Genetic relationship in the ‘Bacillus cereus group’ by rep-PCR fingerprinting and sequencing of a Bacillus anthracis-specific rep-PCR fragment

Authors


Daniele Daffonchio, Dipartimento di Scienze e Tecnologie Alimentari e Microbiologiche, Università degli Studi di Milano, via Celoria 2, 20133, Milano, Italy (e-mail: daniele.daffonchio@unimi.it).

Abstract

Aims: To evaluate the genetic relationship in the Bacillus cereus group by rep-PCR fingerprinting.

Methods and Results: A collection of 112 strains of the six species of the B. cereus group was analysed by rep-PCR fingerprinting using the BOX-A1R primer. A relative genetic distinctness was found among the species. Cluster analysis of the rep-PCR patterns showed clusters of B. thuringiensis strains quite separate from those of B. cereus strains. The B. anthracis strains represented an independent lineage in a B. cereus cluster. The B. mycoides, B. pseudomycoides and B. weihenstephanensis strains were clustered into three groups at some distance from the other species. Comparison of sequences of AC-390, a typical B. anthracis rep-PCR fragment, from 27 strains of B. anthracis, B. cereus, B. thuringiensis and B. weihenstephanensis, representative of different clusters identified by rep-PCR fingerprinting, confirmed that B. anthracis diverges from its related species.

Conclusions: The genetic relationship deduced from the rep-PCR patterns indicates a relatively clear separation of the six species, suggesting that they can indeed be considered as separate units.

Significance and Impact of the Study: rep-PCR fingerprinting can make a contribution in the clarification of the genetic relationships between the species of the B. cereus group.

Introduction

The genetic relationship of the species of the Bacillus cereus group is a subject of debate, and is still far from clear. Five species of the group, B. anthracis, B. cereus, B. mycoides, B. thuringiensis and B. weihenstephanensis (Lechner et al. 1998), have a marked impact on human activity, three with negative effects and two with positive. Bacillus anthracis and B. cereus are well-known pathogens of mammals, including humans, and B. weihenstephanensis grows at temperatures as low as 4°C, posing a threat to the conservation of cold-stored food (Mayr et al. 1999). The most useful species of the group, B. thuringiensis, is a biological insecticide used extensively throughout the world (Schnepf et al. 1998), while B. mycoides strains can improve plant growth (Petersen et al. 1995). A sixth species, B. pseudomycoides, has been described (Nakamura 1998).

Given the effect, the five above mentioned bacteria have on human activity, and the wide application of B. thuringiensis in the field, a clear evaluation of the taxonomic and phylogenetic relationship of the species is necessary. Several authors have concluded that B. anthracis, B. cereus and B. thuringiensis belong to a single species, a conclusion reached on the basis of multilocus enzyme electrophoresis, the sequencing of discrete protein-coding genes (Helgason et al. 2000) and the presence of an S-layer (Mignot et al. 2001). An analysis based on the heteroduplexes formed between the 16S–23S rRNA internal transcribed spacers (ITS) has shown that B. thuringiensis can be differentiated from B. cereus, and that B. anthracis represents an independent lineage diverging from B. cereus (Daffonchio et al. 2000; Cherif et al. 2003).

In the rep-PCR fingerprint typing method (Versalovic et al. 1991; Martin et al. 1994; Versalovic et al. 1994), a single primer targets repetitive regions scattered throughout the bacterial genome to give a PCR product profile; such a profile is generally specific to a given strain. Repetitive regions present higher variability than other genomic regions, and can be used to analyse the genetic relationship between strains (Van Belkum et al. 1998; Kim et al. 2002). In fact, long-range rep-PCR enabled Brumlik et al. (2001) to distinguish 105 B. anthracis strains from related species of the B. cereus group and, recently, rep-PCR plus primer BOX-A1R was used to identify a specific DNA marker for B. anthracis (Cherif et al. 2002). Also Kim et al. (2002) used BOX-A1R-based rep-PCR to establish the genetic relationship between 17 strains of the B. cereus group. They found that B. anthracis could be separated from the closely related species, but the genetic relationship among the other species of the B. cereus group was not well described, because of the small number of strains.

The aim of this work was to establish the genetic relationship between the species of the B. cereus group, basing the work on the rep-PCR fingerprinting of 112 strains of the six species. Strains of the species B. anthracis, B. cereus, B. thuringiensis and B. weihenstephanensis selected from different rep-PCR fingerprinting clusters were further characterized by sequencing AC-390, a rep-PCR fragment typical of B. anthracis identified in previous work as homologous to ywfK a trascriptional regulator of B. subtilis (Cherif et al. 2002).

Materials and methods

Bacterial strains, DNA extraction, rep-PCR and pattern cluster analysis

A total of 112 strains of the six species of the B. cereus group were used (Table 1). Strain cultivation and DNA extraction procedures are reported elsewhere (Borin et al. 1997; Daffonchio et al. 1998a,b, 1999a,b, 2000; Cherif et al. 2002).

Table 1.  Strains analysed in this study, rep-PCR pattern type number and relevant characteristics
Species (number of strains)rep-PCR haplotype number*Strains with the same rep-PCR haplotype Strain source†Relevant characteristic(s) of strain(s) [reference(s)]
  1. *rep-PCR haplotype number represented in Figs 1 and 2.

  2. †Institution from which the strains were obtained. BGSC, Bacillus Genetic Stock Center; B. thuringiensis strains HD2 and HD868 were kindly provided by D. R. Zeigler; CBS, Centre de Biotechnologie de Sfax, Tunisia; BUPM strains were kindly provided by S. Jaoua; DBS-UJ, Department of Biological Sciences, University of Jordan, Amman, Jordan; DISTAM, Dipartimento di Scienze e Tecnologie Alimentari e Microbiologiche, Milan, Italy; DSMZ, Deutsche Sammlung von Mikroorganismen unt Zellkulturen, GmbH, Braunschweig, Germany; IPP, Institute Pasteur, Paris, France; total DNA of B. anthracis strain was kindly provided by M. Mock; LMT, Laboratoire de Microbiologie, Faculté des Sciences de Tunis, Tunisia; NRRL, Agricultural Research Service Culture Collection, Peoria, IL, USA; strains were kindly provided by L. K. Nakamura; WSBC, Weihenstephan Bacillus Collection, Weihenstephan, Germany; B. weihenstephanensis strains were kindly provided by S. Scherer.

Bacillus anthracis (17)50779IPPpXO1/2+ (Ramisse et al. 1996); strains isolated in France in 1984 (Patra et al. 2002)
51582IPPpXO1+/2+ (Ramisse et al. 1996); strains isolated in France (Patra et al. 2002)
52300IPPpXO1+/2+ (Ramisse et al. 1996)
53256, 282, 846IPPpXO1+/2+ (Ramisse et al. 1996); strains isolated in the French Pyrenees in 1994 (256, 846) and 1995 (Patra et al. 2002)
54227, Cepanzo, 832, 957, Davis TE702, 6602IPPStrains 227 and 832 were pXO1+/2+, 957 and Cepanzo pXO1+/2, 6602 and Davis TE702 pXO1/2+ (Ramisse et al. 1996); strains 227 and 957 were vaccinal strains isolated in France (Patra et al. 2002); strain 832 was isolated in France in 1979 (Patra et al. 2002)
55  376IPPpXO1+/2+ (Ramisse et al. 1996)
567700IPPpXO1/2 (Ramisse et al. 1996)
576769IPPpXO1+/2+ (Ramisse et al. 1996); strain isolated in French Pyrenees and Alps in 1997 (Patra et al. 1998)
584229IPPpXO1/2+ (Ramisse et al. 1996)
59663IPPpXO1+/2+ (Ramisse et al. 1996); strain isolated in France in 1984 (Patra et al. 2002)
Bacillus cereus (19)6MydDISTAMNot available
11360DSMZIsolated from garden soil
12cer1DISTAMNot available
13bc2DISTAMNot available
14626DSMZProduces l-leucine dehydrogenase
1531TDSMZB. cereus type strain
16cer5DISTAMIsolated from rice (Daffonchio et al. 1998b)
17cer3DISTAMIsolated from candies (Daffonchio et al. 1998b)
18my1DISTAMIsolated from ultrahigh-temperature milk (Daffonchio et al. 1998b)
19IO200LMTIsolated from Norwegian sea
20PO1DISTAMIsolated from ultrahigh-temperature milk (Daffonchio et al. 1998b)
21  345DSMZGrowth at 7°C (Daffonchio et al. 1999a)
4746321, 6127DSMZStrain 6127 produces penicillinase
48  487DSMZNot available
49cer4DISTAMIsolated from rice (Daffonchio et al. 1998b)
61  351DSMZIsolated from soil
63bc1DISTAMIsolated from marble, Venice, Italy (Daffonchio et al. 1998b)
74cer6DISTAMIsolated from tomato sauce (Daffonchio et al. 1998b)
Bacillus mycoides (21)1NRS306NRRLIsolated from soil (Nakamura and Jackson 1995)
2NRS319NRRLIsolated from soil (Nakamura and Jackson 1995)
3MycHDISTAMNot available
4303, TP2DSMZ, DISTAMIsolated from soil. Strain 303 grows clockwise on acid and on alkaline media
5BmMedDISTAMIsolated from alkaline soil, Italy
8299DSMZIsolated from soil; growth anticlockwise at pH 7 and clockwise at pH 5
92048TDSMZB. mycoides type strain; growth at 7°C (Daffonchio et al. 1999a)
10B14828NRRLNot available
76BmFDISTAMNot available
77309DSMZProduces a yellow diffusible pigment
78384DSMZGrowth at 7°C (Daffonchio et al. 1999a)
79A81DISTAMIsolated from soil, Italy
80BifDISTAMIsolated from soil, Italy
81NdrDISTAMIsolated from soil, Italy
82BmSDISTAMNot available
83B615NRRLIsolated from soil (Nakamura and Jackson 1995)
90G2DISTAMIsolated from garden soil, Italy
91Nov2DISTAMIsolated from maize rhizosphere, Italy
92Nov1DISTAMIsolated from maize rhizosphere, Italy
96G1DISTAMIsolated from garden soil, Italy
Bacillus pseudomycoides (8)84A82DISTAMIsolated from soil, Italy
85BD10NRRLIsolated from soil (Nakamura and Jackson 1995)
86BD14NRRLIsolated from soil (Nakamura and Jackson 1995)
89B346NRRLIsolated from soil (Nakamura and Jackson 1995)
93TP1DISTAMIsolated from soil, Italy
94B617TNRRLB. pseudomycoides type strain (Nakamura 1998)
95CADISTAMIsolated from soil, Italy
97B618NRRLIsolated from soil (Nakamura and Jackson 1995)
Bacillus thuringiensis (43)22BMG1.6LMTIsolated from soil, Tunisia
23Bt13LMTB. thuringiensis subsp. pakistani
24Bt55LMTB. thuringiensis subsp. galleriae
25BX16LMTIsolated from hypersaline soil in Tunisia; displays antifungal activity against Fusarium sp.
26BUPM21, BUPM22, BUPM23, BUPM25, BUPM26CBSIsolated from soil, Sfax region, Tunisia
27BMG1.9LMTIsolated from soil, Tunisia
28BMG1.7LMTIsolated from forest soil, Tunisia; produces the bacteriocin thuricin 7 (Cherif et al. 2001)
29Bt33, HD1, AlDISTAMBt33 and HD1 B. thuringiensis subsp. kurstaki
30Bt9LMTB. thuringiensis subsp. tolworthi
31Bt7LMTB. thuringiensis subsp. aizawai
32Bt1LMTB. thuringiensis subsp. thuringiensis
33BUPM30CBSIsolated from soil, Sfax region, Tunisia
345724DSMZIsolated from commercial B. thuringiensis israelensis product
355725DSMZIsolated from commercial B. thuringiensis kurstaki product.
36BMG1.1LMTIsolated from soil, Tunisia
37Bt44LMTB. thuringiensis subsp. dendrolimus
38Bt14LMTB. thuringiensis subsp. israelensis
392046TDSMZB. thuringiensis type strain subsp. thuringiensis
40BMG1.2LMTIsolated from soil, Tunisia
41Bt10LMTB. thuringiensis subsp. darmstadiensis
42BMG1.4LMTIsolated from surface water, Tunisia
43BMG1.3LMTIsolated from surface water, Tunisia
44BMG1.5LMTIsolated from surface water, Tunisia
46BMG1.8LMTIsolated from soil, Tunisia
60Hc13DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from dead insects (Khyami-Horani et al. 1996, 1999)
62Bt19BGSCOriginal strain ID: HD868; B. thuringiensis subsp. tochigiensis; produces the bacteriocin tochicin (Paik et al. 1997)
64Ht39DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from water (Khyami-Horani et al. 1996, 1999)
65Hc24DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from chicken faeces (Khyami-Horani et al. 1996, 1999)
66Hc32DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from infested plant leaves (Khyami-Horani et al. 1996, 1999)
67Hc35DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from water (Khyami-Horani et al. 1996, 1999)
68Hc17DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from chicken faeces (Khyami-Horani et al. 1996, 1999)
69Hc15DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from dead insects (Khyami-Horani et al. 1996, 1999)
70Hc45DBS-UJDisplays entomocidal activity against Culiseta sp. larvae. Isolated in Jordan from water (Khyami-Horani et al. 1996, 1999)
71Hc16DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from dead insects (Khyami-Horani et al. 1996, 1999)
72Ht51DBS-UJDisplays entomocidal activity against Culiseta sp. larvae. Isolated in Jordan from soil (Khyami-Horani et al. 1996, 1999)
73Hc36DBS-UJDisplays entomocidal activity against Culex sp. larvae. Isolated in Jordan from water (Khyami-Horani et al. 1996, 1999)
75BUPM33CBSIsolated from soil, Sfax region, Tunisia
B. weihenstephanensis (4)710208WSBCGrowth at 7°C (Lechner et al. 1998; Pruss et al. 1999); isolated from milk (Mayr et al. 1999)
4510204TWSBCB. weihenstephanensis type strain; growth at 7°C (Lechner et al. 1998; Pruss et al. 1999); isolated from milk (Mayr et al. 1999)
8710202WSBCGrowth at 7°C (Lechner et al. 1998; Pruss et al. 1999); isolated from milk (Mayr et al. 1999)
8810201WSBCGrowth at 7°C (Lechner et al. 1998; Pruss et al. 1999); isolated from milk (Mayr et al. 1999)

The BOX-PCR were performed using BOX-A1R primer as already described (Urzìet al. 2001; Cherif et al. 2002).

Computer-assisted analysis of the rep-PCR fingerprinting patterns was performed using the Diversity DatabaseTm Fingerprinting Software (Bio-Rad, Milan, Italy). The banding patterns were acquired from ethidium bromide-stained gels with the Gel Doc 2000 image system (Bio-Rad) and stored on disk as TIFF files. The ‘rolling disk’ background subtraction method was applied to each gel and a database containing all the gel images was created. The bands from all the gels were automatically detected and normalized using the 50-bp DNA ladder (Amersham Pharmacia Biotech, Milano, Italy) as the molecular size marker. A band set including all the polymorphic fragments was created, and each band in each lane was compared with the band set. The similarity between strains was determined by the band-sharing coefficient calculated by the formula of Jaccard, and strain clustering was performed by the unweighted pair group method with arithmetic averages (UPGMA), using ntsys software (Daffonchio et al. 2000).

Sequencing of AC-390 fragment

A 342-bp fragment, named AC-390, a rep-PCR marker previously identified in B. anthracis and homologous to the ywfK gene of B. subtilis (Cherif et al. 2002), was obtained from selected strains of B. anthracis, B. cereus, B. thuringiensis and B. weihenstephanensis by PCR and sequenced in both directions using primers YWFK-f (5-GAAAATGGCCGGATGAGT-3) and YWFK-r (5-GACGTTGAAACATTTATGCA-3) as previously described (Cherif et al. 2002).

The obtained sequences were subjected to neighbour-joining analysis to assess the phylogenetic relationship of the species of the B. cereus group. The alignment of the sequences was checked manually and corrected, and similarity values were determined using jalview software (http://circinus.ebi.ac.uk:6543/jalview).

Results

rep-PCR polymorphism and genetic relationships in the B. cereus group

The entire collection of 112 strains of the six species of the B. cereus group (Table 1) was analysed by rep-PCR using the BOX-A1R primer and a protocol optimized in a previous study (Cherif et al. 2002). The reproducibility of the rep-PCR patterns was evaluated by repeating PCR for several strains using DNA templates obtained from two independent DNA extractions. An example of this experiments, relative to four B. cereus strains (cer4, 487, 6127 and 46321) is shown in Fig. 1. Comparison of the resulting patterns indicated a very high reproducibility of the rep-PCR protocol (Fig. 1, panel A lanes 18–21 and panel D lanes 1–4). Figure 1 shows representative rep-PCR pattern types of the six species. The rep-PCR patterns were shown to be very discriminative and allowed us to identify 97 pattern types (Table 1). The number of bands per profile varied from 1 to 20 and the apparent molecular weight ranged from 100 to more than 2000 bp.

Figure 1.

Example of rep-PCR pattern variability observed in the Bacillus cereus group. Lane M, 50-bp ladder. The 250, 500 and 1000-bp bands of the ladder are indicated. (a) Lanes 1–17, B. anthracis 7700 (refer for rep-PCR haplotype in Table 1: B56), 6769 (B57), 4229 (B58), 6602 (B54), Cepanzo (B54), Davis TE702 (B54), 957 (B54), 227 (B54), 300 (B52), 779 (B50), 832 (B54), 663 (B59), 376 (B55), 846 (B53), 256 (B53), 582 (B51), 282 (B53); lanes 18–21, B. cereus cer4 (B49), 487 (B48), 6127 (B47), 46321 (B47). (b) Lanes 1–28 B. thuringiensis, Bt1 (B32), Bt7 (B31), Bt9 (B30), Bt33 (B29), HD1 (B29), Bt55 (B24), Al (B29), BMG1.7 (B28), BMG1.9 (B27), BUPM21 (B26), BUPM30 (B33), BX16 (B25), 2046T (B39), HD868 (B62), BMG1.6 (B22), Bt14 (B38), Bt44 (B37), BMG1.1 (B36), BMG1.2 (B40), BMG1.8 (B46), Bt13 (B23), 5724 (B34), 5725 (B35), BUPM33 (B75), Bt10 (B41), BMG1.3 (B43), BMG1.4 (B42), BMG1.5 (B44). (c) Lanes 1–3, B. cereus bc2 (B13), cer1(B12), 360 (B11); lane 4, B. weihenstephanensis, 10204T (B45); lane 5, B. cereus myd (B6); lanes 6–23, B. mycoides 2048T (B9), 299 (B8), NRS306 (B1), NRS319 (B2), MycH (B3), B14828 (B10), BmMed (B5), TP2 (B4), 303 (B4), BmS (B82), A81 (B79), 309 (B77), 384 (B78), bmf (B76), B615 (B83), Ndr (B81), Bif (B80), G1 (B96); lanes 24–27, B. pseudomycoides, CA (B95), TP1 (B93), B617T (B94), B618 (B97). (d) Lanes 1–9, B. cereus 487 (B48), 6127 (B47), 46321 (B47), cer4 (B49), Bc1 (B63), 626 (B14), cer5 (B16), 31T (B15), 351 (B61); lane 10, B. mycoides G2 (B90); lanes 11–15, B. cereus cer3 (B17), my1 (B18), 345 (B21), IO200 (B19), po1 (B20); lane 16, B. pseudomycoides A82 (B84); lane 17, B. cereus cer6 (B74); lanes 18–22 B. pseudomycoides BD10 (B85), BD14 (B86), B346 (B89), Nov1 (B92), Nov2 (B91); lanes 23–25 B. weihenstephanensis, 10201 (B88), 10202 (B87), 10208 (B7)

The B. anthracis strains had a relatively uniform rep-PCR pattern with major bands of about 130, 400, 630 and 770 bp. The 400-bp fragment was relatively specific to B. anthracis and had a sequence homologous to ywfK, a transcriptional regulator of B. subtilis (Cherif et al. 2002). Only minor band variations were observed between the strains and none of these were correlated with the presence/absence of the pXO1 and pXO2 virulence plasmids. The B. cereus strains showed wide polymorph profiles but no signature bands could be identified. Some B. cereus strains and most of the B. thuringiensis strains had two bands of about 670 and 1070 bp and several bands with an apparent molecular weight higher than 2000 bp. The B. mycoides strain showed two main pattern types, one very simple with a main band of variable length around 570 bp, found also in some B. cereus and B. weihenstephanensis strains, the other more complex with several bands that co-migrated with the corresponding bands in B. pseudomycoides and B. weihenstephanensis.

Computer-assisted analysis of the rep-PCR profiles was performed using Diversity DatabaseTm Fingerprinting Software (Bio-Rad). By comparing rep-PCR patterns obtained from independent DNA extractions and PCR, we estimated that the error in clustering as a result of pattern reproducibility was negligible, not exceeding 5%. Figure 2 shows the dendrogram generated by UPGMA cluster analysis. Four major sub-clusters (A.1, A.2, B.1 and B.2) were identified, clusters A and B being separated out quite early. Cluster A grouped the B. mycoides (sub-cluster A.1) and B. cereus (sub-cluster A.2) strains while cluster B was divided into two major sub-clusters. Sub-cluster B.1 included strains of B. mycoides, B. pseudomycoides and B. weihenstephanensis. Sub-cluster B.2 showed wide variability and was the most complex, including 70 of the 97 rep-PCR patterns of the strain collection. In sub-cluster B.2 of Fig. 2 were highlighted four groups of patterns at different branching levels, denominated B.2.a to B.2.d and including relevant strains and species. The first group (B.2.a) included only strains of B. mycoides. The B. cereus and B. thuringiensis strains were divided into several groups: the main group of B. cereus was the cluster B.2.b, while that of B. thuringiensis was cluster B.2.c that included 24 patterns of B. thuringiensis (30 strains) and one of B. weihenstephanensis. The remaining strains of B. cereus and B. thuringiensis were grouped in cluster B.2.d, particularly sub-cluster B.2.d.1 that accounted for B. thuringiensis, and sub-cluster B.2.d.2, that included B. cereus. A branch of sub-cluster B.2.d.2 named B.2.d.3 included all the B. anthracis strains.

Figure 2.

Genetic relationship between Bacillus cereus group strains as described by the unweighted pair group method with arithmetic averages (UPGMA) cluster analysis of the rep-PCR patterns. The percentage of similarity between the rep-PCR patterns was calculated using the Jaccard coefficient. (b) rep-PCR haplotype number (see Table 1); N, number of isolates for each rep-PCR haplotype (see Table 1). Acronyms used to indicate the B. cereus group species: BA, B. anthracis; BC, B. cereus; BM, B. mycoides; BP, B. pseudomycoides; BT, B. thuringiensis; BW, B. weihenstephanensis. The dots with letter/number designations (see text) were drawn on the dendrogram nodes where clusters of relevant strains or species are evident. The stars indicate rep-PCR haplotypes for which the ywfK marker was sequenced

Genetic relationship based on the sequence of a rep-PCR marker specific to B. anthracis

The genetic relationship between B. anthracis, B. cereus, B. thuringiensis and B. weihenstephanensis was also assessed by sequencing AC-390, a B. anthracis-specific rep-PCR marker previously identified as homologous to the transcriptional regulator ywfK of B. subtilis (Cherif et al. 2002). The sequencing was performed in 27 strains of the four species selected from the different branches of the dendrogram in Fig. 2. For the amplification of ywfK we used the same primers designed in (Cherif et al. 2002). Using these primers, all attempts at obtaining the PCR product from the B. mycoides and B. pseudomycoides strains failed, indicating that these two species harbour a different sequence at the primer sites.

The genetic relationship of the four species of the B. cereus group based on the sequences of the AC-390 fragment is shown in the neighbour-joining tree of Fig. 3. All but two of the strains of B. cereus and B. thuringiensis were intermingled in branch A of the tree, confirming that it is very difficult to discriminate these two species by analysing a single locus. In branch B of the ywfK neighbour-joining tree, there was the identification of two sequence clusters, namely B1 and B2, in which were placed B. anthracis and B. weihenstephanensis strains, respectively. In branch B of the tree were also placed B. cereus strain PO1 and B. thuringiensis strain BUPM33.

Figure 3.

Phylogenetic relationship between strains of the species of the Bacillus cereus group determined by neighbour-joining analysis of the ywfK sequences (Cherif et al. 2002). Before each strain name the species abbreviation is given (BA, B. anthracis; BC, B. cereus; BS, B. subtilis; BT, B. thuringiensis; BW, B. weihenstephanensis). The dots with letter/number designations (see text) were drawn on the dendrogram nodes where separated clusters and sub-clusters are evident. Bar indicate 2% of phylogenetic distance

Discussion

From the dendrogram topology described in Fig. 2 several considerations can be made: our rep-PCR data confirmed the data of Kim et al. (2002) who found B. anthracis to diverge from the other species in the genus Bacillus; our use of a greater collection of strains of the six B. cereus group species gives strength to the findings of their data. Bacillus anthracis was found to form a sub-group within the B. cereus/B. thuringiensis group, indicating that although this species can be considered closely related to B. cereus and B. thuringiensis it actually diverges from them, sharing only 32% of similarity with the closest B. cereus strains (haplotypes 47, 48 and 49 in Fig. 2). These data are also in agreement with the genetic relationship described by ITS fingerprinting (Daffonchio et al. 2000) and by the sequencing of the long ITS-containing tRNA (Cherif et al. 2003).

Almost all the 19 B. cereus strains had different band patterns and were distributed over several branches of the dendrogram (Fig. 2). Some strains were grouped in cluster A together with several strains of B. mycoides, others were put into cluster B.2.b, and those remaining went into cluster B.2.d.2, the cluster that included cluster B.2.d.3 of B. anthracis. This distribution suggests that the B. cereus strains are characterized by a relatively variable genome, possibly due to a relatively frequent horizontal gene transfer within the species itself and with the other species of the group. This hypothesis is substantiated by recent multilocus enzyme electrophoresis data from sympatric B. cereus populations that show a panmictic population structure (Vilas-Boas et al. 2002).

Of the two main groups of B. thuringiensis strains, one included several known subspecies as well as different Tunisian strains isolated from different environments. Almost all the Jordan isolates were grouped together in a separate branch of the dendrogram, indicating that these strains, which all derive from the same geographical regions and that display mosquitocidal activity (Khyami-Horani et al. 1996, 1999), are relatively homogeneous. This grouping confirms the findings of a similar relationship revealed by ITS-PCR fingerprinting (Daffonchio et al. 2000).

The strains of B. mycoides, B. pseudomycoides and B. weihenstephanensis were confirmed in their relative genetic distance from the other species of the B. cereus group. The rep-PCR data support the distinctness between B. mycoides and B. pseudomycoides.

The genetic relationship described by the rep-PCR fingerprinting patterns was confirmed by the sequencing of a discrete locus, the AC-390 fragment. The sequence data support the idea that B. anthracis diverges from B. cereus and B. thuringiensis and hence represents an independent genetic lineage in the B. cereus group. Also B. weihenstephanensis was confirmed to be quite distant from B. cereus/B. thuringiensis, suggesting that, in general, this bacterium is an intermediate form between B. cereus/B. thuringiensis and B. mycoides. In fact B. weihenstephanensis shares colony morphology with B. cereus/B. thuringiensis, while with B. mycoides it shares the psychrotolerant phenotype and several genetic features such as ITS sequences and structure (Daffonchio et al. 2000; Cherif et al. 2003), signature sequences in several genetic loci like 16S and 23S rRNA (Lechner et al. 1998; Pruss et al. 1999), and the cold shock protein cspA (Francis et al. 1998). All attempts at obtaining the AC-390 PCR product from the B. mycoides and B. pseudomycoides strains failed, confirming the divergence of these two species from the other species of the B. cereus group.

The AC-390 sequence data confirm the impossibility of discriminating B. cereus from B. thuringiensis by comparing the sequence of a single discrete genetic locus. Although it has been shown that sympatric populations of B. cereus and B. thuringiensis have a higher level of recombination within a species than between species, it has been proposed that these two species cannot be considered sexually isolated (Vilas-Boas et al. 2002). Interspecies genetic exchange, even at a low level, would determine the wide genetic diversity and the species intermingling observed in the neighbour-joining tree, breaking down the clonal genetic structure in favour of a panmictic one. The impossibility of finding a distinct, exclusive cluster for the different species of the B. cereus group, neither in the sequence analysis of the ywfK fragment (e.g. the presence of B. cereus PO1 and B. thuringiensis BUPM33, respectively, close to the clusters where B. weihenstephanensis and B. anthracis are placed) nor in the rep-PCR fingerprinting analysis, supports the observation (Vilas-Boas et al. 2002) that the species in the B. cereus group are not completely sexually isolated, and that genetic exchange probably plays a significant role in the structuring and evolution of the populations of these bacteria in the environment.

In conclusion, it has been shown that rep-PCR analysis is a useful tool to describe the genetic diversity and genetic relationship of closely related bacteria such as members of the B. cereus group. An analysis of the rep-PCR patterns of these bacteria has confirmed their wide genetic variability, a variability that was further corroborated by the sequences of a rep-PCR marker typical of B. anthracis rep-PCR patterns. In general, the genetic relationship, from the point of view of the rep-PCR patterns, indicates a relatively clear separation of the six species into quite homogeneous clusters, suggesting that they can indeed be considered as separate units, the level of recombination within each unit being higher than that between the units themselves. In any case, the finding of intermingled stains in several units, revealed by both rep-PCR fingerprinting and ywfK sequencing, suggests that a certain degree of recombination does take place between the species (Vilas-Boas et al. 2002).

Acknowledgements

The study was supported by the Italian Ministry for University and Scientific Research within the project ‘Risposta della comunità microbica del suolo a differenti pressioni antropiche: effetti su struttura, dinamica e diversità della microflora’ (Cofin 2000), and from the INTAS-International Association for the promotion of co-operation with scientists from the New Independent States of the former Soviet Union within the project ‘An epidemiological study of outbreaks of B. anthracis in Georgia’ (INTAS-01-0725). AC was supported by a grant from the Direction Generale de Recherche Scientifique et Technologique of the Ministere de l'Education Superieure of Tunisia.

We thank M. Mock, G. Patra, S. Jaoua, L. K. Nakamura, S. Scherer, R. Mayr and D. R. Zeigler for kindly providing Bacillus strains and DNA. The manuscript was edited by Barbara Carey.

Ancillary