Uncovering the arsenal of class II bacteriocins in salivarius streptococci

Abstract

Facing the antibiotic resistance crisis, bacteriocins are considered as a promising alternative to treat bacterial infections. In the human commensal Streptococcus salivarius, the production of unmodified bacteriocins (or salivaricins) is directly controlled at the transcriptional level by quorum-sensing. To discover hidden bacteriocins, we harnessed here the unique molecular signatures of salivaricins not yet used in available computational pipelines and performed genome mining followed by orthogonal reconstitution and expression. From 100 genomes of S. salivarius, we identified more than 50 bacteriocin candidates clustered into 21 groups. Strain-based analysis of bacteriocin combinations revealed significant diversity, reflecting the plasticity of seven independent loci. Activity tests showed both narrow and broad-spectrum bacteriocins with overlapping activities against a wide panel of Gram-positive bacteria, including notorious multidrug-resistant pathogens. Overall, this work provides a search-to-test generic pipeline for bacteriocin discovery with high impact for bacterial ecology and broad applications in the food and biomedical fields.

Fig. 1. Schematic representation of the six steps of the multiparametric bioinformatics pipeline.

Red boxes represent the ComR-box and blue arrows indicate the presence of downstream open reading frame(s). Leader sequences, mature sequences, and ribosome binding sites are shown as green squares, orange arrows, and purple dots, respectively.

Fig. 2. Distribution of salivaricins among oral streptococci.

a Occurrence of the 21 prototypical salivaricins in S. salivarius compared to three streptococcal species of in the oral cavity. The number in each box indicates the frequency of observation of the bacteriocin among all analyzed strains of the species. The analysis was performed on S. salivarius, S. thermophilus, S. pneumoniae, and S. pyogenes for 99, 78, 137, and 261 genomes, respectively. b Phylogenetic tree and heatmap of sequence identity (%) of natural BlpK variants from various streptococcal strains. The streptococcal groups are indicated on the right (Sal., salivarius; Pyo., pyogenic). BlpK sequences were aligned using MUSCLE and the tree was then generated with MEGA11 using the neighbor-joining method (10,000 bootstrap replicates). Bootstrap values are given at each node. c Alignment of natural BlpK variants. The alignment was generated with Clustal Omega and residues were colored (RasMol code) according to physicochemical properties.

Fig. 3. Composition of salivaricin cocktails identified in S. salivarius.

a Schematic representation of the 37 salivaricin cocktails identified in seven loci from 100 S. salivarius genomes. A blue box indicates the presence of the bacteriocin. The hierarchical clustering in bacteriocin content of the different strains is shown on the left. The C letter on the right indicates that the cocktail was identified in at least one strain for which a closed genome is available. b Occurrence (%) of the number of salivaricins among S. salivarius strains. c Occurrence (%) of each individual salivaricin in S. salivarius.

Fig. 4. Comparison of the seven salivaricin loci in S. salivarius.

Panels (a–g) display a representative variability in the seven different bacteriocin loci. A red box indicates the presence of a ComR-box; arrows represent genes encoding bacteriocins (blue), immunity peptides (green), bacteriocin leader sequences without the mature part (orange), fragments of the BlpRH system (purple), homologues of BlpI (pink), and other unrelated products (black). Dotted lines are spaces that were artificially added to align conserved genes in order to better visualize the genetic reorganization of each locus.

Fig. 5. Activity assays performed with individual bacteriocins produced by S. salivarius.

a Test strain (Δslv5) for in vivo production of salvaricins. The gene fragment encoding the mature part of a salivaricin (orange arrow box) was fused to expression-secretion signals of blpK (P_blpK-blpK’; green rectangle). The addition of the XIP inducer (blue circles) activates ComR, which will in turn stimulate the production of ComS (positive feedback loop via PptAB-mediated ComS export and Opp/Ami-mediated XIP import) and ComA (bacteriocin exporter, red arrow box). Concomitantly, the ComR-XIP complex activates the production of the hybrid bacteriocin by binding to P_blpK. b Spectra of salivaricins against 12 Gram-positive bacteria belonging to the Streptococcus, Lactococcus, Enterococcus, Staphylococcus, and Listeria genera. The list of indicator strains is reported in Supplementary Table 8. In case of S. pneumoniae and S. aureus, strains PMEN31 and ATCC 6538 were used as indicator strains, respectively. Green, orange, and white boxes refer to high, weak, and absence of activity. c Spot-on-lawn assays with the 21 producer strains used to generate the table shown in panel (b). Single-peptide producers of class IIa bacteriocins were mixed. High (inhibition zone > 0.5 mm) and weak (inhibition zone ≤ 0.5 mm) activities are highlighted by green and orange lines, respectively. Producers are grouped into three categories: no activity, narrow spectrum, and large spectrum. Strain Δslv5 without bacteriocin activity is used as a negative control (Ctl⁻). Each inhibition spot is representative of three biologically independent experiments performed in the same conditions (Supplementary Data 2).

Fig. 6. Activity assays performed against antibiotic-resistant strains of S. aureus and S. pneumoniae.

Spot-on-lawn assays against antibiotic-resistant strains of S. pneumoniae (a) and S. aureus (b). The antibiotic resistance profile of strains is provided in Supplementary Table 8. High (inhibition zone > 0.5 mm) and weak (inhibition zone ≤ 0.5 mm) activities are surrounded by green and orange lines, respectively. Δslv5 mutant deprived of bacteriocin activity is used as a negative control. Each inhibition spot is representative of three biologically independent experiments performed in the same conditions (Supplementary Data 3).

Fig. 7. Activity assays of modified salivaricin cocktails produced by S. salivarius HSISS4.

a Spot-on-lawn assays (left panels) of HSISS4 derivatives (wild-type and Δslv5) producing PsnJ, PsnK, or PsnL against four Gram-positive bacteria. The list of indicator strains is reported in Supplementary Table 8. In case of S. aureus, strain ATCC 6538 was used as indicator strain. Strains Δslv5 and HSISS4 (wild-type) are used as negative (Ctl⁻) and positive (Ctl⁺) controls, respectively. Bacteriocin producers are surrounded by green lines. A schematic representation of inhibitory spectra from HSISS4 cocktails is shown in the right panel. Boxes with a minus sign indicate that no activity was detected. Boxes surrounded in green indicate high activity (inhibition zone > 0.5 mm). Boxes filled in green indicate a modified activity due to the presence of the additional salivaricin in the cocktail of wild-type HSISS4; light and dark green colors refer to additive and synergic effects, respectively. Each inhibition spot is representative of three independent experiments performed in the same conditions (Supplementary Data 4). b Inhibition zone (mm) of HSISS4 wild-type (dark gray bars) or Δslv5 (light gray bars) derivatives producing PsnJ, PsnK, or PsnL against four Gram-positive bacteria (list of strains in Supplementary Table 8). Inhibition zones were measured with the ImageJ software. Dots show the values of biologically independent experiments (n = 3); mean values ± standard deviations. Statistical analysis was performed using One-Way ANOVA with Tukey’s correction. ***, P-Value < 0.001 (Supplementary Data 5).

Fig. 8. Global prey spectra and structure prediction of class II salivaricins.

a Global overview of the overlapping activity spectra of active salivaricins against streptococci and Firmicutes. The activity against L. lactis (dairy strain) is not included here since considered as not relevant for the ecology of the digestive tract. b Structure prediction by AlphaFold 2 of monopeptides and two-peptide bacteriocins. Predicted 3D structures are organized according to their subclasses (IId, IIb, and IIa) in class II bacteriocins and shared structural elements (α helices and β strands). The presence of a predicted disulfide bond (C2) is also indicated. The color of the bacteriocin name in green, orange, and black refers to a large spectrum of prey species, a narrow spectrum of prey species, and the absence of identified activity, respectively. Monopeptides are in dark gray, while dipeptides are in green and pink. Cysteine, positively charged, and negatively charged residues are colored in yellow, red, and blue, respectively.