Evolution of lasR mutants in polymorphic Pseudomonas aeruginosa populations facilitates chronic infection of the lung

Chronic infection with the bacterial pathogen Pseudomonas aeruginosa often leads to coexistence of heterogeneous populations carrying diverse mutations. In particular, loss-of-function mutations affecting the quorum-sensing regulator LasR are often found in bacteria isolated from patients with lung chronic infection and cystic fibrosis. Here, we study the evolutionary dynamics of polymorphic P. aeruginosa populations using isolates longitudinally collected from patients with chronic obstructive pulmonary disease (COPD). We find that isolates deficient in production of different sharable extracellular products are sequentially selected in COPD airways, and lasR mutants appear to be selected first due to their quorum-sensing defects. Polymorphic populations including lasR mutants display survival advantages in animal models of infection and modulate immune responses. Our study sheds light on the multistage evolution of P. aeruginosa populations during their adaptation to host lungs.

Reviewer #1 (Remarks to the Author): This is an interesting manuscript by Zhao and colleagues tracking the emergence and behaviors of lasR mutants in people who have chronic P. aeruginosa infections.The most interesting observation is the idea that the presence of both WT and LasR mutant cells attenuates the pathogenicity of the bacteria in a way that cannot be explained by either type in a monoinfection (Fig 4).
Major issues: -The authors have chosen to interpret their data through the prism of cooperation and cheating in bacterial populations, focusing on public goods.They provide little evidence, if any, that is consistent with the idea that there is a social dilemma in the lungs of people with COPD.In fact, there is some evidence (eg, PMID: 24866798) that in the context of lung infections, lasR mutants are better adapted to the environment and are not cheaters.
-Consistent with this issue it is not clear how the experiments in Figure 3 are relevant to the evolution within the lung.That the mutations occur in a sequential way that stabilizes cooperation in laboratory conditions does not mean that this was the selective pressure for the mutations.
-The SNP data and the trees drawn suggest the possibility that some of the patients might have harbored multiple lineages of P. aeruginosa, which makes the analysis of mutational change over time rather difficult.

Minor points:
-The figures are hard to read and I had difficulty distinguishing between some of the colors (and figuring out what the control group was) -Did the authors adjust their statistical tests for multiple comparisons in the cases where there are multiple comparisons?-The introduction should be reorganized to introduce quorum sensing and QS regulation of extracellular products before a discussion of social cheating.
Reviewer #2 (Remarks to the Author): The authors collect longitudinal samples of P. aeruginosa from COPD patients and look at the evolution of social behaviors, with a primary focus on QS and pyoverdine production.They perform in vitro competition experiments and look at host colonization and immune response in a mouse model, and gene expression in human samples also subjected to mass spec.They infer that the sequential evolution of social cheats on different traits can stabilize a heterogeneous population and affect host response.
It is great to see this type of analyses on clinical isolates.However, I found the manuscript difficult to read, with many evolutionary statements that I don't understand, or that are wrong.Examples are (see also below): "modifying the adaptability" eg.l 85: we hypothesize that further evolution of the polymorphic P. aeruginosa population may mainly concentrate on modifying the adaptability of lasR mutants to facilitate the persistent colonization of the whole populationand l 233: while the adaptability of lasR mutants would be adjusted by the accumulation of other mutations during further evolution in COPD airways -what does this mean?130-134: These results suggested that QS-controlled extracellular proteases might be a vitally important public goods that could bring more direct benefits (in terms of nutrient acquisition) for the survival of P. aeruginosa in vivo, followed by siderophores, and the intraspecific competitions of which would be more readily to cause the sequential invasion of individuals deficient in producing the products.-unclear to me L.465-466: will be introduced to stabilize the population structure and to promote the population fitness during chronic lung infection.Selection doesn't work to stabilize population structure and population fitness.
The naming of isolates makes it difficult to remember what is what, and to see that not all isolates from the same patient are of the same clone type -and therefore not sequentially evolved from the same ancestor.Consider renaming strains to also include clone type, and perhaps even phenotype?I truly appreciate the challenges of working with clinical isolates.However, the generality of the findings is not clear: the sequential accumulation of mutants in first lasR and then pyoverdine is based on two patients, where selected isolates of one are analysed in more detail.Additionally, the isolates analysed in co-culture were not sampled at the same time and only assumed to be cooccurring.This is not entirely unreasonable as they were collected within a relatively short time frame, however, it needs to be stated explicitly.453: To achieve the above purpose, this may also be a direction of lasR mutant evolution ?455: and the evolution of multiple independent routes of lasR mutants in COPD can promote the colonization of multiple strains in different infection niches.S3: Why compare all isolates back to w1a when some are different clone types?In the case of transmission you could compare these with each other?801: when comparing two different clone types I don't think it makes sense to list the SNPs etc as accumulated mutations Figure 4: in the W1C and W1A mono-infections, a number of animals die between day 2-4 -could this impact the CFU counts in B? Figure 4B & C: I would argue that W1c alone does best -all hosts are alive and CFU counts are steady -so is it really a cheat in vivo?
Figure S1: what are the sample sizes at the different time points?In patient W, all isolates still produce pyoverdine?How do they grow in iron limited media?Zhao et al. have investigated social cheating of P. aeruginosa during chronic infection in the lungs of patients with COPD, specifically focusing on the role of lasR mutants.This study uses a great number of techniques to explore their research question, and the use of longitudinal samples from the same patients is really interesting.However, it would be strengthened by more in depth analysis of some of the data generated to provide more novel insight into COPD P. aeruginosa infections over time.

Major comments:
• It would be good to include a table with information for the mutations identified with WGSwhich genes are these in?Are all the numbers reported for non-synonymous mutations?Just because there are different numbers of SNPs/indels, it doesn't necessarily mean that these are more important without knowing the location and effect of these mutations.
• The quality of the genome assemblies for the isolates used as a reference to call SNPs for the later isolates needs to be reported.If these are a poor quality or the quality greatly differs between patient reference isolates, this could be a source of variation for the data.
• L155 -what is meant by mutations harmful to gene function?There is little information of the specific mutations found outside of lasR, I think more evidence is needed to make this statement.
• The phylogeny and groups are confusing -P.aeruginosa phylogenetic groups are well reported, with PAO1 being part of group 1 and PA14 part of group 2, the groups assigned in this study are not consistent with this.It would also be useful to include the PA7 reference strain in the phylogeny, as it is part of the 3rd phylogenetic group, to see where the isolates in this study sit in the phylogeny in relation to this strain, and would put the isolates from this study into context with standard P. aeruginosa phylogeny groups.
• Why were isolates from only one patient selected to phenotypically investigate?It would strengthen the work to look across multiple patients to confirm this isn't only seen in one case.
• It would be useful to repeat competition assays in a more realistic medium (i.e.sputum mimicking), as P. aeruginosa behaves differently in different in vitro conditions, and standard laboratory media is known to be a poor representation of chronic P. aeruginosa infection.

Minor comments:
• The introduction section has some very long sentences (e.g.L88-92), it would be much easier to read if these were broken up into smaller sentences.
• If patients were negative for PA for the last year, why were these selected and how was PA isolated from samples of PA-negative patients?• Why were isolates selected from those specific patients for further analysis?This needs to be explained, even if it was just at random it would be good to state how this selection was performed.
• The isolate IDs are quite hard to follow, it would be easier to understand the overall manuscript if they were more intuitive, or include a table with information for each isolate ID to refer back to whilst reading.
• More method details are needed for the RNA seq analysis from the mouse experiments, the aligners etc. used for bacterial RNA seq analysis are typically not suitable for mammalian RNA seq analysis.
• Need to add references for the bioinformatics packages/programmes used.L83 -better to say 'populations, comprising of' rather than 'population comprises' L93 -think this line is missing the word population L105 -define BALs acronym here as it is the first time it is used and may not be known by all readers Fig 1c -bigger graphs would be easier to read -maybe 2x2 instead?

Authors' point-by-point responses to reviewers' comments:
Reviewer #1 (Remarks to the Author): This is an interesting manuscript by Zhao and colleagues tracking the emergence and behaviors of lasR mutants in people who have chronic P. aeruginosa infections.The most interesting observation is the idea that the presence of both WT and LasR mutant cells attenuates the pathogenicity of the bacteria in a way that cannot be explained by either type in a monoinfection (Fig 4).

Authors' responses:
We highly appreciate the warm and insightful comments and constructive suggestions of this expert reviewer, and also sincerely thank him/her for the time and effort in reviewing this manuscript.

Major issues:
1.The authors have chosen to interpret their data through the prism of cooperation and cheating in bacterial populations, focusing on public goods.They provide little evidence, if any, that is consistent with the idea that there is a social dilemma in the lungs of people with COPD.In fact, there is some evidence (eg, PMID: 24866798) that in the context of lung infections, lasR mutants are better adapted to the environment and are not cheaters.Consistent with this issue it is not clear how the experiments in Figure 3 are relevant to the evolution within the lung.That the mutations occur in a sequential way that stabilizes cooperation in laboratory conditions does not mean that this was the selective pressure for the mutations.
Authors' responses: Thanks for the comments.Yes, compared to the relatively stable and controllable laboratory conditions, clearly proving the detail of bacterial public goods game in vivo is still a challenge.
By developing an ex vivo porcine lung (EVPL) model, Harrison et al. (PMID: 24866798) reported that the lasR mutant grew better than the wild-type (WT) P. aeruginosa in vivo but showed no significant growth advantage during co-infection.However, we think that the statement of "lasR mutants are better adapted to the lung environment and are not cheaters" still remains to be discussed.First of all, the better growth of isogeneic lasR mutant with deficient quorum-sensing (QS) system than WT P. aeruginosa in vivo is kind of counter-intuitive, and this might be due to the absence of an active immune system in the EVPL model.Furthermore, the application of 1:1 (or 60% WT/40% lasR mutant) co-infection in EVPL model can better verify the coexistence of lasR mutant and WT strains, but may be inappropriate to study the invasion The experiments in Fig. 3 of our present study explored the in vitro competition of the isolates W1a, W4b, and W8a, which belong to the same lineage from the same patient and carry the mutations occurred in a sequential way.The isolate W1c from the same patient was included as a highly evolved lasR mutant that carried several additional mutations, albeit it belongs to a different lineage.These results demonstrated the influence of the sequentially occurred mutations on the interaction of P. aeruginosa isolates in the framework of public goods game.We admit that these experiments just provide a paradigm of bacterial interaction in vitro by using differentially evolved isolates from the host.It is hard (or currently impossible) to directly identify the transition of bacterial interaction according to the change of selection pressure in the lung environment.Alternatively, we further investigated the dynamic change of the proportion of each isolate during co-culture and co-infection to verity the outcome of the bacterial competition.We found that the subsequently emerged mutants could invade the original isolate, and social cooperation successed both in vitro and in vivo.In the revised version, this conclusion was further validated by repeating all the experiments using the isolates N1a, N5c-lasR, and N7d-lasRpvdS from the patient N, and by culturing the four isolates from patient W in artificial sputum medium (ASM) (Figs.S10-S14).
2. The SNP data and the trees drawn suggest the possibility that some of the patients might have harbored multiple lineages of P. aeruginosa, which makes the analysis of mutational change over time rather difficult.
Authors' responses: Thanks for the comments.Yes, as we have stated in the main text, 'our results clearly demonstrated the cross-transmission of P. aeruginosa among patients and confirmed that COPD airways were frequently cocolonized by polymorphic P. aeruginosa' (Lines 145-147).After comparative genome analysis, we have clarified the mutational change of each P. aeruginosa lineage from patient W and listed the sequentially evolved mutations in Fig. 2b and Table 1. 3. The introduction should be reorganized to introduce quorum sensing and QS regulation of extracellular products before a discussion of social cheating.

Authors' responses:
Thanks for the suggestion.The Introduction section has been carefully reorganized as suggested by this expert reviewer.

Reviewer #2 (Remarks to the Author):
The authors collect longitudinal samples of P. aeruginosa from COPD patients and look at the evolution of social behaviors, with a primary focus on QS and pyoverdine production.They perform in vitro competition experiments and look at host colonization and immune response in a mouse model, and gene expression in human samples also subjected to mass spec.They infer that the sequential evolution of social cheats on different traits can stabilize a heterogeneous population and affect host response.
It is great to see this type of analyses on clinical isolates.However, I found the manuscript difficult to read, with many evolutionary statements that I don't understand, or that are wrong.Examples are (see also below): Authors' responses: We highly appreciate the warm and insightful comments and constructive suggestions of this expert reviewer, and also sincerely thank him/her for the time and effort in reviewing this manuscript and the positive recommendation.The whole manuscript has been carefully rewritten according to the suggestions of this reviewer.
1. "modifying the adaptability" eg.l 85: we hypothesize that further evolution of the polymorphic P. aeruginosa population may mainly concentrate on modifying the adaptability of lasR mutants to facilitate the persistent colonization of the whole population -and l 233: while the adaptability of lasR mutants would be adjusted by the accumulation of other mutations during further evolution in COPD airways -what does this mean?Authors' responses: Thanks for the valuable comments.Sorry for bring any inconveniences to this expert reviewer caused by our inappropriate wording.We have revised the sentences as 'we hypothesize that further evolution of the polymorphic P. aeruginosa population may mainly concentrate on selecting the lasR mutants that can facilitate the persistent colonization of the population' (Lines 75-77), 'The evolution of the polymorphic P. aeruginosa was characterized by selecting different lasR mutants…(Line 83)', '…the accumulation of other mutations might be associated with the enhanced adaptability of lasR mutants during further evolution' (Lines 212), '…is associated with the accumulation of mutations in the lasR mutants…(Lines 447), etc.
2. 130-134: These results suggested that QS-controlled extracellular proteases might be a vitally important public goods that could bring more direct benefits (in terms of nutrient acquisition) for the survival of P. aeruginosa in vivo, followed by siderophores, and the intraspecific competitions of which would be more readily to cause the sequential invasion of individuals deficient in producing the products.
-unclear to me Authors' responses: Thanks for the valuable comments.Sorry for bring any inconveniences to this expert reviewer because of our unclear writing.This sentence has been revised as 'Therefore, these data revealed that the P. aeruginosa isolates deficient in producing the costly and sharable extracellular products were selected in a sequential way.P. aeruginosa lasR mutants deficient in producing extracellular proteases might be selected first in COPD airways, while the lasR mutants carrying additional mutations in the pvdS gene might be selected during further evolution' (Lines 118-122).
3. L.465-466: will be introduced to stabilize the population structure and to promote the population fitness during chronic lung infection.Selection doesn't work to stabilize population structure and population fitness.
Authors' responses: Thanks for the valuable comments.This sentence has been revised as 'The population structure can be stabilized by the formation of a cascaded public goods game mediated by QS-controlled extracellular products and siderophores during chronic lung infection' (Lines 448-450).

The naming of isolates makes it difficult to remember what is what, and to see that not all isolates from
the same patient are of the same clone type -and therefore not sequentially evolved from the same ancestor.Consider renaming strains to also include clone type, and perhaps even phenotype?Authors' responses: Thanks for the valuable suggestion.Yes, as we have written in the main text, the P. aeruginosa isolates from patient W could be divided into at least three main lineages.To enhance the distinguishability of the isolates used in each experiment, we have renamed the isolates throughout the manuscript by adding the information of MLST and the relevant phenotypes in the revised version.
5. I truly appreciate the challenges of working with clinical isolates.However, the generality of the findings is not clear: the sequential accumulation of mutants in first lasR and then pyoverdine is based on two patients, where selected isolates of one are analysed in more detail.Additionally, the isolates analysed in co-culture were not sampled at the same time and only assumed to be co-occurring.This is not entirely unreasonable as they were collected within a relatively short time frame, however, it needs to be stated explicitly.
Authors' responses: Thanks for the positive and valuable comments of this expert reviewer, and sorry for bring any misunderstandings to this reviewer because of our unclear naming on the isolates and writing.
Actually, all the isolates used for the evolutionary trajectory analysis and co-culture/co-infection assays in the present study were from the same patient W. The sequentially evolved isolates W1a, W4b and W8a belong to the same lineage.The highly evolved lasR mutant W1c, which was co-isolated with W1a, belongs to another lineage.We have renamed the isolates by adding the information of clone type to enhance the traceability of them.
Furthermore, we would like to clarify that there was a huge number of P. aeruginosa clones after the clinical samples from each sampling period were cultured on LB plates.Identifying the genetic composition of all the isolates from each sampling period by using whole-genome sequencing is a huge task and too expensive.Thus, it is too difficult to trace the presence of a specific isolate in host lungs over time.As we have stated in the Materials and Methods section, only 'The colonies with apparent differences in shape, color, size and surface states were picked out…'.The isolates from each sampling period of different patients and representing the phenotypes of others in each subclade of the isolate typing were selected for genomic identification.Although we did not prove the co-occurrence of the isolates W1a, W4b and W8a, the detection of isolates producing extracellular protease like W1a, the isolates harbored lasR mutation and lasR-pvdS double mutation after 4 months in patient W (Fig. 1c-f, and Fig. S1c), might suggest a strong probability of the coexistence of them.Moreover, we further performed a set of parallel experiments by using the isolates from patient N and had comparable genetic and phenotypic characters to those from patient W, and obtained similar results to those presented in Fig. 3 (Fig. S10 and S11).

25: heterogeneous population, not individuals
Authors' responses: Thanks for the suggestion.We have corrected this in the revised version (Line 20).
Authors' responses: Thanks for the suggestion.We think the word 'unnecessary' here is fine.Because not all the COPD patients enrolled in the present study always need the endoscopic surgery, which was only applied to the patients with breathing difficulties.
490: what is meant with: The single colony of identified Authors' responses: Thanks for the comment.This sentence has been revised as 'The single colony of P. aeruginosa clinical isolates…' (Line 474).

505: DNA singular
Authors' responses: Thanks for the suggestion.We have corrected this in the revised version (Line 489).518: was siderophore production measured in iron limited or iron supplemented media?Authors' responses: Thanks for the valuable comment and sorry for the writing error.The production of siderophores was determined by measuring the cell densities of P. aeruginosa cultured in M9-CAA medium supplemented with 100 qg/mL of Transferrin (strict iron-limitation medium) for 24 h (Line 499).

560: mention also monocultures
Authors' responses: Thanks for the suggestion.We have added the statement of 'The growth status of monocultured isolates in QS-required and not required media was determined also' in the revised version (Lines 544-545).S3: Why compare all isolates back to w1a when some are different clone types?In the case of transmission you could compare these with each other?801: when comparing two different clone types I don't think it makes sense to list the SNPs etc as accumulated mutations Authors' responses: Thanks for the valuable comments.After comparative genomic analysis, we had also realized that the isolates from the same patients were belonged to different clone types.We agree that it is inappropriate to compare different types of isolates back to the initial isolates.Therefore, to check the accumulation of mutations in the isolates with the same clone type, we have performed an additional analysis by comparing W4a and W6b to W1c (Subgroup 1) and comparing W5b and W7f to W3a (Subgroup 2).The results are provided in the new Table 1.Authors' responses: Thanks for the comments.Yes, we also agree that the highly evolved lasR mutant W1c alone might do best, because this isolate also harbored a large set of loss-of-function mutations in several pathoadaptive genes.As we had shown in Fig. 3l, W1c could indeed invade W1a in mouse lungs from an initial proportion of 1% to a proportion comparable to W1a.Therefore, we further studied the influence of the coexistent W1a and W1c on the immune responses of mice.W1a had a strong pathogenic ability, while W1c lost most of the lethality but adapted to persistent colonization.Even W1c caused no death of mice, we were attracted by the finding that either the mono-infection of W1a or W1c caused a remarkable expression change (increase or decrease) of immune-related genes in mice compared to the uninfected control, while the expression change of these genes in the mice of co-infection group was relatively lower.Moreover, the clearance of P. aeruginosa isolates (pure extracelluar protease-negative) from the lung of patient C after 12 months also indicates that the persistent colonization of lasR mutants in vivo might require the WT strains.In patient W, all isolates still produce pyoverdine?How do they grow in iron limited media?Authors' responses: Thanks for the important comments.We have just found that these data were missed when we were copying the data from Excel to GraphPad Prism.We have checked all the data and revalidated the phenotypes and genotypes of the isolates and regenerated the figures.Sorry for this mistake and we would like to thank this expert reviewer for his/her careful and rigorous scientific attitude.
Figure 2B: what are the black arrow heads?Where do the boxes belong to?
of WT P. aeruginosa by lasR mutant.A previous study by Rumbaugh et al. (2009.Current Biology 19, 341-345) had clearly demonstrated the in vivo invasion of lasR mutant towards WT P. aeruginosa.They found that the frequency of lasR mutant during the co-infection with WT P. aeruginosa would increase from 1% to 10%, or even to 35% in the chronic infection.Moreover, the work of Köhler (2009.Proc Natl Acad Sci U S A. 106:6339-6344) and Andersen (2015.Proc Natl Acad Sci U S A. 112:10756-10761) also provide clinical evidence to show the cheating of lasR or pvdS mutants in the presence of wild-type cells.
figures are hard to read and I had difficulty distinguishing between some of the colors (and figuring out what the control group was) Authors' responses: Thanks for the comments.We have clarified the meaning of the colors and the control group in the figure legends in the revised version.2. Did the authors adjust their statistical tests for multiple comparisons in the cases where there are multiple comparisons?Authors' responses: Thanks for the comments.Multiple comparisons were performed by comparing the data to a designated group using one-way ANOVA with Tukey post-hoc tests using a 95% confidence interval.

Figure 2B :
Figure 2B: what are the black arrow heads?Where do the boxes belong to?Authors' responses: Thanks for the comments.The black arrow heads were omitted in the revised version.The meaning of genes in the boxes has been explained in the figure legends in Lines 777-779.

Figure 2B & Table
Figure 2B & TableS3: Why compare all isolates back to w1a when some are different clone types?In the

Figure 4 :
Figure 4: in the W1c and W1a mono-infections, a number of animals die between day 2-4 -could this impact the CFU counts in B? Authors' responses: In the experimental operation of Fig. 4a and Fig. 4b, the different number of agar bead-embedded bacterial cells we used.In Fig. 4a, we administered a lethal dose of bacteria (2.0 × 10 6CFUs in 50 qL of sterile saline) and part of the mice died within 2-4 days.While in Fig.4b, sublethal dose (0.5 × 10 6 CFUs in 50 qL of sterile saline) was used and only few mice died.We have revised the infected number of bacteria in the part of "Mouse models" in the Method (Line 581).

Figure
Figure 4B & C: I would argue that W1c alone does best -all hosts are alive and CFU counts are steady -so is it really a cheat in vivo?

Figure S1 :
Figure S1: what are the sample sizes at the different time points?Authors' responses: Thanks for the comments.The numbers of isolates at different sampling time points of each patient have been provided in the legend of Fig. S1.

Figure S5: 8
Figure S5: 8 replicates in C? Authors' responses: Thanks for the comment.Yes, this experiment was performed on the 96-well plate with 8 replicates for each culture.