Phenotypic and Genotypic Characterization of Enterobacteriaceae Producing Oxacillinase-48–Like Carbapenemases, United States

Oxacillinase (OXA)–48–like carbapenemases remain relatively uncommon in the United States. We performed phenotypic and genotypic characterization of 30 Enterobacteriaceae producing OXA-48–like carbapenemases that were recovered from patients during 2010–2014. Isolates were collected from 12 states and not associated with outbreaks, although we could not exclude limited local transmission. The alleles β-lactamase OXA-181 (blaOXA-181) (43%), blaOXA-232 (33%), and blaOXA-48 (23%) were found. All isolates were resistant to ertapenem and showed positive results for the ertapenem and meropenem modified Hodge test and the modified carbapenem inactivation method; 73% showed a positive result for the Carba Nordmann–Poirel test. Whole-genome sequencing identified extended-spectrum β-lactamase genes in 93% of isolates. In all blaOXA-232 isolates, the gene was on a ColKP3 plasmid. A total of 12 of 13 isolates harboring blaOXA-181 contained the insertion sequence ΔISEcp1. In all isolates with blaOXA-48, the gene was located on a TN1999 transposon; these isolates also carried IncL/M plasmids.

Oxacillinase (OXA)-48-like carbapenemases remain relatively uncommon in the United States. We performed phenotypic and genotypic characterization of 30 Enterobacteriaceae producing OXA-48-like carbapenemases that were recovered from patients during 2010-2014. Isolates were collected from 12 states and not associated with outbreaks, although we could not exclude limited local transmission. The alleles β-lactamase OXA-181 (bla OXA-181 ) (43%), bla OXA-232 (33%), and bla OXA-48 (23%) were found. All isolates were resistant to ertapenem and showed positive results for the ertapenem and meropenem modified Hodge test and the modified carbapenem inactivation method; 73% showed a positive result for the Carba Nordmann-Poirel test. Whole-genome sequencing identified extended-spectrum β-lactamase genes in 93% of isolates. In all bla OXA-232 isolates, the gene was on a ColKP3 plasmid. A total of 12 of 13 isolates harboring bla OXA-181 contained the insertion sequence ΔISEcp1. In all isolates with bla OXA-48 , the gene was located on a TN1999 transposon; these isolates also carried IncL/M plasmids.
T he prevalence of carbapenem-resistant Enterobacteriaceae (CRE) has been increasing in the United States since 2000 (1,2). This finding is problematic because treatment options for CRE infection are limited, and these infections are associated with a higher mortality rate than are infections with carbapenem-susceptible Enterobacteriaceae (3). Enterobacteriaceae might be resistant to carbapenems by a variety of mechanisms, the most concerning of which is production of carbapenemases (4). Although the Klebsiella pneumoniae carbapenemase is the most common carbapenemase reported in the United States, there have been reports of several other carbapenemases including the metallo-β-lactamases and, more recently, oxacillinase (OXA)-48-like carbapenemases (1,(5)(6)(7)(8)(9)(10).
The first description of isolates with β-lactamase OXA-48-like (bla OXA-48 -like) genes in the United States was from a surveillance study in 2013, which incidentally reported 2 K. pneumoniae isolates (6). This description was followed shortly afterward by a report of 2 clinical K. pneumoniae isolates with bla OXA-48 -like genes in patients from 1 institution in Virginia who had traveled internationally (7). More recently, CRE with bla OXA-232 genes have been isolated in the United States (8). The Centers for Disease Control and Prevention (CDC) has collected multiple isolates harboring bla OXA-48 -like genes from patients in the United States (19). We report the genotypic and phenotypic characterization of those isolates.

Collection of Isolates
Isolates are submitted to CDC for antimicrobial susceptibility testing (AST) for many reasons, including outbreak response, AST confirmation, and surveillance studies. Surveillance studies include the Multi-Site Gram-Negative Surveillance Initiative, which is part of the Emerging Infections Program, and the Sentinel Study (5,20). All Enterobacteriaceae isolates received for AST at CDC during June 1, 2010-October 31, 2012, with reduced susceptibility to carbapenems (MIC >1 μg/mL for any carbapenem), a positive modified Hodge test result, and a PCR-negative result for bla K. pneumoniae carbapenemase were retrospectively screened for bla OXA-48 -like genes (n = 115). During November 1, 2012-September 30, 2014, all Enterobacteriaceae received at CDC were routinely tested for bla OXA-48 -like genes by real-time PCR (n = 1,399). Submitting institutions were characterized by state and US Department of Health and Human Services (HHS) region (https://www.hhs.gov/ ash/about-ash/regional-offices/index.html).
We randomly selected K. pneumoniae isolates 1, 11, and 23, encoding bla OXA-181 , bla OXA-232 , and bla OXA-48 , respectively, as internal reference strains and sequenced them by using Single Molecule Real-Time Technology (Pacific Biosciences, Menlo Park, CA, USA) in addition to Illumina sequencing ( Table 2). We extracted and purified gDNA by using the MasterPure Complete DNA and RNA Kit (Epicenter, Madison, WI, USA), according to the manufacturer's recommended protocol. We generated 10-kb libraries by using the SMRTbell Template Prep Kit 1.0 (Pacific Biosciences) and sequenced libraries by using C4v2 Chemistry on the RSII Instrument (Pacific Biosciences). We assembled data by using Hierarchical Genome-Assembly Process Oxacillinase-48-Like Carbapenemases, United States version 3.0 (Pacific Biosciences) and generated clean consensus sequences by using Quiver (28). We deposited all raw sequencing reads, Pacific Biosciences assemblies, and MIC results in GenBank under BioProject PRJNA296771. We determined multilocus sequence types for each specimen by mapping clean Illumina reads to allele sequences (http://www.pubmlst.org) by using SRST2 software (Illumina) (29). We described antimicrobial resistance genotype profiles from assembled Illumina and Pacific Biosciences contigs by using SSTAR V1.0 (30) in combination with the ARG-ANNOT (31) and ResFinder (32) repositories.
We used the PlasmidFinder database (http://www. genomicepidemiology.org/) to detect plasmid replicon sequences among Illumina and Pacific Biosciences contigs to estimate the plasmid composition of each isolate (33). In addition, we predicted insertion sequences that might be associated with spread of antimicrobial resistance genes by using ISfinder (34). For isolates with bla OXA-48 , we estimated the copy number of IS1R insertion sequences for determining Tn1999 variants by using blastn (https://blast. ncbi.nlm.nih.gov/Blast.cgi) and SPAdes K-mer coverage output (26,(34)(35)(36). The clonality of our plasmids was also assessed, as was the location of bla OXA-48 -like genes (online Technical Appendix, https://wwwnc.cdc.gov/EID/ article/24/4/17-1377-Techapp1.pdf). Because of a cluster of isolates from 1 state in this study, a phylogenetic tree and single-nucleotide polymorphism (SNP) tree matrix were produced by using RAxML version 8 (37) (online Technical Appendix).

Transformation Experiments
We randomly transformed 10 selected isolates (3 with bla OXA-48 , 4 with bla OXA-181 , and 3 with bla OXA-232 ) for transformation experiments to better characterize plasmids harboring bla OXA-48 -like genes. We subcultured parent isolates on trypticase soy agar containing 5% sheep blood, placed them in 50 mL of tryptic soy broth containing ertapenem (1 μg/mL), and incubated them overnight at 35°C. We extracted plasmid DNA by using Plasmid Midi Kits (QIAGEN, Valencia, CA, USA), according to the manufacturer's protocol. We digested intact plasmid DNA and gDNA with HindIII (New England Biolabs, Ipswich, MA, USA) and separated this DNA by electrophoresis on a 0.9% agarose gel.
We transformed 500 ng of plasmid DNA from each isolate into E. coli DH10B cells (Invitrogen, Carlsbad, CA, USA) by electroporation and incubated at 35°C for 2 h. Potential transformants were plated on Luria-Bertani agar containing ertapenem (1 μg/mL) and incubated overnight at 35°C. Four colonies from each transformant plate were screened for bla OXA-48 -like genes by using PCR. Transformant plasmid DNA was digested and separated by gel electrophoresis along with digested parent plasmid DNA to ensure that transformant plasmids were also present in parental cells.
When we compared transformants with parent strains, most of which harbored multiple plasmids and numerous resistance genes, transformants were confirmed to carry only 1 plasmid and typically showed greater susceptibility to extended-spectrum cephalosporins but retained resistance to >1 carbapenem. As confirmed by WGS, we found that ESBL genes were not typically present on the same plasmid as bla OXA-48 -like genes; only 1 transformant (23T) carried a plasmid harboring bla CTX-M-14b on the IncL/M plasmid carrying bla OXA-48 . Similar to the parent strain, strain 23T showed increased MICs to cephalosporins and carbapenems, although the carbapenem MICs were lower than both the parent strain and other transformants carrying only an OXA-48-like carbapenemase ( Table 4). None of the plasmids harboring bla OXA-48 -like genes encoded additional carbapenemases.
We identified no SNPs when we compared Illumina and Pacific Biosciences genome sequences for the same isolate for isolates 1, 11, and 23. This finding indicates that Pacific Biosciences sequences can be used as a mapping reference. We compared Illumina sequence data for the remaining clinical isolates, which were not subjected to Pacific Biosciences sequencing, against the Pacific Biosciences genomes according to bla OXA-48 -like allele. For all 10 isolates containing bla OXA-232 , the gene was co-located with the ColKP3 replicon gene and a ΔISEcp1 upstream insertion sequence (upstream of bla OXA-232 ) on an ≈6 kb contig. Pacific Biosciences sequence analysis of isolate 23 confirmed the presence of bla OXA-48 on transposon Tn1999.2; bla OXA-48 was found on a variant of transposon Tn1999 in all instances. In 3 isolates (23, 28, and 29), coverage of the IS1R insertion sequence was similar to the overall assembly coverage suggestive of the Tn1999.2 variant identified in isolate 23 by Pacific Biosciences sequencing. However, in 4 isolates (14, 21, 25, and 30), coverage of the IS1R insertion sequence was much higher than the overall assembly coverage, indicating multiple occurrences of this locus, suggestive of a different Tn1999 variant. Of the 13 isolates containing bla OXA-181 , 12 had an upstream insertion sequence ΔISEcp1. In isolate 1, which was sequenced by using Pacific Biosciences technology, bla OXA-181 was confirmed as being chromosomally located. Finally, given the geographic association of several isolates carrying bla OXA-181 , we created a phylogenetic tree and SNP matrix table for the 7 K. pneumoniae isolates from 1 state in HHS region 9 (Table 5; Figure 3).

Discussion
The increasing prevalence of CRE in the United States poses a challenge to patients, clinicians, and public health. The diversity of carbapenemases, including the OXA-48like enzymes reported in this study, is an ongoing diagnostic challenge to clinical microbiology laboratories because of the variety of phenotypes displayed by isolates producing different, and sometimes multiple, carbapenemases. OXA-48 has been described as the phantom menace because of its subtle phenotype in the absence of co-resistance mechanisms (12).
In this study, all isolates with bla OXA-48 -like genes showed resistance to ertapenem, and most showed intermediate resistance or resistance to meropenem, ceftriaxone, ceftazidime, and cefepime. Three tests for carbapenemase production were performed on the isolates in this study. The modified Hodge test, performed for ertapenem or meropenem, and the mCIM showed positive results for all isolates with bla OXA-48-like genes. The Carba Nordmann-Poirel test showed positive results for 73% of all isolates, which is consistent with other studies that have shown that this test had a sensitivity of 72%-76% for OXA-48-like carbapenemase producers (45,46). All isolates in this study would be identified as CRE by the current CDC and Council of State and Territorial Epidemiologists definitions (https://www.cdc.gov/hai/ organisms/cre/definition.html) (47).
The 10 isolates that harbored bla OXA-232 were all found on a small ColKP3 plasmid, and this association has been reported by Potron et al. (43). Likewise, the 7 isolates producing OXA-48 carried bla OXA-48 on a similar genetic environment to those reported (44,48,49). Isolate 23, which was sequenced by using Illumina and Pacific Biosciences technology, harbored bla OXA-48 on an IncL/M plasmid. The other 6 isolates, which were sequenced only by using Illumina technology, all had the IncL/M replicon gene. In addition, bla OXA-48 was always associated with a variant of transposon TN1999, as discerned on the basis of the copy number of IS1R insertion sequences (36). Because these IS1R sequences are identical and duplicated, Illumina technology often fails to assemble these as separate loci but instead produces a single locus with high coverage. Comparing coverage of the IS1R insertion sequence to the overall coverage of the assembly sequence enabled us to estimate the presence of the TN1999 variant by using isolate 23 as the reference. In 12 of 13 isolates with bla OXA-181 , we found an upstream ΔISEcp1 element inserted upstream of the bla OXA-181 cassette. bla OXA-181 is often associated with ISEcp1, which might facilitate its spread (50).
The transformation experiment helped to clarify our understanding of the plasmids harboring bla OXA-48 -like genes. Transformation experiments were successful for each of the parent strains carrying bla OXA-48 or bla OXA-232 . Carbapenem and penicillin MICs were not different between the parent and transformant, but transformant MICs were comparatively lower for cephalosporins and aminoglycosides. This finding supports the genotypic data, which indicated that ESBL genes and other β-lactamase genes did not cotransfer with the plasmid encoding bla OXA-48 -like genes. One transformant (23T) did not have decreased cephalosporin MICs when compared with its parental strain, which is consistent with Pacific Biosciences sequencing of this isolate, which showed bla CTX-M-14b to be on the same IncL/M plasmid as bla OXA-48 . The unsuccessful transformation attempts of bla OXA-181 -containing strains 1, 2, 26, and 27 was explained by WGS evidence that bla OXA-181 was chromosomally located in isolate 1.
This study had several limitations. The collection of isolates in this study might not be representative of all isolates with bla OXA-48 -like genes in the United States. There is also a reporting bias because only isolates sent to CDC were included. CDC receives isolates as part of outbreak investigations, surveillance studies, and to confirm AST results, but there is no national requirement to submit carbapenemase-producing isolates. Thus, unusually resistant isolates are more likely to be sent to the CDC and included in this study. Also, no prevalence rates of Enterobacteriaceae with bla OXA-48 -like genes in the United States can be inferred because there is not an evaluable denominator. In addition, almost all the isolates we studied were clinical isolates; colonizing isolates might have different phenotypic characteristics.
Another limitation is that the 10 isolates selected for the transformation experiment and the 3 isolates selected for Pacific Biosciences sequencing might not have been representative of the other isolates in this collection. Ideally, all isolates would have been sequenced by using Pacific Biosciences technology and been a part of the transformation experiment, but this testing was not performed because of limited resources. In addition, the decisions regarding which isolates to select for transformation experiments and sequencing by using Pacific Biosciences technology were made before WGS was complete. In retrospect, it would have been better to select bla  isolates that were hypothesized to be on a plasmid for the transformation experiment; instead, chromosomal bla OXA-181 isolates were selected. Thus, the bla OXA-181 gene loci for the isolates in this study are inconclusive.
In summary, the continued increase of CRE in the United States is a major problem, and the increasing prevalence of OXA-48-like carbapenemases is also concerning. We found Enterobacteriaceae in the United States with bla OXA-48 -like genes on similar mobile genetic elements to those described elsewhere and that displayed relatively resistant AST profiles. The first step in continued detection of CRE producing these and other carbapenemases is identifying all carbapenem resistance among Enterobacteriaceae, including resistance to ertapenem. Future prospective investigations are needed to determine the true prevalence of OXA-48-like carbapenemases in the United States. I n 1965, the transferability of ampicillin resistance was reported, and the plasmid-encoded mechanism of resistance for 2 Salmonella sp. isolates from the United Kingdom and 1 Escherichia coli isolate from Greece was determined. Resistance (R) factors from Salmonella sp. isolates were designated R1818 and R7268 (R7268 encoding the current TEM-1). The E. coli isolate and its plasmid were named TEM (encoding the current TEM-2) because the isolate was recovered from a feces culture of an Athenian patient named Temoniera in 1963.