Establishment of a Simple and Quick Method for Detecting Extended-Spectrum β-Lactamase (ESBL) Genes in Bacteria

Abstract

Extended-spectrum β-lactamase (ESBL) genes that render bacteria resistant to antibiotics are commonly detected using phenotype testing, which is time consuming and not sufficiently accurate. To establish a better method, we used phenotype testing to identify ESBL-positive bacterial strains and conducted PCR to screen for TEM (named after the patient Temoneira who provided the first sample), sulfhydryl reagent variable (SHV), cefotaxime (CTX)-M-1, and CTX-M-9, the 4 most common ESBL types and subtypes. We then performed multiplex PCR with 1 primer containing a biotin and hybridized the PCR products with gene-specific probes that were coupled with microbeads and coated with a specific fluorescence. The hybrids were linked to streptavidin-R-phycoerythrins (SA-PEs) and run through a flow cytometer, which sorted the fluorescently dyed microbeads and quantified the PEs. The results from single PCR, multiplex PCR, and cytometry were consistent with each other. We used this method to test 169 clinical specimens that had been determined for phenotypes and found 154 positive for genotypes, including 30 of the 45 samples that were negative for phenotypes. The CTX-M genotype tests alone, counting both positive and negative cases, showed 99.41% (168/169) consistency with the ESBL phenotype test. Thus, we have established a multiplex-PCR system as a simple and quick method that is high throughput and accurate for detecting 4 common ESBL types and subtypes.

INTRODUCTION

ESBLs are a family of enzymes capable of hydrolyzing penicillins, cephalosporins, and monobactam antibiotics, whereas the host bacteria would still be sensitive to cephamycins, carbapenem antibiotics, and inhibitors of the enzymes.^1–3 As ESBL genes are harbored by plasmids, but not by the bacterial genome, they can easily be taken by bacteria and thus, easily render other bacteria resistant to the antibiotics.⁴

The third generation of cephalosporins is abused worldwide, especially in China, in infected hospital patients. As an adverse consequence, acquired drug resistance has emerged to enterobacteriaceae, which has gradually undermined the efficacy of cephalosporin, causing widespread infections in hospitals, and has thus become an increasingly serious problem for public health.^{5, 6} There are many members in the ESBL gene family with different drug-resistant properties and distinctive epidemic patterns that result not only from only different micropathogens with different virulence but also from differing antibiotic uses across different countries and geographic areas.⁷ Nomenclature of β-lactamases is complicated and not entirely rational, as the enzymes have been named after a preferred substrate, biochemical properties, genes, bacteria, patients, hospitals, or states, as well as from the initials of their authors.^8–10 The most common ESBL genotypes include the TEM, SHV, CTX-M (active on CTX), and oxacillin (OXA; active on OXA).^10–13

Distribution of ESBL genes varies greatly among countries or geographic areas. In Europe, the CTX-M type has already taken the dominant seat from the TEM and SHV types,⁵ whereas in China, despite the disparities among reports from different areas,^{4, 13–15} the CTX-M is the most common type, followed by the SHV, whereas the TEM type is uncommon.¹⁴ For instance, the CTX-M-3 and CTX-M-14 subtypes are the most common in the Zhejiang Province and in Beijing.^{16, 17}

Currently, ESBLs are commonly determined indirectly using the phenotype testing in the hospital laboratory, which involves time-consuming and arduous bacterial culture and is not sufficiently accurate. On the other hand, direct determination of the ESBL genes in bacteria, especially with a quick and high-throughput approach, such as a flow cytometry, has not yet been a practical test in the hospital laboratory. Therefore, we have developed a quick, high-throughput, and more-accurate method that directly detects the 4 most common ESBL types and subtypes, i.e., the TEM, SHV, CTX-M-1, and CTX-M-9, in bacteria from clinical samples.

MATERIALS AND METHODS

Bacterial Sources

A total of 172 bacterial strains that had been determined for drug-resistant phenotypes in our clinical practice were collected, including 169 strains of Escherichia coli and 3 strains of Klebsiella pneumoniae (H36–H38). Among them, the H1–3, H5, H7–34, H39–141, and N1–11 strains of E. coli were isolated from clinical specimens cultured in our departments, whereas the E1–17 and D1–5 strains of E. coli were isolated from the clinical specimens preserved in our hospitals. The ATCC25922, a detection index strain of E. coli, recommended by the National Health Quality Control Standard of China, and another E. coli strain H35 were obtained from the Health Ministry of China in Beijing. This study was approved and abided by the Research Ethics Committee of the Baiyun Hospital affiliated to the Guizhou Medical University Hospital.

Single PCR

After drug-resistance tests, i.e., phenotype tests, we extracted plasmids from a positive strain, H1–52, as it contained the 4 most common ESBL genes, i.e., the TEM, SHV, CTX-M-1, and CTX-M-9. The plasmids were used as templates in single PCR to amplify fragments of these 4 genes. The primers, with their sequences listed in Table 1, were chosen according to published studies,¹⁸ were verified for their specificity using the Basic Local Alignment Search Tool (BLAST) approach, and were synthesized by Sangon Biotech (Shanghai, China). The reaction was carried out in a volume of 25 µl, containing 1 µl of each primer (from a 10 µM stock solution), 1 µl plasmid DNA as template, and 12.5 µl 2× Taq PCR MasterMix (KT201; Tiangen Biotech, Beijing, China), with H₂O to make up the volume. After denaturation at 94°C for 3 min, PCR was carried out in a Thermal Cycler (Life Express, vended by Hangzhou Bioer Technology, Hangzhou, China) with 30 cycles at 30 s at 94°C, 30 s at 55°C, and 60 s at 72°C, followed by a final elongation at 72°C for 5 min. PCR products were electrophoresed using 1× Tris-borate-EDTA buffer (89 mM Tris, pH 7.6, 89 nN boric acid, and 2 mM EDTA) in a 3% agarose gel containing 0.5 µg/ml ethidium bromide (EB) for ∼1 h, with slight adjustments among individual electrophoresis. The EB-visualized band was photographed. The PCR products were also sent to SinoGenoMax (Beijing, China) for purification and sequencing to confirm their identity.

TABLE 1.

Single PCR primer sequences

Genotype	Primer sequence,a 5′ → 3′	Tm, °C	Product size, bp
TEM	Forward: ATGAGTATTCAACATTTCCGTGTC	56.8	851
	Reverse: TTAATCAGTGAGGCACCTATCTC	58.4
SHV	Forward: TATATTCGCCTGTGTATTATCTCCC	54.2	856
	Reverse: GTTAGCGTTGCCAGTGCTCG	58.9
CTX-M-1	Forward: ATGGTTAAAAAATCACTGCGCCAG	58.6	869
	Reverse: CCGTCGGTGACGATTTTAGCCG	63.8
CTX-M-9	Forward: GGTGACAAAGAGAGTGCAACGG	61.9	868
	Reverse: CCCTTCGGCGATGATTCTCGC	63.9

Tm, Temperature. ^aBLAST against the Nucleotide Collection databases of the National Center for Biotechnology Information of the United States (NCBI) indicates that the sequences of all probes and primers are 100% homologous to the targeted region of the corresponding genes without significant similarity to other ESBL gene sequences.

Multiplex PCR

Positive plasmids confirmed by the results of the above single PCR and ensuing sequencing were used as the templates in multiplex PCR. Primers, as well as probes to be used in the ensuing hybridization for the aforementioned 4 ESBL genes, were designed according to a reported study,¹⁴ but their specificity was checked using BLAST against the nucleic acid collection of the NCBI. All primers and probes were confirmed to be specific among ESBL genes with 100% similarity to the corresponding genes (Table 2). Moreover, 5 or more consecutive nucleotides at the 5′ or 3′ end of the primers or probes had no homolog in the same gene and in any other ESBL gene; thus, the possibility is ruled out that these primers or probes may serve as another primer to amplify a second fragment of the same gene or to amplify another gene. For each gene, the 5′ end of 1 primer of the 2 primers was linked with biotin, whereas the 5′ end of each probe was coupled with an amino (NH₂) group.

TABLE 2.

Primer and nucleic acid probe sequences after adjustment

Genotype	Primer and probe sequence,a 5′ → 3′	Tm, °C	Product size, bp
TEM	Forward: Biotin-CGTGTCGCCCTTATTCCCT	66	253
	Reverse: GACCGAGTTGCTCTTGCC	63.7
	Probe: NH2-TTGCTCACCCAGAAACGCT	65.9
SHV	Forward: Biotin-TGTATTATCTCCCTGTTAGCCAC	60.8	205
	Reverse: CGCCGCAGAGCACTACTTT	65.5
	Probe: NH2-TGGGAAAGCGTTCATCGG	67.2
CTX-M-1	Forward: Biotin-ATGGTTAAAAAATCACTGCGCCAG	58.6	869
	Reverse: CCGTCGGTGACGATTTTAGCCG	63.8
	Probe: NH2-CACCCAGCCTCAACCTAA	65.3
CTX-M-9	Forward: Biotin-GGTGACAAAGAGAGTGCAACGG	61.9	868
	Reverse: CCCTTCGGCGATGATTCTCGC	63.9
	Probe: NH2-TACCCAGCCGCAACAGAA	67.5

^aBLAST against the Nucleotide Collection databases of the NCBI indicates that the sequences of all probes and primers are 100% homologous to the targeted region of the corresponding genes without significant similarity to other ESBL gene sequences. For each gene, an NH₂ group was conjugated to the 5′ end of the forward primer to allow it to form an amide bond with a microbead.

Primers and probes for the commonly reported ESBL genes in the TEM and SHV types were also synthesized according to the literature.^{7, 13–15} Within the TEM type, there were TEM-3, -10, -12, -26, -105, -116, -128, and -143 subtypes, and their corresponding GenBank accession numbers were HM063042.1, KC859395.1, M88143.1, L19940.1, AF516720.1, AJ847364.1, AY368237.1, and DQ075245.1, respectively. Within the SHV type, there were SHV-2, -2a, -5, -11, -12, -26, -28, and -59 subtypes, and their corresponding GenBank accession numbers were JX268750.1, AF462393.1, AF462394.1, DQ166780.3, EU418910.1, KF585139.1, EU441172.1, and AY790341.1, respectively.

Multiplex PCR was conducted in a volume of 25 µl, containing 2.5 µl plasmid DNA, a 10 µl mix of 4 pairs of primer, and 12.5 µl 2× Taq PCR MasterMix, with the same condition as for the above-described single PCR. PCR products were fractioned using electrophoresis in a 3% agarose gel containing EB, as described above. The EB-visualized band was photographed.

Microbead Hybridization

Before being used in multiplex PCR, each gene’s probes were cross-linked with a selected type of microbeads via a covalent binding between the NH₂ group of a probe and the COOH group of a bead, which was conducted by the microbead suppliers (Shanghai Tellgen Life Science, Shanghai, China). For each gene’s probes, the microbeads were coded with a different fluorescein using a relatively new technique under a U.S. patent (Patent No. US6,632,526B1). This technique of fluorescently dyed particles allowed polystyrene latex beads of ∼5.6 μm in diameter to be dyed with different collocations of 2 types of fluorescein, each at a 10-fold dilution, which could produce up to 100 different kinds of microbeads, each kind with a distinct fluorescence. Each kind of microbeads could match 1 gene, and therefore, up to 100 genes could be analyzed simultaneously with, theoretically, each gene read for 100 times by a flow cytometer in a single test. The median fluorescence intensity (MFI) of these 100 reads in a single run could be reported as the final result, thus significantly lowering, if not eliminating, false positivity.

The beads were diluted with 1.5× tetramethylammonium chloride (TMAC) hybridization buffer (4.5 M TMAC, 0.15% sarkosyl, 75 mM Tris-HCl, and 6 mM EDTA, pH 8.0)¹⁹ to achieve ∼150 beads of each florescence/µl, conducted by Shanghai Tellgen Life Science and kept at 4°C in the dark until use. As single-PCR results showed that the H1, H30, H33, H36, and H37 strains contained multiple genotypes and were thus better suited for testing our method, the multiplex-PCR products of these strains, 3 μl each, were mixed with 22 μl TMAC hybridization solution. After denaturation at 95°C for 5 min, the hybridization was carried out at 48°C for 15 min.

SA-PE Solution

SA-PE (purchased from Shanghai Tellgen Life Science) was diluted to a concentration of 2–4 μg/ml with 1× TMAC hybridization solution. A volume of 75 μl SA-PE solution was then added into the above-described hybridization system, followed by incubation at 48°C for another 15 min to allow the biotin in the hybrids to bind to the avidin in the SA-PE to form an avidin-biotin complex (ABC).

Flow Cytometry

The SA-PE-coupled hybrids were run through a Luminex 200 multifunctional flow lattice cytometer (Luminex, Austin, TX, USA), which used a green laser to sort out the beads by their sizes and then used a red laser to quantify the intensity of PE in the sorted hybrids.

Statistical Analyses

The SPSS 18.0 software was used for data analyses. Pair-match design of 4-table χ² test was used to evaluate the reliability of each experiment, whereas the Kappa coefficients were used to evaluate the consistency. P < 0.05 was considered statistically significant.

RESULTS

Screening for the ESBL-Positive Strains

Single PCR was conducted, and the resulting products were sequenced. The resulting sequences were aligned against the GenBank database using the BLAST approach, which resulted in >99% of sequence consistency. The sequencing results showed that H1 had the TEM-1 and CTX-M-14 genes; H30 had the TEM-1, CTX-M-55, and CTX-M-14 genes; and H33 had the CTX-M-15 gene, whereas H36 (a strain of K. pneumoniae) had the SHV-27 gene. As each of our E. coli. strains contained several ESBL genes, they, together with H37 (a strain of K. pneumoniae) that had the SHV-28 gene, were chosen for further study.

Establishment of a Multiplex PCR System

After multiplex PCR for each strain of bacteria was conducted, the PCR product of each strain was hybridized with the corresponding bead-coupled probes, followed by incubation with SA-PE to allow ABC formation. The ABC-coupled hybrids were then run through a flow cytometer, which gave rise to a much stronger MFI (>150) compared with the blank control group (Table 3). An MFI ≥ 150 was used as the cutoff criterion, as recommended by Luminex, the cytometer manufacturer. These results were consistent with the single-PCR products visualized in agarose gel. Hence, by combining a multiplex PCR, a microbead hybridization, an ABC approach, and flow cytometry, we established a method coined as a multiplex-PCR system that could easily detect the 4 most common ESBL types or subtypes. Figure 1 summarizes the procedure of our method.

TABLE 3.

Hybridizing results of multiplex PCR products

Sample	CTX-M-1-P	CTX-M-9-P	SHV-PA1	TEM-P1	Total events
H1	2	5076	20	685	2347
H30	1463	4867	26.5	788	2553
H33	1420	111	9	24	2668
H36	1653	58	2060	47	1969
H37	1418	85	1309	1184	2357
Blank	13	134.5	9	36.5	2374

Florescent signal was expressed as MFI. CTX-M-55 and CTX-M-15 belong to CTX-M-1-P; CTX-M-14 belongs to CTX-M-9-P; SHV-27 belongs to SHV-PA1; TEM-1 belongs to TEM-P1; “Total events,” total bead number of each sample.

FIGURE 1.

A flow chart of our multiplex-PCR system. This system uses 4 pairs of PCR primers, in addition to 4 probes specific for each gene. Each microbead contains a COOH group and is coated with a distinct fluorescence. For each gene, 1 primer is labeled with a biotin, and the probe contains an NH₂ group so that it can complex with a microbead via an amide bond before use. After multiplex PCR, PCR products are hybridized with SA-PE to allow formation of ABC in the hybrids. The hybrids will then be run through a flow cytometer, which uses a green laser to sort out the microbeads, according to their different types of fluorescence, and then uses a red laser to quantify the PE in the ABC.

Evaluation of our Method with Clinical Samples

We verified our method by examining 169 clinically derived strains of E. coli, with 3 strains of K. pneumoniae as positive controls and the E. coli strain ATCC25922 as a negative control. Our phenotype tests showed that 123 of these 169 strains were positive for CTX-M, and that 81 of these 123 were also positive for TEM, with 1 additional strain showing TEM positivity but CTX-M negativity (Table 4). None of the 169 specimens was positive for an SHV phenotype.

TABLE 4.

Statistical comparison between ESBL genotype and phenotype tests of the 169 E. coli strains that we studied

Genotype	ESBL phenotype		Total	Kappa	P
	+	−
CTX-M
+	123	0	123	0.985	1.000
−	1	45	46
Total	124	45	169
TEM
+	82	30	112	0.005	0.195
−	42	15	57
Total	124	45	169

As detailed in Table 5, our genotype tests detected 154 (91.12%) of the 169 strains as positive for ESBL genotypes, significantly >124 (73.37%) that showed positive for ESBL phenotypes. On the other hand, of the 45 strains that were negative for ESBL phenotypes, 30 were positive for the TEM genotype. The results from the 2 CTX-M genotype tests alone, counting both positive and negative cases, showed 99.41% (168/169) consistency with the traditional ESBL phenotype test. The χ² test of 4-table (Table 4) showed that our CTX-M genotype test, but not our TEM genotype test, was consistent with the ESBL phenotype test. Similar to the phenotype tests, our genotype tests did not detect any sulfhydryl reagent variable (SHV) positivity in the 169 strains either.

TABLE 5.

Details of the 154 strains positive for genotype tests

Genotype (type or subtype)	n	%
TEM	31	20.13
CTX-M-1	24	15.58
CTX-M-9	16	10.39
CTX-M-9 + TEM	42	27.27
CTX-M-1 + TEM	25	16.23
CTX-M-1 + CTX-M-9 + TEM	13	8.44
CTX-M-1 + CTX-M-9	3	1.95
Total	154	99.99

Of the 154 strains positive for ESBL genotypes, 71 were single-gene types, including 24 (15.58%) with the CTX-M-1 gene, 16 (10.39%) with the CTX-M-9 gene, and 31 (20.13%) positive for TEM (Table 5). The remaining 83 strains belonged to compound-gene types, including 42 (27.27%) with the CTX-M-9 and TEM genes; 25 (16.23%) with the CTX-M-1 and TEM genes; 13 (8.44%) with the CTX-M-1, CTX-M-9, and TEM genes; and 3 (1.95%) with the CTX-M-1 and CTX-M-9 genes (Table 5). These results showed that the CTX-M and TEM types concomitantly existed in the same strains at a high frequency.

As a strong signal from the cytometer indicated the existence of certain genes, we randomly selected multiplex-PCR products, together with products from some control strains, for electrophoresis to visualize and thus verify the results. All of the expected, specific PCR products were observed (Fig. 2). Positive strains, including the TEM type (H25, H71, H116, N2, N4, and N11), SHV type (H38), CTX-M-1 subtype (H17, H56, and H110), and CTX-M-9 subtype (H10, H47, and H92), were also randomly selected to run single PCR again. The PCR products were sequenced, and the resulting sequences were aligned against the NCBI database using BLAST, which showed that the sequences had >99% homology to the corresponding genes, thus endorsing the multiplex-PCR results. Hence, our sequencing data and gel images of single PCR and multiplex PCR, as well as our data of flow cytometry, were highly consistent with one another, confirming the specificity of the new method that we developed.

FIGURE 2.

Electrophoresis results of multiplex PCR products of clinical bacterial strains visualized in an agarose gel. The expected molecular weights of the bands are described in Tables 1 and 2. M is the D2000 DNA marker, whereas N is the E. coli strain ATCC25922, included as a negative control.

DISCUSSION

It sometimes occurs that ESBL-positive bacterial strains are still sensitive to antibiotics at a certain level in culture but that clinical treatments of the infected patients with these antibiotics are ineffective. We suspect that this inconsistency may be, in part, because currently, determination of the existence of an ESBL in bacteria still relies on the phenotype testing, which is an indirect approach and is not only time consuming and laborious but also less accurate than a direct detection of the genes. Therefore, it is imperative to develop a simple, quick, high-throughput but accurate approach to replace the currently used phenotype test, as an effective treatment of infected patients depends on the correct determination of whether the resistant bacteria really bears an ESBL gene.

As aforementioned, ESBL-producing bacterial strains in China have been reported to be mainly the CTX-M type, which can be categorized further into the CTX-M-1, -2, -8, -9, and -25 subtypes.²⁰ The CTX-M-3 and CTX-M-14 subtypes are the most common and actually belong to the CTX-M-1 and CTX-M-9 subtypes, respectively.¹⁴ As all genes in the TEM type are very similar in the DNA sequence—true also for all genes in the SHV type^{6, 11}—we are unable to design specific primers and thus, are unable to use PCR to distinguish each subtype from the others in the TEM and SHV types. Therefore, for the TEM and SHV types, we can only design primers to amplify a genomic region shared by all genes in each of these types. However, interestingly, we did not detect the SHV genotype, which is consistent with the results from our phenotype tests. It is likely that the SHV type may be uncommon in the Guiyang district, which is located in the southwestern part of China, although this conclusion still awaits further confirmation by studying many more clinical samples.

Our results suggest that our method is not only highly sensitive and accurate but also simple, quick, and high throughput, as it uses fluorescent microbeads that can be detected easily using a flow cytometer. With a relatively new technique of fluorescently dyed particles, we are able to separate microbeads into 100 different types, with each type matched to 1 gene, which allows us to analyze simultaneously up to 100 genes with, theoretically, each gene read for 100 times in a single test. The reporting of the MFI of these 100 reads in a single run as the collective result should significantly lower false positivity. Also significant, our method is quick and semi-high throughput compared with the traditional phenotype test, as a batch of 96 samples can be tested in a single run in only 6 h. In contrast, in the disk-diffusion test, currently used in our clinical practice, the so-called phenotypic confirmatory test of ESBLs, the bacterial culture for a drug-resistance test, takes 16–20 h.

In our clinical trial, the CTX-M genotype analysis, using our new method, has shown a result of 99.41% (168/169) consistency with the traditional ESBL phenotype test. As for the less-reliable results of the TEM genotype tests, we infer that the TEM genes may not be the main cause for the development of the drug resistance, which is supported by our sequence analyses. The strains positive for the TEM genotypes are all the CTX-M genotypes with only 1 exception, and this exception is positive in the TEM genotype and the ESBL phenotype. Therefore, the drug resistance of this strain may be a result of other genes not included in our multiplex PCR, which suggests that there is still room for us to improve our approach further in the future for less-common genotypes.