Genetic Diversity of Virulent Polymyxin-Resistant Klebsiella aerogenes Isolated from Intensive Care Units

This study evaluated the scope and genetic basis of polymyxin-resistant Klebsiella aerogenes in Brazil. Eight polymyxin-resistant and carbapenemase-producing K. aerogenes strains were isolated from patients admitted to the ICU of a tertiary hospital. Bacterial species were identified by automated systems and antimicrobial susceptibility profile was confirmed using broth microdilution. The strains displayed a multidrug resistant profile and were subjected to whole-genome sequencing. Bioinformatic analysis revealed a variety of antimicrobial resistance genes, including the blaKPC-2. No plasmid-mediated colistin resistance gene was identified. Nonetheless, nonsynonymous mutations in mgrB, pmrA, pmrB, and eptA were detected, justifying the colistin resistance phenotype. Virulence genes encoding yersiniabactin, colibactin, and aerobactin were also found, associated with ICEKp4 and ICEKp10, and might be related to the high mortality observed among the patients. In fact, this is the first time ICEKp is identified in K. aerogenes in Brazil. Phylogenetic analysis grouped the strains into two clonal groups, belonging to ST93 and ST16. In summary, the co-existence of antimicrobial resistance and virulence factors is deeply worrying, as it could lead to the emergence of untreatable invasive infections. All these factors reinforce the need for surveillance programs to monitor the evolution and dissemination of multidrug resistant and virulent strains among critically ill patients.


Introduction
Klebsiella aerogenes represents one of the bacterial species causing nosocomial infections; it also causes respiratory infections, urinary tract infections, endocarditis, skin or soft-tissue infections, abdominal infections, and osteomyelitis [1][2][3]. Moreover, it easily acquires numerous genetic mobile elements containing resistance and virulence genes that increase its pathogenicity and lethality. Therefore, this group of pathogens is receiving special attention in clinical practice [4,5].
Polymyxin-resistant Enterobacteriaceae is an urgent threat to human health because polymyxin is regarded as a drug of last resort in the therapeutic management of Gramnegative infections that otherwise have limited or no therapeutic options [6]. Resistance to colistin is typically caused by lipopolysaccharide (LPS) modification with 4-amino-4dexoxy-L-arabinose or phosphoethanolamine. These positively charged molecules reduce the overall negative charge of LPS, leading to a smaller electrostatic interaction with colistin and preventing cell lysis [7]. In addition to colistin resistance due to chromosomal mutations, plasmid-borne colistin resistance has been reported; all these contribute to decreased binding of colistin to LPS [8]. Furthermore, Klebsiella spp. may harbor several virulence factors associated with survival and pathogenesis, including a e integrative conjugative element (ICEs), which is a group of mobile genetic element transferable between bacterial species, associated with hypervirulence in K. pneumoniae strains (e.g., ICE Kp10 encoding the siderophore yersiniabactin and genotoxin colibactin) [9][10][11][12].
Effective treatment of infections caused by these pathogens as well as the implementation of adequate preventive measures to effectively contain the spread remain challenging [13]. Various studies have investigated the mechanisms associated with polymyxinresistance in Gram-negative infections. The polymyxin exposure may triggered the genetic events that lead to gene modifications in the polymyxin-resistant isolates [14][15][16][17][18][19][20][21][22]. However, studies on polymyxin-resistant Enterobacter spp. have been limited compared with those on other Enterobacteriaceae pathogens. In this study, we identified the genetic mechanisms associated with polymyxin resistance in K. aerogenes isolates from a tertiary hospital located in the Central-Western region of Brazil by using whole-genome sequencing.

General Patient Characteristics
Over the study period, a total of 340 cultures were performed and 124 cases of Enterobacteriaceae infection were identified. Twenty-five strains were isolated from patients presented to the adult intensive care unit (ICU). Of these, polymyxin-resistant and carbapenemaseproducing K. aerogenes strains were isolated from eight patients. The majority of patients were female (62.5%), with a median age of 45 years (range, 31-76 years). No significant differences (p > 0.05) in baseline demographics were observed among the patients. All patients had a diagnosis of infectious diseases at the sample collection time. Four of them had bloodstream infections, whereas the remaining patients had a diagnosis of pulmonary and urinary infections. All patients were exposed to antibiotics 30 days prior to the initial detection of the polymyxin-resistant strain (Appendix A Table A1). Outcome analysis revealed that five patients infected with polymyxin-resistant K. aerogenes died of sepsis, giving an infection-attributable high mortality rate of 62.5% (Appendix A Table A1).
However, we could not identify any other article that reported the mortality rate for patients with polymyxin-resistant K. aerogenes. Some studies have evaluated patients with carbapenem-resistant K. aerogenes infection and reported mortality rates of 20% [23], 37.5% [24], and 40.3% [25] for patients with severe sepsis. These results reinforced the criticality of infection control measures to prevent the spread of such bacteria in hospitals.
Regarding phylogenetic analysis and virulome, the core genome phylogeny and MLST analysis identified similar populations and were used to classify the strains into two strongly clonal groups. On the basis of SNPs, a maximum-likelihood phylogenetic tree was constructed, demonstrating that strains were closely related and partitioned into two clades. MLST analysis identified two different sequence types: ST93 (n = 5), belonging to clonal complex 3 (CC3), and ST16 (n = 3). ST93 has been previously described in clinical isolates of K. aerogenes in Brazil [26] and, at the moment, it represents one of the dominant global K. aerogenes clones associated with human infections [11]. On the other hand, to date, there is a single register of ST16 in a clinical isolate from China, from 2014, according to data available at the PubMLST database (https://pubmlst.org/bigsdb?db= pubmlst_kaerogenes_isolates&page=query, accessed on 18 May 2022). The high genetic similarity of strains suggested transfer events between patients. Therefore, measures must be implemented to avoid nosocomial transmission [27].
Virulome analysis revealed the presence of genes encoding yersiniabactin, colibactin, and aerobactin. Additionally, distinctive yersiniabactin (ybt) lineages (ybt 0, ybt 10, and ybt 17) were found associated with two different structural variants of ICEKp (ICEKp4 and ICEKp10) (Figure 1). ICEKp is the most common mobile genetic element associated with virulence in K. pneumoniae isolates, facilitating the spread of virulence genes within the population [9]. To the best of our knowledge, until now, there is only one report of ICEKp in K. aerogenes isolates from New York, USA, which was even identified as ICEKp10 [11]. In fact, ICEKp4 and ICEKp10 seems to be the most widely distributed variants [9]. As ICEKp elements harbor yersiniabactin and colibactin genes, which are considered key bacterial virulence factors, we could even hypothesize that a higher severity of infection with death outcome, as observed in some patients, could be related to the presence of these elements. and aerobactin. Additionally, distinctive yersiniabactin (ybt) lineages (ybt 0, ybt 10, and ybt 17) were found associated with two different structural variants of ICEKp (ICEKp4 and ICEKp10) ( Figure 1). ICEKp is the most common mobile genetic element associated with virulence in K. pneumoniae isolates, facilitating the spread of virulence genes within the population [9]. To the best of our knowledge, until now, there is only one report of ICEKp in K. aerogenes isolates from New York, USA, which was even identified as ICEKp10 [11]. In fact, ICEKp4 and ICEKp10 seems to be the most widely distributed variants [9]. As ICEKp elements harbor yersiniabactin and colibactin genes, which are considered key bacterial virulence factors, we could even hypothesize that a higher severity of infection with death outcome, as observed in some patients, could be related to the presence of these elements.

Antimicrobial Susceptibility Profile, Resistome and Plasmid Incompatibility Groups
Regarding the susceptibility profile (Appendix A Table A2), all strains exhibited resistance to polymyxin B (MIC50 ranging from 8 mg/L to 32 mg/L). Resistance gene profiles varied among strains, with some harboring several beta-lactam, aminoglycoside, tetracycline, quinolones, sulfonamides, trimethoprim, amphenicol, fosfomycin, macrolide and rifamycin resistance determinants. The blaKPC-2 and blaOXA−1 carbapenemase genes were identified in all isolates. In addition, the strains carried the blaCTX−M−15 (75%, n = 6), blaTEM-1B (75%, n = 6) and blaOXA−9 groups (12.5%, n = 1) extended-spectrum β-lactamase (ESBL) encoding genes. Although it was not possible to determine the location of these antimicrobial resistance genes, six different plasmid replicons were identified among these isolates ( Figure 1). This finding suggests the horizontal transmission of at least some of these genes, mainly those encoding resistance to beta-lactams, once they are frequently harbored by plasmids [28].
Plasmid-encoded colistin resistance genes (mcr-like) were not detected. Nonetheless, deleterious point mutations were found in mgrB, pmrA, pmrB, eptA, and arnT genes (Table  A2). All strains exhibited alterations in the mgrB. The alterations were located in two different amino acids positions (M1V and G37S). A second potential mechanism of mutational colistin resistance, due to a substitution at amino acid position (T296S) in eptA, was identified in all ST93 isolates (62.5%). A third potential mechanism of mutational resistance to colistin, due to an amino acid position substitution in pmrAB, was observed in the ST16 isolates (37.5%).

Antimicrobial Susceptibility Profile, Resistome and Plasmid Incompatibility Groups
Regarding the susceptibility profile (Appendix A Table A2), all strains exhibited resistance to polymyxin B (MIC 50 ranging from 8 mg/L to 32 mg/L). Resistance gene profiles varied among strains, with some harboring several beta-lactam, aminoglycoside, tetracycline, quinolones, sulfonamides, trimethoprim, amphenicol, fosfomycin, macrolide and rifamycin resistance determinants. The bla KPC-2 and bla OXA−1 carbapenemase genes were identified in all isolates. In addition, the strains carried the bla CTX−M−15 (75%, n = 6), bla TEM-1B (75%, n = 6) and bla OXA−9 groups (12.5%, n = 1) extended-spectrum β-lactamase (ESBL) encoding genes. Although it was not possible to determine the location of these antimicrobial resistance genes, six different plasmid replicons were identified among these isolates ( Figure 1). This finding suggests the horizontal transmission of at least some of these genes, mainly those encoding resistance to beta-lactams, once they are frequently harbored by plasmids [28].
Plasmid-encoded colistin resistance genes (mcr-like) were not detected. Nonetheless, deleterious point mutations were found in mgrB, pmrA, pmrB, eptA, and arnT genes (Table A2). All strains exhibited alterations in the mgrB. The alterations were located in two different amino acids positions (M1V and G37S). A second potential mechanism of mutational colistin resistance, due to a substitution at amino acid position (T296S) in eptA, was identified in all ST93 isolates (62.5%). A third potential mechanism of mutational resistance to colistin, due to an amino acid position substitution in pmrAB, was observed in the ST16 isolates (37.5%).
Studies have suggested that alterations in mgrB and pmrAB may be responsible for polymyxin resistance in Gram-negative pathogens [13,[29][30][31]. The inactivation of mgrB, which encodes a negative feedback regulator of the PhoQ-PhoP signaling system, was recently demonstrated to be a common mutational mechanism responsible for acquired polymyxin resistance among the clinical isolates of K. pneumoniae, Enterobacter spp., and E. coli [14,15,21,[32][33][34][35]. Mutations in mgrB may be the main determinant for colistin resistance in K. aerogenes [19]. Researchers from Croatia have reported that polymyxin resistance in K. aerogenes occurred due to mgrB present in a wild-type sequence. However, they did not detect the presence of mcr-1 or mcr-2 plasmid genes [36].
Polymyxin resistance is most commonly regulated by two-component systems, including PmrAB and PhoPQ [17]. Alterations in pmrAB or eptA may be a mechanism of colistin resistance, as described among E. coli strains [37]. A French study reported that a G53S substitution in PmrA resulted in a polymyxin resistance phenotype in a K. aerogenes strain [38]. Mutations in pmrB were associated with colistin resistance in A. baumannii [16]. To the best of our knowledge, it is the first time that the mutation of mgrB, responsible for colistin resistance in K. aerogenes, has been detected in Brazil.
Our findings demonstrated that the mechanisms of polymyxin resistance in K. aerogenes appear to be highly diverse. The emergence and spread of polymyxin-resistant strains have been reported, especially in the hospitals where bla KPC-2 is endemic, and the increased consumption of polymyxins has been proven to be a major risk factor for polymyxinresistance development [15,18,20]. Previous studies have reported that high levels of antimicrobials, including polymyxin, are frequently administered in Brazilian ICUs, mainly after bacterial isolates have become resistant to almost all other available antibiotics [39]. We hypothesized that the polymyxin exposure triggered the genetic event that led to gene modifications in the first isolate of each clone. However, because of the very recent release of K. aerogenes MLST profile, there is no sufficient information about the clonal characteristics and global distribution of this lineage. This study is limited by its small sample size; it only included the polymyxin-resistant K. aerogenes strains isolated during the study period. However, the emergence and clonal spread of mutational colistin resistance mediated by three distinct mechanisms over the course of three months is concerning, especially for patients admitted in ICUs.

Bacterial Strains
Polymyxin-resistant K. aerogenes recovered from patients hospitalized at a public tertiary care hospital, in Brazil, from August 2016 to October 2016 were included ( Figure 2). Samples were collected by hospital nurses as part of routine screening used in patient diagnosis. Patient characteristics along with clinical and demographic data were reviewed and entered into Research Electronic Data Capture (Redcap) (Vanderbilt University, Nashville, TN, USA). Data regarding the clinical outcome were reviewed. Death due to any cause or death attributable to infection was assessed. Septic shock was defined as sepsis associated with organ dysfunction, accompanied by persistent hypotension following volume replacement. Multidrug resistant were defined as resistance to one or more antimicrobials from three or more tested categories [41].

Whole-Genome Sequencing (WGS)
Genomic DNA was extracted from fresh cultures by using QIAamp ® DNA Mini Kit (Qiagen, Hilden, Germany). The concentration and purity of DNA were determined using ® Figure 2. Flowchart of the study design.

Bacterial Identification and Antimicrobial Susceptibility Testing
Identification and screening of antimicrobial-resistant bacterial species were performed using Phoenix ® Automated System (BD Diagnostic Systems, Sparks, MD, USA) according to the manufacturer's instructions. After isolation, the susceptibility profile was confirmed and minimal inhibitory concentrations (MICs) of antimicrobials were determined using broth microdilution, following the Clinical and Laboratory Standards Institute guidelines [40]. The following classes of antimicrobials were tested: Cephalosporins (cefotaxime, ceftazidime, ceftriaxone and cefepime), carbapenems (ertapenem, imipenem and meropenem), fluoroquinolones (ciprofloxacin and levofloxacin), aminoglycosides (amikacin and gentamicin), monobactams (aztreonam), and polymyxins (polymyxin B). Multidrug resistant were defined as resistance to one or more antimicrobials from three or more tested categories [41].

Whole-Genome Sequencing (WGS)
Genomic DNA was extracted from fresh cultures by using QIAamp ® DNA Mini Kit (Qiagen, Hilden, Germany). The concentration and purity of DNA were determined using a Qubit ® 2.0 fluorometer and the dsDNA BR Assay Kit (Life Technologies, Carlsbad, CA, USA). Sequencing libraries were prepared using the Nextera library kit (Illumina, San Diego, CA, USA). The prepared libraries were sequenced with 150 bp paired-end reads via IlluminaMiSeq Platform (Illumina, San Diego, CA, USA), as described in a previous study [42]. FastQC 0.11.2 was used to preprocess the reads [43]. Each read set was assembled using SPAdes 3.6.1 [44] with k-mer sizes of 21, 33, 55, 77, 99, and 127 and mismatch correction. These sequences were annotated using Prokka [45]. Species identification was performed using Kraken [46]. The K. aerogenes core genome was defined as the concatenation of coding sequences presenting one copy in all final assemblies [47]. The whole-genome sequences described in this paper have been deposited in ENA (European Nucleotide Archive) (Project: PRJEB25746; accession numbers in Appendix A Table A3).

Bioinformatics Analysis
Single nucleotide polymorphisms (SNPs) were identified through mapping of Illumina reads to a reference genome (Klebsiella aerogenes ATCC 13048). Maximum likelihood phylogenetic trees were constructed using RAxML 8.1.23 [48]. Analyses were performed with 100 bootstrap replicates per run, with a generalized time-reversible model and a gamma distribution to model site-specific rate variation (GTR+Γ substitution model; GTRGAMMA in RAxML); final visualization was performed using FigTree 1.3.1. We selected a single tree with the highest maximum likelihood as the best tree. For the larger tree containing global isolates, clades collapsed manually in R. SRST2 [49] were used to map known alleles and identify multilocus sequence typings (MLSTs) directly from reads according to the K. aerogenes MLST database [50].
Assembled genomes were submitted to ResFinder 4.1 (https://cge.cbs.dtu.dk/services/ ResFinder/, accessed on 6 February 2022) for prediction of acquired antimicrobial resistance genes. Chromosomal genes mgrB, phoP, phoQ, pmrA, pmrB, eptA, and arnT were manually screened for point mutations associated with colistin resistance by using blastn and blastx tools, and Klebsiella aerogenes ATCC 13048 was used as reference genome (GenBank accession number QVMZ00000000.1). PROVEAN v. 1.1.3 software was used to predict the functional effect of each found mutation, considering deleterious mutations as related to the resistant phenotype [51]. Plasmid replicon sequences were identified using ARIBA to screen reads for replicons in the PlasmidFinder database [52]. Virulence genes were identified by comparison of the assembled genome with genes from the Virulence Factor Database (http://www.mgc.ac.cn/VFs/main.htm, accessed on 6 February 2022) and ICEKp variants were determined by using blastx. A ≥98% threshold for sequence identity was used for resistance and virulence genes identification (Figure 1).