Multiplex Real-Time PCR Assay for Detection and Classification of Klebsiella pneumoniae Carbapenemase Gene (bla_KPC) Variants

Abstract

Carbapenem resistance mediated by plasmid-borne Klebsiella pneumoniae carbapenemases (KPC) is an emerging problem of significant clinical importance in Gram-negative bacteria. Multiple KPC gene variants (bla_KPC) have been reported, with KPC-2 (bla_KPC-2) and KPC-3 (bla_KPC-3) associated with epidemic outbreaks in New York City and various international settings. Here, we describe the development of a multiplex real-time PCR assay using molecular beacons (MB-PCR) for rapid and accurate identification of bla_KPC variants. The assay consists of six molecular beacons and two oligonucleotide primer pairs, allowing for detection and classification of all currently described bla_KPC variants (bla_KPC-2 to bla_KPC-11). The MB-PCR detection limit was 5 to 40 DNA copies per reaction and 4 CFU per reaction using laboratory-prepared samples. The MB-PCR probes were highly specific for each bla_KPC variant, and cross-reactivity was not observed using DNA isolated from several bacterial species. A total of 457 clinical Gram-negative isolates were successfully characterized by our MB-PCR assay, with bla_KPC-3 and bla_KPC-2 identified as the most common types in the New York/New Jersey metropolitan region. The MB-PCR assay described herein is rapid, sensitive, and specific and should be useful for understanding the ongoing evolution of carbapenem resistance in Gram-negative bacteria. As novel bla_KPC variants continue to emerge, the MB-PCR assay can be modified in response to epidemiologic developments.

Carbapenems are a class of β-lactam antibiotics with broad-spectrum antibacterial activity, often used to treat infections caused by extended-spectrum β-lactamase (ESBL)-producing Gram-negative bacteria (29). Resistance to carbapenems poses serious challenges in the treatment of such infections, with pan-resistant phenotypes described for an increasing number of Enterobacteriaceae. In particular, Klebsiella pneumoniae carbapenemases (KPCs) constitute a new variant of class A β-lactamase enzymes capable of hydrolyzing all known β-lactam antibiotics and displaying resistance to β-lactamase inhibitors. As with other class A enzymes, they are carried on a variety of plasmids, thereby facilitating horizontal transmission of bla_KPC genes. Since their initial description in 2001, KPC-producing strains of K. pneumoniae and other Enterobacteriaceae have spread rapidly in the New York metropolitan region, with increasing numbers of cases reported across the United States (4, 36, 46). Similar reports have been described in other locations, including Argentina, Brazil, China, Colombia, Demark, Finland, France, Germany, Greece, Hungary, Italy, Israel, Norway, Sweden, Puerto Rico, and the United Kingdom (12, 14, 15, 23–26, 39, 43).

Ten KPC gene variants have been described to date, classified in sequential numeric order from bla_KPC-2 to bla_KPC-11 (Table 1 and Fig. 1). KPC-1 was initially reported in 2001 from a K. pneumoniae strain isolated in North Carolina (48), but subsequent revision of the bla_KPC-1 sequence demonstrated that bla_KPC-1 and bla_KPC-2 are identical (47). As shown in Table 2, the KPC-2 to KPC-11 genes are characterized by nonsynonymous single nucleotide substitutions within four codons (nucleotides [nt] 147, 308, 716, and 814). A hypothetical scenario depicting the stepwise evolution of all subsequent variants from the KPC-2 gene is shown in Fig. 1. International reports of KPC-2 (25), KPC-3 (25), KPC-4 (20, 27, 31–32), KPC-5 (45), KPC-6 (33), KPC-7 (30), KPC-8 (15), and KPC-10 (31) have been described in numerous countries, with KPC-2 and KPC-3 accounting for most epidemic outbreaks (25). Sequence data for KPC-9 and KPC-11 have been deposited in GenBank (http://www.ncbi.nlm.nih.gov), but no clinical information is available as of this writing.

Table 1.

Discovery of bla_KPC variants described to date^a

blaKPC gene	KPC enzyme	Species	Yr isolated	Location	GenBank accession no.	Reference(s)
blaKPC-1b	KPC-1b	Klebsiella pneumoniae	1996	North Carolina	AF297554	48
blaKPC-2	KPC-2	K. pneumoniae	1998–1999	Maryland	AY034847	36
blaKPC-3	KPC-3	K. pneumoniae	2000–2001	New York	AF395881	46
blaKPC-4	KPC-4	Enterobacter cancerogenus	2003	Scotland	AY700571	27
blaKPC-5	KPC-5	Pseudomonas aeruginosa	2006	Puerto Rico	EU400222	44 and 45
blaKPC-6	KPC-6	K. pneumoniae	2003	Puerto Rico	EU555534	33
blaKPC-7	KPC-7	K. pneumoniae	2007–2008	Ohio	EU729727	30
blaKPC-8	KPC-8	K. pneumoniae	2008	Puerto Rico	FJ234412	15
blaKPC-9	KPC-9	Escherichia coli	2009	Israel	FJ624872	Unpublished
blaKPC-10	KPC-10	Acinetobacter baumannii	2009	Puerto Rico	GQ140348	31
blaKPC-11	KPC-11	K. pneumoniae	2010		HM066995	Unpublished

^aSpecies, year, and location of initial report for each variant.

^bbla_KPC-1 and KPC-1 are no longer considered valid designations, as their sequences are identical to those of bla_KPC-2 and KPC-2, respectively (47).

Fig. 1.

Hypothetical stepwise evolution of bla_KPC variants, based on single nucleotide changes and order of discovery. In this scenario, bla_KPC-2 is the prototype from which all subsequent variants have arisen (44) and gives rise by stepwise nonsynonymous substitution to bla_KPC-3 (46), bla_KPC-5 (44), bla_KPC-6 (33), and bla_KPC-11 (GenBank accession no. HM066995). Similarly, bla_KPC-3 gives rise to bla_KPC-7 (30), bla_KPC-8 (15), and bla_KPC-10 (31), while bla_KPC-4 (27) and bla_KPC-9 (FJ624872) each differ by two nucleotides from bla_KPC-2 and bla_KPC-3, respectively. Alternatively, bla_KPC-10 and bla_KPC-11 may be derived from bla_KPC-5, bla_KPC-4 may be derived from either bla_KPC-5 or bla_KPC-6, and bla_KPC-9 may be derived from bla_KPC-8. Nucleotide sequences for each bla_KPC variant are deposited in GenBank (NCBI) under the following accession numbers: AY034847 (bla_KPC-2), AF395881 (bla_KPC-3), AY700571 (bla_KPC-4), EU400222 (bla_KPC-5), EU555534 (bla_KPC-6), EU729727 (bla_KPC-7), FJ234412 (bla_KPC-8), FJ624872 (bla_KPC-9), GQ140348 (bla_KPC-10), and HM066995 (bla_KPC-11).

Table 2.

Nucleotide and amino acid differences between KPC enzymes (bla_KPC variants)

KPC variant	Nonsynonymous substitutiona				Molecular beacon probeb
	nt 147 (aa 49)	nt 308 (aa 103)	nt 716 (aa 239)	nt 814 (aa 272)	147G	308C	716T	716Cc/308T	814C
KPC-2	ATG (Met)	CCG (Pro)	GTG (Val)	CAC (His)	+	+	+	−	+
KPC-3	ATG (Met)	CCG (Pro)	GTG (Val)	TAC (Tyr)	+	+	+	−	−
KPC-4	ATG (Met)	CGC (Arg)	GGG (Gly)	CAC (His)	+	−	−	−	+
KPC-5	ATG (Met)	CGC (Arg)	GTG (Val)	CAC (His)	+	−	+	−	+
KPC-6	ATG (Met)	CCG (Pro)	GGG (Gly)	CAC (His)	+	+	−	−	+
KPC-7	ATA (Ile)	CCG (Pro)	GTG (Val)	TAC (Tyr)	−	+	+	−	−
KPC-8	ATG (Met)	CCG (Pro)	GGG (Gly)	TAC (Tyr)	+	+	−	−	−
KPC-9	ATG (Met)	CCG (Pro)	GCG (Ala)	TAC (Tyr)	+	+	−	+	−
KPC-10	ATG (Met)	CGC (Arg)	GTG (Val)	TAC (Tyr)	+	−	+	−	−
KPC-11	ATG (Met)	CTG (Leu)	GTG (Val)	CAC (His)	+	−	+	+	+

^aNucleotide and amino acid (aa) positions of nonsynonymous substitutions in bla_KPC variants. Single nucleotide polymorphisms used to design molecular beacon probes are underlined, and amino acids specified by each codon are displayed.

^bMolecular beacon probe dyes and quenchers are listed in Table 3. Plus and minus signs denote the presence and absence of the target, respectively.

^cMixture of MB308T and MB716C, each labeled with the same fluorescent dye (Quasar 670). Positive amplification will be observed if either bla_KPC-9 (716C) or bla_KPC-11 (308T) is present.

Worrisomely, KPC gene variants have been described in numerous Gram-negative species, including Klebsiella spp. (3, 36, 49), Enterobacter spp. (20, 27), Escherichia coli (4), Pseudomonas aeruginosa (44), Acinetobacter spp. (31), Salmonella enterica (22), Serratia marcescens (7), Citrobacter freundii (50), Proteus mirabilis (38), and most recently Raoultella spp. (8). The successive appearance of distinct bla_KPC variants in different species is suggestive of adaptation in response to antibiotic pressure and highlights the ease with which plasmid-borne variants can spread among Enterobacteriaceae, thereby accelerating the global dissemination of carbapenem-resistant strains. Consequently, timely identification of KPC variants is critical for understanding the ongoing evolution of carbapenem resistance in Gram-negative bacteria and can inform infection control measures directed against this emerging public health threat.

In this study, a multiplex real-time PCR scheme was developed which can identify and classify all 10 current bla_KPC gene variants (bla_KPC-2 to bla_KPC-11) in a single reaction using molecular beacon probes. Molecular beacons are short oligonucleotide probes with a hairpin stem-loop structure which exhibit fluorescence only when the loop region anneals to a complementary target (41). The presence of the hairpin significantly enhances specificity (40), enabling detection of single nucleotide polymorphisms (SNPs) in an unambiguous and reproducible manner (21). The assay is rapid and simple to perform, with a high degree of sensitivity and specificity, and may be expanded to accommodate the discovery of novel variants.

MATERIALS AND METHODS

Bacterial strains.

K. pneumoniae strains BK26633 and BK26625 were selected as positive controls for bla_KPC from our collection at the Public Health Research Institute Tuberculosis (PHRI TB) Center. Sequence analysis determined that the bla_KPC genes of BK26633 and BK26625 are identical to those of Klebsiella oxytoca strain 3127 (bla_KPC-2, GenBank accession no. AY210886) and K. pneumoniae strain CL 5761 (bla_KPC-3, GenBank accession no. AF395881) (46), respectively. Positive-control strains containing bla_KPC-4 to bla_KPC-8 were created using a site-saturation mutagenesis method as described in reference 28 at the Louis Stokes Cleveland Department of Veterans Affairs Medical Center in Cleveland, OH (see below for details). Control strains E. coli ATCC 25922 and P. aeruginosa ATCC 27853 were acquired from the American Type Culture Collection (ATCC). All other clinical isolates used in this study were selected from the PHRI TB Center strain repository, obtained from 17 hospitals in New York and New Jersey between 1998 and 2010 (five hospitals from New York City, six from northern New Jersey, and six from southern New Jersey).

Molecular beacon design and synthesis.

Molecular beacon probes were designed according to guidelines described at http://www.molecular-beacons.org and obtained from Biosearch Technologies (Novato, CA). Oligonucleotide primers were designed using Primer3 software (34) and obtained from Integrated DNA Technologies (Coralville, IA). BLAST (http://blast.ncbi.nlm.nih.gov) similarity searches were performed to rule out potential cross-reactivity between primers/probes and nonspecific targets. Six molecular beacons were synthesized with the following 5′ modifications: FAM (fluorescein), HEX (hexachlorofluorescein), CAL Fluor Red 610, Quasar 670, and amino (Table 3); molecular beacon probes MB716C and MB308T utilize the same dye (Quasar 670) (Table 2). For probe MB716T, the 5′ amino modification was coupled with fluorophore Atto 425 as described on the ATTO-TEC GmbH website. Briefly, a molecular beacon probe with a 3′ DABCYL [4-(4′-dimethylaminophenylazo)benzoic acid] quencher and a 5′ amino modification was obtained from Biosearch Technologies and then labeled with Atto 425–N-hydroxysuccinimide (NHS) ester (Sigma-Aldrich, St. Louis, MO) in 0.1 M sodium carbonate buffer (pH 9). All molecular beacons were purified by high-performance liquid chromatography (HPLC) as described elsewhere (42).

Table 3.

Oligonucleotide primers and molecular beacon probes used in this study^a

Primer/probeb	Sequence (5′ to 3′)c	Size (bp)	5′ fluorescent dye	3′ quencher
KPC-F107	TCGAACAGGACTTTGGCGGCT	260
KPC-R366	GGACAGCTCCGCCACCGTCATG
MB147G	gcgct CGATGGATACCGGCTCA agcgc		FAM	DABCYL
MB308C	cgcga CTGGTTCCGTGGTCAC tcgcg		HEX	DABCYL
MB308T	cgcga CTGGTTCTGTGGTC tcgcg		Quasar 670d	BHQ-2
KPC-F684	GGCAGTCGGAGACAAAACC	177
KPC-R860	CCCTCGAGCGCGAGTCTA
MB716T	cgcga AACCTGCGGAGTGTATGG tcgcg		Atto 425	DABCYL
MB716C	cgcga AACCTGCGGAGCGTATGG tcgcg		Quasar 670d	BHQ-2
MB814C	cgacg GACAAGCACAGCGAGG cgtcg		CAL Fluor Red 610	BHQ-2

^aFAM, fluorescein; HEX, hexachlorofluorescein; DABCYL, 4-(4′-dimethylaminophenylazo)benzoic acid; BHQ, black hole quencher.

^bOligonucleotide primer sets spanning nucleotide positions 107 to 366 and 684 to 860 were used to amplify two segments, each containing two molecular beacon target positions. Positions 308 and 716 are targeted by more than one molecular beacon probe.

^cMolecular beacon hairpin sequences are shown in lowercase letters. Single nucleotide polymorphism target positions are underlined, as in Table 2.

^dQuasar 670 dye is used for both MB308T and MB716C, as explained in Table 2.

DNA isolation.

DNA was isolated from bacterial colonies using the boiling lysis method described elsewhere (9). Bacterial strains were grown on Luria-Bertani (LB) agar and incubated overnight at 37°C. A loopful of bacterial growth was suspended in 100 μl of sterile double-distilled water (ddH₂O) and boiled for 10 min in a water bath. Following centrifugation at 16,000 × g for 5 min, the supernatant was diluted 10-fold into ddH₂O, and 1 μl was used as the template in a 10-μl PCR. For analytical sensitivity based on bacterial CFU, DNA isolation was performed using the NucliSENS easyMAG system (bioMérieux, Durham, NC). In brief, overnight bacterial cultures were serially diluted 10-fold in LB broth, and 200 μl of each dilution was used to extract total nucleic acid (TNA) according to the manufacturer's instructions. Bacterial TNA was eluted in 50 μl elution buffer and stored at −20°C.

Real-time PCR.

A multiplex real-time PCR scheme using molecular beacons (MB-PCR) was designed for rapid detection of variants bla_KPC-2 to bla_KPC-11 (Table 2). The combination of two sets of primers and six molecular beacons (Table 3) results in a minimum of two specific targets for each bla_KPC gene variant. Real-time PCR amplification was performed using the Stratagene Mx3005P multiplex quantitative PCR system (Agilent Technologies, Santa Clara, CA). PCR was performed in 10-μl reactions consisting of 0.5 U of AmpliTaq Gold DNA polymerase (Applied Biosystems, Foster City, CA), 4 mM MgCl₂, 250 μM each deoxynucleoside triphosphate (dNTP), and 0.5 μM each primer (Table 3) in 1× PCR buffer (Applied Biosystems). Optimized molecular beacon concentrations were as follows: 0.4 μM for MB716T and MB716C; 0.2 μM for MB308C, MB308T, and MB814C; and 0.1 μM for MB147G. Optimal cycling conditions included an initial denaturation step of 95°C for 10 min, followed by 40 cycles of 95°C for 20 s (denaturation) and 58°C for 45 s (annealing and extension). The entire MB-PCR can be performed in less than 1.5 h on the Mx3005P platform. K. pneumoniae strains BK26633 and BK26625 were used as positive controls in each PCR assay, while E. coli ATCC 25922 and P. aeruginosa ATCC 27853 were used as negative controls.

Analytical sensitivity.

The analytical sensitivity of the MB-PCR assay was estimated by serial dilution experiments using purified synthesized DNA templates. At the time the study was performed, only bacterial strains harboring bla_KPC-2 and bla_KPC-3 were available. Overlap extension PCR was therefore used to generate the complete coding sequences of bla_KPC-4 to bla_KPC-11 (17). Briefly, site-directed mutagenesis was performed using oligonucleotide primers containing single nucleotide polymorphisms (SNPs) specific to each bla_KPC variant. Overlapping PCR fragments were then fused together in successive PCR extension reactions. For consistency, the complete coding sequences of bla_KPC-2 and bla_KPC-3 were similarly generated using the same primer set. The resulting synthesized targets (bla_KPC-2 to bla_KPC-11) were further purified using the QIAquick gel extraction kit (Qiagen, Hilden, Germany), and DNA concentrations were measured using the Invitrogen Quant-iT PicoGreen dsDNA assay kit (Molecular Probes, Eugene, OR). DNA templates were serially diluted in ddH₂O, and l μl of each dilution was used for MB-PCR as described above. Limits of detection for each MB-PCR target were then calculated based on the molecular weight and concentration of each target.

In order to estimate analytical sensitivity using bacterial CFU, transformants carrying bla_KPC-4 through bla_KPC-8 were created by site-saturation mutagenesis using the Stratagene QuikChange XL site-directed mutagenesis kit (Agilent Technologies, Santa Clara, CA), as described elsewhere (28). Plasmids pET24a(+)-bla_KPC-4 through pET24a(+)-bla_KPC-8 were electroporated into E. coli DH10B competent cells, following which their respective bla_KPC regions were sequenced in order to verify the presence of the correct variants. Overnight cultures of the resulting transformants, along with K. pneumoniae strains BK26633 (bla_KPC-2) and BK26625 (bla_KPC-3), were harvested by centrifugation and serially diluted 10-fold into sterile LB medium. DNA was isolated from 200 μl of each dilution using the NucliSENS easyMAG instrument as described above, and 1 μl of each sample was used for MB-PCR amplification. The numbers of CFU per suspension (CFU/ml) were estimated using standard plating procedures and were used to calculate MB-PCR limits of detection for molecular beacon probes MB147G, MB308C, MB716T, and MB814C. The analytical sensitivities of MB716C (bla_KPC-9) and MB308T (bla_KPC-11) were not determined by this method, since these targets are present only in bla_KPC-9 and bla_KPC-11, and transformants containing these variants were not created for this study.

Analytical specificity.

Analytical specificities for each molecular beacon probe were initially tested using the synthesized bla_KPC-2 to bla_KPC-11 targets described above. DNA was isolated by boiling lysis from the aforementioned bla_KPC-4 to bla_KPC-8 E. coli DH10B transformants, as well as BK26633 (bla_KPC-2) and BK26625 (bla_KPC-3), and then used as a template to rule out cross-reactivity between different bla_KPC gene variants. In addition, the analytical specificities of all primer/probe combinations were tested using several other species, including E. coli ATCC 25922, P. aeruginosa ATCC 27853, Enterococcus spp. (n = 2 isolates), Staphylococcus spp. (n = 8 isolates), Streptococcus spp. (n = 2 isolates), and Candida albicans (n = 4 isolates). All strains except ATCC 25922 and ATCC 27853 were selected from a previous study (9).

Detection of bla_KPC variants in clinical isolates.

Fifty-three bla_KPC-positive K. pneumoniae isolates were obtained from three hospitals in the New York/New Jersey metropolitan area and tested using the multiplex real-time MB-PCR assay described here. The isolates were collected from multiple infection sites, including blood, wound, sputum, urine, bone, bile, and bronchial fluid (Table 4). Carbapenemase production was determined for all isolates using the modified Hodge test (19), and the presence of bla_KPC was confirmed by conventional PCR (35). Additional Gram-negative bacterial strains (n = 404 isolates) were randomly chosen from the PHRI TB Center collection and also subjected to bla_KPC analysis by the MB-PCR assay. The latter were collected between 1998 and 2010 from 17 hospitals in New Jersey and New York and included K. pneumoniae, K. oxytoca, E. coli, Enterobacter spp., Acinetobacter spp., S. marcescens, P. aeruginosa, and C. freundii (Table 5).

Table 4.

MB-PCR results for 53 bla_KPC-positive K. pneumoniae isolates

Isolate source	No. of isolates with:
	blaKPC-2	blaKPC-3
Blood	2	3
Wound	1	4
Sputum	1	7
Urine	5	25
Othera	0	5

^aIsolated from bone (n = 1 isolate), bile (n = 1 isolate), or bronchial fluid culture (n = 3 isolates).

Table 5.

MB-PCR results for 404 randomly selected clinical Gram-negative isolates

Species	No. of isolates with:		Total no. of isolates
	blaKPC-2	blaKPC-3
Acinetobacter spp.	0	0	67
Citrobacter freundii	0	1	1
Escherichia coli	3	3	35
Enterobacter aerogenes	1	0	1
Enterobacter cloacae	0	3	13
Enterobacter sakazakii	0	0	3
Klebsiella pneumoniae	72	97	230
Klebsiella oxytoca	0	0	3
Pseudomonas aeruginosa	0	0	25
Serratia marcescens	0	0	26

RESULTS

Classification of bla_KPC variants by MB-PCR assay.

All 10 currently described bla_KPC variants were successfully detected using a combination of two oligonucleotide primer sets and six molecular beacon probes (Table 3). Each variant displayed at least two positive amplification signals (Table 2), with four signals displayed by bla_KPC-2 and bla_KPC-11; three signals by bla_KPC-3, bla_KPC-5, bla_KPC-6, and bla_KPC-9; and two signals by bla_KPC-4, bla_KPC-7, bla_KPC-8, and bla_KPC-10. All bla_KPC variants were positive for target 147G, with the exception of bla_KPC-7. Most of the variants could be classified using only four probes/dyes, except for two closely related pairs which required the use of a fifth target. A mixture of MB308T and MB716C was therefore included to distinguish between bla_KPC-5/bla_KPC-11 and bla_KPC-8/bla_KPC-9, respectively (Table 3). Characteristic MB-PCR amplification plots for bla_KPC-2 and bla_KPC-3 are depicted in Fig. 2.

Fig. 2.

Real-time PCR amplification results for bla_KPC-2 (A) and bla_KPC-3 (B), obtained after 40 cycles on the Stratagene Mx3005P multiplex quantitative PCR system. The multiplex real-time MB-PCR assay consists of six molecular beacon probes labeled with five different fluorescent dyes, as indicated in the legend.

Analytical sensitivity.

The sensitivity of the real-time PCR platform was initially evaluated using synthesized bla_KPC-2 to bla_KPC-11 targets. Limits of detection based on DNA copy number were obtained for each molecular beacon using purified synthesized DNA, as described in Materials and Methods. Lower limits were reliably detected within 45 cycles as follows: 5 to 10 copies per PCR for targets MB147G, MB308C, MB308T, and MB814C and 20 to 40 copies for targets MB716T and MB716C. Limits of detection based on bacterial CFU were also estimated using DNA isolated from serial dilutions of bacterial cultures. The multiplex assay was capable of reproducibly detecting 2 CFU per PCR for MB147G, MB308C, and MB814C and 4 CFU for MB716T. The overall detection limit of the MB-PCR assay was therefore approximately 4 CFU per reaction.

Analytical specificity.

All molecular beacon primer/probe sets used in this study displayed good specificity for their respective bla_KPC targets, with no cross-reactivity observed between individual beacons and synthesized targets. Comparable results were obtained using K. pneumoniae strains harboring bla_KPC-2 and bla_KPC-3 and E. coli DH10B transformants carrying bla_KPC-4 through bla_KPC-8. No amplification was observed among bla_KPC-negative Enterobacteriaceae isolates (Table 5), while analysis of other organisms, including E. coli ATCC 25922, P. aeruginosa ATCC 27853, Enterococcus spp., Staphylococcus spp., Streptococcus spp., and Candida albicans yielded negative results, in agreement with BLAST similarity searches performed during the assay design phase.

Detection of bla_KPC variants in clinical isolates.

Fifty-three bla_KPC-positive clinical K. pneumoniae isolates were selected in order to validate the multiplex real-time MB-PCR assay. All isolates were initially shown to produce carbapenemase using the modified Hodge test, with bla_KPC carriage confirmed by conventional PCR. Isolates were obtained from diverse specimen sources, including urine (n = 30), sputum (n = 8), blood (n = 5), wound (n = 5), bronchial fluid (n = 3), bone (n = 1), and bile (n = 1). All 53 isolates were positive by MB-PCR, in agreement with conventional PCR results (Table 4). Among the 53 isolates, 44 (83.0%) harbored the bla_KPC-3 variant, while only 9 (17.0%) harbored bla_KPC-2.

Following validation of the MB-PCR assay, an additional 404 clinical isolates from 17 hospitals in New Jersey and New York were also screened for bla_KPC variants. These isolates, collected from 1998 to 2010, were comprised of several Gram-negative species, including K. pneumoniae, K. oxytoca, E. coli, Enterobacter spp., Acinetobacter spp., S. marcescens, P. aeruginosa, and C. freundii (Table 5). Within this collection, 104 (25.7%) isolates were positive for bla_KPC-3, while 76 (18.8%) harbored bla_KPC-2; no other bla_KPC variants were identified in this study. Among the 230 K. pneumoniae isolates tested, a total of 169 (73.5%) harbored bla_KPC genes, of which 97 (57.4%) were bla_KPC-3 and 72 (42.6%) were bla_KPC-2. In contrast, only six (17.1%) bla_KPC-positive isolates were identified among 35 E. coli strains tested, with bla_KPC-2 and bla_KPC-3 represented equally. Three Enterobacter cloacae isolates also harbored bla_KPC-3 genes, while one E. aerogenes isolate was positive for bla_KPC-2. Similarly, one C. freundii isolate was positive for bla_KPC-3, whereas no bla_KPC variants were found among the rest of the species included in this study. In order to further validate these findings, we sequenced the entire bla_KPC coding regions from 40 randomly selected bla_KPC-positive strains and observed 100% concordance with our MB-PCR results (data not shown).

DISCUSSION

Carbapenem resistance among Gram-negative bacteria is an emerging concern of significant clinical and public health importance worldwide (25). In light of the potential for rapid horizontal and vertical transmission of these genes, prompt recognition of KPC-producing organisms is critical for controlling their spread in nosocomial and long-term-care settings. Unfortunately, detection of KPC carbapenemases by susceptibility testing is challenging, due to the heterogeneous expression of β-lactam resistance by multiple determinants (18, 25). Various studies have reported that carbapenem-resistant bacteria may be incorrectly identified as carbapenem susceptible, resulting in inappropriate selection of therapy (2, 5, 37). In order to overcome such shortcomings when treating infections caused by Enterobacteriaceae, the Clinical and Laboratory Standards Institute (CLSI) has recently lowered the susceptibility breakpoints for meropenem, imipenem, and doripenem to ≤1 μg/ml and to ≤0.25 μg/ml for ertapenem (10). Nevertheless, despite the new breakpoints, susceptibility testing results still vary among different methods (6).

Given these limitations, molecular detection of bla_KPC genes by PCR has been proposed as the gold standard (25) for detection of KPC-bearing organisms. To date, several PCR-based detection methods have been described (35), including two real-time PCR assays (11, 16), as well as a method that uses PCR in conjunction with electrospray ionization mass spectrometry (PCR/ESI-MS) (13). However, most of these methods do not distinguish between different bla_KPC variants, thereby failing to convey information of epidemiologic as well as evolutionary significance. In addition, evidence suggests that some nonsynonymous mutations within bla_KPC variants may be of functional significance as well (44). KPC-3, for instance, exhibits a higher rate of ceftazidime hydrolysis than KPC-2 but has lower affinity for cefoxitin (1). Similarly, the activity spectra of KPC-4 and KPC-5 for certain β-lactam antibiotics may differ from that of KPC-2 (44). KPC enzymes may therefore be evolving through stepwise mutation (Fig. 1) in a manner that alters their functional activity (44). Consequently, identification of specific bla_KPC variants may be relevant to selection of appropriate therapies for infections caused by multidrug-resistant Gram-negative organisms.

Currently, DNA sequencing is the definitive method for identification of bla_KPC variants. However, sequencing is impractical for studies involving large sample sizes, as well as for rapid identification in clinical settings. In contrast, methods such as real-time PCR offer rapid, robust, and cost-efficient alternatives to DNA sequencing for known bla_KPC variants. Previously, Cole et al. (11) developed an assay for bla_KPC detection using TaqMan real-time PCR, with subsequent differentiation of bla_KPC-2 and bla_KPC-3 by restriction enzyme analysis. However, the assay is based on one SNP (nt 944, equivalent to nt 814 in the MB-PCR assay described here), corresponding to an RsaI restriction site in bla_KPC-3. Consequently, as those authors point out, their assay can potentially misclassify bla_KPC-4 to bla_KPC-6 as either bla_KPC-2 or bla_KPC-11, while bla_KPC-7 to bla_KPC-10 may be misclassified as bla_KPC-3. Moreover, this method requires postamplification restriction analysis and electrophoresis (11), which is time-consuming and unsuitable for large-scale studies.

In this study, we describe a multiplex real-time PCR assay for rapid detection and classification of all currently known bla_KPC variants. This platform includes two sets of oligonucleotide primers and six molecular beacon probes, enabling detection and differentiation of 10 bla_KPC gene variants in a single reaction. Each variant exhibits at least two positive amplification curves, increasing the reliability of bla_KPC detection. Using synthesized bla_KPC targets, our MB-PCR assay was successfully able to detect all bla_KPC gene variants, with no cross-reactivity observed between mutant and wild-type targets. Further analysis using E. coli DH10B transformants harboring bla_KPC-4 to bla_KPC-8, as well as bla_KPC-2- and bla_KPC-3-positive K. pneumoniae strains, confirmed the assay's ability to specifically identify these variants. The MB-PCR assay also demonstrated very good analytical sensitivity, detecting as few as 5 to 40 DNA copies and 4 CFU per reaction, similar to the sensitivity reported (1 CFU) by Hindiyeh et al. (16). Lastly, the assay successfully classified bla_KPC variants in 53 K. pneumoniae isolates from diverse clinical specimens, while screening of 404 Gram-negative strains uncovered bla_KPC genes in four additional species. As shown in Table 5, while bla_KPC-2 and bla_KPC-3 may be found in the same species, no other bla_KPC gene variants were detected in this study.

Aside from greater sensitivity and specificity, real-time PCR does not require postamplification analysis, thereby facilitating results with significantly less time and labor. Similarly, the relatively high price of molecular beacons is offset by savings in sample processing costs and labor, as described previously (9). Moreover, based on the regional frequencies of different bla_KPC variants, the MB-PCR scheme may be modified to include a smaller number of molecular beacons. In our region, for example, bla_KPC-2 and bla_KPC-3 appear to be predominant, requiring only four molecular beacons (MB147G, MB308C, MB716T, and MB814C) for definitive identification (Table 2). Such modifications may be of practical significance given the limited availability of real-time PCR platforms capable of recognizing more than four dyes.

In conclusion, we have described a multiplex real-time PCR assay for the identification and classification of bla_KPC gene variants from Gram-negative bacterial isolates. As many as 96 strains may be classified easily within 3 h, including DNA isolation, PCR cycling, and data analysis. The MB-PCR assay is robust, sensitive, and specific, allowing for high-throughput detection and classification of all presently described bla_KPC variants. Nevertheless, as novel bla_KPC variants continue to emerge, the typing system described herein can be modified in response to epidemiologic developments.