The traditional method of identifying SEs is serological typing with antibodies. In general, serological typing is less sensitive to small variations among SEs than DNA-based methods. Several PCR-based methods are available for S. aureus
toxin typing (21
). Most require several separate reactions to distinguish among several ent
genes. More recently, methods for S. aureus
toxin typing by multiplex PCR have been reported (6
). These PCR methods are based on combinations of ent
gene-specific primers or a combination of universal forward primers and specific reverse primers (36
One problem with all present PCR-based methods is that novel or unexpected toxin genes can lead to false-positive or -negative results. For example, we observed that all sea gene-specific primers described in the literature can be used for successful amplification of sep as well. This might lead to the mistaken conclusion that a strain encodes sea and incorrect data about the distributions of ent genes and their roles in food poisoning. Since the relationship between the presence of a specific enterotoxin (or a combination of enterotoxins) and human food poisoning is not clearly understood, there is a need for a reliable and universal method for unambiguous identification of known ent genes and for detection of novel ent genes.
We describe here a combination PCR-microarray assay for detection and identification of ent genes. The analysis is based on PCR amplification of a variable region of almost all known ent genes with a single set of degenerate primers whose sequences correspond to those of the flanking highly conserved regions. The amplicons are then identified by analysis on the oligonucleotide microarray. This combined method takes advantage of the strengths of each technique. PCR amplification is highly sensitive, detecting target genes from genomic DNA even when they are present at low concentrations. DNA-DNA hybridization on the microarray increases the specificity of the assay and allows parallel analysis of multiple sequences simultaneously. In addition, the nonspecific amplicons often seen in PCRs have no effect on the hybridization of the targets with specific oligonucleotide probes.
Microarrays are not in common use in average laboratories today. However, like any new technology, as more applications are developed for the microarray technology, it will become more practical and may well become widely used. In the work described here, the presence of genes for each of 16 ent genes was analyzed by four methods: PCR amplification with primers specific for each of the 16 enterotoxin genes, followed by analysis by gel electrophoresis (Fig. ); PCR amplification with specific primers (Fig. ), followed by analysis by the microarray assay; amplification of the 16 genes with universal primers, followed by microarray analysis (Fig. ); and sequencing of the enterotoxin genes to verify their identities. Thus, using two amplification methods (methods with specific primers and universal primers) as well as three DNA analysis methods (gel electrophoresis, DNA sequencing, and DNA microarray analysis), we verified the performance of the method. Using this array, we have shown that some strains previously analyzed by immunological methods contain additional ent genes not detected by the original assays (Fig. and ).
In our microarray system, we used relatively short oligonucleotides (21 to 30 nucleotides) for three reasons. First, shorter oligonucleotide probe sequences (<25 bp) are often capable of detecting a single-nucleotide mismatch between the target ssDNA and the oligonucleotide probe, which allows detection of minor genetic variants in target genes in a bacterial population. Second, the shorter oligonucleotide probes allow independent testing of several species-specific regions of each gene, enabling effective coverage of the target sequence with more (but shorter) oligonucleotide probes. This reduces the probability of misidentification. Third, short oligonucleotides reduce the cost of chip production.
The redundancy of the testing (the number of spots representing each gene) is one way to reduce the risk of SE misidentification. While only one portion of each SE gene (the variable region shown in Fig. ) is used for the analysis, leaving open the possibility of significant sequence variation in other parts of the genes, such variation is not common.
The main disadvantage of simultaneous PCR amplification of multiple targets is that different copy numbers of the genes in a cell result in different signal intensities on the array. This might be overcome by use of supplemental specific primers to improve the detection of underrepresented amplicons (Fig. and ).
Many of the known SEs have been discovered only in the last few years, including some, such as sep
, that were discovered only through the sequencing of the S. aureus
N315 genome (22
). Given the genetic variability and the spread of the ent
gene family, it is possible that there are other, as yet unknown, ent
genes. However, the similarity in the conserved regions of these genes (Fig. ) suggests that any additional gene family members will share those conserved sequences. Thus, it is possible that this assay might lead to the discovery of additional ent
genes in new strains. The amplicons of the novel ent
genes can be discernible as amplicons that hybridize to common spots but not to any ent
Multiple SE genes are commonly found in S. aureus
). Among 198 S. aureus
isolates implicated in S. aureus
infections in France, 85.4% expressed multiple SEs (19
). Our analysis of these data suggests that the majority (92%) contain multiple SEs, especially the egc
, and sem
). Among the S. aureus
isolates implicated in food-poisoning episodes in Japan, 93% expressed SEs, while only 72.2% of isolates from healthy people expressed SEs (32
One explanation for the presence of multiple toxins in most strains is that these genes are often structurally linked. Several pathogenicity islands have been reported in S. aureus
, including one encoding the toxic shock syndrome toxin (tst
) and s-
-like proteins (13
) and another encoding SE serotypes B, K, and Q (41
). Others have reported pathogenicity islands containing the tst
gene and an open reading frame with sequence similarity to those encoding SEs (25
) and a region contains enterotoxins D and J (42
). In addition, as noted above, a group of five toxin genes (seg
, and sem
) is encoded by the enterotoxin gene cluster, egc
). Interestingly, in the study of 198 clinical isolates by Jarraud et al. (19
), half of the 14 strains that carried only a single enterotoxin gene had the sea
gene, perhaps because it has been shown to be associated with a structurally unstable, possibly mobile, discrete genetic element (8
) that is not part of the egc
In terms of the functionalities of SEs, multiple toxins with diverse spectra of activities may offer the pathogen versatility in terms of the host range. For example, it has been shown that different types of SEs have different emetic response activities in house musk shrews, although it was thought that there are no differences in the emetic response activities of SEA, SEB, SEC, SED, and SEE in humans and primates (18
). In addition, S. aureus
expression of specific ent
genes may depend on the host tissue and may play a role in the adaptation of S. aureus
to the host environment (4
). Some have speculated that some cases of food poisoning result from the simultaneous expression of several enterotoxins in a single pathogenic S. aureus
strain rather than from the expression of a single toxin.
However, it is unclear whether all the toxins are actually expressed and what the biological and clinical effects of multiple toxins might be. The method presented here can detect ent
genes but does not determine whether the gene is expressed or whether the encoded protein is functional. The levels of correlation between the presence of genes that code for the production of SE (as determined by PCR) and the expression of these genes (as determined by enzyme-linked immunosorbent assay) were 100% for SEA and SEE, 86% for SEC, 89% for SED, and 47% for SEB (30
). Thus, the actual presence of the toxin needs to be assessed by an immunological or activity assay.
In summary, the PCR-microarray method described here is a potentially powerful tool for the analysis of S. aureus strains. We used this method to test clinical isolates analyzed previously and found that these isolates frequently carry the genes for numerous toxins, including some of the newly discovered SEs. More studies need to be done to understand the biological regulation and the biological and clinical effects of multiple enterotoxins. Our method has great potential for application in high-throughput screening and accurate genotyping of ent genes, which are especially important in epidemiological studies.