Development of a Serotype-Specific DNA Microarray for Identification of Some Shigella and Pathogenic Escherichia coli Strains

Abstract

Shigella and pathogenic Escherichia coli are major causes of human infectious diseases and are responsible for millions of cases of diarrhea worldwide every year. A convenient and rapid method to identify highly pathogenic serotypes of Shigella and E. coli is needed for large-scale epidemiologic study, timely clinical diagnosis, and reliable quarantine of the pathogens. In this study, a DNA microarray targeting O-serotype-specific genes was developed to detect 15 serotypes of Shigella and E. coli, including Shigella sonnei; Shigella flexneri type 2a; Shigella boydii types 7, 9, 13, 16, and 18; Shigella dysenteriae types 4, 8, and 10; and E. coli O55, O111, O114, O128, and O157. The microarray was tested against 186 representative strains of all Shigella and E. coli O serotypes, 38 clinical isolates, and 9 strains of other bacterial species that are commonly present in stool samples and was shown to be specific and reproducible. The detection sensitivity was 50 ng genomic DNA or 10⁴ CFU per ml in mock stool specimens. This is the first report of a microarray for serotyping Shigella and pathogenic E. coli. The method has a number of advantages over traditional bacterial culture and antiserum agglutination methods and is promising for applications in basic microbiological research, clinical diagnosis, food safety, and epidemiological surveillance.

Shigella and pathogenic Escherichia coli are the major causative agents of diarrhea. E. coli clones are normally classified by a combination of oligosaccharide (O), flagellar (H), and capsular (K) antigens. Shigella clones are classified by the O antigen only, as they lack H and K antigens (14). There are four named species of Shigella classified on the basis of biochemical and O-antigen serological differences: Shigella dysenteriae (consisting of 13 serotypes), Shigella flexneri (consisting of 14 serotypes [including subtypes]), Shigella boydii (consisting of 18 serotypes), and Shigella sonnei (27). Of the estimated 165 million cases of Shigella diarrhea per year, 69% of the episodes occurred in children under 5 years of age, and 1.1 million deaths were attributed to Shigella infections (24). In the United States, most infections are caused by S. sonnei; S. flexneri is the second most common serotype (4, 24).

The E. coli species consist of various serotypes, ranging from highly pathogenic to nonpathogenic strains, which are from normal intestinal flora and are often used as safe laboratory strains (32). There are five major pathotypes of E. coli strains that cause diarrhea in humans: enteropathogenic, enterotoxigenic, enteroinvasive, enteroaggregative, and enterohemorrhagic (28). These pathotypes consist of genetic clones that often correspond to distinct O:H serotypes (15). Serotypes O55, O111, O114, and O128 belong to the group of enteropathogenic E. coli-associated O serotypes (23). E. coli O157:H7 is one of the enterohemorrhagic E. coli strains, and it causes significant illness and represents a serious public health threat worldwide (5).

The O antigen, which consists of repeats of the O unit, is part of the lipopolysaccharide in the outer membrane of gram-negative bacteria and contributes major antigenic variability to the cell surface. In Shigella and E. coli, genes involved in the biosynthesis of the O antigen are normally clustered in the chromosome between two housekeeping genes, galF and gnd, and are classified into three main classes: the nucleotide sugar biosynthesis pathway genes, glycosyltransferase genes, and O-unit processing genes encoding flippase and polymerase (wzx and wzy) (14). Glycosyltransferase genes encode enzymes for the transfer of sugars to build the O unit (12). The role of Wzx is to translocate, or flip, the O units formed at the cytoplasmic face of the inner membrane to the periplasmic face. The O units are then polymerized by Wzy to form a long-chain O antigen at the periplasmic face of the membrane (19). The diverse forms of O antigen are almost entirely due to genetic variation in the O-antigen gene cluster (29). Due to the relatively low similarity of glycosyltransferase, wzx, and wzy genes among different serotypes, their sequences are normally highly specific to individual O antigens (7, 8, 11, 14, 30).

The highly variable nature of the O antigen provides the basis for serotyping, and more than 160 different serotypes (1, 8, 28, 33, 35) have been recognized in E. coli. Traditional serotyping requires the use of a large panel of antisera; moreover, it is subjective and cross-reactive (16). In recent years, PCR assays based on O-serotype-specific genes have been proposed by us and others for molecular typing of many Shigella and E. coli O serotypes (2, 3, 7, 8, 10, 11, 12, 15, 18, 31, 34, 36). Molecular typing has many advantages over the traditional method. However, it is difficult to quantify PCR products and to differentiate bands of similar size in a multiplex PCR mixture.

The oligonucleotide-based microarray assay is an efficient approach for parallel analyses of a large number of specific sequences (37). In this study, we developed a DNA microarray based on the target genes wzx, wzy, and wfaU (encoding glycosyltransferase) of the 15 Shigella and E. coli serotypes: S. sonnei; S. flexneri type 2a; S. boydii types 7, 9, 13, 16, and 18; S. dysenteriae types 4, 8, and 10; and E. coli O55, O111, O114, O128, and O157. The DNA microarray method described in this communication is specific, sensitive, and reliable and serves as a prototype for an array of all serotypes of Shigella and E. coli in our future work.

MATERIALS AND METHODS

Bacterial strains.

The strains used in this study include 186 representative strains of all Shigella and E. coli O serotypes (15), 38 clinical isolates, and 9 strains of Salmonella enterica, Staphylococcus aureus, Bacillus cereus, and Vibrio cholerae (Table 1). Serotypes of the 38 clinical isolates were identified using commercial antisera from the Chengdu Institute of Biological Products, China (data not shown).

TABLE 1.

Shigella and E. coli clinical isolates and strains of other bacterial species

Bacterium	Serotype	No. of strains of each source	Total no.
Shigella and E. coli clinical isolates used for blind testing of the microarray (n = 38)
S. sonnei		1a, 2b	3
S. flexneri	Type 2a	1a, 1b	2
S. boydii	Type 7	1a, 1b	2
	Type 9	1a, 1b	2
	Type 13	1a, 1b	2
	Type 16	1a, 1b	2
	Type 18	1a, 1b	2
S. dysenteriae	Type 4	1a, 1b	2
	Type 8	1a, 1b	2
	Type 10	1a, 1b, 1c	3
E. coli	O55	1a, 1c	2
	O111	1a	1
	O114	1a, 7d	8
	O128	1a	1
	O157	1a, 3d	4
Other bacterial species used to test the specificity of the probes (n = 9)
Salmonella enterica		3a	3
Staphylococcus aureus		1c, 1e	2
Bacillus cereus		2e	2
Vibrio cholerae		2a	2

^aInstitute of Medical and Veterinary Science, Adelaide, Australia.

^bInstitute of Epidemiology and Microbiology, Chinese Academy of Preventive Medicine, Beijing, China.

^cNational Center for Medical Culture Collection, China.

^dRobert Koch-Institut, Germany.

^eInstitute of Microbiology, Chinese Academy of Sciences, China.

Genomic-DNA extraction.

Genomic DNA was extracted from 1.5 ml of overnight broth culture (approximately 10⁹ CFU) using a DNA extraction kit (Tiangen, Beijing, China). The mock stool specimens were prepared as follows. A serial dilution of bacterial culture of each of the 15 serotype strains in the range of 10¹ to 10⁶ CFU per ml was prepared and mixed with 0.3 g of stool specimens from adult volunteers. DNA was extracted with a QIAamp Mini Stool Kit (QIAGEN GmbH, Hilden, Germany). For each strain, at least 10 extractions were repeated to verify the reproducibility of the DNA extraction method.

Identification of O-serotype-specific genes by PCR.

Genomic DNAs were prepared from 186 representative strains of all Shigella and E. coli O serotypes and examined for quality by PCR amplification of the mdh gene coding for malate dehydrogenase, as described previously (30). A total of 13 pools of DNA were made, each containing DNAs from 12 to 19 strains (15). The pools were screened using primers based on the wzx and wzy genes of E. coli O128; S. sonnei; S. flexneri type 2a; S. boydii types 7 and 9; and S. dysenteriae types 4, 8, and 10 (Table 2). The PCR cycles used were as follows: 30 cycles of denaturation at 94°C for 2 min, annealing at 45°C for 30 s, and extension at 72°C for 1 min, with a final extension at 72°C for 5 min. An aliquot of 10 μl of the PCR product was examined on an agarose gel.

TABLE 2.

PCR specificity test for E. coli O128, S. sonnei, S. flexneri type 2a , S. boydii types 7 and 9, and S. dysenteriae types 4, 8, and 10

Serotype	Specific gene	GenBank accession no.	Forward primer sequences (5′-3′)	Reverse primer sequences (5′-3′)	Product size (bp)
E. coli O128	wzx	AY217096	TCTTGCTTATAGCCAGAATT	AATAAACCGACACCGAAA	1,353
			GTGAATCGCAACACTTAT	GCAAACGATAAAGGAGGC	1,035
	wzy		ATGATTTCTTACGGAGTGC	CTCTAACCTAATCCCTCCC	782
			ATTCTGGTATGCGGTGTT	ATAATTGCTGGGTACATC	460
S. sonnei	wzx	AF285971	ATTTCATTAACTCTGCTTGT	ACAACCGCTGCTGACCATT	967
			TTGGTCGGTTTAGATGTG	TCCCTACGAAATAGATGC	516
	wzy		TGAGGTTTCACGTTTCTC	AATAATCCCTAACTGAGCC	817
			CCACCGCAATTATAGTAGT	TACGTATAAAAACCACCGA	687
S. flexneri type 2a	wzx	AY900451	CACTTGTTGGGTATGCTGG	CCGGCAAACAGATTAGAAA	782
			CTGAAGGTTTCGGTGTTTA	ATTACTTACTGTCATCCAACC	584
	wzy		GTGGTGGAAGATTACTGGA	GCTCCAGAAGTGAGGTTAT	1,084
			GTGTCGGTGCGATTATCAG	AAGATAAGCAACATAGGAACT	418
S. boydii type 7	wzx	Laboratory stock	ATTGCTTCCCTATCTTAC	GAGAACTGAGGCTATTTG	685
			TGTGACGACATCCCTATTG	AGTGCTTTATTCAACGCC	528
	wzy		TCCTCGTAGGTCAACTCA	AATACAATCCTGCAACAG	384
			GGCTCTGCATTATCTGTAAC	ATAAACTTACGACCTGAC	365
S. boydii type 9	wzx	AF402315	TTTGTTGGAGGAATGTTGT	TAAACCTTCAGCGACTACA	804
			CTGTTTCTTCCATTATCTGC	TTGTAGATATTTGAGGGT	1,069
	wzy		GCGTTGGTTGGTGAAAGAG	TTCCCACAAATCAAACCA	877
			TAAGCCACGCTATGTTGA	CTCTTAATTGATTTCCCACA	736
S. dysenteriae type 4	wzx	Laboratory stock	AACGATTAGTTGGTTGACA	CAATAAATAGACACGCACC	1,058
			CGACTTTGGGAAATGTGGA	GAAGGGTGGAAAACTGGCC	505
	wzy		TGTATGCTGGTGGAGGACC	GAACCGTATAGCGGAAAAA	263
			TTATGTGGGATATTGCTTC	CATTTTAACCTTCCTTCAT	719
S. dysenteriae type 8	wzx	Laboratory stock	CTGGTATTTCAGTTGTCAC	CAGAAGCAGCGCCAACCG	1,096
			TGGGTAGTTGGGCAACG	AACCATTAATACTTGCGCC	869
	wzy		ATTGGCAACATTCTTTTTCC	CATTGATATAGTTAACACC	1,139
			TACCATGAGTTAAATTAT	GTTATTCCCTAAAGACAC	870
S. dysenteriae type 10	wzx	Laboratory stock	GGAGCATTGGTGGTGT	AGAACGGAAAGTTGGG	718
			TGGCTTGTTATCTGCAGTAT	CTTTTACCAAAACTGACGTG	728
	wzy		GACACTGAAAGACTGGCGTT	AAGAAGGTGTTCCAAGCGTA	623
			CGCTGTTTCTATATTAATTG	AATTGAAGTGACCAGATAAC	707

Primer design.

The sources of the O-antigen gene cluster sequences of the 15 serotypes are listed in Table 3. Based on the O-serotype-specific genes of the 15 serotypes, the 15 compatible primer pairs in a multiplex PCR were designed: 10 primer pairs for the 10 serotypes of Shigella and 5 primer pairs for the 5 serotypes of pathogenic E. coli (Table 3). There was also one primer pair for amplifying the 16S rRNA genes of Shigella and E. coli. All 16 primer pairs were contained in one multiplex PCR.

TABLE 3.

Multiplex PCR primers and oligonucleotide probes used in the study

Serotype	Target gene	GenBank accession no.	Primera sequences (5′-3′)	Product size (bp)	Probeb sequence(s) (5′-3′)
S. flexneri type 2a	wzx	AY900451	1F: CACTTGTTGGGTATGCTGG	782	1-1: GGGCAGTGTTTCCAAGGTTAAGTAACATCAAAGACTTTAA
			1R: CCGGCAAACAGATTAGAAA		1-2: CGGTTGGATGACAGTAAGTAATATCATAAGTCCGGTCATG
S. sonnei	wzy	AF285971	2F: TGAGGTTTCACGTTTCTC	817	2-1: GGCACTGGATTAGGTGTTGCAAATTATGTAAAGGCTA
			2R: AATAATCCCTAACTGAGCC		2-2: TGATTATCGTCGAGTTGAGTTAGTATTTATTGGGGTTGAT
					2-3: TGGTTCTTTTGGTGTCATTTTGCATTGGTAAAGATGAGCT
S. boydii type 7	wzx	Laboratory stock	3F: GTGTTGACTGCTGGATTTC	572	3-1: ATTCCACTTCCATCAATCATATCGTTGATTATTCCTGCTG
			3R: CGATATGATTGATGGAAGTG		3-2: CCCATTGATTTTGCATCAAATAAGTGGCTTTCTTAAAGGA
					3-3: CTTGATACTGATAATGGCGTTGAATAAAGCACTAGTTCCA
S. boydii type 13	wzy	AY369140	4F: AAAGATTGGTAGCGTCGG	847	4-1: TTATCTTGAGTTTAGTCGAATGCAACGAGTAGTCGCG
			4R: TGAAGCCCTGGTAAAGTGC		4-2: TTATTCTCATATTCGAATGTTACTTTTACCAGCGCAACTG
					4-3: GCTGCGGGGAATAGAAATAAGAATCGTCTGTATTACTTTA
S. boydii type 16	wfaU	DQ371800	5F: CCATACGGATAATGTTGAG	698	5-1: TTGTATGTAATCGCAAATACTCGCAGTACATGTGGA
			5R: TCTTTGTCTTCTCGGCTA		5-2: CCTGAGCGATTAGAACAACTGTATCTCGAGTCAAGA
					5-3: GACATGGGAGTGGAATGAAAACTAAAATTGCTGAGGCTTT
					5-4: GTGAGATTGTTGATTGCTTTAGCCGAGAAGACAAAGATAT
S. boydii type 18	wzx	AY948196	6F: CAGGGCACAAACTATCT	1,096	6-1: TTTGTCGTGTCTATTGGTATTGTATTCGGGCAATGG
			6R: TAATTCGGAATGTGCT		6-2: TGGAGTTCCATTATCCGAGTGGTATTTTGGTAACAATAAT
					6-3: TACTCATTAGGCTTGCTATTTACTGGGCTATTATCAGTTT
S. dysenteriae type 4	wzy	Laboratory stock	7F: TTTCTGCTTATTCATTATTG	946	7-1: TCGCATTGCTTGGTATTAGAGCTGGTAGTGTTAATT
			7R: ACTACCAGCTCTAATACC		7-2: GCAGCTTCTCTTTACATACGTGCATTATCAGAAACGGGAA
					7-3: ACTAGTCTAGATAAAATGGCTAGCTCAGACAATCTTTCTG
					7-4: TGGATATTTAATTCGGCTATTCTCTATAGCTAGTGAGCCT
S. dysenteriae type 8	wzy	Laboratory stock	8F: TTCCCTCTTGTTGTATTGA	1106	8-1: GAGGGTGTGTATGGTATGATTGATTACATATTGGAGGC
			8R: ACCTTTATCAATTGCCTCC		8-2: ACTTCAGTTTCTGGGAACGATAAATTCACACGTTTGC
					8-3: TTCCTAAATAATCATCCATTTACTGAGGGTGTGTATGGTA
					8-4: CGCCAATTTATTCTGTGCTATTATCGAAATTCACTTCAGT
S. dysenteriae type 10	wzy	Laboratory stock	9F: CGCTGTTTCTATATTAATTG	706	9-1: ATGGCACTGATAGTAGCGATAAAACTATTTTGATCGGG
			9R: AATTGAAGTGACCAGATAAC		9-2: CGTGCATTAATTGCTATTGTCGTAATTTCTTTCATTGTGG
					9-3: GGAACACCTTCTTGGGATTATTTTACGCAACCACTTATTA
E. coli O55	wzy	AF461121	10F: GGGAGGAGTATTATCATTAC	917	10-1: AGAGTGAGACGAATAATTGGGTGTTATATAAGCTCTC
			10R: ATCAATCTAAAGGTCGGTA		10-2: GCTTTTGGGGATAACATTATCGATAATACCGACCTTTAGA
					10-3: TGAAGCTTTAAAACGTCTAATTATTAGCGGAACTTTCTTG
E. coli O111	wzy	AF078736	11F: TTAATGCGGAGGATCTATT	934	11-1: CGGGGATGATATATTATTTGGTTTCACAGCTTGGTG
			11R: GTAAGCCCGCAAATCAATC		11-2: TTACGGTTCTTTATAAGTATTGGTGTGATAGGAGCATTGG
					11-3: GCTCCTTTCATTGTTGTAAGTTGTTTGTTACTGTTACA
					11-4: TTAAATAACGGCGGACAATATAAGACGTTATATGGACTTC
E. coli O114	wzy	AF573377	12F: ACTTTCCCAAGCCCATTA	852	12-1: TTATTGTATGCTTGTTAGTGCTTGTGCTGATTTGTTTTCT
			12R: CAGCACAAGCACTAACAAG		12-2: TGAGATGCTTAAATTAGGTGGATGGAATGTTAATGGG
					12-3: GTTGTTCATATGCTCAGGGGAAAATCAGAAGAGTATCTAT
					12-4: TGGATGGAATGTTAATGGGTTATTTATTTCAGAAGCATG
E. coli O128	wzy	AY217096	13F: ATGATTTCTTACGGAGTGC	782	13-1: TCTGATCTTGGATTAAGTAAGATGTACCCAGCAA
			13R: CTCTAACCTAATCCCTCCC		13-2: CGGTGTTTTGCAAGAGATATAAAAGAGTTAGCTTTAGCAT
					13-3: GCTAGGTATTTAGCAAATTCAACAGATTTGGCTGACTTTG
E. coli O157	wzy	AF061251	14F: TTGCTGCTGTAGTTTTATTTCTT	555	14-1: CGATTTCTTTCCGACACCAGAGTTAGAAAAGGAATT
			14R: TGATGCTTTATTCCCTGTATTCT		14-2: TAGAGCAAGTTGAAAGTGTTCCATATGTTGTTTCTGAATC
					14-3: GTATGCTCGTTGTTTTATCTAAGTTTAGGACAAGACGGAG
S. boydii type 9	wzy	AF402315	15F: GCGTTGGTTGGTGAAAGAG	877	15-1: AACTGAGTTCACTTATGGTTCGAGAACCTTTACTCCATTT
			15R: TTCCCACAAATCAAACCA		15-2: GGATTTTAATACAACTGAGTTCACTTATGGTTCGAGAACC
	16S rRNA gene	AE000406	16F: GACGGGTGAGTAATGTCTGG	1,251	16-1: CGGGAACTCAAAGGAGACTGCCAGTGATAA
			16R: ATCCACGATTACTAGCGATTCC		16-2: CGGGAACTCAAAGGAGACTGCCAGTGATAAACTGGAG
					17: TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT
					18: TTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTTT_Cy3

^aF, forward primer; R, reverse primer.

^b16-1 and 16-2, positive control probes; 17, negative control probe; 18, positional reference and printing control probe.

Multiplex PCR and labeling of the target genes.

Each multiplex PCR amplification was performed with 30 μl of reaction mixture consisting of 50 to 100 ng of DNA; 1× PCR buffer (50 mM KCl, 10 mM Tris-HCl [pH 8.3]), 0.5 mM MgCl₂, 100 μM concentration of deoxynucleoside triphosphate, 1.5 U Taq DNA polymerase, 0.047 μM each of two primers based on the 16S rRNA gene sequence, 0.14 μM each of primers based on each target gene, and 0.15 nM cyanine dye Cy3-dUTP (Amersham Biosciences UK Ltd., Little Chalfont, England). The reaction parameters were 94°C for 5 min; 35 cycles of 94°C for 30 s, 50°C for 1 min, and 72°C for 1 min; and a final extension at 72°C for 5 min. An aliquot of 2 μl of PCR product was run on an agarose gel to check the amplified DNA, and the rest was stored at −20°C in the dark.

Oligonucleotide probe design.

For each serotype, two to four probes were designed with OligoArray 2.0 based on GenBank and an in-house database of all 33 of the O-antigen gene clusters of Shigella and 175 O-antigen gene clusters of E. coli. Two probes based on the 16S rRNA genes of all known Shigella and E. coli strains were designed as positive controls. A probe containing 40 poly(T) oligonucleotides was used as the negative control. A probe labeled with 3′ Cy3 was used as the positional reference and printing control. Each probe was 5′ amino modified. All of the oligonucleotide probes are listed in Table 3.

DNA array preparation.

The probes were dissolved in 50% dimethyl sulfoxide at a final concentration of 1 μg/μl and printed onto aldehyde group-modified glass slides (CEL Corporation) using SpotArray72 (Perkin-Elmer Corporation). Each probe was spotted in triplicate. The printed slides were dried for 24 h at room temperature, cross-linked by UV cross-linker (UVP Corporation), and stored at room temperature in the dark. Each slide consisted of four microarrays framed with a 12-μl Geneframe (Beijing Capital Biochip Corporation, Beijing, China), which constituted individual reaction chambers. One of the four microarrays was tested with the positive control standard, 100 ng/μl S. dysenteriae type 8 genomic DNA, to ensure that the reagents were effective. Another was tested with the negative control standard, sterile deionized water, to show that the reagents were uncontaminated. The other two were used to detect samples. A schematic diagram of the probe positions on the microarray is shown in Fig. 1.

FIG. 1.

Probe positions on the slide. Numbers 1-1 to 15-2 are the specific probes for the target strains. Numbers 16-1 and 16-2 are the positive control probes based on the 16S rRNA genes of all Shigella and E. coli strains. Number 17 is the negative control probe. Number 18 is the positional reference and printing control probe.

Hybridization process.

An aliquot of 15 μl of labeled PCR product was baked for about 1.5 h at 65°C until it was dry and diluted in 13 μl of hybridization buffer (25% formamide, 0.1% sodium dodecyl sulfate, 6× SSPE [1× SSPE is 0.18 M NaCl, 10 mM NaH₂PO₄, and 1 mM EDTA {pH 7.7}]). After denaturation at 98°C for 5 min, an aliquot of 12 μl of labeled target DNA was hybridized with the probes at 40°C for 16 h. After hybridization, the Geneframe was removed and the slide was washed with solution A (1× SSC [1× SSC is 0.15 M NaCl plus 0.015 M sodium citrate], 0.1% sodium dodecyl sulfate) for 3 min, followed by solution B (0.05× SSC) for 3 min, and finally by solution C (95% ethanol) for 1.5 min. The slide was dried under a gentle air stream before it was scanned. For each DNA, at least three hybridization reactions were replicated to verify the reproducibility of the microarray method.

Data acquisition and automated analysis.

The slide was scanned with a laser beam at 532 nm using the GenePix personal 4100A (Axon Instruments), and two files were generated, one for the images, saved as TIF, and the other for the signal intensity, saved as GPR. The signal-to-noise ratio was calculated for each spot using the Bactarray Analyzer 1.0, developed in-house, with the threshold set at 3.0. For each serotype, two to four probes were used, and each probe was printed in triplicate to eliminate any possible physical defects in the glass slide. A serotype was confirmed and reported when the following conditions existed: (i) the positive standard, the negative standard, the two positive control probes, the negative control probe, and the printing control probe all provided the expected signals and (ii) more than half of all the probes of the given serotype generated positive signals above the signal-to-noise ratio threshold.

RESULTS

Identification of specific genes.

The O-serotype-specific genes of each of the S. boydii types 13, 16, and 18 and E. coli O55, O111, O114, and O157 have been reported previously (7, 12, 13, 15, 22, 33, 36). In this study, we identified the specific genes of the other eight serotypes: S. sonnei; S. flexneri type 2a; S. boydii types 7 and 9; S. dysenteriae types 4, 8, and 10; and E. coli O128. For each of them, two primer pairs based on each of the wzx and wzy genes were designed (Table 2) and used to screen DNA pools containing 186 representative strains of all Shigella and E. coli O serotypes. With primer pairs of S. flexneri type 2a, only the pools containing E. coli O13, O129, and O135, which shared the same O antigen with S. flexneri type 2a (9), in addition to S. flexneri 2a, gave PCR products of the expected size. With the other primer pairs, only the bands of correct size were observed in the pool containing the target strains. Therefore, both the wzx and wzy genes were specific to the eight serotypes. For each serotype, one of the O-serotype-specific genes was chosen as a target gene to identify Shigella and E. coli (Table 3).

Multiplex PCR to amplify the target genes.

Multiplex PCR was used to streamline the overall test while maintaining specificity for individual amplicons. In the presence of one target strain, only the 16S rRNA gene primer pair and one of the serotype-specific primer pairs performed, while none of the other 14 specific primer pairs annealed with the template DNA. Therefore, for each of the 15 serotype strains, two bands were generated from the PCRs: one was the 16S rRNA gene, and the other was the specific gene (Fig. 2). The amplicon size ranged from 555 bp to about 1.25 kb in length (Table 3). The Shigella and E. coli representative strains belonging to other serotypes (except E. coli O13, O129, and O135 [see below]), S. enterica, and V. cholerae generated only the 16S rRNA gene products. Strains of B. cereus and S. aureus failed to generate any amplicon. The results showed that the 15 primer pairs were specific and compatible in one multiplex PCR.

FIG. 2.

Agrose gel electrophoresis of multiplex PCR products. Lanes: Mr, molecular weight standards (lambda DNA/EcoRI plus HindIII marker); A, S. boydii type 7; B, S. boydii type 9; C, S. boydii type 13; D, S. boydii type 16; E, S. boydii type 18; F, S. dysenteriae type 4; G, S. dysenteriae type 8; H, S. dysenteriae type 10; I, S. flexneri type 2a; J, S. sonnei; K, E. coli O55; L, E. coli O111; M, E. coli O114; N, E. coli O128; O, E. coli O157. Two bands were generated from the PCR products: one was the 16S rRNA gene (1.2 kb), and the other was the gene specific to the individual serotype.

Probe specificity.

The DNA microarray was tested using 186 representative strains of all Shigella and E. coli O serotypes, 9 strains of other bacterial species (Table 1), and 40 stool specimens. From 141 oligonucleotide probes initially screened, 52 probes were selected for the microarray: 48 probes for specific genes, 2 probes for positive control, 1 for negative control, and 1 for positional reference and printing control (Table 3). All of the representative strains belonging to the 15 serotypes consistently hybridized to their corresponding probes with 100% specificity. The hybridization results are shown in Fig. 3, arrays 1 to 15. Strains of E. coli O13, O129, and O135 gave the same results as S. flexneri type 2a (Fig. 3, array 1), because they had same O antigen (9). None of the 23 representative strains of Shigella and the 148 representative strains of E. coli belonging to other serotypes, or strains of other bacterial species which are likely present in stool samples, hybridized to the serotype-specific probes on the microarray (Fig. 3, arrays 16 and 17). The 40 stool specimens obtained from 20 adult volunteers reacted only to the 16S rRNA gene probes, not the serotype-specific probes (Fig. 3, array 18), a result that was consistent with the fact that large numbers of nonpathogenic E. coli organisms exist in the stools of healthy people.

FIG. 3.

Microarray differentiation of the pathogens. (1) S. flexneri type 2a and E. coli O13, O129, and O135. (2) S. sonnei. (3) S. boydii type 7. (4) S. boydii type 13. (5) S. boydii type 16. (6) S. boydii type 18. (7) S. dysenteriae type 4. (8) S. dysenteriae type 8. (9) S. dysenteriae type 10. (10) E. coli O55. (11) E. coli O111. (12) E. coli O114. (13) E. coli O128. (14) E. coli O157. (15) S. boydii type 9. (16) Other serotype strains of E. coli or Shigella, Salmonella, and V. cholerae. (17) B. cereus and S. aureus. (18) DNAs from healthy-adult stool specimens.

Double-blind test.

A double-blind test was performed in order to verify the stability and specificity of the microarray. A total of 38 clinical isolates of Shigella and E. coli (Table 1) and 70 mock stool samples were selected to hybridize to the microarray without disclosure of their identities during testing. All the detection results were consistent with those of the conventional methods.

Sensitivity of detection with genomic DNA.

Serial dilutions of the genomic DNAs of 15 serotype representative strains in the range of 1 μg, 100 ng, 50 ng, 10 ng, and 1 ng were used as the templates for multiplex PCR to test the sensitivity of the microarray method. The positive signals could be obtained from the dilutions of 1 μg to 10 ng, while the results were negative or the fluorescence signals were very weak with less than 10 ng. We chose 50 ng as the most suitable DNA quantity for this microarray, and all of the 38 clinical isolates belonging to the 15 serotypes could be detected successfully at this level.

Sensitivity of detection with mock stool specimens.

Pure cultures of each of the 15 serotype representative strains were diluted to 10¹ to 10⁶ CFU per ml, mixed with approximately 0.3 g of fresh stool specimens from healthy people, and tested with the microarray. All of the targets were detected at levels as low as 10⁴ CFU per ml. Some strains, such as E. coli O128, S. sonnei, S. boydii types 16 and 18, and S. dysenteriae types 8 and 10, could be detected successfully at 10³ CFU per ml (Table 4).

TABLE 4.

Sensitivity test results for mock stool specimens

Strain	Test result ata:
	106 CFU/ml	105 CFU/ml	104 CFU/ml	103 CFU/ml	102 CFU/ml	101 CFU/ml
S. sonnei	+	+	+	+	±	−
S. flexneri type 2a	+	+	+	−	−	−
S. boydii type 7	+	+	+	±	−	−
S. boydii type 9	+	+	+	−	−	−
S. boydii type 13	+	+	+	±	±	−
S. boydii type 16	+	+	+	+	±	−
S. boydii type 18	+	+	+	+	±	−
S. dysenteriae type 4	+	+	+	−	−	−
S. dysenteriae type 8	+	+	+	+	±	−
S. dysenteriae type 10	+	+	+	+	±	−
E. coli O55	+	+	+	−	−	−
E. coli O111	+	+	+	−	−	−
E. coli O114	+	+	+	±	−	−
E. coli O128	+	+	+	+	±	−
E. coli O157	+	+	+	−	−	−

^a+, positive signal; −, negative signal; ±, weak or ambiguous signal.

DISCUSSION

Systematic O serotyping of E. coli began in the early 1930s (23), and many studies showed that the O serotypes of E. coli are generally associated with pathogenesis (6, 17, 26, 33, 36). O serotyping became an important tool to classify E. coli in clinical settings. In recent years, some microarray and PCR assays have used toxin genes as targets (20, 23, 25) to identify pathogenic E. coli. However, since mutations, instability, and loss of toxin genes among Shiga-like toxin-producing E. coli are quite common (20, 25), it is not reliable to use the toxin genes, even if a large number of primer pairs are incorporated into the test. wzx, wzy, and glycosyltransferase genes are generally specific to individual O-antigen gene clusters. We previously suggested the application of O-serotype-specific genes for detection and identification of E. coli (14, 30). Here, we demonstrated the feasibility of using a microarray based on wzx, wzy, or transferase genes to identify pathogens from pure culture and more clinically relevant mock stool samples. The DNA microarray described in this communication has paved the way for the establishment of an array of all serotypes of Shigella and E. coli that we are currently working on.

In comparison to the traditional serotyping method, the microarray method is high throughput, specific, and sensitive and also avoids most cross-reactions. Nevertheless, as with traditional serotyping, the microarray has its limits in distinguishing S. sonnei and S. flexneri type 2a from other strains that share the same O-antigen structure and the corresponding O-antigen gene cluster. Among the 46 Shigella serotypes recognized, there are only 33 distinct O antigens. Of the 33 O-antigen forms, 12 are identical to some E. coli O antigens (35). When E. coli O13, O129, or O135 existed in stool samples, the probes for S. flexneri type 2a would have been expected to give positive signals. Similarly, the S. sonnei O-antigen gene cluster is identical in sequence to that of Plesiomonas shigelloides O17 (21). We therefore assumed that when P. shigelloides existed in stool, the probes for S. sonnei would also give false-positive results. Nevertheless, we will try to use other methods or other genes, such as lacZ or cadA (31), which are commonly present in E. coli but absent in Shigella, to differentiate these potential false positives.

The microarray has many advantages over traditional bacterial culture and serotyping methods and is applicable to many fields. First, it facilitates medical detection of Shigella and E. coli from stool samples in a timely fashion. Second, it allows efficient inspection for food contamination. Third, it will be an invaluable tool to sensitively monitor and accurately pinpoint causative bacterial strains to prevent the occurrence or spread of epidemics of bacterial infection.