Evaluation of a DNA microarray for the rapid detection of extended-spectrum β-lactamases (TEM, SHV and CTX-M), plasmid-mediated cephalosporinases (CMY-2-like, DHA, FOX, ACC-1, ACT/MIR and CMY-1-like/MOX) and carbapenemases (KPC, OXA-48, VIM, IMP and NDM)

Abstract

Objectives

Carbapenem-resistant Gram-negative bacilli are reported increasingly and represent an emerging public health concern. Laboratory detection of extended-spectrum β-lactamase (ESBL), plasmid-mediated cephalosporinase (pAmpC) and carbapenemase producers remains a challenge for microbiology laboratories and is important to avoid clinical failure due to inappropriate antimicrobial therapy and to prevent nosocomial outbreaks. We evaluated a novel microarray, the ‘Check-MDR CT103 array’ test (Check-Points, Wageningen, The Netherlands), that employs highly specific DNA markers to identify the β-lactamase genes of ESBLs (TEM, SHV and CTX-M, and discriminates between ESBL and non-ESBL TEM and SHV variants), of pAmpC (CMY-2-like, DHA, FOX, ACC-1, ACT/MIR and CMY-1-like/MOX) and of carbapenemases (KPC, OXA-48, VIM, IMP and NDM).

Methods

One-hundred-and-eighty-seven well-characterized Gram-negative bacilli isolates possessing different bla genes were tested. Total DNAs were extracted using a Qiagen DNA mini kit. The ‘Check-MDR CT103 array’ was used as recommended by the manufacturer.

Results

The system correctly identified representatives of the three ESBL gene families tested, including differentiation between non-ESBL and ESBL TEM and SHV variants. All bla_CTX-M genes were classified into the appropriate family group (i.e. CTX-M-1 group, CTX-M-2 group, CTX-M-9 group and CTX-M-8/25/26 group). In addition, the clinically relevant plasmid-encoded cephalosporinase and carbapenemase genes were also reliably detected. Specificities and sensitivities of 100% were recorded for most bla genes.

Conclusions

The ‘Check-MDR CT103 array’ is a powerful high-throughput tool for rapid identification of ESBL, pAmpC and carbapenemase producers in culture. Because of its rapid performance, this platform is a valuable tool for epidemiological or infection control studies.

clinically relevant β-lactamases, acquired β-lactamases, culture

Introduction

Resistance to β-lactams, including carbapenems, in Gram-negative bacilli is rapidly spreading and is mainly related to the dissemination of genes encoding β-lactamases. Three main types of β-lactamases are of clinical interest: extended-spectrum β-lactamases (ESBLs), plasmid-mediated cephalosporinases (pAmpCs) and carbapenemases.

The vast majority of ESBLs belong to the TEM-, SHV- and CTX-M-type enzymes.^1–3 TEM- and SHV-type ESBLs arise via substitutions in strategically positioned amino acids from narrow-spectrum enzymes, whereas all known CTX-M enzymes have expanded-spectrum activity. Class C cephalosporinases (AmpCs) are chromosome encoded in many species of Enterobacteriaceae. These enzymes can be overproduced and, in association with porin defect, can lead to carbapenem resistance.⁴ Moreover, cephalosporinases can also be plasmid mediated (pAmpCs) and are divided into six groups: the CMY-2 group, the MIR-1 and ACT-1 group, the DHA group, the ACC-1 group, the CMY-1 group (also called MOX-1), and the FOX-1 group. These pAmpCs have been mobilized from the chromosome of natural AmpC-producing species of the Enterobacteriaceae.⁵ Although several mechanisms of carbapenem resistance have been reported, most of these are related to the spread of acquired carbapenem-hydrolysing β-lactamases. The main carbapenemases identified in Enterobacteriaceae belong to Ambler class A (KPCs), class B (VIMs, IMPs and NDMs) and class D (OXA-48-like).⁶

Laboratory detection of ESBL-, pAmpC- and/or carbapenemase-producing bacteria remains a challenge for any laboratory. Detection is routinely based on phenotypic testing,⁷ but these methods may fail to identify β-lactamase producers and also usually delay final report by at least an additional 24 h. Standard PCR and gene sequencing remain the gold standard for β-lactamase genes identification, but they fail to accurately detect the simultaneous presence of more than a single β-lactamase gene of a given family.

The microarray technology has the potential to detect an almost unlimited number of genes within a single reaction. A commercial microarray technology-based DNA test, the ‘Check-Points ESBL/KPC array’ (Check-Points, Wageningen, The Netherlands), has been developed to identify TEM-, SHV- and CTX-M-type ESBLs as well as KPC-type carbapenemases.⁸ A further development of this microarray was recently evaluated to detect additional clinically relevant carbapenemases: OXA-48, VIM, IMP and NDM.⁹ A third array was developed targeting the six pAmpC genes groups previously described, as well as bla_TEM, bla_SHV, bla_CTX-M, bla_KPC and bla_NDM.¹⁰ Here we have evaluated a novel array able to detect all the previously targeted genes, including ESBLs, pAmpCs and carbapenemases, in a single reaction vial.

Materials and methods

A total of 149 Enterobacteriaceae, 28 Pseudomonas spp., 8 Acinetobacter spp. and 2 Aeromonas spp. isolates were tested. Collection included isolates possessing different β-lactamase genes and negative control strains. The majority of the strains had previously been well characterized with respect to their β-lactamase genes.^8,9 In this collection, isolates possessed an average of three different β-lactamase genes (range, one to five). This collection also included reference bacterial isolates displaying a wild-type resistance phenotype, expressing no β-lactamase gene or only the naturally encountered β-lactamase genes, and several isolates producing other resistance genes that were used as negative control strains.^8,9

Whole-cell DNAs were extracted from overnight bacterial cultures using the QiaAmp DNA mini kit (Qiagen, Les Ulis, France). All the DNAs were diluted for the assay, thus avoiding inhibitor problems, especially with non-fermenters. Operators of the test assay were not aware of the PCR and sequencing results of the isolates, thus the interpretation of the assay was not influenced by the genotype of the isolates. Only once all the samples were analysed were the two results compared. Microarray assays were performed according to the instructions of the manufacturer and as previously outlined.^8–10 Briefly, the ‘Check-MDR CT103 array’ uses a methodology called a multiplex ligation detection reaction (LDR) to generate a collection of DNA molecules that are subsequently PCR amplified by means of a single pair of amplimers. The PCR products are next sorted by hybridization to a low-density DNA microarray. Positive hybridization is detected using a biotin label incorporated in one of the PCR primers. Tubes are then inserted in the single-channel ATR03 array tube reader upon completion of the detection reaction, and images are acquired and interpreted with the software supplied by the manufacturer (Check-Points). This software automatically translates the data into presence or absence of a specific β-lactamase gene. The microarray set-up is depicted in Figure 1(a). It includes a number of controls assessing the success of each critical step in the procedure, including ligation specificity and efficiency, PCR amplification, hybridization efficiency, label detection and label quality.

Figure 1.

Typical DNA microarray pictures obtained with the ‘Check-MDR CT103 array’ set-up. The DNA microarray, consisting of unique complementary (cZIP) oligonucleotides targeting individual probes, is fixed at the bottom of a microreaction vial. When hybridization of the PCR-amplified ligation products to the microarray is complete, colorimetric detection of the positive reactions is initiated. Each array defines the typing results of one strain and consists of control spots and specific marker spots, which are numbered from 1 to 96. (a) Theoretical display of the array probes for one strain. HybC, hybridization control; DNA-C, control for Gram-negative bacterial DNA; rxC, reaction control; negC, negative control. (b) Array results for a K. pneumoniae isolate harbouring bla_NDM-1, bla_SHV-1, bla_TEM-1, bla_CTX-M-15 and bla_CMY-23. (c) Array results for a K. pneumoniae isolate harbouring bla_OXA-48, bla_SHV-1, bla_SHV-2a and bla_TEM-1. Positive reactions of spots 15, 16, 18 and 19 denote the presence of both bla_SHV-ESBL and bla_SHV-non-ESBL genes in the isolate.

Results

One-hundred-and-eighty-seven well-characterized Gram-negative bacilli isolates possessing different bla genes were tested. An example of results obtained with two Klebsiella pneumoniae isolates possessing different bla gene patterns is shown in Figure 1 (b and c). Overall, the ‘Check-MDR CT103 array’ correctly identified representatives of the three ESBL gene families tested (SHV, TEM and CTX-M) (sensitivity 100%) (Table 1). For bla_SHV and bla_TEM genes, the array was able to differentiate between non-ESBL and ESBL variants in all isolates. Moreover, the microarray accurately identified both ESBL genes and non-ESBL genes (TEM or SHV) of the same family when both were present in one isolate. The naturally encoded SHV β-lactamase was detected in all K. pneumoniae isolates. As previously shown,⁸bla_OKP and bla_LEN genes, two other families of chromosomal β-lactamase genes encountered in K. pneumoniae isolates, were not detected by the microarray. The chromosomally encoded K1 β-lactamases of Klebsiella oxytoca, which is weakly related to CTX-Ms, was not detected by the array. All bla_CTX-M genes that were detected were classified into the appropriate family group (i.e. CTX-M-1 group, CTX-M-2 group, CTX-M-9 group and CTX-M-8/25/26 group). The software also reports four subgroups belonging to the CTX-M-1 group: CTX-M-1-like, CTX-M-3-like, CTX-M-15-like and CTX-M-32-like. Only bla_CTX-M-1 and bla_CTX-M-15 genes were tested here. For five isolates (12%), discrepancies between the array (that identified bla_CTX-M-1) and the standard PCR amplification and DNA sequencing (that identified bla_CTX-M-15) were found (data not shown).

Table 1.

Sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for the different bla genes detected by the array

Gene	Sensitivity (%)	Specificity (%)	PPV (%)	NPV (%)
blaCTX-Ma	100	100	100	100
blaSHV-ESBLb	100	100	100	100
blaSHV-non-ESBLc	100	100	100	100
blaTEM-ESBLd	100	100	100	100
blaTEM-non-ESBLe	100	100	100	100
blaKPCf	100	100	100	100
blaOXA-48g	95	100	100	99
blaNDMh	100	100	100	100
blaVIMi	100	100	100	100
blaIMPj	100	100	100	100
blaCMY/MOXk	100	100	100	100
blaDHAl	100	100	100	100
blaFOXm	100	100	100	100
blaACCn	100	100	100	100
blaACT/MIRo	100	100	100	100

Further details of the data summarized in this table can be found in Tables S1–S3 (available as Supplementary data at JAC Online).

^aIncludes CTX-M-1, CTX-M-2, CTX-M-9, CTX-M-14, CTX-M-15, CTX-M-19 and CTX-M-24.

^bIncludes SHV-2a, SHV-5 and SHV-12.

^cIncludes SHV-1, SHV-11 and OKP-A.

^dIncludes TEM-3, TEM-24 and TEM-52.

^eIncludes TEM-1, TEM-10 and TEM-24.

^fIncludes KPC-2 and KPC-3.

^gIncludes OXA-48 and OXA-181.

^hIncludes NDM-1, NDM-2 and NDM-3.

ⁱIncludes VIM-1, VIM-2, VIM-3, VIM-4 and VIM-19.

^jIncludes IMP-1, IMP-4, IMP-8 and IMP-13.

^kIncludes CMY-1, CMY-2, CMY-10, CMY-2 CMY-58, CMY-60, CMY-61 and CMY-62.

^lIncludes DHA-1, DHA-2 and DHA-7.

^mIncludes FOX-3.

ⁿIncludes ACC-1.

^oIncludes ACT-1.

Eighty-nine isolates harboured a carbapenemase gene being bla_KPC, bla_OXA-48, bla_NDM, bla_VIM or bla_IMP. All but one of them were detected in all the bacterial isolates tested with the exception of one K. pneumoniae for which bla_OXA-48 gene was not identified (Table 1). Specificities and sensitivities of 100% were recorded for the bla_KPC, bla_VIM, bla_IMP and bla_NDM genes, whereas for the bla_OXA-48 they were 100% and 95%, respectively. Positive and negative predictive values were 100% for bla_KPC, bla_VIM, bla_IMP and bla_NDM carbapenemase genes, and 100% and 99% for bla_OXA-48 genes, respectively.

Using the ‘Check-MDR CT103 array’, all 46 pAmpC-producing isolates tested were correctly identified (sensitivity 100%) (Table 1). A specificity of 100% was recorded for the detection of pAmpC genes of bla_CMY/MOX, bla_DHAbla_FOX, bla_ACC and bla_ACT/MIR types. When testing the array against 29 chromosomally encoded AmpC-producing isolates of Enterobacteriaceae, the microarray gave inconsistent results for the detection of bla_AmpC progenitor. In particular, bla_ACT/MIR was identified in all 11 E. cloacae/Enterobacter asburiae isolates (progenitor of bla_ACT/MIR), but the bla_CMY-2 gene was detected in two of seven Citrobacter freundii isolates (progenitor of bla_CMY-2-like) (28.5%). bla_DHA and bla_ACC genes, respectively, were detected in the only two Morganella morganii and Hafnia alvei isolates tested, which were the progenitors of these enzymes. bla_CMY-1 was not detected in Aeromonas hydrophila. bla_DHA was detected in two A. hydrophila and one H. alvei with a weak signal. Thus, the technique is not suited for detection of naturally encoded cephalosporinases, and proper identification of the isolates is mandatory to correctly interpret the array results.

Overall, none of the other β-lactamase genes that were used as negative controls cross-reacted with the probes used in the assay, suggesting an excellent specificity. Three samples had to be tested twice because of poor DNA extraction.

Discussion

The ‘Check-MDR CT103 array’ is designed for the rapid molecular detection of a wide range of clinically important β-lactamase genes in Gram-negative bacteria, in particular Enterobacteriaceae. Usually detection of β-lactam resistance and asymptomatic carriage of ESBL or carbapenemase producers are based on phenotypic testing and usage of screening culture media.^7,11 But these methods are time-consuming and the presence and identification of β-lactamase genes may be confirmed by molecular methods.

The ‘Check-MDR CT103 array’ is a further improvement of previously commercialized arrays. It is a combination of two arrays, the ‘Check-MDR CT101-Carba array’ and the ‘Check-MDR CT102-AmpC array’, and additionally allowed the subgrouping of important CTX-Ms of group 1. The ‘Check-MDR CT103 array’ correctly identified the targeted genes (bla_ESBLs, bla_pAmpCs and bla_{carbapenemase}), whatever the number of bla genes present in a single isolate. It was found to be specific and sensitive. The co-existence of both ESBL and non-ESBL enzymes of the same family may be difficult to detect with the usual methods, such as sequencing. The array was able to discriminate between TEM- and SHV-ESBL and non-ESBL variants and to detect both of them in the same strain. Although several mechanisms of carbapenem resistance have been reported, most of them involve β-lactamases. These β-lactamases may be a cephalosporinase or an ESBL with a very low level of carbapenem-hydrolysing activity combined with decreased permeability due to porin loss or alteration,⁴ or true carbapenem-hydrolysing β-lactamases.⁶ Differentiation of these two mechanisms may have clinical consequences for the therapeutic management of patients and prevention of nosocomial outbreaks. The ‘Check-MDR CT103 array’ was capable of distinguishing them. As with all molecular tools, this assay only identifies known genes, and does not preclude the existence of novel genes in the bacteria isolated. Therefore, proper interpretation, especially when the result is negative, needs to be done in accordance with antibiotic susceptibility results. This is especially the case with minor ESBLs and novel carbapenemases such as GES-type enzymes, which seem to be isolated more frequently.²

The ‘Check-MDR CT103 array’ has only been tested on bacterial cultures and not directly on clinical samples. It is likely that the sensitivity of the method may be insufficient to detect bla genes in the latter case. New diagnostic tools have been developed to detect relevant bla genes directly from swabs or blood cultures, such as the Hyplex^® technology. However, Hyplex^® ESBL ID targets a limited number of genes in one reaction. The analysis time of these new methods (6 h for 40 samples with microarray, 2–4 h for Hyplex^®) is relatively short compared with the barely predictable time scale of classical PCR followed by sequencing. Another advantage is their automation. For microarrays, DNA extraction and PCR are automated stages; only the hybridization step remains to be done by hand, but it is very easy to perform. The software provided by the manufacturer facilitates the interpretation of results and the method is easy to implement. The use of microarray technology on a daily routine basis is hampered by the high price : analysis ratio (catalogue price 80 euros per sample compared with <20 euros for classical PCR methods). However, when sequencing is required for TEM and SHV alleles, and often conjugation experiments to separate SHV-1 present on the chromosome of K. pneumoniae from the plasmid-encoded variants, the price and time to result of classical techniques is increased. This technology requires instruments available in numerous laboratories (automated DNA extraction instrument, thermocycler), but also specific equipment (thermomixer, tube reader including software) that costs ∼12 000 euros. Due to its rapid performance, and in order to decrease the cost of equipment, the microarray could be used in epidemiological or infection control studies in which large collections of isolates need to be characterized rapidly.

Funding

This work was funded mostly by a grant from INSERM, France, and by a grant from the European Community (7th Framework Programme FP7/2007-2013 under grant agreement no. 241742).

Transparency declarations

None to declare.

Supplementary data

Tables S1–S3 are available as Supplementary data at JAC Online (http://jac.oxfordjournals.org).

Acknowledgements

We thank Check-Points for providing the material necessary for the study and Dr Aneta Karczmarek for technical support.