Monitoring of Pseudomonas aeruginosa mutational resistome dynamics using an enrichment panel for direct sequencing of clinical samples

Abstract

Background: Pseudomonas aeruginosa is a major cause of hospital-acquired and chronic infections, characterised by an extraordinary capacity to develop antimicrobial resistance through the selection of chromosomal mutations, leading to treatment failure. Here, we designed and tested a hybridisation-based capture system for the enrichment of genes of interest before sequencing to monitor resistant populations genomics directly from clinical samples.

Methods: A panel for enrichment before sequencing of close to 200 genes related to P. aeruginosa antimicrobial resistance, multilocus sequence typing, mutability or virulence was designed, synthesised (KAPA HyperCap, Roche) and initially validated in vitro using a multidrug-resistant ST175 isolate and representative isolates from major P. aeruginosa clades. In vivo testing included ventilator associated pneumonia by MDR P. aeruginosa in ICU (3-10 sequential samples from 3 patients) and chronic respiratory infection by hypermutable P. aeruginosa in cystic fibrosis (8 sequential samples from a single patient covering a 4-year period). Results from direct sequencing with the enrichment panel were compared with those of whole genome sequencing (WGS) and phenotypic profiling of 10 isolated colonies per sample.

Findings: In vitro assays confirmed the selectivity of the enrichment panel and the correct identification of the vast mutational resistome of ST175, including specific mutations even when introduced in a 1:100 proportion. In vivo performance was at least equivalent to sequencing 10 colonies per sample, including the accurate identification of the sequence types and the basal and acquired mutational resistome. To note, specific resistance mutations, such as those in ampC leading to resistance to novel β-lactams, could be traced even at frequencies of 1%. Moreover, the coselection of mutator populations and antibiotic resistance mutations, predicted in theoretical and in vitro studies, was evidenced in vivo.

Interpretation: This proof-of-concept study demonstrates that resistance genomics of P. aeruginosa can be analysed directly from clinical samples, determining not only a considerable reduction in turnaround time and cost from a diagnostics perspective, but also an unprecedented potency for accurate monitoring of in vivo population dynamics in bacterial infections.

Funding: Instituto de Salud Carlos III, Ministerio de Ciencia e Innovación and Unión Europea-NextGenerationEU.

Keywords: Antimicrobial resistance development; Chronic infections; Cystic fibrosis; Nosocomial infections; Pseudomonas aeruginosa; Resistome.

Fig. 1

Wet-lab and dry-lab workflow overview of the in vitro and in vivo validation of the P. aeruginosa resistome enrichment panel.

Fig. 2

Results obtained on the in vitro evaluation of the enrichment panel. (a) PAO1 reference genome coverage and sequencing depth with and without adding the capture hybridisation step during the library preparation workflow. (b) Sequencing depth comparison of major resistome mutations. Number of aligned bases supporting the indicated mutation and number of total aligned bases in the specific position are shown.

Fig. 3

Summary of results for patient ICU-06. A total of six respiratory samples collected during this episode were studied (days 1, 7, 9, 13, 28 and 30). The key basal and acquired resistome (transferable and mutational) is shown, comparing the results obtained with the enrichment panel with those obtained from the analysis of the saved individual colonies. Samples lacking isolated colonies (or when their number is lower than 10) are represented with blank spaces. For the enrichment panel, the percentage of aligned bases supporting each indicated mutation are shown, along with the sequencing depth at that specific position. Transferable resistance determinants were not analysed with the enrichment panel. For the saved colonies, the ST, the detected mutations, and the antibiotic susceptibility profile are shown. P. aeruginosa load in each of the respiratory samples, determined by quantitative cultures and 16S rRNA sequencing, are also indicated. Finally, targeted treatments for P. aeruginosa are represented. This patient was admitted at Son Espases ICU on 28th January 2020 with an S. aureus infective endocarditis for urgent valve surgery; after surgery, the patient required mechanical ventilation and on 24th February 2020 (day 1), they developed a MDR-P. aeruginosa VAP. Ceftolozane/tazobactam was initiated on day 3 and extended to day 15, when treatment needed to be changed to ceftazidime plus inhaled colistin due to lack of ceftolozane/tazobactam stock at the hospital.

Fig. 4

Summary of results for patient ICU-08. Three respiratory samples were collected in this patient (days 1, 3 and 10). The key basal and acquired resistome (transferable and mutational) is shown, comparing the results obtained with the enrichment panel with those obtained from the analysis of the saved individual colonies. Samples lacking isolated colonies (or when their number is lower than 10) are represented with blank spaces. For the enrichment panel, the percentage of aligned bases supporting each indicated mutation are shown along with the sequencing depth at that specific position. Transferable resistance determinants were not analysed with the enrichment panel. For the saved colonies, the ST, the detected mutations, and the antibiotic susceptibility profile are shown. P. aeruginosa load in each of the respiratory samples, determined by quantitative cultures and 16S rRNA sequencing, are also indicated. Finally, targeted treatments for P. aeruginosa are represented. This patient was admitted on 8th June 2020 presenting a multi-organic failure and a community acquired pneumonia (E. coli and S. pneumoniae) for which ceftriaxone was initiated. The patient worsened requiring mechanic ventilation and, on day 12th June (day 1), a MDR-P. aeruginosa VAP was diagnosed. Ceftolozane/tazobactam was initiated on day 2 and maintained till day 10 when respiratory symptoms resolved.

Fig. 5

Summary of results for patient ICU-09. Ten respiratory samples were collected from this patient (days 1, 8, 11, 21, 31, 39, 45, 53, 62 and 78). The key basal and acquired resistome (transferable and mutational) is shown, comparing the results obtained with the enrichment panel with those obtained from the analysis of the saved individual colonies. For the enrichment panel, the percentage of aligned bases supporting each indicated mutation are shown along with the sequencing depth at that specific position. Transferable resistance determinants were not analysed with the enrichment panel. For the saved colonies, the ST, the detected mutations and the antibiotic susceptibility profile are shown. P. aeruginosa load in each of the respiratory samples, determined by quantitative cultures and 16S rRNA sequencing, are also indicated. Finally, targeted treatments for P. aeruginosa are represented. ICU-09 was a onco-haematological patient who acquired a severe COVID-19 pneumonia with S. pneumoniae superinfection in late September 2020, requiring ICU admission and mechanical ventilation during 7 days. After recovering from this event, the patient was admitted at the Pneumology Department, but respiratory symptoms worsened again and the patient was re-admitted to ICU, requiring again mechanical ventilation. On 5th November 2020 (+1 month of ICU re-admission) a MDR P. aeruginosa was isolated from a respiratory sample and treatment with ceftolozane/tazobactam was initiated for 15 days. Unfortunately, respiratory symptoms did not improve and MDR P. aeruginosa could not be eradicated, despite the administration of several antibiotic treatments and the patient was exitus.

Fig. 6

Summary of results for patient CF-02. A total of eight respiratory samples from this patient, obtained from August 2019 to February 2023, were collected (days 1, 58, 91, 185, 275, 744, 1243 and 1303). A summary of the key basal and acquired resistome (transferable and mutational) is shown, comparing the results obtained with the enrichment panel with those obtained from the analysis of the saved individual colonies. For the enrichment panel, the percentage of aligned bases supporting each indicated mutation are shown along with the sequencing depth at that specific position. Transferable resistance determinants were not analysed with the enrichment panel. For the saved colonies, the ST, the detected mutations, and the antibiotic susceptibility profile are shown. P. aeruginosa load in each of the respiratory samples, determined by quantitative cultures and 16S rRNA sequencing, are also indicated. Finally, chronic (black) and targeted treatments for P. aeruginosa exacerbations (blue) received during this period are indicated.

Fig. 7

Analysis of the number mutations detected in patient CF-02. Bars indicate the total number of mutations detected in each respiratory sample. Numbers of these mutations detected only from the colonies, only from the enrichment panel or in both approaches are also indicated. This figure was plotted using the ggplot2 package in R v4.3.2 and RStudio v2023.09.1 + 494.

Fig. 8

Analysis of the overtime dynamics of the selection of ampC mutations and PDC variants in patient CF-02. a. Number of colonies showing each of the different PDC variants detected for each of the respiratory samples analysed. b. Percentages of aligned bases obtained from the totalpileup files yielding each of the ampC mutations detected with the enrichment panel in each of the respiratory samples. These figures were plotted using the ggplot2 package in R v4.3.2 and RStudio v2023.09.1 + 494. c. 3D structure of PAO1 AmpC (PDB-4GZB) indicating the location of the detected mutations. AmpC β-lactamase domains (SVSK, YSN, KTG) and the Ω-loop (residues 229–247) are coloured in black. Representations were performed with the PyMOL Molecular Graphic System v2.3 (http://www.pymol.org/pymol).

Fig. 9

Overtime dynamics of mutS deficient mutator lineages and their association with antibiotic resistance mutations in patient CF-02. Percentage of aligned bases obtained from the totalpileup files for each of the mutations detected with the enrichment panel in each of the respiratory samples are represented. The study of the association between mutS and resistance genes (oprD, dacB, ampC, ftsI and pmrB) mutations in the time series was performed using the cross-correlation algorithm ccf, included in the core distribution of R (R v4.3.2 (RStudio version 2023.09.1–494). Cross-correlation coefficient values with a confidence interval over 95% (P value < 0.05) calculated with stats available in R v4.3.2 are also shown. This figure was plotted using the ggplot2 package in R v4.3.2 and RStudio v2023.09.1 + 494.