role of multiple bacterial species in the pathogenesis of BV have posed major challenges for developing realistic polymicrobial in vitro biofilm models [50,51]. To date, the majority of in vitro studies only address mono-or dual-species biofilms, and are focused on Gardnerella

Discussion
The dynamic and complex nature of the vaginal microbiota and the likely role of multiple bacterial species in the pathogenesis of BV have posed major challenges for developing realistic polymicrobial in vitro biofilm models [50,51]. To date, the majority of in vitro studies only address mono-or dual-species biofilms, and are focused on Gardnerella species [37,52]. Herein, we describe, for the first time, an in vitro biofilm composed of G. vaginalis, A. vaginae, and P. bivia, three highly relevant BV-associated species [28,30]. In a polymicrobial community, bacterial species interact extensively with each other and these interactions might also determine the structure and composition of multi-species biofilms [53]. It is, thus, reasonable to assume that in triple-species BV-Pathogens 2021, 10, 247 8 of 15 associated biofilms the microbial interactions become more complex than in dual-or mono-species biofilms [54].
Our experimental model follows the hypothesis that Gardnerella is the early colonizer in BV [9], to which we later allowed co-incubation with A. vaginae and P. bivia. In vitro experimental data supporting the role of some Gardnerella species to be the early colonizers is derived from the fact that some Gardnerella isolates have a significantly higher ability to adhere to epithelial cells than many other BV-associated species [52,55,56] and are also able to displace vaginal lactobacilli [57], a pivotal step in the development of a characteristic multi-species biofilm. It is also important to bear in mind that the recent reclassification of G. vaginalis [15] in multiple Gardnerella species might have important implications in BV etiology, as a particular species might have a different role in BV development, although this still needs to be further explored [19,21,58]. Herein, we selected a type strain of G. vaginalis since its complete genome is available, and thus, it is well genotypically characterized, being used in several studies associated with BV [15,58]. Regarding other BV-associated species, we selected A. vaginae and P. bivia to be included in this study, due to data suggesting they may be more than bystanders [9,32]. Indeed, in a prospective vaginal microbiome study Muzny and colleagues showed that the mean relative abundance of P. bivia, Gardnerella spp., and A. vaginae became significantly higher in cases four days before (P. bivia), three days before (Gardnerella spp.), and on the day of (A. vaginae) incident BV onset [28]. Based on this study, the authors suggested that together with virulent G. vaginalis, P. bivia, and A. vaginae may have potential key roles in the induction and development of incident BV [9]. Nevertheless, knowledge about the microbial relationships between these three bacterial species remains scarce. In order to shed new light on this aspect, we analyzed the effect of A. vaginae and P. bivia on biofilm formation and its impact on G. vaginalis pathogenicity. Similar to what is described in oral infections, in which the interactions between microbial communities have tremendous importance for the development of oral disease [59], we hypothesize that G. vaginalis, A. vaginae, and P. bivia establish a network of interactions that affect the development of the BV-associated biofilm, a key hallmark of BV. Curiously, under the tested conditions, inoculation or co-inoculation of the different species on the pre-formed G. vaginalis biofilm did not significantly enhance the amount of total biofilm biomass. However, despite widespread utilization of the CV staining method for biofilm quantification used in this study, an inherent limitation of this method is that total biomass comparison between species is not possible, since different species produce distinct biofilm matrices and have different cell sizes [60,61]. Thus, interpretation of these data should be made with reservation. Nevertheless, as shown in our CLSM and PNA-FISH quantitative data, dual-and triple-species biofilms contained significant numbers of each species, with G. vaginalis as the most prominent species, similar to in vivo studies [11,22]. Interestingly, our PNA FISH data suggest that different microbial relationships are established in dualand triple-species biofilms. This is inferred by the distinct bacterial composition observed in multi-species biofilms, in particular the triple-species biofilm where the relative composition of G. vaginalis and P. bivia increased while A. vaginae decreased. This is even more relevant since P. bivia formed a weak mono-species biofilm and grew slower than A. vaginae and G. vaginalis in New York City III (NYC III) medium, as shown previously [38]. It is important to mention that several factors influenced the bacterial integration in biofilm, including the growth rates of the bacterial species [62] and the ability to adhere to a surface [63,64] and to each other [65,66]. Interestingly, despite the slower growth rate, P. bivia was better fit to grow in the biofilm than A. vaginae in our tested conditions. This suggests that G. vaginalis modifies the local environment, making it more favorable for the growth of P. bivia, which might be a result of an ammonia flow mechanism as proposed by Pybus and colleagues [29].This further demonstrates that different bacterial species influence the growth of other species, and likely have an impact in BV etiology. It is also interesting to observe that the planktonic fraction of the dual-and triple-species biofilm did not always coincide with the biofilm consortia, suggesting that the specific microenvironment of the biofilms provide different competitive advantages, as shown elsewhere [67,68].
It is noteworthy that synergistic effects were, however, observed when we analyzed changes in G. vaginalis gene expression. The expression of the HMPREF0424_0821 transcript, which encodes a glycosyltransferase, was significantly up-regulated in the multi-species consortia. Glycosyltransferases have been described as key enzymes required for biofilm maintenance [48], and may also have a putative role for cell surface glycoconjugates which has been proposed to shape vaginal microbiota-host interactions [69]. This enhancement was only significant in the triple-species biofilm model, highlighting, therefore, the importance of the interplay between multiple bacterial species in the development of BV. Studies to assess the role of glycosyltransferase on biofilm formation and virulence have been conducted for other species, namely for Streptococcus mutans, a bacterium responsible for the initiation of tooth decay [70]. It was shown that glycosyltransferase mediated biofilm matrix dynamics and virulence. Interestingly, the deletion of the glycosyltransferase gene resulted in no change in overall biofilm biomass, however, the mutant strain originated an altered biofilm architecture. Concurrently, the mutant was less virulent in an in vivo rat model of dental caries [70]. Corroborating the relevance of glycosyltransferase in biofilms, we have recently shown that this gene was up-regulated in Gardnerella spp. in 15 different dual-species consortia [37]. Together, these data emphasize the importance of other BV-associated species in G. vaginalis virulence, and consequently, in host infection.
While we have previously shown that P. bivia could enhance G. vaginalis vly expression two-fold, when a dual-species biofilm was grown in supplemented brain-heart infusion (sBHI) medium [37], in NYC III medium vly expression was not changed. Regarding the role of sld, some early studies pointed out that this gene is strongly linked with the development of a biofilm [71,72], however, in our tested conditions, no relevant changes were verified when G. vaginalis was cultivated in mono-or multi-species biofilms, corroborating our previous findings [37]. Such evidence supports a recent study that postulated that sld does not likely have a role in establishing or maintaining the biofilm [73]. It should be pointed out that in this study we chose to use NYC III medium, since sBHI was not appropriate to induce mono-species biofilms from either P. bivia or A. vaginae [38]. Since bacterial gene expression is strongly influenced by media conditions [74,75], we cannot exclude that the optimal growth conditions provided by NYC III could also somehow be affecting gene expression by G. vaginalis. Indeed, by comparing the data from this study with our previous findings [37], we observed that using NYC III, the base level of G. vaginalis vly expression was approximately 10-fold higher than in comparison with sBHI, which could explain the differences observed in both studies. On account of the fact that synergistic effects often occur when one bacterium is providing some advantage to another [76,77], by using an optimal biofilm-inducing media such as NYC III, it might be possible that some synergistic effects were masked. Evidence to support this hypothesis can be observed by the fact that several strains of A. vaginae easily die out when grown in sBHI medium, but maintain viability when co-cultured with G. vaginalis [78]. However, in NYC III, A. vaginae is able to survive alone. Furthermore, for the other tested genes, with exception of sld, we also noted a higher base level (2-to 5-fold) of expression in NYC III as compared with sBHI. Despite these differences, a similar effect on G. vaginalis transcriptomic profile was observed for the remaining tested genes, as compared with our previous dual-species biofilm study [37]. However, further studies are required to elucidate the influence of the growth media in gene expression by G. vaginalis. In triple-species biofilms, we observed a downregulation of these genes when compared to G. vaginalis mono-species biofilms, however, this requires future study to understand the molecular mechanisms involved in antimicrobial resistance and evasion of the immune system. Taken together, our data suggest that microbial relationships between co-infecting bacteria can influence the development of a polymicrobial biofilm, a marker of BV. However, more research is needed to provide a better mechanistic insight into the complex interplay between G. vaginalis, A. vaginae, and P. bivia, and their eukaryotic hosts. While the choice of a very rich medium was used to guarantee that all the tested bacteria were able to grow in vitro in a biofilm phenotype, trying to determine these microbial interactions in conditions more similar to the vaginal environment might provide novel insights, as we previously demonstrated that growing Gardnerella spp. in a medium simulating genital tract secretions and complemented with components of the host immune system had a significant impact on the growth and biofilm formation [79]. Understanding the molecular basis and biological effect of these microbial interactions and microbial-host interactions may provide novel information necessary to define more effective and goal-oriented treatment in BV and improve women's reproductive health.

Biofilm Formation and Biomass Quantification by the Crystal Violet Method
Dual-and triple-species biofilms were initiated by inoculating a 10 7 colony-forming units (CFU)/mL bacterial suspension of G. vaginalis into 24-well tissue culture plates (Orange Scientific, Braine L'Alleud, Belgium) and by incubating the plates for 24 h, at 37 • C, and under anaerobic conditions. After 24 h, planktonic cells were removed, and 10 7 CFU/mL of each bacterial species, A. vaginae or P. bivia (for dual-species biofilms) and A. vaginae, P. bivia (for triple-species biofilms), were inoculated in the pre-formed G. vaginalis biofilms and incubated for another 24 h. Of note, we first adjusted the bacterial concentration of the bacterial suspension to 9 × 10 7 CFU/mL due to the limit of detection of the microplate reader and then we diluted to 1 × 10 7 CFU/mL, confirming this concentration by CFUs. At 620 nm, 9 × 10 7 CFU/mL of G. vaginalis corresponds to an optical density (OD) = 0.15; for A. vaginae an OD = 0.11, and for P. bivia an OD = 0.16. Mono-species biofilms of G. vaginalis were grown as a control for 48 h, in which fresh medium was added to the respective wells after the first 24 h of biofilm formation. In addition, we also performed 24-h mono-species biofilm growth for A. vaginae and P. bivia to examine their individual ability to grow in the tested conditions. To quantify the biomass of mono-, dual-, and triple-species biofilms, we used the crystal violet (CV) method [80]. In brief, after the fixation step with 100% (v/v) methanol (Thermo Fisher Scientific) for 20 min, biofilms were stained with CV solution at 1% (v/v) (Merck, Darmstadt, Germany) for 20 min. Each well was washed twice with phosphate-buffered saline, and bound CV was released with 33% (v/v) acetic acid (Thermo Fisher Scientific). To estimate total biofilm biomass, the OD of the resulting solution was measured at 595 nm. Biofilm assays were repeated three times on separate days, with four technical replicates assessed each time.

In Vitro PNA Gard162 and AtoITM1 Probes Specificity and Efficiency
Although the specificity of PNA Gard162 [39], and AtoITM1 [40] probes have been previously tested, we also evaluated the probe specificity for the bacterial species used in this study. Thus, we performed an experiment in order to detect any possible crosshybridization. The evaluation of PNA FISH hybridization was based on a qualitative score, as previously described [39]: (−) absence of hybridization, (++) moderate hybridization, (+++) good hybridization, and (++++) optimal hybridization.
We then carried out an experiment to analyze the efficiency of both PNA probes. As such, we performed several dilutions from pure bacterial suspensions obtained from mono-species biofilms and their planktonic fractions. To determine the efficiency of each probe, the same sample was hybridized with a species-specific probe and then stained with 4 -6-diamidino-2-phenylindole (DAPI, 2.5 µg/mL) to account for nonhybridizing bacteria; defined as double staining. After the double staining, the bacteria were enumerated at two different wavelengths at the same position within the sample. Based on both data, we performed a correlation between the PNA counts and the DAPI counts that allowed us to obtain the equations shown in Figure 2. All experiments were performed in triplicate.

Quantification of Bacterial Populations in Dual-and Triple-Species Biofilms and Their Planktonic Fraction by PNA FISH
The bacterial population within the 48-h multi-species biofilms and in their planktonic fraction was discriminated using the peptide nucleic acid fluorescence in situ hybridization (PNA FISH) method, as previously described [37]. Briefly, after fixing the biofilm suspension, a PNA probes specific for G. vaginalis (Gard162) and for A. vaginae (AtoITM1) were added to each well of epoxy-coated microscope glass slides (Thermo Fisher Scientific). An additional staining step was done at the end of the hybridization procedure, covering each glass slide with DAPI (2.5 µg/mL). Microscopic visualization was performed using filters capable of detecting the PNA Gard162 probe (BP 530-550, FT 570, LP 591 sensitive to the Alexa Fluor 594 molecule attached to the Gard162 probe), the PNA AtoITM1 probe (BP 470-490, FT500, LP 516 sensitive to the Alexa Fluor 488 molecule attached to the AtoITM1 probe), and DAPI (BP 365-370, FT 400, LP 42). Twenty fields were randomly acquired in each sample. The number of bacteria was counted using ImageJ Software [59]. To reduce any possible overestimation due to the use of DAPI as the probe efficiency was not 100%, we then applied the equations from Table 2 to obtain more accurate relative values. These assays were repeated three times on separate days.

Confocal Laser Scanning Microscopy Analysis of Biofilm Bacterial Distribution
To analyze the bacterial distribution of dual-and triple-species biofilms, the biofilm structure was evaluated by confocal laser scanning microscopy (CLSM) using the PNA Gard162 and AtoITM1 probes coupled to DAPI staining, as described above. For this experiment, biofilms were formed on an 8-well chamber slide (Thermo Fisher Scientific™ Nunc™ Lab-Tek™, Bohemia, NY, USA) at 37 • C under anaerobic conditions for 48 h with the replacement of NYC III medium at 24 h of growth and the addition of the respective BV-associated bacteria. The CLSM images were acquired in an Olympus™ FluoView FV1000 (Olympus, Tokyo, Japan) confocal scanning laser microscope, using a 40× objective. Microscopic visualization was performed using lasers capable of detecting the PNA Gard162 probe (Laser 559, excitation wavelength 559 nm, emission wavelength 618 nm, BA 575-675, sensitive to the Alexa Fluor 594 molecule attached to the Gard162 probe), the PNA AtoITM1 probe (Laser 488, excitation wavelength 488 nm, emission wavelength 520 nm, BA 505-540, sensitive to the Alexa Fluor 488 molecule attached to the AtoITM1 probe), and DAPI (Laser 405, excitation wavelength 405 nm, emission wavelength 461 nm, BA 430-470). Images were acquired with 640 × 640 resolutions of each surface analyzed. All assays were repeated three independent times with two technical replicates.

G. vaginalis Gene Expression Quantification in Mono-, Dual-, and Triple-Species Biofilms
Gene expression of six potential Gardnerella virulence genes, specifically vaginolysin (vly), sialidase (sld), glycosyltransferase, type II (HMPREF0424_0821), multidrug ABC transporter (HMPREF0424_1122), bacitracin transport, ATP-binding protein BcrA (HM-PREF0424_0156), and a transcript that encodes a Rib-protein (HMPREF0424_1196) was determined in 48-h mono-, dual-, and triple-species biofilms. Total RNA was extracted using an E.Z.N.A. ® Bacterial RNA Kit (Omega Bio-tek, Norcross, GA, USA) with minor changes, as previously optimized [81]. Next, genomic DNA was degraded with one step of DNase treatment (Fermentas, Vilnius, Lithuania) following the manufacturer's instructions. RNA concentration, purity, and integrity were determined, as previously described [82]. The same amount of total RNA (300 ng/µL) was reverse transcribed using the RevertAid™ First Strand cDNA synthesis kit (Fermentas), as previously optimized [83], and gene-specific reverse transcription primers as a priming strategy. Quantitative PCR (qPCR) was prepared by mixing 5 µL of iQ SYBR green supermix (Bio-Rad, Hercules, CA, USA), 2 µL of 1:100 diluted cDNA, 0.5 µL of 5 µM forward and reverse primers (Table S1), and water up to 10 µL. The run was performed in a CFX96 TM thermal cycler (Bio-Rad) with the following cycling parameters: 3 min at 95 • C, followed by 45 cycles of 10 s at 95 • C, 10 s at 60 • C, and 15 s at 72 • C. Reaction efficiency was determined by the dilution method [84]. It is of note that at 60 • C, all sets of primers used (Table S1) had similar efficiencies. In addition, the analysis of the melting curves confirmed the presence of a single peak, providing evidence for the specificity of the tested primers. Normalized gene expression was determined by using the delta C t method (E ∆Ct ), a variation of the Livak method, where ∆C t = C t (reference gene) − C t (target gene) and E stands for the reaction efficiency experimentally determined. A non-reverse transcriptase control was included in each reaction. All assays were repeated at least three independent times with three technical replicates.

Statistical Analysis
The data were analyzed using the statistical package GraphPad Prism version 6 (La Jolla, CA, USA) by paired t-test, two-way ANOVA (Sidak's multiple comparison test) and Mann-Whitney U test for the data that did not follow a normal distribution according to the Kolmogorov-Smirnov test. Values with p < 0.05 were considered statistically significant.
Supplementary Materials: The following are available online at https://www.mdpi.com/2076-0 817/10/2/247/s1, Figure S1: An example of fluorescence microscopy pictures representing the specificity of Gard162 probe, Figure S2: An example of fluorescence microscopy pictures representing the specificity of AtoITM1 probe, Figure S3: An example of orthogonal views of BV-associated biofilms by confocal laser scanning microscopy (CLSM). Table S1: Primers used in qPCR experiments.  The funders had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.

Conflicts of Interest:
Christina A. Muzny, MD, MSPH has received research grant support from Lupin Pharmaceuticals, is a consultant for Lupin Pharmaceuticals and BioFire Diagnostics, and has received honoraria from Elsevier, Abbott Molecular, Cepheid, Becton Dickinson, Roche Diagnostics, and Lupin. She is currently on the scientific advisory board for PhagoMed. The other authors declare no conflict of interest.