High-Level Carbapenem Resistance in OXA-232-Producing Raoultella ornithinolytica Triggered by Ertapenem Therapy

OXA-232 is an OXA-48-group class D β-lactamase that hydrolyzes expanded-spectrum cephalosporins and carbapenems at low levels. Clinical strains producing OXA-232 are sometimes susceptible to carbapenems, making it difficult to identify them in the clinical microbiology laboratory.

ABSTRACT

OXA-232 is an OXA-48-group class D β-lactamase that hydrolyzes expanded-spectrum cephalosporins and carbapenems at low levels. Clinical strains producing OXA-232 are sometimes susceptible to carbapenems, making it difficult to identify them in the clinical microbiology laboratory. We describe the development of carbapenem resistance in sequential clinical isolates of Raoultella ornithinolytica carrying bla_OXA-232 in a hospitalized patient, where the ertapenem MIC increased from 0.5 μg/ml to 512 μg/ml and the meropenem MIC increased from 0.125 μg/ml to 32 μg/ml during the course of ertapenem therapy. Whole-genome sequencing (WGS) analysis identified loss-of-function mutations in ompC and ompF in carbapenem-resistant isolates that were not present in the initial carbapenem-susceptible isolate. Complementation of a carbapenem-resistant isolate with an intact ompF gene resulted in 16- to 32-fold reductions in carbapenem MICs, whereas complementation with intact ompC resulted in a 2-fold reduction in carbapenem MICs. Additionally, bla_OXA-232 expression increased 2.9-fold in a carbapenem-resistant isolate. Rapid development of high-level carbapenem resistance in initially carbapenem-susceptible OXA-232-producing R. ornithinolytica under selective pressure from carbapenem therapy highlights the diagnostic challenges in detecting Enterobacteriaceae strains producing this inefficient carbapenemase.

BACKGROUND

Carbapenems are last-resort agents in the management of infections caused by multidrug-resistant Gram-negative pathogens in critically ill patients. Multiple carbapenem resistance mechanisms have emerged in recent years, jeopardizing our ability to treat these difficult infections. KPC, NDM, and OXA-48 are the most prevalent carbapenemases in the family Enterobacteriaceae globally. In North America, KPC is most common, but strains producing NDM and OXA-48-group carbapenemases have been increasingly reported (1).

OXA-48-group carbapenemases hydrolyze penicillins at a high level but demonstrate lower levels of hydrolysis for expanded-spectrum cephalosporins and carbapenems than KPC (2). For this reason, clinical strains producing OXA-48-group enzymes may test susceptible to carbapenems and cephalosporins unless extended-spectrum β-lactamases (ESBLs) are coproduced, making it difficult to identify them in the clinical microbiology laboratory (3). OXA-48 was the first OXA-48-group carbapenemase to be identified from a carbapenem-resistant Klebsiella pneumoniae strain in Turkey in 2001 (2). Since then, it has disseminated through countries in the Middle East, Europe, and North Africa (3). Furthermore, several close variants of OXA-48, such as OXA-181 and OXA-232, have been reported frequently in Enterobacteriaceae strains (4, 5). As of December 2017, 146 unique Enterobacteriaceae strains producing OXA-48-group enzymes, including OXA-48, OXA-181, and OXA-232, had been reported to the CDC in the United States (1, 6).

OXA-232 is a variant of the OXA-48 carbapenemase that was identified from clinical isolates in India in 2012 (4). This enzyme possesses hydrolytic activities against penicillins, cephalosporins, and carbapenems that are comparable to OXA-48 and by itself does not confer carbapenem resistance as defined by the CLSI and EUCAST. However, carbapenem resistance is manifested when it is expressed in an Escherichia coli strain that also lacks OmpF and OmpC porin channels (4). OXA-48-group carbapenemases are most commonly produced by Klebsiella pneumoniae, E. coli, and Enterobacter spp.; however, the presence of OXA-48 has been reported in other Enterobacteriaceae species, such as Citrobacter, Providencia, and Raoultella spp. (7).

Raoultella spp. are encapsulated Gram-negative aerobic bacilli belonging to the order Enterobacteriales and the family Enterobacteriaceae. Raoultella ornithinolytica, previously known as Klebsiella ornithinolytica, is often misidentified as Klebsiella oxytoca by conventional biochemical identification methods (8–10). However, with implementation of matrix-assisted laser desorption ionization–time-of-flight mass spectrometry (MALDI-TOF MS) in clinical laboratories, it is increasingly recognized as a significant nosocomial pathogen associated with biliary and urinary tract infections, as well as bacteremia (11). R. ornithinolytica strains that have acquired ESBL genes, as well as carbapenemase genes such as bla_KPC, bla_NDM, bla_IMP, and bla_OXA-48, have been reported (12, 13).

In this report, we describe the rapid development of high-level carbapenem resistance in sequential clinical isolates of Raoultella ornithinolytica carrying bla_OXA-232, which were initially susceptible to carbapenems and were therefore not recognized as carbapenemase-producing organisms.

RESULTS

Resistance phenotype of sequential R. ornithinolytica isolates.

Sequential isolates of R. ornithinolytica YDC775-2, YDC775-9, and YDC775-15 were obtained over an 18-month period from a patient with recurrent abdominal abscess.

The patient was initially hospitalized due to obstructive cholangitis in the setting of biliary tree malignancy, for which he was treated with a 3-week course of piperacillin-tazobactam therapy. The patient subsequently underwent resection of the malignancy and 3 days postoperatively developed intra-abdominal abscess. Initial percutaneous drainage of the abscess recovered isolate YDC775-2, which was reported to be susceptible to carbapenems and cephalosporins and resistant only to piperacillin-tazobactam as determined using MicroScan WalkAway (imipenem MIC, <1 μg/ml; meropenem MIC, <1 μg/ml; piperacillin-tazobactam MIC, >64 μg/ml). Retrospective antimicrobial susceptibility testing conducted by a research lab confirmed carbapenem and cephalosporin susceptibility and piperacillin-tazobactam resistance (ertapenem MIC, 0.5 μg/ml; meropenem MIC, 0.125 μg/ml; cefotaxime MIC, 0.25 μg/ml; ceftazidime MIC, 0.5 μg/ml; cefepime MIC, 0.125 μg/ml; and piperacillin-tazobactam MIC, >64 μg/ml). In response to these results, cefepime therapy was initiated, and the patient underwent exploratory laparotomy with drainage of the fluid collection on day 10 after the initial surgery. He was subsequently transitioned to ertapenem therapy for ease of administration. However, abdominal fluid collection recurred, and on day 30 after initial surgery and 10 days after ertapenem initiation, a new drain was placed, yielding isolate YDC775-9 from the fluid. Isolate YDC775-9 was resistant to imipenem and meropenem as determined by MicroScan (imipenem MIC, >8 μg/ml; meropenem MIC, >8 μg/ml). Retrospective testing showed significant increases in ertapenem and meropenem MICs to 512 μg/ml and 32 μg/ml, respectively. In addition, isolate YDC775-9 showed increased MICs to cefotaxime and cefepime of 4 μg/ml and 2 μg/ml, respectively. The patient underwent several abscess drainage procedures, and ertapenem was switched to ciprofloxacin based on susceptibilities (ciprofloxacin MIC, <2 μg/ml). These interventions led to eventual resolution of abscess. Eighteen months later, YDC775-15 was isolated from recurrent abdominal abscess after 3 weeks of empirical ertapenem treatment. It showed reversion of ertapenem and meropenem MICs to 1 μg/ml and 0.5 μg/ml, respectively. Additionally, MICs to cefotaxime and cefepime were 0.5 μg/ml and 0.25 μg/ml, respectively, which were comparable to those of the initial isolate. The modified carbapenem inactivation method (mCIM) test performed retrospectively was positive for carbapenemase production in all isolates, including the carbapenem-susceptible ones. The patient was treated with tigecycline based on tigecycline Etest susceptibility results (MIC, 0.5 μg/ml). Additionally, he underwent repeat abscess drainage with no further recurrence. The β-lactam MICs of the sequential isolates are shown in Table 1. Figure S1 in the supplemental material summarizes the clinical course.

TABLE 1.

MICs of R. ornithinolytica clinical isolates^a

R. ornithinolytica isolate	MIC (μg/ml)
	Ertapenemb	Meropenemc	Cefepimed	Cefotaximee	Ceftazidimef
YDC775-2	0.5	0.125	0.125	0.25	0.5
YDC775-9	512	32	2	4	1
YDC775-15	1	0.5	0.25	0.5	1

^aAll isolates were isolated from abdominal fluid.

^bThe breakpoint for ertapenem approved by the Clinical and Laboratory Standards Institute is ≤0.5 μg/ml for susceptible and ≥2 μg/ml for resistant.

^cThe breakpoint for meropenem approved by the Clinical and Laboratory Standards Institute is ≤1 μg/ml for susceptible and ≥4 μg/ml for resistant.

^dThe breakpoint for cefepime approved by the Clinical and Laboratory Standards Institute is ≤2 μg/ml for susceptible and ≥16 μg/ml for resistant.

^eThe breakpoint for cefotaxime approved by the Clinical and Laboratory Standards Institute is ≤1 μg/ml for susceptible and ≥4 μg/ml for resistant.

^fThe breakpoint for ceftazadime approved by the Clinical and Laboratory Standards Institute is ≤4 μg/ml for susceptible and ≥16 μg/ml for resistant.

Whole-genome sequencing.

There were 4 and 9 core genome single-nucleotide polymorphisms (SNPs) between the initial isolate YDC775-2 and isolates YDC775-9 and YDC775-15, respectively. There was no evidence of gene gain or loss between the isolates. The findings indicated persistence and evolution of the same strain over the course of the patient’s illness rather than reinfection with another R. ornithinolytica strain or mixed infection with multiple strains.

β-lactamases of R. ornithinolytica isolates.

The genome sequences of the serial R. ornithinolytica isolates were examined for the presence of β-lactamase genes and other carbapenem resistance determinants. bla_ORN-1, the intrinsic, chromosomal β-lactamase gene of R. ornithinolytica, and plasmidic bla_OXA-232 were the only β-lactamase genes present. While the kinetic parameters of enzyme ORN-1 have not been established, it shares 94% amino acid sequence similarity with enzyme PLA-1 from Raoultella planticola, which is known to possess weak hydrolytic activity against penicillins and cephalosporins, including cefotaxime and cefepime, but not imipenem (14).

bla_OXA-232 was located on a 6.1-kb ColE-like plasmid, which had >99% sequence identity with plasmid 4 of K. pneumoniae PittNDM01 (CP006802), a clinical strain previously reported from the same hospital, with only 2 nucleotide differences. The plasmids in the three isolates had identical sequences. Nonetheless, expression of bla_OXA-232 was 2.9-fold higher in carbapenem-resistant isolate YDC775-9 than in carbapenem-susceptible isolates YDC775-2 and YDC775-15 (P = 0.02 by the Kolmogorov-Smirnov test). There was no fold difference in plasmid copy number as determined by quantitative reverse transcription-PCR (qRT-PCR). Upon examination of the promoter sequences of bla_OXA-232, no variations were identified across the study isolates. The source of differential expression of bla_OXA-232 therefore remained unclear in this investigation.

Porin analysis.

Manual curation of SNPs and indels revealed the presence of single nucleotide variants in the ompC and ompF sequences in association with the carbapenem resistance phenotype. Compared with YDC775-2, isolate YDC775-9 had an early stop codon in ompC at amino acid position 112 and a deletion in ompF, which led to a frameshift and early termination at amino acid position 250. YDC775-15, which was intermediate to ertapenem and susceptible to meropenem, possessed ompC and ompF sequences that were identical to the initial carbapenem-susceptible isolate YDC775-2 (Table 2).

TABLE 2.

Correlation between the carbapenem resistance phenotype and isolate genotype

	YDC775-2	YDC775-9a	YDC775-15
Hospital day	Day 1	Day 28	Day 639
Ertapenem MIC (μg/ml)	0.5	512	1
β-Lactamases	blaORN1a, blaOXA-232	blaORN1a, blaOXA-232	blaORN1a, blaOXA-232
ompC	Wild type	C336A	Wild type
Effect		Tyr112Stop
ompF	Wild type	ΔG565	Wild type
Effect		Frameshift, stop at AA 250
Other resistance genes	fosA, oqxA, oqxB, qnrB19	fosA, oqxA, oqxB, qnr19	fosA, oqxA, oqxB, qnrB19

^aOXA-232 expression was 2.9-fold higher in isolate YDC775-9 than in isolates YDC775-2 and YDC775-15.

Contribution of ompC and ompF mutations to carbapenem resistance.

To clarify the contribution of the observed ompC and ompF mutations on the carbapenem MICs, carbapenem-resistant isolate YDC775-9, possessing ompC and ompF mutations leading to truncated porin proteins, was complemented with intact ompC and ompF genes from carbapenem-susceptible isolate YDC775-2. Complementation with intact ompC had minimal impact on carbapenem MICs, with the ertapenem MIC decreasing by 2-fold from 512 μg/ml to 256 μg/ml and the meropenem MIC decreasing from 32 μg/ml to 16 μg/ml. On the other hand, complementation with intact ompF led to a significant decrease of the ertapenem MIC from 512 μg/ml to 16 μg/ml and the meropenem MIC from 32 μg/ml to 2 μg/ml. Complementation with ompC in addition to ompF did not result in a further decrease of carbapenem MICs. The results of this complementation experiment are summarized in Table 3.

TABLE 3.

MICs of R. ornithinolytica complemented with OmpC and OmpF from YDC775-2

R. ornithinolytica isolate	MIC (μg/ml)
	Ertapenem	Meropenem
YDC775-9	512	32
YDC775-9 + OmpC	256	16
YDC775-9 + OmpF	16	2
YDC775-9 + OmpC + OmpF	32	4
YDC775-9 + pMQ132	512	32
YDC775-9 + pMQ450	512	64

DISCUSSION

OXA-48-group carbapenemases are increasingly reported as one of the three major groups of carbapenemases along with KPC and NDM, but their low hydrolytic activity toward expanded-spectrum cephalosporins and carbapenems makes it difficult to detect them in the absence of other coexisting β-lactam resistance mechanisms. The case described here summarizes challenges faced by clinicians, as the initial R. ornithinolytica isolate, despite carrying bla_OXA-232, was susceptible to all carbapenems and oxyimino-cephalosporins but developed high-level resistance to carbapenems within 10 days of the start of carbapenem therapy. Carbapenem therapy failure in bla_OXA-48-containing carbapenem-susceptible isolates has been reported in both in vitro studies and clinical studies, but the exact mechanisms of failure have not been reported (15–17).

We ascribe loss-of-function mutation in ompF in combination with increased expression of bla_OXA-232 to be the main driver of carbapenem resistance in R. ornthinolytica, with additional incremental contributions of loss-of-function mutations in ompC. There was no additive or synergistic effect in restoring both porins on the reduction of the carbapenem MIC in isolate YDC775-9. This is in contrast to previous work on K. pneumoniae, where OmpK36, the OmpC homolog, was reported to be the main contributor to carbapenem resistance (18). It is possible that this discrepancy is due to a difference in porin structure or expression levels between the species, as OmpC and OmpF of R. ornithinolytica share only 90% amino acid identity with OmpK36 and OmpK35 of K. pneumoniae, respectively. Alternatively, OmpC complementation may not have provided sufficient protein production to restore the phenotype. While we saw increased expression of OmpC and OmpF in complemented mutants compared to YDC775-9, it is possible that increased expression did not translate into full restoration of production of functional porins.

OXA-232 is still uncommon in the United States, and in hindsight, the only clue of its presence in the initial isolate was resistance of the initial isolate to piperacillin-tazobactam, despite susceptibility to cephalosporins and carabapenems. Of note, mCIM preformed retrospectively was positive in the initial ertapenem-susceptible isolate. It is possible that this knowledge could have allowed for prompt initiation of appropriate antimicrobial therapy and possibly have altered the clinical course. However, the initial isolate did not meet CLSI criteria for carbapenemase testing (imipenem or meropenem MIC, 2 to 4 μg/ml; ertapenem MIC, 2 μg/ml) since OXA-48-group carbapenemases are not common in the United States. Even using the EUCAST guidelines, isolate YDC775-2 barely misses the indication for carbapenemase testing with a meropenem MIC of 0.125 μg/ml, when the recommended cutoff value for screening is a meropenem MIC of >0.125 μg/ml. Early molecular diagnostics may allow for prompt identification of these resistance genes and guide appropriate selection of therapy as more novel β-lactam/β-lactamase inhibitor combinations are used for treatment of infection caused by carbapenem-resistant Enterobacteriaceae. This case serves as a great illustration of the need for clinicians to be familiar with the local epidemiology of resistance mechanisms and vigilant for uncommon resistance mechanisms.

In conclusion, we report rapid development of high-level carbapenem resistance in bla_OXA-232-carrying R. ornithinolytica that was initially susceptible due to increased bla_OXA-232 expression and ompC and ompF mutations during treatment with carbapenem. This suggests that ertapenem should be used with caution in the treatment of infections with bla_OXA232-carrying organisms despite the susceptibility profile, as resistance can emerge rapidly.

MATERIALS AND METHODS

Bacterial strains.

R. ornithinolytica isolates YDC775-2, YDC775-9, and YDC775-15 were isolated sequentially over the course of 18 months from a patient who was cared for at a hospital in western Pennsylvania. All isolates were obtained from the drainage of abdominal abscess. Initial identification and primary antimicrobial susceptibility testing (AST) were performed by the clinical microbiology laboratory using MALDI-TOF and MicroScan, respectively.

Antimicrobial susceptibility testing.

The MICs of cephalosporins (cefotaxime, ceftazidime, and cefepime) and carbapenems (ertapenem and meropenem) were determined using the broth microdilution procedure in 96-well plates as stipulated by the CLSI, except that selective markers (gentamicin or chloramphenicol) were included in the medium to prevent plasmid loss from transformants. E. coli ATCC 25922 was used as the quality control strain. The modified carbapenem inactivation method (mCIM) was performed per CLSI guidelines to screen for carbapenemase production (19).

Whole-genome sequencing.

Three R. ornithinolytica clinical isolates were subjected to whole-genome sequencing on a NextSeq 500 instrument (Illumina, San Diego, CA) at the Microbial Genomics Sequencing Center (MiGS) at the University of Pittsburgh. Sequencing libraries were prepared as described with minor modifications (20). The reads were assembled de novo using the CLC Genomics Workbench v11 (CLC Bio, Aarhus, Denmark). Additionally, YDC775-2 was sequenced on a PacBio RS II instrument (Pacific Biosciences, Menlo Park, CA, USA) at the Yale Center for Genome Analysis (YCGA). With RS II, a total of 145,262 subreads with an average read length of 13,039 bp were obtained. De novo assembly of the reads using the Hierarchical Genome Assembly Process (HGAP) v3.0 available in the SMRT Analysis v2.3 software generated two contigs, the chromosome and plasmid, with an average coverage of 268×. The contigs were then circularized and polished using the resequencing protocol in SMRT Analysis with two passes to reach the final consensus accuracy of 100%. Nucleotide variants were identified by mapping the NextSeq sequencing reads from each isolate to the annotated closed genome of YDC775-2. Variants at positions with a minimum coverage of 25× and occurring at frequencies higher than 95% were included. Pairwise single-nucleotide polymorphism (SNP) differences between the genomes were identified using Snippy (https://github.com/tseemann/snippy). Mutations in the ompC and ompF genes were confirmed with Sanger sequencing using gene-specific primers. Antimicrobial resistance genes were identified using ResFinder (https://cge.cbs.dtu.dk/services/ResFinder/). For the gene content analysis, contigs were compared directly using BLAST searches. A gene was considered present in an isolate if hits with greater than 80% identity covered at least 80% of the gene. The reads were aligned to genes that were present in other isolates but missing in the isolates of interest using Short Read Sequence Typing for Bacterial Pathogens (SRST2) v2 and were not considered missing if there were at least 10 reads aligned that covered 80% of the gene.

The draft genome assemblies of isolates YDC775-2, YDC775-9, and YDC775-15 generated by NextSeq sequencing have been deposited in the NCBI database under accession numbers RJTT00000000, RJTR00000000, and RJTO00000000, respectively. The Sequence Read Archive (SRA) accession numbers for the isolates are SRR8159600, SRR8159602, and SRR8159597. The complete sequences of the YDC775-2 chromosome and plasmid generated by RSII can be found under GenBank accession numbers CP033683 and CP033682.

Quantitative reverse transcription-PCR (qRT-PCR).

Quantitative reverse transcription PCR was performed using the Power SYBR green RNA-to-CT 1-step kit (Thermo Fisher). Twenty microliter reactions were performed in a 96-well format, and reaction mixtures contained 200 nM final concentrations of either rpoB, bla_OXA-232, ompC, or ompF primers (Table S1) and the reaction mix diluted to 1×. RNA was then added to a final amount of 50 ng per reaction. The cycle conditions were as follows: 1 cycle at 50°C for 30 min, 1 cycle at 95°C for 10 min, and 40 cycles at 95°C for 15 s and 60°C for 1 min. Melting curves were generated by 80 cycles at 60°C for 15 s with 0.5°C increments. The results were analyzed using the comparative threshold cycle (C_T) method (2^–ΔΔCT), where relative amounts of RNA transcript were normalized to that of the rpoB gene (21). Changes in fluorescence were monitored using an Applied Biosystems 7500 real-time system (Thermo Fisher). Controls without reverse transcriptase were included in each run. qRT-PCRs were performed in triplicate at least three times with comparable results.

Copy number analysis of bla_OXA-232.

The copy number of bla_OXA-232 was quantified with real-time PCR using a PowerUP SYBR Green Master Mix qPCR kit (Thermo Fisher Scientific). Twenty microliter reactions were performed in a 96-well format, and reactions contained 50 ng of DNA. A negative-control reaction mixture lacking DNA template was performed. The quantitative PCR was processed in an Applied Biosystems 7500 real-time system (Thermo Fisher Scientific) as follows: one cycle at 50°C for 2 min, then one cycle at 95°C for 2 min, and then 40 repeated cycles of 15 s denaturation at 95°C and 1 min annealing and extension at 60°C. Melting curves were generated by 80 cycles at 50°C for 10 s with 0.5°C increments. qPCRs were performed in triplicate at least three times with comparable results. The results were calculated using the ΔΔC_T method, where relative amounts of bla_OXA-232 were compared to amounts of the rpoB gene as an internal standard.

Generation of R. ornithinolytica YDC775-9 transformants.

The genes ompC and ompF from R. ornithinolytica YDC775-2 (carbapenem susceptible) were amplified from bacterial DNA using the PCR primers listed in Table S1. Amplicons were digested with XbaI and HindIII and ligated into shuttle vectors pMQ132 (ompC) and pMQ450 (ompF). The inserts were sequenced on both strands. R. ornithinolytica YDC775-9 was transformed using electroporation, and the transformants were selected on lysogenic broth (LB) agar containing 10 μg/ml gentamicin for pMQ132 (ompC), and 30 μg/ml chloramphenicol for pMQ450 (ompF). Empty vector controls were used for the transformation experiments. Expression of the complemented ompC and ompF genes in transformants was confirmed using qRT-PCR (data not shown).