Article https://doi.org/10.1038/s41467-022-34101-2 Gut to lung translocation and antibiotic mediated selection shape the dynamics of Pseudomonas aeruginosa in an ICU patient Received: 17 January 2022 Rachel M. Wheatley 1,8, Julio Diaz Caballero1,8, Thomas E. van der Schalk 2 , Fien H. R. De Winter 3, Liam P. Shaw1, Natalia Kapel1, Claudia Recanatini 4 , Accepted: 13 October 2022 Leen Timbermont 2, Jan Kluytmans4, Mark Esser 5, Alicia Lacoma6, Cristina Prat-Aymerich 4,6, Antonio Oliver7, Samir Kumar-Singh 2,3, Surbhi Malhotra-Kumar 2 & R. Craig MacLean 1 Check for updates 1234567890():,; 1234567890():,; Bacteria have the potential to translocate between sites in the human body, but the dynamics and consequences of within-host bacterial migration remain poorly understood. Here we investigate the link between gut and lung Pseu- domonas aeruginosa populations in an intensively sampled ICU patient using a combination of genomics, isolate phenotyping, host immunity profiling, and clinical data. Crucially, we show that lung colonization in the ICU was driven by the translocation of P. aeruginosa from the gut. Meropenem treatment for a suspected urinary tract infection selected for elevated resistance in both the gut and lung. However, resistance was driven by parallel evolution in the gut and lung coupled with organ specific selective pressures, and translocation had only a minor impact on AMR. These findings suggest that reducing intestinal colonization of Pseudomonas may be an effective way to prevent lung infections in critically ill patients. Bacteria often colonise multiple anatomical sites in human hosts, but P. aeruginosa is an opportunistic pathogen that is a major cause the dynamics of within-host translocation and its consequences for of healthcare-associated infections worldwide7,8, most notably in pathogenesis and host adaptation remain poorly understood1–3. For patients with compromised immunity9,10. Pseudomonas is not con- example, advances in microbiome profiling methods have shown that sidered to be a typical member of the gut microbiome, and intestinal the gut microbiome can transmit to the lungs of critically ill patients4,5, colonisation with Pseudomonas is associated with an increased risk of and translocation is associated with poorer outcomes in mechanically developing lung infections11–13 and mortality14. Gut colonisation ventilated patients6. While gut-to-lung translocation has been usually precedes lung infection, and the same strain is often found in demonstrated at the microbiome level, the dynamics and con- the gut and lungs, suggesting that the gut acts as a reservoir of sequences of translocation for individual pathogens and antibiotic Pseudomonas that can be transmitted to the lung and other infection resistance are not well understood. sites15–17. However, direct evidence for gut-to-lung transmission of 1 University of Oxford, Department of Biology, Oxford, UK. 2Laboratory of Medical Microbiology, Vaccine and Infectious Disease Institute, Faculty of Medicine, University of Antwerp, Wilrijk, Belgium. 3Molecular Pathology Group, Laboratory of Cell Biology and Histology, Faculty of Medicine, University of Antwerp, Wilrijk, Belgium. 4Julius Center for Health Sciences and Primary Care, University Medical Center Utrecht, Utrecht University, Utrecht, The Netherlands. 5 Microbial Sciences, BioPharmaceuticals R&D, AstraZeneca, Gaithersburg, MD, USA. 6Microbiology Department, Hospital Universitari Germans Trias i Pujol, Institut d’Investigació Germans Trias i Pujol, CIBER Enfermedades Respiratorias, Universitat Autònoma de Barcelona, Badalona, Spain. 7Servicio de Micro- biología, Hospital Universitari Son Espases, Instituto de Investigación Sanitaria Illes Balears (IdISBa), Palma de Mallorca, Spain. 8These authors contributed equally: Rachel M. Wheatley, Julio Diaz Caballero. e-mail: craig.maclean@biology.ox.ac.uk Nature Communications | (2022)13:6523 1 Article https://doi.org/10.1038/s41467-022-34101-2 P. aeruginosa is lacking, and it is possible that intestinal carriage patient over a 30-day period. We used phylogenetic approaches to simply reflects an innate susceptibility to Pseudomonas infection or test for translocation, and a combination of genomic and phenotypic proximity to a source of Pseudomonas that can independently colo- methods to study the link between AMR and within-host nise the lung and gut. transmission. One of the major challenges of dealing with Pseudomonas aeru- ginosa is antibiotic resistance. P. aeruginosa has high levels of intrinsic Results resistance to antibiotics18–20 and a remarkable ability to evolve resis- Clinical timeline tance de novo in hospitalised patients21–23. In classic population genetic The focal patient was admitted to ICU of Hospital Universitari Ger- models, migration can accelerate evolutionary adaptation at a local mans Trias i Pujol in Badalona, Spain with a primary diagnosis of scale by increasing the genetic diversity that selection acts on24. In this seizure. Mechanical ventilation was started on ICU admission and case, translocation could play a role in antibiotic resistance by moving was continued for a total duration of 39 days. The patient was resistance determinants between bacterial colonisation sites. In an immediately treated with amoxicillin clavulanate, which is not active extreme example, bacterial populations in one organ (for example, the against P. aeruginosa, due to suspected aspiration of gut) could act as a source of resistant mutants that are then dis- oropharyngeal or gastric contents into the lower respiratory tract seminated to other organs (i.e. the lung). (bronchoaspiration). The patient was enrolled in ASPIRE-ICU trial25 at To test the importance of gut-to-lung transmission in Pseudo- 48 h post admission (hereafter day 1). Meropenem was started on day monas colonisation and antimicrobial resistance (AMR), we carried 12 and continued for 10 days to treat a suspected urinary tract out an in-depth case study on a single intensively sampled ICU infection. Over the course of stay in ICU, a total of 52 P. aeruginosa Fig. 1 | Clinical timeline and resistance phenotyping. A Timeline of patient (orange) and lung (green). Each isolate had a median meropenem MIC calculated sampling, showing samples that tested positive or negative for P. aeruginosa from n = 3 biologically independent replicates. Meropenem resistance increased colonisation by culturing. Sampling points from which isolates were collected are over time, and P. aeruginosa isolates from the final lung sample were above the highlighted with a green ring. The patient was treated with amoxicillin clavulanate EUCAST clinical breakpoint for meropenem resistance (red dashed line). Amox- from 2 days prior to enrolment until day 6 and with meropenem from day 12 to day icillin clavulanate resistance was not measured as this antibiotic is not active against 21. B Meropenem minimum inhibitory concentration (MIC) (mean ± s.e.m μg/mL) P. aeruginosa. Source data are provided as a Source Data file. for isolates (n = 4–12, as labelled on plot) from each sampling point from the gut Nature Communications | (2022)13:6523 2 Article https://doi.org/10.1038/s41467-022-34101-2 isolates were collected, from endotracheal aspirate (ETA) samples treatment, and meropenem resistant P. aeruginosa ultimately colo- (n = 12) and peri-anal swabs (n = 40) up until day 30 (Fig. 1), which was nised the lung. This complex clinical timeline suggests that the end-point of the ASPIRE-ICU trial. Culture screening of patient translocation between the gut and lung may have occurred (Fig. 1), blood samples from day 2, day 11 and day 21 all reported no growth of but clinical data and isolate phenotypes alone provide limited P. aeruginosa. P. aeruginosa colonisation was detected in the insights into the underlying drivers of within-host translocation lungs at day 1. Gut colonisation was detected following meropenem and AMR. Nature Communications | (2022)13:6523 3 Article https://doi.org/10.1038/s41467-022-34101-2 Fig. 2 | Genome sequencing and phylogenetic analysis. A Phylogenetic recon- meropenem for each isolate is presented with filled black circles in log2 scale of the struction of lung (n = 12) and gut (n = 40) isolates rooted using P. aeruginosa PA1, minimum inhibitory concentration (MIC). D The topology of the tree suggested five another ST782 clinical isolate sampled from a respiratory tract infection72, as the distinct groups based on the identification of small polymorphisms. E The accu- outgroup. Putatively adaptive polymorphisms in genes or pathways showing par- mulation of variants over time (mean ± s.e for isolates (n = 4–12, as labelled on plot) allel evolution are annotated on the phylogeny. Protein altering mutations are from each sampling point from the gut (orange) and lung (green) suggests within- shown in black and silent mutations are shown in light grey. A polymorphism in a host evolution of a clone rather than recurrent episodes of colonisation. F Dated known multi-drug efflux pump regulator (mexR) is also highlighted. Variation in the genealogy of isolates constructed with BactDating73. The inferred instance of gut to presence/absence of a 190kB genomic island is shown, and inferred losses of the lung transmission and the acquisition of meropenem resistance mutations have genomic island are identified with blue triangles in the tree. B Isolate name, lung been annotated on the genealogy. Source data are provided as a Source Data file. (green) or gut (orange) origin, and day in study of collection. C Susceptibility to Genomic insights into pathogen colonisation and evolution loss of the genomic island (Fig. 2), suggesting that the loss of this island To characterise the genetic diversity within this patient, we used long was not driven by antibiotic treatment. and short read sequencing to construct a hybrid assembly for a single Bacterial phylogenies are a powerful tool to detect transmission isolate, yielding a ~6.3 Mb ST782 reference genome distributed across events, particularly when combined with isolate sampling dates3,32,33. 5 contigs. Short-read sequences of lung (n = 12) and gut (n = 40) iso- To reconstruct within-host translocation, we used our time scaled lates were mapped to this reference genome, and we identified poly- isolate genealogy to infer translocation events (Fig. 2F). Secondary morphic SNPs (n = 17), indels (n = 7), and variation in presence/absence lung colonisation was driven by the growth of a clone with mutations of a 190 kb genomic island (Supplementary Table 1 and Supplemen- in the oprD porin, which is a key carbapenem sensitivity determinant, tary Fig. 1). and mexR, which regulates the expression the MexAB-OprM multi- The genetic diversity found in this patient could reflect either (i) drug efflux pump34 (Fig. 2A – lineage 3). This lineage is nested within a recurrent colonisation/infection by multiple clones or (ii) within-host broader clade of gut isolates, providing strong evidence of gut to lung evolution of a single clone. To discriminate between these processes, transmission (Fig. 2F). The dated genealogy suggests the MRCA of we reconstructed the phylogeny of isolates using P. aeruginosa PA1, a lineage 3 existed between day 18 and day 24 (Supplementary Fig. 2)26, closely related ST782 genome, as an outgroup (Fig. 2A–D). All of the giving an approximate time frame for gut to lung translocation. isolates were closely related and the number of variants per isolate However, this genealogy does not provide any insight into whether correlated strongly with the day of isolate collection, supporting the lineage 3 acquired the carbapenem resistance mutations before or idea that within-patient diversity emerged as a result of in situ evolu- after transmission to the lung. Interestingly, a single lineage 3 isolate tion of a single founding clone (Fig. 2E; r2 = 0.62, F1,50 = 82, P < 0.0001). was recovered from the gut. The presence of this lineage in the lung No other patients within the ASPIRE-ICU cohort at this hospital were and gut implies that either meropenem resistance evolved in the gut colonised by P. aeruginosa ST782 during the trial, providing further prior to transmission to the lung, or that this lineage secondarily support for the within-host diversification as opposed to repeated transmitted from the lung to the gut after evolving meropenem colonisation. resistance in the lung. We estimated the onset of colonisation using a time-scaled gen- Beyond this clear-cut case of gut to lung translocation, inferred ealogy of isolates with the most-recent common ancestor (MRCA) of patterns of translocation depend strongly on the assumptions made all isolates predicted to be at least 3 weeks prior to ICU admission regarding initial colonisation by the ancestral clone. If Pseudomonas (Fig. 2F; Supplementary Fig. 2; MRCA from BactDating: 22–74 days initially colonised the gut, then the phylogeny implies that initial lung before day 0)26.The rate of evolution in this patient was ~18 ± 10 SNPs/ colonisation by lineage 1 was driven by an earlier gut-to-lung translo- year26, which is higher than the typical evolutionary rate of bacterial cation event that could have been associated with broncoaspiration pathogens of 1–10 SNPs/year3. However, this elevated evolutionary upon ICU admission. Alternatively, the ancestral clone may have rate is in line with the rate reported from another patient in this trial23, colonised the lungs. Under this model of host colonisation, the phy- highlighting the high in vivo mutation rate of P. aeruginosa in critically logeny suggests that lung to gut transmission occurred in lineages 2 ill patients. This phylogenetic approach, which strongly supports and 5, implying that at least one episode of translocation occurred at continuous host colonisation, suggests that culture-based approaches some point prior to ICU admission. Finally, it is possible that the (i.e. Fig. 1A) have limited ability to detect P.aeruginosa colonisation. ancestral clone independently colonised both the lungs and gut. Signatures of parallel evolution provide a simple way to identify According to this model of host colonisation, it is not necessary to putative beneficial mutations that underpin adaptation to the novel invoke any additional translocation (beyond the clear gut to lung environment of the human host3,27. Parallel evolution occurred in 3 translocation in lineage 3) in order to explain the phylogenetic dis- genes or operons that have functional roles in resistance to carbape- tribution of lung and gut isolates. Unfortunately, it is not possible to nem antibiotics (oprD)27, alginate biosynthesis (algW, algL)28, and clearly differentiate between these models because the lack of pre-ICU selenocysteine biosynthesis (selA and selB)29. These putative pathoa- admission isolates hinders our ability to reconstruct the early history daptive mutations accounted for 7 of the 24 variants, providing strong of Pseudomonas colonisation in this patient. All three models of host evidence for rapid adaptation to the host environment. Interestingly, 3 colonisation lead to a similar number of inferred translocation events of these 7 mutations were synonymous, suggesting that transcription (1–3), suggesting that they are similarly parsimonious from a phylo- efficiency may have been a key target of selection30. The presence of genetic perspective. Culturing patient samples found evidence of lung the variable genomic island in the outgroup and in lung isolates (Fig. 2, colonisation prior to gut colonisation (Fig. 1A), but we argue that this lineage 5) implies that island was lost on two independent occasions, culture data provides limited insights into early host colonisation, suggesting that loss of this element was adaptive. Inferring the selec- which likely occurred >20 days prior to ICU admission. A further tive advantage of large scale deletions is difficult, but it is worth noting challenge of using culture data to infer host colonisation is that the that this island carries pyoverdine biosynthesis genes that are selected limits of detection from culturing samples from different tissues are against in the host environment31. Isolates in possession of the geno- unknown, and it is conceivable that ETA samples and peri-anal swabs mic island showed similar levels of meropenem resistance (~0.5 μg/mL simply differ in their sensitivity to detect Pseudomonas in the lung MIC) to isolates from the same phylogenetic lineage (lineage 5) with and gut. Nature Communications | (2022)13:6523 4 Article https://doi.org/10.1038/s41467-022-34101-2 Fig. 3 | Cytokine concentrations were measured in ETA samples collected over indicates meropenem treatment window on timeline (day 12 to day 21). Source data the course of the study at days 4, 9, 12, 17, 19 and 24. The following cytokines are provided as a Source Data file. were measured: A IL-4, B IL-33, C Fractalkine, D IL-22, E IL-8. Yellow shading Immune response to lung colonisation idea that secondary lung colonisation occurred at some point between Lung colonisation provides P. aeruginosa with the opportunity to day 12 and day 17, and they suggest that the host immune response establish infection by adhering to the mucosal surface and penetrating may have prevented colonisation from progressing to pneumonia. the epithelial barrier, leading to the development of pneumonia, which is associated with a very high mortality rate in ICU patients7. To Drivers of antibiotic resistance investigate the role of host immunity in preventing infection, we P. aeruginosa has high levels of intrinsic antibiotic resistance and a measured the abundance of a panel of host immune effectors in remarkable ability to evolve increased resistance under antibiotic samples of endotracheal aspirate (ETA; Fig. 3). The dated genealogy treatment18,21. Given the possibility of translocation between the gut suggests that secondary lung colonisation by lineage 3 occurred and lung, we next sought to understand the relative contributions of between day 18 and day 24. Interestingly, ETA samples from days 17 migration, mutation and selection to the origin and spread of mer- and 19 were associated with spikes in the expression of IL-33, Frac- openem resistance in this patient. The phylogeny clearly shows that talkine, and IL-435,36, which have previously been to shown to enhance elevated meropenem resistance evolved on 2 separate occasions due the clearance of P. aeruginosa37. Colonisation was also associated with to mutations in oprD and mexR (Fig. 2, lineages 3 and 4). However, it is a > 10-fold increase in the concentration of IL-22 (day 19 sample), which challenging to follow the dynamics of meropenem resistance muta- protects against infections caused by attaching and effacing bacterial tions using isolates alone due to the limited number of isolates pathogens by increasing mucous production and by limiting excessive sequenced (n = 52) and the gaps in the sampling of isolates. To try and inflammation mediated by neutrophil influx38,39. Measuring cytokine overcome this problem, we combined isolate sequencing data with levels in ETA samples from the lungs provides a direct measurement of amplicon sequencing of oprD using DNA extracted directly from ETA immune response, but one concern over this approach is that it is samples and peri-anal swabs, some of which were not screened for possible for individual samples to give high concentrations of all isolates according to ASPIRE-ICU protocol (Fig. 4A, B). Amplicon cytokines, for example as a result of patient dehydration. However, in sequencing revealed that Pseudomonas was present in the lung from this case levels of IL-8 remained essentially constant across samples, day 17 onwards, which coincides with the inferred time of secondary supporting the idea that spikes of protective cytokines were not an lung colonisation from the isolate phylogeny (Fig. 2F) and cytokine artefact (Fig. 3E). These cytokine data provide further support of the profiling (Fig. 3). oprD amplicon sequencing from DNA isolated from Nature Communications | (2022)13:6523 5 Article https://doi.org/10.1038/s41467-022-34101-2 Fig. 4 | Evolution and transmission of meropenem resistance. A, B Dynamics of biologically independent replicates (Source Data), and at least three isolates from oprD variants, as determined by isolate (circle) and oprD amplicon (diamond) each lineage were measured. Aerobic growth (D) was measured as exponential sequencing data. For the two sampling points (gut day 23 and gut day 25) where growth rate in standard culture conditions. Isolates are colour coded according to both amplicon and isolate sequencing was carried out, the mean of the two fre- phylogenetic lineage, as defined in Fig. 2. Values plotted for each point (i.e. isolate) quencies is shown (star). Measurements of SNP frequency from isolate sequencing are calculated from n = 7 biologically independent replicates (Source Data), and at and amplicon sequencing were strongly correlated (R2 = 0.9963; Supplementary least three isolates from each lineage were measured. oprD mutations were asso- Fig. 3). The yellow area shows the window of meropenem treatment and the red ciated with impaired growth under anaerobic conditions (P = 0.010), but not area shows a conservative minimum detection limit of variants from amplicon aerobic conditions (P = 0.950), as judged by a nested ANOVA. Data from isolates sequencing due to the error rate of nanopore sequencing. C, D Growth of isolates from different lineages are shown together because fitness measures did not differ with an oprD variant (Δ oprD) compared to isolates with the wild-type oprD back- between lineages nested within oprD genotype (P > 0.5). All statistical tests for this ground. Anaerobic growth (C) was measured as OD595 after 72 hours growth in analysis are two-tailed. Source data are provided as a Source Data file. anaerobic broth. Values plotted for each point (i.e. isolate) are calculated from n = 3 peri-anal swabs was less successful for reasons that are not clear, and fixation under meropenem treatment. The frequency of this mutation we only obtained amplicon sequencing data from days 23 and 25. ultimately declined in the gut due to the expansion of a carbapenem Importantly, the frequency of oprD variants measured by amplicon sensitive lineage (Fig. 2A – lineage 5). We speculate that this lineage and isolate sequencing from samples that yielded both isolates and may have survived carbapenem treatment by either colonising a Pseudomonas DNA was essentially identical (Supplementary Fig. 3), region of the gut with low carbapenem toxicity or by forming validating the use of amplicon sequencing to measure changes in allele persister cells. frequency. Isolate sequencing revealed a second frameshift mutation in oprD The most common oprD mutation in sequenced isolates (nt559Δ1) that was linked to a mutation in the mexR effux pump (nt1160Δ1) arose in the main clade of intestinal isolates (Fig. 2, lineage transcriptional regulator. Isolate sequencing revealed the presence of 4) and was never convincingly detected in lung samples (Fig. 4B). This the oprD nt559Δ1/mexRT305C lineage in gut samples from day 25 and combination of isolate and amplicon sequencing results provides lung samples from day 30, providing strong evidence that this resis- strong evidence that this mutation arose in the gut and swept to near tant lineage transmitted between colonisation sites. Amplicon Nature Communications | (2022)13:6523 6 Article https://doi.org/10.1038/s41467-022-34101-2 sequencing first detected this mutation as a polymorphism (frequency Migration increases genetic variation24, suggesting that within- approx. 20%) in the lung at day 20 (Fig. 4A), suggesting that this host translocation may accelerate bacterial adaptation to mutation arose in the lung following the transmission of lineage 3 from antibiotics1–3,47–50. In this case, resistance was driven by the spread of the gut. According to this model, the presence of rare oprD nt559Δ1 independent lineages in the gut and lung that were adapted to local variants (in isolates and amplicons) at day 25 reflects secondary lung to differences in antibiotic concentration. We found some evidence of gut transmission. However, we emphasise that it is not possible to translocation of resistant lineages, but the impact of within-host completely exclude the possibility that oprD nt559Δ1 arose in the gut migration on resistance was weak compared to selection, leading to and then transmitted to the lung where it spread rapidly by positive the emergence of a highly structured meropenem resistant pathogen selection. population1–3,49. We speculate that the high in vivo mutation rate of Antibiotic concentrations vary between host tissues, and it is Pseudomonas was key to shaping local adaptation to antibiotic selec- unclear to what extent pathogen populations adapt to local variation tion across tissues, and that within-host transmission is likely to pro- in selective pressures associated with antibiotic treatment. Mer- vide a more important source of resistance at smaller spatial scales2, or openem achieves higher concentrations in lung tissues than in the when mutation rate is low. gut40, suggesting that selection for resistance is stronger in the lung Hospital acquired infections caused by epidemically successful than in the gut. Isolates from the lung-associated oprD nt559Δ1/ MDR and XDR strains of P. aeruginosa have become a serious problem mexRT305C lineage had higher meropenem resistance (MIC = 16 μg/ worldwide19, and there is an urgent need to develop new antibiotics to mL, s.e. = 0 μg/mL, n = 7) than isolates from the gut-associated oprD treat infections caused by these strains. At the same time, the incred- nt1160Δ1 lineage (MIC = 4 μg/mL, s.e. = 0 μg/mL, n = 19; Fig. 2C), which ible ability of Pseudomonas to evolve resistance to antibiotic is consistent with the idea that selection for meropenem resistance treatment18,21,23,47 highlights the need to develop novel approaches to varies between organs. prevent or treat Pseudomonas infections. Our study suggests that A key challenge in evolutionary studies of AMR is to understand preventing gut colonisation or gut to lung transmission may be an how resistance can be maintained in pathogen populations in the effective strategy for preventing Pseudomonas infection in critically ill absence of continued antibiotic use41,42. In this case, oprD nt559Δ1 patients51–54. resistance remained stable in the lung following antibiotic treatment, but the frequency of oprD nt1160Δ1 declined in the gut. To test the role Methods of selection in the stability of resistance, we measured the growth rate Clinical timeline of isolates of all 5 major phylogenetic lineages in anaerobic culture The patient was admitted to the intensive care unit (ICU) of Hospital medium (Fig. 4C) and aerobic culture medium (Fig. 4D), which reca- Universitari Germans Trias i Pujol in Badalona, Spain with a primary pitulates one of the physiological differences between the gut and diagnosis of seizure. This patient was recruited as part of an observa- lung. Meropenem resistant lineages (Δ oprD) were not associated with tional multicenter European epidemiological cohort study (ASPIRE- decreased growth rate under aerobic conditions, suggesting that the ICU25), which was conducted according to the principles of the oprD and mexR mutations have little, if any, associated costs under Declaration of Helsinki, in accordance with the Medical Research these conditions (Fig. 4D; F1,12 = 0.0032, P = 0.956). In contrast, both Involving Human Subjects Act and local guidelines in the participating meropenem lineages were associated with decreased growth under countries. The study protocol was approved by the Research Ethics anaerobic conditions, suggesting that fitness costs associated with Committee of the Germans Trias i Pujol University Hospital and par- oprD mutations drove the loss of resistance in the gut (Fig. 4C; ticipants gave written informed consent. F1,12 = 9.27, P = 0.010). At the time of ICU admission, the patient did not suffer from pneumonia or any other active P. aeruginosa infection (APACHE-II Discussion score = 22 and Glasgow Coma scale = 3). No antibiotic use was The goal of this project was to understand the link between gut and reported in the two weeks prior to hospital admission. After 48 h of lung Pseudomonas colonisation in a single patient. By combining ICU admission, informed consent was obtained and the patient was clinical and genomic data, we were able to demonstrate a clear cut case enrolled in the ASPIRE-ICU study (day 1)25. Mechanical ventilation was of gut to lung transmission while the patient was in ICU. Whilst it is started on ICU admission and was continued for a total duration of difficult to generalise the findings of a single case study, these findings 39 days. Amoxicillin/clavulanic acid (1000 mg IV q8h for 8 days) was support the idea that gut to lung transmission may be a major driver of started on ICU admission for bronchoaspiration, the suspected P. aeruginosa respiratory tract colonisation in critically ill patients11,12,43. inhalation of oropharyngeal or gastric contents into the lower Carbapenem antibiotics such as meropenem are key to the respiratory tract. Meropenem (1000 mg IV q8h for 10 days) was treatment of P. aeruginosa infections23,44,45, and carbapenem-resistant started on day 12 to treat a suspected urinary tract infection. Patient P. aeruginosa has been identified as an important threat by the World endotracheal aspirate (ETA) and peri-anal swab samples were col- Health Organisation and the Centers for Disease Control and Preven- lected and screened for P. aeruginosa isolates via selective plating tion. In this patient, meropenem treatment for a suspected urinary until day 3023. Patient ETA samples were first blended (30,000 rpm, tract infection drove the repeated evolution of resistance, providing a probe size 8 mm, steps of 10 s, max 60 s in total), diluted 1:1 v/v with poignant example of the importance of ‘bystander selection’ for Lysomucil (10% Acetylcysteine solution) (Zambon S.A, Belgium) and AMR46. Ultimately, selection led to the emergence of a population of incubated for 30 min at 37 °C with 10 s vortexing every 15 min. highly resistant bacteria that persisted in the lung in the absence of Selective plating to screen for P. aeruginosa was carried out using antibiotic treatment, suggesting that the respiratory tract may act as a CHROMID P. aeruginosa Agar (BioMérieux, France) and blood agar source of carbapenem resistant Pseudomonas that can transmit to (BBL®Columbia II Agar Base (BD Diagnostics, USA) supplemented other body sites and potentially to other patients. Isolate sequencing, with 5% defibrinated horse blood (TCS Bioscience, UK)). Matrix- amplicon sequencing and immunological profiling all support the idea Assisted Laser Desorption Ionization-Time of Flight Mass Spectro- that gut to lung translocation coincided with meropenem treatment. metry (MALDI-TOF MS) was used to identify up to 12 P. aeruginosa This association may have arisen due to chance, but it is also possible colonies per sample, which were stored at −80 °C until further use. that antibiotic treatment facilitated gut to lung transmission, for This resulted in a total of 52 P. aeruginosa isolates collected from example by eliminating commensal lung bacteria that protected endotracheal aspirate samples (n = 12) and peri-anal swabs (n = 40). against Pseudomonas colonisation. In addition, patient blood cultures from day 2, day 11 and day 21 were screened at the local laboratory, all reported no growth (negative for Nature Communications | (2022)13:6523 7 Article https://doi.org/10.1038/s41467-022-34101-2 P. aeruginosa). The patient was discharged from the ICU and trans- Each isolate was typed using the Pseudomonas aeruginosa multi- ferred to a general medical ward on day 41. locus sequence typing (MLST) scheme from PubMLST (last accessed on 11.06.2021)62. Resistance phenotyping All isolates were grown from glycerol stocks on Luria-Bertani (LB) Long-read sequencing Miller Agar plates overnight at 37 °C. Single colonies were then Two isolates (EP717 (day 1 lungs) and EP623 (day 17 gut)) were inoculated into LB Miller broth for 18–20 h overnight growth at 37 °C sequenced using the Oxford nanopore MinION platform with a FLO- with shaking at 225 rpm. Overnight suspensions were serial diluted to MIN106 flow-cell and SQK-LSK109 sequencing kit. EP717 had sequen- ~5 × 105 CFU/mL. Resistance phenotyping to meropenem was carried cing coverage of 141X and EP623 of 233X. Raw reads were basecalled out as minimum inhibitory concentration (MIC) testing via broth using guppy v. 0.0.0 + 7969d57 and reads were demultiplexed using microdilution as defined by EUCAST recommendations55,56, with the qcat v. 1.1.0 (https://github.com/nanoporetech/qcat). Resulting alteration of LB Miller broth for growth media and the use of P. aeru- sequencing reads were assembled using unicycler v. 0.4.863, which ginosa PAO1 as a reference strain. Resistance to meropenem was used SAMtools v. 1.964, pilon v. 1.2359, and bowtie2 v. 2.3.5.165, in hybrid assayed along a 2-fold dilution series between 0.25–64 μg/mL. We mode with respective illumina reads. The EP717 assembly had a N50 of defined growth inhibition as OD595 < 0.200 and we calculated the MIC 1,797,327 for a total of 6,217,789 bases distributed in 11 contigs. The of each isolate as the median MIC score from three biologically inde- EP623 assembly had a N50 of 6,133,283 for a total of 6,330,243 bases pendent assays of each isolate (Source Data). distributed in 5 contigs. Growth assays Variant calling P. aeruginosa isolates were grown from glycerol stocks on LB Miller To identify mutations and gene gain/loss during the infection, short- Agar plates overnight at 37 °C. Single colonies were then inoculated length sequencing reads from each isolate were mapped to each of the into LB Miller broth for 18–20 h overnight growth at 37 °C with long-read de novo assemblies with BWA v. 0.7.1766 using the BWA-MEM shaking at 225 rpm. Overnight suspensions were serially diluted to an algorithm. Preliminary SNPs were identified with SAMtools and OD595 of ~0.05 within the inner 60 wells of a 96-well plate equipped BCFtools v. 1.9. Low-quality SNPs were filtered out using a two-step with a lid. To assess growth rate under standard aerobic conditions, SNP calling pipeline, which first identified potential SNPs using the isolates were then grown in LB Miller broth at 37 °C and optical following criteria: (1) Variant Phred quality score of 30 or higher, (2) At density (OD595 nm) measurements were taken at 10-min intervals in least 150 bases away from contig edge or indel, and (3) 20 or more a BioTek Synergy 2 microplate reader set to moderate continuous sequencing reads covering the potential SNP position. In the second shaking. Growth rate (Vmax; mOD/min) was calculated as the max- step, each preliminary SNP was reviewed for evidence of support for imum slope of OD versus time over an interval of ten consecutive the reference or the variant base; at least 80% of reads of Phred quality readings, and we visually inspected plots to confirm that this cap- score of 25 or higher were required to support the final call. An tured log-phase growth rate. We measured the growth rate of all 52 ambiguous call was defined as one with not enough support for the gut and lung isolates with a minimum of seven biological replicates reference or the variant, and, in total, only one non-phylogenetically to assess the relationship between meropenem resistance and fitness informative SNP position had ambiguous calls. Indels were identified (Source Data). Throughout growth assays a media control (to control by the overlap between the HaplotypeCaller of GATK v. 4.1.3.067 and for contamination) and a PAO1 control (to control for replicate breseq v. 0.34.068. The maximum parsimony phylogeny was con- plates) were included. We measured anaerobic growth using an structed based on high-confidence SNPs. To construct a dated gen- anaerobic jar (Thermo ScientificTM OxoidTM AnaeroJarTM base jar) ealogy of isolates, we dated the internal nodes of this tree using system with anaerobic gas generating sachets (Thermo ScientificTM bactdate in BactDating v1.1.026 (updateRoot = TRUE, minbralen=0.1) OxoidTM AnaeroGenTM sachets). An Oxoid Resazurin indicator strip after first converting multifurcating nodes to binary nodes. Phylo- was placed in the jar as an indicator to confirm generation of an genies were plotted with ggtree v3.0.469. anaerobic environment. For growth measurements, single colonies The variable genome was surveyed using GenAPI v. 1.09870 based were inoculated into LB Miller broth in the wells of a 96-well plate on the prokka annotation of the short-read de novo assemblies. The and placed in the anaerobic jar for 72 h, after which plates were presence or absence of genes in the potential variable genome was removed and OD595 was measured. For the comparison of oprD reviewed by mapping the sequencing reads to the respective genes variant (Δ oprD) isolates to wild-type oprD background (WT oprD) with BWA v.0.7.1766. isolates, growth measurements were taken for a minimum of three isolates (and a minimum of three biological replicates) selected as Amplicon Sequencing of oprD representatives from each phylogeny group to generate a mean Amplicon sequencing of the oprD gene was carried out to quantify the growth measurement for each Δ oprD and WT oprD group. To test for presence of the two key oprD variants observed in the isolate an association between oprD mutations and impaired growth we sequencing in whole gDNA samples that were available from the lung used a nested ANOVA that included main effects of oprD (ie either and gut of this patient. DNA was extracted from the ETA and peri-anal WT or Δ oprD, 1 df) and phylogenetic lineage nested within oprD (5 swab samples using a ZymoBIOMICS DNA Miniprep Kit (Zymo lineages shown in Figs. 2 and 3 df). Research, CA USA). On these extracted gDNA samples, a PCR amplification strategy Illumina sequencing using barcoded primers to amplify the oprD gene (1489 bp product All isolates were sequenced in the MiSeq or NextSeq illumina plat- length) and add sample specific DNA barcodes was followed71. This was forms yielding a sequencing coverage of 21X–142X. Raw reads were performed with PCR in Q5 High-Fidelity Master Mix (New England quality controlled with the ILLUMINACLIP (2:30:10) and SLI- BioLabs) and using a universal reverse primer and sample specific DINGWINDOW (4:15) in trimmomatic v. 0.3957. Quality controlled forwards primers containing 12 nt barcodes listed in Supplementary reads were assembled for each isolate with SPAdes v. 3.13.158 with Table 271. The temperature profile was 98 °C for 30 s, followed by 30 default parameters. These assemblies were further polished using cycles of 98 °C for 10 s, 70 °C for 30 s, 72 °C for 40 s, followed by a final pilon v. 1.2359 with minimum number of flank bases of 10, gap margin extension of 72 °C for 2 min. The barcoded oprD PCR products were of 100,000, and kmer size of 47. Resulting contigs were annotated pooled and sequenced on an Oxford nanopore MinION platform using based on the P. aeruginosa strain UCBPP-PA1460 in prokka v. 1.14.061. a FLO-MIN106 flow-cell and the SQK-LSK109 Ligation Sequencing kit. Nature Communications | (2022)13:6523 8 Article https://doi.org/10.1038/s41467-022-34101-2 Amplicon sequencing raw reads were basecalled using guppy v. 3. Didelot, X., Walker, A. S., Peto, T. E., Crook, D. W. & Wilson, D. J. 0.0.0 + 7969d57. This yielded 163,766 reads with an estimated read Within-host evolution of bacterial pathogens. Nat. Rev. Microbiol. length N50 of 2.64 kb. The data was demultiplexed allowing 2/ 14, 150–162 (2016). 12 sequencing errors in the barcode sequence and a maximum of 1 4. Dickson, R. P. et al. Enrichment of the lung microbiome with gut error in the downstream and upstream 4-mer. To identify the genotype bacteria in sepsis and the acute respiratory distress syndrome. Nat. of each read, we searched for the 11-mer sequence including the var- Microbiol. 1, 1–9 (2016). iant base and 5 bases downstream and upstream from this position. 5. Stanley, D. et al. Translocation and dissemination of commensal Using this conservative approach, we recovered 32–43% of the reads bacteria in post-stroke infection. Nat. Med. 22, 1277–1284 (2016). (Supplementary Table 3). 6. Dickson, R. P. et al. Lung microbiota predict clinical outcomes in critically ill patients. Am. J. Respir. Crit. Care Med. 201, Cytokine measurements 555–563 (2020). After ETA was blended, 0.5 g of the sample was diluted 1:1 with 7. Kang, C.-I. et al. Pseudomonas aeruginosa bacteremia: risk factors Sputolysin (Merck, Overijse, Belgium), vortexed and incubated at for mortality and influence of delayed receipt of effective anti- room temperature for 15 min. Samples were then centrifuged for microbial therapy on clinical outcome. Clin. Infect. Dis. 37, 5 min at 2000×g at room temperature. Supernatant was stored at 745–751 (2003). −80 °C until further processing. Levels of interleukin (IL-)4, IL-33, IL- 8. De Bentzmann, S. & Plésiat, P. The Pseudomonas aeruginosa 22, IL-8 and fractalkine were measured with the Mesoscale Discovery opportunistic pathogen and human infections. Environ. Microbiol. platform (Rockville, MD, USA) following the manufacturer’s instruc- 13, 1655–1665 (2011). tions. In brief, the plate was coated with capturing antibodies for 1 h 9. Fazeli, H. et al. Pseudomonas aeruginosa infections in patients, with shaking incubation at room temperature followed by washing hospital means, and personnel’s specimens. J. Res. Med. Sci. 17, off the plate. Samples were loaded and incubated for 1 h, after which 332 (2012). the plate was washed and incubated with detection antibodies. A 10. Yang, L. et al. Evolutionary dynamics of bacteria in a human host final wash was performed and MSD reading buffer 2x was applied environment. Proc. Natl Acad. Sci. USA 108, 7481–7486 (2011). before reading the plate in the QuickPlex SQ 120 (Rock- 11. Bonten, M. J. M., Bergmans, D., Speijer, H. & Stobberingh, E. E. ville, MD, USA). Characteristics of polyclonal endemicity of Pseudomonas aerugi- nosa colonization in intensive care units - implications for infection Reporting summary control. Am. J. Respir. Crit. Care Med. 160, 1212–1219 (1999). Further information on research design is available in the Nature 12. Gomez-Zorrilla, S. et al. Prospective observational study of prior Research Reporting Summary linked to this article. rectal colonization status as a predictor for subsequent develop- ment of pseudomonas aeruginosa clinical infections. Antimicrob. Data availability Agents Chemother. 59, 5213–5219 (2015). All clinical data analyzed for this patient as part of the study are 13. Fujitani, S., Sun, H. Y., Yu, V. L. & Weingarten, J. A. Pneumonia due to included in this article. Isolates can be obtained from the corre- pseudomonas aeruginosa part I: epidemiology, clinical diagnosis, sponding author for research use via an MTA subject to permission and source. Chest 139, 909–919 (2011). from the ASPIRE research committee. The source data are provided 14. Marshall, J. C., Christou, N. V. & Meakins, J. L. The gastrointestinal with this paper and have been deposited in the Oxford Research tract. The "undrained abscess" of multiple organ failure. Ann. Surg. Archive for Data (“DOI: 10.5287/bodleian:r1ekRa9wE” [https://doi.org/ 218, 111 (1993). 10.5287/bodleian:r1ekRa9wE]). 15. Bertrand, X. et al. Endemicity, molecular diversity and colonisation All sequencing data generated in this study has been deposited in routes of Pseudomonas aeruginosa in intensive care units. Intensive the NCBI short-read archive (“PRJNA802704”) and all data on isolates Care Med. 27, 1263–1268 (2001). can be found at: “SRR17868883”, “SRR17868884”, “SRR17868885”, 16. Okuda, J. et al. Translocation of Pseudomonas aeruginosa from the “SRR17868886”, “SRR17868887”, “SRR17868888”, “SRR17868889”, intestinal tract is mediated by the binding of ExoS to an Na, “SRR17868890”, “SRR17868891, “SRR17868892”, “SRR17868893”, K-ATPase regulator, FXYD3. Infect. Immun. 78, 4511–4522 (2010). “SRR17868894”, “SRR17868895”, “SRR17868896”, “SRR17868897”, 17. Markou, P. & Apidianakis, Y. Pathogenesis of intestinal Pseudomo- “SRR17868898”, “SRR17868899”, “SRR17868900”, “SRR17868901”, nas aeruginosa infection in patients with cancer. Front. Cell. Infect. “SRR17868902”, “SRR17868903”, “SRR17868904”, “SRR17868905”, Microbiol. 3, 115 (2014). “SRR17868906”, “SRR17868907”, “SRR17868908”, “SRR17868909”, 18. Breidenstein, E. B. M., de la Fuente-Nunez, C. & Hancock, R. E. W. “SRR17868910”, “SRR17868911”, “SRR17868912”, “SRR17868913”, Pseudomonas aeruginosa: all roads lead to resistance. Trends “SRR17868914”, “SRR17868915”, “SRR17868916”, “SRR17868917”, Microbiol. 19, 419–426 (2011). “SRR17868918”, “SRR17868919”, “SRR17868920”, “SRR17868921”, 19. Horcajada, J. P. et al. Epidemiology and treatment of multidrug- “SRR17868922”, “SRR17868923”, “SRR17868924”, “SRR17868925”, resistant and extensively drug-resistant pseudomonas aeruginosa “SRR17868926”, “SRR17868927”, “SRR17868928”, “SRR17868929”, infections. Clin. Microbiol. Rev. 32, e00031-19 (2019). “SRR17868930”, “SRR17868931”, “SRR17868932”, “SRR17868933”, 20. Lopez-Causape, C., Cabot, G., del Barrio-Tofino, E. & Oliver, A. The “SRR17868934” Source data are provided with this paper. versatile mutational resistome of Pseudomonas aeruginosa. Front. Microbiol. 9, 685 (2018). Code availability 21. Fish, D. N., Piscitelli, S. C. & Danziger, L. H. Development of resis- Code used for analyses are publicly available via Github [“https:// tance during antimicrobial therapy - a review of antibiotic classes github.com/juliofdiaz/Wheatley_DiazCaballero_etal”]. and patient characteristics in 173 studies. Pharmacotherapy 15, 279–291 (1995). References 22. Chung, H., et al. Rapid expansion and extinction of antibiotic 1. Jorth, P. et al. Regional isolation drives bacterial diversification resistance mutations during treatment of acute bacterial respira- within cystic fibrosis lungs. Cell Host Microbe 18, 307–319 (2015). tory infections. Nat. Commun. 13, 1231 (2022). 2. Chung, H. et al. Global and local selection acting on the pathogen 23. Wheatley, R. et al. Rapid evolution and host immunity drive the rise Stenotrophomonas maltophilia in the human lung. Nat. Commun. and fall of carbapenem resistance during an acute Pseudomonas 8, 14078 (2017). aeruginosa infection. Nat. Commun. 12, 12 (2021). Nature Communications | (2022)13:6523 9 Article https://doi.org/10.1038/s41467-022-34101-2 24. Hartl, D. & Clark, A. Principles of Population Genetics, 4th edn 45. Yayan, J., Ghebremedhin, B. & Rasche, K. Antibiotic resistance of (Sinauer Associates, 2007). Pseudomonas aeruginosa in pneumonia at a single university hos- 25. Paling, F. P. et al. Rationale and design of ASPIRE-ICU: a prospective pital center in Germany over a 10-year period. PLoS ONE 10, cohort study on the incidence and predictors of Staphylococcus e0139836 (2015). aureus and Pseudomonas aeruginosa pneumonia in the ICU. BMC 46. Tedijanto, C., Olesen, S. W., Grad, Y. H. & Lipsitch, M. Estimating the Infect. Dis. 17, 643 (2017). proportion of bystander selection for antibiotic resistance among 26. Didelot, X., Croucher, N. J., Bentley, S. D., Harris, S. R. & Wilson, D. J. potentially pathogenic bacterial flora. Proc. Natl Acad. Sci. USA 115, Bayesian inference of ancestral dates on bacterial phylogenetic E11988–E11995 (2018). trees. Nucleic Acids Res. 46, e134–e134 (2018). 47. Folkesson, A. et al. Adaptation of Pseudomonas aeruginosa to the 27. Li, H., Luo, Y.-F., Williams, B. J., Blackwell, T. S. & Xie, C.-M. Structure cystic fibrosis airway: an evolutionary perspective. Nat. Rev. and function of OprD protein in Pseudomonas aeruginosa: from Microbiol. 10, 841–851 (2012). antibiotic resistance to novel therapies. Int. J. Med. Microbiol. 302, 48. Lieberman, T. D. et al. Parallel bacterial evolution within multiple 63–68 (2012). patients identifies candidate pathogenicity genes. Nat. Genet. 43, 28. Albrecht, M. T. & Schiller, N. L. Alginate lyase (AlgL) activity is 1275–U1148 (2011). required for alginate biosynthesis in Pseudomonas aeruginosa. J. 49. Young, B. C. et al. Severe infections emerge from commensal Bacteriol. 187, 3869–3872 (2005). bacteria by adaptive evolution. Elife 6, 25 (2017). 29. Gursinsky, T., Jäger, J., Andreesen, J. R. & Söhling, B. A selDABC 50. Young, B. C. et al. Evolutionary dynamics of Staphylococcus aureus cluster for selenocysteine incorporation in Eubacterium acid- during progression from carriage to disease. Proc. Natl Acad. Sci. aminophilum. Arch. Microbiol. 174, 200–212 (2000). USA 109, 4550–4555 (2012). 30. Lebeuf-Taylor, E., McCloskey, N., Bailey, S. F., Hinz, A. & Kassen, R. 51. Liberati, A. et al. Antibiotic prophylaxis to reduce respiratory tract The distribution of fitness effects among synonymous mutations in infections and mortality in adults receiving intensive care. Cochrane a gene under directional selection. Elife 8, e45952 (2019). Database Syst. Rev. 2009, CD000022 (2000). 31. Andersen, S. B., Marvig, R. L., Molin, S., Johansen, H. K. & Griffin, A. 52. Wittekamp, B. H. et al. Decontamination strategies and bloodstream S. Long-term social dynamics drive loss of function in pathogenic infections with antibiotic-resistant microorganisms in ventilated bacteria. Proc. Natl Acad. Sci. USA 112, 10756–10761 (2015). patients: a randomized clinical trial. JAMA 320, 2087–2098 (2018). 32. De Maio, N., Wu, C.-H. & Wilson, D. J. SCOTTI: efficient recon- 53. De Smet, A. et al. Decontamination of the digestive tract and oro- struction of transmission within outbreaks with the structured pharynx in ICU patients. N. Engl. J. Med. 360, 20–31 (2009). coalescent. PLoS Comput. Biol. 12, e1005130 (2016). 54. Plantinga, N. L. et al. Selective digestive and oropharyngeal 33. Klinkenberg, D., Backer, J. A., Didelot, X., Colijn, C. & Wallinga, J. decontamination in medical and surgical ICU patients: individual Simultaneous inference of phylogenetic and transmission trees in patient data meta-analysis. Clin. Microbiol. Infect. 24, infectious disease outbreaks. PLoS Comput. Biol. 13, 32 (2017). 505–513 (2018). 34. Adewoye, L., Sutherland, A., Srikumar, R. & Poole, K. The mexR 55. EUCAST. http://www.eucast.org. The European Committee on repressor of the mexAB-oprM multidrug efflux operon in Pseudo- Antimicrobial Susceptibility Testing. EUCAST Reading Guide for monas aeruginosa: characterization of mutations compromising Broth Microdilution, Version 11.0 (2021). activity. J. Bacteriol. 184, 4308–4312 (2002). 56. EUCAST. http://www.eucast.org. The European Committee on 35. Hart, P. H. et al. Potential antiinflammatory effects of interleukin 4: Antimicrobial Susceptibility Testing. Breakpoint Tables for inter- suppression of human monocyte tumor necrosis factor alpha, pretation of MICs and Zone Diameters, Version 11.0 (2021). interleukin 1, and prostaglandin E2. Proc. Natl Acad. Sci. USA 86, 57. Bolger, A. M., Lohse, M. & Usadel, B. Trimmomatic: a flexible trim- 3803–3807 (1989). mer for Illumina sequence data. Bioinformatics 30, 36. Bielen, K. et al. Mechanical ventilation induces interleukin 4 secre- 2114–2120 (2014). tion in lungs and reduces the phagocytic capacity of lung macro- 58. Bankevich, A. et al. SPAdes: a new genome assembly algorithm and phages. J. Infect. Dis. 217, 1645–1655 (2018). its applications to single-cell sequencing. J. Comput. Biol. 19, 37. Jain-Vora, S. et al. Interleukin-4 enhances pulmonary clearance of 455–477 (2012). Pseudomonas aeruginosa. Infect. Immun. 66, 4229–4236 (1998). 59. Walker, B. J. et al. Pilon: an integrated tool for comprehensive 38. Aujla, S. J. et al. IL-22 mediates mucosal host defense against Gram- microbial variant detection and genome assembly improvement. negative bacterial pneumonia. Nat. Med. 14, 275–281 (2008). PLoS ONE 9, e112963 (2014). 39. Zheng, Y. et al. Interleukin-22 mediates early host defense against 60. He, J. et al. The broad host range pathogen Pseudomonas aerugi- attaching and effacing bacterial pathogens. Nat. Med. 14, nosa strain PA14 carries two pathogenicity islands harboring plant 282–289 (2008). and animal virulence genes. Proc. Natl Acad. Sci. USA 101, 40. Nicolau, D. P. Pharmacokinetic and pharmacodynamic properties of 2530–2535 (2004). meropenem. Clin. Infect. Dis. 47, S32–S40 (2008). 61. Seemann, T. Prokka: rapid prokaryotic genome annotation. Bioin- 41. San Millan, A. S. et al. Positive selection and compensatory adap- formatics 30, 2068–2069 (2014). tation interact to stabilize non-transmissible plasmids. Nat. Com- 62. Jolley, K. A., Bray, J. E. & Maiden, M. C. Open-access bacterial mun. 5, 11 (2014). population genomics: BIGSdb software, the PubMLST. org website 42. Andersson, D. I. & Hughes, D. Antibiotic resistance and its cost: is it and their applications. Wellcome Open Res. 3, 124 (2018). possible to reverse resistance? Nat. Rev. Microbiol. 8, 63. Wick, R. R., Judd, L. M., Gorrie, C. L. & Holt, K. E. Unicycler: resolving 260–271 (2010). bacterial genome assemblies from short and long sequencing 43. Gomez-Zorrilla, S. et al. Antibiotic pressure is a major risk factor for reads. PLoS Comput. Biol. 13, e1005595 (2017). rectal colonization by multidrug-resistant pseudomonas aerugi- 64. Li, H. et al. The sequence alignment/map format and SAMtools. nosa in critically Ill Patients. Antimicrob. Agents Chemother. 58, Bioinformatics 25, 2078–2079 (2009). 5863–5870 (2014). 65. Langmead, B. & Salzberg, S. L. Fast gapped-read alignment with 44. Papp-Wallace, K. M., Endimiani, A., Taracila, M. A. & Bonomo, R. A. Bowtie 2. Nat. Methods 9, 357 (2012). Carbapenems: past, present, and future. Antimicrob. Agents Che- 66. Li, H. & Durbin, R. Fast and accurate short read alignment with mother. 55, 4943–4960 (2011). Burrows–Wheeler transform. Bioinformatics 25, 1754–1760 (2009). Nature Communications | (2022)13:6523 10 Article https://doi.org/10.1038/s41467-022-34101-2 67. McKenna, A. et al. The Genome Analysis Toolkit: a MapReduce S.M.K. and R.C.M. contributed to project conception and study design. framework for analyzing next-generation DNA sequencing data. R.M.W., J.D.C., S.K.S., S.M.K. and R.C.M. wrote and revised the Genome Res. 20, 1297–1303 (2010). manuscript. 68. Deatherage, D. E. & Barrick, J. E. Engineering And Analyzing Multi- cellular Systems. p. 165-188 (Springer, 2014). Competing interests 69. Yu, G., Smith, D. K., Zhu, H., Guan, Y. & Lam, T. T. Y. ggtree: an R The authors declare no competing interests. package for visualization and annotation of phylogenetic trees with their covariates and other associated data. Methods Ecol. Evol. 8, Additional information 28–36 (2017). Supplementary information The online version contains 70. Gabrielaite, M. & Marvig, R. L. GenAPI: a tool for gene absence- supplementary material available at presence identification in fragmented bacterial genome sequen- https://doi.org/10.1038/s41467-022-34101-2. ces. bioRxiv https://doi.org/10.1101/658476 (2019). 71. Ståhlberg, A. et al. Simple, multiplexed, PCR-based barcoding of Correspondence and requests for materials should be addressed to R. DNA enables sensitive mutation detection in liquid biopsies using Craig MacLean. sequencing. Nucleic Acids Res. 44, e105–e105 (2016). 72. Li, G., et al. Genomic analyses of multidrug resistant Pseudomonas Peer review information Nature Communications thanks Camilo Bar- aeruginosa PA1 resequenced by single-molecule real-time bosa and the other anonymous reviewer(s) for their contribution to the sequencing. Biosci. Rep. 36, e00418 (2016). peer review of this work. 73. Didelot, X., Croucher, N. J., Bentley, S. D., Harris, S. R. & Wilson, D. J. Bayesian inference of ancestral dates on bacterial phylogenetic Reprints and permissions information is available at trees. Nucleic Acids Res. 46, e134 (2018). http://www.nature.com/reprints Acknowledgements Publisher’s note Springer Nature remains neutral with regard to jur- This research was supported by Wellcome Trust Grant (106918/Z/15/Z) isdictional claims in published maps and institutional affiliations. and the Innovative Medicines Initiative Joint Undertaking under COMBACTE-MAGNET (Combatting Bacterial Resistance in Europe- Open Access This article is licensed under a Creative Commons Molecules against Gram-negative Infections, grant agreement no. 115737) Attribution 4.0 International License, which permits use, sharing, and COMBACTE-NET (Combatting Bacterial Resistance in Europe-Net- adaptation, distribution and reproduction in any medium or format, as works, grant agreement no. 115523), resources of which are composed of long as you give appropriate credit to the original author(s) and the financial contribution from the European Union’s Seventh Framework source, provide a link to the Creative Commons license, and indicate if Program (FP7/2007-2013) and EFPIA companies’ in kind contribution. We changes were made. The images or other third party material in this thank the Oxford Genomics Center (funded by Wellcome Trust Grant article are included in the article’s Creative Commons license, unless 203141/Z/16/Z) for the generation and initial processing of Illumina indicated otherwise in a credit line to the material. If material is not sequence data. We thank the local ASPIRE ICU research team for their included in the article’s Creative Commons license and your intended contribution to this project at Hospital Universitari Germans Trias i Pujol use is not permitted by statutory regulation or exceeds the permitted and the WP3A working group for assistance with project. use, you will need to obtain permission directly from the copyright holder. To view a copy of this license, visit http://creativecommons.org/ Author contributions licenses/by/4.0/. R.M.W., J.D.C., T.E.vdS., F.H.R.D.W., L.P.S., N.K., C.R., L.T. and A.L. con- tributed to data acquisition and analysis. J.K., M.E., C.P.A., A.O., S.K.S., © The Author(s) 2022 Nature Communications | (2022)13:6523 11