A New Phylogenetic Cluster of Cereulide-Producing Bacillus cereus Strains

Phenotypic and molecular studies have established that cereulide-producingstrains of Bacillus cereus are a distinct and probably recentlyemerged clone within the Bacillus population. We analyzed aset of B. cereus strains, both cereulide producers and nonproducers,by multilocus sequence typing. Consistent with earlier reports,nonproducers demonstrated high heterogeneity. Most cereulide-producingstrains and all flagellar antigen type H1 strains were allocatedto the known sequence type of exclusively emetic B. cereus strains.Several cereulide-producing strains, however, were recoveredat a new phylogenetic location, all of which were serotype H3or H12. We hypothesize that the group of cereulide producersis diversifying progressively, probably by lateral transferof the corresponding gene complex.

INTRODUCTION

Top

Introduction

Bacillus cereus is often associated with food contamination,food spoilage, and food poisoning. The clinical presentationtakes one of two forms, emetic and diarrheal. It is well establishedthat the emetic type of food poisoning is associated with asingle heat-stable exotoxin, cereulide (2, 8, 22). Diarrhealfood poisoning, on the other hand, can be caused by severalthermolabile enterotoxins (6).

Agata et al. have shown that most cereulide-producing strainsbelong to serovar H1, with the exception of a few that belongto serotypes H3 and H12 (1). Thus, cereulide producers appearto be a homogeneous group. Phenetic and molecular typing studieshave established that cereulide-producing B. cereus strainsbelong to a highly homogeneous cluster (9), which was also suggestedby earlier reports of identical ribopatterns of 16S and 23SrRNA sequences from emetic strains (5, 15, 17). Multilocus sequencetyping (MLST) has assigned all analyzed cereulide-producingstrains, isolated from cases of both food poisoning and theenvironment, to single-sequence type (ST) 26 (16). Recently,however, a few cereulide-producing strains have been reportedto possess unique ribopatterns and to demonstrate a variationin the housekeeping gene adk (4).

This unexpected finding leaves open the question of how widelythe capacity for emetic toxin production is distributed amongB. cereus strains. Should these findings be considered minordiscrepancies, or is heterogeneity among cereulide producersreal? To address these questions, we analyzed B. cereus strainsisolated from a variety of sources, both cereulide-producingand cereulide-nonproducing strains, using MLST, and comparedthe phylogenetic allocation of the cereulide-producing strainswith their serotypes.

MATERIALS AND METHODS

Top

Materials and Methods

Bacterial strains and serotyping. A collection of 52 strains of B. cereus, 25 cereulide-producingstrains and 27 cereulide-nonproducing strains, was used in thisstudy. The strains were collected from cases of emetic and diarrhealfood poisoning, during regular food hygiene inspections, andfrom the environment (river, water, and soil sources). Thesestrains were isolated between 1974 and 2005, mainly in Japan(Tables 1 and 2). Flagellar antigen serotyping was performedby the H antigen agglutination method described by Taylor andGilbert (19).

View this table:
[in this window]
[in a new window]

TABLE 1. Characteristics of cereulide-nonproducing strains of B. cereus

View this table:
[in this window]
[in a new window]

TABLE 2. Characteristics of cereulide-producing strains of B. cereus

Toxin analyses. Cereulide production was tested by the vacuole response assayas described by Agata et al. (3). Enterotoxin production wastested by using a reversed passive latex agglutination B. cereusenterotoxin test kit (Oxoid, Basingstoke, United Kingdom) forHBL (hemolytic enterotoxin), using instructions provided bythe manufacturer.

Molecular analyses. DNA was extracted according to the instructions for using aPromega Wizard genomic DNA purification kit (Promega K. K.,Tokyo, Japan) after culture growth in 5 ml Luria-Bertani broth.MLST was performed as described on the B. cereus MLST Web site(http://pubmlst.org/bcereus). PCR amplification was performedon an Astec PC 802 thermal cycler (Astec, Fukuoka, Japan) asfollows: 94°C for 5 min, 35 cycles of denaturation at 94°Cfor 1 min, annealing for 1 min (the recommended annealing temperaturefor each primer set was used), and extension at 72°C for1 min, with final extension at 72°C for 7 min. The PCR ampliconswere extracted following agarose gel electrophoresis using aQIAquick gel extraction kit (QIAGEN K.K., Tokyo, Japan). Sequencingwas performed using a GenomeLab dye terminator cycle sequencingQuick Start kit (Beckman Coulter, Fullerton, CA) with the sameprimers used for the amplification, in both forward and reversedirections, and the reaction products were separated using aBeckman Coulter CEQ 2000 XL DNA analysis system (Beckman Coulter,Fullerton, CA). DNA sequences were assembled using Sequenchersoftware (Gene Codes, Ann Arbor, MI).

Data analyses. The completed sequences were compared with those in the B. cereusMLST database. All new sequences were submitted to the database,and relevant data are available at http://pubmlst.org/bcereus.Concatenated sequences of all STs were downloaded from the database,and a neighbor-joining (NJ) tree was constructed using ClustalX (20) and visualized using Njplot (14). Split decompositionanalysis was performed using the Web version of Splits Treesoftware on the MLST Web site (http://pubmlst.org) (12). eBURSTanalysis was performed using the Web tool at http://www.mlst.net/(18).

RESULTS

Top

Results

All cereulide-producing strains of B. cereus were serotypedinto one of three groups, H1, H3, and H12. In contrast, manycereulide-nonproducing strains could not be assigned to anyof the 18 established serotypes described by Taylor and Gilbert(19), except for a few strains belonging to serotype H11. Whentested for enterotoxin production, all cereulide-nonproducingstrains were HBL positive, whereas all cereulide-producing strainswere HBL negative (Table 1 and 2).

MLST revealed that cereulide-nonproducing strains were highlydiverse, and 13 new STs were identified (data not shown). Fourstrains were allocated to ST 142. Another six strains were recoveredin ST 24, and four of them were serotype H11, but not all H11strains were ST 24. MLST analysis of cereulide-producing strainsseparated them into four STs (Table 3). Sequence types 165,164, and 144 were new to the database. Two of these STs (ST165 and 144) were single-nucleotide variants of their parenttypes, ST 26 and ST 164, respectively (data not shown). It isnoteworthy that all strains allocated to ST 26 and its variantST 165 were serotype H1. All strains in ST 144 were serotypeH3, and all strains in ST 164 were serotype H12 (Table 3).

View this table:
[in this window]
[in a new window]

TABLE 3. Changes in the allelic profiles of cereulide-producing strains of B. cereus

Phylogenetic analysis placed all emetic toxin-producing strainsin Clade 1, as designated by Priest, et al. (16), clusteredat two distinct loci: the previously known ST 26 with the newlyidentified variant ST 165 and the new ST 144/ST 164 locus (Fig.1). The new cluster ST 144/ST 164 is located in the neighboringlineage to B. anthracis, distinct from the previously identifiedcluster at ST 26. Cereulide-nonproducing strains, from boththe environment and from cases of food poisoning, were distributedin Clade 2, as designated by Priest, et al. (16), with onlytwo exceptions. These strains had no significant relationshipswith each other and did not form a recognizable cluster (Fig.2). We further subjected the group of emetic strains to eBURSTanalysis (implementation of the BURST [based upon related sequencetypes] algorithm) to look for phylogenetic links in their recentevolutionary history (18). However, the two clusters did notlink even when the number of identical loci for group definitionwas decreased from the standard 6 out of 7 to 3 out of 7 (Fig.3).

View larger version (10K):
[in this window]
[in a new window]

FIG. 1. NJ phylogenetic tree of the B. cereus group population as represented in the B. cereus MLST database. The figure shows Clade 1, as designated by Priest, et al. (16). Arrows indicate the position of the STs of cereulide-producing strains analyzed in this study.

View larger version (19K):
[in this window]
[in a new window]

FIG. 2. Split decomposition analysis of the group of cereulide-nonproducing strains of B. cereus analyzed in this study.

View larger version (10K):
[in this window]
[in a new window]

FIG. 3. eBURST analysis of cereulide-producing strains of B. cereus analyzed in this study. FREQ, frequency (number of strains at each ST); SLV, single-locus variant; DLV, double-locus variant; TLV, triple-locus variant; SAT, satellites, more distantly related strains. Number of isolates, 24; number of STs, 4; number of resamplings for bootstrapping, 1,000; number of loci per isolate, 7; number of identical loci for group definition, 3; number of groups, 1. Group 1 number of isolates, 24; number of STs, 4; predicted founder, none.

DISCUSSION

Top

Discussion

Cereulide-nonproducing strains of B. cereus isolated from casesof food poisoning or the environment showed a high level ofgenetic diversity. Consistent with previously published results(16), they were distributed mainly in Clade 2, in close phylogeneticrelationship to Bacillus thuringiensis. On the contrary, publishedreports have thus far assigned all analyzed cereulide-producingstrains to a single phylogenetic cluster (9, 16). Based on MLSTanalysis, these strains have been assigned exclusively to ST26 (16). Our study identified a strain with a new single mutationin ST 26, creating the new variant ST 165. We also recoveredseveral cereulide-producing strains whose genetic sequenceswere at a location distinct from ST 26, in the lineage mostclosely related to the lineage populated by B. anthracis. Analysisby eBURST showed that this second cluster of cereulide-producingstrains represents an independent group, with no detectablephylogenetic relationship to the previously reported cluster.

Previous studies have based their conclusions on strains fromvarious European locations and on a few strains from North America(9, 16). To balance this, our collection of cereulide producersconsisted predominantly of Japanese strains, in an effort toshed some light on genetic diversity in a geographically distinctregion. Interestingly, all non-Japanese strains in our collection(total of three) fell into the newly identified cluster, togetherwith a few Japanese strains. This distribution shows the rarityof ST 144/ST 164 strains and the predominance of ST 26 emeticstrains worldwide.

Comparison of MLST analysis to flagellar typing provided a linkbetween phylogenetic allocation and serotype. All strains ofserotype H1 belonged to ST 26, while the newly identified clusterwas formed exclusively of strains with serotypes H3 and H12.Our findings provide a link between phylogenetic location andflagellar serotype insofar as the clusters are distinguishableby their flagellar H serotype. It appears not only that clusterST 26 (serotype H1) strains outnumber those of the new clusterST 144/ST 164 (serotype H3/H12) but also that cluster ST 26includes almost all food poisoning-associated strains. Thiscorrelates with the observations by Agata et al. that serotypeH1 strains produce higher amounts of cereulide (1).

Recently published complete sequences of the cereulide synthetasegene cluster (7) and plasmid profiling work in emetic strains(11) as well as our unpublished data have shown that the operonresponsible for cereulide production is situated on a largeplasmid. Thorsen et al. recently reported that B. weihenstephanensisstrains could produce cereulide (21). The fact that the capacityfor cereulide production is shared by two phylogenetically distinctB. cereus clusters and by B. weihenstephanensis strains suggeststhat the plasmid can be subjected to lateral transfer.

It has been hypothesized that cereulide-producing strains, likeB. anthracis, represent a recently emerged clone in the B. cereuspopulation (9). So far, no substantial heterogeneity has beenfound among strains in the B. anthracis population (10, 13).The data presented here suggest that cereulide-producing strainsare progressively diversifying.

ACKNOWLEDGMENTS

This work was supported by a Grant-in-Aid for Scientific Research(17590388, 14370090 and 16590356) and by Research for the Futureprogram from the Japanese Society for the Promotion of Science(JSPS-RFTF00L01411).

This publication made use of the Bacillus cereus Multi LocusSequence Typing Website (http://pubmlst.org/bcereus/) developedby Keith Jolley and sited at the University of Oxford (Jolleyet al. 2004; BMC Bioinformatics 5:86). The development of thissite has been funded by the Wellcome Trust.

FOOTNOTES

* Corresponding author. Mailing address: Department of Bacteriology, Nagoya University Graduate School of Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya 466-8550, Japan. Phone: 81-52-744 2106. Fax: 81-52-744 2107. E-mail: mohta{at}med.nagoya-u.ac.jp

Published ahead of print on 21 February 2007.

REFERENCES

Top