Parallel Evolution of Pseudomonas aeruginosa during a Prolonged ICU-Infection Outbreak

ABSTRACT Most knowledge about Pseudomonas aeruginosa pathoadaptation is derived from studies on airway colonization in cystic fibrosis; little is known about adaptation in acute settings. P. aeruginosa frequently affects burned patients and the burn wound niche has distinct properties that likely influence pathoadaptation. This study aimed to genetically and phenotypically characterize P. aeruginosa isolates collected during an outbreak of infection in a burn intensive care unit (ICU). Sequencing reads from 58 isolates of ST1076 P. aeruginosa taken from 23 patients were independently mapped to a complete reference genome for the lineage (H25338); genetic differences were identified and were used to define the population structure. Comparative genomic analysis at single-nucleotide resolution identified pathoadaptive genes that evolved multiple, independent mutations. Three key phenotypic assays (growth performance, motility, carbapenem resistance) were performed to complement the genetic analysis for 47 unique isolates. Population structure for the ST1076 lineage revealed 11 evolutionary sublineages. Fifteen pathoadaptive genes evolved mutations in at least two sublineages. The most prominent functional classes affected were transcription/two-component regulatory systems, and chemotaxis/motility and attachment. The most frequently mutated gene was oprD, which codes for outer membrane porin involved in uptake of carbapenems. Reduced growth performance and motility were found to be adaptive phenotypic traits, as was high level of carbapenem resistance, which correlated with higher carbapenem consumption during the outbreak. Multiple prominent linages evolved each of the three traits in parallel providing evidence that they afford a fitness advantage for P. aeruginosa in the context of human burn infection. IMPORTANCE Pseudomonas aeruginosa is a Gram-negative pathogen causing infections in acutely burned patients. The precise mechanisms required for the establishment of infection in the burn setting, and adaptive traits underpinning prolonged outbreaks are not known. We have assessed genotypic data from 58 independent P. aeruginosa isolates taken from a single lineage that was responsible for an outbreak of infection in a burn ICU that lasted for almost 2.5 years and affected 23 patients. We identified a core set of 15 genes that we predict to control pathoadaptive traits in the burn infection based on the frequency with which independent mutations evolved. We combined the genotypic data with phenotypic data (growth performance, motility, antibiotic resistance) and clinical data (antibiotic consumption) to identify adaptive phenotypes that emerged in parallel. High-level carbapenem resistance evolved rapidly, and frequently, in response to high clinical demand for this antibiotic class during the outbreak.

1. Materials and methods (lines 131-132). More information should be provided on how the phylogenetic trees were generated. The software used to generate the trees is listed, but information should also be given about the type of trees that were made (distance method?). 2. Materials and methods (lines 150-153). It is stated that oprD mutations were determined by PCR followed by Sanger sequencing. Why were the whole-genome sequences not used for this purpose? 3. Results: lines 184-188. Nonsynonymous vs synonymous mutations are provided in Table S2, but a discussion of how many SNPs were synonymous and how many were nonsynonymous should be provided in the Results section. Only nonsynonymous SNPs should be included in the subsequent discussions of genes that were mutated and potentially pathoadaptive, as synonymous SNPs would not be expected to have an effect on gene products. 4. Fig. 1: Some "core" isolates, from which all the other isolates were hypothesized to evolve from, were cultured several months after several of their supposed descendants. Some discussion of how this may have occurred should be provided. 5. . Likewise, it should be clarified whether "pathoadaptive" genes were defined as those that were mutated at least twice (as stated in the text) or were mutated at least twice with nonsynonymous changes, which would be required if the genes are truly pathoadaptive. 6. Fig. S3B. An explanation for the two pairs of isolates with relatively low pairwise identity should be provided. If all these isolates shared a single ancestor, why are these two pairs relatively dissimilar? 7. Fig. 5. In panel A, this reviewer is not familiar with the terms "sense" and "nonsense" being used in this way. In panel B, different colors should be used for the different lines to make it easier to distinguish them.
Reviewer #2 (Comments for the Author): Interesting work analyzing the evolution of Pseudomonas aeruginosa during a prolonged outbreak in burn ICU. There are however some comments for the authors consideration: 1. Lines 140-144. Why were only the MICs of imipenem and meropenem determined? Would be informative to add other relevant antipseudomonal agents.
2. Related to this, other relevant mutations related to effux pumps expression (mexR, mexZ, mexS) or AmpC expression (dacB, mpl, ampR) are seen and should be interpreted since many of the mutated genes are related to the P. aeruginosa mutational resistome (see for instance López-Causapé C front Microbiol 2018).
3. Lines 240-245. The interpretation of EUCAST breakpoints and definitions is incorrect, particularly "intermediately susceptible" shall not be used since it is misleading. 4. One relevant phenotype that could/should be studied in the context of adaptation in burn wound infections is the capacity to form biofilms. 5. It might be informative to discuss your results together with those of Wheatley R Nat Commun 2021 Staff Comments:

Preparing Revision Guidelines
To submit your modified manuscript, log onto the eJP submission site at https://spectrum.msubmit.net/cgi-bin/main.plex. Go to Author Tasks and click the appropriate manuscript title to begin the revision process. The information that you entered when you first submitted the paper will be displayed. Please update the information as necessary. Here are a few examples of required updates that authors must address: • Point-by-point responses to the issues raised by the reviewers in a file named "Response to Reviewers," NOT IN YOUR COVER LETTER.
• Upload a compare copy of the manuscript (without figures) as a "Marked-Up Manuscript" file. • Each figure must be uploaded as a separate file, and any multipanel figures must be assembled into one file. For complete guidelines on revision requirements, please see the journal Submission and Review Process requirements at https://journals.asm.org/journal/Spectrum/submission-review-process. Submissions of a paper that does not conform to Microbiology Spectrum guidelines will delay acceptance of your manuscript. " Please return the manuscript within 60 days; if you cannot complete the modification within this time period, please contact me. If you do not wish to modify the manuscript and prefer to submit it to another journal, please notify me of your decision immediately so that the manuscript may be formally withdrawn from consideration by Microbiology Spectrum.
If your manuscript is accepted for publication, you will be contacted separately about payment when the proofs are issued; please follow the instructions in that e-mail. Arrangements for payment must be made before your article is published. For a complete list of Publication Fees, including supplemental material costs, please visit our website.
Corresponding authors may join or renew ASM membership to obtain discounts on publication fees. Need to upgrade your membership level? Please contact Customer Service at Service@asmusa.org.
Thank you for submitting your paper to Microbiology Spectrum. We would like to thank the reviewers for their positive comments and helpful suggestions. We have performed additional experiments and included more clinical data as requested. Below, we have included a point-by-point response to each comment.

Reviewer #1
The manuscript describes detailed genomic and phenotypic analyses of 58 ST1076 Pseudomonas aeruginosa isolates from an outbreak in the burn unit of a single medical center. A total of 16 genes were found to have acquired mutations independently (i.e., in at least two lineages from the outbreak), suggesting that they were pathoadaptive (conferred a fitness advantage). The most frequently mutated gene was oprD, and these mutations were shown to occur in the context of frequent carbapenem use in the ICU. Reduced growth rate and reduced motility were also frequently observed. The manuscript is clearly and concisely written and presents interesting findings on how P. aeruginosa evolves over the course of an outbreak. A few clarifications and modifications would increase its significance.
We appreciate the positive response of reviewer #1.

Materials and methods (lines 131-132).
More information should be provided on how the phylogenetic trees were generated. The software used to generate the trees is listed, but information should also be given about the type of trees that were made (distance method?).
More information about the type of tree generated has been added to the Material and methods section (lines 126-128).

Materials and methods (lines 150-153)
. It is stated that oprD mutations were determined by PCR followed by Sanger sequencing. Why were the wholegenome sequences not used for this purpose?
Mutations in oprD were originally called using whole-genome sequencing data and then validated by PCR and Sanger sequencing.
3. Results: lines 184-188. Nonsynonymous vs synonymous mutations are provided in Table S2, but a discussion of how many SNPs were synonymous and how many were nonsynonymous should be provided in the Results section. Only nonsynonymous SNPs should be included in the subsequent discussions of genes that were mutated and potentially pathoadaptive, as synonymous SNPs would not be expected to have an effect on gene products.
We agree with the reviewer. The number of SNPs that were synonymous and nonsynonymous has been added to the Results section accordingly (lines 180-183). Fig. 1: Some "core" isolates, from which all the other isolates were hypothesized to evolve from, were cultured several months after several of their supposed descendants. Some discussion of how this may have occurred should be provided.

4.
The bacterial isolates were persisting on surfaces in the ICU throughout the outbreak. It is possible that by the time each respective isolate was collected, independent, niche specific evolution had already occurred. Some discussion has been added to the text (lines: 190-194). [194][195]. Likewise, it should be clarified whether "pathoadaptive" genes were defined as those that were mutated at least twice (as stated in the text) or were mutated at least twice with nonsynonymous changes, which would be required if the genes are truly pathoadaptive.

Lines
We agree with this point and we have modified the definition of pathoadaptive genes as suggested by the reviewer (ie., mutated with nonsynonymous changes at least twice independently). One of the 16 genes classified originally as pathoadaptive (rhlR) did not fulfil the criteria of the new definition. Therefore, both the text and the Fig. 2 have been modified accordingly.
6. Fig. S3B. An explanation for the two pairs of isolates with relatively low pairwise identity should be provided. If all these isolates shared a single ancestor, why are these two pairs relatively dissimilar?
The mean pairwise identity provided in Fig. S3B is a measure of the quality of the reads and the quality of the mapping. Therefore, it does not reflect the similarity of the isolates. 7. Fig. 5. In panel A, this reviewer is not familiar with the terms "sense" and "nonsense" being used in this way. In panel B, different colors should be used for the different lines to make it easier to distinguish them.
The terms "sense" and "nonsense" have been replaced in the figure legend. Similarly, the colors for the different lines in panel B have been changed as requested.

Reviewer #2
Interesting work analyzing the evolution of Pseudomonas aeruginosa during a prolonged outbreak in burn ICU. There are however some comments for the authors' consideration: We thank reviewer #2 for the interest on our work.
1. Lines 140-144. Why were only the MICs of imipenem and meropenem determined? Would be informative to add other relevant antipseudomonal agents.
We originally focused on the MICs of imipenem and meropenem due to the striking genetic signature at the oprD locus. In response to the reviewer's comment, we obtained the antibiotic susceptibility profiles for other relevant anti-Pseudomonas agents, including penicillins, cephalosporins, quinolones and colistin. Data collected showed emergence of resistance also against other antibiotic classes. This information has been added to the Supplemental material (Fig. S4) and some commentary added to the text (248-251).
2. Related to this, other relevant mutations related to effux pumps expression (mexR, mexZ, mexS) or AmpC expression (dacB, mpl, ampR) are seen and should be interpreted since many of the mutated genes are related to the P. aeruginosa mutational resistome (see for instance López-Causapé C front Microbiol 2018).
We agree with the reviewer. We added some text and also the reference (lines 248-251).
3. Lines 240-245. The interpretation of EUCAST breakpoints and definitions is incorrect, particularly "intermediately susceptible" shall not be used since it is misleading.
We thank the reviewer for pointing out this important correction. Isolates that we previously characterized as "intermediately susceptible" have been adapted according to the new definition of "I" (i.e., "susceptible increased exposure") throughout the text and in Fig. 5. 4. One relevant phenotype that could/should be studied in the context of adaptation in burn wound infections is the capacity to form biofilms. This is certainly an interesting point -In response to the comment, we evaluated biofilm formation for each of the 47 unique P. aeruginosa isolates. Biofilm capacity was assessed using a microtiter plate assay with crystal violet staining (1). The assay was performed in technical duplicates and in biological triplicates. To determine the differences in biofilm capacity, we compared the biomass produced by the core isolate (ie., H26076) with the biomass produced by each unique isolate in the collection. Our data did not reveal a statistically significant difference in biofilm forming capacity for any of the isolates (p ≥ 0.9899 for each comparison, using Kruskal Wallis with Dunn's multiple comparison test for non-normally distributed data). Indeed, the methodological approach has well known limitations. We feel that the biofilm formation capacity of the isolates may warrant further investigations using more elegant methods (2), however this falls outside the scope of the current report.