Environmental pH and peptide signaling control virulence of Streptococcus pyogenes via a quorum-sensing pathway

Bacteria control gene expression in concert with their population density by a process called quorum sensing, which is modulated by bacterial chemical signals and environmental factors. In the human pathogen Streptococcus pyogenes, production of secreted virulence factor SpeB is controlled by a quorum-sensing pathway and environmental pH. The quorum-sensing pathway consists of a secreted leaderless peptide signal (SIP), and its cognate receptor RopB. Here, we report that the SIP quorum-sensing pathway has a pH-sensing mechanism operative through a pH-sensitive histidine switch located at the base of the SIP-binding pocket of RopB. Environmental acidification induces protonation of His144 and reorganization of hydrogen bonding networks in RopB, which facilitates SIP recognition. The convergence of two disparate signals in the SIP signaling pathway results in induction of SpeB production and increased bacterial virulence. Our findings provide a model for investigating analogous crosstalk in other microorganisms.

1. Did the authors bioinformatically examine other quorum sensing regulators (particularly if structure is known) for the presence of similar key motifs (histidine switch). This might be an informative aspect to include in the discussion of this work. Figure 1E, were the WT GAS present at significantly higher CFU loads compared to speB mutant GAS? At either/both pH?

For
3. Figure 2F. I had some initial difficulty interpreting the differences between extracellular and intracellular pH. It might be clearer for the reader if you use only 1 decimal point for the intracellular pH (same as extracellular) or you plot out as a bar graph presentation? What do you think is going on at pH8, where there is also a 0.3 pH change? Buffering capacity of the cell kicks in?
4. Figure 4G. The speB transcript levels were measured at 24 h and SIP L6A was no different to other mutants, and very distinct from WT. Do the authors think that if this experiment were conducted at a later timepoint (48 h or 72 h) that there would be a clear difference?
5. I would expect that the authors would have considered this point, but was there any attempt made to undertake the crystallisation study (RopB-CTD + SIP) at lower pH to directly visualize structural changes? 6. Bottom of page 15. The authors mention that the Y182 mutation did not result in protein misfolding. How was this accomplished? Was CD spectrometry (or other technique) undertaken to confirm? 7. For virulence experiments ( Figure 4E and 6I), can the authors provide information in the Figure legends on the individual dose for each WT and mutant represented in the figure panels?
Finally, the authors work is exciting and novel. This paper provides a detailed model for the regulatory control of an important GAS virulence determinant.

Mark Walker
Reviewer #2 (Remarks to the Author): In this manuscript, Do and colleagues explore how pH influences SpeB production via a quorum sensing (QS) peptide (SIP) interaction with the global regulator RopB. Notably, pH has been shown to control other QS pathways in other bacteria but this is the first study to discover specific molecular mechanisms that help explain this phenomenon. This study is detailed and happily, many times I asked an experimental question the authors went on to answer it. The real power in the data is that it includes genetic, phenotypic, biochemical and in vivo demonstrations of pH regulation of SpeB production (via the SIP-RopB pathway). As mentioned before, there has been no obvious mechanism for pH-dependent control of QS in bacteria and this paper should spur others in the field to examine pH control of QS in other bacteria. My comments on the paper can be found below: In some cases, statistics (as shown with *, etc) are used, and in other cases they are not. Examples where statistics may help follow: Figure 1B, Figure 2A, Figure 3D and 3F. Consistency would help the reader determine what is significant and what is not. Figure S4B: The affinity for SIP binding to RopB is obviously increased at lower pH (reduced Kd). The dynamic range (total change in mP) is reduced with pH. Can the authors explain why this has happened?

Minor comments
The line "Each component of the SIP signaling pathway must be functional for a wild-type virulence phenotype" is repeated on page 4. I appreciate that the authors have explored in great detail pH mediated regulation of SIP-RopB and have an excellent working model (Fig. 7). SpeB production is very complex and there is a limit to how much detail can be provided (a few excellent reviews are cited). However, do the authors believe that pH could influence other areas of the pathway e.g. the endopeptidase PepO, or CovRS?
Reviewer #3 (Remarks to the Author): This review covers mainly the structural biology portion of the study. The authors previously determined the crystal structure of apo RopB. Here, they report the structure of RopB bound to its activating peptide SIP. The proposed model of RopB activation is moderately supported by the structural data: -The proposed acidity-induced conformational change mediated by the histidine switch is not actually observed in this structure. -The activating peptide seems to be bound to a low-binding-affinity conformation of the protein.
-The molecular mechanism by which SIP binding promotes RopB binding to DNA is still unknown, as the DNA-binding domain of the protein is absent from the structure. On the other hand, the proposed model is more strongly supported by the accompanying biochemical and biological data. Several issues need to be addressed: # 1 -There seems to be a major error in the built structure of the protein (as well as in the authors' previously published structure of apo RopB in the same crystal form, PDB 5DL2). Two chains of RopB are found in the asymmetric unit (ASU), A and B. They are almost completely built but both are missing residue 180. The segments 159-179 and 181-200 form antiparallel helices, and in the current model, residue 180 appears to form a loop between the two helices. This helixloop-helix segment is located very close to its crystallographic symmetry mate from the adjacent ASU (which also contains two chains of RopB, A' and B'). There is no electron density at the implied position of residue 180. However, there is clear continuous electron density from the helix 159-179 leading into helix 181'-200' of the adjacent ASU, and vice versa from helix 159'-179' of the adjacent ASU leading into helix 181-200. Thus, residues 159-200 form one long continuous helix, and half of the protein chain was built into the wrong ASU. Attached are pictures of the original and corrected arrangements. One full RopB chain is in green and its crystallographic symmetry mate, in blue. Residues 179 and 181 are indicated by small and large spheres, respectively. This analysis applies to chain B as well. The model should be rearranged to correct this, or an explanation should be provided. This potential error actually has little impact on the structural analysis, but it would imply that the SIP binding site and the histidine switch are composed of residues from both copies of the crystallographic dimer. If the authors do apply this correction, they should also consider correcting their previously reported apo RopB structure.
# 2 -There are a few Ramachandran outliers or borderline values in the backbone angles of the SIP structure. Notably, a cis peptide bond was modeled at position 6-7. As the electron density for the peptide is not clearly defined at this resolution, this should very likely be corrected to a trans bond.
# 3 - Figure 3A: Do the authors mean "composite omit electron density map"? Could they verify that this is indeed a 2fo-fc map displayed at the 1-sigma level? The map file that they provided, although being a refinement output map and not a composite omit one, shows weak fragmented density for the peptide at the 1-sigma level, and only shows a similar density to Figure 3A when contoured at 0.5sigma.
# 4 -Section Structural Basis of SIP Recognition by RopB: -"stabilizing polar contacts between the side chains of N152 of helix a6, N192 of helix a8, and K278 of helix a12 of RopB and the peptide backbone of SIP" The side chains of N152 and K278 are quite far from the peptide backbone, as seen in Figure 3B and in the coordinates file provided. Rather, the hydrogen bonds between RopB and SIP involve the side-chains of Y224 and N192, as listed correctly in Figure S6A. All the residues mentioned throughout this paragraph should be double-checked and Figure 3B modified accordingly.
-"hydrophobic contacts between the side chains of SIP amino acids and the side chains of F155 of helix a6, I195 of helix a8, and M267, F268, and Y271 of helix a12" V191 also establishes hydrophobic contacts with the peptide.
# 5 - Figures 4 and S7, and the related sections in the Results: One of the mutants analyzed is N152A. However as mentioned in #4, N152 is quite far from the peptide (over 5 Angstroms from the L5 side chain and even further from the SIP backbone, as seen in Figure 3B and in the coordinates file provided). The authors should explain this or remove this mutant from the Results. # 6 -Section Structural Basis of SIP Recognition by RopB: "However, no direct interaction was observed between the side chain of L6 of SIP (SIP-L6) and RopB. Together, these data indicate that each amino acid of SIP except the side chain of L6 is required for sequence-specific SIP recognition by RopB." Residues 1-4 of SIP establish few contacts with RopB, and those are weak non-specific Van der Waals interactions between hydrophobic side chains of the peptide and polar or charged residues of the protein. The structure alone does not seem to explain the importance of this portion of SIP for its recognition by RopB, although the biochemical and biological data do support the importance of all these residues. The binding affinity of SIP to RopB is around 10nM, and although the peptide was present at relatively high concentration during crystallization, its average crystallographic B-factor is two times higher than the protein's, indicating conformation heterogeneity or partial occupancy. Also, SIP is absent from the second RopB copy in the asymmetric unit. The authors propose a pH-induced conformational change in the protein that increases its affinity for the peptide. Based on all these points, have they considered the possibility that the current structure (at neutral pH) may represent a low-binding-affinity conformation in which RopB can still associate with SIP but without forming many of the specific interactions implied by their biochemical data that would result in nanomolar affinity? The authors should discuss this point if they deem it relevant. # 7 -Related to #6, there should be a figure, at least in the supplementary, comparing the overall structure (for example in ribbon form) of the SIP-bound chain A crystallographic dimer with that of chain B and with the dimers from the previously published apo RopB structure. This would illustrate that the conformations of apo RopB and SIP-bound RopB from this neutral-pH structure are almost indentical.
# 8 -Have the authors considered a potential role for residue E185 in the histidine switch? It is located close to H144, is highly conserved and could form a salt bridge with the latter at moderately acidic pH. The pH dependency of the interaction between H144 with Y182 is unclear, as tyrosine is uncharged and can act both as hydrogen bond donor and acceptor, and can participate in aromatic stacking. The data does implicate Y182, but there may be a larger conformational change at acidic pH, than only rotation of Y182 towards H144 or vice versa. In addition, if the authors implement the correction suggested in #1, this would mean that H144 interacts with a residue (Y182 or maybe E185) from the second subunit of the dimer, with possible implications in dimer formation or stability for the histidine switch.
# 9 -As the main novelty of the structure is the complex between RopB and SIP, it could be interesting to compare it to related proteins. For example, the TPR-containing, quorum-sensing regulator PlcR also undergoes a conformational change upon binding to a signaling peptide (PMID 23277548), albeit not pH-induced. Such a comparison could give clues as to how SIP binding to RopB stimulates DNA binding by the protein.
# 10 -Minor points: -In the third paragraph of the introduction, this sentence is duplicated: "Each component of the SIP signaling pathway must be functional for a wild-type virulence phenotype" -The Materials and Methods section is not very informative and could be re-written. Instead of listing every type of experiment followed by "details are provided in the supplementary", the authors should mention the key points here, and have one introductory sentence referring to the supplementary for the detailed Materials and Methods. - Figure 3B: the gray side chains should have their oxygen and nitrogen atoms colored in red and blue, like the other residues in this panel.
- Figure 5A: the line connecting the base of the SIP-binding pocket of the two subunits does not stand out against the background of the rest of the image. The base of the pocket could be marked in some other way. Also, the residues forming the histidine switch, displayed as all-atom spheres, are difficult to distinguish from each other. They could be shown as sticks, or maybe one sphere per residue. - Figure S8A: the source of the full-length RopB model is not indicated. Is it from a previous publication? If not, details should be provided in the supplementary methods.
- Table S3: the Wilson B-factor for the dataset should be provided in the Data Collection section.
-At several places in the manuscript, the authors refer to the pH-induced conformational change in RopB. As this rearrangement was not actually observed in the present structure, it should be specified as putative / proposed / probable / likely… Alexei Gorelik, McGill University