Hydroxyphenylacetic acids trigger predation in S. salivarius

ComR(-like) and structurally-related Rgg sensors are predominant members of the RRNPPA superfamily in streptococci [21,22,31], which are highly influenced by external signals and cell physiology. These parameters could either control the expression of the ComR/Rgg-encoding genes [10,32] or affect the formation of the active ComR/Rgg-peptide complexes, such as through the modulation of inducing peptide availability or the production of interfering peptides (Fig 1A) [11,33–35]. To explore the regulation and activation of the three paralogous sensors, ComR, ScuR, and SarF, from S. salivarius (Ssa) HSISS4, we designed a screening approach based on six luminescent reporter fusions inserted in an ectopic locus (Fig 1B). While a first set of fusions aims to monitor the activation of the promoters of each regulator gene (P_comR, P_scuR, and P_sarF, fused to luxAB genes), a second set reports the impact of these regulators on the activation of the downstream genes (Fig 1A and 1B). The chosen promoters were (i) P_comX as a proxy of competence that is specifically induced by ComR, (ii) P_slvX as a proxy of predation that is induced by both ComR and ScuR, and (iii) P_sptA controlling a bacitracin-like ABC transporter that is activated by ScuR and SarF [7] (Fig 1B). Altogether, these six fusions will allow us to monitor any direct or indirect effect of a physiological stimulus that would activate one or more of these three regulatory systems in S. salivarius.

Since carbon/nitrogen sources were previously shown to impact the activity of Rgg systems in various streptococci [32,36], we screened 192 molecules (e.g., simple/complex sugars, amino acids and derivatives, and organic acids) using phenotype microarrays (Biolog plates PM1 and PM2A) with the six luminescent reporter strains (S1 Table and S1 Fig). As preliminary tests showed limited growth of S. salivarius across a wide range of carbon sources, chemically-defined medium (CDM) was supplemented with 0.15% glucose. This concentration of glucose allows an initial growth (OD₆₀₀~0.75), which could be increased if the tested carbon source is sequentially consumed (diauxic growth) or co-metabolized. While no major increase in luminescence was noted for P_comR, P_scuR, P_sarF, P_comX, and P_sptA reporter strains compared to the control (C^-, CDM with 0.15% glucose alone), two carbon sources induced luminescence with the P_slvX reporter strain (S1 Fig). Interestingly, 4-hydroxyphenylacetic acid (4HPAA) and 3-hydroxyphenylacetic acid (3HPAA) triggered a ~50- and ~20-fold upregulation of P_slvX, respectively (Fig 1C). In addition, P_slvX activation is not due to a higher expression of any of the three regulator genes and does not imply a coactivation of competence, as shown by the absence of P_comX induction (Fig 1C). Because predation by S. salivarius through bacteriocin production requires co-induction of the bacteriocin exporter ComA, encoded within the comS–comA operon (Fig 1A), we next monitored activation of this locus. P_comS was also induced by the two compounds, indicating coordinated activation of the bacteriocin exporter together with its cognate bacteriocins (using P_slvX as a proxy) (Fig 1C).

Together, these results show that predation via the activation of genes encoding bacteriocins and their exporter is stimulated by two small organic acids, 4HPAA and 3HPAA, which are not produced by S. salivarius but are known as byproducts or intermediates in the catabolism of aromatic amino acids of many anaerobic bacteria ([37] and KEGG database).

4HPAA induces predation through ComR

Since the presence of 4HPAA has been reported as a biomarker of microbial dysbiosis in the digestive tract due to the proliferation of proteolytic bacteria [30], this prompted us to further investigate how 4HPAA stimulates the activation of bacteriocin genes. To confirm the activation of P_slvX, the dynamics of its induction by 4HPAA during growth was compared to the XIP pheromone (Fig 2A). Whereas XIP addition triggered a sharp and transient activation peak during early exponential growth, as previously reported [8], 4HPAA induced a more sustained activation throughout exponential growth, consistent with a distinct mode of signal perception. We therefore examined whether these two modes of activation could interact in an additive, synergistic, or competitive manner. When a low concentration of XIP (5 nM) was combined with 4HPAA (1 mM), bacteriocin gene expression displayed a prolonged activation profile, reflecting the combined contributions of both signals (S2A Fig). Interestingly, under these conditions, 4HPAA partially antagonized XIP signaling, leading to an approximately 2-fold reduction of the XIP-induced response, without markedly affecting the overall level of bacteriocin expression (S2B Fig). At higher XIP concentrations (25 nM), this antagonistic effect of 4HPAA was no longer observed (S2C Fig). However, increasing the concentration of 4HPAA (7.5 mM) restored the inhibitory effect (S2D Fig), indicating a strong and dose-dependent interplay between the two inducers. Given that P_slvX could be activated by either ComR or ScuR, we examined the impact of 4HPAA on various deletion mutants (Fig 2B). The deletion of comR (ΔcomR) led to a complete absence of P_slvX induction by 4HPAA. In contrast, the deletion of comS (ΔcomS) did not alter the 4HPAA response, indicating that neither the XIP precursor nor the mature XIP is required for this regulatory pathway. In addition, the deletion of scuR and sarF genes (ΔscuR-sarF) had no impact on 4HPAA stimulation. Altogether, these results show that 4HPAA specifically activates the cytoplasmic sensor ComR.

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 2. 4HPAA activates predation through ComR in salivarius streptococci.

A. Luminescence (RLU/OD₆₀₀) over time of the S. salivarius (Ssa) P_slvX-luxAB reporter fusion without inducer, with 4HPAA (1 mM), or with XIP (5 nM). B. Fold increase in total luminescence between 4HPAA induction (10 mM) and non-induced conditions for the S. salivarius P_slvX reporter strain (WT) and its derivative mutants (ΔcomR, ΔscuR-sarF, and ΔcomS). C. Bacteriocin assays with wild type (WT) and mutants P₃₂-scuR (positive control), Δslv5 (negative control), ΔcomR, and ΔscuR-sarF without inducer (−), with 1 μM XIP (ComR-specific), 1 μM sBI7 (ScuR/SarF-specific), or 10 mM 4HPAA. Lactococcus lactis IL1403 was used as an indicator strain. D. Bacteriocin assays with increasing concentrations of 4HPAA (0 to 10 mM) on wild type (WT) and ΔcomR mutant (negative control). Indicator strain as in panel C. The bottom graph shows the size of inhibition zones (mm). E. and F. Maximum luminescence in response to a 4HPAA gradient (0 to 10 mM) for S. salivarius (Ssa) P_slvX-luxAB (ComR_Ssa) (panel E), S. thermophilus (Sth) P_comS-luxAB ΔcomS (ComR_Sth), and its isogenic strain comR::comR_Sve (ComR_Sve) (panel F). Data presented in panels A, B, D, E, and F are mean values of biological (panels A, B, E, and F) or technical replicates (panel D) ± standard deviation (error bars in panel B and D, and light color zones in panels A, E, and F). In panel D, the inhibition zones induced by 4HPAA were statistically compared to the WT without 4HPAA using One-Way ANOVA with Dunnett’s test (** P < 0.01; **** P < 0.0001; ns, non-significant). The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003718.g002

Bacteriocin assays were also performed to confirm that 4HPAA effectively induces a genuine predation response. We tested the wild-type strain, a mutant strain with all five bacteriocin loci deleted (Δslv5; negative control), a mutant constitutively expressing bacteriocins (P₃₂-scuR; positive control), the ΔcomR mutant, and the double ΔscuR-sarF mutant. Four experimental conditions were evaluated: no inducer, addition of XIP (1 µM), sBI7 (1 µM), or 4HPAA (10 mM) (Fig 2C). With XIP addition, a ComR-specific inducing peptide, both the wild type and the ΔscuR-sarF mutant produced inhibition halos, but not the ΔcomR mutant. Conversely, with the addition of the sBI7 peptide, which is ScuR/SarF-specific, the wild type and the ΔcomR mutant exhibited inhibition halos, but not the ΔscuR-sarF mutant. Notably, the addition of 4HPAA phenocopies the XIP effect, confirming that the 4HPAA response is strictly ComR-dependent. We also evaluated bacteriocin production in response to an increasing concentration of 4HPAA (Fig 2D). In the wild type, an inhibition halo was detected at a concentration of 0.5 mM, and its size gradually increased up to 10 mM. Given that ComR activation by XIP leads to a coupling between predation and competence [8], we also evaluated the capacity of 4HPAA to induce natural DNA transformation in S. salivarius. 4HPAA-induced cells (10 mM) failed to generate any detectable transformant in standard conditions (S2E Fig), corroborating the absence of P_comX activation by 4HPAA (Figs 1C and S2F).

Altogether, these results show that 4HPAA is exclusively perceived by ComR through a pathway that is independent of the XIP pheromone. Moreover, the selective activation of predation via ComR without competence induction is an exclusive feature of the 4HPAA-inducing response.

4HPAA induction is ComR-specific among salivarius streptococci

S. salivarius, Streptococcus thermophilus (Sth), and Streptococcus vestibularis (Sve) are closely related species forming the salivarius group [1]. This is evident from the high similarity among their respective ComR proteins. Indeed, ComR_Ssa and ComR_Sth display a very high level of identity (95%) and are both activated by the same XIP peptide (XIP_Ssa/Sth, LPYFAGCL) [19]. ComR_Sve is more distantly related (83% of identity with ComR_Ssa) and is strictly induced by a variant XIP peptide (XIP_Sve, VPFFMIYY) [16]. To test the HPAA response, we used three reporter strains, each encoding a different homolog of ComR: S. salivarius P_slvX-luxAB for ComR_Ssa (positive control), S. thermophilus P_comS-luxAB (ΔcomS) for ComR_Sth, and a variant version of the previous strain where ComR_Sth was exchanged by ComR_Sve (comR_Sth::comR_Sve) [16]. The two isogenic strains of S. thermophilus were previously validated for the strict response (P_comS activation) of their ComR version to XIP_Sth and XIP_Sve, respectively [16]. We tested a gradient of 4HPAA (0–10 mM) on these three reporter strains (Fig 2E and 2F). For ComR_Ssa and ComR_Sth reporter strains, an increase in luminescence started in the range of 100–500 µM with a maximum achieved at ~8–10 mM. Interestingly, the third reporter strain encoding ComR_Sve did not respond to 4HPAA at any tested concentration (Fig 2F).

Since the more distant ortholog of ComR (ComR_Sve) is insensitive, these results show that the 4HPAA response is not only ComR-dependent but also ComR-specific, which points towards a direct interaction between 4HPAA (or a derivative compound) and ComR for its activation.

4HPAA activates ComR in vitro

There was no prior evidence for direct interactions between RRNPPA regulators and organic acids as inducers. Therefore, we were prompted to investigate the direct binding of 4HPAA to ComR in vitro. Since ComR_Sth and ComR_Ssa behave similarly in vivo, we used ComR_Sth based on its extensive biochemical characterization [16,17,25]. We performed Electrophoretic Mobility Shift Assays (EMSA) with purified ComR_Sth (and ComR_Sve used as a negative control), a fluorescent ^Cy3P_comS DNA probe, and 4HPAA (or XIP_Sth used as a positive control) as inducer of complex assembly [16,17] (Figs 3A and S3). In these experiments, we added 4HPAA at a single concentration directly into the running buffer to ensure a homogenous distribution of the molecule. We used gradients of ComR_Sth and ComR_Sve (noting that an excess of ComR leads to its precipitation) along with a fixed amount of the DNA probe and performed migration in buffers with and without 4HPAA (10 mM). As a positive control, we also used conditions with a fixed concentration of ComR_Sth and a gradient of XIP_Sth to visualize the position of the ComR·XIP complex. Notably, ComR_Sth in the presence of 4HPAA led to the formation of a specific complex with similar mobility compared to the XIP_Sth positive control (Fig 3A). Conversely, ComR_Sve, which does not respond to 4HPAA in vivo (Fig 2D), displayed no high molecular size complex (S3 Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 3. Direct binding of 4HPAA into the peptide pocket of ComR.

A. Mobility shift assays of the comS promoter probe (40 ng) conducted with a gradient of purified ComR_Sth (gray triangles, 2:2 dilutions from 4 μM) (left) or with a fixed concentration of ComR_Sth (2 μM) and a gradient of XIP_Sth (gray triangles, 25, 50, 200, and 1,000 nM) (right), without (top) or with (bottom) 10 mM 4HPAA. Probes (P) indicated by a black arrow are Cy3-conjugated DNA fragments of 40 bp. Vertical back lines indicate the position of specific complexes and stars label complexes formed in the presence of 4HPAA alone. The control condition without ComR_Sth is indicated with a minus sign (−). B. In silico prediction of 4HPAA in complex with ComR_Sth holoform (right) compared to the native XIP-ComR_Sth complex (PDB 5JUB) (left). The top-ranked docked 4HPAA (blue) is mimicking the positioning of XIP-L8 (white). ComR residues T90, K100, and D167 are highlighted in light green, as they were predicted to directly interact with 4HPAA (yellow dotted lines). ComR residues F171 and Y174 are shown since they are essential for the interaction with XIP. C. In vivo luminescence response (total RLU/OD₆₀₀) of ComR_Sth wild-type (WT) and variants K100A, T90A, and Y174A/F171A to 4HPAA (10 mM) or XIP (5 nM). Experiments were performed with S. thermophilus P_comS-luxAB ΔcomS. D. In vivo luminescence response of ComR_Sth wild-type (WT), ComR_Sth penta-mutant (5× ComR*_Sth) responding only to XIP_Sve, and comR_Sve WT to 4HPAA (10 mM) or XIP (5 nM). Same genetic background as in panel C. Dots, bars, and error bars in panels C and D show biological triplicates, mean values, and standard deviations, respectively. All variants were statistically compared to ComR_Sth WT using one-way ANOVA with Dunnett’s test (* P < 0.05; ** P < 0.01; **** P < 0.0001). The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003718.g003

Together, these results showed the direct binding of 4HPAA to ComR, resulting in the formation of an active complex able to bind DNA. To the best of our knowledge, a chemical compound has never been reported to directly activate a regulator of the RRNPPA family.

4HPAA is predicted to occupy the XIP C-terminus binding site for ComR activation

To investigate how 4HPAA could interact with ComR, we ran in silico docking predictions using the EADock DSS engine from SwissDock [38]. We used the anionic form of 4HPAA and compared docking predictions for the crystal structures of ComR_Sth and ComR_Sve holoforms (XIP-activated), and ComR_Sth apoform. Remarkably, the top-ranked prediction (total energy of −17.7 kcal/mol) for holo-ComR_Sth positioned the 4HPAA molecule deeply in the XIP binding pocket right at the site occupied by the pheromone C-terminal leucine (XIP-L8) (Fig 3B). By comparison, the predicted docking of 4HPAA in the same configuration within the XIP-binding pocket of holo-ComR_Sve is less stable (total energy of −14.8 kcal/mol), and such positioning is not predicted in apo-ComR_Sth. The carboxylate group of docked 4HPAA is predicted to form a salt bridge with the side chain ammonium group of residue K100 from helix α7 and to H-bond with T90 from loop α6-α7 of ComR_Sth, as previously reported for the C-terminal carboxylate group of XIP-L8 (Fig 3B) [17]. In addition, the hydroxyl group of 4HPAA is predicted to H-bond with D167 from helix α10, which could enhance its stabilization in the pocket (Fig 3B). Finally, N208 is contacting the 4HPAA aromatic ring in an adequate orientation to establish a weak H-bond with the π cloud as an acceptor (NH...πbond) [39].

We further modeled ab initio the positioning of 4HPAA in the predicted structures of various ComR proteins using the protein-ligand prediction tool Chai-1 [40] (S4 Fig). Based on the conservation of the four residues reported above (S5 Fig), we selected ComR_Sth (used here as a positive control), ComR_Ssa, ComR_Sve, and ComR of Streptococcus mutans (ComR_Smu), which show perfect conservation of these residues, as well as the more distant ComR proteins from Streptococcus pyogenes (ComR_Spy) and Streptococcus suis (ComR_Ssu), which contain substitutions (i.e., K100R, D167E/R, or N208K/D) (S5 Fig). While 4HPAA was predicted to adopt a similar binding mode in ComR_Sth and ComR_Ssa, consistent with docking results obtained for holo-ComR_Sth, a distinct positioning of the ligand was observed in all other ComR homologs analyzed (S4 Fig). This difference likely reflects a suboptimal architecture of the 4HPAA-binding pocket (e.g., ComR_Sve and ComR_Smu) and/or the lack of conservation of several key interacting residues (e.g., ComR_Spy and ComR_Ssu). Altogether, these in silico predictions suggest that 4HPAA may preferentially bind to the XIP-binding pocket of ComR_Sth and ComR_Ssa, whereas structural divergence in other ComR homologs likely prevents productive ligand accommodation.

We previously reported that mutating residues at the bottom (T90, K100; interaction with XIP-L8) or the top (F171/Y174; interaction with XIP-L1) of the XIP binding pocket in ComR_Sth abolished the XIP_Sth-mediated activation (Fig 3B) [17,25]. Mutants T90A and K100A lost their ability to induce luminescence in response to 4HPAA, mirroring our previous observations with XIP [17] (Fig 3C). In contrast, the double Y174A/F171A mutant exhibited a 4-fold increase in light emission upon 4HPAA activation compared to the control, despite a loss of inducibility by XIP (Fig 3C). These effects are partially recapitulated in the single F171A mutant (S6 Fig). The F171A and Y174A/F171A mutants were previously shown to display a higher basal level of activation without XIP addition [25], indicating a less stringent conformational control that could explain their better activation by the remote binding of 4HPAA. In a previous work, we also generated a ComR_Sth variant (5× ComR*) with five substitutions (R92G, V205A, S248G, S289K, and I290T) that responds exclusively to XIP_Sve due to a remodeling of the peptide binding pocket [16]. Our luminescence assays showed that, as observed with ComR_Sve, this ComR_Sth variant is unable to respond to 4HPAA (Fig 3D). A dissection of individual substitutions in the XIP binding pocket showed that the R92G substitution recapitulates the loss of 4HPAA induction, as previously reported for its key role in the XIP_Sth-mediated activation mechanism [17] (S6 Fig). In addition, the S289K substitution showed an interesting behavior, being more reactive to 4HPAA and less sensitive to XIP_Sth than their respective controls (S6 Fig). The lysine at position 289 could favor initial encountering with 4HPAA to facilitate its activation effect.

Overall, this analysis supports a model where 4HPAA partially mimics the XIP activation mechanism by substituting XIP-L8 for key interactions with residues T90 and K100 in the peptide binding pocket of ComR. This alternative mode of activation is further supported by the 4HPAA ability to activate ComR mutants that are not or are weakly responding to XIP.

Carboxylic acid derivatives of bulky hydrophobic amino acids are ComR inducers

In our investigation of 4HPAA selectivity for ComR activation, we tested several structural variants of this molecule for their in vivo activation capacity with a specific focus on three key aspects: (i) the carboxylate function and its associated negative charge, (ii) the polarity around the aromatic ring, and (iii) the overall size of the molecule (Fig 4A). For variations of the carboxylate moiety, we examined tyramine (featuring a positively charged ammonium at neutral pH) and methyl-4HPAA (a neutral methylester derivative), but none of them elicited a luminescence signal (Fig 4B). This aligns with our 4HPAA-binding prediction, which suggests that the carboxylate is critical for interactions with residues T90 and K100 inside the XIP binding pocket (Fig 3B). In addition, two α-hydroxy-carboxylates, (R)-(−)-mandelic acid and D-(+)-3-phenyllactic acid, also failed to induce luminescence (Fig 4). This is in line with the structural model that predicts steric hindrance upon substituting the pro-R α-hydrogen by a bulkier group or extending the length of the carboxylate moiety. For the polarity around the aromatic ring, phenylacetic acid (PAA), which lacks a hydroxyl group, or compounds with similar steric hindrance, like 3HPAA and 3-fluoro-4HPAA (Fig 4), were able to produce luminescence, although at ~2-fold lower levels (orange bars, Fig 4B). This indicates that the presence, position, or polarity of the hydroxyl group plays a secondary role in ComR activation, suggesting that 4HPAA interaction with residue D167 is not critical. Furthermore, pentafluoro-PAA failed to induce luminescence (Fig 4B), further supporting the potential NH...π H-bond stabilization since fluorine atoms will decrease the electron density of the πcloud [39]. We also examined compounds with a larger steric hindrance than 4HPAA. The presence of one or two methoxy groups in the meta position of the aromatic ring, as in homovanillic acid and 4-hydroxy-3,5-dimethoxy-PAA, or the addition of a second aromatic ring, as in 1-naphthylacetic acid, resulted in inactive compounds (Fig 4).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 4. Extended activation of ComR by bulky derivatives of hydrophobic amino acids.

A. Structure of 4HPAA variants with modifications of the carboxylic moiety (charge and length; blue rectangle), phenyl group substitutions (orange rectangle), steric hindrance (green rectangle); and hydrophobic amino acid derivatives (gray). Modified chemical groups compared to 4HPAA are highlighted in red. The names of chemicals in bold green indicate inducing molecules. The initial and new inducing molecules are labeled with * and ⁺, respectively B. Luminescence assays (total RLU/OD₆₀₀) with 4HPAA structural variants (10 mM) shown in panel A. C. Luminescence assays with organic acids derived from bulky hydrophobic amino acids (10 mM) shown in panel A. Derivatives of Tyr, Phe, Trp, Ile, and Leu induced a signal. In panels B and C, experiments were performed with S. thermophilus P_comS-luxAB ΔcomS. Dots, bars, and error bars show biological triplicates, mean values, and standard deviations, respectively. All variant molecules were statistically compared to the control condition without organic acid (minus sign) using one-way ANOVA with Dunnett’s test (*** P < 0.001; **** P < 0.0001; ns, non-significant). The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003718.g004

Since 4HPAA and PAA are two carboxylic acids derived from aromatic amino acids (Tyr and Phe), we also tested equivalent derivatives from the whole set of hydrophobic amino acids for ComR_Ssa/Sth activation (Figs 4C and S7). In addition to 4HPAA and PAA, the bulkiest compounds, indole-3-acetic acid (IAA; Trp derivative), 3-methylvaleric acid (3MVA, Ile derivative), and 4-methylvaleric acid (4MVA, Leu derivative), were also able to increase the luminescence signal (Fig 4C), while isovaleric acid (Val derivative) with one carbon less than MVA and other amino acid derivatives of shorter length did not (S7 Fig). Among these three activating compounds, we chose 4MVA for EMSAs, which confirmed its ability to activate ComR as observed for 4HPAA (S3 Fig).

Together, these in vivo assays with 4HPAA structural variants strengthen the docking model of ComR activation, regarding (i) the key role played by the carboxylate group and hydrophobic side chain, (ii) the simulating effect of the hydroxyl group in the para position of the aromatic ring, and (iii) the size fitting of the molecule into the binding pocket. Moreover, we pinpoint five amino acid derivatives that could be of biological relevance as signaling molecules since most of them were previously identified in various human fluids associated with microbial dysbiosis [28–30,41–44].

Porphyromonas gingivalis culture supernatant triggers ComR activation

S. salivarius is a dominant commensal species of the oral cavity and the upper part of the small intestine [1,27]. In cases of microbial dysbiosis, the proliferation of gram-negative bacteria (e.g., Prevotella sp. and Porphyromonas sp.) in the oral cavity or Gram-positive bacteria (e.g., Clostridium sp.) in the small intestine was reported to incompletely catabolize aromatic amino acids, leading to increased levels of 4HPAA, PAA, or IAA in saliva, urine, or feces [30]. To evaluate if one of these bacteria could accumulate a substantial level of those compounds in their culture supernatants, we tested the well-studied opportunistic pathogen P. gingivalis of the oral cavity [45]. P. gingivalis W83 was grown under anaerobic conditions in enriched BHI medium (BHIe) until late stationary growth. Filtered supernatants were then added (25% v/v of CDM) to cultures of S. thermophilus and S. salivarius reporter strains (P_comS-luxAB and P_slvX-luxAB, respectively) for luminescence assays (Fig 5A). The S. thermophilus reporter strain was strongly activated by P. gingivalis culture supernatants (unextracted Pg SN, Fig 5A), while the BHIe medium alone was inefficient to stimulate light emission. In addition, culture supernatants of P. gingivalis grown in BHIe enriched with Phe or Tyr significantly increased the luminescence signal (~3-fold), supporting the hypothesis that byproducts of aromatic amino-acid catabolism contribute to activation (S8A Fig). We also validated that the supernatant effect was ComR-dependent and ComS/XIP-independent by using ΔcomR and ΔcomS mutants of the reporter strain, respectively (Fig 5B). To exclude the presence of an inducing peptide in the culture supernatant that could mimic XIP, we also used an oligopeptide transporter mutant (Δami/opp), which remains similarly inducible as the control strain (Fig 5B). Intriguingly, the S. salivarius reporter strain did not respond in the same conditions, except for a weak activation with aromatic amino acid-enriched BHIe medium (S8B Fig). We identified that the BHIe medium alone has a quenching effect on the activation of the ComRS system through a downregulation of comR expression (P_comR-luxAB) (S8C Fig). To counteract this effect, we used an isogenic reporter strain containing a P_xyl2-comR cassette (xylose-inducible) [10]. At a low comR expression level (xylose 0.2%), the S. salivarius reporter strain was activated by the different culture supernatants as shown above with S. thermophilus (S8D Fig).

Download:

• PPT
  PowerPoint slide
• PNG
  larger image
• TIFF
  original image

Fig 5. P. gingivalis culture supernatant triggers ComR activation.

A. Luminescence assays (total RLU/OD₆₀₀) without addition (−), with 1 mM 4HPAA (positive control), and 25% (vol/vol) of BHIe medium (negative control), filtered P. gingivalis supernatant (Pg SN), ethyl acetate-extracted BHIe (negative control), and ethyl acetate-extracted Pg SN in CDM. Experiments were performed with S. thermophilus (Sth) P_comS-luxAB ΔcomS. B. Luminescence assays (total RLU/OD₆₀₀) of ΔcomS, ΔcomR and Δopp mutants in presence of 25% (vol/vol) of Pg SN in CDM. The three mutants are derivatives of S. thermophilus P_comS-luxAB. C. Separation by capillary electrophoresis of ethyl acetate-extracted P. gingivalis supernatant. Electro-chromatograms display a mixture of 4HPAA (3 mM), IAA (2 mM), and PAA (1 mM) used as standards (top left); extracted Pg SN (top right); addition of 0.5 mM 4HPAA to extracted Pg SN (bottom left); and addition of 0.5 mM PAA to extracted Pg SN (bottom right). D. Bacteriocin assays with increasing concentrations of 4HPAA and PAA (each from 0 to 5 mM) on S. salivarius (Ssa) wild type (WT) and ΔcomR mutant (negative control). Lactococcus lactis IL1403 was used as an indicator strain. The bottom graph shows the size of inhibition zones (mm). In panels A, B, and D, dots, bars, and error bars show biological (panels A and B) or technical (panel D) triplicates, mean values, and standard deviations, respectively. In panel A, all tested conditions were statistically compared to the control condition without external addition (minus sign). In panel B, the ΔcomS and Δopp mutants were statistically compared to the ΔcomR mutant. In panel D, all tested conditions with the WT were statistically compared to the control condition without external addition of 4HPAA and PAA. Statistical analyses used a one-way ANOVA with Dunnett’s test (** P < 0.01; **** P < 0.0001; ns, non-significant). The data underlying this Figure can be found in S1 Data.

https://doi.org/10.1371/journal.pbio.3003718.g005

Since these results converged towards the presence of aromatic amino-acid derivatives in the supernatant, we performed an ethyl acetate extraction at low pH to selectively extract hydrophobic acids such as HPAA, PAA, and IAA [46] for their analysis by capillary electrophoresis. We showed that the extracted fraction was still able to generate a specific luminescence signal (extracted Pg SN, Fig 5A) and appeared to contain a mixture (~1:1) of HPAA and PAA, as shown by electro-chromatograms with appropriate standards (Fig 5C). Assuming that those two compounds are the main inducing molecules in the unextracted supernatant of P. gingivalis, we evaluated their individual concentration to be in the range of ~4–5 mM based on a dose-response curve with equimolar concentrations of HPAA and PAA (S8E Fig). To validate that a 1:1 mixture of HPAA and PAA can induce bacteriocin production in S. salivarius wild type, bacteriocin assays were performed with increasing concentrations of both compounds. Remarkably, growth inhibition was detected at a ~10-fold lower concentration than found in the P. gingivalis filtered supernatant (Fig 5D).

Together, these results show that the well-known amino acid degrader, P. gingivalis [47,48], secretes enough aromatic amino acid derivatives in the extracellular growth medium to activate predation in salivarius streptococci.

Carbon/nitrogen sources, organic acids, and synthetic peptides

Carbon/nitrogen sources from Biolog plates PM1 and PM2A are listed in S1 Table. Organic acids and their suppliers are listed in S2 Table. All organic acids were diluted in water to 100 mM and neutralized with an equimolar amount of NaOH. Synthetic octapeptides XIP_Sth (LPYFAGCL) and sBI7 (LPFWLILG) (purity of 95%) and salivaricins BlpK, SlvV, SlvW, SlvX, SlvY, SlvZ, and PsnL [6] were supplied by Peptide 2.0 (Chantilly, VA, USA) and resuspended in 100% dimethyl sulfoxide (DMSO). Final concentration was quantified using a Nanodrop apparatus (ThermoFisher Scientific).

Bacterial strains, plasmids, and oligonucleotides

Bacterial strains, plasmids, and oligonucleotides used in this study are listed and described in the Supporting information (S3 and S4 Tables).

Growth conditions

Escherichia coli TOP10 (Invitrogen) were cultivated with shaking at 37°C in LB (Lysogeny Broth). S. salivarius HSISS4 and S. thermophilus LMD-9 derivatives were grown at 37°C without shaking in M17 (Difco Laboratories, Detroit, MI) or in CDM [58] supplemented with 1% (w/v) glucose (M17G, CDMG, respectively), except when specified. The Sodium β-glycerophosphate (19 g/l) buffer used in M17 was also added to CDM to increase growth and luminescence induction. Lactococcus lactis IL1403 was grown in M17G at 30°C without shaking. S. gordonii LMG17843 was grown in liquid M17G or M17G agar plates supplemented with 5% (v/v) defibrinated horse blood (bioMérieux, France) at 37°C. Agar 1.5% (w/v) was added into M17 and LB plates. Solid plates inoculated with S. salivarius cells were incubated anaerobically in jars (Oxoid AnaeroGen 2.5 L, Thermo Fischer Scientific) at 37°C. Ampicillin (250 µg ml⁻¹), spectinomycin (200 µg ml⁻¹), chloramphenicol (5 µg ml⁻¹), or erythromycin (10 µg ml⁻¹) were added as required. P. gingivalis W83 (ATCC308) cultures were restarted from frozen samples on Columbia agar plates supplemented with yeast extract (5 g l⁻¹), defibrinated horse blood (5% v/v), hemin (25 mg l⁻¹), and menadione (10 mg l⁻¹) at 37°C under a 5% CO₂ atmosphere for 2–3 days. Afterwards, colonies were transferred into liquid Brain Heart Infusion medium supplemented with yeast extract (5 g l⁻¹), hemin (25 mg l⁻¹), and menadione (10 mg l⁻¹) (BHI_e) and incubated for 3–5 days. When needed, 15 mM tyrosine or 15 mM phenylalanine was added to BHIe to stimulate the catabolism of aromatic amino acids. Culture supernatants of cells grown until an OD₆₀₀ of ~1.9 were collected by centrifugation (5,000 g for 15 min) and then filtered (0.22 µm). This supernatant was kept at −80°C before its addition to reporter strain cultures.

Construction of mutants and reporter strains

Null mutants were constructed by exchanging (double homologous recombination) the coding sequences (CDS) of target genes (sequence between start and stop codons) for an erythromycin resistance cassette using natural transformation as reported before [8]. Integration of the antibiotic resistance cassette at the right location was subsequently checked by PCR. The promoters of regulator genes (i.e., comR, scuR, and sarF) were fused to the luxAB reporter genes and inserted with a chloramphenicol resistance cassette at the permissive tRNA threonine locus (HSISS4_r00061) by double homologous recombination. All DNA fragments (S5 Table) were amplified by PCR using the Phusion high-fidelity polymerase (Thermo Fischer Scientific) following a protocol as recommended by the manufacturer. Overlapping PCR products were transferred in competence-induced HSISS4 derivatives, erm and luxAB-cat cassettes were amplified from pGIUD0855ery and pJIMcat, respectively.

Luciferase assays

Overnight precultures were diluted at a final OD₆₀₀ of 0.05. Culture samples (300 μl) were incubated in the wells of a sterile covered black microplate with a transparent bottom (Greiner Bio-One, Alphen a/d Rijn, the Netherlands). Growth (OD₆₀₀) and luciferase (Lux) activity (expressed in relative light units, RLU) were monitored at 10-min intervals during 24 h in a Hidex Sense multimode reader (Hidex, Turku, Finland). Total or maximum specific Lux activity was obtained by dividing Lux activity by the OD₆₀₀ for each measurement and summing all the data obtained over time or selecting the maximum value, respectively. Experimental values represent the averages of at least three independent biological replicates. Statistical analyses of multiple comparisons to the control mean were performed with one-way ANOVA with Dunnett’s test. Standard deviations and P values were calculated.

For the screening of carbon/nitrogen sources using Phenotype Microarrays (Biolog plates PM1 and PM2A), commercial plates were unsuitable for luminescence assays. The carbon/nitrogen sources from the original plates were initially dissolved in 100 µl of CDM without glucose and incubated for 10 min at 37°C with agitation to ensure complete dissolution. Subsequently, this solution was transferred to the wells of a sterile black microplate with a transparent bottom. A preculture of the reporter strain was added to obtain a final OD₆₀₀ of ~0.05. Glucose from a 25% (w/v) stock solution was added to achieve a final concentration of 0.15% (v/v), and the total volume in each well was adjusted to 300 µl with additional CDM. Growth and luminescence were then monitored as reported above.

Bacteriocin assays

The spot-on-lawn (multilayer) detection method [7] used to test bacteriocin production by S. salivarius was performed as follows. Overnight cultures of producer strains were diluted in fresh M17G medium and grown until mid-log phase (OD₆₀₀ of ∼0.5). In parallel, we casted plates with a bottom feeding layer (M17G, 1.5% agar) supplemented with inducing molecules (XIP_Sth, sBI7, 4HPAA, or 4HPAA + PAA in equimolar ratio), when required. Next, an overnight culture of L. lactis IL1403, used as an indicator strain, was incorporated in pre-warmed soft M17G medium (0.3% agar) at a final OD₆₀₀ of 0.05 and casted as a top layer. Finally, we spotted 3 μl of the producer strains on the top layer. Plates were incubated overnight in anaerobic conditions before analysis of the inhibition zones surrounding the producer colonies. Inhibition zones (mm) were measured with the ImageJ software. The size of the inhibition zones was calculated by subtracting the radius of the bacterial spot from the radius of the inhibition halo.

A variation of the above procedure was applied to test the activity of chemically-synthesized salivaricins against P. gingivalis W83 and S. gordonii LMG17843. Stationary-phase cultures (2–3 days) of P. gingivalis W83 or diluted overnight cultures (OD₆₀₀ of 0.05) of S. gordonii LMG17843 were spread evenly with sterile cotton swabs on their respective solid media using the lawn method. Onto these lawns, 2 μl of synthetic bacteriocin (100 μM) was spotted. Plates were incubated overnight under appropriate growth conditions until lawns were established, after which inhibition zones at the application sites were assessed.

Natural DNA transformation assays

S. salivarius overnight precultures in CDMG were diluted in the same medium at a final OD₆₀₀ of 0.05. After an incubation of 75 min at 37°C, 4HPAA (10 mM) or XIP_Sth (5, 200, or 500 nM) and PCR-amplified DNA fragments containing an erm gene (ΔscuR::erm) surrounded by 1.5-kb recombination arms (1 μg) were added. Cells were grown for 4 h at 37°C before plating on selective M17G agar and incubation in anaerobic conditions.

ComR purification

E. coli TOP10 strains containing plasmids pBADcomR_Sth-streptag [59] and pBADcomR_Sve-streptag [16] were used for the purification of ComR_Sth-StreptagII and ComR_Sve-StreptagII proteins, respectively. Protein purification and storage were performed as previously described [16]. Protein purity was analyzed by SDS-PAGE and protein concentration was measured using a Nanodrop apparatus (Thermo Fisher Scientific).

Electrophoretic mobility shift assays

EMSA assays were performed as previously described [7,16]. Twofold serial dilutions of purified ComR protein (initial concentration of 4 μM) were mixed with a 40-bp dsDNA fragment (40 ng) carrying the ComR box of P_comS coupled to the Cy3 fluorophore in the binding buffer. In parallel, a fixed concentration of purified ComR protein (2 μM) was mixed with a gradient dilution of XIP_Sth (5, 50, 200, and 1,000 nM) together with ^Cy3P_comS (40 ng). Negative controls were performed in the absence of protein. Samples were incubated at 37°C for 10 min prior to separation on a native 4% to 20% gradient gel (iD PAGE gel; Eurogentec). Carboxylic acids were added to the running buffer (MOPS) at a final concentration of 10 mM. DNA complexes were detected by fluorescence on the Amersham Typhoon biomolecular imager (Cytiva) with bandpass excitation and emission filters of 595/25 nm (Cy3). The double-stranded DNA fragment was obtained from annealing of single-stranded Cy3-labeled (at the 5′ end) and unlabeled oligonucleotides.

Supernatant extraction and analysis by capillary electrophoresis

Considering the water solubility of 4HPAA at different pHs and its solubility in ethyl acetate [46], a dedicated extraction protocol was developed to extract 4HPAA and closely related molecules from P. gingivalis culture supernatants. Initially, 4 ml of culture supernatant was acidified using 0.1 ml of concentrated hydrochloric acid (HCl, 37% w/w). Subsequently, 2 ml of ethyl acetate was incorporated into the solution, followed by thorough mixing and a 10-min decantation period. The organic phase was then separated and mixed with a sodium hydroxide (NaOH) solution (2 ml, 0.01 M). After a further 10-min period of decantation, the aqueous phase was isolated, neutralized with HCl (37% w/w), and filtered (0.22 µm).

For analytical characterization, the extracted sample was separated using a capillary electrophoresis apparatus (Capel 105M, Lumex Instruments), with boric acid (0.1 M, pH ~8.5) as the background electrolyte. A bare fused silica capillary of 54/46 cm in total/effective length with an internal diameter of 50 μm from Agilent was used. Samples were injected at a pressure of 30 mbar for 5 s. A voltage of 25 kV was applied throughout the analysis. UV absorbance at 210 nm was used for detection. The capillary temperature was maintained at 20°C during all steps. The analyses were conducted on the sample alone and in the presence of either 0.5 mM 4HPAA or 0.5 mM PAA, allowing for the assessment of the extraction efficiency and the selectivity of the protocol towards 4HPAA and closely related molecules.

4HPAA docking and structure prediction or visualization

For in silico docking predictions of 4HPAA in ComR, the EADock DSS engine from SwissDock was used with default parameters [38]. Docking was performed with the crystal structure of 4HPAA (CCDC 274674) in its anionic form. The holoforms of ComR_Sth (PDB: 5JUB, chain A) and ComR_Sve (PDB: 6HUA, chain A), and the apoform of ComR_Sth (PDB: 5JUF, without SO₄ and H₂O molecules) were used as targets. Ab initio modeling of ComR·4HPAA complexes were performed with the protein-ligand prediction tool Chai-1 using default parameters (https://lab.chaidiscovery.com) [40] using 4HPAA in SMILES format and the primary amino-acid sequences of ComR from S. thermophilus LMD-9 (GenBank: ABJ65625.1), S. salivarius HSISS4 (GenBank: ALR79229.1), S. vestibularis F0396 (GenBank: EFQ60116.1), Streptococcus mutans UA159 (GenBank: AAN57849.1), Streptococcus pyogenes M1 (GenBank: XXA72633.1), and Streptococcus suis P1/7 (GenBank: CAR44181.1). Structure predictions of ComR mutants with AlphaFold 2 were obtained from the AlphaFold CoLab notebook using default parameters [60]. The figures with structural elements were prepared by using the graphic software PyMol (http://www.pymol.org/). The multiple amino-acid sequence alignment was performed with the PRALINE program using default parameters (https://www.ibi.vu.nl/programs/) [61].