Molecular Epidemiology of KPC-Producing Escherichia coli: Occurrence of ST131-fimH30 Subclone Harboring pKpQIL-Like IncFIIk Plasmid

Abstract

Of 20 Klebsiella pneumoniae carbapenemase (KPC)-producing Escherichia coli isolates identified at hospitals in western Pennsylvania, 60% belonged to the epidemic ST131-fimH30 subclone. IncFIIk was the most common replicon type for the bla_KPC-carrying plasmids (n = 8). All IncFIIk plasmids possessed a scaffold similar to that of pKpQIL, and seven of them were borne by ST131-fimH30 isolates. IncN plasmids conferred resistance to trimethoprim-sulfamethoxazole, and IncA/C plasmids conferred resistance to gentamicin. Three bla_KPC-carrying plasmids (IncA/C and IncN) possessed bla_SHV-7/12 and qnrA1 or qnrS1.

TEXT

Carbapenem resistance in Enterobacteriaceae is now recognized as a major challenge facing hospitals and long-term-care facilities in many parts of the world, including the United States (1). The problem is most acute in Klebsiella spp., where carbapenem resistance reached approximately 12% among health care-associated isolates in U.S. hospitals in 2010 (2). While less prevalent than in Klebsiella pneumoniae, K. pneumoniae carbapenemase (KPC) production in other species of Enterobacteriaceae is increasingly reported, presumably as bla_KPC-carrying plasmids are acquired from K. pneumoniae by these non-Klebsiella species and then propagate (3). Of particular concern is the acquisition of bla_KPC by Escherichia coli, which is primarily a community pathogen and may potentially serve as the vehicle for the spread of KPC-mediated carbapenem resistance into community settings. Data on the molecular epidemiology of KPC-producing E. coli remain relatively scarce. We recently reported 13 cases of KPC-producing E. coli infection and colonization identified at hospitals in Pennsylvania (4). Building upon our initial report, we conducted further characterization of the microbiological features and molecular epidemiology of 20 unique KPC-producing E. coli isolates identified by the end of 2012.

E. coli clinical isolates with reduced susceptibility to ertapenem were collected from the clinical microbiology laboratory at the University of Pittsburgh Medical Center between 2008 and 2012. The isolates were included in the study if the presence of bla_KPC was confirmed by PCR (4). The 13 isolates reported in our previous case series were also included (4). Eleven isolates possessed bla_KPC-2, and the remaining 9 isolates carried bla_KPC-3. All isolates were nonsusceptible to ertapenem and variably resistant to the other carbapenems when interpreted by the updated Clinical and Laboratory Standards Institute (CLSI) breakpoints (Table 1) (5, 6). Most of the KPC-producing E. coli clinical isolates were nonsusceptible to piperacillin-tazobactam, cefotaxime, ceftazidime, and aztreonam, but 11 isolates were susceptible to cefepime. All isolates were susceptible to amikacin. All isolates were susceptible to colistin and tigecycline when interpreted by the European Committee on Antimicrobial Susceptibility Testing (EUCAST) breakpoints (version 4; http://www.eucast.org) (Table 1).

TABLE 1.

Antimicrobial susceptibility of the 20 KPC-producing E. coli isolates^a

Agent	No. (%) of isolates
	Susceptible	Intermediate	Resistant
Ertapenem	0 (0)	2 (10)	18 (90)
Imipenem	8 (40)	7 (35)	5 (25)
Meropenem	9 (45)	2 (10)	9 (45)
Doripenem	13 (65)	6 (30)	1 (5)
Piperacillin-tazobactam	3 (15)	10 (50)	7 (35)
Cefotaxime	1 (5)	3 (15)	16 (80)
Ceftazidime	6 (30)	4 (20)	10 (50)
Cefepime	11 (55)	8 (40)b	1 (5)
Aztreonam	0 (0)	2 (10)	18 (90)
Ciprofloxacin	6 (30)	0 (0)	14 (70)
Sulfamethoxazole-trimethoprim	3 (15)	0 (0)	17 (85)
Gentamicin	13 (65)	0 (0)	7 (35)
Tobramycin	13 (65)	2 (10)	5 (25)
Amikacin	20 (100)	0 (0)	0 (0)
Doxycycline	10 (5)	7 (35)	3 (15)
Minocycline	19 (95)	1 (5)	0 (0)
Tigecycline	20 (100)	0 (0)	0 (0)
Colistin	20 (100)	0 (0)	0 (0)

^aCarbapenem MICs were determined by the agar dilution method. The other MICs were determined by the broth microdilution method.

^bSusceptible dose dependent.

Multilocus sequence typing was performed on the clinical isolates using the Achtman protocol (7). Twelve of the 20 isolates (60%) belonged to sequence type 131 (ST131). Two isolates belonged to ST648, an ST which has been found widely among extended-spectrum β-lactamase (ESBL)-producing isolates from humans as well as wild birds, livestock, and companion animals worldwide (8). The other STs were from single isolates (Table 2). Further subtyping of the 12 ST131 isolates was conducted by pulsed-field gel electrophoresis (PFGE) as well as fimH typing and gyrA or parC typing (4, 9, 10). PFGE revealed the presence of two clusters in a dendrogram generated by Bionumerics software (Applied Maths, Austin, TX) (see Fig. S1 in the supplemental material). The first cluster included 7 isolates identified between 2008 and 2011, and the second cluster included 5 isolates identified after 2011. By applying the interpretive criteria of Tenover et al. (11), all 7 isolates in the first cluster were considered closely or possibly related to each other. In the second cluster, 3 of the 5 isolates were closely or possibly related to each other. Nonetheless, all 12 ST131 isolates had the fimH30 allele upon fimH typing (10), and sequences of the quinolone resistance-determining regions (QRDRs) of gyrA and parC (9) showed that all had the 1AB allele for gyrA and 1aAB allele for parC, which has been shown to be the principal allele combination among E. coli ST131-fimH30 isolates in the United States (9). KPC-producing E. coli ST131 isolates have been reported on several occasions (12, 13). Our findings suggest that the clone, in particular the fimH30 subclone, may be a major vehicle of bla_KPC-carrying plasmids in E. coli but replacement or evolution of the strains within this subclone may have occurred over time.

TABLE 2.

Characteristics of bla_KPC-carrying plasmids from E. coli^a

Characteristic	IncFIIk (n = 8)	IncN (n = 4)	ColE (n = 2)	IncFIA (n = 1)	IncFIA-FIB (n = 1)	IncA/C (n = 1)	IncA/C-N (n = 1)	Unknown (n = 2)
Estimated size (kb)	100–130	50–70	<50	70	120	170	200	<50–60
KPC allele	KPC-2	KPC-2 (1), KPC-3 (3)	KPC-3	KPC-2	KPC-3	KPC-3	KPC-3	KPC-2 (1), KPC-3 (1)
Other ESBLs		SHV-12 (1)				SHV-7	SHV-7
Drug coresistance		SXT (4), NAL (1)			TET	GEN, NAL	GEN, SXT, NAL
ST of the host E. coli	131 (7), 404 (1)	131 (1), 648 (2), 964 (1)	131 (1), 372 (1)	131	131	3866	2521	131 (1), 536-like (1)

^aValues in parentheses indicate the number of isolates. SXT, trimethoprim-sulfamethoxazole; NAL, nalidixic acid; TET, tetracycline; GEN, gentamicin.

PCR was performed to detect the presence of additional extended-spectrum β-lactamase genes, including bla_CTX-M, bla_SHV, and the plasmid-mediated AmpC β-lactamase gene bla_CMY (4). Four isolates possessed bla_SHV (3 bla_SHV-7 and 1 bla_SHV-12, both encoding ESBL), 2 isolates possessed bla_CTX-M-14, and 1 isolate had bla_CMY-44 (14).

E. coli DH10 transformants harboring bla_KPC-carrying plasmids were generated as described previously (4). The sizes of the plasmids were estimated using the S1 nuclease PFGE method (15). Replicon typing of the plasmids was conducted as described by Carattoli et al. (16), which was supplemented by additional pairs of primers (17). IncFII plasmids were further identified as IncFIIk by the method of Villa et al. (18). As a result, 8 plasmids were assigned to IncFIIk, followed by IncN (4 plasmids), ColE (2 plasmids), IncFIA (1 plasmid), IncFIA-FIB (1 plasmid), IncA/C (1 plasmid), and IncA/C-N (1 plasmid) (Table 2). ST131-fimH30 isolates were associated with plasmids of various incompatibility groups (FIIk, FIA, FIA-FIB, N), with the ST131-fimH30-IncFIIk combination observed in 7 of the 12 isolates.

Antimicrobial susceptibility of the E. coli transformants was determined for ertapenem, imipenem, ceftotaxime, ceftazidime, gentamicin, trimethoprim-sulfamethoxazole, nalidixic acid, tetracycline, and fosfomycin by use of the standard Kirby-Bauer disk method (BBL Sensi-Disc, BD, Sparks, MD) (6, 19). Coresistance to trimethoprim-sulfamethoxazole was conferred by 5 plasmids (4 IncN and 1 IncA/C-N), whereas that to gentamicin was conferred by 2 plasmids (IncA/C-N and IncA/C) (Table 2). Tetracycline resistance was conferred by only 1 plasmid (IncFIA-FIB). Three bla_KPC-carrying plasmids conferring reduced susceptibility to nalidixic acid carried qnrA1 or qnrS1. These three plasmids also carried bla_SHV (bla_SHV-7 and qnrA1 on the IncA/C-N and IncA/C plasmids and bla_SHV-12 and qnrS1 on the IncN plasmid). Location of the bla_SHV genes on the same plasmids with bla_KPC was confirmed by DNA hybridization experiments (data not shown).

IncFIIk was the most predominant incompatibility group observed among the bla_KPC-carrying plasmids in our series, and their sizes ranged approximately between 100 and 130 kb. Restriction digestion analysis revealed highly similar banding patterns for the 8 IncFIIk plasmids, using either HindIII or PstI as the restriction enzyme (Fig. 1). With HindIII, all plasmids appeared identical except for the plasmid from isolate 16, which was missing a single band. This band appeared to coincide with a 3,567-bp section in pKpQIL downstream of Tn4401 carrying bla_KPC (nucleotides 18028 to 21594 in the sequence of GenBank accession no. GU595196). With PstI, the plasmid from isolate 16 lacked a single band which likely corresponded to an 8,834-bp section in pKpQIL (nucleotides 29342 to 38175), including the mercury resistance locus downstream of bla_TEM. These differences, as well as more minor ones which are not fully resolved by restriction digestion, likely accounted for the variation in the estimated plasmid sizes. Nonetheless, the overall banding patterns showed that the 8 IncFIIk plasmids shared a highly similar scaffold.

FIG 1.

Restriction digestion profile of IncFIIk plasmids carrying bla_KPC in E. coli transformants. Lanes: M, markers; 1, isolate 3; 2, isolate 6; 3, isolate 7; 4, isolate 10; 5, isolate 12; 6, isolate 15; 7, isolate 16; 8, isolate 18. Lanes 1 to 8, HindIII digestion; lanes 9 to 16, PstI digestion. The plasmid from isolate 7 (lanes 3 and 11) is nearly identical to pKpQIL based on high-throughput sequencing.

The IncFIIk plasmid from isolate 7 was subjected to high-throughput sequencing on an Ion PGM instrument (Life Technologies, Carlsbad, CA). The reads were assembled using the NextGENe de novo assembler (SoftGenetics, State College, PA). After an initial screening for the closest match, the contigs were aligned with pKpQIL (GenBank accession no. GU595196) (20) using Mauve (21). As a result, both the IncFIIk and IncFIBk replicons were present in this plasmid as in pKpQIL and its related plasmids (22). The comparison revealed that 96.8% of the contigs from the isolate 7 plasmid aligned with pKpQIL, with 9 single-nucleotide polymorphisms observed in the aligned regions. Six of them were nonsynonymous, and the rest were in intergenic regions. One of the nonsynonymous polymorphisms corresponded to amino acid position 272 of KPC, which differentiated KPC-2 and KPC-3; pKpQIL had histidine defining KPC-3, and the IncFIIk plasmid from isolate 7 had tyrosine defining KPC-2. pKpQIL was characterized as a plasmid playing a role among multiple STs, including ST258, during the national outbreak of KPC-producing K. pneumoniae in Israel (23, 24). It was also one of the first bla_KPC-encoding plasmids to be fully sequenced and characterized (20). pKpQIL-like plasmids have since been shown to be widespread (22, 25). In our study, 7 of the pKpQIL-like IncFIIk plasmids were harbored by ST131-fimH30 isolates collected between 2009 and 2012, making this the most common subclone-plasmid combination that has persisted over time and by inference perhaps one with high potential to spread further among the STs and plasmids identified so far in this species.

KPC-producing E. coli remains relatively rare in comparison with carbapenemase-producing K. pneumoniae. However, the spread of bla_KPC from K. pneumoniae to E. coli, which by nature is a community pathogen, is a potential public threat which requires close monitoring. Polyclonal, plasmid-mediated spread of KPC-producing E. coli has been reported previously (26, 27). Our study suggests that E. coli ST131-fimH30 and the pKpQIL-like IncFIIk plasmid may represent a high-risk subclone-plasmid combination for the future spread of bla_KPC in E. coli.