Écrire un commentaire

The manuscript by Borde et al. discusses an important bacterial adaptation. It reveals how the type I ABC transporter YhdWXYZ exhibits functional plasticity linked to its substrate-binding protein (SBP), YhdW. The authors report that in Escherichia coli K-12, lacking a full-length YhdW, deletion of the transporter genes increases susceptibility to antibiotics lomefloxacin and mecillinam, pointing to a role in antibiotic homeostasis. The same deletion in Citrobacter rodentium, which encoding a transporter system with a complete SBP protein, does not alter antibiotic susceptibility. Purified C. rodentium YhdW binds asparagine with low μM affinity, yet, in vivo, the transporter does not appear to be a primary asparagine importer under the laboratory conditions tested. In mice, YhdWXYZ contributes to early colonization and persistence, as the deletion mutants showed reduced bacterial loads and altered intestinal pathology.

Together, the data supports the idea that the function of a transporter can diversify even within a conserved system. It combines multidisciplinary approaches, including comparative genomics, bacterial genetics, reporter assays, HPLC-based drug quantification, biochemical characterization of YhdW and mouse infection with histological analysis. The main limitations are the indirect evidence for the role of the SBP, the lack of direct transport analyses, some ambiguity in the intracellular lomefloxacin quantification, and the lack of complementation for the in vivo data. Addressing these points would further strengthen an already interesting piece of work.

Major Points

● Role of the substrate-binding protein in the functional shift

A central conclusion of the paper is that loss of a functional SBP shifts YhdWXYZ toward antibiotic homeostasis. This is an appealing idea, but the evidence is indirect. It rests on comparing E. coli K-12 with C. rodentium, which differs in species background and in the lomefloxacin concentration used (0.075 vs 0.25 µg/mL). Therefore, factors other than the SBP could explain the different phenotypes. Several experiments could make the link more direct. Restoring a full-length yhdW (such as the one from C. rodentium or from an E. coli strain like O157:H7 Sakai) in E.coli K-12 would help address whether the SBP is what changes the transporter's role. If the SBP drives the shift, the phenotype of the ΔyhdWXYZ mutant might be expected to change in the restored strain. For E. coli K-12, the authors screened a panel of antibiotics, such as tetracycline, puromycin, lomefloxacin, mecillinam, and ampicillin, that likely have quite different mechanisms of action. However, C.rodentium was only tested in presence of lomefloxacin and tetracycline. A similar broad antibiotic screen in C. rodentium could help rule out effects that the two antibiotics tested may not reveal. Together these analyses would strengthen the main claim and better support the conclusion that the transporter is not involved in antibiotic susceptibility in this species.

● Intracellular lomefloxacin quantification Figure 3A shows considerable variability at the later time points. The authors suggest that this likely reflects higher cell mortality in the mutant population. It would be helpful to clarify whether the per-cell values were normalized to CFU, since this could affect the interpretation if mutant viability drops. The figure legend also mentions sampling every 10 min, while the plot and Methods list 10, 20, 30, and 45 min, so aligning these descriptions would make the experiment easier to follow. Adding more time points might also reduce the variability and strengthen the conclusion that deletion of yhdWXYZ promotes lomefloxacin accumulation. It would be interesting to see whether deleting yhdW alone has the same effect, since a ΔyhdW strain is already available from the infection experiments.

● Complementation of the in vivo phenotype

The infection experiments are central to the manuscript's conclusion that YhdWXYZ contributes to colonization and persistence. Unfortunately, the phenotype is not complemented, and the strains were not co-infected. Fecal loads of the mutants are lower than wild type mainly at early time points. The histological comparisons rest on small groups. Reintroducing yhdW and yhdWXYZ in the respective mutants would help confirm that the effect is due to the deleted genes and not to secondary changes, such as polar effects on neighboring genes or second-site mutations introduced during strain construction. A competitive index experiment, in which wild type and mutant are co-infected at a 1:1 ratio, may also be a sensitive way to detect small fitness differences. The authors may also consider moderating the title and abstract wording on host colonization if complementation is not feasible, since this would match the strength of the current evidence.

● Sequence identities The sequence identities reported in the section "YhdWXYZ is not involved in antibiotic susceptibility in C. rodentium" may not capture the full picture. The authors nicely present the distribution and genetic variation of the yhdWXYZ genes across Enterobacteriaceae. Yet Figure 1B shows only whether each gene is present, absent, pseudogenized, or interrupted by an insertion sequence. A sequence alignment of the homologs discussed would be a helpful addition, particularly for the protein species mentioned in the text, such as C. rodentium YhdW (used experimentally), BztA from B. ovis, AapJ from R. leguminosarum, and the related systems from B. abortus. It would also be helpful to see the identity for YhdW, since it is the most relevant component to the proposed functional shift. It would be especially interesting to see how the transmembrane domains encoded by yhdX and yhdY have diverged in lineages that lost a functional SBP, since this might reveal changes in specificity and selectivity. Such an analysis could further support the proposed diversification of the transporter's function and might help visualize some of the common patterns mentioned in the Discussion.

Minor Points

● Only female mice were used in the infection experiments, and the manuscript does not mention the reasoning. It would improve clarity to briefly explain this choice in methods. If feasible, including a male group could also help rule out sex-dependent effects on the results.

● PAS (periodic acid-Schiff) staining is listed in the Methods as one of the approaches used on the colon samples, but the results are not shown or discussed. H&E reveals inflammation and crypt elongation, whereas PAS would add information about the mucus barrier and goblet cells, such as whether the epithelium has lost some of its secretory, protective cells. This could help explain the differences in colonization and persistence between the wild type and the mutants. Including these results, or noting why they were omitted, would further support the proposed contribution of YhdWXYZ to infection.

● Some field-specific abbreviations, such as EPEC, EHEC, and LEE, are not defined. Spelling them out at first use (enteropathogenic E. coli, enterohemorrhagic E. coli, and locus of enterocyte effacement) would make the text easier to follow for readers from outside the field and help place the results in context.

● The Discussion raises the possibility that YhdWXYZ is capable of bidirectional transport, but this idea rests solely on homology to the Aap system from R. leguminosarum. The authors might consider phrasing it more cautiously or, if they prefer, leaving it out. Either option would keep the focus on the paper's main message, which is how strikingly homologous systems can differ.

● Figure 6 and the text describing it are not fully consistent. First, the text refers to Figure 5C where Figure 6C seems to be meant. Second, the competition assay in Figure 6C shows time points of 3, 6, and 24 h, while the text mentions 2, 6, and 24 h. Third, the reference to Figure 5A for growth in minimal medium seems to refer to Figure 6A. Aligning the figure and the text would make the information easier to follow.

● The growth curves in Figure 6A and 6B are shown for a single representative replicate, and for 6B this reflects only two independent experiments (n = 2). Figures S1 and S2 present similar data as the mean and SD of all replicates, so a comparable presentation here could work well. If that is not feasible, the individual curves could be added as a supplementary figure. This would help readers see the reproducibility of the data.

● In the fluorescence thermal shift assay (FTSA) screen in Figure 5A, asparagine produced the largest increase in melting temperature, but cysteine, glutamine, and serine also showed notable shifts. Some of these amino acids were followed up in dose-dependent FTSA or isothermal titration calorimetry (ITC) experiments. The authors might consider expanding the discussion of why these amino acids stood out in the screen, and how they chose the ligands for ITC. Since glutamine and serine also shifted in FTSA, it would be interesting to see them tested by ITC alongside asparagine, valine, and cysteine. It would also be helpful to include the ITC isotherms for all experiments as a supplementary figure, so readers can see the fit quality and reproducibility across the replicates.

● For Figure 5B, the text states that other amino acids showed minimal or no effect in the secondary screen, while the corresponding panel shows a significant increase in Tm for cysteine, so the wording could be adjusted to reflect this. Additionally, the x-axis of Figure 5B is labeled in nM, whereas the legend gives mM, so aligning the two would avoid confusion. Overall, these modifications would improve the clarity.

● Figure 4 examines the sensitivity of C. rodentium to lomefloxacin and tetracycline. For E. coli K-12, the authors screened a broader panel (tetracycline, puromycin, lomefloxacin, mecillinam, and ampicillin), which covers antibiotics with quite different mechanisms of action. A similar screen in C. rodentium could help rule out effects that the two antibiotics tested may not reveal and would strengthen the conclusion that the transporter is not involved in antibiotic susceptibility in this species. Mecillinam might be an especially informative addition, since it was one of the two hits in E. coli.

● In E. coli K-12, a frameshift in yhdW causes premature termination of translation. This produces a 78-amino-acid protein that can no longer act as a substrate-binding protein (SBP). The authors report that this product shows no homology to other ORFs, but they do not speculate on whether the shorter protein might have a new function. A predicted structure of the truncated protein could be used to search for similarly folded proteins that are not SBPs, which might offer clues. This could open interesting new avenues and further increase the impact of the work.

● The authors show that deletion of yhdWXYZ increases susceptibility to lomefloxacin and mecillinam. It would be interesting to hear their thoughts on why these two antibiotics, and not the three others tested (tetracycline, puromycin, and ampicillin), led to a significant reduction in mutant viability. A short discussion of the antibiotics' features, such as their targets, mechanisms of action, or physicochemical properties, could help explain why different antibiotic families affect this strain differently.

● The discussion of the work's impact could be extended a little further. The authors connect their findings to the adaptability of bacterial transport systems in antibiotic resistance and host-pathogen interactions. Including some broader implications, such as potential pharmaceutical uses of a changed physiological role in an ABC transporter, would help underscore why these findings matter to the field.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author used a generative AI tool (Claude) solely for final language editing to improve the flow and readability of the review. The scientific content, analysis, and suggestions are the author's own, and the author reviewed all edited text.

I reviewed this preprint as part of the Transparent and Accountable Peer Review training program. This report was written with the support of my research mentor.

Competing interests

The author declares that they have no competing interests.

Use of Artificial Intelligence (AI)

The author declares that they did not use generative AI to come up with new ideas for their review.