A novel serovar of Shigella dysenteriae from patients with diarrhoea in Bangladesh

Shigellosis is one of the major diarrhoeal diseases in Bangladesh and other developing countries. It has been estimated that more than 95 000 children <5 years of age die of shigellosis annually in Bangladesh (Kotloff et al., 1999). Shigella is spread by direct faecalhandoral contact wherever personal hygiene is compromised (Weissman et al., 1975) and outbreaks are difficult to control because of the low infectious dose of the organisms (DuPont et al., 1989).

The genus Shigella comprises four subgroups that historically have been considered species (Enterobacteriaceae Subcommittee of the Nomenclature Committee of the International Association of Microbiological Societies, 1958): subgroup A is referred to as Shigella dysenteriae, subgroup B as Shigella flexneri, subgroup C as Shigella boydii and subgroup D as Shigella sonnei. Speciation of Shigella is not easily achieved based on biochemical properties alone, and definitive identification requires serotyping to be performed (Ewing et al., 1958; Ewing & Lindberg, 1984). Based on somatic O antigens, S. dysenteriae is subdivided into 15 serogroups (Bopp et al., 2003) and 2 provisional serogroups (Coimbra et al., 2001; Melito et al., 2005) and S. flexneri into 6 serogroups with 14 subserotypes (Bopp et al., 2003). S. boydii comprises 19 serogroups. Ewing (1986) acknowledged 17 serogroups, which included two provisional serogroups represented by reference strain 2710-54 as S. boydii 16 and 3615-53 as S. boydii 17. Gross et al. (1980) reported a new provisional serogroup isolated in Bangladesh and designated serogroup 18 with the reference strain E10163. Recently, provisional S. boydii serovar E16553, designated serogroup 19, has been reported by several groups from diverse geographical origins (Gross et al., 1982; Ansaruzzaman et al., 2005); and serogroup 20 was recently proposed and confirmed as novel by Woodward et al. (2005).

In a previous study, we found that a significant number of strains isolated between 1999 and 2002 had biochemical properties typical of Shigella but could not be serotyped following the current serotyping scheme (Talukder et al., 2003). These strains were designated Shigella-like organisms (SLOs) and approximately 3 % of Shigella infections at ICDDR,B hospital are caused by the such organisms (Talukder et al., 2003). The prevalence of SLOs has increased over the years, accentuating the necessity for extensive characterization of these strains. In a retrospective study, we tested all SLOs with all available antisera specific for Shigella including existing, provisional and atypical types. Isolates were also tested for plasmid profiles as most Shigella serotypes contain a wide array of plasmids, some of which are serotype specific (Talukder et al., 2003). This revealed a cluster of seven isolates with identical plasmid patterns and which were not agglutinated by Shigella antisera. We describe here the properties of these isolates with respect to activity in biochemical tests, the presence of pathogenicity-associated genes, chromosomal DNA profiles and reactivity with a serum raised against one of their number.

Isolates. Seven isolates from 153 SLOs isolated from patients with diarrhoea attending the Dhaka Treatment Centre operated by ICDDR,B in Bangladesh were selected for this study. These isolates were non-reactive with available Shigella antisera and were isolated in the clinical microbiology laboratory with standard biochemical and microbiological methods (World Health Organization, 1987). All isolates were tested further in API 20E kits (bioMérieux). Strain E112707-96, provisional S. dysenteriae serotype, was received from Dr Thomas Cheasty, ESYV Reference Laboratory (Laboratory of Enteric Pathogens, Health Protection Agency, Centre for Infections, London, UK). YSH6000, S. flexneri 2a (Talukder et al., 2002), and an Escherichia coli strain (ATCC 25922) lacking the 140 MDa invasive plasmid and susceptible to all antibiotics were used as positive and negative controls, respectively, in the Sereny test, Congo red binding ability test, and in the PCR assay for detection of ipaH and Shigella enterotoxin genes (set).

Biochemical characterization. Isolates were subjected to a range of biochemical tests by conventional methods as described by Ewing (1986).

Serotyping. Isolates were tested in the local laboratory with commercially available antisera (Denka Seiken) specific for all recognized Shigella serovars and with antisera raised against serovars S. dysenteriae 13, 14 and 15. They were subcultured on MacConkey agar (Difco, Becton Dickinson) and incubated for 1820 h at 37 °C. Slide agglutination tests were performed as described previously (Talukder et al., 2002). Isolates were also tested in the ESYV reference laboratory with all available serogroups of Shigella and E. coli. An antiserum was raised in rabbits against a heat-killed suspension of a representative isolate (KIVI 162) according to the protocol of Ewing & Lindberg (1984). This antiserum was tested against other SLOs, S. dysenteriae serotypes 13, 14 and 15, S. boydii serotypes 19 and 20, and enteroinvasive E. coli serotypes O28ac, O29, O124, O136, O143, O144 and O164.

Agglutinin-absorption tests. When cross-reactions were found, reciprocal absorptions were performed to determine the extent of the antigenic relationship between the strains (Wathen-Grady et al., 1985). Serological cross-reactions were considered significant if the heterologous titre was ≥1/16 of the homologous titre.

Antimicrobial susceptibility. Susceptibility to antimicrobial agents was determined on MuellerHinton agar by the disc diffusion method following the guidelines of the Clinical and Laboratory Standards Institute (2004). The antimicrobial discs used were ampicillin (10 µg), tetracycline (30 µg), mecillinam (25 µg), nalidixic acid (30 µg), sulfamethoxazoletrimethoprim (25 µg), ciprofloxacin (5 µg) and ceftriaxone (30 µg). E. coli ATCC 25922 and Staphylococcus aureus ATCC 25923 were used as control strains.

Keratoconjunctivitis assay. This test was performed according to the procedure of Sereny (1957). Briefly, an overnight culture of bacteria, of approximately 10¹⁰ viable cells in 20 µl PBS, was introduced into the conjunctival sac of the guinea pig. The other eye served as the control. The guinea pigs were observed daily for 72 h and inflammatory responses were graded. The experimental procedures were approved by the animal experimentation ethics committee of the ICDDR,B.

Congo red binding ability. Trypticase soy broth containing 0.3 % yeast extract with 1.5 % agar supplemented with 0.01 % Congo red (Sigma) was used to study the pigment binding ability of isolates according to the procedures of Sasakawa et al. (1986).

Detection of virulence genes by PCR. DNA was prepared from the seven isolates and examined by PCR for the presence of the invasion-associated locus (ial) and the invasion plasmid antigen H (ipaH) of Shigella (Vargas et al., 1999). PCR to demonstrate the presence of Shiga toxin 1 (stx₁) and Shigella enterotoxins (set1 and sen) was performed according to published procedures (Bonnet et al., 1998; Vargas et al., 1999). Primer sequences are listed in Table 1; reactions were performed in a thermal cycler (PTC-200; MJ Research) and amplification products were separated by 1 % agarose gel electrophoresis and stained with ethidium bromide.

Table 1. Primers and conditions used for identification of virulence genes of Shigella spp

Plasmid DNA. Plasmid DNA was prepared according to the alkaline lysis method of Kado & Liu (1981) with some modifications (Talukder et al., 2002). The molecular mass of the plasmid DNA bands was assessed by comparison with the mobility of known molecular mass plasmids in E. coli PDK-9, R1, RP₄, Sa and V517 in agarose gels (Talukder et al., 2002).

PFGE. Intact agarose-embedded chromosomal DNA was prepared, digested with XbaI (Gibco-BRL), and restriction fragments were resolved in a contour-clamped homogeneous electric field (CHEF-DRII) apparatus (Bio-Rad) following the procedure of Talukder et al. (2002). Pulse times were 16 s for 9 h, 340 s for 12 h, and 570 s for 13 h. The gel was stained, destained and photographed on a gel documentation system and bands were sized against a 48.51000 kb molecular mass ladder (Bio-Rad).

Biochemical characterization
All isolates showed biochemical reactions typical of Shigella species. They were oxidase- and catalase-negative, and did not ferment D-mannitol, sucrose, lactose, rhamnose, raffinose, xylose, trehalose or dulcitol. All utilized arginine and were lysine- and ornithine-negative. API 20E tests identified all isolates as Shigella species.

Serotyping
All isolates were negative in agglutination tests for Shigella and E. coli serotypes and reacted only with the antiserum raised against KIVI 162. This was confirmed by the ESYV reference laboratory in London, but the isolates did react with an antiserum against a provisionally recognized S. dysenteriae serotype, E112707-96 (unpublished data).

Antibiotic susceptibility
All isolates were resistant to trimethoprimsulfamethoxazole, tetracycline and ampicillin and were susceptible to nalidixic acid, ciprofloxacin, ceftriaxone and mecillinam.

Test for invasiveness
Congo red dye was bound by all isolates and they caused keratoconjunctivitis in the guinea pig eye within 48 h.

Virulence pathogenicity
All isolates were positive for ial, ipaH and sen genes and were negative for stx₁ and set1 genes. In contrast, the provisional strain E112707-96 was negative for ial and sen genes and only positive for ipaH.

Plasmid DNA
All isolates harboured the 140 MDa invasive plasmid and two other plasmids of approximately 2.7 and 2.0 MDa in size. This plasmid pattern was different from that of other Shigella serotypes tested, including the provisional strain E112707-96, which lacked the 140 MDa plasmid but contained another plasmid of approximately 65 MDa.

PFGE
All isolates were indistinguishable from each other by their XbaI restriction pattern. This pattern differed markedly from that of other serotypes, including the provisional S. dysenteriae serotype E112707-96 (Fig. 1).

(101K): Fig. 1. PFGE patterns of XbaI-digested chromosomal DNA from seven SLOs and provisional S. dysenteriae serotype E112707-96. Lanes: 1 and 10, lambda molecular mass ladders; 28, SLO isolates; 9, provisional S. dysenteriae serotype E112707-96.

SLOs are consistently isolated from the hospitalized diarrhoeal population in Bangladesh. During the period 19992002, S. flexneri accounted for 60 % of the total Shigella infections followed by S. boydii (17 %), S. dysenteriae (12 %) and S. sonnei (8 %) then SLOs (3 %) (Talukder et al., 2003). Serological classification is insufficient for the identification of novel Shigella species and a polyphasic approach utilizing biochemical, serological and molecular characterization is advocated (Ansaruzzaman et al., 2005). In this study, we characterized seven SLO isolates recovered over 3 years in hospitalized patients with diarrhoea (Table 2). They all had biochemical properties compatible with S. dysenteriae and had an antimicrobial resistance profile (ampicillin, tetracycline and trimethoprimsulfamethoxazole) common among S. flexneri and S. dysenteriae isolates in Bangladesh (Talukder et al., 2002). Among the serotypes of Shigella, S. dysenteriae type 1 is generally the most resistant to antibiotics and appears to readily acquire resistance to newly introduced antibiotics for Shigella infection (Talukder et al., 2004). Indeed, all S. dysenteriae type 1 isolates in Bangladesh in 2004 and onwards are resistant to ciprofloxacin (Talukder et al., 2004). However, none of the seven SLO isolates was resistant to ciprofloxacin.

Table 2. Characteristics of the SLOs studied , Negative; +, positive.

Although pathogenicity tests are not fundamental criteria for the classification of members of the Enterobacteriaceae, the findings of a positive Sereny test, the presence of a 140 MDa plasmid and the identification of ial and ipaH genes provide strong evidence in support of a classification in the genus Shigella. Two novel toxins, Shigella enterotoxin 1 (ShET-1) encoded by chromosomal gene set1 and Shigella enterotoxin 2 (ShET-2) encoded by sen (located in the 140 MDa plasmid), have been associated with the pathogenicity of many Shigella species (Nataro et al., 1995; Noriega et al., 1995). To date, S. dysenteriae 1 is the only serotype that has been reported to produce Shiga toxin 1 and some E. coli strains produce Shiga-like toxin 1 and 2. Strains belonging to the other serotypes of the Shigella group have not been shown to produce Shiga toxin.

Serotyping is considered to be essential for the confirmation of an organism as a member of the genus Shigella. An antiserum raised against one of the isolates in this study (KIVI 162) reacted only with the seven isolates but not with other Shigella or E. coli reference strains. Among E. coli pathotypes, enteroinvasive E. coli are the most similar in biochemical and virulence properties to Shigella, which can sometimes makes differentiation difficult. However, the SLO isolates did react with antiserum to an existing provisional S. dysenteriae serotype, designated E112707-96 (unpublished data), and this strain was strongly agglutinated by serum KIVI 162, supporting their antigenic identity.

A common genetic lineage for the SLO isolates was indicated by their similarity in plasmid and chromosomal DNA profiles. In the latter, one isolate exhibited the loss of a major band and the appearance of two minor bands in comparison with the others, but this difference was insufficient to distinguish between the isolates by the criteria of Tenover et al. (1995).

In summary, the SLO isolates characterized in this report exhibited traits characteristic of the genus Shigella in biochemical, serological and molecular properties. No serological cross-reactivity was observed with other Shigella and E. coli strains, except for an unpublished provisional S. dysenteriae serotype, E112707-96, isolated and identified by the E. coli reference laboratory in 1996. These isolates possess an antigenic epitope that sets this group apart from all other recognized strains of Shigella and we propose that they represent a novel emerging serovar of S. dysenteriae, type strain KIVI 162.

This study was shared by the United States Agency for International Development (USAID) under Cooperative Agreement No. HRN-A-00-96-90005-00 and the ICDDR,B: Centre for Health and Population Research, which is supported by countries and agencies who share its concern for the health problems of developing countries. Current donors providing unrestricted support include: Australian International Development Agency (AusAID), Australia; Canadian International Development Agency (CIDA), Canada; Centre endowment fund; Department for International Development (DFID), UK; Government of Bangladesh (GoB); Government of Sri Lanka; Kingdom of Saudi Arabia (KSA); Government of the Netherlands; Swedish International Development Cooperative Agency (Sida), Sweden; and Swiss Development Cooperation (SDC), Switzerland. ICDDR,B acknowledges with gratitude the commitment of USAID to the Centres research efforts.