Abstract

Objectives

Fast and adequate detection of extended-spectrum β-lactamases (ESBLs) is crucial for infection control measures and the choice of antimicrobial therapy. The aim of this study was to develop and evaluate a novel ESBL assay using ligation-mediated amplification combined with microarray analysis to detect the most prevalent ESBLs in Enterobacteriaceae: TEM, SHV and CTX-M.

Methods

Analysis of the Lahey database revealed that the vast majority of TEM and SHV ESBLs differ from non-ESBL variants in three amino acid positions. TEM ESBLs have at least one of the following amino acid substitutions: R164S/H/C, G238D/N/S and E104K. In SHV ESBLs, one or more of the following substitutions is observed: D179A/N/G, G238S/A and E240K. Oligonucleotide probes were designed to detect these substitutions, covering 95% of ESBL TEM variants and 77% of ESBL SHV variants. In addition, probes were designed to distinguish between CTX-M groups 1, 2, 9 and 8/25. For evaluation of the assay, 212 Enterobacteriaceae isolates with various β-lactamases were included (n = 106 ESBL positive).

Results

The sensitivity of the microarray was 101/106 (95%; 95% CI 89%–98%), and the specificity 100% (95% CI 97%–100%) using molecular characterization of ESBLs by PCR and sequencing as reference. Assay performance time was 8 h for 36 isolates.

Conclusions

This novel commercially available DNA microarray system may offer an attractive option for rapid and accurate detection of CTX-M, TEM and SHV ESBL genes in Enterobacteriaceae in the clinical laboratory.

Introduction

Rapid and adequate extended-spectrum β-lactamase (ESBL) detection is crucial for infection control measures and the choice of antimicrobial therapy. Because phenotypic detection of ESBLs is time consuming and the results may be difficult to interpret, a faster and accurate detection method is desirable. A microarray-based genotypic test may meet these demands. However, published data on microarrays for ESBL detection are sparse.

The aim of this study was to develop and evaluate a genotypic assay, designed to identify the most prevalent ESBLs in the clinical setting (TEM, SHV and CTX-M).1–3 TEM and SHV ESBLs differ from their ancestral non-ESBL TEM and SHV penicillinases by one or more amino acid substitutions.4–6 Genotypic detection of TEM and SHV ESBL mutations therefore requires identification of single nucleotide polymorphisms (SNPs). CTX-M ESBLs can be divided into five clusters, i.e. groups 1, 2, 8, 9 and 25, based on their sequence similarity.7

In this study, ligation-mediated amplification combined with detection of amplified products on a microarray was used to detect the various CTX-M groups and the ESBL-associated SNPs in TEM and SHV variants (Figure 1).

Figure 1

Principle of the microarray system (adapted from Wattiau et al.9). (a) Probe arms hybridize to the target sequence resulting in a double-stranded molecule with a single-stranded nick. This nick will subsequently be ligated by the DNA ligase in the case of a perfect double-stranded match, connecting the two probe arms (upper left). Probe arms will not be connected in the case of a mismatch (upper right). White boxes indicate generic primer-binding sites that are identical in each target-specific probe. The black box indicates the ZIP sequence, specific for each probe, and complementary to its cZIP counterpart on the microarray. (b) Only successfully ligated probe arms will be amplified by PCR using a single primer pair, generating biotinylated products, because one of the primers has a 5-biotin moiety. A single pair of primers will co-amplify all target-specific probes, because they share the generic primer-binding sites. (c) Detection of each target-specific amplification product at a specific position on the microarray is achieved through unique ZIP codes (boxes with various patterns). PCR products are hybridized to the microarray and visualized using colorimetric detection with streptavidin–horseradish peroxidase and tetramethyl benzidine.

Figure 1

Principle of the microarray system (adapted from Wattiau et al.9). (a) Probe arms hybridize to the target sequence resulting in a double-stranded molecule with a single-stranded nick. This nick will subsequently be ligated by the DNA ligase in the case of a perfect double-stranded match, connecting the two probe arms (upper left). Probe arms will not be connected in the case of a mismatch (upper right). White boxes indicate generic primer-binding sites that are identical in each target-specific probe. The black box indicates the ZIP sequence, specific for each probe, and complementary to its cZIP counterpart on the microarray. (b) Only successfully ligated probe arms will be amplified by PCR using a single primer pair, generating biotinylated products, because one of the primers has a 5-biotin moiety. A single pair of primers will co-amplify all target-specific probes, because they share the generic primer-binding sites. (c) Detection of each target-specific amplification product at a specific position on the microarray is achieved through unique ZIP codes (boxes with various patterns). PCR products are hybridized to the microarray and visualized using colorimetric detection with streptavidin–horseradish peroxidase and tetramethyl benzidine.

Materials and methods

Probe design

To identify essential SNPs associated with the ESBL phenotype, sequences of all listed TEM variants (n = 175) and SHV variants (n = 128) in the Lahey database (http://www.lahey.org/Studies/) as of 6 September 2009 were related to phenotypes described in the literature (www.pubmed.gov). In line with previous reports, it was concluded that 95% (84 of 88) of the TEM variants with an established ESBL phenotype have one or more amino acid substitution at Ambler's position 104, 164 or 238, and 77% (27 of 35) of ESBL SHV variants have substitutions at position 179, 238 or 240 [Tables S1 and S2, available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/)].4,6

Based on these findings, 10 oligonucleotide probes [Figure S1, available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/)] were designed for identification of TEM variants (3 probes for non-ESBL variants, 5 for ESBL variants and 2 for discrimination of TEM-116). The probes detect the following ESBL-associated amino acid substitutions: E104K; R164S/C/H; and G238S. Four rare ESBL TEM variants (TEM-126, -157, -164 and -169) will not be detected because they lack these substitutions. The microarray system reports TEM-116 as ESBL positive although the Lahey database categorizes TEM-116 as an unknown phenotype, and publications on the phenotype of TEM-116 are controversial.

For identification of SHV variants, 10 oligonucleotide probes (Figure S1) were designed (3 probes for non-ESBL variants, 1 for exclusion of SHV-10 and 6 for ESBL variants) to detect the following ESBL-associated substitutions: D179A/N/G; G238S/A; and E240K. The eight ESBL SHV variants that are missed by the microarray are SHV-16, -27, -38, -40, -41, -42, -57 and -70. However, data on the ESBL phenotype of SHV-27, -40, -41 and -42 are controversial.

Four probes were designed for discrimination between CTX-M groups CTX-M-1, CTX-M-2, CTX-M-9 and CTX-M 8/25. The detection of the CTX-M-8 and -25 groups was combined because of the low prevalence of these gene clusters.

Ligation-mediated amplification and microarray analysis

The ESBL Array (Check-Points B.V., Wageningen, The Netherlands) was supplied as a kit. Microarray images were generated using a microarray reader (ArrayTube Reader, ClonDiag Chip Technologies, Jena, Germany) connected to a computer running dedicated software for analysis of the images. The software indicates whether a TEM, SHV or CTX-M ESBL or a combination is detected and specifies the CTX-M group. By visual inspection of the microarray pictures, the accuracy of individual probes in identifying specific SNPs was determined. Three isolates can be analysed in parallel on one microarray [Figure S2, available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/)]. The performance time of the microarray was 8 h per 36 isolates (3 h DNA isolation; and 5 h ligation, amplification and detection).

Test isolates

A total of 212 genotypically and phenotypically well-characterized Enterobacteriaceae were tested [Tables S3 and S4, available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/)]. The collection contained 106 ESBL-positive and 106 ESBL-negative isolates (1 isolate per patient) of human (132) and veterinary (80) origin. An overview of the ESBL genes present is displayed in Table 1. With this collection, 20 of the 24 probes for ESBL detection could be evaluated, while 4 probes detecting TEM R164C and SHV D179A/N/G could not be tested.

Table 1

Detection of ESBL-associated substitutions by microarray in 106 ESBL-positive isolates (110 genes, since 4 isolates harboured 2 ESBL genes)

    Number of ESBL substitutions detected/number of expected ESBL substitutions
 
TEM ESBL genes in collection ESBL-associated substitutions detectable by array Number of isolates ESBL-positive result in array E104K R164S R164H G238S 
TEM-18 E104K 1/1 — — — 
TEM-9, -63 E104K, R164S 3a 3/3 1/3a — — 
TEM-6, -43 E104K, R164H 2/2 — 2/2 — 
TEM-3, -4, -52 E104K, G238S 11 11 11/11 — — 11/11 
TEM-8 E104K, R164S, G238S 1/1 1/1 — 1/1 
TEM-5, -7, -10, -12 R164S 4b — 4/6b — — 
TEM-19, -20, -72 G238S 4b — — — 4/5b 
Sensitivity of individual probes    18/18 (100%) 6/10 (60%) 2/2 (100%) 16/17 (94%) 

 
    Number of ESBL substitutions detected/number of expected ESBL substitutions 

 
SHV ESBL genes in collection ESBL-associated substitutions detectable by array Number of isolates ESBL-positive result in array  G238S G238A E240K 

 
SHV-2, -2a, -3 G238S 11 11  11/11 — — 
SHV-4, -5, -12 G238S, E240K 22 22a  21/22a — 22/22 
SHV-18 G238A, E240K  — 1/1 1/1 
SHV-57 not detectable by array  — — — 
Sensitivity of individual probes     32/33 (97%) 1/1 (100%) 23/23 (100%) 

 
    Number of specific CTX-M groups detected/number of expected CTX-M groups 

 
CTX-M genes in collection CTX-M group Number of isolates ESBL-positive result in array CTX-M-1 CTX-M-2 CTX-M-9 CTX-M 8/25 

 
1, 3, 10, 15, 28 CTX-M-1 21 21/21 21/21 — — — 
2, 5 CTX-M-2 6/6 — 6/6 — — 
9, 14, 16, 27 CTX-M-9 16 16/16 — — 16/16 — 
8, 39 CTX-M-8/25 2/3b — — — 2/3b 
Sensitivity of individual probes    21/21 (100%) 6/6 (100%) 16/16 (100%) 2/3 (67%) 
    Number of ESBL substitutions detected/number of expected ESBL substitutions
 
TEM ESBL genes in collection ESBL-associated substitutions detectable by array Number of isolates ESBL-positive result in array E104K R164S R164H G238S 
TEM-18 E104K 1/1 — — — 
TEM-9, -63 E104K, R164S 3a 3/3 1/3a — — 
TEM-6, -43 E104K, R164H 2/2 — 2/2 — 
TEM-3, -4, -52 E104K, G238S 11 11 11/11 — — 11/11 
TEM-8 E104K, R164S, G238S 1/1 1/1 — 1/1 
TEM-5, -7, -10, -12 R164S 4b — 4/6b — — 
TEM-19, -20, -72 G238S 4b — — — 4/5b 
Sensitivity of individual probes    18/18 (100%) 6/10 (60%) 2/2 (100%) 16/17 (94%) 

 
    Number of ESBL substitutions detected/number of expected ESBL substitutions 

 
SHV ESBL genes in collection ESBL-associated substitutions detectable by array Number of isolates ESBL-positive result in array  G238S G238A E240K 

 
SHV-2, -2a, -3 G238S 11 11  11/11 — — 
SHV-4, -5, -12 G238S, E240K 22 22a  21/22a — 22/22 
SHV-18 G238A, E240K  — 1/1 1/1 
SHV-57 not detectable by array  — — — 
Sensitivity of individual probes     32/33 (97%) 1/1 (100%) 23/23 (100%) 

 
    Number of specific CTX-M groups detected/number of expected CTX-M groups 

 
CTX-M genes in collection CTX-M group Number of isolates ESBL-positive result in array CTX-M-1 CTX-M-2 CTX-M-9 CTX-M 8/25 

 
1, 3, 10, 15, 28 CTX-M-1 21 21/21 21/21 — — — 
2, 5 CTX-M-2 6/6 — 6/6 — — 
9, 14, 16, 27 CTX-M-9 16 16/16 — — 16/16 — 
8, 39 CTX-M-8/25 2/3b — — — 2/3b 
Sensitivity of individual probes    21/21 (100%) 6/6 (100%) 16/16 (100%) 2/3 (67%) 

aThe failure to detect the R164S substitution in two TEM-9-harbouring isolates did not result in an ESBL-negative test result due to detection of the E104K substitution in these isolates. Similarly, the failure to detect the G238S substitution in an SHV-12-producing isolate did not result in an ESBL-negative test result due to identification of the E240K substitution.

bThe R164S mutation was not demonstrated in one TEM-5-producing isolate and one TEM-7-producing isolate. The G238S substitution was not detected in a TEM-72-positive isolate resulting in a false ESBL-negative array result.

The ESBL-negative isolates included Enterobacteriaceae with plasmid-mediated and chromosomal AmpC production, OXA β-lactamases, non-ESBL TEM or SHV variants, ampicillin-susceptible Escherichia coli isolates and seven Klebsiella oxytoca K1 hyperproducers.

Phenotypic characterization

To compare the results of the microarray system with the ESBL phenotype, the Phoenix automated system with the NMIC-ID75 card (Becton Dickinson Diagnostic Systems, Baltimore, MD, USA) was used to determine MICs. Confirmatory tests for ESBL production were performed if the ESBL MIC of ceftriaxone or ceftazidime was >1 mg/L. As confirmation, three ESBL Etests (AB Biodisk, Solna, Sweden) with ceftazidime, cefotaxime and cefepime, each ±clavulanic acid, were used.

DNA isolation and sequencing

DNA isolation was performed using Nucleospin Tissue Columns (Macherey-Nagel, Düren, Germany) according to the instructions of the manufacturer. The presence of TEM, SHV and CTX-M was determined using PCR and sequencing of the same batch of DNA as used in the microarray [references in Table S5, available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/)].

Statistics

Test characteristics (sensitivity and specificity) are presented with 95% confidence intervals (CIs).

Results

Test characteristics of the microarray

Using sequencing of the ESBL gene and the corresponding phenotype in the Lahey database as the reference test, the sensitivity of the microarray was 101/106 (95%; 95% CI 89%–98%) and the specificity was 100% (95% CI 97%–100%).

A false-negative result was obtained in five ESBL-positive isolates. One SHV-57-positive isolate was not detected, because SHV-57 lacks the ESBL-associated SNPs detected in the array system. In one TEM-5 and one TEM-7 isolate, the R164S mutation was not detected, and in one TEM-72 isolate the G238S mutation was not detected. Finally, a CTX-M-39 gene was not detected in one isolate (Table 1). In the isolates with a false-negative result, the array test was repeated twice, including a new DNA extraction. Since PCR and sequencing were also repeated (with a positive result) on the same DNA batches, plasmid instability was not a likely explanation for these negative results.

Accuracy of individual probes in identifying specific SNPs

In total, 8 of the 110 ESBL genes in the isolate collection were not, or incompletely, detected. For five isolates this resulted in a false-negative result, while for three isolates (two TEM-9-positive isolates and one SHV-12-positive isolate) the microarray result was not influenced because another ESBL substitution was detected (Table 1).

The specificity of the probes was high. For only two isolates were spots detected not in accordance with the sequencing results. An additional TEM E104K spot was detected in a TEM-19 isolate and a TEM G238S spot in a TEM-18 isolate.

Test characteristics of the phenotypic assay

The sensitivity of phenotypic ESBL testing was 104/106 (98%; 95% CI 93%–99%) and the specificity 98/106 (92%; 95% CI 86%–96%).

In one CTX-M-9-producing Enterobacter cloacae isolate, all three ESBL Etests were non-determinable, and in one E. coli isolate, containing an SHV-2 ESBL and an ACC-1 plasmid-mediated AmpC, the ESBL Etests were negative. Using the microarray system, CTX-M-9 and SHV-2 were correctly detected in these isolates.

As expected, the seven K1-hyperproducing K. oxytoca isolates were ESBL positive in the phenotypic tests, but negative in the microarray system. One E. coli isolate displayed positive ESBL Etests, although no TEM, SHV or CTX-M genes were detected by either PCR or the microarray system.

Discussion

This study describes the first commercially available microarray system enabling the detection of ESBLs belonging to the three most prevalent ESBL families: TEM, SHV and CTX-M. Analysing 212 well-characterized Enterobacteriaceae isolates, this test had a sensitivity of 95% and a specificity of 100%.

A microarray has advantages in comparison with phenotypic testing for ESBL production. First, the results could be obtained within the same working day, whereas phenotypic confirmatory tests require an overnight incubation. Secondly, the microarray is accurate for species producing β-lactamases that may interfere with phenotypic tests for ESBL production, such as K1-hyperproducing K. oxytoca and AmpC-producing isolates. Thirdly, the microarray identifies ESBL families and provides information on the SNPs in the TEM and SHV ESBL gene(s), which may be useful for infection control purposes.

The assay also has some disadvantages. An intrinsic limitation of this type of genotypic assay is the limited number of ESBLs (or ESBL families) that may be detected. The array was designed to include the most prevalent ESBLs, resulting in coverage of 95% (84/88) of the TEM ESBLs and 77% (27/35) of the SHV ESBLs described in the Lahey database. Recently another microarray (not commercially available) was described enabling detection of 99% of the TEM ESBLs and 94% of the SHV ESBLs in the Lahey database.8 The flexibility of the microarray system presented here, however, allows easy extension of the assay with additional probes.

Another issue is the potential sensitivity for cross-contamination from previous amplification products. To prevent this, the use of two separate rooms proved to be imperative (one room for DNA isolation and ligation, and one for amplification, hybridization and detection).

Another limitation of this study was the fact that 4 of the 24 probes in the microarray system were not evaluated, because isolates with the target mutations were not available. However, these probes have been tested successfully by the manufacturer on artificial DNA sequences (data not shown).

In conclusion, this microarray accurately detects and identifies the three most prevalent ESBL gene families in Enterobacteriaceae. The simple and straightforward protocol makes this system a promising tool for detection of ESBLs in a clinical microbiology laboratory.

Funding

This work was supported by the University Medical Centre Utrecht, The Netherlands, Check-Points B.V., Wageningen, The Netherlands, and by SenterNovem, an agency of the Dutch Ministry of Economic Affairs.

Transparency declarations

A. K. is an employee of Check-Points B.V., Wageningen, The Netherlands. P. V. is an employee/shareholder of Check-Points B.V., Wageningen, The Netherlands. All other authors: none to declare.

Supplementary data

Figures S1 and S2 and Tables S1, S2, S3, S4 and S5 are available as Supplementary data at JAC Online (http://jac.oxfordjournals.org/).

Acknowledgements

We acknowledge Neil Woodford for his comments on the manuscript prior to submission.

References

1
Bush
K
Extended-spectrum β-lactamases in North America, 1987–2006
Clin Microbiol Infect
 , 
2008
, vol. 
14
 
Suppl 1
(pg. 
134
-
43
)
[PubMed]
2
Canton
R
Novais
A
Valverde
A
, et al.  . 
Prevalence and spread of extended-spectrum β-lactamase-producing Enterobacteriaceae in Europe
Clin Microbiol Infect
 , 
2008
, vol. 
14
 
Suppl 1
(pg. 
144
-
53
)
[PubMed]
3
Pitout
JD
Laupland
KB
Extended-spectrum β-lactamase-producing Enterobacteriaceae: an emerging public-health concern
Lancet Infect Dis
 , 
2008
, vol. 
8
 (pg. 
159
-
66
)
4
Bradford
PA
Extended-spectrum β-lactamases in the 21st century: characterization, epidemiology, and detection of this important resistance threat
Clin Microbiol Rev
 , 
2001
, vol. 
14
 (pg. 
933
-
51
)
5
Ikryannikova
LN
Shitikov
EA
Zhivankova
DG
, et al.  . 
A MALDI TOF MS-based minisequencing method for rapid detection of TEM-type extended-spectrum β-lactamases in clinical strains of Enterobacteriaceae
J Microbiol Methods
 , 
2008
, vol. 
75
 (pg. 
385
-
91
)
6
Knox
JR
Extended-spectrum and inhibitor-resistant TEM-type β-lactamases: mutations, specificity, and three-dimensional structure
Antimicrob Agents Chemother
 , 
1995
, vol. 
39
 (pg. 
2593
-
601
)
[PubMed]
7
Bonnet
R
Growing group of extended-spectrum β-lactamases: the CTX-M enzymes
Antimicrob Agents Chemother
 , 
2004
, vol. 
48
 (pg. 
1
-
14
)
8
Leinberger
DM
Grimm
V
Rubtsova
M
, et al.  . 
Integrated detection of extended-spectrum-β-lactam resistance by DNA microarray-based genotyping of TEM, SHV, and CTX-M genes
J Clin Microbiol
 , 
2010
, vol. 
48
 (pg. 
460
-
71
)
9
Wattiau
P
Weijers
T
Andreoli
P
, et al.  . 
Evaluation of the Premi®Test Salmonella, a commercial low-density DNA microarray system intended for routine identification and typing of Salmonella enterica
Int J Food Microbiol
 , 
2008
, vol. 
123
 (pg. 
293
-
8
)

Supplementary data