Multiple sequence alignment of the hypervariable regions of streptococcal M proteins. The top panel shows a MARCOIL coiled coil prediction [43] for full-length M3 protein. Positions of signal and wall anchor cleavage sites are indicated by S and W, respectively. The PARF motif (position indicated by pink bar), previously suggested to be required for collagen binding, resides in a region of M3 that is predicted to not adopt coiled coil topology. In the alignment, sequence similarity is indicated by grey shading for residues conserved or similar in at least 75% of sequences. 100% conserved residues are highlighted by black shading. Uniprot identifiers are shown to the left of the alignment. Sequences are ordered by alignment score. The proteins are from the following species: Streptococcus pyogenes: A0A0H2UWN1 (M3), A0A0E1EQ89 (M3.2), M4HZY1 (M133), M4I010 (M31), P19401 (M12), M4I038 (M228), Q840T7 (M229), Q54840 (M55), M4HZT2 (M222); Streptococcus dysgalactiae subsp. equisimilis: Q1KQ01, Q9L4N1, Q00720, Q1KQ03, Q5YB85, Q4ZGP4, D0EZI1, W0T3Y6, W0T3A4; Streptococcus equi subsp. zooepidemicus: I7AXP7.

Binding of full-length M3 protein fused to GST to immobilised CLC II and III peptides. ELISA signal based on GST antibody is plotted against peptides spanning the tropocollagen (triple-helical) regions of collagens II and III, as well as positive (collagen II) and negative (GPP10, BSA) controls. Signals for three peptides chosen for further binding experiments are highlighted in red.

Effect of amino acid classes on binding of M3 to CLCs.

1H,15N HSQC spectra for monomeric and disulfide-bond stabilized dimeric M3-NTD.

ITC binding curves for M3-NTD interactions with CLC peptides. Top panels show heat responses to repeated injections of M3-NTD into collagen peptide solutions, with baselines shown in red. Bottom panels shown integrated signals (enthalpy changes) plotted against the molar ratio of binding partners. Where non-linear fitting gave meaningful results, the fits are shown as blue lines, and dissociation constants (KD) are specified.

Data collection and refinement statistics

Crystal structure of M3-NTD (PDB 8p6k). A) Ribbon diagram of the covalently stabilized M3-NTD dimer. Monomers are shown in blue and grey. The regions previously associated with collagen binding (PARF motif) are highlighted in red. N-termini (Asp42) and C-termini (Cys151) are labelled N and C, respectively. The three helices of the blue monomer are labelled H1-3. The C-terminal disulfide bond is shown as sticks. B) Ribbon diagram model for the full extracellular region of M3, predicted by AlphaFold3 [49] and colored by per-residue confidence (pLDDT). C) Conserved leucine, isoleucine (Ile60) and glycine (Gly103) residues define the T-junction structure of M3-NTD. D) Role of conserved residues around the PARF motif in stabilizing the T-bar region of M3-NTD. Polar contacts are shown as dashed lines.

AlphaFold3 predictions for M proteins of GAS, SDSE and SESZ. A) Overlay of the experimental M3-NTD structure (magenta) with a predicted structure for the N-terminal 230 residues of mature M3 (coloured by pLDDT). B) Predicted structures for the N-terminal 230 residues of other M proteins included in the sequence alignment (Figure 1). C) Predicted structures for the N-terminal 230 residues of M1 and M28 proteins, which are not known to interact with collagens.

Structural basis of collagen binding by M3 (PDB 8p6j). A) Crystal structure of the complex of M3-NTD (subunits shown in light and dark grey cartoon representation) with collagen-derived peptide JDM238. Two binding registers are observed bound to equivalent sites of the M3-NTD dimer, indicated by shades of magenta and cyan. B) Space-filling models of views from the top of the T-bar of M3-NTD (top) and looking down the stem towards the bottom of the T-bar (bottom). The PARF motif is shown in tones of red, otherwise coloring is as in A.

Collagen triple helix/M3-NTD interface. A) Residues of M3-NTD (grey cartoon representation) that form the collagen binding site are shown as sticks and are labelled. The collagen peptide triple helix is shown in stick and surface representation in shades of magenta. The sequence is shown indicating the staggered arrangement of the chains in the triple helix. B) Superposition of the two peptide binding sites to highlight conservation of collagen binding mode despite sequence deviation between the two binding registers. Tyr96 and Trp103 are shown as sticks. The two monomers of M3-NTD are shown in light and dark shades of grey. Collagen peptides of the two binding modes are shown in magenta and cyan, with five equivalent sidechains shown as sticks. The equivalent sequences of the binding sites are shown. Water molecules in the binding interface are shown as blue spheres. In the zoomed-in image on the bottom right, only one of the collagen peptide chains is shown per complex for clarity. Hydrogen bonds are shown as grey dashed lines.

Effect of collagen type I on biofilm formation by necrotizing soft tissue infections strains. A) Quantitative analysis of biofilm formation on polystyrene surface with or without human type I collagen. The inoculum of each bacterial strain was 105 CFU per well. Significance was determined by one-way ANOVA with Tukey’s post-hoc test. ****, P<0.0001; ***, P< 0.001; **, P< 0.01; *, P< 0.05 B) Varying effect of collagen concentration. The inoculum of each bacterial strain was 105 CFU per well. C) Inoculum effects on biofilm of strain 2028. D) Effect of concentration of coating collagen on strain 2028 biofilm (100 CFU per well). A-D) Results are shown as Mean + SE. All assays were repeated at least three times in triplicate. E) Confocal microscopic analysis of biofilm formation 48 h after incubation. Fluorescence staining (WGA-Alexa 488, DAPI, and Nile red) of biofilm on uncoated and collagen-coated glass slides. Scale bars indicate 50 μm.

Competition analysis of biofilm formation of GAS isolate 2028 (100 CFU per well) in the presence of M3-NTD. Results are shown as Mean + SE. All assays were repeated three times in triplicate. Significance was determined by one-way ANOVA, followed by Tukey’s post-hoc test ****, P<0.0001; ***, P< 0.001.

Colocalization of collagen with emm3 GAS in patients’ biopsies. Frozen biopsy sections were stained with anti-GAS (green), anti-collagen IV (red) and DAPI (blue). Scale bars indicate 50 μm.

Colocalization of collagen with emm3 GAS in 3D skin tissue model. Skin tissue models were infected with GAS strains 2006 and 2028. At 8, 24 and 48 hours after infection, frozen sections were stained with anti-GAS (green), anti-collagen IV (red) and DAPI (blue). The merged images are shown.

List of mutagenic primers.

The Figure shows the number of specific residues or residue classes per CLC peptide (y-axis) plotted versus rank group (x-axis). Groups are defined as low affinity (A450 = 0 to 0.25), medium affinity (A450 = 0.25 to 0.5) and high affinity (A450 = 0.5 to 0.5) in the solid phase binding assays. Statistical difference between numbers in each group was determined by non-parametric testing as defined in Methods.

Additional AlphaFold3 predicted structures of proteins included in the sequence alignment (Figure 1). The N-terminal 230 residues of mature proteins were included in predictions (colored by pLDDT). For M133 only the first 196 residues have been modelled to highlight of the effect of the deletion of residues that are integral to the T-shape structure of M3.

AlphaFold3 predicted structures of M proteins in complex with a model collagen peptide (GPO8). Two chains of the N-terminal 230 residues of mature proteins and, for clarity, only three collagen peptide chains were included in predictions. M proteins are shown in cartoon, GPO8 peptide in surface representations, respectively (coloured by pLDDT).

Biofilm formation of emm3 GAS. A) Quantitative analysis of biofilm formation on polystyrene surface with or without human type I collagen. The inoculum of each bacterial strain was 105 CFU per well. This assay was repeated at four times in triplicate and results are shown as Mean + SE. Significance was determined by one-way ANOVA with Tukey’s post-hoc test. ****, P<0.0001; ***, P< 0.001; **, P< 0.01; *, P< 0.05. B) Confocal image of emm3 GAS stained with DAPI and M3-specific antibodies. BCW, B- and C-repeat and wall-spanning regions of M3. Scale bars indicate 5 μm.