Structural basis for collagen recognition by the Streptococcus pyogenes M3 protein and its involvement in biofilm

Reviewer #1 (Public review):

Summary:

Wojnowska et al. report structural and functional studies of the interaction of Streptococcus pyogenes M3 protein with collagen. They show through X-ray crystallographic studies that the N-terminal hypervariable region of M3 protein forms a T-like structure and that the T-like structure binds a three-stranded collagen-mimetic peptide. They indicate that the T-like structure is predicted by AlphaFold3 (with varying confidence level) in other M proteins that have sequence similarity to M3 protein and M-like proteins from group C and G streptococci. For some, but not all, of these related M and M-like proteins, AlphaFold3 predicts complexes similar to the one observed for M3-collagen. Functionally, the authors show that emm3 strains form biofilms with more mass when surfaces are coated with collagen, and this effect can be blocked by an M3 protein fragment that contains the T-structure. They also show the co-occurrence of emm3 strains and collagen in patient biopsies and a skin tissue organoid.

Strengths:

The paper is well-written and the data presented is mostly sound.

Weaknesses:

However, a major limitation of the paper is that it is almost entirely observational and fails to draw a causal relationship. This is mainly due to the near-total absence of mutational studies.

Reviewer #2 (Public review):

Streptococcus pyogenes, or group A streptococci (GAS) can cause diseases ranging from skin and mucosal infections, to plasma invasion, and post-infection autoimmune syndromes. M proteins are essential GAS virulence factors that include an N-terminal hypervariable region (HVR). M proteins are known to bind to numerous human proteins; a small subset of M proteins were reported to bind collagen, which is thought to promote tissue adherence. In this paper, the authors characterize M3 interactions with collagen and its role in biofilm formation. Specifically, they screened different collagen type II and III variants for full-length M3 protein binding using an ELISA-like method, detecting anti-GST antibody signal. By statistical analysis, hydrophobic amino acids and hydroxyproline were found to positively support binding, whereas acidic residues and proline negatively impacted binding (Table 1). The authors applied X-ray crystallography to determine the structure of the N-terminal domain (42-151 amino acids) of M3 protein (M3-NTD). M3-NTD dimmer (PDB 8P6K) forms a T-shaped structure with three helices (H1, H2, H3), which are stabilized by a hydrophobic core, inter-chain salt bridges and hydrogen bonds on H1, H2 helices, and H3 coiled coil. The conserved Gly113 serves as the turning point between H2 and H3 (Figure 5). The M3-NTD is co-crystalized with a 24-residue peptide, JDM238, to determine the structure of M3-collagen binding. The structure (PDB 8P6J) shows that two copies of collagen in parallel bind to H1 and H2 of M3-NTD. Among the residues involved in binding, conserved Try96 is shown to play a critical role supported by structure and isothermal titration calorimetry (ITC). The authors also apply a crystal-violet assay and fluorescence microscopy to determine that M3 is involved in collagen type I binding, but not M1 or M28 (Figure 9). Tissue biopsy staining indicates that M3 strains co-localize with collagen IV-containing tissue, while M1 strains do not. The authors provide generally compelling evidence to show that GAS M3 protein binds to collagen, and plays a critical role in forming biofilms, which contribute to disease pathology. This is a very well-executed study and a well-written report relevant to understanding GAS pathogenesis and approaches to combatting disease; data are also applicable to emerging human pathogen Streptococcus dysgalactiae. One caveat that was not entirely resolved is if/how different collagen types might impact M3 binding and function. Due to the technical constraints, the in vitro structure and other binding assays use type II collagen whereas in vivo, biofilm formation assays and tissue biopsy staining use type I and IV collagen; it was unclear if this difference is significant. One possibility is that M3 has an unbiased binding to all types of collagens, only the distribution of collagens leads to the finding that M3 binds to type IV (basement membrane) and type I (varies of tissue including skin), rather than type II (cartilage).

Author response:

Many thanks for assessing our submission. We are grateful for the reviews and recommendations that will inform a revised version of the paper, which will include additional data and modified text to take into account the reviewers’ comments.

We appreciate Reviewer #1’s suggestion regarding the use of mutational work to demonstrate that collagen binding is indeed dependent on the T-shaped fold. However, we believe that this approach is neither feasible nor necessary for our study. Instead, we propose to measure collagen binding to a monomeric form of M3, which preserves all residues including the ones involved in binding, but cannot form the T-shaped structure. This will achieve the same as unravelling the T fold through mutations, but at the same time removes the risk of directly affecting binding through altering residues that are involved in both binding and definition of the T fold.

Structural biology is by its nature observational, which is not a limitation but the very purpose of this approach. Our study goes beyond observing structures. We identify a critical residue within a previously mapped binding site, and demonstrate through mutagenesis a causal link between presence of this residue on a tertiary fold and collagen binding activity. We will firm up our mutational experiments with a characterisation of the M3 Tyr96 variants to confirm that these mutations did not affect the overall fold. We further demonstrate that the interaction between M3 and collagen promotes biofilm formation as observed in patient biopsies and a tissue model of infection. We show that other streptococci, that do not possess a surface protein presenting collagen binding sites like M3, do not form collagen-dependent biofilm. We therefore do not think that criticising our study for being almost entirely observational is justified.

We thank Reviewer #2 for the thorough analysis of our reported findings. The main criticism here concerns the question if binding of emm3 streptococci would differ for different types of collagen. We will address this point in the revised manuscript. Our collagen peptide binding assays together with the structural data identify the collagen triple helix as the binding site for M3. While collagen types differ in their functions and morphology in various tissues, they all have in common triple-helical tropocollagen regions (with very high sequence similarity) that are non-specifically recognised by M3. Therefore, our data in conjunction with the body of published work showing binding of M3 to collagens I, II, III and IV suggest it is highly likely that emm3 streptococci will indeed bind to many if not all types of collagen in the same manner. Whether this means all collagen types, in the various tissues where they occur, are targeted by emm3 streptococci is a very interesting question, however one that goes beyond the scope of our study.