Comparative genome analysis of colistin-resistant OXA-48-producing Klebsiellapneumoniae clinical strains isolated from two Iranian hospitals

Background Carbapenemase-producing Klebsiella pneumoniae (CP-KP) is becoming extensively disseminated in Iranian medical centers. Colistin is among the few agents that retains its activity against CP-KP. However, the administration of colistin for treatment of carbapenem-resistant infections has increased resistance against this antibiotic. Therefore, the identification of genetic background of co-carbapenem, colistin-resistance K. pneumoniae (Co-CCRKp) is urgent for implementation of serious infection control strategies. Methods Fourteen Co-CCRKp strains obtained from routine microbiological examinations were subjected to molecular analysis of antimicrobial resistance (AMR) using whole genome sequencing (WGS). Results Nine of 14 K. pneumoniae strains belonged to sequence type (ST)-11 and 50% of the isolates had K-locus type 15. All strains carried blaOXA-48 except for P26. blaNDM-1 was detected in only two plasmids associated with P6 and P26 strains belonging to incompatibility (Inc) groups; IncFIB, IncHI1B and IncFII. No blaKPC, blaVIM and blaIMP were identified. Multi-drug resistant (MDR) conjugative plasmids were identified in strains P6, P31, P35, P38 and P40. MICcolistin of K. pneumoniae strains ranged from 4 to 32 µg/ml. Modification of PmrA, PmrB, PhoQ, RamA and CrrB regulators as well as MgrB was identified as the mechanism of colistin resistance in our isolates. Single amino acid polymorphysims (SAPs) in PhoQ (D150G) and PmrB (R256G) were identified in all strains except for P35 and P38. CrrB was absent in P37 and modified in P7 (A200E). Insertion of ISKpn72 (P32), establishment of stop codon (Q30*) (P35 and P38), nucleotides deletion (P37), and amino acid substitution at position 28 were identified in MgrB (P33 and P42). None of the isolates were positive for plasmid-mediated colistin resistance (mcr) genes. P35 and P38 strains carried iutA, iucD, iucC, iucB and iucA genes and are considered as MDR-hypervirulent strains. P6, P7 and P43 had ICEKp4 variant and ICEKp3 was identified in 78% of the strains with specific carriage in ST11. Conclusion In our study, different genetic modifications in chromosomal coding regions of some regulator genes resulted in phenotypic resistance to colistin. However, the extra-chromosomal colistin resistance through mcr genes was not detected. Continuous genomic investigations need to be conducted to accurately depict the status of colistin resistance in clinical settings. Supplementary Information The online version contains supplementary material available at 10.1186/s12941-021-00479-y.


Introduction
Carbapenemase-producing Klebsiella pneumoniae (CP-KP) has been established as a major cause of healthcare-associated infection in many geographic areas, with high morbidity and mortality [1]. Infection with CP-KP is a serious clinical problem because it is difficult to treat using conventional antibiotics. Resistance to carbapenems can be mediated by various mechanisms including the production of carbapenemases, including K. pneumoniae carbapenemase (KPC), New Delhi metallo-β-lactamase (NDM), and oxacillinase-48 (OXA-48), production of extended spectrum β-lactamases (ESBL) plus porins and hyperproduction of Ambler class C (AmpC) β-lactamase [2]. Among major carbapenemases, OXA-48 carbapenemase is currently the one that is spreading the most rapidly in the Middle East and other countries worldwide [3][4][5]. Polymyxins especially colistin are among the few agents that retain its activity against CP-KP, and they are considered as the key components against severe infections caused by these superbugs. Increasing administration of colistin for treatment of CP-KP infections has contributed to the emergence of acquired resistance against this antibiotic [6]. Resistance to colistin is mostly associated with LPS modification (result from mutations in pmrA/pmrB, phoP/phoQ, mgrB, crrB, and ramA genes) as well as overproduction of efflux-pumps (mediated by kpnE/kpnF and mutations of acrB) on the chromosomal level. Additionally, the production of phosphor-ethanolamine transferase encoded by plasmid-mediated colistin resistance (mcr) genes results in transferable colistin-resistance [7]. In Iran, where the carbapenemase is becoming extensively disseminated among clinical Gram-negative isolates, increasing rate of colistin-resistant K. pneumoniae is a critical matter. Therefore, the identification of genetic background of co-carbapenem, colistin-resistance K. pneumoniae (Co-CCRKp) is urgent for implementation of serious infection control strategies.
Genomic studies based on whole genome sequencing (WGS) accurately identify multi-drug resistant (MDR) and hypervirulent clones in outbreaks and provide data about diversity and antimicrobial resistance (AMR) reservoir of K. pneumoniae [8]. Furthermore, such studies are capable to determine the circulation of clinically important sequence types (ST) and spreading of major AMR genes across various clonal lineages and among hospitalized patients and community carriers [9]. Therefore, the global and regional awareness of antibiotic resistance genetic determinants is critical to combat the spread of highly resistant K. pneumoniae and decrease the number of victims.
In this study, we conducted a comparative genomic analysis of 14 colistin-resistant OXA-48-producing K. pneumoniae isolated from different wards of two Iranian hospitals with focus on their AMR genetic reservoir using WGS.

Bacterial isolates
In the period between January 2014 to March 2016, one hundred and thirty-eight carbapenem-resistance K. pneumoniae isolates were obtained from routine microbiological examinations on clinical samples (e.g., urine and bronchial aspirate) from two medical centers in two provinces of Iran (hospital A, a 496-bed university hospital, located in Tehran and hospital B, a 800-bed university hospital, located in Isfahan.). Conventional biochemical examinations were used for identification of the isolates. Genus and specie of isolates were confirmed by PCR-sequencing of 16s rRNA [10].

Antimicrobial susceptibility testing
Antimicrobial susceptibility was determined using Kirby-Bauer disk diffusion method according to the clinical and laboratory standards institute (CLSI) guideline [11]. The minimal inhibitory concentration (MIC) of carbapenems (imipenem, meropenem and ertapenem) were determined by gradient test strips (Liofilchem, Italy). Broth microdilution method was utilized to determine the MIC of colistin using colistin sulfate (Merck, Germany). According to CLSI M100-S30, a MIC = 2 µg/ml was interpreted as intermediate susceptibility, whereas a MIC of ≥ 4 µg/ml was considered as resistance [11]. E. coli ATCC 25922 was used as a control strain for antimicrobial susceptibility testing.

De novo assembly
The quality scores of the FASTQ paired-end files were checked using FastQC software [12]. Trimming and de novo assembly of short-read sequences was performed using Trimmomatic version 0.40 and SPAdes version 3.15.2, respectively [13] with k-mer = 99. The quality of contig assembly was assessed using QUAST software [14].

Chromosomal characterization Genetic features
The DFAST genetic analysis of purified contigs showed that chromosome length of 14 K. pneumoniae strains scaled from minimum 4,941,238 bp (for P38) to maximum 5,349,088 bp (for P37). The number of CDSs varried from 4694 to 5118. The highest N50 was calculated 215,119 belonging to P38. The GC content percentage for all isolated was 57%. The number of rRNA and tRNA varied from three to six and 51 to 80, respectively. See Additional file 1: Table S1.
Single amino acid polymorphisms (SAPs) Single amino acid substitutions of chromosomally-encoded proteins involved in AMR were detected by comparing related coding regions of each strains with K. pneumoniae NTUH-K2044 (NC_012731) as the reference colistinsusceptible strain using paired-wise alignment. SAPs resulted in resistance to cephalosporins; PBP3, elfamycin; EF-Tu, fosfomycin; UhpT, fluoroquinolones; ParC, GyrA and GyrB and multiple drugs; MarR were identified.    Table S2.

The bla OXA-48 and bla NDM-1 genetic environment
The gene annotation of OXA-48-producing plasmids showed that bla OXA-48 was flanked upstream/downstream by the LysR family transcriptional factor (lysR) in almost all strains. Additional file 3: Figure S1A. In P31, bla OXA-48 was flanked downstream by insertion sequence (IS) element transposase and bla CTX-M- 15 . In P32, two bla OXA-48 genes were identified on two distinc regions which environmet for one of them was similar to P31. The latter was flanked upstream by IS6 family transposase. In P33, the lysR and bla OXA-48 complex was flanked downstream by three hypotetical genes and a transposase.

Discussion
Our understanding about the genetic background of K. pneumoniae has increased worldwide. However, few WGS-based studies have focused on clinical K. pneumoniae isolates in Iran and consequently; the detalied genomic knowledge of cirulating STs in clinical settings is still limited in our country. WGS data for 14 clinical Co-CCRKp strains isolated between 2014 and 2016 in Iran were analyzed to provide a comparative genetic background that will help us to upgarde our information regarding epidemiology of AMR determinants. One of the most clinically important AMR genes in K. pneumoniae isolates is class D β-lactamase bla OXA-48 [15]. The bla OXA-48 gene has been carrying by various plasmids Inc types including IncL/M, IncN, and IncA/C [16]. However, the results of our plasmid replicon typing shows that bla OXA-48 is carried by various incompatibility groups including IncL, Col440I, IncFIB and mainly IncFII. The genetic analysis of bla OXA-48 environment in our strains indicates that the LysR family transcriptional factor gene is closely related to bla OXA-48 carriage in K. pneumoniae. Detection of bla OXA-48 in transferrable plasmids is a clinical emergency as it can rapidly be widespread among other strains and even other Enterobacterales mediating highly resistant outbreaks in healthcare settings. P6, P31, P35, P38 and P40 are considered clinically significant strains as they contain conjutaive plasmids carrying bla OXA-48 carbapenemase. This suggests that carbapenem-resistance soon will disseminate in ICU and other wards among Iranian medical settings. Also, the carriage of a conjugative OXA-48-producing plasmids in an outpatient is highly troublesome as it implies the higher prevalence rates of carbapenemase in the community in near future. In the presented study, we reported the first extra-chromosomal carriage of two bla OXA-48 in strain P32. The double carriage of bla OXA family by a single strain is rarely reported and is a matter of concern due to more rapid transmission of this significant carbapenemase. Sherchan and colleagues reported two copies of bla OXA-181 on ST147 K. pneumoniae from Nepal in 2020 which was previously reported from Pakistan, the United Arab Emirates and Korea [17]. P6 and P26 strains were detected as the only NDM-1 producing strains in our study indicating lower rate of bla NDM-1 carriage compared to other β-lactamase such as bla CTX-M-15 and bla TEM-1 . However, higher prevalence of bla NDM-1 was reported form other Middle Eastern countries specifically Egypt and Saudi Arabia [18,19]. In the study by Ghaith et al. [20] 52.2% of K. pneumoniae strains isolated from neonatal ICU in Cairo were NDM-1 producers. The prevalence of bla NDM-1 among Enterobacterales is mediated by rapid dissemination of conjugative plasmids [16]. In our study, none of the NDM-1 producing plasmids were conjugative. bla NDM-1 gene has been detected on plasmids of various incompatibility groups including IncF, IncA/C, IncL/M, IncH, IncN, and IncX3 or untypeable [16]. However, both of NDM-1 producing plasmids in our study belonged to IncFIB and IncHI1B. Co-carriage of bla NDM-1 and bla OXA-48 genes was seen in strain P6. Solgi et al. investigated the ESBLs carriage of 71 clinical carbapenem-resistant Enterobacterales in an Iranian hospital in Tehran. In this study, among 62 bacterial isolates, 46% and 37% of the isolates harbored bla NDM-1 and bla OXA-48 , respectively and co-carriage of bla NDM-1 and bla OXA-48 was detected in 16% of the isolates. Also, the plasmid incompatibility types IncFII and IncA/C were identified among the NDM-1 producing isolates, while only IncL/M was detected among OXA-48 producers [16]. Co-existence of the bla NDM-1 and bla OXA-48 genes among CR-KP isolates was previously reported from other Middle Eastern countries such as Egypt. In a recent study by El-Domany and colleagues [21], 50 isolates co-carrying bla NDM-1 and bla OXA-48 were reported among 230 K. pneumoniae clinical isolates. In this study, the rate of bla NDM-1 and bla OXA-48 was reported 70.0% and 52.0%, respectively. In our study, ST11, ST147, ST893 and ST101 were bla OXA-48 and bla NDM-1 carriers in agreement with previous reports of OXA-48 and NDM-1 producing K. pneumoniae from Iran [16,22]. These consensus reports suggest successful circulation of mentioned STs in Iran. The clonal carriage of bla OXA-48 and bla NDM-1 among K. pneumoniae strains may be region-specific in a single geographical territory despite high burden of individual transits. Accordingly, a significant association was reported for ST199 and ST152 with bla OXA-48 and bla NDM-1 carriage form Saudi Arabia, respectively [23,24].
Production of OXA-48 in K. pneumoniae plasmids seems to be typically concurrent with other ESBLs mainly bla CTX-M and bla TEM. In one study, 88 of 94 of K. pneumoniae isolates in an Iranian hospital harbored bla SHV, bla CTX-M-15 and bla TEM-1 concurrently while only one and two isolates solely carried bla CTX-M-15 and bla SHV , respectively [15]. The transmission of major ESBLs among K. pneumoniae has been reported in conjugative IncL/M and IncFII plasmids with ability of interspecies transfer [23,25]. Accordingly, in our extra-chromosomal analysis, IncFII was a major plasmid replicon type and IncL was detected in two conjugative ESBL producer plasmids.
Resistance to colistin has been increasingly reported in the world including the Middle East region. A high resistance rate of 16.9% was reported between 2015 and 2016 from Iran [26]. Totally, 524 colistin-resistant K. pneumoniae isolates were reported from Turkey and Iran between 2013 and 2018 [27]. Jafari et al. [28] reported an increase of colistin resistance up to 50% in carbapenemresistant K. pneumoniae isolates.
In the presented study; amino acid substitution, premature termination, deletion and insertion of IS element were identified in the coding region of MgrB as well as PmrA/PmrB, PhoP/PhoQ, RamA and CrrB regulators. MgrB inactivation was identified in 42% of the strains including P32, P33, P35, P37, P38 and P42. Substitution of cysteine at position 28 of MgrB with tryptophan was detected in P33 and P42. Position 28 is considered as an important region for amino acid substitution due to the key role of disulfide bonds of cysteine residue in MgrB functionality [29]. Similar SAPs such as C28F, C28Y and C28S has been previously reported in several different studies [6,30]. In addition, truncation of MgrB at position 30 (Q30*) was identified in P35 and P38. Position 30 is highly prone to be modified in long exposure to colistin and therefore; (Q30*) has been commonly reported in other studies [29]. Insertional inactivation of MgrB was identified in only one strain (P32) mediating by insertion of ISKpn72 element (belonging to IS4 family) in coding region of MgrB. Zhang et al. recently demonstrated the phenotypic switch of colistin-susceptible K. pneumoniae strains to colistin-resistant ones by horizontal transfer of an ISKpn72 carrying-plasmid. Therefore, in spite of chromosomal origin of MgrB; its modification can be plasmid-mediated and depended on conjugation processes. However, this phenomenon is not probable in our study because both plasmids of P32 were non-conjugative [31]. In this study, crrB gene was not carried by the P37 chromosome. The absence of crrB was previously reported in the study by Jayol et al. [32] which may be due to the differences in the lateral acquisition of the crrAB operon in K. pneumoniae. Thomas et al. [33] claimed that the lack of CrrB leads to significantly higher MIC colistin and bacterial  [37,38]. Therefore, the mechanism of colistin resistance needs to be carefully interpreted.
The prevalence of hypervirulence among MDR K. pneumoniae is increasing worldwide [39]. In a metaanalysis conducted by Sanikhani et al. [40] the rate of reported hypervirulent K. pneumoniae isolates was determined 21.7% globally which majority of them were from China. In our study, P35 and P38 carried non-conjugative hypervirulence plasmids belonging to repB plasmid replicon type and considered as MDR-hypervirulent strains. This may suggest that P35 and P38 possess hypervirulent background which acquired MDR plasmids through conjugative processes. The rate of hypervirulence was 14% in our study and both strains were positive for iutA, iucD, iucC, iucB and iucA. Few studies have reported hypervirulent K. pneumoniae in Iran. Pajand et al. investigated the presence of hypervirulence genes in K. pneumoniae clinical isolates. peg344, iucA, rmpA, rmpA2, iroB1, iroB2 and iutA were detected among all carbapenem-resistant isolates with 1.8%, 1.8%, 1.8%, 1.8%, 7.3%, 12.7% and 18.2% rate of carriage, respectively. The results of their study indicated that the prevalence of iutA was even higher in NDM-1 producing isolates [41]. Taraghian et al. detected 11 MDR-hypervirulent strains among 105 urinary tract K. pneumoniae isolates. In their study, ESBLs were identified in all hypervirulent strains and their carriage by hypervirulent stains were significantly higher compared to the classical ones [42]. In addition, Sanikhani et al. [43] reported the prevalence rate of 21.38% among 477 K. pneumoniae clinical isolates with high percentage of MDR and high-level resistance to imipenem.
The ICEKp analysis highlighted that the majority of our strains (78%) carried ICEKp3. According to an investigation by Lam et al. among 2498 K. pneumoniae genomes belonging to 37 different STs; ICEKp4, ICEKp3, ICEKp10 and ICEKp5 were the most common types of ICEKp. The results of this study also revealed that ICEKp variants can be integrated more frequently in specific STs compared to the other ones. The authors showed that almost 40% of the ICEKp positive K. pneumoniae strains belonged to CC258 [44]. Similarly, another study from South America indicated that the rate of ICEKp carriage is higher in strains belonging to ST11 and ST340. In our study, almost all ST11 strains harbored ICEKp3. This specific carriage of ICEKp3 in K. pneumoniae ST11 was previously reported by a Turkish study in 2019 [45].

Conclusion
In this study, we assembled the whole-genomes of 14 colistin-resistant OXA-48-producing K. pneumoniae referred to two Iranian hospitals as well as their plasmids. The prevalence of bla OXA-48 in our strains was very high as well as bla SHV, bla CTX-M-15 and bla TEM-1 . MDR conjugative plasmids as well as hypervirulent plasmids were detected in some strains. Amino acid substitution, premature termination, deletion and insertion of an IS element were identified in the coding region of MgrB. SAPs in PmrA, PmrB, PhoQ, RamA and CrrB regulators were involved in colistin resistance. Pan-genomic investigations can provide data on AMR reservoir of highlyresistant K. pneumoniae at clinical and even community levels and ultimately provide a realistic status of AMR countrywide.