Pathogenomic Sequence Analysis of Bacillus cereus and Bacillus thuringiensis Isolates Closely Related to Bacillus anthracis

Abstract

Bacillus anthracis, Bacillus cereus, and Bacillus thuringiensis are closely related gram-positive, spore-forming bacteria of the B. cereus sensu lato group. While independently derived strains of B. anthracis reveal conspicuous sequence homogeneity, environmental isolates of B. cereus and B. thuringiensis exhibit extensive genetic diversity. Here we report the sequencing and comparative analysis of the genomes of two members of the B. cereus group, B. thuringiensis 97-27 subsp. konkukian serotype H34, isolated from a necrotic human wound, and B. cereus E33L, which was isolated from a swab of a zebra carcass in Namibia. These two strains, when analyzed by amplified fragment length polymorphism within a collection of over 300 of B. cereus, B. thuringiensis, and B. anthracis isolates, appear closely related to B. anthracis. The B. cereus E33L isolate appears to be the nearest relative to B. anthracis identified thus far. Whole-genome sequencing of B. thuringiensis 97-27and B. cereus E33L was undertaken to identify shared and unique genes among these isolates in comparison to the genomes of pathogenic strains B. anthracis Ames and B. cereus G9241 and nonpathogenic strains B. cereus ATCC 10987 and B. cereus ATCC 14579. Comparison of these genomes revealed differences in terms of virulence, metabolic competence, structural components, and regulatory mechanisms.

While Bacillus anthracis, Bacillus cereus, and Bacillus thuringiensis are closely related members of the B. cereus group (22), individual isolates exhibit differences in terms of host range and virulence. B. anthracis is the causal agent of anthrax, a zoonotic disease that can be lethal to humans. B. cereus is a ubiquitous soil organism and an opportunistic human pathogen most commonly associated with food poisoning (10). B. thuringiensis is an insect pathogen that is widely used as a biopesticide (36). Here we report the sequencing and comparative analysis of the genomes of two members of the B. cereus group, B. thuringiensis 97-27 subsp. konkukian serotype H34, isolated from a necrotic human wound (17), and B. cereus E33L, which was isolated from a swab of a zebra carcass in Namibia (P. C. B. Turnbull, personal communication). To facilitate the comparison of these two isolates with other members of the B. cereus group, we compiled a core genome of over 3,900 B. cereus group genes. Comparison of these genomes revealed differences in terms of virulence, metabolic competence, structural components, and regulatory mechanisms (see Table 1, below), supporting the idea that differential regulation modulates virulence rather than simple acquisition of virulence factor genes. Our analysis of the genome sequences of B. thuringiensis 97-27 and B. cereus E33L provides insight into the evolutionary relationships among these B. cereus group organisms, as well as the molecular mechanisms contributing to their host range and virulence.

TABLE 1.

Major phenotypic characteristics of B. cereus group genomes

Characteristic	B. anthracis Ames	B. cereus ATCC 10987	B. cereus ATCC 14579	B. thuringiensis 97-27	B. cereus E33L
Plasmid	pXO1 (189 kb), pXO2 (96 kb)	pBc10987 (208 kb)	Linear phage-like pBClin15 (15 kb)	pBT9727 (77 kb)	pE33L466 (466 kb), pE33L54 (54 kb), pE33L9 (9 kb), pE33L8 (8 kb), pE33L5 (5 kb)
Tripartite lethal toxin	Present	Absent	Absent	Absent	Absent
rRNA	33 copies	12 copies	39 copies	39 copies	39 copies
Urease gene cluster	Absent	Present	Absent	Absent	Absent
Xylose utilization genes	Absent	Present	Absent	Absent	Absent
Capsule synthesis
Polysaccharide capsule	Absent	Presenta	Presenta	Presenta	Absent
Polyglutamic acid capsule	Presenta	Absent	Absent	Absent	Absent
Flagellar genes	Fragmental	Intact	Intact	Intact	Intact
N-Acetylgalactosamine degradation	Absent	Present	Absent	Absent	Present
Functional PlcR	Absent	Presentb	Presentb	Presentb	Presentb
Phaged	4	4 (1 degenerate)	6 (1 linear plasmid)	7	18
Arginine degradation
Arginase genes	Present	Present	Present	Present	Present
Arginine deiminase genes	Absent	Present	Present	Present	Absent
B.cereus repeat 1 (bcr1)e	10 copies	72 copies	56 copies	19 copies	22 copies
Restriction enzymes	Absent	Presentc	Presentc	Partial	Absent
Enterotoxins
Hemolytic enterotoxin HBL	Absent	Present	Absent	Present	Absent
Nonhemolytic enterotoxin NHE	Present	Present	Present	Present	Present
Diarrheal toxin BceT	Absent	Absent	Present	Present	Absent
cry genes	Absent	Absent	Absent	Absent	Absent

^aThe capsule biosynthesis region of 20 kb is partially similar in both B. cereus genomes.

^bplcR in B. anthracis Ames contains a frameshift that results in a truncated and nonfunctional protein. The plcR gene in the B. cereus strains is full length, apparently functional, and can act as a regulatory protein.

^cThere are four unique restriction-modification systems in B. cereus ATCC 10987 and three in B. cereus ATCC 14579. B. thuringiensis 97-27 and B. anthracis have a CDS weakly similar to the 5-methylcytosine-specific Mrr endonuclease.

^dThe phages are not conserved in sequence or genomic location in the genomes studied.

^ebcr1 is a 160-bp repeated DNA sequence with unknown function overwhelmingly overrepresented in intergenic regions of the B. cereus group organisms.

MATERIALS AND METHODS

Sequencing of the B. thuringiensis 97-27 and B. cereus E33L genomes.

The random shotgun method of cloning, sequencing, and assembly was used. Large (40-kb; B. thuringiensis 97-27 only), median (8-kb), and small (2.5- to 3.5-kb) insert libraries were sequenced for these genome projects with an average success rate of 90% and average high-quality read lengths of 643 and 621 nucleotides for B. thuringiensis 97-27 and B. cereus E33L, respectively. The completed genome sequences of B. thuringiensis 97-27 and B. cereus E33L contained 134,054 and 141,352 reads, respectively, achieving an average of 19.3- and 18.7-fold sequence coverage per base. After assembly, gaps between contigs were closed by editing, primer walking library clones, or PCR amplifications.

Annotation.

Gene predictions were obtained using Glimmer (7, 35), and tRNAs were identified using tRNAScan-SE (25). Basic analysis of the gene predictions was performed by comparing coding sequences against the PFam, BLOCKS, and Prodom databases. Gene definitions and functional classes were added manually by a team of annotators using BLAST results in addition to information from the basic analysis.

Sequence analysis.

We compared the genomes at the nucleotide level using genome alignment tools such as MUMmer2 (8), ACT (http://www.sanger.ac.uk/Software/ACT/), and Pipmaker (37). To obtain a list of orthologs in the B. thuringiensis 97-27 and B. cereus E33L genomes, we wrote a perl script that determines bidirectional best hits as follows. Genes g and h are considered orthologs if h is the best BLASTP hit for g and vice versa, with e-values less than or equal to 10⁻¹⁵. A gene is considered strain specific if it has no hits with an e-value of 10⁻¹⁵ or less.

To identify IS elements in B. thuringiensis 97-27 and B. cereus E33L and compare them to IS elements present in other B. cereus group members, all known IS elements were used as query sequences and used with BLAST against the genomes of three strains of B. anthracis (Ames, A2012, and Sterne), B. thuringiensis 97-27, B. cereus E33L, and B. cereus (ATCC 14579).

Tandem repeats were identified in B. thuringiensis 97-27 and B. cereus E33L genomes using the Tandem Repeats Finder (4) with the threshold set for a minimum alignment score of 50.

AFLP.

Amplified fragment length polymorphism (AFLP) analysis of the microbial DNAs was accomplished as previously described (18). Briefly, each of the DNA preparations was digested with EcoRI and MseI, and the resulting fragments were ligated to double-stranded adapters and then amplified by PCR using +0/+0 primers. Selective amplifications using the +1/+1 primer combination of 6-carboxyfluorescein-labeled EcoRI-C and MseI-G resulted in products that were mixed with a solution containing DNA size standards (Genescan-500 from Applied Biosystems Inc., Foster City, CA; and MapMarker-400 from BioVentures, Inc., Murfreesburo, TN), both labeled with N,N,N,N-tetramethyl-6-carboxyrhodamine. Following a 2-min heat denaturation at 90°C, the reaction mixtures were loaded onto a 5% Long Ranger DNA sequencing gel (Cambrex Bio Science, Rockland, ME) and visualized on an ABI 377 automated fluorescent sequencer (Applied Biosystems, Inc., Foster City, CA). Each set of AFLP experiments also included as a sample B. anthracis Vollum DNA, which was used as an internal control to allow comparison of results from different gels run at different times. GeneScan analysis software (Applied Biosystems, Inc., Foster City, CA) was used to determine the lengths of the sample fragments by comparison to the DNA fragment length size standards included within each sample.

Data analysis of the microbial DNAs was as previously described (39). DNA fragment sizes between 100 and 500 bp from triplicate data (derived from three lanes from three different gels) for each sample were combined. Fragment sizes that appeared in all three replicates were used to represent the sample, and the peak heights for the fragment sizes were averaged. This “averaged” sample was then used to compare to other “averaged” samples. A hierarchical agglomerative clustering routine using group averages was used to determine which fragments among the samples had similar lengths. A decision rule was added to this clustering routine that limited the allowable number of fragments within a cluster to equal the number of samples being compared and limited the maximum acceptable range of fragments sizes for a cluster to a preset value. Similarities between samples were measured using the Jaccard coefficient. Dendrograms were produced using the similarity matrix and the unweighted pair-group mean average method (F. J. Rohlf, NTSYS-PC numerical taxonomy and multivariate analysis system, version 1.8; Exeter Software, Setauket, N.Y.).

Nucleotide sequence accession numbers.

The sequences of the B. thuringiensis 97-27 and B. cereus E33L genomes and plasmids can be accessed using the GenBank accession numbers AE017355, CP000001, CP000040, CP000041, CP000042, CP000043, CP000044, and CP000047.

RESULTS

General genome features.

The 5.31-Mb genome of B. thuringiensis 97-27 comprises two replicons: a circular chromosome, encoding at least 5,198 open reading frames, and the pBT9727 plasmid (see Fig. S1A in the supplemental material). The 5.84-Mb genome of B. cereus E33L comprises six replicons: a circular chromosome, encoding at least 5,682 open reading frames, and five plasmids (see Fig. S1B in the supplemental material). B. thuringiensis 97-27 and B. cereus E33L have broad similarities to and share a high degree of synteny with B. anthracis Ames (33), B. cereus ATCC 14579 (21), and B. cereus ATCC 10987 (32). Within the B. cereus group, B. anthracis, B. cereus E33L, and B. thuringiensis 97-27 are part of a distinct cluster which contains many pathogenic organisms (18) (Fig. 1).

FIG. 1.

An AFLP-based tree of B. anthracis, B. cereus, and B. thuringiensis isolates. These 48 isolates are representative of the branches identified when over 300 isolates of B. anthracis, B. thuringiensis, and B. cereus were examined by AFLP. Yellow highlighted isolates have fully sequenced genomes, blue indicates B. thuringiensis isolates, black shows B. cereus isolates, and red indicates B. anthracis isolates.

As illustrated in Fig. S2 in the supplemental material, a total of 3,917 putative proteins are shared among B. anthracis Ames, B. cereus ATCC 14579, B. thuringiensis 97-27, and B. cereus E33L using as a criterion whether genes were bidirectional best hits in BLAST searches. Comparison of the genomes of B. cereus E33L and B. thuringiensis 97-27 with B. anthracis Ames and B. cereus 14579 also identified strain-specific genes in each organism. Of the 5,682 predicted B. cereus E33L proteins, 253 of the chromosomally encoded genes and 416 of the plasmid genes are unique. Of the 5,197 predicted B. thuringiensis 97-27 proteins, 307 of the chromosomally encoded genes and 66 of the plasmid genes are unique. B. cereus E33L and B. anthracis Ames are the closest pair and share the highest number (221) of common proteins.

Virulence genes.

The chromosomally encoded virulence genes in B. thuringiensis 97-27 and B. cereus E33L are common to the B. cereus group of bacteria (14). Neither B. thuringiensis 97-27 nor B. cereus E33L has the highly characterized B. anthracis toxin genes (pag, lef, and cya) encoded on pX01 or the cap genes encoded by pXO2 (26, 27). Nonetheless, our results indicate that both B. thuringiensis 97-27 and B. cereus E33L share a set of virulence factors common to the members of the B. cereus group. These common virulence genes include the three nonhemolytic enterotoxin genes (nheABC), two channel-forming type III hemolysins, a perfringolysin O (listeriolysin O), a phosphatidyl-inositol-specific and a phosphatidyl-choline-preferring phospholipase, RNA polymerase sigma-B factor, and a p60 family extracellular protease. These last five genes are homologous to virulence genes encoded by the gram-positive pathogen Listeria monocytogenes 12. B. thuringiensis 97-27 and B. cereus E33L also have a gene encoding cytotoxin K, which was previously identified in B. cereus (ATCC 14579). While B. cereus E33L lacks the hbl operon, which is suspected as a primary factor in diarrheal B. cereus food poisoning (3), B. thuringiensis 97-27 has the hbl operon containing the hemolytic enterotoxin genes hblCDBA also found in B. cereus (ATCC 14579) (Fig. 2A). Interestingly, the hbl gene cluster consists of the hemolytic enterotoxin (hblCDBA) and other genes encoding the spore germination proteins gerIABC, as well as other related proteins, that are ordered and oriented in a way that suggests their expression is coordinated by the transcriptional regulator TrrA. This gene cluster is part of a large, approximately 17.7-kb, 11-gene insertion (Fig. 2B). A degenerate ISRso11 transposase fragment is found at the presumed insertion boundary region, and direct repeats that overlap with the C-terminal UvrC-like protein were identified. This observation suggests a mechanism for the acquisition of these virulence factors in B. thuringiensis 97-27, B. cereus 14579, and B. cereus G9241.

FIG. 2.

A. Comparison of the gerI and hbl operon regions in B. cereus, B. thuringiensis, and B. anthracis. The light blue area between the two groups indicates that these regions share a high level of identity. A conserved region consists of five contiguous genes in B. anthracis Ames, including l-asparaginase (BA3137), ans operon repressor (BA3138), degenerate ISRso11 transposase (BA3139), UvrC-like protein (BA3140), and amino acid permease (BA3141). B. Flanking region of insertion boundary. The orthologs of genes are shown as arrows of the same color. BT9727_2896/BA3140 encode UvrC-like proteins. BT9727_2885.1/BA3139 encode degenerate ISRso11 transposase. Yellow blocks denote the direct repeats found around the insertion boundary. The red triangle indicates the genes of the C-terminal UvrC-like protein fragment.

The opportunistic pathogenicity of B. cereus and B. thuringiensis may depend on the secretion of nonspecific extracellular virulence factors in response to transcriptional activation by PlcR (34). However, in all B. anthracis strains, the plcR gene is inactivated by a frameshift mutation which creates an early stop codon (1). In other B. cereus isolates, the plcR gene product up regulates the transcription of genes encoding enterotoxins, proteases, phospholipases, metabolic enzymes, proteins involved in motility and chemotaxis, proteins involved in sporulation, DNA metabolism, transcriptional regulators, and a variety of transporters by binding to a specific upstream motif (1, 21, 34). The genes encoding PlcR appear intact in B. thuringiensis 97-27 and in B. cereus E33L. Analyzing the upstream sequences of coding regions for PlcR binding motifs identified genes likely to be activated by PlcR in B. thuringiensis 97-27 and B. cereus E33L. We found motifs upstream of most of the genes previously identified as potential members of a plcR regulon in B. cereus (21) (see Table S1 in the supplemental material). Of particular interest are genes encoding probable virulence factors. In this respect, we found that the nonhemolytic enterotoxin genes (nheA, nheB, and nheC) in both B. thuringiensis 97-27 and B. cereus E33L contained upstream PlcR motifs. In B. thuringiensis 97-27, B. cereus E33L, and B. cereus (ATCC 14579), there are PlcR motifs upstream of cytotoxin K and several proteases, including collagenase, bacillolysin, enhancin, aminopeptidase Y, and peptidase T. In addition, we found PlcR motifs upstream of the phospholipase C and phosphatidylinositol-specific phospholipase C genes in all three genomes. Another gene that has an upstream PlcR motif in B. thuringiensis 97-27, B. cereus E33L, and B. cereus (ATCC 14579) is error-prone DNA polymerase IV. This gene was previously suggested to induce adaptive point mutations that may affect pathogenicity. These observations (21) support the hypothesis that differences in virulence among B. anthracis, B. cereus, and B. thuringiensis are predominately due to alterations in gene expression rather than simple gain or loss of gene functions.

As B. thuringiensis isolates usually contain cry, cyt, and/or vip genes encoding insecticidal crystalline toxins, we compiled a list of 131 cry and cyt gene sequences (6), as well as the sequences of vip3A (11), vip3V (9), vip1Ac and vip2Ac (38), and other more recently identified cry gene sequences and blasted these sequences against the B. thuringiensis 97-27 genome. There were no full-length hits to any of the query sequences. For the most part, the partial hits had low identities (under 40 to 50%) with the query sequences. Manual examination of the annotated B. thuringiensis 97-27 genes that were partial hits and further analysis of these genes did not reveal any obvious candidates for cry, cyt, or vip genes. Most of these B. thuringiensis 97-27 genes had >80% amino acid sequence identity to other B. cereus group (noninsecticidal) genes. So, we are confident that the genome of our current isolate contains no homologs of the known cry, cyt, or vip genes. However, another possibility is that the plasmid encoding these genes was lost during culture.

Capsule biosynthetic genes.

Many microbial pathogens produce polymeric capsules that provide protection against host immune systems during the invasion process. Bacillus species can produce both polysaccharide capsules that are common to many gram-positive and gram-negative species and the less common polyglutamic acid capsule. A summary of the capsule biosynthetic content in the sequenced members of the B. cereus group is provided in the supplemental material (see Fig. S3). In B. anthracis, three pXO2-encoded genes, capB, capC, and capA, are required for synthesis of the polyglutamic acid capsule, and this structure plays a key role in the virulence of this organism (12). To date, all B. cereus group strains appear to contain a weak homolog of the pXO2 capA gene. This is also true in B. cereus E33L, which contains a putative protein with 32% identity to capA. However, B. cereus E33L does not have a polyglutamic acid capsule (P. C. B. Turnbull, personal communication), nor does it appear to encode any genes involved in polysaccharide capsule synthesis (Table 1). Interestingly, the B. thuringiensis 97-27 genome and B. cereus 14579 encode a homolog of a member of a polysaccharide capsule synthesis pathway recently identified on a plasmid of the pathogenic B. cereus G9241 (19).

Sporulation and germination.

In anthrax, spores are the agent of infection. Spore formation occurs in response to nutrient limitation in the environment. In B. subtilis, sporulation is initiated by a deficiency in carbon or nitrogen (29) and is linked to changes in the expression of genes for degradative enzymes, such as alpha amylase, neutral protease, and alkaline protease (20). The B. subtilis spore coat is composed of at least 30 polypeptides, homologs of many of the B. subtilis spore coat protein genes present in B. anthracis (23). We found differences in the number and composition of genes encoding spore coat proteins among B. thuringiensis 97-27, B. cereus E33L, B. cereus ATCC 14579, and B. anthracis (see Table S2 in the supplemental material). Both B. anthracis and B. cereus spores germinate in response to l-alanine and ribosides. The germination response to l-alanine and ribosides requires proteins of the gerA family (5). B. thuringiensis 97-27, B. cereus E33L, B. cereus 569, and B. cereus ATCC 14579 have a gerI operon that is involved in an inosine-induced germination. The gerI operon is homologous to the gerA family operons of B. subtilis (5) and the gerH operon in B. anthracis (41). Similarly, the gerQ operon encodes germinant receptors that respond to inosine (2). B. thuringiensis 97-27 encodes gerQ, while B. cereus E33L does not.

Carbohydrate and amino acid utilization.

Like B. cereus 14579 and B. anthracis (21, 33), B. thuringiensis 97-27 and B. cereus E33L appear predisposed to an environment rich in protein, having fewer genes for carbohydrate catabolism and more genes for amino acid metabolism (Table 2). For example, there are 12 carbohydrate polymer degradation genes in B. anthracis Ames, B. cereus 14579, and B. thuringiensis 97-27 and 23 in B. cereus E33L, compared to 41 in B. subtilis. The B. cereus group also appears to have reduced numbers of sugar-specific phosphoenolpyruvate-dependent phosphotransferase system genes. In contrast, members of the B. cereus group have an expanded capacity for amino acid and peptide utilization. For example, there are 18 to 23 genes encoding peptide/amino acid ABC transporter-ATP binding proteins in the B. cereus group, compared to 7 in B. subtilis. There are six to nine genes for LysE family amino acid efflux system proteins in B. cereus group members and only two in B. subtilis. In addition to the expanded number of peptidase and protease genes in B. cereus group species, 52 to 55 genes encode proteins involved in amino acid and amine catabolic pathways compared to 34 in B. subtilis. These observations suggest that proteins, peptides, and amino acids may be a preferred nutrient source for all members of the B. cereus group, which is consistent with the observations made previously (21, 32).

TABLE 2.

Protein and sugar utilization genes in the B. cereus group

Utilization and function	No. of genesa
	B. subtilis	B. anthracis Ames	B. cereus 14579	B. cereus E33L	B. thuringiensis 97-27
Amino acid and peptide utilization
Peptide ABC transporter-ATP binding protein	7	18	23	20	18
Branched chain amino acid transporter	4	11	11	13	14
LysE/RhtB/CadD amino acid efflux system	2	6	8	9	8
Peptidase	30	64	91	92	90
Protease	24	50	49	61	52
Amino acid and amine catabolism	34	52	55	55	55
(BA0242) tyrosine degradation	No	Yes	Yes	Yes	Yes
Epr Bpr AprX protease	Yes	No	No	No	No
Sugar utilization
PTS-sugar transporter	25	19	18	23	20
Carbohydrate polymer degradation	41	12	12	23	12
Mannose, arabinose, rhamnose catabolic pathway	Yes	No	No	No	No

^aNo and yes indicate the absence and presence, respectively, of genes in the pathway.

Although the B. cereus group species are closely related, variations in the sugar catabolism pathways are observed. Of particular note, B. cereus E33L has 11 extra genes for carbohydrate polymer degradation compared with B. anthracis Ames. Most of these are located on the large plasmid pE33L466. One of the most significant differences between B. cereus E33L and other isolates in the Bacillus cereus group is the large number of carbohydrate utilization gene clusters organized as operons on the pE33L466 plasmid (Fig. 3). These genes encode enzymes for myo-inositol degradation, galactose utilization, and pectin and gellan degradation. It is worth noting that the region of pE33L466 containing these four interlocked gene clusters is flanked by IS elements that were probably involved in their mobilization and integration into pE33L466. Figure 3 illustrates the metabolic pathways in which the products of these genes participate.

FIG. 3.

Schematic presentation of phosphotransferase system-catalyzed sugar uptake and phosphorylation in B. cereus E33L, showing possible metabolic pathways catalyzed by the products of genes in this polymorphic locus. Steps along the pathways are catalyzed by the gene products specified near the corresponding arrow.

Antibiotic resistance.

The ecological niche and potential virulence of B. thuringiensis 97-27 and B. cereus E33L may be expanded through the presence of two lantibiotic resistance operons that are not present in B. cereus or B. anthracis. These include a mersacidin resistance operon consisting of mrsR2, mrsK2, mrsF, mrsG, and mrsE and a salivaricin resistance operon consisting of salY, salK, and salR. B. thuringiensis 97-27 and B. cereus E33L have all of the genes in the mersacidin operon, while B. anthracis strains A2012, Ames, and Sterne only have mrsF. Although B. thuringiensis 97-27 and B. cereus E33L have all of the genes in the salivaricin and mersacidin resistance operons, they do not encode the mrsA gene to produce mersacidin or the salA gene to produce salivaricin. Therefore, these organisms can detect the presence of mersacidin and salivaricin produced by other bacteria but do not encode the capability to produce these lantibiotics themselves. Instead, the response may include increased expression of genes encoding other lantibiotics or virulence factors as previously suggested (40).

DISCUSSION

There is considerable debate in regard to the systematic classification of members of the B. cereus group. Historically, these organisms were classified into three species (B. cereus, B. thuringiensis, and B. anthracis) on the basis of distinct phenotypic differences that defined them. For example, the isolation of an organism from an animal with anthrax resulted in the designation of B. anthracis. While the relationship between these organisms is still not clearly understood, recent molecular approaches (15, 18, 31, 33) have revealed extensive similarities between genomes and relatively few consistent differences warranting the segregation of isolates into discrete species classified as B. anthracis, B. cereus, and B. thuringiensis. One unifying concept that has emerged from nucleic acid sequence analyses is that the B. cereus group has evolved as asexually derived clonal populations (15, 18, 28, 30). This has allowed most of the vast number of isolates from this group to be subdivided into consistent phylogenetic clusters.

In this classification scheme (30), B. thuringiensis 97-27 and B. cereus E33L are both members of the anthracis lineage and are descended from ancestral clones that are very distinct from the tolworthi, kurstaki, sotto, and thuringiensis lineages. Importantly, the anthracis lineage provides a molecular-based distinction that separates commercially important B. thuringiensis strains from pathogenic B. anthracis.

The B. thuringiensis 97-27 chromosome and plasmid lacked the typical cry, cyt, and vip genes encoding the insecticidal proteins characteristic of strains that are known to produce entomopathic toxins. The original B. thuringiensis designation for this isolate was due to the discovery of crystals in the initial characterization of the strain (17). However, in a subsequent publication, a second isolate (from the same patient) lacking crystalline toxin was mentioned as a spontaneous mutant (16). It is possible that the plasmid(s) encoding the crystalline toxin genes was lost spontaneously during culture; this has been documented for other B. thuringiensis plasmids (13, 24). Certainly, B. thuringiensis 97-27 is distinct from other known B. thuringiensis isolates in that it is suspected of causing human morbidity (17) resulting in severe tissue necrosis. It was subsequently demonstrated to cause lethal infection in laboratory mice (16). The phylogenetic lineage placement, subsequent laboratory diagnostics, and our comparative sequence analysis suggest that B. thuringiensis 97-27 is more like a pathogenic B. cereus strain than an insecticidal strain.

Both B. thuringiensis 97-27 and B. cereus E33L have homologs of chromosomal virulence genes found in other members of the B. cereus group. Consequently, the isolation of B. thuringiensis 97-27 from a rare case of disease and the presence of common B. cereus group chromosomal virulence genes make it likely that this organism is an opportunistic pathogen. While B. cereus E33L came from a carcass swab, it is probably an environmental isolate and not the cause of death. The relationships between members of the B. cereus group are nonlinear and complex, likely resulting from cycles of isolation and niche expansion facilitated, at least in part, by horizontal gene transfer mechanisms. While the germination of Bacillus anthracis spores or its vegetative growth may be limited to nutritionally rich environments like that found in a mammalian host, the rapid death of the host resulting from vegetative growth would limit the opportunity for genetic exchange and would result in the homogeneity observed in sequenced strains of this species. In contrast, the capacity for vegetative growth outside of an infected host or nonlethal infection provides an opportunity for genetic exchange and niche expansion. The sequences of the two B. cereus group members presented here provide fertile ground to study the evolution of host range and virulence.