Stepwise evolution of pandrug-resistance in Klebsiella pneumoniae

1The “golden era” when modern medicine saved lives through antibiotic treatment is under serious threat. In 2013, the Centers for Disease Control and Prevention (CDC) released a landmark report on “Antibiotic Resistance Threats”. Three microorganisms were tagged as posing a threat level of urgent – Clostridium difficile, carbapenem-resistant Enterobacteriaceae (CRE) and drug-resistant Neisseria gonorrhoeae. CRE, which include organisms such as Klebsiella pneumoniae and Escherichia coli, are resistant to almost all currently available antibiotics. Almost 50% of patients who develop bloodstream infections with these organisms die from the infection. In healthcare settings, carbapenem resistant Enterobacteriaceae have increased sharply over the past decade. Carbapenem resistance is typically mediated by the production of beta-lactamases, and patients with CRE infections are treated with last-resort antibiotics such as colistin.

2The CDC and the European Centre for Disease Prevention and Control (ECDC) have jointly developed definitions for multidrug-resistant, extensively drug-resistant and pandrug-resistant bacteria. Pandrug-resistance implies non-susceptibility to all commercially available antibiotics relevant to the treatment of a particular bacterial infection. Although there has been an anecdotal report of probable pandrug-resistance in K. pneumoniae, no such isolates have been comprehensively analyzed.

3In this manuscript, we describe the genetic basis of pandrug-resistance in a K. pneumoniae isolate using single molecule real-time (SMRT) sequencing. We show that a genetic element conferring resistance to carbapenem antibiotics has been acquired and mobilized, leading to insertional inactivation of a gene that results in resistance to colistin. Overall, our analysis provides a comprehensive description of a pandrug-resistant K. pneumoniae isolate at the whole genome level.

4An 87 year old man, hospitalized in the United Arab Emirates in April 2014, was found to be colonized with multidrug-resistant Klebsiella pneumoniae. The isolate grew from urine and a pre-sacral pressure area but blood cultures were sterile. Susceptibility testing by way of a commercial semi-automated method (Vitek, bioMérieux) showed resistance to all antibiotics tested. The urinary isolate (strain MS6671) was therefore sent to a reference laboratory for further testing. Other K. pneumoniae isolates with this antibiotic resistance phenotype were not detected at the index patient’s hospital.

5MS6671 was found to be non-susceptible to all antibiotics tested, which includes cephalosporins, penicillins, carbapenems, aztreonam, aminoglycosides, ciprofloxacin, colistin, tetracyclines, tigecycline, chloramphenicol, trimethoprim-sulfamethoxazole and fosfomycin (Table 1). Thus, the isolate can truly be described as pandrug-resistant.

6The complete genome of K. pneumoniae MS6671 consists of a circular chromosome 5,402,900 base-pairs in length with an average G-C content of 57%, five circular plasmids and a linear plasmid prophage (Supplementary Table S1). The sequence type of the isolate was ST147. The chromosome of MS6671 is highly similar to K. pneumoniae NTUH-k2044; a hypervirulent strain associated with liver abscess and meningitis, with most variation attributable to differences in their mobile genetic element (MGE) content (Supplementary Figure 1). Further details of the complete genome are provided in the Supplementary Results.

7In order to determine the genetic basis of pandrug-resistance, we interrogated the genome to identify acquired and intrinsic resistance genes. The majority of acquired antibiotic resistance genes were located on the chromosome, with most beta-lactamase and aminoglycoside resistance genes carried within two copies of a class 1 integron or as part of mobile elements that incorporate the ISEcp1 insertion sequence (Table 1, Supplementary Table S2 and Supplementary Results). Mutations in gyrA and parC that have previously been linked with fluoroquinolone resistance (GyrA Ser83Ile and ParC Ser80Ile) were identified. Fosfomycin resistance was mediated by a chromosomally encoded copy of fosA. Mutations in chromosomal genes encoding major outer membrane porins (OmpK35 and OmpK36) were also identified. A novel variant of the ompK36 gene was encoded on the chromosome. The amino acid sequence change is located in loop 3 (L3) of the porin, which constitutes the porin channel eyelet. L3 mutations have previously been associated with increased resistance to carbapenems. Additionally, ompK35 has been disrupted by IS insertion. Inactivation of ompK35 has been associated with increased resistance to a number of different classes of antibiotics, including quinolones and cephalosporins. Genes encoding three beta-lactamases, including an extended-spectrum beta-lactamase (ESBL) and a carbapenemase, were detected at different genomic locations – blaSHV-36, blaCTX-M-15, blaOXA-181.

8OXA-181 is an oxacillinase capable of hydrolysing carbapenems. Three copies of an ISEcp1-blaOXA-181 transposon were identified throughout the chromosome (Fig. 1). One of these insertions has resulted in the inactivation of the mgrB gene, a negative regulator of phoPQ. Insertions in mgrB have previously been shown to cause colistin resistance in K. pneumoniae clinical isolates. Examination of the DNA flanking the ISEcp1-blaOXA-181 transposons shows that the primary insertion site is within MS6671_10430, followed by intra-chromosomal transposition of ISEcp1-blaOXA-181 and a 37 bp fragment of MS6671_10430 to two other locations in the genome (Fig. 2). The three transposons are bracketed by imperfect 14 bp inverted repeats and flanking 5 bp direct repeats (TATCT, TGAAA and TATAA), providing direct evidence for their transposition activity (Supplementary Table S3).

9In a similar fashion, a single copy of ISEcp1-blaCTX-M-15 has inserted into ompK35, leading to inactivation this gene (Fig. 1). The ISEcp1-mediated mobilisation and transposition of blaCTX-M-15, blaOXA-181 and other clinical relevant beta-lactamase resistance genes, including blaCMY and blaACC, has been reported previously.

10A class 1 integron was identified on the chromosome encoding multiple antibiotic resistance genes (arr-3, aac(6′)-Ib-cr, rmtF, catB1) (Fig. 1). These genes result in resistance to rifampin, all aminoglycosides and chloramphenicol. A near-identical copy of this integron was also found on one of the six plasmids (Supplementary Figure S2).

11This is the first genomic analysis of a pandrug-resistant CRE isolate, as defined by the rigorous CDC/ECDC assessment criteria. With the advantage of long-reads provided by SMRT sequencing we were able to identify the genomic context of multiple resistance elements. In contrast to short-read technologies, SMRT sequencing allows complex resistance elements to be properly characterized. This technology platform was used to investigate the German E. coli O104:H11 outbreak and more recently to identify plasmid-borne resistance in a large-scale study of CRE following an outbreak at the National Institute for Health Clinical Center. Critically, elucidation of the complete K. pneumoniae MS6671 genome using long-read sequencing enabled the context of multiple, identical carbapenem resistance elements to be determined. Based on this analysis we propose a model for the development of pandrug-resistance in this K. pneumoniae isolate, whereby mobile resistance determinants are responsible for driving additional resistance. In this example, ISEcp1 carrying the blaOXA-181 carbapenem resistance gene has inserted three times in the chromosome, with one event causing colistin resistance by insertional inactivation of mgrB.

12ISEcp1-like insertion sequences are the most common genetic element associated with blaCTX-M, blaCMY and blaACC genes and have more recently been associated with blaOXA-181. By recognizing a variety of DNA sequences as right inverted repeats (IRR), ISEcp1s are capable of mobilising adjacent genes and inserting at new location. Similar to previous reports on the hydrolytic activities of OXA-181, elevated MICs for ertapenem, imipenem, meropenem and doripenem were observed for MS6671, indicating hydrolytic activity of OXA-181 towards these carbapenems and a possible blaOXA-181 copy number effect (Table 1). Notably, doripenem resistance was higher than previously reported. The ompK36 variant encoded by MS6671 has previously been associated with increased resistance to doripenem and doripenem-colistin, and may contribute to the elevated MIC for doripenem observed in MS6671. Porin deficient E. coli expressing OXA-48-like beta-lactamases have also been shown to have elevated MICs towards carbapenems.

13Inactivation of mgrB has recently been associated with resistance to colistin, and appears to be the most common mechanism for polymyxin resistance in K. pneumoniae. Specifically, disruption of mgrB results in over expression of the phoPQ signaling system and of the pmrHFIJKLM operon which controls modification of LPS, the target of polymyxin antibiotics. Insertional inactivation of mgrB with IS5-like or IS1 elements has been previously reported, however, the present study is the first to show colistin resistance caused by insertion of a carbapenem resistance element itself. While we cannot rule out the possibility that this mechanism may have occurred in other colistin-resistant K. pneumoniae carrying ISEcp1-blaOXA-181, the generation of a complete genome sequence of MS6671 provides unequivocal evidence for this novel insertion event.

14We also found a fourth ISEcp1 element encoding an ESBL (CTX-M-15), which was inserted within the outer membrane porin gene ompK35. Disruption of ompK35 reduces the permeability of the outer membrane and mutants lacking this porin have increased resistance to quinolones, tetracyclines, beta-lactams and chloramphenicol. Pan-aminoglycoside resistance was mediated by the rRNA methyltransferase RmtF, which was encoded on both chromosomal and plasmid copies of a class 1 integron. We did not have access to investigational antibiotics, such as plazomicin, eravacycline, ceftazidime/avibactam or ceftolozane/tazobactam, to assess their potency against MS6671.

15MS6671 was defined by multi-locus sequence typing (MLST) as being ST147. K. pneumoniae ST147 was first described in Hungary in 2008. Subsequently, it has been frequently associated with carbapenem resistance, with ST147 producing KPC well described in Greece and Italy. For example, epidemics of VIM-producing carbapenem resistant K. pneumoniae ST147, and KPC-2-producing carbapenem resistant K. pneumoniae have been reported in Greece. Notably, ST147 isolates carrying both blaVIM and blaKPC-2 genes were identified. Carbapenem resistant ST147 carrying the blaNDM-1 gene have been isolated in Iraq, Switzerland, Canada and the United Kingdom. In MS6671, carbapenem resistance was most likely mediated by the beta-lactamase OXA-181, possibly in combination with permeability defects as has been reported previously in other strains. ST147 K. pneumoniae producing OXA-181 have been previously reported from the Indian sub-continent. Clearly, in addition to the KPC-producing, carbapenem-resistant K. pneumoniae ST258 clone, ST147 also represents a clone of K. pneumoniae with a potential for global significance.

16Fortunately, in six months there have been no further isolates with this resistance phenotype at the index patient’s hospital. However, the occurrence of this strain in the Arabian Gulf is of great significance. OXA-48-like-producing K. pneumoniae are frequent in this region. It is unknown if this strain originated in the index patient, in another patient at the same hospital or was imported from another hospital, perhaps in another country. There are a large number of expatriates in the Gulf region, and travel to the Indian sub-continent, Europe and the United States is frequent. The potential for international transfer of multidrug-resistant bacteria emphasizes the need for global surveillance efforts as one part of a strategy to control antibiotic resistance.

17In summary, we have provided the first report of a pandrug-resistant isolate of CRE using high-resolution genome data. The CDC has denoted CRE as an urgent threat. The emergence of this highly resistant strain, in a clone that has proven capable of causing outbreaks, raises this threat level even higher.

18The K. pneumoniae isolate (hereafter referred to as MS6671) was sent to a reference laboratory (University of Queensland, Centre for Clinical Research) where confirmatory susceptibility testing was performed in order to determine the minimal inhibitory concentrations of all antibiotics used by the Centers for Disease Control and Prevention (CDC) and European Centre for Disease Prevention and Control (ECDC) in defining a pandrug-resistant isolate. Susceptibility of most tested antibiotics was determined using Etests and following the breakpoints of the European Committee for Antibiotic Susceptibility Testing (EUCAST), except for cefazolin, cefoxitin, cefotetan, tetracycline, doxycycline, and minocycline which were determined using the breakpoints of the Clinical and Laboratory Standards Institute (CLSI). The minimum inhibitory concentration of colistin (sulfate, Sigma-Aldrich) was determined by broth microdilution in cation-adjusted Mueller-Hinton broth (Oxoid).

19Pacific Biosciences (PacBio) RS II Single-Molecule Real Time (SMRT) sequencing of K. pneumoniae MS6671 was performed using ~4 μg of the genomic DNA sheared using g-TUBETM (Covaris®) into fragments size targeted at 10 kb. Purification of the sheared DNA was then carried out using 0.45-fold volume of washed Agencourt AMPure XP magnetic beads (Beckman Coulter Inc.). SMRTbell template libraries were subsequently prepared using the commercial Template Preparation Kit from Pacific Biosciences Inc. that involved steps of DNA end repair, adapters ligation followed by exonuclease digestion of incompletely ligated products. Next, 0.83 nM of the libraries were then annealed with sequencing primers followed by binding to 50 nM of P4 DNA polymerase, as provided in the Template Binding Kit from Pacific Biosciences Inc. For enhanced loading efficiency, 15 pM of the bound complexes were immobilized into Magbeads (Pacific Biosciences Inc.) prior to loading into the sequencing zero-mode waveguides (ZMWs). Duration for the sequence collection was set at 180 minutes with stage start option. Reads with length that were less than 50 bp were filtered off upon acquisition of the sequencing data and minimum polymerase read quality was set at 0.75.

20De novo genome assembly of PacBio SMRT reads from the K. pneumoniae MS6671 genome was performed using the hierarchical genome assembly process (HGAP) from the PacBio SMRT analysis software suit (version 2.2.0), with default parameters and a seed read length cut-off of 5 kb. Following assembly, all contigs were screened for duplicate sequences at their 3′ and 5′ ends. Overlapping sequences were manually trimmed and joined based on sequence similarity. Individual contigs with duplicate sequences on their 5′ and 3′ ends were manually trimmed and circularised. Following circularisation the chromosome and plasmid sequences were polished using quiver whereby the raw reads were mapped back to the chromosome and plasmid sequences to validate the assembly and resolve any remaining sequence errors. Non-circularised chromosomal and plasmid contigs were closed using primers designed on their 5′ and 3′ ends. The amplified PCR products were sequenced by the Australian Genome Research Facility and their sequences were manually integrated into the assembly.

21Gene calling and automatic functional annotation of the complete MS6671 chromosome and plasmids was performed using Prokka (Prokka: Prokaryotic Genome Annotation System - http://vicbioinformatics.com/) identifying 5,054 putative coding regions on the chromosome with an additional 644 putative coding regions distributed amongst the 5 plasmids and linear plasmid prophage. The complete annotated genome sequence has been deposited at the European Nucleotide Archive (Bioproject: PRJEB7538, Accessions: LN824133-LN824139).

22Initial identification of antimicrobial resistance genes from the complete PacBio assembly was performed using ResFinder (version 2.0). Additional screening for antimicrobial resistance genes was performed by comparison (BLASTp; sequence identity > =40%; E-value < =0.0001) of all predicted coding regions against the Antibiotic Resistance Genes Database (ARDB) and the Comprehensive Antimicrobial Resistance Database (CARD). Antimicrobial resistance genes were then subject to manual inspection to improve their functional annotation, correct start sites and identify point mutations, which may contribute to a resistant phenotype. Finally, resistance gene loci were screened for known insertion sequences and integrons by comparison against the ISFinder database and Integrall, respectively.

23How to cite this article: Zowawi, H. M. et al. Stepwise evolution of pandrug-resistance in Klebsiella pneumoniae. Sci. Rep. 5, 15082; doi: 10.1038/srep15082 (2015).