Direct interactions with commensal streptococci modify intercellular communication behaviors of Streptococcus mutans

Inhibition of cell signaling by commensal streptococci

To study how S. mutans ComRS signaling could be impacted by the presence of a competing species, we empirically optimized a dual-species model system (Fig. 1a) in which a strain of S. mutans carrying the promoter regions of comS or comX (PcomS, PcomX) fused to a codon-optimized green fluorescence protein (gfp) reporter gene could be cocultured with wild-type strains of Streptococcus gordonii DL1, Streptococcus sanguinis SK150, or S. sp. A12. All experiments were performed in chemically defined medium (CDM) [38, 39] because activation of the ComRS circuit occurs spontaneously in CDM as cell density increases, with no need for addition of synthetic XIP or overexpression of the gene for the XIP precursor (comS) (Supplementary Fig. 1). CDM is also heavily buffered with phosphate, which is advantageous because ComRS signaling is optimal at neutral pH values [40, 41]. The buffer also prevents the generation of strongly acidic conditions by S. mutans, which is detrimental to the comparatively acid-sensitive commensal Streptococcus spp.

When the PcomX::gfp reporter strain was cocultured with wild-type S. mutans UA159 (control), robust ComRS signaling was observed as cell density increased (Fig. 1b). However, when cocultured with a competitor Streptococcus spp., no signal from the S. mutans reporter could be detected above background levels; i.e., the nonspecific fluorescence generated by an S. mutans strain that did not contain a copy of the gfp gene. The lack of fluorescence in the cocultures with commensals was not due to growth inhibition of S. mutans as the reporter strain constituted 10 ± 3%, 37 ± 5%, or 54 ± 3% of the total colony forming units (CFUs) recovered after 18 h of coculturing with S. gordonii DL1, S. sanguinis SK150, or S. sp. A12, respectively (Fig. 1c). The quantity of S. mutans cells in the commensal cocultures compared favorably with the recovery of the reporter strain (54 ± 5%) in coculture with wild-type S. mutans UA159. Of note, the fact that equal proportions of reporter and wild-type S. mutans were recovered from cocultures demonstrated that the presence of the GFP gene fusion did not compromise the fitness of the reporter strain, further verified by growth rate comparisons between wild-type and reporter strains (Supplementary Fig. 1).

Two strategies were implemented to try to recover active ComRS signaling by the reporter strain during cocultivation with commensal streptococci. First, synthetic XIP was added to the cocultures to a final concentration of 5 µM just prior to the beginning of the fluorescence monitoring phase of the experiments, and cocultures were observed as above. No detectable fluorescence signal was recorded above background in the cocultures, with or without exogenously added XIP (Fig. 1d). Second, a plasmid encoding a copy of the XIP precursor comS under the control of a highly expressed constitutive promoter (P23) [42] was introduced into the S. mutans reporter strain; we previously reported that overexpression of comS could strongly activate PcomX [28]. However, no increase in GFP expression was observed in cocultures of the comS overexpressing strain with the commensals, whereas signaling was greatly enhanced when cocultured with strain UA159 as a control (Fig. 1e).

To ensure these observations were not limited to only planktonic growth conditions, we examined S. mutans ComRS signaling in cocultured biofilm populations. While almost all cells harboring the PcomX::gfp reporter were GFP-positive in the control biofilms (coculture of the reporter with wild-type S. mutans), confocal imaging of biofilms containing competitor streptococci uniformly showed that almost no S. mutans cells were expressing detectable GFP (Fig. 2a). However, in some frames (0.22 × 0.22 mm frames, ~30,000 S. mutans cells per frame), a small number of cells (1–3 cells per frame) were GFP-positive. When 3D renderings of these areas within the biofilm were constructed, GFP-positive cells were found close to the substratum (Fig. 2b and Movie S1, same area of biofilm as top panel of Fig. 2b). Also, PcomX-active cells were not necessarily confined to distinct S. mutans microcolonies, and in some cases could be seen adjacent to the competitor streptococci, which carried a constitutively expressed red fluorescent protein (DsRed2) for their identification. To quantify the different types of cells in the biofilm populations, we physically dispersed the biofilms by sonication and analyzed the populations by flow cytometry (Supplementary Fig. 2). About 1 in 10,000 S. mutans cells counted displayed activation of PcomX within the biofilms, which was similar to the proportions of GFP-expressing cells in planktonic growth conditions (Supplemental Table 1).

Commensal signaling inhibition is dependent on cell contact

Changes in phenotypes that are observed when two different species of bacteria are cocultured can usually be induced by secreted molecules from one of the bacterial strains [1]. We suspected that molecule(s) secreted by the competitor strains are required for shutting down cell–cell signaling in S. mutans. To explore this hypothesis, we cultured the competitors individually overnight and collected the supernatant fluids after centrifugation. The supernates were then filter sterilized, pH adjusted from ~6.3 to 7.0 with NaOH, and carbohydrate was added back to achieve a final concentration of added glucose to 20 mM. We then inoculated our reporter strain into the commensal supernates and monitored fluorescence activity (Fig. 3a). Surprisingly, ComRS signaling was readily observed in all supernates. In fact, reporter activity tended to be higher in the supernates of competitors compared to controls.

In another experiment to confirm these results, we grew competitor and our reporter strains together in a transwell apparatus, so that both bacterial strains shared the same growth medium, but were physically separated by 0.4 µm pore size polycarbonate membrane that would allow passage of small molecule(s) between the chambers (Fig. 3b). Even in the transwell system, cell signaling was robust in cocultures containing competitor species. This result is consistent with data showing that the proximity of live commensal cells with S. mutans prior to signal activation is required for the signaling inhibition.

Impairment of S. mutans cell signaling by oral commensals is conserved across species

We next screened a collection of low-passage oral streptococci that had been previously genome sequenced [43] to determine whether the ability to inhibit S. mutans ComRS signaling was conserved across commensal species and to assess whether the presence or absence of certain genes might contribute to inhibition of peptide signaling. Ten different low-passage clinical isolates of S. gordonii, ten isolates of S. sanguinis, and five isolates of S. sp. A12-related organisms [19] were cocultured with our S. mutans ComRS signaling reporter. The S. sp. A12-related organisms included strains classified as A12-like (A13 and BCC21), as Streptococcus australis (G1 and G2), or as Streptococcus parasanguinis (A1). Interestingly, significant production of GFP by S. mutans was evident when cultured with one isolate of S. sanguinis (BCC64) and with three isolates that were classified as A12-related (BCC21, G1 and G2) (Fig. 4a). However, these results were most likely due to the inability of these isolates to grow well within the CDM medium during the course of the experiment (Supplementary Fig. 3). In fact, after 18 h of monitoring, these isolates comprised <0.01% of the total CFUs recovered. Conversely, all commensal strains that grew well in CDM (achieved an OD₆₀₀ > 0.1 after 12 h as monitored using a Bioscreen system, see Supplementary Fig. 3) inhibited PcomX activation. Thus, if a commensal strain could grow in CDM, even somewhat poorly, it could completely inhibit ComRS signaling.

We also tested several genomically and phenotypically diverse isolates of S. mutans [44, 45], both in coculture with our PcomX::gfp reporter in the UA159 background (Supplementary Fig. 4) and against competitor Streptococcus spp., after transformation of the S. mutans strains with the PcomX reporter plasmid (Fig. 4b). Various levels of spontaneous activation of the PcomX::gfp reporter were observed among the different S. mutans strains in monocultures in CDM, consistent with recent reports showing strain-dependent differences in S. mutans peptide signaling [46]. One isolate, Smu107 (R221), had undetectable levels of GFP in monoculture in CDM alone. All others showed activity above baseline. However, when cocultured with S. sp. A12, ComRS signaling was inhibited to an extent similar to that observed with strain wild-type UA159. Therefore, the ability to obstruct ComRS signaling is conserved among isolates of S. gordonii, S. sanguinis, and A12-related streptococci, and inhibition by commensals is similarly conserved in genomically diverse isolates of S. mutans.

Relatively small proportions of live commensal streptococci can inhibit signaling

To verify that the ability of the competitor species to grow (viability) was required for inhibition of peptide signaling, we used two different treatments of the competitor species S. sp. A12 after it was grown to mid-exponential phase: 80 °C for 0.5 h in a heating block (Fig. 5a) or treatment with 4% paraformaldehyde for 1 h at ambient temperature (Fig. 5b). After treatment, the inactivated commensal cells were washed and resuspended in fresh CDM and then mixed with the S. mutans reporter strain to begin the experiment. With heat-treated cells, some ComRS signal activity was evident, but not near the levels seen with S. mutans-only controls. However, when the paraformaldehyde-treated cells were used, the competitor did not inhibit signaling and fluorescence, with levels being similar to the S. mutans-only control. Importantly, we determined that there was a greater number of live cells, by plating and counting CFUs, for the competitor after heat treatment, compared to paraformaldehyde fixing (Supplementary Fig. 5), which likely explains the difference in effects on PcomX activation. These results support that metabolically active and growing competitors are required for S. mutans ComRS signaling obstruction.

Based on the intermediate inhibitory effects seen with reduced proportions of a live competitor on our reporter strain, i.e. with heat-treated cells, we tested whether some minimal proportion of live competitor was required to exert effects on ComRS signaling. We utilized S. sp. A12 and varied the percentage of S. mutans and S. sp. A12 in the cocultures, after determining that the proportions of cells recovered after 18 h were similar to the proportions in the initial inocula (Supplementary Fig. 6). Complete inhibition of S. mutans ComRS signaling occurred when S. sp. A12 constituted ≥6.3% of the initial inoculum (Fig. 5c). At 3.1 or 1.6% of S. sp. A12, reporter activity was detectable, but at lower levels than when no S. sp. A12 was present. No difference in S. mutans reporter activity was observed when <1% of the inoculum was S. sp. A12.

Finally, we determined if timing of the introduction of the competitor commensal to the coculture affected peptide signaling behavior. In this experiment, we inoculated the reporter strain at three different dilutions (1:50, 1:66, and 1:100) to allow ComRS signaling to initiate at different time points during the incubation period, since activation requires that S. mutans attain a threshold cell density. Simultaneously, we began growing a culture of the competitor (for this experiment S. gordonii DL1) and control (S. mutans UA159) separate from our reporter strain. Fluorescence of the PcomX::gfp reporter was actively monitored, and at a time point (4.5 h) when fluorescence was fully detected (Fig. 5d), beginning to be detected (Fig. 5e) or had not yet been detected (Fig. 5f) depending on the dilution used, the competitor was added and fluorescence activity and optical density were monitored (see Supplementary Fig. 7A). Interestingly, in the case where cell–cell signaling had already been activated (Fig. 5d), or was beginning to be detected (initial stages of ComRS activation) (Fig. 5e), addition of competitor did little to dampen reporter activity. However, when competitor was added at the time point when there was no evidence yet of activation (Fig. 5f), the presence of competitor significantly impaired com gene activation, as seen at the 12 h time point (Supplementary Fig. 7B).

Transcriptome profiling of dual-species interactions

To determine if the proximity-dependent effects on S. mutans of encountering oral competitor streptococci was confined to ComRS signaling and genetic competence activation, we performed transcriptome profiling by RNA-Seq of S. mutans grown in CDM under three different conditions: (1) growth in its own (S. mutans) spent supernatant fluid, (2) growth in spent supernates of a competitor (S. sp. A12; treated similarly to Fig. 3a), or (3) cocultured in fresh CDM medium directly with competitor (S. sp. A12). When comparing the growth of S. mutans in its own spent supernates against competitor spent supernates, we found 88 S. mutans genes differentially expressed (Log₂ fold change ≥ (−)1.5, −log10 p value ≥ 4), which included upregulation of the zinc transport system and several amino acid ABC transporters, along with downregulation of the TnSmu1 genomic island (Fig. 6a and Supplementary Table 2). A more substantial effect was seen when we analyzed the transcriptomes of S. mutans grown in direct cocultivation with S. sp. A12 compare to S. sp. A12 supernates alone (Fig. 6b and Supplementary Table 3). In this case, 140 genes were differentially expressed in S. mutans, including upregulation of one of the CRISPR gene clusters and, as would be expected, downregulation of the entire genetic competence regulon in cells grown directly with S. sp. A12. Principal component analysis of transcriptome data from these three conditions displayed a wide separation of the tested groups, demonstrating that there is a unique transcriptomic response by S. mutans when it is grown directly with a competitor, as opposed to cultivation in spent supernates of the  competitor (Supplementary Fig. 8).

To determine if these transcriptomic responses were specific to competition with S. sp. A12 or represented a generalized response to  commensal streptococci, as well as to confirm our RNA-Seq data set, we performed qRT-PCR on harvested RNA from cocultures of S. mutans grown with either S. sp. A12, S. gordonii DL1, or S. sanguinis SK150, or from S. mutans grown in monoculture. Two unique core genes of S. mutans [44] that were not differentially expressed in our RNA-Seq experiment, SMU.996 and Smu.1616c, were used to normalize the amount of S. mutans present at the time of harvest between all cocultured samples (Supplementary Table 4). In total, eight genes were probed that represented both upregulated (Fig. 6c) and downregulated (Fig. 6d) genes found during our RNA-Seq experiment. Remarkably, S. mutans displayed the same genetic response in all three cocultures with different competitor species, including upregulation of cas3 and the gene for a secreted glucosyltransferase (gtfC) required for sucrose-dependent biofilm formation [47, 48]. Aside from a significant decrease in comX expression, bacteriocin gene expression was also impacted in all three cocultures, including a decrease in comE (bacteriocin-related response regulator) and cipB (ComE-regulated bacteriocin). We propose that these selected probed genes in part represent potential larger transcriptomic changes of a conserved GEP by S. mutans to the presence of competitors that has not been previously reported.

Inhibition of cell signaling is a novel antagonism mechanism by oral streptococci

Oral commensal streptococci, such as S. gordonii and S. sanguinis, antagonize S. mutans through different mechanisms, including hydrogen peroxide production [15, 16] and secretion of proteases that degrade signaling molecules [18, 19, 49]. To determine if these known antagonistic pathways were responsible for contact-dependent inhibition of com gene activation, we cocultured our reporter strain with S. gordonii (Fig. 7a) or with S. sp. A12 (Fig. 7b) that carried deletions in known genes involved in antagonism. Specifically, for hydrogen peroxide production, we tested competitors that carried a deletion of spxB, encoding pyruvate oxidase [16, 18]. No recovery of S. mutans ComRS signaling was observed when cocultured with the spxB mutants, suggesting hydrogen peroxide production was not required for obstruction of signaling. To further rule out the effects of oxygen metabolism as a potential mechanism for inhibition, we grew the cocultures anaerobically, yet saw no changes in ComRS signaling inhibition (Fig. 7c). In fact, reporter actively was significantly lower in anaerobic conditions, compared to the controls. Two different peptidases of S. sp. A12 degrade S. mutans signaling molecules. Sgc (an apparent orthologue of S. gordonii Challisin (49)) has CSP degrading activity [18], whereas PcfO of A12, encoded by an apparent orthologue of pepO in multiple Streptococcus spp., can degrade S. mutans XIP [19]. However, mutation of either gene had no significant effect on ComRS signaling interference in our PcomX reporter assay. We additionally analyzed S. mutans gene expression via qRT-PCR with six differently expressed genes, as seen in Fig. 6, grown in coculture with S. sp. A12 and its mutant derivatives (Supplementary Fig. 9). We saw no significant changes in the upregulated genes among the cocultures with the S. sp. A12 mutants, but did see significant differences in comX gene expression (~1.9-fold increase) with A12∆pcfO and a ~16-fold increase in comE expression in the S. mutans coculture with A12∆sgc, suggesting both proteases do contribute to the dampening of S. mutans peptide signaling, albeit not with any apparent measurable effect on PcomX induction based on the reporter assays. Finally, we also proposed that the competitors could be internalizing S. mutans XIP as a potential nutritional source in the peptide-free CDM medium. However, we saw no change when the opp homolog of S. gordonii (old NCBI locus tag SGO_1712) or of S. sp. A12 (ATM98_08725) was deleted. Collectively, then, we conclude that there are novel and mechanistically uncharacterized strategies used by genetically diverse commensal streptococci to impair S. mutans cell–cell signaling that also lead to a unique transcriptome response that is distinct from monocultures or when S. mutans is exposed to supernates of commensal streptococci.

Source: Ecology - nature.com