Mobile Carbapenemase Genes in Pseudomonas aeruginosa

Carbapenem-resistant Pseudomonas aeruginosa is one of the major concerns in clinical settings impelling a great challenge to antimicrobial therapy for patients with infections caused by the pathogen. While membrane permeability, together with derepression of the intrinsic beta-lactamase gene, is the global prevailing mechanism of carbapenem resistance in P. aeruginosa, the acquired genes for carbapenemases need special attention because horizontal gene transfer through mobile genetic elements, such as integrons, transposons, plasmids, and integrative and conjugative elements, could accelerate the dissemination of the carbapenem-resistant P. aeruginosa. This review aimed to illustrate epidemiologically the carbapenem resistance in P. aeruginosa, including the resistance rates worldwide and the carbapenemase-encoding genes along with the mobile genetic elements responsible for the horizontal dissemination of the drug resistance determinants. Moreover, the modular mobile elements including the carbapenemase-encoding gene, also known as the P. aeruginosa resistance islands, are scrutinized mostly for their structures.


INTRODUCTION
Pseudomonas aeruginosa is a non-fermentative and aerobic Gram-negative bacillus that is one of the leading causes of severe health care-associated infections targeting immunocompromised patients (Rice, 2008). The bacterial species is an opportunistic pathogen not only for humans but also for plants and animals. P. aeruginosa is metabolically versatile, and it has an enormous ability for adaptation to different conditions with genome plasticity (Shen et al., 2006). There are diverse opinions whether the pangenome of P. aeruginosa is still open or closed to acquire foreign genes (Klockgether et al., 2011;Mosquera-Rendon et al., 2016). The accessory genome is often composed of genes involved in virulence to human hosts and antimicrobial resistance, resulting in a high risk of mortality and a high rate of multidrug resistance (Moradali et al., 2017). P. aeruginosa has a median genome size value of 6.7 Mbp, a median number of 6,016 coding sequences, and 66.1% GC on average (NCBI, 2020).
Although beta-lactams are one of the most commonly used antimicrobial drug classes for P. aeruginosa infection, antipseudomonal beta-lactam drugs are limited because of the species' intrinsic resistance due to the interplay of chromosomal beta-lactamases (Livermore and Yang, 1987), a low outer membrane permeability (Angus et al., 1982), and the constitutive expression of efflux pump systems (Li et al., 1995). The beta-lactam regimens for P. aeruginosa infection include antipseudomonal penicillins in combination with a beta-lactamase inhibitor, i.e., piperacillin-tazobactam and ticarcillin-clavulanic acid; antipseudomonal cephalosporins alone or in combination with beta-lactamase inhibitors, i.e., ceftazidime (with avibactam), ceftolozane (with tazobactam), cefoperazone (with sulbactam), and cefepime (Slack, 1981); and carbapenems, i.e., imipenem and meropenem. Among those, carbapenems are the preferred choice against multidrug-resistant P. aeruginosa. In recent years, the rate of carbapenem resistance in P. aeruginosa has increased worldwide and has become of great concern since it significantly restricts the therapeutic options for patients (El Solh and Alhajhusain, 2009). Carbapenem resistance in P. aeruginosa is caused by chromosomal substitutions resulting in membrane permeability alterations through porin loss and efflux pump overexpression, together with intrinsic beta-lactamase derepression, and the acquisition of the genes for carbapenemases (Livermore, 1992;Masuda et al., 2000;Lister et al., 2009).
In this review, we summarized carbapenem resistance in P. aeruginosa. The first half of this review outlines the worldwide epidemiology and the second half discusses the mechanisms of carbapenem resistance and the mobile genetic elements responsible for the horizontal dissemination of resistance determinants. All of the information in the review was collected and analyzed from the National Database of Antibiotic Resistant Organisms 1 using the Reference Gene Catalog ver. 2020-09-22.2 2 for the carbapenemases and the Genome Database (last updated on 22 September 2020) 3 for the complete P. aeruginosa genomes.

CARBAPENEMS FOR ANTIPSEUDOMONAL TREATMENT AND RESISTANCE IN P. aeruginosa
Carbapenems for the Treatment of Patients With P. aeruginosa Infection Beta-lactams act by binding to and inactivating the penicillinbinding proteins (PBPs), which have an essential role for the completion of peptidoglycan biosynthesis through their dual activity as a transglycosylase and transpeptidase. Among the beta-lactams, carbapenems are the most effective against Gram-positive and Gram-negative bacteria, presenting a broad spectrum of antibacterial activity. Replacing the sulfur atom at the C-1 position of the penicillin backbone by a carbon atom (the red dot of the carbapenem backbone in the box in Figure 1) allows exceptional stability against most enzymes inactivating beta-lactams (Papp-Wallace et al., 2011).

Carbapenem Resistance in P. aeruginosa
The rate of carbapenem resistance in P. aeruginosa varies worldwide (Figure 3). According to the Antimicrobial Testing Leadership and Surveillance program by Pfizer in 2018 (last updated on September 14, 2020) (Pfizer, 2020), the rate of resistant clinical strains (-R) of P. aeruginosa by continent was the lowest in Oceania (imipenem-R in 7.1% and meropenem-R in 5.1% of 99 isolates from Australia), and the highest was in the Middle East (imipenem-R in 27.9% and meropenem-R in 19.5% of 226 isolates from four participating countries). In descending order, the median resistance rates were 30.7% in South America (the lowest imipenem-R and meropenem-R both in 12.5% of 24 isolates from the Dominican Republic and the highest imipenem-R in 49.3% and meropenem-R in 75.3% in 75 isolates from Chile, among nine participating countries), 28.0% in Europe (the lowest 0.0% of 19 isolates from Finland and the highest imipenem-R in 48.5% and meropenem-R in 44.8% of 194 isolates from Russia, among 24 participating countries), 24.4% in North America (imipenem-R in 21.4% and meropenem-R in 18.3% of 197 isolates from Canada and imipenem-R in 27.4% and meropenem-R in 15.5% of 588 isolates from the United States), 22.8% in Africa (the lowest imipenem-R in 13.2% and meropenem-R in 15.8% of 38 isolates from Nigeria and the highest imipenem-R in 21.4% and meropenem-R in 19.4% of 98 isolates from South Africa, among three participating countries), and 18.1% in Asia (the lowest imipenem-R and meropenem-R both in 8.0% of 75 isolates from Japan and the highest imipenem-R in 33.2% and meropenem-R in 25.1% of 386 isolates from China, among 10 participating countries).

Epidemic High-Risk P. aeruginosa Clones
The current P. aeruginosa high-risk clones, which meet both requirements of global dominance and association with the multidrug-resistant phenotype, include 10 P. aeruginosa lineages belonging to ST111, ST175, ST233, ST235, ST244, ST277, ST298 (CC445), ST308, ST357, and ST654 (Del Barrio-Tofino et al., 2020). The multidrug-resistant P. aeruginosa ST111, ST175, and ST235 have been identified to carry genomic islands (Roy Chowdhury et al., 2017). While all the 10 high-risk clones are relevant to MBL production, ST235 and ST111 are by far the most worrisome carbapenemase producers, associated not only with class B but also with class A and FIGURE 1 | Carbapenem drugs. The backbone of carbapenems is in a box. The C-1 position replaced from the sulfur atom in the penicillin backbone is indicated with a red dot. The important R1 and R3 positions are indicated with red letters. Ertapenem with a bulky R3 residue, which does not have enough affinity to be active against Pseudomonas aeruginosa, is presented with the other three carbapenems having antipseudomonal activity. D carbapenemases. The widespread P. aeruginosa ST235 clone is often associated with poor clinical outcomes due to its multidrug resistance and virulence factors, representatively the cytotoxin ExoU causing necrotic cell death (Sato et al., 2003;Roy Chowdhury et al., 2016;Yoon et al., 2019). The second dominant P. aeruginosa clone is ST111, which has been identified in all six continents except Oceania (Del Barrio-Tofino et al., 2020).

Chromosomal Mutation-Derived Carbapenem Resistance
Pseudomonas aeruginosa can acquire resistance to carbapenems by chromosomal mutations (Lister et al., 2009). Loss of the outer membrane protein OprD, which is a channel for imipenem penetration (Margaret et al., 1989), is associated with a reduced susceptibility to carbapenems, mostly imipenem (Farra et al., 2008). Early reports have underlined OprD deficiency as the predominant mechanism of carbapenem resistance in P. aeruginosa (Margaret et al., 1989;Kohler et al., 1999). The overexpression of efflux pump systems, such as MexAB-OprM, by mutation at the regulatory region contributes directly to the resistance to meropenem (Kohler et al., 1999;Masuda et al., 2000) and mutational derepression of the chromosomal cephalosporinase AmpC, especially the extendedspectrum cephalosporinases (Rodriguez-Martinez et al., 2009a), and plays a part in carbapenem resistance (Quale et al., 2006;Rodriguez-Martinez et al., 2009b). The combination of porin loss, efflux pump overexpression, and chromosomal cephalosporinase derepression is able to confer high-level resistance to carbapenems, and P. aeruginosa could have elevated imipenem and meropenem MICs up to 256 and 128 mg/L, respectively (Chalhoub et al., 2016).

Enzymatic Mechanisms of Carbapenem Resistance
Before 1990, the only known mechanism of carbapenem resistance was mutations occurring in the chromosome. Following the first identification of an MBL-producing P. aeruginosa clinical strain (Watanabe et al., 1991;Minami et al., 1996), a retrospective screening of P. aeruginosa identified the bla IMP−1 gene in 1992 in Japan (Senda et al., 1996a). Subsequent outbreaks due to the transferable drug resistance conferred by the gene were reported (Senda et al., 1996b). The bla VIM−1 gene encoding the Verona integron-encoded MBL (VIM) subtype 1 in P. aeruginosa clinical strain was identified in 1997 in Italy in a P. aeruginosa clinical isolate (Lauretti et al., 1999). And the carbapenem-resistant P. aeruginosa spread rapidly through the contribution of mobile genetic elements and high-risk clones. Thus far, class A, B, and D carbapenemases have been identified in P. aeruginosa, and the class B MBL enzyme is the most prevalent (Queenan and Bush, 2007).

Class A Beta-Lactamases
The class A beta-lactamases include serine at amino acid (aa) 70 at the active site and the general base Glu-166 is involved in the catalytic process, which makes a difference from the other serine beta-lactamases of classes C and D (Matagne et al., 1999). In P. aeruginosa, the Klebsiella pneumoniae carbapenemase (KPC) and the Guiana extended-spectrum beta-lactamase (GES) belonging to the class A beta-lactamases with carbapenemase activity have been identified. The class A carbapenemases actively hydrolyze carbapenems and are partially inhibited by clavulanic acid.
Klebsiella pneumoniae carbapenemase was first discovered in a K. pneumoniae clinical isolate from North Carolina, United States, in 1996, presenting a specific pattern of resistance to penicillins, extended-spectrum cephalosporins, and aztreonam (Yigit et al., 2001). The first KPC-producing P. aeruginosa isolate was identified in Colombia in Villegas et al. (2007), and FIGURE 2 | A stereoview of PBP3 of Pseudomonas aeruginosa complexed with meropenem (PDB ID, 3PBR) and the interaction of the meropenem in the ligand pocket of PBP3 (Han et al., 2010). The structure of PBP3 is colored by secondary structure, and the meropenem is in a ball-and-stick presentation. The molecular surface of PBP3 in the binding pocket is presented with the interacting amino acid residue complex with meropenem in a ball-and-stick presentation. subsequent reports of the pathogen followed all over the world, including America (Akpaka et al., 2009;Poirel et al., 2010;Robledo et al., 2011;Jacome et al., 2012;Ramirez et al., 2013;Kazmierczak et al., 2016a;Walkty et al., 2019), Asia (Ge et al., 2011;Paul et al., 2015;Falahat et al., 2016;Hagemann et al., 2018), and Europe (Figure 4).

Class B Beta-Lactamases
The class B beta-lactamases are also known as "metallo-" betalactamases because they need divalent cations, usually Zn 2+ ions, as a metal cofactor to hydrolyze beta-lactams. Although class B beta-lactamases are subclassified as B1, B2, and B3 based on structural and functional points (Frere et al., 2005), we will discuss subclass B1, the only dominant subclass in P. aeruginosa. The B1 subclass contains the largest number of clinically relevant acquired MBLs, not only in P. aeruginosa but also in Enterobacterales and other Gram-negative nonfermenters. MBLs bind two Zn 2+ atoms for optimal hydrolysis. Zn 2+ ion ligands bind at 3H (His-His-His) and DCH (Asp-Cys-His) sites, and the binding of di-Zn 2+ plays a critical role in hydrolyzing beta-lactam substrates (Frere et al., 2005;Moran-Barrio et al., 2016). Consequently, the carbapenemase activity of MBLs is diminished in the presence of a chelator of Zn 2+ and other divalent cations, i.e., ethylenediaminetetraacetic acid (EDTA). The substrate profile of the MBLs includes penicillins, cephalosporins, and carbapenems, but excludes monobactams. The acquired MBL genes, located mostly within a class 1 integron as gene cassettes, have been found in various bacterial species, including P. aeruginosa.
Metallo-beta-lactamase are the most prevalent type of carbapenemases produced by P. aeruginosa clinical isolates. VIMs are the most disseminated, followed by imipenemases (IMPs). New Delhi MBLs (NDMs) have also been identified.  São Paulo MBL (SPM), and Seoul imipenemase (SIM) was also reported.

Class D Beta-Lactamases
Class D beta-lactamases belong to the superfamily of serine betalactamases with a unique carboxylated Lys-73 responsible for the beta-lactam hydrolysis activity (Golemi et al., 2001). The carbapenem-hydrolyzing class D beta-lactamases (CHDLs) were first described in A. baumannii and published by Paton et al. (1993). CHDLs are serine beta-lactamases with a relatively weak activity against carbapenems and are poorly inhibited by EDTA or clavulanic acid. Among a total of 12 groups of CHDLs, three groups-OXA-40-like, OXA-48-like, and OXA-198-like-have been identified in P. aeruginosa.

MOBILE GENETIC ELEMENTS ASSOCIATED WITH CARBAPENEMASE-ENCODING GENES
The acquired genes encoding carbapenemases are associated with a plethora of mobile genetic elements, such as plasmids, gene cassettes of integrons, transposons, and genomic islands (Kung et al., 2010). Mobile genetic elements have the ability to move from genome to genome by transformation, conjugation, and transduction, presenting intracellular and intercellular mobility (Roberts et al., 2008).

Pseudomonas aeruginosa Plasmids Carrying the Carbapenemase-Encoding Genes
In general, the plasmids carrying the carbapenemase-encoding genes in P. aeruginosa belong to the distinct incompatibility groups from those in Enterobacterales. Among the 13 known incompatibility types of IncP, IncP-2-type plasmids are classic types frequently identified in P. aeruginosa (Korfhagen et al., 1976). Among the 207 complete genomes of P. aeruginosa of the Genome Database (NCBI, 2020), five genomes include a plasmid carrying one or two carbapenemase-encoding genes; two of the five plasmids are of the IncP-2 incompatibility type, while the other three are untypable: two bla KPC−2 genes are harbored by untypable plasmids, which are almost the same (identical nucleotide sequences, except for a 1-bp gap difference between the 57,053-and 57,052-bp plasmids), one bla VIM−1 is harbored by an untypable plasmid, one IncP-2 plasmid harbors bla IMP−45 , and one IncP-2 plasmid harbors both the bla VIM−1 and the bla IMP−45 genes.
Among the plasmids in P. aeruginosa having an incomplete genome, the 31,529-and 38,939-bp IncP-6 plasmids carrying the bla KPC−2 gene have been identified in Colombia and China, respectively (Naas et al., 2013;Dai et al., 2016), and the 7,995-bp IncU plasmid including the bla KPC−2 gene has been identified in Colombia (Naas et al., 2013). An untypable 3,652bp plasmid harboring the bla KPC−2 gene was identified in Brazil (Galetti et al., 2016).
Recently, a Pseudomonas plasmid lineage carrying the MBL genes has been reported (Di Pilato et al., 2019). A retrospective analysis revealed that the plasmid lineage has been identified since the 1990s, mostly in Europe. While the plasmid does not belong to a recognized plasmid type, the type 4 secretion system components classified the plasmids as MOBF11 or MPFT plasmid families. The MBL genes in the plasmid were identified as gene cassettes of the class 1 integron In70 (Di Pilato et al., 2019).

Carbapenemase-Encoding Gene-Associated Transposable Units
The carbapenemase-encoding genes are frequently included in transposable elements, which are often associated with the FIGURE 6 | Transposable units identified in Pseudomonas aeruginosa carrying the carbapenemase-encoding genes. The red arrow indicates the genes for antimicrobial resistance, and those with yellow letters inside indicate the genes for carbapenemase. Yellow arrows depict insertion sequences, and those in orange indicate transposase/resolvase. insertion sequences with a common region (ISCRs) being responsible for the rapid transmission of bacterial multidrug resistance (Figure 6; Toleman et al., 2006). Typically, the ISCR element lacks flanking inverted repeats (IRs) and the integration does not produce direct repeat (DR) sequences (Diaz-Aroca et al., 1987). Rolling circle has been suggested for the transmission mechanism of ISCRs (Figure 7), and the transposition method allows a polarized transfer of the ISCR elements to mobilize adjacent DNA sequences in varied sizes (del Pilar Garcillan- Barcia et al., 2001).
The bla CAM and bla FIM genes were identified in transposable elements associated with ISCR14 and ISCR19, respectively, located downstream from the gene. The bla SPM gene was identified in P. aeruginosa accompanied with the chaperone groEL gene and flanked by a pair of ISCR4 elements. The bla AIM gene was identified in a transposable element associated with the upstream ISCR15. The bla HMB−1 gene was in a transposable unit, namely, Tn6345 (Supplementary Table 1; Pfennigwerth et al., 2017), flanked by IS1595 upstream and ISCR27-like downstream (Toleman et al., 2006). For the bla NDM gene, the composite transposon Tn125 flanked by a pair of ISAba125 copies at both ends has been identified in A. baumannii (Bontron et al., 2016). The bla NDM−1 geneassociated transposable element is composed of a truncated Tn125. ISCR is located upstream from the ISAba1, and downstream of the bla NDM−1 gene, the ble gene for bleomycin resistance and the trpF gene for phosphoribosylanthranilate isomerase are followed (Sole et al., 2011).
The Tn4401 carrying the bla KPC gene has been found in diverse bacterial hosts, including P. aeruginosa (Cuzon et al., 2011; Figure 6). Tn4401 is a 10-kb transposon composed of genes encoding a transposase and a resolvase, and the bla KPC gene together with two insertion sequences (ISs), ISKpn6 and ISKpn7 (Figure 6), and transposition of Tn4401 occurs through the mechanism of copy-and-paste replicative transposition (Figure 7; Grindley, 1983). Tn4401 includes nine isoforms from a to I differing in the sequences upstream of the bla KPC gene (Naas et al., 2012;Bryant et al., 2013;Cheruvanky et al., 2017;Araujo et al., 2018;Schweizer et al., 2019). While both Tn4401a and Tn4401b are prevalent in Enterobacterales Yoon et al., 2018), Tn4401b is the only isotype identified in P. aeruginosa. Seven Tn4401 elements were found through the restricted BLAST against the species P. aeruginosa, and all were Tn4401b: two Tn4401b copies in a chromosome (GenBank accession CP029605) and each Tn4401b copy in five plasmids (GenBank accessions MN082782.1, CP027168.1, CP029092.1, KC609323.1, and EU176013.1).
FIGURE 7 | Mechanisms of replicative transposition of the transposons. (A) Insertion sequence with a common region (ISCR)-mediated rolling circle replicative transposition is involved in rolling circle replication. ISCR elements lack terminal inverted repeats, and a single copy of the element is able to transpose adjacent DNA sequences (Toleman et al., 2006). (B) A Tn3-mediated copy-and-paste replicative transposition requires both a transposase and a resolvase. The transposon is replicated, joining the donor and the recipient in a cointegrate, which is resolved to give the donor and the recipient of the transposon (Grindley, 1983).

Class 1 Integrons Carrying Carbapenemase-Encoding Gene Cassettes
Integrons are assembly platforms comprising an intI gene for a site-specific tyrosine recombinase, an attI for a primary recombination site, the promoter Pc for transcription, and an assemblage of passenger genes composing a gene cassette array (Figure 8; Collis et al., 1993). The IntI integrase recognizes the attC site of the gene cassette and the promoterless gene cassette is inserted as a linear form in the integron (Figure 8; Collis and Hall, 1992). In addition to the catalysis of attC × attC, the integrase catalyzes attI × attC, leading to the gene cassette integration into the attI site. A successive integration of the gene cassettes occurs downstream of the resident Pc promoter (Hall and Collis, 1995). The expression of the gene cassette is dependent on the sole promoter and the level of gene expression depends on the distance from the sole promoter Pc (Coyne et al., 2010). By nature, the first few cassettes are expressed and the rest of the array exists as a reservoir of standing genetic variation (Cambray et al., 2010). Rearrangement of the order of gene cassettes affects the resistance phenotype of the bacterial host (Hall and Collis, 1995;Rowe-Magnus et al., 2002).

Carbapenemase-Encoding Gene Cassettes
The class A carbapenemase-encoding gene bla GES , the MBL genes bla VIM , bla IMP , and bla GIM−1 , and the class D gene bla OXA−198 were identified as a gene cassette composing a class 1 integron. In the integron database INTEGRALL ( 4 last updated on 10 December 2020) (Moura et al., 2009), a total of 812 class 1 integrons of 282 different gene cassette arrays were identified for the organism P. aeruginosa. Among them, 191 class 1 integrons of 148 different arrays carried one or two carbapenemase-encoding gene cassettes (Supplementary Table 2). The most prevalent bla VIM genes were identified as gene cassettes in 98 arrays of the 130 class 1 integrons. Among them, 13 carried only the bla VIM The open box indicates the attI locus. The integron is organized as a functional platform including the intI1 gene and the attI locus and a cassette array assembled through the acquisition of gene cassettes structured with an open reading frame and an attC locus. Expression of the gene cassettes is dependent on the common promoter, Pc, and the level of expression depends on the distance from the Pc. The IntI1 integrase binds to the attC locus of the excised gene cassette to help circularize the cassette (Collis et al., 1993;Hall and Collis, 1995). cassette; the others harbored additional gene cassettes encoding aminoglycoside-modifying enzymes, mostly the aacA genes and less frequently the aadA or aadB genes. All but five carried the bla VIM gene cassette in the first position of the array (62.3%) or in the second position of the array (33.1%).
The second most dominant bla IMP cassettes were identified in 50 class 1 integrons of 40 different arrays, and 14 of the integrons harbored a single gene cassette. All but two harbored the bla IMP gene cassette in the first of the array (70.0%) or in the second of the array (22.9%). Less frequently, the bla GIM−1 gene cassette was identified in six class 1 integrons of five different arrays all in the first of the array, and three bla GES−5 cassette-associated and one of each of the bla GES−8 , bla GES−9 , and bla GES−15 cassetteassociated class 1 integrons were identified.

Mobilized Class 1 Integrons
Basically, class 1 integrons are immobile per se, and a set of functional transposition modules is needed for transposition (Martinez et al., 2012). The Tn402 family transposon Tn5090 is a good example of a functional transposon giving mobility to a class 1 integron (Figure 9; Radstrom et al., 1994). The Tn402 family transposon is a Mu-related transposon that has been identified in the broad-host IncP plasmid R751 conferring trimethoprim resistance (Jobanputra and Datta, 1974;Shapiro and Sporn, 1977). The transposon family has two modules: a transposition module composed of the transposase TniA, the ATP-binding protein TniB, the transposition auxiliary protein TniQ-resolution site res, the serine resolvase TniC, and a class 1 integron carrying antimicrobial resistance gene cassettes. Transposition of the Tn402 family transposon involves a TniABQ-dependent cointegrate formation and a site-specific serine resolvase-dependent resolution (Radstrom et al., 1994). The transposon has 25-bp inverted repeats, and the integration generates 5-bp DRs. The transposon targets the res site of Tn21 subfamily transposons as well as resolution sites found on plasmids (Minakhina et al., 1999). The Tn21-like Tn1403, Tn6060, Tn6162, and Tn6249, which nest inside the Tn402-like, carry a class 1 integron possessing the bla VIM−1 gene cassette (Di Pilato et al., 2015).
In the transposon repository database (Tansirichaiya et al., 2019), a total of 20 transposons carried the gene encoding carbapenemases (Supplementary Table 1), and of those, 19 transposons are associated with class 1 integrons harboring the bla VIM (n = 9), bla IMP (n = 6), bla GES (n = 2), bla SIM (n = 1), and bla DIM (n = 1) gene cassettes. The 207 complete P. aeruginosa genomes from the Genome Database included a total of 29 chromosomal class 1 integrons with a carbapenemase-encoding gene cassette. Among those, 20 integrons included a part or the entire set of the TniABQR transposition module of the Tn402-like.

Resistance Islands Harboring the Carbapenemase-Encoding Genes
Genomic islands carrying many foreign genes are useful for adaptation by providing multiple fitness-associated elements to P. aeruginosa in a single event of horizontal gene transfer (Kung et al., 2010). Genomic islands are often free to move in and out of the chromosome, and pKLC102, a 100-kb plasmid identified in P. aeruginosa C strain, is an example of this (Klockgether et al., 2004). The plasmid pKLC102 was found simultaneously in the Arrows indicate open reading frames with filled colors that differ by function: yellow, transposition; orange, class 1 integrase; red, antimicrobial resistance; green, heavy metal resistance (Radstrom et al., 1994;Minakhina et al., 1999;Shi et al., 2018).
FIGURE 10 | Schematic presentation of the structure of genomic islands belonging to the three groups. The schematic structure is presented for the three groups of genomic islands: (A) Group 1 resistance islands equipping the tyrosine-based integrase gene at the 5' terminal and the conjugative transfer machinery gene cluster at the 3' terminus, (B) Group 2 resistance islands carrying a transposition module of a whole or a partial TniABQR component, (C) and the others. The genes are indicated with open arrows and colored based on function: genes for integration and recombination of the genomic island are in yellow, genes for antimicrobial resistance (AMR) are in red, genes for heavy metal resistance (HMR) are in green, the intI1 gene is in orange, conjugative transfer module mostly composed of the type 4 pili assembly genes is in black, and the core gene is in gray. The open box indicates inverted repeats (IRs) left and right, named the attL and attR loci for integration conjugative elements. chromosome as a genomic island being integrated into the 3 end of the tRNA Lys gene in favor of an att site.
Genomic islands are typically inserted at the 3 end of a transfer RNA (tRNA) gene; however, they are also targeted elsewhere in the chromosome (Klockgether et al., 2011). Genomic islands are easily differentiated from the core genome by their atypical G+C contents, differing from the typical P. aeruginosa G+C content of 65-67% and atypical oligonucleotide usage. Genomic islands typically harbor the genes for factors involved in mobility, such as integrases, transposases, ISs, and other components responsible for biological processes. Genomic islands are categorized accordingly to their main characteristics determined by the gene content, such as pathogenicity, symbiosis, metabolic, fitness, or drug resistance. Thus far, 25 genomic islands have been identified: two P. aeruginosa pathogenic islands, PAPI-1 and PAPI-2; 17 P. aeruginosa genomic islands, PAGI-1 to PAGI-17; five Liverpool epidemic strain genomic islands, LESGI-1 to LESGI-5; and a plasmid-origin genomic island, pKLC102. As an effort to group the genomic islands, Kung et al. (2010) suggested two families of genomic islands by conserved function and synteny of the backbone genes, and Klockgether et al. (2011) proposed that any known genomic islands originated from an ancestry based on conserved orthologs. However, no scheme is publicly endorsed.
In addition, the PA143/97 genomic island, which was manufactured to include the Tn6249 carrying two class 1 integrons harboring the bla VIM−2 gene cassette, was reported (Martinez et al., 2012;Di Pilato et al., 2015).
Pseudomonas aeruginosa is able to obtain multidrug resistance at once through resistance islands. Since this review focuses on carbapenem resistance in P. aeruginosa, the exploration is restricted to genomic islands harboring carbapenemaseencoding genes.
Among the 207 complete genomes of P. aeruginosa extracted from the Genome Database, a total of 38 chromosomes harbor resistance islands carrying the carbapenemase-encoding gene, and eight of those carry two resistance islands. In total, 45 resistance islands, sized between 8,858 and 117,103 bp, were analyzed. The G+C contents varied from 55.5 to 65.5%, which were lower than those of the chromosomes, which were from 65 to 67%, and the integration sites varied.
The 45 resistance islands carrying the carbapenemaseencoding gene were classified into three groups by the transposition module ( Table 2). Group 1 resistance islands furnish the tyrosine-based integrase gene at the 5' terminal and the conjugative transfer machinery gene cluster at the 3 terminus (Figure 10). Group 1 can be further grouped into two subgroups: group 1a, with a core structure of ICE Tn4371 mostly targeting OprD of OccD4/OpdT tyrosine or adenylate cyclase ExoY, and group 1b, which resembles the known multidrugresistant genomic islands PAGI-13, PAGI-15, and PAGI-16 targeting tRNA Gly . Resistance islands belonging to group 1a had G+C contents between 64.5 and 65.1%, and the size was 44-74 kbp. The bla SPM−1 and bla NDM−1 genes carried by the group 1a resistance islands were all associated with IS91-like ISCRs composing unit transposons, and neither the other antimicrobial resistance determinants nor the heavy metal resistance-associated genes were identified in the resistance island. Meanwhile, the G+C contents of group 1b ranged between 60.5 and 64.2%, and the size ranged from 37 to 117 kbp. The upper cluster of group 1b, including four genomic islands ( Table 2), targeted tRNA Gly at the locus PA0714 of the genome of P. aeruginosa PAO1 (NCBI RefSeq, NC_002516.2). Characteristically, the genomic islands belonging to the cluster always possess the mercury resistance gene cluster, which is likely derived from Tn501,  a The molecular phylogenetic trees were constructed using the aligned amino acid sequences of the integration module, either the tyrosine-type integrase/recombinase or the concatenated TniABQR for transposition, using the maximum likelihood method implemented in the PhyML program (v. 3.0 aLRT) with the WAG matrix and a gamma correction for variable evolutionary rates (Guindon and Gascuel, 2003). A total of 100 bootstrap experiments are carried out; the red dots at the node indicate perfect robustness. b If the chromosome includes two genomic islands, each case is indicated with numbers 1/2 and 2/2 in brackets following the GenBank accession number. c Integration sites are indicated using the locus tag of the genome of P. aeruginosa PAO1 (NCBI RefSeq, NC_002516.2). d In0 is indicated with "0." e Antimicrobial resistance genes located in a class 1 integron and transposons are indicated, and the genes encoding carbapenemase are in boldface. and the carbapenemase-encoding genes are carried by class 1 integrons as a gene cassette. The eight genomic islands belonging to the next cluster of subgroup 1b targeted tRNA Gly either at the locus PA2583 or PA2817 of the PAO1 genome. A quarter of the genomic islands possess a class 1 integron, and the acquisition of the carbapenemase-encoding gene is mostly due to ISCR elements. Pieces of the heavy metal gene cluster have been observed. The group 2 resistance islands contain a transposition module composed of a whole or a partial TniABQR component. The resistance islands had obviously lower G+C contents, between 55.5 and 62.0%, and the sizes were diverse, from 8.8 to 105 kbp. The composition of the resistance islands in group 2 corresponds to that of the Proteus genomic island PGI-1 (Mac Aogain et al., 2016). The resistance islands are constructed through the accumulated assemblage on a Tn402 backbone. Such an assemblage of transposons is commonly observed in the AbaRtype resistance islands in A. baumannii, which evolved from the Tn6019 backbone (Krizova et al., 2011).
In the case of the remaining two resistance islands, categorization is unavailable since only two cases of possible clonal relations are available from the GenBank database. Neither a specific tyrosine-based recombinase nor a transposition component has been identified in the resistance islands; however, they are flanked by 20-bp inverted repeats.

CONCLUSION
Carbapenems represent a valuable therapeutic option for patients with infections caused by multidrug-resistant P. aeruginosa. It is ironic that carbapenem resistance, especially that conferred by carbapenemase production, is closely related to multidrug resistance, highlighting the role of modular mobile units carrying multiple antimicrobial resistance determinants. The molecular epidemiology of antimicrobial resistance has been studied by traditional methodologies based on PCR and Sanger sequencing to identify the resistance genes and to distinguish the fundamental mobile genetic elements carrying the gene, i.e., gene cassettes of integrons, transposons, and plasmids. The present era of massive next-generation sequencing, mostly the long-read sequencing, allows resolving a wide range of complex genome regions, such as modular mobile units associated with genes for antimicrobial resistance, also known as resistance islands. Such an extensive analysis has been carried out for limited numbers of the P. aeruginosa genome mostly for the genomic islands unrelated to antimicrobial resistance determinants. Among the plenty genomic islands in the genome of P. aeruginosa, resistance islands have a meaning beyond genome plasticity. Such a modular mobile unit harboring antimicrobial resistance determinants is able to disseminate by itself and capture an alien gene for resistance, which means the resistance islands and high-risk clones are the A to Z of acquisition of multidrug resistance. Our trial to classify P. aeruginosa resistance islands needs improvement with more cases for resistance islands.
The global spread of the carbapenem-resistant P. aeruginosa is one of the major global public health challenges, and the epidemiological scenario is often associated with the circulation of carbapenemase-encoding genes linked with (i) the endemic carbapenemase gene and (ii) the carbapenem usage in clinical settings. The KPC-producing P. aeruginosa in KPC-endemic United States, the SPM-producing P. aeruginosa in SPMendemic Brazil, and the NDM-producing P. aeruginosa in NDMendemic India exemplify well the first linkage. The second linkage is illustrated through the dominance of IMP-6-producing P. aeruginosa in South Korea, in which meropenem usage is approximately twice more in clinical settings than that of imipenem. Since IMP-6 has greater hydrolyzing activity to meropenem than to imipenem compared to the other subtypes of IMP enzymes, producing the IMP-6 subtype is favorable to the bacterial host. It emphasizes that the carbapenemaseproducing organisms should be controlled regardless of the bacterial host, and control includes both surveillance study and antimicrobial stewardship. Needless to say, more attention is needed to be paid to the emergence and spread of the highrisk P. aeruginosa clones, together with their enzymatic/nonenzymatic carbapenem resistance. Though continuing efforts are being made to develop the beta-lactamase inhibitors in order to preserve the efficacy of beta-lactam drugs including carbapenems, it is fruitful just for the serine beta-lactamases. Development of the inhibitors for MBLs, which are frequently produced by P. aeruginosa, is eager to be accelerated.
Despite the efforts to control the spread of carbapenemresistant P. aeruginosa, a conclusive solution to the issue is still far from being accomplished. Surveillance study for the drugresistant pathogen is essential and global collaboration using harmonized methods is important for a practical comparison of the outputs. In addition, to fight against the drug-resistant pathogen, we need to understand how the pathogens acquire resistance determinants. Taking the advantages of up-to-date techniques, assessment of the bacterial genome should be carried out, not only for the mobile genetic elements carrying a carbapenemase-encoding gene but also for the genomic islands. Furthermore, through the technique, the mobile genetic elements should be investigated extensively, and it would allow a comprehensive grasp of the dissemination of drug resistance.

AUTHOR CONTRIBUTIONS
SJ conceived and supervised the study, performed the data evaluation and confirmation, and finalized and edited the manuscript. E-JY carried out the analysis and data validation, and drafted the manuscript. Both authors contributed to the article and approved the submitted version.

FUNDING
This work was supported by a fund (NRF-2018R1C1B6002674) from the National Research Foundation of Korea. The funders had no role in the design of the study, in the collection, analysis, or interpretation of data, in the writing of the manuscript, or in the decision to publish to the results.