Identification of Receptor Binding Proteins of Yersinia Phage φR1-37 and Enterocoliticin That Use the Same Bacterial Surface Receptor

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript extensively describes a tailocin of Yersinia enterocolitica O:3 (enterocoliticin) from sequence to function, as well as identification of the receptor binding site (RBP) and a pentasaccharide from the receptor LPS, and their complex. Similarly, some aspects of the biology of the bacteriophage φR1-37 were also treated, especially concerning the RBP site. This study is well performed and the conclusions are convincing. 

Only a small part of the work is puzzling me. In the docking experiment between the pentasaccharide and the RBP, the authors eliminated the first solution (based on the best free energy of association) for the second saying "However, for biological plausibility of the Orf39 trimeric interface, the slightly lower-scoring Mode 2 deltaG = −6.273 kcal/mol) was selected as the representative binding conformation." Some more information should be given concerning the non-preferred first solution. 

Finally, the coordinates of the RBP/pentasccharide complex should be deposited, e.g. in Zenodo or other depository. 

Author Response

See the Reply to reviewers PDF

Author Response File: Author Response.pdf

Reviewer 2 Report

Comments and Suggestions for Authors

This manuscript describes the identification and characterization of a tailocin (enterocoliticin) from Yersinia enterocolitica strain 29930, identifies Gp298 as the receptor-binding protein (RBP) of phage φR1-37, and proposes an 80 residue conserved motif shared between Gp298 and the tailocin Orf39 as a putative carbohydrate-binding site. The experimental work, including tailocin purification, mass spectrometry, adsorption and blocking assays, and EOP measurements, is technically sound and provides useful information on phage and tailocin host interactions. This work is timely - there is significant interest in using phages or phage-derived entities like tailocins for treating antibiotic resistant infections.

A major limitation of the mechanistic interpretation is the absence of experimental validation of the proposed carbohydrate-binding motif. While the modelling and docking analyses are suggestive, the manuscript does not include functional tests demonstrating that the 80-85 aa region is important for receptor binding. The glycan docking is highly speculative, as it relies on predicted protein structures and modelled oligosaccharides without experimental validation. The problem is that the manuscript describes the docking results in mechanistic terms (e.g. specific hydrogen bonds and energetic contributions) that are not supported by any experimental evidence. Targeted point mutagenesis of residues predicted to contact the glycan, truncation or domain-swapping experiments to test sufficiency of the motif, simple glycan competition assays, or direct binding measurements (e.g. SPR, ITC) using purified Gp298 and defined oligosaccharides or LPS would really strengthen the central claim. At minimum, the authors should clearly frame the proposed binding mechanism as a hypothesis and remove detailed discussion of interactions and energetics.

The authors interpret the highly conserved 80 amino acid region shared between Gp298 and Orf39 as the receptor-binding pocket responsible for carbohydrate specificity. However, BLAST searches showed that this region is broadly conserved across phages infecting a wide range of bacterial genera that possess highly diverse outer membrane and LPS architectures. The extensive conservation across hosts with markedly different cell envelope compositions is difficult to reconcile with a primary role as a specificity-determining carbohydrate-binding pocket, which would typically be expected to diverge in accordance with host glycan diversity. The authors need to address how they think receptor specificity is encoded within such a conserved motif or temper claims regarding specificity.

Table 3 shows that phage R1-37 doesn’t plate on strains with serotypes 25,26,44 and O5. Yet the authors test these strains in the blocking assay (Figure 6). This doesn’t make sense – if the phages are unable to form plaques on these strains you can’t tell if they are blocked at the cell surface or not.

Given that the complete genome sequence of Y. enterocolitica 29930 is available and that the Cos141 insert is identical to the corresponding chromosomal locus, the paper could be greatly simplified and made easier to follow by simply using the bacterial genome (GenBank) annotation rather than the cosmid ORF numbering. While Table S4 provides a detailed annotation of the cosmid genes, it would be more informative (and better for long term continuity of this work) to present the corresponding chromosomal locus tags and gene names from the 29930 genome. In addition, the detailed discussion of the cosmid identification and annotation could be shortened or removed, as it adds length but no insight beyond the genome-level sequence.

The statement about tailocins lacking head assembly genes (line 567–569) needs a citation. Also note that many tailocins have associated lysis genes. It would be helpful to compare with R- and F-pyocins in Pseudomonas and cite better-characterized systems to orient the reader.

Edman degradation results are redundant given the later, more comprehensive MS data. The authors should consider removing the Edman section to streamline the narrative.

In Fig 1 the authors show enterocoliticin produced heterologously in E. coli. An image of particles produced in the native Yersinia strain (29930) should also be shown to corroborate the statement that they are the same dimensions (line 526).

I found the EOP experiments (starting at line 600) to be confusing and I had to read it several times to understand what was being done. It would be more straightforward to just report the phage titers on the various strains.

Figure 3 would benefit from clearer normalization of the adsorption data. Plotting absolute PFU values of unadsorbed phage makes the y-axis difficult to interpret without knowing the initial phage input and the differences are difficult to interpret as the axis goes from 0 to 5x10⁵ as the first marking. Expressing the data as percent adsorbed would improve clarity here. And are these differences statistically significant? 

The description of Gp298 homology to both T4 long (Gp37) and short (Gp12) tail fibres is confusing as written and will be difficult for a non-expert to understand. It would help to clarify that this reflects shared modular domains rather than functional equivalence to both fibre types.

Figure 5A does not appear to directly demonstrate failed trimerization of Gp298 expressed from pCP-1 (line 650). The basis for this conclusion should be clarified. Is it just lack of / low expression?  

The authors state that “analysis of the supernatant revealed a significant reduction in free phage particles in the presence of DB-coupled O:3 LPS (Figure 6A)” but no statistics are presented. Figure 6C similarly states significant differences without stats.

Author Response

See the Reply to reviewers PDF

Author Response File: Author Response.pdf

Round 2

Reviewer 2 Report

Comments and Suggestions for Authors

All of my previous concerns have been addressed. Congratulations to the authors on their work.