Antibiotic multi-resistance is a major threat to public health because continuous emergence, worldwide spread, and increasing prevalence (17). Unlike highly host adapted pathogens and symbionts undergoing genome reduction, as a versatile environmental organism, P. aeruginosa continually expands its genomic repertory (50). With a high-risk ST-111 profile, PaeAG1 is a critical organism due to its resistance to multiple antibiotics, including carbapenems (10). In order to compare PaeAG1 genome against other P. aeruginosa complete sequences, we performed a pan-genome analysis with 211 genomes. We were able not only to reveal that whole genome sequences could separate clones by ST profile (MLST), but also identification of unique, core and accessory genes was achieved. Clear clusters were found for all the MLST profiles (Fig. 1), including the ST-111 (Fig. 1, green, 10 genomes) and ST-235 (orange, including some strains with unknown ST profile) high-risk clones. The closest genomes to PaeAG1 were Pae-RIVM-EMC2982 and Pae-Carb0163.
Other pan-genome analysis in P. aeruginosa also found clusters than could be identified by the ST profile (51, 52). MLST is the most convenient method nowadays for the assessment of the phylogenetic relationships among the species in the genus Pseudomonas (53). This is possible not only at clinical setting but also in other environments (54, 55).
While multiple comparative genomic analyses (many using a pan-genome approach) have been reported for P. aeruginosa (39, 51, 60, 52–59), most of them include incomplete, fragmented or draft genomes, or sequences of few genes. In 2015, complete genomes were used in a similar approach, but only 17 genomes were available (NCBI), which only three corresponded to high-risk clones (57). Thus, our analysis provides an up-date of the general status of relationships of the 211 available complete genomes by pan-genome analysis.
In relation to gene content among all strains, we identified a total of 2726 genes as part of the core-genome (> 99% strains). This value and gene distribution of accessory-genome and unique elements in our study (histogram in Supplementary file 2 Pan-genome analysis results) is comparable to another similar approach, in which 2503 core genes were identified in a very robust approach (58). Other studies have suggested a higher number of core-genome genes (4000–5300) (52, 55–57), which it is similar to our results with 4659 genes if we include the soft-core-genome (> 95% strains). Other cases using > 1000 genomes have reported < 700 core genes (59, 60). In general, differences can be explained by selected genomes (number, origin, completeness) and implemented algorithms (55, 56). The relatively high number of conserved genes in the core-genome of P. aeruginosa can be associated with the ability to conquer multiple environments and to facilitate infectious capability towards a large set of hosts (57). According to functional analysis, 42 KEGG pathways (energy metabolism, nucleic acids, amino acids, ribosomal activity, and many others) were found as part of the enriched routes for all the core genes, with functions that are in line with other similar pan-genome studies (57, 58).
P. aeruginosa genome is composed of a mosaic structure including the large core-genome (57), into which regions of genomic plasticity lead to the insertion of block of genes belonging to the accessory genome (50). In the case of PaeAG1 and other ST-111 strains, genome sequence is around 1.0 Mb longer that the reference genome Pae-PAO1, difference that is reflect as genomic islands distributed along the genome.
In order to determinate the presence/absence of the 57 PaeAG1 genomic islands in other strains, a comparative analysis was done using 15 related genomes (ST-111 and closest genomes in the cluster of the pan-genome analysis, Fig. 1). Pae-RIVM-EMC2982 and Pae-Carb0163 (closest genomes to PaeAG1) had the most similar profiles carrying 41 out the genomic islands. As highlighted in Results, many genomic islands formed clusters (GICs, Figs. 2 and 3), including the genomic islands clusters harboring the two integrons (GICVIM−2 and GICIMP−18). Genomic islands groups have been reported before as integrative and conjugative elements or ICEs (16), but ICEs in PaeAG1 (using ICEberg 2.0 platform, https://db-mml.sjtu.edu.cn/ICEfinder/ICEfinder.html) overlap with other GICs but none with GICVIM−2 and GICIMP−18. Since size of the core-genome and its content is not well known (57), prediction methods are required to define accessory regions, but outcome depends on algorithms (56), which could explain differences and the GICs.
On the other hand, this prominent number of genomic islands in PaeAG1 and other ST-111 strains can be explained due to the absence of a functional CRISPR-Cas system (bacterial defense system against foreign DNA) and consequent high number of successful events of horizontal gene transfer (16). This genome plasticity of individual strains represents an advantage for P. aeruginosa to fit the needs for survival in virtually any environment (50). In addition, as reported for other high-risk clones (16), exclusive genomic islands are frequently found in P. aeruginosa (50). GI49 (harboring the IMP-18-carrying integron) is an example of an exclusive island able to horizontally transfer resistance genes.
In this sense, the bacterial accessory genome, including class 1 integrons, can have important clinical implications such as wide distribution and spread of antimicrobial resistance (36, 37). Since class 1 integrons are considered defective transposons, dissemination is achieved by their association with functional transposons, genomic islands and plasmids (39). In the context of carbapenems resistance, genes encoding for MBLs are usually found as gene cassettes in class 1 integrons (34, 36). This allows a rapid dissemination in the clinical setting due to the selective pressure by the use of antibiotics (24), which is aggravated due to this antibiotic represents the last therapeutic source to treat P. aeruginosa infections (11). While multiple studies correlate antibiotic resistance and the presence of integrons, genetic context surrounding class 1 integrons is often not investigated in P. aeruginosa, as remarked before (39).
Carbapenem resistance in PaeAG1 was demonstrated to be explained by activity of two MBLs (VIM-2 and IMP-18) (11), each gene harbored in two independent class 1 integrons (11, 12). With the aim of describing the genomic context associated with the two integrons, we characterized the architecture of genomic regions defined by GICVIM−2 and GICIMP−18.
For GICVIM−2 (composed of GI27 to G30), using Pae-PAO1 as reference, it was shown that insertion occurred in a specific region in the middle of PA2229 gene (Fig. 4), a hypothetical protein with unknown function. Evaluation of the sequence showed that GICVIM−2 is also present in Pae-RIVM-EMC2982 (100% coverage and 99.99% identity) and Pae-Carb0163 (100% coverage and 99.87% identity) at chromosomal level. However, a study including these strains showed that VIM-2-carrying integron and surrounding regions (~ 30 Kb, equivalent to GICVIM−2) were shared with a plasmid of ST-446 P. aeruginosa S04-90 with 99% identity. Based on identity, mobilization of the fragment between plasmids and chromosomes may have occurred recently (27).
In the same study, analysis of genome landscape showed that the regions (equivalent to GICVIM−2) corresponded to a DNA segment acting as a composite transposon, composed of four different transposons (Tn402, Tn21-like, a disrupted and another complete Tn4661). The class 1 integron carrying VIM-2 is contained in the Tn402 transposon (27, 61). Evolutionary details are completely explained in (27). GICVIM−2 carries the genes involved in its own transposition module (transposases such as TniB and TnpA) and mercury resistance module, as described in other similar transposons and insertion sequences (27, 36, 38, 39, 62). Presence of gene cassettes unrelated to the antibiotic resistance can be result of anthropogenic settings (38) and selection pressures in environments polluted with heavy metals and other substances such as mercury, arsenic and disinfectants (63). Thus, ability of strains adapted to toxic environments to thrive in hospital settings can be promoted by co-selection of antibiotic and metal resistance genes (16).
Regarding the VIM-2-carrying integron, this element is an In59-like integron. In59 was first reported two decades ago in France (21) and then worldwide (11, 25, 27, 61). The general architecture of VIM-2-carrying integron reveals that has classical genes of a class 1 integron, such as intI1 (integrase), attC, the gene cassette (in this case carrying VIM-2 and aacA29 genes), and sul1. Among all the 211 strains, VIM-2 was only present into PaeAG1 and the two closest genomes (all ST-111). Differences in aacA29 genes defined the aacA29e allele found in Pae-Carb0163 (27), in contrast to aacA29a-b in PaeAG1, all coding for aminoglycoside acetyltransferases. Since GICVIM−2 sequence and architecture is completely found in two VIM-2+/ST-111 strains, VIM-2-carrying integron (In59-like) can be considered old-acquaintance element in a well-known genomic context.
Additionally, genomic context defined by GICIMP−18 was also analyzed. Using Pae-PAO1 as reference, it is shown that GICIMP−18 insertion occurred in a specific point (prrH) between PA4704 and PA4705 (Fig. 5). This region contains three genes for regulatory small RNAs (prrF1, prrH and prrF2) are found, which are involved in iron homeostasis under iron-depleted conditions (64) or to avoid iron toxicity (65).
While complete GICIMP−18 (composed of GI48-GI49 genomic islands) was not found in none of other strains, GI48 section was found in Pae-97 strain (ST-234, with a class 1 integron), a close genome to ST-111 group (Figs. 1 and 3). Sequences comparison of GICIMP−18 and Pae-97 showed 77% coverage and 99.92% identity. Gene composition of GICIMP−18 includes endonucleases and recombinases module, the class1 integron, transposase TniB and hypothetical proteins.
In relation to the integron harbored in GICIMP−18, the IMP-18-carrying element is composed of the intI1, the gene cassette (carrying IMP-18, gcuD and OXA-2), aacA4 and sul1. In another strain, similar genes with another arrangement (orderly IMP-18, a disrupted aacA43, OXA-2 and gcuD) were reported for the first time in the In706 integron in 2012 (31). Pae-97 contains a class 1 integron, but with a different arrangement with IMP-1 allele (without OXA nor gcuD genes). Other studies found multiple strains carrying both IMP-18 and OXA-2 (without gcuD nor aacA4) in Mexican isolates as part of In169 (24) and In1215 (33) integrons, including some located in plasmids.
Since there is a lack of information about the genomic context of many IMP-carrying integrons (such as region GICIMP−18, unlike GICVIM−2), and the particular architecture of the class 1 integron in PaeAG1 with the gene cassette IMP-18/gcuD/OXA-2/aacA4, we consider that this IMP-18-carrying integron (registered as In1666) is a novel element that we report here for the first time. An in-depth further analysis to describe the possible origin of this and the other PaeAG1 genomic islands is required, including an exhaustive sequence mining to identify other mobile elements (such as transposons and insertion sequences) and particular evolutionary steps to reach the current architecture.
Jointly, identification of the landscape of the genomic context defined by GICVIM−2 and GICIMP−18, provides insights about the dissemination and evolution of mobile elements, in this particular case for integrons carrying MBLs. Since MBL-producing P. aeruginosa is able to produce epidemic outbreaks and responsible for the dissemination of carbapenemase resistance worldwide (23), it is worrisome that strains such as PaeAG1 are able to circulate among Costa Rican hospitals. This can be correlated with the high prevalence of carbapenem resistant strains in Costa Rica, many carrying VIM or IMP genes (11). Future works are necessary to trigger the surveillance system in order to evaluate if other circulating strains carry these two elements, to identify its possible dissemination and hence carry out an adequate infection control program in medical centers.