GBS promoted the formation of GV biofilm
In our in vitro conditions, we observed that G. vaginalis and GBS were able to form mono-species biofilms, and no differences were noted in the total biomass between these two 48-hours biofilms. However, the addition of GBS to G. vaginalis led to a significant enhancement of biomass in the 48-hours dual-species biofilms (Figure 1A). SEM was employed to examine the structure and interactions in static biofilms formed by G. vaginalis and GBS. As illustrated in Figure 1B, G. vaginalis biofilms were arranged in clusters, while those of GBS were diffuse in the mono-species biofilms. A closer view, from the side, of the 48-hours dual-species biofilms revealed G. vaginalis remained arranged in clumps, with GBS tending to be arranged around G. vaginalis. It is noteworthy that a substantial amount of amorphous matrix, likely EPS (indicated by arrows), was visible on its surface. To confirm the results obtained with the CV assay and visualize the spatial distribution and different architectures of the tested dual-species biofilms, we analyzed various z-stacks under diverse treatment conditions using FISH/CLSM. CLSM generally confirmed that the biofilm of GBS was thicker than GV, albeit without statistical significance (Figure 1C). Notably, as shown in Figure 1C, the presence of GBS promoted biomass in dual-species biofilms compared to G. vaginalis mono-species biofilms, consistent with the CV results. Throughout the co-culture period, the biomass of both GBS and GV in the dual-species biofilm significantly increased compared with mono-species biofilms. Consequently, we can conclude that GBS and GV may act as bidirectional promoters.
Transcriptomic analysis
Nine samples were sequenced using the Illumina HiSeqTM 2500, following the removal of adapters and low-quality reads. A total of 7.3 G and 7.1G of valid data was obtained in the M vs. GBS and M vs. GV groups, respectively, with the sequencing error rate not exceeding 0.03%. In both analyses, Q20 exceeded 90%, and Q30 surpassed 85%, indicating that the samples were free from contamination and that the RNA-Seq quality was good (Table S1, S2). Tables S3 and S4 display the comparison between the sample and the reference genome, illustrating that all the multiple mapping rates were less than 10%. The transcriptomic data exhibited minor variations among replicates at intensity values (Figure S1A, D), with the R2 between the samples consistently exceeding 92% (Figure S1B, E), demonstrating a high reproducibility of transcriptome profiles. Furthermore, in the principal component analysis (PCA) analysis, the data from mono-species and dual-species biofilms were distinguishable (Figure S1C, F).
Biofilm formation is a complex, dynamic process, typically involving surface attachment, biofilm maturation, and biofilm dispersion 41. Subsequent analysis of the transcriptomes uncovered a substantial number of genes exhibiting differential expression in mono- and dual-species biofilms. In comparison with GBS mono-biofilms, approximately 10.48% of the S. agalactiae genome (246 genes) displayed significant differences (P < 0.05 & FoldChange≥1.5) in dual-species biofilms, with 99 transcripts being upregulated and 147 downregulated (Figure 2A). Moreover, DEGs were subjected to row and column clustering (q value ≤ 0.05), depicting the gene expression level and the similarity of expression patterns (Figure S2A). Similarly, when compared with GV biofilms, a total of 437 DEGs were identified in dual-species biofilms, meeting the criteria of P value ≤ 0.05 and FoldChange ≥ 1.5. Among these, 206 genes were significantly up-regulated, while 231 genes were significantly down-regulated in their expression.
To explore the function of these nonredundant DEGs during dual-species biofilm formation, we conducted a functional analysis using GO and KEGG pathway terms. The 246 DEGs in GBS were allocated to 257 GO terms, encompassing 152 biological processes (BP), 19 cellular components (CC), and 86 molecular functions (MF). The most prevalent groups identified in CC were associated with “membrane”. The top three BP categories observed were linked to “phosphoenolpyruvate-dependent sugar phosphotransferase system (PTS)”, “carbohydrate transport”, and “organic substance transport”. Furthermore, the predominantly enriched MF terms indicated that the transcripts are mainly affiliated with transmembrane transporter activity (especially carbohydrate, active, and organic anion transmembrane transporter activity) and protein-N(PI)-phosphohistidine-sugar phosphotransferase activity (Figure 3A). Subsequently, the functional categorization of these 246 genes by their assigned KEGG classification is depicted in Figure 3B. Parallel to the GO enrichment, the most abundant groups identified in the KEGG pathway were “PTS” and “Starch and sucrose metabolism”. Additionally, numerous annotated transcripts were clustered within groups related to signal communication, such as “Quorum sensing” and “Two-component system”. The 437 DEGs in GV were associated with 279 GO terms, including 172 BPs, 18 CCs, and 86 MFs. In contrast to GBS, the GO enrichment analysis of GV primarily focused on ribosome-related BPs, biosynthetic processes, and signal transduction (Figure 3C). The most notable GO terms included “structural constituent of ribosome”, “structural molecule activity”, and “ribosome”. Similarly, KEGG analysis indicated that DEGs were significantly enriched for those involved in “metabolic pathways”, “ribosome”, and “Biosynthesis of secondary metabolites” (Figure 3D). Notably, GV also exhibited enrichment in both “Quorum sensing” and “Two-component system”. This enrichment analysis suggests substantial information exchange or mutual utilization of metabolites between GBS and GV during the dual-species biofilm reaction.
Proteomic analysis
The impact of co-culture on the proteome of GBS and GV was assessed using a liquid chromatography-mass spectrometry (LC-MS) approach. In the comparison of dual-species biofilms with GBS biofilms, a total of 5420 peptides and 919 proteins were identified. Similar to the findings in the transcriptomic data, the proteomes from the dual-species biofilm samples formed distinct clusters separate from the GBS biofilm proteomes (Figure S3A). Moreover, the CV in Figure S3B indicates high sample reproducibility. Likewise, a total of 9103 peptides and 1042 proteins were identified in dual-species biofilms compared with GV biofilms. The PCA analysis demonstrated significant differences in protein abundance between the two biofilms (Figure S3C). Additionally, Figure S3D illustrates good repeatability in this comparison. As depicted in Figure S4A, a total of 672 proteins were detected in both the M group and GBS group, while 22 proteins were exclusively identified in GBS mono-species biofilms and 10 in dual-species biofilms. Similarly, a total of 914 proteins were detected in both the M group and GV group, with only 1 protein being detected in GV biofilm and 3 in dual-species biofilms.
Label-free quantification was utilized to assess the relative abundance of the proteome in mono- and dual-species biofilms. A 1.5-fold threshold and t-test (p < 0.05) were employed to determine valid protein changes. Among the proteins identified from dual-species biofilms and quantified, 134 were up-regulated and 89 were down-regulated in GBS (Figure 2C). In comparison to GV biofilms, 61 were down-regulated and 54 were up-regulated in dual-species biofilms (Figure 2D). Furthermore, the hierarchical clustering of significant proteins illustrated the arrangement of biological replicates under different biofilm conditions, indicating that GBS and G. vaginalis exhibited distinct proteomic patterns influenced by co-culture (Figure S2C and S2D). Regarding the subcellular localization of differentially expressed proteins (DEPs), 41, 20, and 4 proteins in the comparison of GBS and M groups were predicted as cytoplasmic proteins, cell membrane proteins, and extracellular proteins, respectively (Figure S4B). Similarly, in the comparison of GV and M groups, 15, 6, 4, and 1 proteins were predicted as cell membrane proteins, cytoplasmic proteins, extracellular proteins, and cell wall proteins, respectively (Figure S4C). These predictions revealed that GBS DEPs were mainly located in the cytoplasm, while GV DEPs were primarily situated in the cell membrane.
According to the GO analysis using Blast2GO, the differentially expressed proteins were categorized into three cellular functions classified as BP (103 pathways), MF (85 pathways), and CC (3 pathways), which were confirmed by Fisher’s exact test. Specifically, 86 proteins were predicted to modulate cellular processes, and 76 proteins were implicated in regulating metabolic processes. In terms of MF, 20 proteins possessed a structural constituent of the ribosome, and 3 proteins exhibited glycerone kinase activity. Unlike the transcriptome, the most prevalent group identified in CC was the ribosome (Figure 4A). Furthermore, enrichment analysis using the KEGG database revealed 46 pathways in which GBS differentially expressed proteins are involved in dual-species biofilms. The majority of pathways were related to metabolism or degradation, encompassing Glycerolipid, Tyrosine, Thiamine metabolism, Phenylalanine metabolism, and fatty acid, Chloroalkane and chloroalkene, Lysine, and Styrene degradation (Figure 4B). The enrichment analysis indicated that GBS under two growth conditions triggered various BPs related to phenotypes.
We also conducted statistical analysis on the distribution of quantified proteins in the comparison of GV and M groups using GO secondary annotation classification, encompassing three major classes: BP (34), CC (10), and MF (34). The significant GO terms for BP included “macromolecule catabolic process” “cellular catabolic process” and “cellular macromolecule catabolic process” (Figure 4C). Among the up-regulated proteins, the top three GO terms for CC comprised “integral component of membrane” “beta-galactosidase complex” and “membrane”. Based on KEGG pathway enrichment analysis, “mismatch repair” “nucleotide excision repair” and “ribosome enrichment” were most pronounced in G. vaginalis during co-culture (Figure 4D). In this process, the most upregulated proteins were involved in the “Citrate cycle” “Sphingolipid metabolism” and “Galactose metabolism”, while the major downregulated proteins were associated with the “Ribosome” “Nucleotide excision repair” and “Acarbose and validamycin biosynthesis”. The functional diversity of the DEPs (not limited to the previous enumeration) was related to the complex regulatory network that controls biofilm formation.
Correlation Analysis
The Pearson correlation coefficients between gene expression and protein abundance for GBS and GV were 0.034 and 0.064, respectively (Figure S5). This result indicated that, under the given conditions (mono or dual-species biofilms), there is essentially no correlation between genes and proteins that are down-regulated or up-regulated in the same direction between the transcriptome and proteome datasets. Furthermore, we identified a number of genes related to GBS and GV biofilm formation, epithelial cell adhesion, and virulence factors from the sequencing results and compared the trends of these genes in the transcriptome and proteome analyses (Table 1 and Table 2).
AI-2 increased significantly in dual-species biofilms
To validate the transcriptomic and proteomic results, GBS selected 17 virulence factors, while GV selected 16, which may be associated with biofilm formation and epithelial cell adhesion for qPCR. The gene IDs and primers of these proteins are shown in Tables S5 and S6. The qPCR results showed that the expression of most genes related to biofilm formation and cell adhesion was not significantly elevated in GBS and GV. Additionally, the expression trends of the proteome and transcriptome in both groups were generally consistent with the qPCR results. Notably, our results indicated that in dual-species biofilms, the expression levels of luxS were greatly up-regulated (Figure 5A).
Next, we constructed the luxS mutant strain (ΔGBS) of GBS. As depicted in Figure S6A, the upstream (Line 1) and downstream (Line 2) regions of the luxS gene, as well as erm (Line 3) and pSET4s (Line 4), were amplified by PCR. Similarly, the recombinant plasmid (plasmid pSET4s-luxS) was verified using the primers M13F (5’-TGTAAAACGACGGCCAGT-3’) and M13R (5’-CAGGAAACAGCTATGACC-3’). Lines 1, 2, 4, and 5 (2483bp) in Figure S6B represent the recombinant plasmids in this study. Finally, the validation of luxS mutants is shown in Figure S6C, where 1 and 4 (483bp) correspond to wild strains and 2, 3, and 5 represent mutant strains.
AI-2 activities in cell-free supernatants from batch cultures of GBS wild type, ΔGBS, GV, GV+GBS (M), and (GV+ΔGBS) ΔM were measured using the V. harveyi AI-2 bioluminescence induction assay, which provides an indication of AI-2 concentration. AI-2 activity was detected in the cell-free supernatant of wild-type GBS, GV, ΔM, and M groups after 48 hours of biofilm formation (Figure 6A). It is also notable that the fluorescence intensity of all groups decreased in the first 3 hours. The trend of decreasing fluorescence intensity of the M group and the positive control remained consistent, with the M group exhibiting significantly higher fluorescence intensity than the other groups (refer to Figure 6B). At the point of lowest fluorescence intensity for the negative control (3.5 hours), the fluorescence intensity of GBS, GV, and ΔM was greater than the negative control but less than the positive control, indicating that the samples being tested contained AI-2, albeit at a concentration lower than that of the positive control. Simultaneously, we observed that no AI-2 was detected in the ΔGBS group, while the concentrations of AI-2 in the GV and ΔM groups were comparable, suggesting that GV did not produce AI-2 in the co-culture model (Figure 6B).
LuxS/AI-2 of GBS promotes GV biofilm formation
To evaluate the viability of the luxS gene in biofilm development, a CV assay was employed to assess the thickness of biofilms under different culture conditions. As depicted in Figure 7A, the biomass of the ΔGBS biofilm decreased significantly following the knockout of the luxS gene (P < 0.05).
Concurrently, we observed that wild-type GBS uniformly covered most of the surface with biofilm (refer to Figure 1 c2). In contrast, the biofilm formed by the luxS mutant GBS, as examined by a three-dimensional confocal stack, appeared to cover a smaller surface area (Figure 7c1). The same phenomenon can be seen in Figure 7 d1-d3, where the bacterial arrangement in the ΔGBS biofilm was sparse and loose. To provide evidence for the involvement of AI-2 in biofilm formation, various concentrations ranging from 0.76nM to 7.6µM of exogenous AI-2 molecules were used to complement the luxS mutant. As illustrated in Figure 7B, the biofilm biomass of the luxS mutant reached the same level as that of wild-type GBS when the exogenous AI-2 concentration reached 76nM. Figures 7c2 and c3 depict ΔGBS with the addition of 76nM and 0.76µM AI-2, respectively. As the concentration of AI-2 increased (0.76µM), ΔGBS aligned more tightly and the biofilm thickness increased significantly, accompanied by the production of a small amount of EPS (Figure 7 d4-d6).
In the dual-species biofilm culture system, the biofilm biomass of the ΔGBS co-cultured with GV also decreased significantly compared to that of the M group (Figure 8B). This observation is further corroborated by the CLSM images in Figure 8A, where both GBS and GV in ΔM exhibited much thinner biofilms than those in the M group (Figure S7). Similarly, when the concentration of exogenous AI-2 reached 0.76µM, the total biofilm biomass of ΔM reached the same level as that of the M group (Figure 8B). Furthermore, the biofilm formed by adding 7.6µM AI-2 to ΔM was significantly thicker than the 48-hours biofilm of the M group (Figure 8 a4-6). The same result can be observed in SEM (Figure 8C). These results indicate that AI-2 not only promoted the growth of GBS biofilm but also facilitated GV biofilm development in the co-culture model. To further verify the effect of AI-2 on GV, exogenous AI-2 was added to GV and GV+ΔGBS biofilms, and genes in GV that might be related to biofilm formation were verified by RT-qPCR. We can see from Figure 8D that two genes, murG and GAVG_RS05105, were significantly increased after the addition of AI-2. Therefore, we speculate that exogenous AI-2 could lead to elevated murG and GAVG_RS05105 expression in GV, thus affecting GV biofilm formation and leading to recurrence of BV.