The Antibacterial Effect of Cannabigerol toward Streptococcus mutans Is Influenced by the Autoinducers 21-CSP and AI-2

Bacteria can communicate through an intercellular signaling system referred to as quorum sensing (QS). The QS system involves the production of autoinducers that interact with their respective receptors, leading to the induction of specific signal transduction pathways. The QS systems of the oral cariogenic Streptococcus mutans regulate the maturation of biofilms and affect its virulent properties. We have previously shown that the non-psychoactive compound cannabigerol (CBG) of the Cannabis sativa L. plant has anti-bacterial and anti-biofilm activities towards S. mutans. Here we were interested in investigating the effect of the two QS systems ComCDE and LuxS on the susceptibility of S. mutans to CBG and the anti-QS activities of CBG. This was assessed by using various comCDE and luxS mutant strains and complementation with the respective autoinducers, competence stimulating peptide (CSP) and (S)-4,5-dihydroxy-2,3-pentandione (DPD, pre-AI-2). We found that S. mutans comCDE knockout strains were more sensitive to the anti-bacterial actions of CBG compared to the WT strain. Exogenously added 21-CSP prevented the anti-bacterial actions caused by CBG on the ΔcomC, ΔcomE and ΔluxS mutants, while having no effect on the susceptibility of the WT and ΔcomCDE strains to CBG. Exogenously added DPD increased the susceptibility of WT and ΔluxS to CBG. Vice versa, CBG significantly reduced the 21-CSP-induced expression of comCDE genes and ComE-regulated genes and suppressed the expression of luxS with concomitant reduction in AI-2 production. DPD induced the expression of comCDE genes and ComE-regulated genes, and this induction was repressed by CBG. 21-CSP alone had no significant effect on luxS gene expression, while ΔcomCDE strains showed reduced AI-2 production. In conclusion, our study shows that the susceptibility of S. mutans to CBG is affected by the ComCDE and LuxS QS pathways, and CBG is a potential anti-QS compound for S. mutans. Additionally, we provide evidence for crosstalk between the ComCDE and LuxS QS systems.


Introduction
Bacteria can communicate through an intercellular signaling mechanism known as quorum sensing (QS) [1]. Several phenotypes in bacteria are controlled by the QS system. These phenotypes range from basic cell mobility to more complex communal interactions including biofilm formation, production of virulence factors, modulation of metabolic activity, production of extracellular polymeric substance (EPS), acquisition of nutrients, competence of genetic material, resistance to antibiotics and the generation of secondary metabolites [2,3].
The QS systems commonly consist of secreted autoinducers which interact with their respective receptors on the bacteria cell surface, leading to the induction of distinct signal transduction pathways when their concentrations exceed the threshold [4]. This the anti-bacterial effect of CBG. Thus, the QS pathways may modulate the susceptibility of S. mutans to CBG. CBG itself prevented the expression of luxS, comCDE and ComEregulated genes, indicating that this compound possesses anti-QS properties. Additionally, we provide evidence that there is a crosstalk between the LuxS and ComCDE systems in S. mutans, where DPD (pre-AI-2) induced the expression of comCDE and ComE-regulated genes, and a lack of comCDE resulted in reduced AI-2 production. The results obtained from this study provide novel insights into the molecular pathways regulating the susceptibility of S. mutans to CBG and adds another layer to the anti-bacterial action mechanism of CBG.

Bacterial Strains, Culture Conditions and End Point Planktonic Growth Assay
A starter culture of S. mutans UA159 (WT), S. mutans ∆luxS (TW26), S. mutans ∆comC, S. mutans ∆comE and S. mutans ∆comCDE were incubated in brain heart infusion broth (BHI, Acumedia, Lansing, Michigan, USA) for 20-24 h at 37 • C in the presence of 5% CO 2 until an OD 600nm of 1.0-1.2 was obtained [32]. The S. mutans ∆luxS (TW26) was generously provided by Professor Robert Burne (University of Florida, Gainesville, FL, USA) [37]. S. mutans ∆comC, S. mutans ∆comE and S. mutans ∆comCDE mutants were kindly provided by Professor Howard Kuramitsu (State University of New York, Albany, NY, USA) [38]. For the planktonic growth assays, the bacterial cultures were adjusted to an OD 600nm of 0.1 in fresh BHI and exposed to various concentrations of CBG (0.75, 1.25, 2.5 and 5 µg/mL) or respective concentrations of ethanol (0.0075-0.05%) and seeded in a volume of 200 µL in a sterile tissue-grade transparent flat-bottom 96-well microplate (Corning, Incorporated, Kennebunk, ME, USA). Untreated bacteria served as an additional control. After a 24 h incubation at 37 • C, the absorbance was measured at 595 nm by a Tecan M200 infinite microplate reader (Tecan Trading AG, Männedorf, Switzerland). The bacterial viability is expressed as percentage in comparison to control sample according to the following formula: % Viability = (OD treated sample )/(OD control sample ) × 100.

Growth Kinetics Studies
The influence of DPD (pre-AI-2)/21-CSP on planktonic growing S. mutans strains was examined by performing kinetics studies. Overnight cultures of S. mutans UA159, S. mutans ∆luxS (TW26), S. mutans ∆comC, S. mutans ∆comE and S. mutans ∆comCDE bacteria were made to an OD 600nm of 0.1 in fresh BHI medium and exposed to varying concentrations of CBG (0.75-5 µg/mL) in the presence or absence of 21-CSP (1 µg/mL) or DPD (5 µM). The bacteria were then cultivated in a transparent flat-bottom 96-well microplate (Corning) at a volume of 200 µL, and the bacterial growth (OD 595nm ) was measured every 30 min for 20 h in a Tecan M200 microplate reader (Tecan Trading AG, Männedorf, Switzerland) at 37 • C [32].

Investigation of AI-2 Production
AI-2 production by S. mutans under different concentrations of CBG was determined by using Vibrio harveyi reporter strain MM77 in a bioluminescence assay, similar to the method of Shemesh et al. [39]. MM77 was kindly provided by Professor Bonnie Bassler (Princeton University, Princeton, NJ, USA) [40]. The wild-type (WT) and mutant S. mutans strains (OD 595nm = 0.1) were allowed to grow in BHI media for 4 h in order to reach the early logarithmic growth phase in the absence or presence of varying concentrations of CBG (0.75-5 µg/mL) and/or 21-CSP (1 µg/mL). The condition medium (CM) was collected after removing the bacteria by centrifugation and was passed through a 0.22 µm pore size sterile PVDF filter (Millex-GV, Merck, Darmstadt, Germany). The CMs were stored at −20 • C until future use. Bioluminescence assay was performed as described previously by Aqawi et al. [35]. Briefly, V. harveyi MM77 (luxLM::Tn5, luxS::CmR; lacking both AI-1 and AI-2) were grown for 20-24 h at 30 • C in AB medium with constant shaking until reaching stationary phase (OD 600nm = 0.7). The MM77 bacteria at a starting OD of 0.035 were incubated at 30 • C in AB medium containing 10% (v/v) S. mutans CM in white optical 96-well flat-bottomed plates (µCLEAR CELLSTAR, Greiner Bio-One, Frickenhausen, Germany). MM77 bacteria incubated in 10% fresh BHI medium served as a negative control and MM77 incubated with 5 µM DPD served as a positive control. Luminescence and absorbance (OD 595nm ) were measured in parallel each 30 min for 48 h using the Tecan M200 infinite microplate reader (Tecan Trading AG, Switzerland). Luminescence values were normalized by OD values of each measurement time point to adjust for any differences in V. harveyi growth rates. The resulting luminescence was then divided on the OD 600nm of the various S. mutans cultures used for collecting the CMs, to correct for different numbers of bacteria. The data are expressed as the ratio of luminescence/OD, and calculations for the area under the curve (AUC) were performed.

RNA Isolation
Overnight cultures of the various S. mutans strains were adjusted to OD 600nm = 0.1 in BHI and incubated in the absence or presence of 1.25 µg/mL CBG and/or 21-CSP (1 µg/mL). After a 4 h incubation at 37 • C, the bacteria were resuspended in 1 mL RNA protect reagent (Qiagen, Hilden, Germany) for 5 min. Following centrifugation (1500× g, 15 min, 4 • C), the pellets were stored at −80 • C. On the day of RNA extraction, the pellets were resuspended in 350 µL RLT lysis buffer (Qiagen, Hilden, Germany), placed into microcentrifuge tubes containing 1 mm glass beads followed by beating 3 times at a speed of 4.5 m/s for 45 s at 5 min intervals using the Bio 101 FastPrep FP120 cell disruption system (Savant Instruments, Inc., Holbrook, NY, USA). The samples were further processed for RNA isolation using the RNeasy MINI kit (Qiagen, Hilden, Germany) according to the manufacturer's instructions including an on-column DNase digestion step [35]. The RNA concentration and purity were analyzed by a Nanodrop (Nanovue, GE Healthcare Life Sciences, Buckinghamshire, UK). Samples that had an OD 260 /OD 280 ratio of 1.8-2 and an OD 260 /OD 230 ratio above 2 were used for cDNA synthesis. The RNA was reverse transcribed into cDNA using the qScript cDNA synthesis kit (QuantaBio, Beverly, MA, USA) [35].

Quantitative Real-Time PCR
Quantitative real-time PCR was performed as described by Aqawi et al. [35]. Each reaction was prepared by mixing 10 ng cDNA with 300 nM of respective forward and reverse primers (Supplementary Table S1) and Power Sybr Green PCR Master mix (Applied Biosystems, Warrington UK). Triplicates were done for each gene of each sample. The amplification was done in a Bio-Rad CFX Connect Real-time system using the Bio-Rad CFX Maestro program with 40 cycles of 15 sec at 95 • C and 1 min at 60 • C. The 16S rRNA was used as a housekeeping gene and the changes in gene expression were calculated using the 2 −∆∆Ct method. The control was set to 1 for each gene, and gene expression is presented as relative values.

Statistical Analysis
Each experiment was conducted in triplicate. The student's t-test was used for statistical analysis of the collected data. A p value of less than 0.05 in comparison to control was considered statistically significant.

Streptococcus Mutans Strains Deficient in the comCDE QS System Showed Increased Sensitivity to CBG
To explore the effect of quorum sensing on the susceptibility of S. mutans to the anti-bacterial activity of CBG, S. mutans UA159 (WT) and different QS-knockout strains (S. mutans ∆luxS, S. mutans ∆comC, S. mutans ∆comE and S. mutans ∆comCDE) were exposed to increasing concentrations of CBG (0.75-5 µg/mL) or the respective ethanol dilutions and the OD 595nm was monitored after a 24 h incubation ( Figure 1). Untreated S. mutans from each strain served as another control. We observed that the different comCDE-knockout strains showed higher susceptibility to CBG than the WT strain, while the ∆luxS strain showed similar susceptibility to CBG as the WT (Figure 1), suggesting that the ComCDE signaling pathway may promote bacterial survival. The MIC of CBG on the WT and ∆luxS strains was found to be 2.5 µg/mL, while that of the ∆comC and ∆comE strains was 1.25 µg/mL. The triple ∆comCDE mutant strain showed an even higher susceptibility than each of the single knockouts, with a MIC of 0.75 µg/mL. Altogether, these data suggest that the anti-bacterial effect of CBG is influenced by the ComCDE quorum sensing system. Figure 1. Streptococcus mutans strains deficient in the ComCDE QS system showed increased sensitivity to CBG. The viability of different S. mutans strains (WT, ∆luxS, ∆comC, ∆comE and ∆comCDE) after a 24 h incubation with increasing doses of CBG (0.75-5 µg/mL) as measured by OD 595nm . n = 3; * p < 0.05; ** p < 0.01.

21-CSP Prevented the Anti-Bacterial Effect of CBG on the ∆comC and ∆comE Strains, but Not on ∆comCDE
To test the effect of the 21-CSP signal on the anti-bacterial effect of CBG on the different S. mutans mutant strains, the different strains were exposed to CBG (0.75-5 µg/mL) in the absence or presence of 1 µg/mL of 21-CSP, and the bacterial growth was analyzed in a kinetic study by measuring changes in the optical density. 21-CSP alone caused a minor increase in the bacterial growth of the WT strain ( Figure 2A) and did not affect the bacterial growth in the presence of CBG at the tested concentrations (0.75-5 µg/mL) (Figures 2A and S1A). 21-CSP significantly prevented the delay in the log phase growth of the ∆luxS strain caused by 1.25 µg/mL CBG ( Figure 2B). Notably, 21-CSP antagonized the anti-bacterial effect induced by 1.25 µg/mL CBG on ∆comC ( Figure 2C) and ∆comE ( Figure 2D), suggesting that 21-CSP induces a cell survival response that opposes the CBG anti-bacterial effects. The anti-bacterial effect of CBG against the ∆comCDE mutant remained unchanged despite the presence of 21-CSP ( Figure 2E). Since the ∆comE mutant responded to 21-CSP, this indicates that the survival response induced by 21-CSP is mediated by a different pathway than the classical ComCDE cascade. However, since 21-CSP had no effect on the triple ∆comCDE mutant ( Figure 2E), it is likely that the signal depends on the ComD receptor. It is noteworthy that at the higher CBG concentrations (2.5 and 5 µg/mL), 21-CSP exerted no effect on any of the S. mutans mutant strains ( Figure S1). This might be due to the activation of cell death pathways in the bacteria upon higher concentrations. 3.3. DPD Increased the Susceptibility of WT and ∆luxS, but Not of ∆comC or ∆comE, to CBG To examine the effect of the AI-2 signal on the anti-bacterial effect of CBG on the different S. mutans mutant strains, the strains were exposed to CBG (0.75-5 µg/mL) in the absence or presence of 5 µM of DPD (pre-AI-2), and the bacterial growth was analyzed in a kinetic study by measuring changes in the OD. We found that AI-2 increased the susceptibility of the WT and the ∆luxS strains to the anti-bacterial effect of CBG ( Figure 3A,B).
The combined treatment of DPD with 1.25 µg/mL CBG completely prevented the bacterial growth of these two strains ( Figure 3A,B). At 0.75 µg/mL CBG, DPD delayed the log growth phase, and the bacteria did not reach the OD of the control bacteria even after a 20 h incubation ( Figure 3A,B). DPD did not significantly affect the sensitivity of the ∆comC ( Figure 3C) and ∆comE ( Figure 3D) to CBG, and surprisingly, DPD actually inhibited the anti-bacterial activity of 1.25 and 0.75 µg/mL CBG on the ∆comCDE strain ( Figure 3E). DPD had no observable effects on the bacterial growth when using higher CBG concentrations (2.5 and 5 µg/mL) ( Figure S2), which might be due to the already strong anti-bacterial activity at these concentrations.

CBG Antagonizes the 21-CSP-Induced Gene Expression of nlmA, nlmB and nlmC/cipB
Next, we were interested in studying the effect of CBG on comCDE gene expression and the expression of genes induced by 21-CSP. The different S. mutant strains were exposed to 1.25 µg/mL CBG in the absence or presence of 1 µg/mL 21-CSP for 4 h, and then the changes in gene expression were studied by real-time qPCR. While CBG alone had some variable effects on comCDE gene expression in the WT strain, it reduced their expression in the ∆luxS, ∆comC and ∆comE strains ( Figure 4A-E). The comA gene expression was also reduced by CBG in these three strains as well as in the ∆comCDE strain ( Figure 4A-E), suggesting that its suppression by CBG is independent of the ComCDE QS cascade. The addition of 21-CSP led to the induction of comCDE genes in the WT, ∆luxS and ∆comC strains, while no effect was seen on the ∆comE and ∆comCDE strains ( Figure 4A-E), which confirms that this induction is dependent on the regulatory ComE transcription factor. Notably, CBG strongly reduced the 21-CSP-induced expression of comCDE genes ( Figure 4A-E). To further confirm that CBG has an inhibitory effect on the ComCDE QS system, its effect on ComE-regulated genes was studied. CBG at 1.25 µg/mL had only a mild inhibitory effect on the expression of the nlmA, nlmB and nlmC genes in the WT strain ( Figure 5A-E), and it consistently reduced the expression of these genes in the four mutant bacterial strains ( Figure 5A-E). CBG strongly inhibited the 21-CSP-induced expression of these genes in the WT, ∆luxS and ∆comC strains ( Figure 5A-E). As expected, 21-CSP did not induce the expression of the ComCDE-regulated genes in the ∆comE and ∆comCDE strains ( Figure 5A-E). Since the expression of nlmA, nlmB and nlmC genes were repressed by CBG in the ∆comE and ∆comCDE strains, this indicates that the repression is independent of ComE, or the CBG-mediated suppression is downstream to ComE. There are some indications in the literature that suggest a crosstalk between the LuxS and ComCDE QS systems, but so far substantial data are lacking. We therefore decided to use our system to further investigate this possible crosstalk. We first studied the effect of 5 µM of DPD (pre-AI-2) on the expression of the comCDE and comA gene expression following a 4 h incubation. We found that DPD induced the expression of both the comCDE and comA genes in the WT strain with a stronger effect on comA expression ( Figure 6A). This is reflected in the enhanced expression of the nlmA, nlmB and nlmC genes ( Figure 7A). In the ∆luxS strain, DPD significantly induced the expression of the comD, comA and nlmC genes (Figures 6B and 7B). The comCDE genes and ComE-regulated genes were not induced by DPD in the ∆comC strain ( Figures 6C and 7C), while a 1.5-2-fold induction was observed in the ∆comE strain in the presence of AI-2 ( Figure 6D), suggesting a dependency on CSP, and involvement of both ComE-dependent and ComE-independent pathways. Surprisingly, DPD inhibited the expression of the ComE-regulated genes in the ∆comCDE strain ( Figure 7E), suggesting for an AI-2 induced repression in the absence of ComCDE. CBG inhibited the DPD-induced gene expression in the WT, ∆luxS and ∆comE strains ( Figure 6A,B,D). Notably, 21-CSP and DPD did not affect luxS expression in either of the mutant strains ( Figure 8A,B), while CBG significantly reduced the luxS gene expression in all conditions ( Figure 8A,B).

CBG Significantly Reduced the Production of AI-2
To study the effect of 21-CSP and/or CBG on AI-2 production by S. mutans, the various strains were incubated with different concentrations of CBG for 4 h in the absence or presence of 1 µg/mL 21-CSP, and then the V. harveyi MM77 reporter strain was exposed to 10% of the collected conditioned medium (CM) and the luminescence emission was monitored in a kinetic assay. The OD 600nm of the S. mutans samples were measured at the time of CM collection, and the bioluminescence obtained was divided by this OD to correct for the different cell densities. As expected [41], the double luxM, luxS null MM77 mutant lacking both AI-1 and AI-2 did not emit bioluminescence by itself ( Figure S3A,B), and as a control, exogenously added DPD (pre-AI-2) induced a strong bioluminescence response ( Figure S3A,B). No difference was seen in the DPD (pre-AI-2)-induced bioluminescence when 21-CSP was added as a control ( Figure S3A,B), indicating that 21-CSP did not interfere with the bioassay. Additionally, the presence of 0.125 µg/mL CBG did not interfere with the bioassay ( Figure S4). BHI was added as a control and caused no bioluminescence of the MM77 cells ( Figure S3A,B). Different S. mutans strains showed different basal bioluminescence. In comparison to WT, the ∆comE strain exhibited a 68 ± 0.5% reduction in the bioluminescence, while ∆comC and ∆comCDE exhibited a reduction of 56 ± 3% and 82 ± 0.5%, respectively (Figure 9), suggesting that the ComCDE QS pathway may affect genes involved in the secretion of AI-2. Due to the absence of the AI-2-producing gene luxS in the ∆luxS strain, no bioluminescence was emitted (Figure 9). 21-CSP reduced the AI-2 production in the WT strain by 37 ± 4% ( Figure 10A,B), while significantly increasing the expression in the ∆comC strain by 50 ± 1.6% ( Figure 10C,D). 21-CSP had no significant effect on the AI-2 production in the ∆comE and ∆comCDE strains ( Figure 10E-H), suggesting that the 21-CSP-induced AI-2 expression depends on the ComDE signal transduction pathway. CBG at 1.25 µg/mL reduced the AI-2 production in WT, ∆comC and ∆comE strains ( Figure 10A-F), while barely having any effect on the ∆comCDE strain ( Figure 10G,H), suggesting that ComD might be involved in the CBG-mediated downregulation of AI-2 production. Further studies are required to prove this hypothesis.

Discussion
Targeting QS pathways has gained popularity during the last several years because of its many applications within medicine and agriculture [42][43][44]. By interrupting the communication between bacteria, QS-regulated virulence factors can be suppressed [45]. This has important clinical implications for treating infectious diseases, especially those caused by antibiotic resistant strains [46].
Several plant species and even bacteria themselves have been found to produce anti-QS substances [44,47,48]. A classic example is the halogenated furanones produced by the Australian red alga (Delisea pulchra), which inhibit the QS system of the marine bacterium Serratia liquefaciens [49] as well as other QS systems in Gram-negative bacteria [50]. The anti-QS compounds can act at several levels [44,51]: (i) They might prevent the binding of the autoinducer to its receptor (e.g., phytol binds to CviR of Chromobacterium violaceum) [52]. (ii) They might inhibit the phosphorelay induced by the autoinducer receptor by either preventing the phosphorylation of downstream mediators or by repressing the expression of the receptor (e.g., cinnamaldehyde represses the expression of LasB, RhlA and PqsA in Pseudomonas aeruginosa) [53]. (iii) They might interact with the transcriptional regulator, thereby preventing the activation of the target genes (e.g., LasR antagonists) [35,54,55]. (iv) They may prevent the synthesis of the autoinducer (e.g., carvacrol inhibits LasI AHL synthetase of Pseudomonas aeruginosa) [56]. (v) Finally, they might cause the degradation of the autoinducer (e.g., AHL-degradation enzymes of Rhodosporidium and Rhodococcus species) [44,57].
Various plant species including the Cannabis sativa L. plant have received much attention due to their therapeutic potential against microorganisms [24,58]. CBG, a non-psychoactive cannabis constituent was found to exert anti-quorum sensing properties against Vibrio harveyi [35]. With regards to S. mutans, CBG was found to exert anti-bacterial and antibiofilm properties [32,33]. CBG exerts a bacteriostatic effect on S. mutans that is affected by the initial bacterial cell density. It also affects properties of the membrane and structure, causing immediate membrane hyperpolarization, increase in the membrane permeability, reduction in the fluidity of the membrane and accumulation of mesosome-like membrane structures [32]. Moreover, CBG prevented the decrease in pH that is usually caused by S. mutans, thereby preventing is cariogenic property [32]. In addition to the anti-bacterial effects of CBG, it also demonstrated anti-biofilm activities. CBG directly inhibited the formation of biofilms by acting as an anti-bacterial agent, and indirectly by acting on metabolic pathways that regulate the formation of biofilms. CBG also reduced essential biofilmregulating gene expression, prevented the production of EPS and caused an induction of reactive oxygen species (ROS) production [33]. Due to the involvement of QS in biofilm formation, we have here investigated the anti-QS activity of CBG on S. mutans.
S. mutans has several two-and three-component QS systems in which post-translationally modified oligopeptides are used as autoinducers [1,17,59]. One of these autoinducers is the 21-competence stimulating peptide (21-CSP) produced as a precursor by the comC gene, that during the export through the ABC ComAB transporter is cleaved into the mature 21-CSP [1,19]. 21-CSP can be further cleaved into 18-CSP by SepM [19,60] (Figure 11). The mature CSPs bind to the membrane ComD receptor leading to the induction of a phosphorelay with consequent phosphorylation and activation of the transcriptional regulator ComE [17]. Another family of autoinducers is the autoinducer-2 (AI-2) whose production relies on the luxS gene [61,62]. AI-2 is a chemical compound (furanosyl borate) formed from its precursor DPD (4,5-dihydroxy-2,3-pentanedione) [61,62]. AI-2 is acclaimed as a 'universal autoinducer' as it facilitates interspecies communication [63]. Figure 11. The anti-QS activity of CBG on S. mutans. CBG prevents the transcription of the comC, comD, comE, comA, nlmA, nlmB, nlmC and luxS genes, thus preventing the CSP and AI-2-induced QS signal transduction pathways. When we exposed the QS-mutant S. mutans strains (WT, ∆luxS, ∆comC, ∆comE and ∆comCDE) to various CBG concentrations (0.75-5 µg/mL), different sensitivities to its anti-bacterial action were observed. Strains deficient in the ComCDE QS system showed increased sensitivity to CBG in comparison to the WT strain. This suggests that the antibacterial activity of CBG is affected by the ComCDE QS system. To further investigate the effect of the 21-CSP signal on the anti-bacterial effects of CBG, we incubated S. mutans WT and QS-knockout strains in the presence of CBG and 21-CSP. We observed that exogenously added 21-CSP prevented the anti-bacterial effect of CBG on the ∆comC and ∆comE strains, but not of the WT strain which constitutively expresses CSP. These findings suggest that 21-CSP induces a cell survival response that opposes the anti-bacterial action of CBG.
Since the ∆comE mutant responded to the survival signal of 21-CSP, it is likely that the survival response induced by 21-CSP is mediated by a different pathway than the classical ComCDE cascade. Based on the finding that 21-CSP had no effect on the triple ∆comCDE mutant, which differs from the ∆comE and ∆comC mutants by lacking the ComD receptor, we propose that the survival signal depends on the ComD receptor which propagates a downstream survival signaling pathway independent of ComE.
We next examined how CBG affects the early expression of genes regulated by the ComCDE QS system. We chose to follow the expression of the ComE-regulated nlmA (SMU.150), nlmB (SMU.151) and nlmC (cipB;SMU.1914) genes, that encode for mutacins (bacteriocins) that affect autolysis and cell viability [64]. Following a 4 h incubation with 1.25 µg/mL CBG, there was a strong repression of the three comCDE genes as well as the ComE-regulated genes nlmA-C in all strains except for the WT. It should be noted that the ∆luxS strain responded to 1.25 µg/mL CBG with an 8 h delay in the log growth phase, and all the comCDE mutant strains showed growth arrest during the 20 h incubation period with this CBG concentration, while there was no growth inhibition on the WT strain. Thus, the reduced response of 1.25 µg/mL CBG on the tested gene expression in the WT strain goes along with its lower sensitivity to CBG. Another possibility is that there might be feedback mechanisms between the two QS systems, both systems being present in the WT, while only one of them in the mutant strains.
Since CBG down-regulated the nlmA-C genes also in the ∆comCDE strain, the CBGmediated gene repression seems to be independent of ComE, or it acts downstream of ComE. The addition of 21-CSP to the S. mutans strains significantly induced the expression of the comCDE genes (2-4-fold) and ComE-regulated genes (15-200-fold) in WT, ∆luxS and ∆comC which have functional ComCDE systems, while, as expected, no response was observed in the ∆comE and ∆comCDE strains. CBG prevented the 21-CSP-induced expression of ComE-regulated genes in the WT, ∆luxS and ∆comC strains, suggesting that CBG antagonizes the 21-CSP signal transduction pathway. An alternative explanation is that the CBG-mediated repression of the nlmA-C genes could not be overcome by 21-CSP. Since 21-CSP antagonized the anti-bacterial effects of CBG on S. mutans while the nlmA-C genes are still repressed by CBG in the presence of 21-CSP, it is likely that these genes are not involved in either the 21-CSP-induced survival or the anti-bacterial action of CBG.
The comA gene that is transcribed from a different operon than the comCDE operon [65] was also significantly upregulated by both 21-CSP and DPD. Its induction by DPD was dependent on the ComCDE QS system and strongly inhibited by CBG, again pointing to an anti-QS activity of CBG.
The susceptibility of WT and ∆luxS to the anti-bacterial activity of CBG was increased when DPD was added. However, this addition had no effect on the ∆comC and ∆comE strains, suggesting that also this effect depends on the ComCDE pathway. DPD had no anti-bacterial effect by its own. To our surprise, DPD inhibited the anti-bacterial activity of 1.25 and 0.75 µg/mL CBG on the ∆comCDE strain. The opposite effect of DPD in the presence or absence of the ComCDE system indicates that a mild activation of the ComCDE pathway by AI-2 together with other effects induced by AI-2 promote the anti-bacterial effect of CBG, while in the absence of ComCDE, AI-2 provides pro-survival signals. It is known that AI-2 modulates the expression of an extensive range of genes in S. mutans [21]. Previous studies have shown that DPD reduces the sensitivity of a ∆luxS Streptococcus intermedius strain to antibiotics [66].
To get a better understanding of the AI-2-induced signals, we analyzed the expression level of comCDE genes, ComE-regulated genes and luxS in all our strains after a 4 h incubation with 1.25 µg/mL CBG and/or exogenous DPD. The expression of comCDE genes and ComE-regulated genes in WT, ∆luxS and ∆comE, but not in ∆comC and ∆comCDE, were induced upon the addition of DPD, suggesting that AI-2 requires the 21-CSP encoded by comC for the induction of these genes. There has previously been a vague indication for a crosstalk between the AI-2 and ComCDE system, where an RNA sequencing analysis of a ∆luxS Streptococcus suis strain showed reduced expression of the CSP gene [67]. Our study supports that there is a crosstalk between these two QS systems.
It should be noted that the induction of nlmA, nlmB and nlmC (cipB) by AI-2 was milder (2.5-4.5-fold in WT) than that of 21-CSP (14-90-fold in WT), which is not surprisingly as the latter is the autoinducer of the ComCDE system in S. mutans [68]. In the absence of the ComCDE QS system, DPD repressed the expression of nlmA, nlmB and nlmC (cibB). The opposite effect of DPD on the gene expression in the ∆comCDE strain in comparison to the WT and ∆luxS strain is reflected in bacterial survival in the presence of CBG. Notably, CBG prevented the DPD-mediated gene induction, indicating that CBG also antagonizes the AI-2 signals. CBG alone reduced luxS gene expression in all strains, suggesting a direct gene suppressive effect of CBG.
By using a bioluminescence assay based on the AI-1-and AI-2-deficient V. harveyi strains, the anti-QS effects of CBG were further proven. This was done by detecting the amount of AI-2 secreted by the various S. mutans strains in the presence or absence of CBG and/or 21-CSP. CBG inhibited AI-2 production in all strains tested, and the simultaneous presence of 21-CSP could not overcome this repression. It should be noted that CBG at 1-5 µg/mL inhibited the QS of V. harveyi that was related to CBG-induced LuxO expression and activity with a consequent downregulation in the LuxR gene [35]. In our bioassay, the S. mutans condition medium was diluted 1:10, resulting in a final CBG concentration of 0.125 µg/mL which is below that required for the anti-QS effect on V. harveyi.
In summary, the present study shows that the two QS systems LuxS and ComCDE affect the susceptibility of S. mutans to CBG, and CBG acts as an anti-quorum sensing compound that represses the expression of comCDE, luxS and ComE-regulated genes (Figures 11 and 12). We have further provided evidence that there is a crosstalk between the AI-2 and ComCDE QS systems.