A Novel Deletion Mutation in pmrB Contributes to Concurrent Colistin Resistance in Carbapenem-Resistant Escherichia coli Sequence Type 405 of Clinical Origin

We report the first clinical Escherichia coli strain EC3000 with concomitant chromosomal colistin and carbapenem resistance. A novel in-frame deletion, Δ6-11 (RPISLR), in pmrB that contributes to colistin resistance was verified using recombinant DNA techniques. Although being less fit than the wild-type (WT) strain or EC3000 revertant (chromosomal replacement of WT pmrB in EC3000), a portion of serially passaged EC3000 strains preserving colistin resistance without selective pressure raises the concern for further spread.

ABSTRACT

We report the first clinical Escherichia coli strain EC3000 with concomitant chromosomal colistin and carbapenem resistance. A novel in-frame deletion, Δ6-11 (RPISLR), in pmrB that contributes to colistin resistance was verified using recombinant DNA techniques. Although being less fit than the wild-type (WT) strain or EC3000 revertant (chromosomal replacement of WT pmrB in EC3000), a portion of serially passaged EC3000 strains preserving colistin resistance without selective pressure raises the concern for further spread.

TEXT

The main colistin resistance mechanism in Gram-negative bacteria is modification of lipopolysaccharide (LPS), which reduces the electrostatic affinity between LPS and positively charged colistin (1, 2). Because of the extensive use of colistin for treatment of carbapenem-resistant Enterobacteriaceae (CRE) infections, global reports of colistin resistance in Enterobacteriaceae continue to emerge (3). A more worrisome phenomenon is the occurrence of colistin resistance among CREs (3–8). These strains further complicate the already-challenging treatment options for CREs and are associated with high mortality in infected patients (9). In our previous study, one carbapenem-resistant Escherichia coli strain, designated EC3000, belonged to sequence type (ST) 405 and was found to have concomitant colistin resistance (10). Current reports reveal that Klebsiella pneumoniae accounts for the majority of strains with this resistance phenotype. Another important species of Enterobacteriaceae, E. coli, is less frequently reported (8, 11). Therefore, the aim of this work was to characterize the colistin resistance mechanism, overall fitness cost, and colistin resistance stability of strain EC3000.

The E. coli strains and plasmids used in this study are listed in Table 1. All experiments were repeated on three separate occasions and presented as means ± SD. The EC3000 isolate was from a wound culture from a 41-year-old man after colistin use. The carbapenem resistance mechanism of this isolate was the CMY-2 β-lactamase in combination with OmpC porin loss. The MICs against meropenem and imipenem were 4 and 8 μg/ml, respectively (10). According to CLSI, colistin resistance of EC3000 was confirmed by broth microdilution, which yielded an MIC of 8 mg/liter. Moreover, the EC3000 strain showed resistance to most of the antibiotics tested with the Vitek 2 system (Table 2). Only amikacin, gentamicin, tigecycline, and trimethoprim-sulfamethoxazole retained in vitro activity, with the breakpoints of tigecycline and colistin interpreted according to EUCAST. In the literature, such highly drug-resistant phenotypes have been reported exclusively in K. pneumoniae isolates and have been associated with high mortality (9). The case we present, along with previously reported cases, revealed an alarming phenomenon in which high levels of drug resistance have spread to other species of the Enterobacteriaceae family (8, 11).

TABLE 1.

Strains and plasmids used in this study

Strain or plasmid	Relevant characteristica	Source/referenceb
Strain
EC3000	Clinical isolate with colistin resistance; in-frame deletion in pmrB (△6-11 [RPISLR])	NIHH
S17-1λpir	hsdR recA pro RP4-2 (Tc::Mu; Km::Tn7) (λpir)	(16)
MG1655	WT	ATCC
ATCC 25922	WT	ATCC
ECS01	Clinical isolate susceptible to colistin	This study
ECS02	Clinical isolate susceptible to colistin	This study
ECS03	Clinical isolate susceptible to colistin	This study
ECS04	Clinical isolate susceptible to colistin	This study
ECS05	Clinical isolate susceptible to colistin	This study
ECS06	Clinical isolate susceptible to colistin	This study
ECS07	Clinical isolate susceptible to colistin	This study
ECS08	Clinical isolate susceptible to colistin	This study
C3000_ΔpmrBΔ6-11	1,088-bp deletion in pmrBΔ6-11 locus of EC3000	This study
EC3000_ΔpmrBΔ6-11 (pCRII-TOPOpmrBΔ6-11)	EC3000_ΔpmrBΔ6-11 complemented with pCRII-TOPOpmrBΔ6-11	This study
EC3000 revertant	EC3000 background; WT PmrB replacement strain of EC3000	This study
Plasmid
pCRII-TOPO	TOPO cloning vector; Ampr Kanr	Invitrogen
pUT-KB	Suicide vector; Kanr	(15)
pUT-KB-KOpmrB	1,919-bp fragment containing 1,088-bp deletion in PmrB from E. coli EC3000 cloned into pUT-KB	This study
pUT-KB-pmrB	1,919-bp fragment containing entire pmrB gene from E. coli MG1655 cloned into pUT-KB	This study

^aAmp^r, resistance to ampicillin; Kan^r, resistance to kanamycin.

^bNIHH, National Health Research Institutes; ATCC, American Type Culture Collection.

TABLE 2.

MICs of antibiotics against E. coli strain EC3000 and colistin-susceptible E. coli strains ECS01 to ECS08

Strain	MIC (μg/ml) ofa:
	AMP	TZP	CRO	CAZ	FEP	IPM	GEM	AMK	LVX	CIP	TGC	SXTb	CST
EC3000	>32	>128	>64	>64	16	8	<1	<2	>8	>4	<0.5	<20	8
ECS01	>32	>128	<1	<1	<1	<0.25	>16	<2	>8	>4	<0.5	<20	<0.5
ECS02	4	<4	<1	<1	<1	<0.25	<1	<2	<0.12	<0.25	<0.5	<20	<0.5
ECS03	>32	64	>64	>64	>64	<0.2	>16	4	>8	>4	<0.5	<20	<0.5
ECS04	>32	<4	<1	4	<1	<0.25	<1	<2	4	1	<0.5	>320	<0.5
ECS05	>32	4	<1	<1	<1	<0.25	<1	<2	0.5	<0.25	<0.25	<20	<0.5
ECS06	>32	<4	<1	<1	<1	<0.25	<1	<2	<0.12	<0.25	<0.5	<20	<0.5
ECS07	>32	16	<1	<1	<1	<0.25	>16	4	>8	>4	<0.5	>320	<0.5
ECS08	>32	<4	<1	<1	<1	<0.25	<1	<2	>8	>4	<0.5	<20	<0.5
ATCC 25922c	8	<4	<1	<1	<1	<0.25	<1	4	<0.12	<0.25	<0.5	<20	<0.5

^aAMP, ampicillin; TZP, piperacillin-tazobactam; CRO, ceftriaxone; CAZ, ceftazidime; FEP, cefepime; IPM, imipenem-cilastatin; GEM, gentamicin; AMK, amikacin; LVX, levofloxacin; CIP, ciprofloxacin; TGC, tigecycline; SXT, trimethoprim-sulfamethoxazole; CST, colistin.

^bVITEK 2 SXT MICs are reported as sum of trimethoprim and sulfamethoxazole MICs, which are present in ratio of 1:19. SXT resistance breakpoint in this system is >80 μg/ml.

^cQuality control strain of antibiotic susceptibility test.

EC3000 did not contain mcr-1 to mcr-9 genes, according to use of specific primer (12, 13). The pmrA, pmrB, phoP, phoQ, and mgrB genes, which were thought to be involved in chromosomally encoded colistin resistance, were sequenced using the previously described primers shown in Table S1 in the supplemental material (14). We then compared the sequences of the mgrB, phoP, phoQ, pmrA, and pmrB genes from EC3000 and eight collected colistin-susceptible clinical isolates (ECS01 to ECS08) with the wild-type (WT) strain MG1655. As shown in Table 3, the unique amino acid deletions of 6 amino acid residues in PmrB were found only in EC3000 and therefore may play a role in colistin resistance.

TABLE 3.

Amino acid alterations in MgrB, PhoP, PhoQ, PmrA, and PmrB in E. coli EC3000 and clinical E. coli isolates compared with E. coli MG1655

Strain	Amino acid alteration ina,b:
	MgrB	PhoP	PhoQ	PmrA	PmrB
EC3000		I44L		S29G, G144S	H2R, Δ6-11 (RPISLR), D283G
ECS01		I44L	R6H	S29G	H2R, D283G, Y315F
ECS02	V8A	I44L		S29G, T31S, I128N, G144S	H2R, E123D, D283G, V351I
ECS03	V8A	I44L		S29G, T31S, I128N, G144S	H2R, E123D, D283G, V351I
ECS04		I44L	S138T	S29G	H2R, S138N, D283G
ECS05	V8A	I44L		S29G, T31S, I128N, G144S	H2R, E123D, D283G, V351I
ECS06	V8A	I44L	R6H	S29G, T31S, I128N, G144S	H2R, E123D, D283G, V351I
ECS07	V8A	I44L		S29G, T31S	H2R, E123D, D283G, V351I
ECS08	V8A	I44L		S29G, T31S, I128N, G144S	H2R, E123D, D283G, V351I

^aBoldface, amino acid substitutions found only in colistin-resistant E. coli isolates.

^bThe one-letter code for the amino acid designation is used.

To investigate further, strain EC3000_ΔpmrB_Δ6-11 (pmrB deletion mutant) and EC3000 revertant (chromosomal replacement of the WT pmrB in EC3000) derived from EC3000 were generated via homologous recombination as described by use of primers designed in this study and shown in Table S1 (15, 16). Strain EC3000_ΔpmrB_Δ6-11 (pCRII-TOPOpmrB_Δ6-11) was constructed after transformation with a pCRII-TOPO TA vector containing a mutated pmrB (Δ6-11) according to the manufacturer’s protocol (Invitrogen, Carlsbad, CA). Previous studies demonstrated that elevated expression of the pmrHFIJKLM operon increased modification of LPS and contributed to colistin resistance (1). Therefore, the expression of pmrK and pmrB in EC3000, MG1655, and EC3000-derived strains was assessed using reverse transcription-quantitative PCR (RT-qPCR) as described to confirm a contributing role of pmrB in colistin resistance (17). Primers used for RT-qPCR are listed in Table S1 (17, 18). As shown in Table 4, EC3000 had higher pmrB and pmrK expression than the WT strain. The EC3000 revertant restored colistin susceptibility, with an MIC of 0.25 μg/ml. Furthermore, pmrB and pmrK expression was reduced to nearly basal levels. Strain EC3000_ΔpmrB_Δ6-11 restored colistin susceptibility together with decreased pmrK expression, and the complemented strain EC3000_ΔpmrB_Δ6-11 (pCRII-TOPOpmrB_Δ6-11) was, again, resistant to colistin and showed upregulated pmrK and pmrB expression. Taken together, these data confirm that the novel deletion between positions 6 and 11 (RPISLR) in PmrB confers colistin resistance in EC3000. Our study results strengthen the hypothesis that changes in PmrB are important for chromosomally encoded colistin resistance in E. coli infection (18). The 6-amino-acid deletion in PmrB we presented could not be categorized into any function domains by SMART analysis (http://smart.embl.de/). A similar deletion mutation in PmrB has been reported in another colistin-resistant E. coli isolate (19). The exact mechanism of colistin resistance requires further investigation.

TABLE 4.

Colistin MIC and expression levels of pmrB and pmrK genes of different E. coli strains and transformants carrying pCRII-TOPOpmrB_Δ6-11

Strain	Chromosomal pmrB status	pCRII-TOPOpmrB status	Colistin MIC (mg/liter)	Relative expression levela
				pmrB	pmrK
MG1655	WT	None	0.25	1	1
EC3000	Δ6-11 (RPISLR)	None	8	15.48 ± 9.36	127.47 ± 35.14
EC3000_ΔpmrBΔ6-11	Deleted	pmrBΔ6-11 (RPISLR)	0.25	<0.01	3.25 ± 1.08
EC3000_ΔpmrBΔ6-11 (pCRII-TOPOpmrBΔ6-11)	Deleted	None	8	1.76 ± 0.14	10.18 ± 2.14
EC3000 revertant	WT	None	0.25	0.89 ± 0.43	2.21 ± 0.25

^aExpression levels (fold change) were normalized against the value obtained with MG1655. Values are means ± SD.

Fitness cost of EC3000 was then analyzed and compared to strain MG1655 and the EC3000 revertant. Growth curves of EC3000, MG165, and EC3000 revertant were determined using a Bioscreen C MBR (Oy Growth Curves Ab, Helsinki, Finland). EC3000 showed the lowest growth rates in the stationary phase (Fig. 1). Competitive coculture was performed, and the competition index (CI), defined as EC3000/MG1655 and EC3000/EC3000 revertant, was calculated as described (20). The CI values indicated that EC3000 showed significant fitness defects compared with E. coli MG1655 and EC3000 revertant (Fig. 2). Altogether, EC3000 showed the highest fitness cost of all tested strains. After serial passage of EC3000 (1:100 dilution) alone in colistin-free LB broth for 7 days, we observed that 32 ± 17.2% of the tested cells retained the colistin-resistant phenotype. Colistin-susceptible revertants in serially passaged EC3000 strains may have resulted from compensatory mutations, which were studied previously (20, 21).

FIG 1.

Bacterial growth curves of E. coli strains EC3000, MG1655, and EC3000 revertant. The optical density at 600 nm of each strain was measured every 20 min to determine the growth. Data illustrated are representative of three independent experiments.

FIG 2.

Relative in vitro competition assay determined over 24 h. The competition index (CI) was calculated as the number of E. coli EC3000 isolates divided by the number of E. coli MG1655 and EC3000 revertant isolates, respectively. The mean CI values from three independent experiments are shown in parentheses. Diamonds, CI obtained in each replicate.

The fitness cost study indicated that mutated pmrB significantly impacts the fitness of strain EC3000. Nevertheless, EC3000 was not completely overgrown by the colistin-susceptible EC3000 revertant or strain MG1655 in competitive cocultures. Furthermore, a portion of cultured EC3000 cells retained the colistin-resistant phenotype after 7 days of daily passage in colistin-free medium. The results raise concerns that clinically selected colistin-resistant organisms, once emerged, have the potential to persist in patients and in hospital environments, resulting in subsequent transmission. Indeed, this troublesome phenomenon has been observed in colistin-resistant K. pneumoniae isolates in many countries (5, 7).

In conclusion, we report the first chromosomal carbapenem- and colistin-resistant E. coli strain, EC3000, after previous colistin exposure. We confirmed that a novel deletion mutation in the pmrB gene contributes to colistin resistance. Despite decreased fitness, the resistance phenotype of the cultured EC3000 strain was still present after serial passage, which may further spread in hospital settings. Judicious colistin use and strict infection control measures are required to reduce such occurrence and spread of highly drug-resistant organisms.

Accession number(s).

The complete nucleotide sequences of the pmrB gene obtained from EC3000 have been deposited in the GenBank nucleotide database under accession number MN719180.