Carbapenem Resistance Determinants Acquired through Novel Chromosomal Integrations in Extensively Drug-Resistant Pseudomonas aeruginosa

ABSTRACT Two novel blaDIM-1- or blaIMP-1-containing genomic islands (GIs) were discovered by whole-genome sequence analyses in four extensively drug-resistant (XDR) Pseudomonas aeruginosa isolates from inpatients at a tertiary hospital in Ghana. The strains were of sequence type 234 (ST234) and formed a phylogenetic clade together with ST111, which is recognized as a global high-risk clone. Their carbapenem resistance was encoded by two Tn402-type integrons, In1592 (blaDIM-1) and In1595 (blaIMP-1), both carrying complete tni mobilization modules. In1595 was bound by conserved 25-bp inverted repeats (IRs) flanked by 5-bp direct repeats (DRs) associated with target site duplication. The integrons were embedded in two GIs that contained cognate integrases and were distinguished by a lower GC content than the chromosomal average. PAGI-97A (52.659 bp; In1592), which encoded a P4-type site-specific integrase of the tyrosine recombinase family in its 3′ border, was integrated into tRNA-Pro(ggg) and bracketed by a 49-bp perfect DR created by 3′-end target duplication. GIs with the same structural features, but diverse genetic content, were identified in 41/226 completed P. aeruginosa genomes. PAGI-97B (22,636 bp; In1595), which encoded an XerC/D superfamily integrase in its 5′ border, was inserted into the small RNA (sRNA) PrrF1/PrrF2 locus. Specific insertions into this highly conserved locus involved in iron-dependent regulation, all leaving PrrF1 intact, were identified in an additional six phylogenetically unrelated P. aeruginosa genomes. Our molecular analyses unveiled a hospital-associated clonal dissemination of carbapenem-resistant ST234 P. aeruginosa in which the XDR phenotype resulted from novel insertions of two GIs into specific chromosomal sites.


RESULTS
XDR P. aeruginosa containing both bla IMP-1 and bla DIM-1 . Four P. aeruginosa strains originating from a collection of MDR clinical Gram-negative bacteria sampled in Ghana in 2015 (32) were selected for further studies due to their carbapenem-resistant phenotype. They were isolated from four different patients located in different wards at the same hospital during a 6-week time span (see Table S1 in the supplemental material). Antimicrobial susceptibility testing demonstrated resistance to agents from all categories of antipseudomonal antimicrobials except polymyxins, with some MIC values differing slightly between the strains (Table 1). According to standard criteria (33), the four strains were all categorized as XDR.
Whole-genome sequencing analyses of the four selected strains (Table 2) substantiated the phenotypic findings. All four strains encoded two carbapenemases (DIM-1 and IMP-1) revealing 100% identity to the corresponding prototypes, although three synonymous mutations were observed in bla IMP-1 . Additionally, they encoded two extended-spectrum-b-lactamases, OXA-10 and OXA-129, the latter being most closely related to OXA-5, which is known as a narrow-spectrum b-lactamase (34). Moreover, the strains contained mutations in GyrA (T83I) and ParC (S87L) that are typically involved in high-level fluoroquinolone resistance and are nearly universally associated with P. aeruginosa high-risk clones (16). Other acquired resistance genes included aadA1, aph(39)-IIb, aacA4, crpP, catB7, arr2, sul1, and dfrB5; in addition, aadA6 and qnrVC1 were present only in strains 97 and 140.
Independent acquisition of the carbapenemase-encoding determinants. An overall low prevalence of bla DIM-1 -positive (n = 4) and bla IMP-1 -positive (n = 15) P. aeruginosa genomes (listed in Table S2 in the supplemental material) were found by protein BLAST search in the NCBI database. Of particular interest, genome 1334/14 (bla DIM-1 ) from Poland belonged to the same sequence type determined for the four genomes in our study (ST234), although it was shown to carry bla DIM-1 on a genetic structure not present in the Ghanaian strains (35). The other genomes with bla DIM-1 or bla IMP-1 were assigned to 12 different STs, with none encoding both genes, implicating dissemination by horizontal transfer rather than by clonal spread.
P. aeruginosa core genome phylogeny (Fig. 1) confirmed a wide distribution of bla DIM-1 -and bla IMP-1 -encoding strains into various subclades of phylogenetic main groups 1 and 2. The genomes of the four Ghanaian strains formed a cluster, with genomes 1334/14 and AR_0103 (ST964; bla IMP-1 ) present in paraphyletic groups. Together with the ST111 genomes, they constitute a phylogenetic clade with a total of 10 strains carrying bla IMP-1 and/or bla DIM-1 (Fig. 1). Relatedness addressed by core genome single-nucleotide polymorphisms (SNPs) confirmed clonality between the genomes from our study (25 to 190 SNPs), whereas more distant relations to 1334/14 (;1,500 SNPs), AR_0103 (;21,500 SNPs), and AG1 (ST111; ;25,500 SNPs), representing the other defined paraphyletic groups in the ST111 clade, were confirmed. Taken together, these results point to multiple independent acquisitions of bla DIM-1 and bla IMP-1 followed by clonal expansions, also within the ST111 clade.
Tn402-type class 1 integrons containing bla DIM-1 or bla IMP-1 . The bla IMP-1 -and bla DIM-1 -containing contigs were identical for all four genomes (2,659 and 3,093 nt, respectively) and gave limited information regarding genetic context. To obtain further insight into the acquisition of these MBL-encoding genes, one of the genomes was completed by Oxford Nanopore Technologies (ONT) long-read sequencing. The genome of strain 97 had a chromosome size of 6,925,889 bp, a GC content of 65.89%, and contained no plasmids.
A majority of the resistance genes were carried by five novel class 1 integrons (see Table S3 in the supplemental material) identified at different positions of the chromosome (Fig. 2). Of particular interest were In1592 (8,945 bp) and In1595 (8,592 bp), which contained bla DIM-1 and bla IMP-1 , respectively. Both carried a conserved tni module (92% identical to each other) encoding complete arrays of transposition genes (tniR, tniQ, tniB, and tniA) originally detected in Tn402-type integrons (36,37). Moreover, In1595 was delineated by 25-bp inverted repeats (IRs), 100% identical to the IRs reported for the complete versions of Tn402-related elements (36). The presence of 5-bp DRs   The tree is based on closed genomes with the addition of all bla IMP-1 and bla DIM-1 whole-genome sequence data (n = 243) available in GenBank. Phylogenetic main groups 1, 2, and 3 (red, blue, and green, respectively) and members of the sequence type 111 (ST111)-containing clade are highlighted (yellow). For each genome, assembly number, multilocus sequence type (MLST, innermost ring; color codes are shown for STs assigned to $5 genomes), and presence of bla IMP-1 (triangle), bla DIM-1 (black square), or both (star) are shown. In the outer ring, genomic islands (GIs) integrated into small RNA (sRNA) PrrF1/PrrF2 (circles) or tRNA-Pro (squares) are shown with color codes referring to the type of recombinase/ integrases (XerD or Int) or to the intG phylogroup, respectively. The tree was midpoint routed and metadata were added using iTOL (https://itol.embl.de/).
(TCAAA) bracketing the IRs strongly indicate a small target duplication associated with a transposition event. Single copies of the conserved 25-bp sequence were detected in the 39 or 59 ends of the four additional integrons (Table S3). Two chromosomally integrated carbapenemase-encoding genomic islands. To investigate their insertion sites as well as genetic content, the bla DIM-1 -and bla IMP-1 -containing genetic elements were delineated by sequence alignments. Among the completed genomes phylogenetically related to strain 97, 1334/14 was selected for pairwise alignments due to the high DNA identity in the flanking regions of both In1592 and In1595, as revealed by BLAST analyses. Flanked by 100% identical DNA, two FIG 2 Chromosome of P. aeruginosa strain 97 used as reference for BLAST alignments. In the circularized chromosome, CG skew and GC content are indicated (two innermost circles), as well as annotated coding sequences (CDS; red arrows, outermost circle) with acquired resistance genes (lilac). Arcs showing GIs (orange), integrons (green), intact (dark gray), or questionable (light gray) prophages are shown in the next circles. In the BLAST comparisons with the selected genomes from phylogenetic main groups 1 (red), 2 (blue), and 3 (green), DNA identity (50 to 100%) to the reference is visualized by the given color codes, with missing regions appearing as white. The map was constructed using BRIG software (http://brig.sourceforge.net/). distinct insertions were identified in strain 97 compared to 1334/14, PAGI-97A (52,659 bp) and PAGI-97B (22,636 bp) (Fig. 3). As is characteristic for GIs, their GC contents of 54.7% (PAGI-97A) and 57.3% (PAGI-97B) were markedly different from the genome average, and GC peaks corresponding to the two GIs (positions 2,622,067 to 2,674,677 and 5,905,810 to 5,928,446) could be identified in the chromosome map (Fig. 2). We did not recognize any conjugation genes among their annotated coding sequences (CDS), and no origin of transfer was identified in their DNA sequences, showing that these GIs do not belong to the ICEs.
NCBI BLAST alignments using the chromosome of strain 97 as a reference (Fig. 2) revealed additional sections with no DNA identity to 1334/14. These can be interpreted as RGP, known to cluster at limited chromosomal loci in P. aeruginosa (25). Other putative RGP were identified when comparing to less related genomes from phylogenetic main groups 1 (AG1; PAO1), 2 (PA14), and 3 (PA7). Several of these regions corresponded to prophages, one of which was situated in the same highly variable region of the chromosome as PAGI-97A. BLAST comparisons showed high DNA identity of the region immediately downstream of PAGI-97A among the ST234 genomes available in the PATRIC database (n = 11), suggesting a recent acquisition of this island. PAGI-97B, on the other hand, was embedded in a more conserved region of the chromosome.
Site-specific integration of PAGI-97A into tRNA-Pro(ggg). Bacterial tRNA genes have commonly been recognized as target sites for chromosomal integration of various genetic elements by site-specific recombinases (38). Here, the tRNA-Pro(ggg) gene locus, which was present in a single copy, was identified as the insertion point for PAGI-97A (Fig. 3A). In accordance with previously described tRNA integrations (39)  Novel Integrations of Genomic Islands in P. aeruginosa Antimicrobial Agents and Chemotherapy the GI. The GI encoded a cognate integrase, IntG, containing a C-terminal P4-type sitespecific integrase of the tyrosine recombinase family, as recognized by NCBI Conserved Domain Database (CDD), in its 39 border. Additionally, the GI carried resistance genes contained by In1592 (bla DIM-1 ) and In1591 (sul1, qnrVC1, qacED1, and aadA6), genes encoding a toxin-antitoxin system (vapBC) and a viral recombinase (yqaJ), insertion sequence (IS) elements, and, additionally, genes encoding mostly unknown hypothetical proteins, for a total of ;50 CDS. NCBI BLAST search showed that this GI was unique to the four genomes in our study. However, an examination of the completed P. aeruginosa genomes (n = 226) available in the NCBI database, identified chromosomal elements with structural features similar to PAGI-97A, but differing in genetic cargo, in a total of 41 strains (listed in Table S2 in the supplemental material). Specific for this group of GIs was that they carried intG in the 39 terminus and were delineated by the 49-bp DR identified at different distances (21.2 to 102.9 kbp) from tRNA-Pro(ggg). In the core genome tree, they were found in a variety of clades in phylogenetic main groups 1 and 2 (Fig. 1), implicating horizontal spread.
To explore the dissemination and relatedness of these GIs with regard to genetic cargo as well as their cognate integrases, we constructed phylogenetic trees based on their complete nucleotide sequence or on intG only (Fig. 4). The tree revealed an overall high level of diversity of the GIs (Fig. 4A), with several distinct clusters as well as unique ones, like PAGI-97A from strain 97. GIs from the same clusters were mainly present in phylogenetically highly related genomes (indicated in Fig. 1) and might be regarded as clonally disseminated homologs. Interestingly, within some of these genome clusters, which also belong to the same ST, strains where the specific GI is absent can be observed. This implicates recent gain or loss of the GI and therefore integration/excision activities.
Additional comparisons of the GIs were conducted through phylogeny of their cognate integrase (Fig. 4B), defining clade A with clusters (I and II) and subclusters (i and ii), and clade B. Highly related GIs contained intG from either clade B or one of the clade A subclusters, and further subgrouping of A-Iii correlated exactly to GI subclusters c1 to c7 (Fig. 4A). However, the overall relatedness of the GIs, including that of the c1 to c7 clusters, did not correlate with the intG phylogeny, indicating independent evolution and clonal relatedness. Moreover, GIs from some of the groups (B, A-Iii, and A-IIi-c3) were found in unrelated genomes, which suggests that these were independently acquired.
The DNA identity was $99% within clade B and the clade A subclusters of intG and $92.3% between the clade A subclusters. Surprisingly, intG from clades A and B displayed no detectable DNA homology in BLAST alignment. At the amino acid level, they revealed 40% identity and, importantly, shared the C-terminal P4-like integrase domain. These findings disclose additional complexity of the evolutionary relationship of this group of GIs.
Taken together, these GIs share a mechanism for site-specific integration that is mediated by their inherent integrase, which renders tRNA-Pro(ggg) as their specific recognition site and thereby a location for genome plasticity in P. aeruginosa.
The conserved sRNA PrrF1/PrrF2 loci as a genomic integration site for PAGI-97B. In its 59 border (Fig. 3B), a XerC/D recombinase containing a C-terminal DNA breaking-rejoining catalytic domain, known to be involved in unidirectional site-specific DNA recombination, was recognized by CDD search. Additional genetic content included In1595 (bla OXA-10 , aacA4, and bla IMP-1 ), a type I restriction system (hsdR and hsdM), and hypothetical proteins with unknown functions, for a total of ;20 CDS. The GI was integrated exactly downstream of the PrrF1-encoding gene of the intergenic PrrF1/PrrF2 locus (Fig. 5A), which is involved in iron-dependent homeostasis (40). The locus encodes the two independently expressed small RNA (sRNA; 92% identical) and forms imperfect DRs of 159/156 bp separated by 53 bp. In strain 97, the 59-end part of the locus, including PrrF2, was missing due to the insertion of PAGI-97B.
Although NCBI BLAST search confirmed that the PrrF1/PrrF2 locus is highly conserved in P. aeruginosa, integrations (;20 to 50 kb) were detected in six additional, phylogenetically unrelated genomes. DNA alignments that included the wild-type locus represented by 1334/14 enabled a precise detection of insertion point and delineation and comparison of these GIs (Fig. 5B). Two more genomes had GIs (AG1, 21,154 bp; NCTC13715, 31,074 bp) that carried xerD in the 59 end, inserted into the same site (Fig. 5B). They exhibited extensive sequence identity to PAGI-97B, except for the integron gene cassette regions. AG1 belongs to the ST111 clade, whereas NCTC13715 is phylogenetically unrelated to strain 97.
It is noteworthy that in AG1, the complete 59-end part of the PrrF1/PrrF2 locus was detected downstream of the insertion. This could implicate that the GIs originally were integrated between PrrF1 and PrrF2, and the downstream part of the locus was lost thereafter in strain 97 and NCTC13715. It is also of interest that truncations of the 59-end part of the locus were discovered at a position identical to that of the GI integration site in the genomes of Pa127 (GenBank assembly accession number GCF_002205355.1) and ATCC 27853 (accession number GCF_001687285.1). This might be explained by loss of integrated DNA by homologous recombination between the two DRs. In PA59, DNA (47,568 bp) was inserted in the same manner as in AG1, although this GI carried int in the 59 end and had no sequence identity to the ones described above. The alignments showed, however, extensive DNA identity, including that of the 59-end int, to the corresponding insertions in the phylogenetically unrelated genomes of E90 (43,995 bp), NCGM2.S1 (47,189 bp), and VRFFP04 (60,277 bp). Adding to the complexity, two of these GIs were inserted into a site almost identical to PAGI-97B, exactly downstream of PrrF2, truncating the last 8 bp of the second DR of this locus, while in VRFFP04, the GI was inserted upstream of the last codon of PrrF2 (Fig. 5A).
In summary, we have identified two phylogenetically unrelated groups of GIs, encompassing different enzymes anticipated to be involved in their integration into the same highly conserved intergenic locus, which forms imperfect DRs. The locus was disrupted downstream of either PrrF1 or PrrF2, keeping one or both of these 90% identical regulatory sRNA genes intact.

DISCUSSION
The global dissemination of carbapenemase-producing P. aeruginosa is facilitated by an interplay between mobile genetic elements and successful clones (12,31,41). Our study identified dissemination of an XDR ST234 clone phylogenetically related to the ST111 high-risk clone, having two horizontally acquired carbapenemase-encoding determinants, bla DIM-1 and bla IMP-1 , embedded in the chromosome. DIM is a relatively rare MBL and was initially described in a clinical isolate of Pseudomonas stutzeri in The Netherlands (42), and thereafter in clinical isolates of Pseudomonadaceae, Enterobacteriaceae, Burkholderiaceae, and Comamonadaceae in Sierra Leone (43), and Pseudomonas putida in China (44). Isolation of an additional clinical P. stutzeri strain in India with bla DIM-1 contained in a genetic environment different from that of the index strain led to the suggestion of an unidentified environmental source of this MBL (45). Coproduction of more than one MBL has recently been reported in clinal P. aeruginosa strains in Iran (46) but is not commonly found.
Unlike the class 1 integrons that are generally associated with carbapenemaseencoding genes in P. aeruginosa (9,13,41,47), both bla DIM-1 and bla IMP-1 were carried by integrons possessing complete tni modules with boundaries defined by inverted repeats, features typical for Tn402-like class 1 integrons (36). For In1595, the observed 5-bp target duplication indicates a recent acquisition within PAGI-97B in strain 97, which is also supported by the absence of a corresponding integron in a homologous GI in P. aeruginosa AG1. The four tni genes enable active transposition to res sites of Tn21 subfamily transposons but also to resolution sites found on plasmids, and are implicated in integron dissemination (10,48,49). Integrons of this type, also termed "preclinical class 1 integrons," are suggested to be key evolutionary intermediates still circulating in the environment and capable of acquiring gene cassettes (50). Variants denoted "unusual" class 1 integrons, due to the lack of the 39 conserved sequence (qacED1-sul1) found in the vast majority of class 1 integrons in clinically relevant bacteria, have been reported in clinical P. aeruginosa and P. putida isolates, where they contained bla VIM-2 and aacA4 (51,52).
GIs play a key role in prokaryotic genome plasticity and are recognized in P. aeruginosa as a major factor in acquisition and dissemination of multiple antibiotic resistance genes (31,35,(53)(54)(55), as well as in virulence traits (56)(57)(58). Our bioinformatic analyses revealed that both chromosomal regions containing the carbapenemase-encoding integrons exhibited GI features (57,59,60), including size, decreased GC content, presence of integrase/recombinase genes, and, for PAGI-97A, integration into a tRNA gene and flanking by DRs. In line with our findings, GIs usually carry genes encoding hypothetical proteins whose functions are unknown (30,61). The two GIs were uniquely detected in the Ghanaian clone and are absent in highly related genomes, including 1334/14, which belongs to the same ST234 lineage, and have 100% protein identity within 5,700 CDS (35), which argues for a recent acquisition by horizontal transfer from an unknown donor source.
Mobile genetic elements that integrate into the host chromosomes either carry their own DNA integration machineries or exploit machineries already existing in the bacterial host (10). Integrases associated with GIs form a separate clade within the tyrosine recombinase family and are frequently located at the 59 border of the GI, close to the tRNA locus or the respective attachment site (59,60,62). In P. aeruginosa, GI integrations have been detected in defined tRNA genes, including tRNA-Gly, tRNA-Thr, and tRNA-Phe, but also in structural genes such as hrpB (53,57,58,63). Our study revealed tRNA-Pro(ggg) as an additional site for specific GI integrations. Although these GIs belonged to several phylogroups based on genetic content, they all encoded P4-type integrases of the tyrosine recombinase family in their 39 rather than 59 border, and were bracketed by 49-bp DRs resulting from 39-end tRNA-Pro(ggg) target duplications. With these specific features in common, they might be regarded as a distinct family of GIs with specificity determined by their common P4 integrase. In line with our observations of phylogenetically related genomes revealing presence or absence for a specific GI in this site is the reported instability due to the flanking DRs, which are often homologous to phage attachment sites and promote integration and excision (64,65).
Three genetically related XerD-encoding GIs were inserted into the highly conserved sRNA PrrF1/PrrF2 locus. This intergenic locus encodes tandem sRNAs, which are .95% identical to each other and independently expressed, with an overlapping central function in the regulation of iron homeostasis and pathogenesis in P. aeruginosa (40,66). Despite the GIs, their important regulatory functions might be restored due to the insertion sites directly downstream of the PrrF1 gene. In the AG1 chromosome, the complete 39 part of the locus is present downstream of the GI, which positions the two imperfect DRs that constitute the PrrF1 and PrrF2 genes at the GI boundaries, where they might facilitate the Xer-mediated excision and integration. It can be speculated that truncation after integration explain the absence of the 39 part of the locus downstream of the GIs in strains 97 and NTC13715, and that the DRs themselves might be of importance for the integration activity. However, the finding of an unrelated group of GIs inserted into the locus at additional positions, and where XerD is replaced by a different integrase, add complexity that needs to be resolved by future studies.
Main conclusions. The presence of class 1 integrons with complete tni mobilization modules located on GIs in high-risk P. aeruginosa strains are of great clinical concern with regard to the dissemination of carbapenem resistance. Moreover, the molecular characterization of the two carbapenemase-encoding GIs and their chromosomal integration sites has added two novel locations for RGP in P. aeruginosa and enabled the identification of additional GIs with integration specificities toward these sites. Finally, our findings emphasize the importance of GIs in the dissemination of antibiotic resistance determinants in P. aeruginosa. Bacterial characterization and selection. Relevant patient data and sources of specimens for the four carbapenem-resistant P. aeruginosa strains selected for this study are shown in Table S1 in the supplemental material. The strains were isolated from four patients in three different wards and originated from a collection of 200 clinical, nonduplicate, Gram-negative bacterial isolates sampled from inpatients at the Komfo Anokye Teaching Hospital in Ghana between February and August 2015 (32). Isolates were identified by the Vitek 2 (bioMérieux) automated system and confirmed by matrix-assisted laser desorption ionization-time of flight mass spectrometry (MALDI-TOF MS; Bruker Daltonic Gmbh). Antimicrobial susceptibility testing was conducted by broth microdilution (Sensititre, Trek Diagnostic Systems) and interpreted according to European Committee on Antimicrobial Susceptibility Testing (EUCAST) guidelines (67,68).

MATERIALS AND METHODS
DNA sequencing. For Illumina sequencing, DNA was extracted using the PureLink Microbiome DNA purification kit according to the instructions of the manufacturer (Invitrogen). DNA libraries were generated using the Nextera kit and sequenced on an Illumina NextSeq 550 system at the Genomics Support Centre Tromsø (Norway). For long-read sequencing by MinION (Oxford Nanopore Technologies; ONT), genomic DNA was purified using the Qiagen Genomic-tip 100/G kit, and a DNA library prepared using a rapid kit (SQK-RAD001) and sequenced on a R9.4 flow cell (FLO-MIN106) supplied by ONT, all according to the manufacturer's instructions.
Global phylogeny was constructed using Parsnp v1.2 with the "-c" flag enabled, using strain 97 as the reference on all the completed P. aeruginosa genomes, together with bla IMP-1 and bla DIM-1 -containing genomes (n = 243, listed in Table S2 in the supplemental material). The genome assemblies used in the phylogenetic and comparative analyses, including corresponding metadata, were downloaded (16 April 2020) from the Pathosystems Resource Integration Center (PATRIC) database (https://www.patricbrc .org/). FigTree (http://tree.bio.ed.ac.uk/software/figtree/) was used to visualize and curate the phylogenetic trees, and metadata were coupled to specific genomes using iTOL (https://itol.embl.de/).
Accession number(s). The raw read sequences and the assembled whole-genome contigs for strains 97, 130, 140, and 142 have been deposited in GenBank under BioProject accession number PRJNA411997. The closed genome of strain 97 has accession number NZ_CP031449.2.

SUPPLEMENTAL MATERIAL
Supplemental material is available online only. SUPPLEMENTAL FILE 1, PDF file, 0.3 MB.

ACKNOWLEDGMENTS
We are grateful to the Komfo Anokye Teaching Hospital staff, especially the technical staff in the microbiology laboratory for their support during sample collection and preliminary identification, as well as the physicians and nursing staff in various wards for their cooperation. We thank the staff of National Health Laboratory Services (NHLS), Albert Luthuli Hospital, South Africa, for further bacteria identification and the initial antimicrobial susceptibility testing. We are grateful to Bjørg Christina Haldorsen for performing the broth microdilution and to Bettina Aasnaes and Ellen Josefsen for technical assistance with DNA sequencing, all three at the Norwegian National Advisory Unit on Detection of Antimicrobial Resistance, Department of Microbiology and Infection Control, University Hospital of North Norway, Tromsø, Norway. We acknowledge the Genomics Support Centre Tromsø (GSCT) at the Arctic University of Norway for performing the Illumina sequencing.
T.P., J.J., Ø.S., and A.S. conceived and designed the study. T.P. and J.J. conducted the research and investigation process, analyzed the data, and prepared the figures. N.A. collected and identified the P. aeruginosa strains and performed initial antimicrobial susceptibility testing, supervised by S.Y.E., U.G., and A.O.-O. The manuscript draft was prepared by T.P. and J.J., and all authors contributed to reviewing and editing and approved submission of the manuscript for publication.
S.Y.E. is chairperson of the Global Respiratory Infection Partnership and a member of the Global Hygiene Council, both sponsored by unrestricted educational grants from Reckitt and Benckiser (Pty.), United Kingdom. The other authors have no conflicts of interest to declare.