INTRODUCTION
Infections caused by Shigella continue to be a major public health problem with an estimated annual incidence of 160 million cases worldwide [Reference Kotloff1]. Several epidemiological studies indicate that S. flexneri 2a and S. sonnei are the most predominant Shigella isolated in both developing and industrialized countries [Reference Kotloff1, Reference Niyogi2]. Considering the global burden of Shigella, the difficulties in implementing preventive and control measures and emerging antibiotic resistance, WHO has given high priority for vaccine development programmes against Shigella [Reference Levine3, Reference Phalipon, Mulard and Sansonetti4]. However, this development requires the capacity to identify the most prevalent species and serotypes in different geographic locations.
Routine microbiological identification of Shigella, including serotyping, is a multiple-step technique that usually takes 3–5 days [Reference Niyogi2]. Multiplex PCR (mPCR) assays, which detect several virulence markers in a single PCR reaction, are becoming the method of choice for rapid, specific and sensitive detection of diarrhoeagenic pathogens in both developing and industrialized countries [Reference Vidal5, Reference Gomez-Duarte, Bai and Newell6]. For Shigella, several mPCR have been described, but they do not discriminate between Shigella species or serotypes, nor do they differentiate Shigella from the closely related pathogen enteroinvasive Escherichia coli (EIEC) [Reference Aranda, Fagundes-Neto and Scaletsky7–Reference Villalobo and Torres10]. To improve the specificity of these tests, genetic markers exclusively present in Shigella spp. and/or serotypes need to be identified.
Pathogenicity islands (PAI) are discrete genetic elements often inserted adjacent to tRNA genes, which encode virulence genes and mobile genetic elements such as integrase genes [Reference Schmidt and Hensel11, Reference Hacker and Kaper12]. The Shigella-specific she PAI is located in the chromosome next to the pheV tRNA gene and has been found mostly in S. flexneri 2a but rarely in other serotypes [Reference Al-Hasani13]. This PAI encodes a P4 phage-like integrase (int gene), and proteins associated with Shigella pathogenicity such as Shigella enterotoxin 1 (ShET-1) and a cytopathic autotransporter protease, SigA [Reference Fasano14, Reference Al-Hasani15]. Using in silico analyses with available Shigella genome sequences, we identified that the 5′-end of the she PAI, which includes the sigA gene, but not the 3′-end is present in S. sonnei Ss046 genome sequence. Moreover, this analysis showed that the she PAI is completely absent in S. flexneri serotype 5b strain 8401 genome sequence (Fig. 1 a). With this information, we designed primers to develop a mPCR that specifically differentiates S. flexneri and S. sonnei. In addition, and because of their importance as virulence factors in the development of live attenuated vaccines against Shigella, specific primers were included in this mPCR to detect ShET-1 and ShET-2 enterotoxin genes, set and sen genes, respectively [Reference Kotloff16, Reference Kotloff17]. The mPCR described here provides a new tool not only useful for identification of a presumptive Shigella isolate in a diagnostic laboratory but also for epidemiological surveillance of the most prevalent Shigella serotypes and Shigella-associated virulence determinants.
Shigella spp. detection. (a) Alignment analysis comparing the she pathogenicity island (PAI) insertion region on the genome sequence of S. flexneri 2a strain 2457T with S. sonnei strain Ss046 and S. flexneri 5b strain 8401 (top panel). Arrows indicate localization of She1, She16, Int1R and Int2R primers. Magnification of the integrase gene region of S. flexneri 2a strain 2457T and S. sonnei strain Ss046, indicating the recognition site of primer Int1R and Int2R (bottom panel). (b) Agarose gel electrophoresis showing mPCR products obtained with the four primers described above simultaneously using S. flexneri 2a (lane 1, 1676-bp fragment), S. sonnei (lane 2, 1097-bp fragment) and S. flexneri non-2a (lane 3, 401-bp fragment) strains as template. L, Molecular size markers (1 kb plus ladder from Invitrogen).
METHODS
Bacterial strains
We evaluated 304 Shigella strains, corresponding to 129 strains of S. sonnei, 99 of S. flexneri 2a, 72 of S. flexneri non-2a (1 of S. flexneri 1a, 4 of S. flexneri 1b, 18 of S. flexneri 2b, 14 of S. flexneri 3a, 1 of S. flexneri 3b, 7 of S. flexneri 3c, 8 of S. flexneri 4a, 2 of S. flexneri 5b, 12 of S. flexneri 6, 1 of S. flexneri X and 4 of S. flexneri Y) and 4 of S. boydii. These strains were isolated from stool samples of Chilean children aged <14 years with acute diarrhoea, collected from 1995 to 2007. Bacteria were cultured and identified by conventional biochemical methods and serotyping (Denka Seiken Co., Japan). We also analysed 79 Shigella strains from the French National Reference Centre for Escherichia coli and Shigella Collection, Institut Pasteur, France (Table 1). These strains were mostly isolated from stools during the period 2004–2007; they represent a set of different serotypes coming from different geographic regions (Europe, Asia, Africa, South America). Serotype distribution of the 79 strains was as follow: S. sonnei (3), S. flexneri 2a (4), S. flexneri non-2a (38), S. boydii (17), and S. dysenteriae (17). S. flexneri 2a strain 2457T and S. sonnei ATCC 25922 strain were used as Shigella reference strains.
Shigella strains from different geographic region used in this study
* The probable original source is associated to travellers' diarrhoea when the source is different from France.
† Serotype described by Matsushita et al. [Reference Matsushita18].
The specificity of the mPCR was tested using the following bacteria obtained from clinical samples: Salmonella group A (1), Salmonella group B (7), Salmonella group C (1), Salmonella group D (14), Klebsiella pneumoniae (3), Hafnia spp. (1), Proteus spp. (1), Yersinia enterocolitica (1), Citrobacter freundii (1), Vibrio parahaemolyticus (1), Enterobacter cloacae (1), Acinetobacter baumannii (1), 3 Pseudomonas aeruginosa (3), Campylobacter jejuni (2), and Listeria monocytogenes (1). We also tested the following diarrhoeagenic E. coli strains: EIEC (1), enterohaemorragic E. coli (EHEC) (4), enteropathogenic E. coli (EPEC) (4), enteroaggregative E. coli (EAEC) (3), diffuse adherent E. coli (DAEC) (3), Shiga toxin-producing E. coli (STEC) (1), and enterotoxigenic E. coli (ETEC) (3).
Multiplex PCR design
To develop a mPCR to specifically differentiate S. flexneri and S. sonnei, specific markers were sought at the she PAI. Genome sequences available for S. flexneri 2a strains 2457T and 301 [Reference Jin19, Reference Wei20], S. sonnei strain Ss046 (GenBank accession no. CP000038) and S. flexneri 5b strain 8401 (GenBank accession no. CP000266.1) were compared at the she PAI insertion site, the pheV tRNA gene. This analysis indicated that a homologous int gene in S. sonnei, located at the 5′-end of the she PAI, can be differentiated from the S. flexneri 2a P4-like int gene using primers Int1R and Int2R. This analysis also showed that in the S. flexneri 5b genome sequence the she PAI is not present in pheV gene boundary and its absence can be detected using primers She1 and She16. The recognition sites of these primers are shown in Fig. 1 a. Alignment analyses using Shigella genome sequences described above and the primers She1, She16, Int1R, Int2R showed that a 1676-bp and a 1097-bp fragment would be detected exclusively in S. flexneri 2a (with She1 and Int1R primers), and S. sonnei (with She1 and Int2R primers), respectively. Moreover, a predicted 401-bp fragment would be detected only in S. flexneri 5b (with She1 and She16 primers) indicating the absence of the she PAI in this reference strain. The final mPCR included the primers described above plus three sets of primers specific for sen, set and virF genes (Table 2). Primer3 software (http://frodo.wi.mit.edu/cgi-bin/primer3/primer3_www.cgi) was used for design of the primers.
Primers used in this study
* Amplicon obtained with primer She1.
Multiplex PCR protocol
One colony from a McConkey or Salmonella-Shigella (SS) agar plate was suspended in 200 μl of 1% Triton X-100 solution, and boiled for 10 min. The mPCR reaction was performed with a 25-μl mixture, containing 1 μl of boiled lysate as DNA template, 3 mm MgCl2, 400 μm (each) deoxynucleoside triphosphate, the 10 primers simultaneously (Table 2), and 1 U of Taq polymerase (Invitrogen, USA). Optimal mPCR reaction was performed using an initial 2 min denaturation step at 95°C, 30 cycles at 95°C for 1 min, 56°C for 30 s, and 72°C for 2·5 min, with a final extension at 72°C for 10 min. PCR products were analysed by electrophoresis in 2% agarose gels stained with ethidium bromide.
Determination of mPCR sensitivity
Cultures of ∼1× 108 colony-forming units (c.f.u.)/ml of S. flexneri 2a 2457T, S. sonnei ATCC 25922 or a clinical isolate of S. flexneri serotype 3 were serially diluted in phosphate buffered saline (PBS) (pH 7·4). Two hundred microlitres of each diluted culture was boiled for 10 min and 1 μl of the lysate was used as a template for the mPCR reaction. The number of c.f.u. was determined in 100 μl of each diluted culture on McConkey agar. The sensitivity of the assay was defined as the lowest c.f.u. of Shigella per mPCR reaction that yielded positive amplification of all the markers expected.
RESULTS
Shigella spp. detection
To develop a mPCR to specifically differentiate S. flexneri and S. sonnei, we searched for specific markers at the she PAI. The comparative analyses between S. flexneri 2a strain 2457T, S. sonnei strain Ss046 and S. flexneri 5b strain 8401 genome sequences and the location of primers Int1R, Int2R, She1 and She16 are detailed in Fig 1 a. Using these four primers simultaneously, a band of ∼1650 bp was detected exclusively in the S. flexneri 2a 2457T reference strain used as a template (Fig. 1 b, lane 1). A ∼1000-bp fragment was amplified only in S. sonnei ATCC 25922 (Fig. 1 b, lane 2). In addition, the mPCR was tested with 10 Chilean strains of S. flexneri non-2a randomly selected, amplifying a ∼400-bp single band in all the strains assayed (Fig. 1 b, lane 3). Non-specific bands were not detected. The sequencing of the mPCR products amplified above corresponded to the expected size fragments obtained by the alignment analyses with Shigella genome sequences.
Based on the detection of she PAI associated markers S. flexneri and S. sonnei reference strains were differentiated using four specific primers in the same PCR reaction. Moreover, in all S. flexneri non-2a Chilean strains assayed a 401-bp fragment was amplified indicating lack of the she PAI insertion at pheV gene.
mPCR evaluation in Shigella isolates from clinical samples
To test the mPCR including primers She1, She16, Int1R and Int2R, lysates were tested from isolated colonies of 304 Shigella strains obtained from Chilean children previously identified by serotyping. All 99 S. flexneri 2a strains yielded the 1676-bp fragment specific for this serotype, according to sequence analysis. Similarly for all 129 strains of S. sonnei, the 1097-bp fragment was detected. Interestingly, the 401-bp fragment was amplified in 58/72 (81%) S. flexneri non-2a strains. Of the 14 S. flexneri non-2a strains lacking the 401-bp fragment, two were S. flexneri serotype Y and amplified a band similar to the 1676-bp fragment specific for S. flexneri 2a; the other 12 strains corresponded to S. flexneri serotype 6. No amplicons were detected in Chilean S. boydii isolates.
To determine the universal applicability of the mPCR, 79 Shigella strains isolated from other geographical regions were tested (Table 1). Table 3 shows that in all (3/3) S. sonnei strains and all (4/4) S. flexneri 2a strains the expected 1097-bp and 1676-bp fragments, respectively, were detected. However, the 1676-bp fragment was also found in 10 strains of S. flexneri non- 2a used as a template (6 strains corresponding to serotype 2b, 1 to serotype 3a, 1 to serotype 4a and 2 to serotype Y). From the remainder of the S. flexneri non-2a strains (28/38), the 401-bp fragment was detected in 22 strains, and no amplification products were found in any of six S. flexneri serotype 6 tested. No amplicons were detected using S. boydii and S. dysenteriae.
Amplification products obtained with primers She1, Int1F, Int2F and She16 using Shigella strains as a template
* All strains correspond to S. flexneri serotype 6.
Detection of Shigella virulence determinants
To enhance the utility of this mPCR as a tool for detection of Shigella associated-virulence determinants, specific primers were incorporated to detect set, sen and virF genes. The set gene (ShET-1 enterotoxin) is encoded chromosomally within the she PAI [Reference Al-Hasani13]; sen (ShET-2 enterotoxin) and virF (Shigella virulence regulator) genes are encoded in the 220-kb virulence plasmid [Reference Nataro21, Reference Hale22].
The different amplification patterns using the ten primers simultaneously are shown in Fig. 2. For S. sonnei three amplification patterns were found (Fig. 2, lanes 1–3); in contrast, all S. flexneri strains that harboured the she PAI displayed a unique amplification pattern (Fig. 2, lane 4). Four patterns were observed for S. flexneri she PAI-negative strains (Fig. 2, lanes 5–8); three of these strains were characterized by amplification of the 401-bp fragment, indicating the absence of she PAI in the pheV boundary (Fig. 2, lanes 5–7); the fourth pattern, distinguished by the sole amplification of virF and sen markers was exclusive to S. flexneri serotype 6 (Fig. 2, lane 8). All S. boydii and S. dysenteriae strains also presented this amplification pattern (virF + and sen +), indicating the presence of the virulence plasmid. Interestingly, for S. sonnei and S. flexneri non-2a a potential loss of the 220-kb virulence plasmid was detected, since no amplification of virF and sen markers was observed (Fig. 2, lanes 3 and 7).
Agarose gel electrophoresis showing the amplification patterns to discriminate Shigella spp. using the ten primers in the mPCR reaction. L, Molecular size markers (1 kb plus ladder from Invitrogen); lanes 1–3, S. sonnei; lane 4, S. flexneri harbouring she pathogenicity island (PAI); lanes 5–8, S. flexneri she PAI-negative strains; lanes 9–12, diarrhoeagenic E. coli (lane 9, EIEC; lane 10, EAEC; lane 11, STEC; lane 12, EHEC/EPEC).
Analysis of the 383 Shigella strains showed that the sen gene marker was present in 308 (80%) of them whereas the set gene marker was found only in Shigella strains that harboured the she PAI (115/383) (Table 4).
Frequency of the virulence-determinant markers in 383 Shigella isolates analysed
Specificity and sensitivity of the mPCR
The mPCR proved to be specific for Shigella and was negative with all other species tested with the exception of diarrhoeagenic E. coli strains which displayed some amplified products which are probably related to the PAI present in these pathogens (Fig. 2, lanes 9–12). For the EIEC strain, which is known to harbour the Shigella 220-kb virulence plasmid, amplicons compatible with the sen and virF genes were detected, displaying the same pattern observed for S. flexneri 6, S. boydii and S. dysenteriae (Fig. 2, lanes 8, 9). For EAEC strains, the set marker was amplified (Fig. 2, lane 10), as previously described [Reference Vila23]. In both of these E. coli pathotypes no amplification for S. flexneri 2a or S. sonnei integrase was obtained, which allows for the differentiation of these pathogens from S. flexneri (with the exception of S. flexneri 6) or S. sonnei strains. For STEC, a band similar to the 1676-bp fragment specific for S. flexneri she PAI-positive int gene was detected, but not the set gene (Fig. 2, lane 11). Finally, for EHEC and EPEC, a band similar to the 1097-bp fragment specific for S. sonnei int gene was amplified, displaying the amplification pattern observed for S. sonnei lacking the virulence plasmid (Fig. 2, lane 12). Sensitivity tests with S. flexneri 2a, S. sonnei and S. flexneri non-2a cultures revealed the expected amplicons in 100 c.f.u. of organisms, and repeated tests with freshly prepared bacterial template gave the expected amplification products for all strains.
DISCUSSION
Classical methods for determining the presence of Shigella are time-consuming and labour-intensive; therefore, the detection of several virulence markers in a single PCR reaction represents an important advance in the identification of this microorganism [Reference Mackay24, Reference Settanni and Corsetti25]. Most of the mPCR assays to identify Shigella so far described are confined to the detection of the ipaH gene, a marker present in all Shigella isolates as well as in EIEC strains [Reference Aranda, Fagundes-Neto and Scaletsky7, Reference Thong9, Reference Riyaz-Ul-Hassan26–Reference Brandal28]. The lack of specific markers to recognize a particular species or serotype has limited the design of new mPCRs. In this study, we designed a mPCR based on the detection of markers present on the she PAI. This PAI correspond to a 46-kb segment mostly found in S. flexneri 2a [Reference Al-Hasani13] and is partially present in the S. sonnei genome (Fig. 1 a). Using bioinformatic analyses the integrase gene located in the 5′-end of this PAI can be differentiated from its homologous gene present in S. flexneri 2a with specific primers (Fig. 1 a). Using this approach, four primers were designed to differentiate S. flexneri harbouring the she PAI and S. sonnei by the specific amplification of a 1676-bp and a 1097-bp fragment, respectively. Moreover, a 401-bp fragment was found that indicated the absence of the insertion of the she PAI in the pheV boundary in most (73%) of S. flexneri non-2a obtained from different regions worldwide.
Also included in the mPCR reaction were specific primers for ShET-1 and ShET-2 enterotoxin genes and the 220-kb Shigella virulence plasmid marker virF gene. Recently, vaccine trials using live attenuated Shigella strains suggested that ShET-1 or ShET-2 or both contribute to human disease, indicating their importance as virulence factors [Reference Kotloff16, Reference Kotloff17]. Thus ten primers were incorporated in the same PCR reaction and used to test a large collection of Shigella strains. A high prevalence of the ShET-2 marker was found, in agreement with previous reports [Reference Yavzori, Cohen and Orr29–Reference Noriega31]. Considering that 16% of the strains were virF negative (63/383 strains), which suggests the lack of the 220-kb virulence plasmid might probably be due to long storage [Reference Schuch and Maurelli32], the prevalence of the ShET-2 marker could be even higher. On the other hand, the presence of the she PAI, indicated by the amplification of the 1676-bp fragment and the set gene marker, was found in all S. flexneri 2a strains as well as in a minority of S. flexneri non-2a strains, a result that is in accord with previous studies [Reference Al-Hasani13, Reference Yavzori, Cohen and Orr29, Reference Roy33]. Interestingly, the absence of the set gene in all S. flexneri strains not harbouring the she PAI was correlated with the amplification of the 401-bp fragment, with the exception of all S. flexneri 6 strains assayed. The lack of this fragment in S. flexneri 6 isolates might be explained by insertion of a long DNA sequence or changes in the pheV boundary that affect the amplification with primers She1 and She16. For S. boydii and S. dysenteriae no amplification of she PAI markers was evident, but these serotypes exhibited an amplification pattern similar to S. flexneri 6 where the virulence plasmid markers sen and virF genes were detected.
A DNA microarray targeting O-serotype-specific genes to detect all 34 distinct O-antigen forms of Shigella was recently developed [Reference Li34, Reference Li35]. Even though this technique proved to be specific, sensitive and reproducible, its application as a diagnostic or epidemiological tool is difficult, even in industrialized countries, in view of the elevated cost, instruments and qualified personnel necessary to perform this technique [Reference Call36]. The mPCR described here might offer a more practical approach for rapid, easy and affordable identification of Shigella, particularly in developing countries where Shigella incidence is high and resources are limited. Although our mPCR was tested on pure cultures rather than food or clinical samples, the application of this technique using a single colony grown on selective media from food or stool samples might reduce the time of the identification of Shigella compared to conventional methods.
The mPCR described here represents a new approach based on the identification of several serotypes of clinical or epidemiological importance. As the assay is not based exclusively on the detection of genes present in the 220-kb virulence plasmid, this assay might allow the differentiation between S. sonnei and most of S. flexneri serotypes from EIEC strains. In addition, as Shigella vaccine development is mostly focused on S. flexneri 2a and S. sonnei [Reference Levine3, Reference Phalipon, Mulard and Sansonetti4], the mPCR described here may prove to be a valuable tool in epidemiological studies to identify specifically the most frequent Shigella isolates and contribute to the surveillance of the virulence determinants ShET-1 and ShET-2 enterotoxins.
ACKNOWLEDGEMENTS
This work was supported by grant FONDECYT 1040539, grant ADI-08/2006 from ‘Programa Bicentenario de Ciencia y Tecnología’ (CONICYT, Chile) and The World Bank to C.S.T, and Cooperation Program INSERM/CONICYT (A. Phalipon-V. Prado). We thank Drs Miguel O'Ryan, James Nataro, Juan Carlos Salazar, and Carlos Santiviago for careful review of the manuscript and helpful discussions. We thank Carlos Blondel for invaluable help in sequence data analyses and generation of graphic material. We are extremely grateful to Professor Michael McClelland at the Sidney Kimmel Cancer Center in San Diego, CA, USA, for support, ideas and laboratory space during preliminary work towards the multiplex PCR assay described here. M.J.F. was supported by a CONICYT Ph.D. fellowship.
