2008-Guinebretiere Et Al 2008 EM10 851-865

You might also like

Download as pdf or txt
Download as pdf or txt
You are on page 1of 15

Environmental Microbiology (2008) 10(4), 851–865 doi:10.1111/j.1462-2920.2007.01495.

Ecological diversification in the Bacillus cereus Group

Marie-Hélène Guinebretière,1* Group using (i) molecular data from Fluorescent


Fabiano L. Thompson,2,3 Alexei Sorokin,4 Amplified Fragment Length Polymorphism patterns,
Philippe Normand,5 Peter Dawyndt,2,6 ribosomal gene sequences, partial panC gene
Monika Ehling-Schulz,7 Birgitta Svensson,8 sequences, ‘psychrotolerant’ DNA sequence signa-
Vincent Sanchis,9 Christophe Nguyen-The,1 tures and (ii) phenotypic and descriptive data from
Marc Heyndrickx10 and Paul De Vos2 range of growth temperature, psychrotolerance and
1
Institut National de la Recherche Agronomique, thermal niches. Seven major phylogenetic groups
UMR A408 Sécurité et Qualité des Produits d’Origine (I to VII) were thus identified, with ecological differ-
Végétale, INRA, Domaine Saint-Paul, Site Agroparc, ences that provide evidence for a multiemergence of
F-84914 Avignon Cedex 9, France. psychrotolerance in the B. cereus Group. A moderate
2
Laboratory for Microbiology, and BCCM/LMG Bacteria thermotolerant group (VII) was basal to the meso-
Collection, Ghent University, Ghent 9000, Belgium. philic group I, from which in turn distinct thermal
3
Federal University of Rio de Janeiro, Institute of lineages have emerged, comprising two mesophilic
Biology, Department of Genetics, Ilha do fundão, Caixa groups (III, IV), an intermediate group (V) and two
postal 68011, CEP 21944-970, Rio de Janeiro, Brazil. psychrotolerant groups (VI, II). This stepwise evolu-
4
Institut National de la Recherche Agronomique, tionary transition toward psychrotolerance was par-
Laboratoire de Génétique Microbienne, Centre de ticularly well illustrated by the relative abundance
Recherche de Jouy-en-Josas, 78352 Jouy-en-Josas of the ‘psychrotolerant’ rrs signature (as defined by
cedex, France. Pruss et al.) copies accumulated in strains that varied
5
Université Lyon 1, CNRS, UMR 5557, Ecologie according to the phylogenetic group. The ‘psychro-
Microbienne, Lyon, F-69003 Villeurbanne, France. tolerant’ cspA signature (as defined by Francis et al.)
6
Department of Applied Mathematics and Computer was specific to group VI and provided a useful way to
Science, Ghent University, Krijgslaan 281, S9 B-9000 differentiate it from the psychrotolerant group II. This
Ghent, Belgium. study illustrates how adaptation to novel environ-
7
Lehrstuhl für Mikrobielle Ökologie, Department für ments by the modification of temperature tolerance
Grundlagen der Biowissenschaften, Technische limits has shaped historical patterns of global eco-
Universität München, Weihenstephaner Berg 3, logical diversification in the B. cereus Group. The
D-85354 Freising, Germany. implications for the taxonomy of this Group and for
8
Swedish Dairy Association, Scheelevaegen 18, 223 63 the human health risk are discussed.
Lund, Sweden.
9
Institut National de la Recherche Agronomique, Unité
Introduction
Génétique Microbienne et Environnement, INRA, La
Minière, 78285 Guyancourt, Cedex, France. Organisms of the Bacillus genus are ubiquitous bacteria
10
Ministry of the Flemish Community, Center for found both above and below the Earth’s surface. The
Agricultural Research, Department of Animal Product Bacillus cereus Group strains display a wide diversity. Six
Quality, Brusselsesteenweg 370, B-9090 Melle, closely related species have so far been described in this
Belgium. taxonomic Group: Bacillus anthracis, Bacillus thuringien-
sis, Bacillus mycoides, Bacillus pseudomycoides, Bacillus
weihenstephanensis and B. cereus (sensu stricto).
Summary
Some literature data indicate that these species are not
The Bacillus cereus Group comprises organisms that strictly based on genomic divergence (Guinebretiere and
are widely distributed in the environment and are of Sanchis, 2003; Helgason et al., 2004; Hill et al., 2004;
health and economic interest. We demonstrate an Priest et al., 2004), but rather on subjective consideration
‘ecotypic’ structure of populations in the B. cereus of practical usefulness such as virulence (B. anthracis,
B. thuringiensis), physiology (B. weihenstephanensis),
morphology (B. mycoides and B. pseudomycoides) and
Received 24 June, 2007; accepted 15 October, 2007. *For
correspondence. E-mail guinebre@avignon.inra.fr; Tel. (+33) 432 72 ill-defined features (B. cereus sensu stricto) (Pirttijarvi
25 24; Fax (+33) 432 72 24 92. et al., 2000; Guinebretiere and Sanchis, 2003). Theses
© 2007 The Authors
Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd
852 M.-H. Guinebretière et al.

species range in practical importance. Bacillus anthracis related to the species currently defined in the B. cereus
is known to be the causal agent of anthrax. Bacillus Group. One hypothesis is that in the B. cereus Group,
cereus (sensu stricto) includes pathogenic strains that distinct genetic groups can possess their own growth
may cause food poisoning, eye infections and periodontal range specificities. Addressing this hypothesis might give
disease in humans (Drobniewski, 1993; Helgason et al., microbiologists a better understanding of extant B. cereus
2000). By contrast, certain B. cereus strains are used as Group diversity and its special ability to adapt to widely
probiotics. Spore and crystal toxin preparations from diverse habitats. Thus, the objectives of the present work
B. thuringiensis are used as commercial biopesticides were (i) to establish the existence of thermal groups
(Schnepf et al., 1998). Bacillus weihenstephanensis, based on a genotypic structure and (ii) to position psy-
B. mycoides and B. pseudomycoides have not been chrotrophs, food pathogens and the current phenotypic
described as food poisoning agents, but their toxigenic species in this ‘ecotypic structure’.
potential remains uncertain (Pruss et al., 1999a; Stenfors For this purpose, we undertook an extensive character-
et al., 2002). ization of 425 independent strains selected to cover the
Strains of the B. cereus Group sensu lato seem to adapt global phenotypic and genotypic diversity of the entire
to very different habitats, from cold to hot thermal environ- B. cereus Group, including all six currently defined species
ments, spanning alpine to temperate soils (von Stetten and, in particular B. weihenstephanensis, B. mycoides
et al., 1999), and refrigerated foods to dehydrated foods and B. pseudomycoides, of which little is known. MLST is
(Pirttijarvi et al., 1998; Guinebretiere and Nguyen-The, generally an excellent tool for genetic population studies,
2003). As a consequence, a very broad thermal range for but to characterize 425 strains and avoid biases due to
growth temperature in the B. cereus Group strains has potentially inadequate MLST schemes, we chose instead
been recorded in the literature, ranging from 4°C to 50°C to associate a technique based on genome-wide blind
(Anderson, 1997). Although the range of growth tempera- markers, the Fluorescent Amplified Fragment Length Poly-
ture varies between strains, it has not yet been associated morphism (fAFLP), and the phylogeny of ribosomal genes
with particular species or groups, except for the psychro- and of an appropriate housekeeping gene, the panC gene.
tolerant species B. weihenstephanensis whose growth Fluorescent Amplified Fragment Length Polymorphism is a
range has been first defined from below 7°C up to 43°C reliable tool that has successfully been adapted to the
(Lechner et al., 1998) and later revised from 7°C up to 38°C study of strains in the B. cereus Group (Keim et al., 1997;
(von Stetten et al., 1999). Adaptation to this very broad Ticknor et al., 2001; Hill et al., 2004). Derived from a pre-
range of growth temperatures in the B. cereus Group has vious study (Candelon et al., 2004), the panC gene
important implications for the Food Industry. Psychrotoler- sequence analysis showed high congruence with fAFLP
ant strains are a recurrent problem in perishable foods results in preliminary tests. While ribosomal sequence data
during refrigerated storage and distribution, particularly in were useful for distinguishing moderately divergent
dairy products (Larsen and Jorgensen, 1997; Te Giffel populations of B. cereus Group into separate sequence
et al., 1997) and cooked chilled foods (Carlin et al., 2000; clusters, panC gene provided a better opportunity for dis-
Guinebretiere et al., 2001), where they may induce spoil- tinguishing very closely related ecological populations.
age or raise safety concerns. Mesophilic B. cereus strains Using this global approach, we reconstituted the global
can persist on hot dairy equipment such as pasteuriser genotypic and phylogenetic structure of the whole
(Svensson et al., 2000). They can also survive the drastic B. cereus Group, without a priori and biases, and assessed
processing of dehydrated foods and subsequently con- the robustness of this model using different clustering
taminate diverse foodstuffs via dehydrated ingredients methods and bootstrap analyses. Finally, the resulting
(Becker et al., 1994; Guinebretiere and Nguyen-The, phylogenetic groups were characterized for their growth
2003). Numerous cases of food poisoning have been temperature limits and for the presence of known ‘psychro-
attributed to cooked foods or infant foods containing boiled tolerant signatures’ (Francis et al., 1998; Pruss et al.,
or reconstituted dehydrated ingredients, e.g. spaghetti, 1999b) to define their thermal growth properties.
rice, noodles, macaroni, milk powder, infant formula and
dehydrated potato purées (Holmes et al., 1981; Kramer
and Gilbert, 1989; Shinagawa, 1990). Results
Temperature is one of the most important environmen-
Reconstructing the phylogenetic structure of the
tal factors to which microorganisms have to respond. The
B. cereus Group strains
existence of diverse growth temperature ranges in the
B. cereus Group and psychrotolerant strains that cannot Fluorescent Amplified Fragment Length Polymorphism
be identified as B. weihenstephanensis (Stenfors and banding patterns were found to consist of a mean of
Granum 2001), suggests there may be different ecologi- 68 ⫾ 6 fragments in the size range of 50–536 bp. Com-
cal populations associated with this character, not directly parison and classification of data using the Dice similarity

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
B. cereus ecological diversification 853
c
% strains with : Total No. of strains received as : No. of strains with No. of food
A B Genetic Genetic rrs cspA Range of
No. of Bm Bm Rhizoidal Growth poisoning
50 60 70 80 90 100
subgroup group a b growth T°C Bc Bt Ba Bw Bp d
sign sign strains (p) (m) colonies at 7°C strains
BC10 19 0 0 0 0 0 17 2 19 0 0
I 14 0 > 10-43 <
BC13 9 0 0 0 0 0 5 4 9 0 0

BC6 II 88 0 > 7-40 < 33 26 7 0 0 0 0 0 0 25 7

BC12 15 9 6 0 0 0 0 0 0 0 2

BC05 34 33 1 0 0 0 0 0 0 0 18
III 37 0 > 15-45 <
BC09 19 18 1 0 0 0 0 0 0 0 4

BC08 28 10 5 12 0 0 1 0 0 0 5

BC04 0 36 13 23 0 0 0 0 0 0 0 8

BC03 IV 59 0 > 10-45 < 36 26 9 0 0 1 0 0 0 0 11

BC07 0 29 8 21 0 0 0 0 0 0 0 3

BC11 V 94 0 > 8-40 < 17 6 11 0 0 0 0 0 0 1 2

BC01 100 104 9 12 0 22 61 0 0 57 101 0

VI 100 > 5-37 <

BC02 100 39 8 0 0 11 17 3 0 20 35 0

BC14 VII 0 0 > 20-50 < 2 2 0 0 0 0 0 0 0 0 1

Fig. 1. Genetic diversity of the B. cereus group.


A. Simplified dendrogram showing the genotypic relationships between bacterial strains of the B. cereus group, based on fAFLP data analysis.
The percentage similarities (represented by the horizontal scale) were calculated using the Dice similarity coefficient and clustering by UPGMA
(Sneath and Sokal, 1973). The resulting fAFLP groups (DU groups) formed at 60% similarity level are represented in grey. Numbers on
branches are bootstrap values. The genetic subgroups (BC01 to BC14) are schematically represented in black on the dendrogram and are
issue from a non-hierarchical classification using the Bin Class software. BpT, B. pseudomycoides DSM 12442T = lMG 18993T; Ba, B. anthracis
CEB 94–0040; BcT, B. cereus ATCC 14579T = lMG 6923T; BtT, B. thuringiensis CIP 53.137T = lMG 7138T; BwT, B. weihenstephanensis WSBC
10204T = lMG 18989T; BmT, B. mycoides CIP 103472T = lMG 7128T.
B. Distribution and characteristics of strains in each genetic group. a‘psychrotolerant’ rrs sequence signature; b‘psychrotolerant’ cspA sequence
signature; cBc, B. cereus; Bt, B. thuringiensis; Ba, B. anthracis; Bw, B. weihenstephanensis; Bm, B. mycoides; Bp, B. pseudomycoides (p):
psychrotolerant (m): mesophilic; daccording to the rapid screening test.

coefficient and the Unweighted Pair Group Method of congruent (Table 1). New relationships among strains
Arithmetic Averages (UPGMA) clustering method gener- within each DU group were disclosed by the BinClass
ated 12 robust groups at a 60% similarity level (DU groups software, delimiting robust subgroups within each DU
represented in grey in Fig. 1) among the 425 strains of the group (Table 1). Finally, taking into account the two clas-
B. cereus Group (Table 2, Fig. 1). This cut-off level (60%) sification methods, we established a genetic structure of
represented the maximum level for which all resulting the B. cereus Group, which was composed of seven
fAFLP groups with no exception were highly robust, major genetic groups (I to VII), themselves subdivided into
showing bootstrap values above 80% (Fig. 1). BinClass 14 genetic subgroups (Table 1, Fig. 1). The good overall
classification performed on the same data set generated correspondence between the two classification methods
14 groups (BC groups, shown in black in Fig. 1). Despite demonstrates the high robustness of the seven major
the major conceptual difference between the two genetic groups and subgroups. In what follows, this geno-
clustering approaches (one hierarchical and one non- typic structure represented by seven major genetic
hierarchical method), the obtained results are highly groups will be used as reference.

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
854 M.-H. Guinebretière et al.

The genetic group I is a well-separated and robust

Major genetic
entity (Fig. 1) validated by high bootstrap values (above

groupsb
95%), along with the small genetic group VII (bootstrap
values 100%) and a third complex composed of the five

VII
IV

VI
III

V
II
I
remaining genetic groups II, III, IV, V and VI (bootstrap
Total of
strains

26
values 100%). In this complex, the relationship between
2
31
3
2
94
101
17
139
5
3
2
425
the five genetic groups and their subgroups remained
unresolved. panC phylogeny (Fig. 2) strongly correlates
BC14

201.3
565.7
with the major genetic groups defined by AFLP analysis,
1d

2
7
4
confirming the robustness (bootstrap values > 90%) of the

106.4
327.5
five major genetic groups (II, III, IV, V and VI) and high-
BC13

9
lighting relatedness between subgroups inside each
genetic group. It also confirmed that the genetic group II
BC10

132.7
429.8
17
2

19
was phylogenetically close to the genetic group III, and
the genetic group IV to the genetic group V. The phyloge-
BC06

206.3
604.2 netic relationships between the five genetic groups (II, III,
31
2

33

IV, V and VI) were not completely resolved, notably the


position of the phylogenetic group VI in this set.
BC09

177.6
522.0
2
17

19

The position of group VI in the topology shown by the


fAFLP dendrogram (Fig. 1) showed it clearly separated
BC12

184.2
541.0

from groups II, III, IV, V. Furthermore, the group VII


15

15

seemed to be basal to all groups. To elucidate this, a


phylogeny of the 16S (rrs) gene and Intergenic tran-
BC08

152.3
470.0
28

28

scribed spacer situated between the rrs and the 23S (rrl)
genes (ITS1) (Normand et al., 1996) was constructed
BC05

157.1
485.5

(Fig. 3), based on sequence analysis from one to two


34

34

representative strains of each group (the strains closest


BC07

184.5
559.3

to the HMO of BC classes). A strain (AF2), not included


29

29

in the AFLP study but presenting similar genetic traits


as strains of group VII (Fagerlund et al., 2007), was
Table 1 Congruence of the BinClass classification and Dice-UPGMA clustering methods.

BC04

166.8
523.4
36

36

included in this study to check its phylogenetic related-


Groups obtained from the clustering method Dice-UPGMA at 60% similarity level.

ness with group VII strains. The resulting phylogenic


BC03

179.9
523.6

topology of the different groups posits group VII as basal


36

36

to the others, supported by strong bootstrap results. In


Major genetic groups issue from the combination of the two methods.

total coherence with fAFLP topology, the next to emerge


149.6
466.5
BC11

17

17

is group I (supported by a low bootstrap value of 51%)


and then, the group VI (supported also by a low boot-
BC02

188.6
561.3

strap value of 43%). The remaining groups (II, III, IV and


35
1
3

39

V) are difficult to position relative to one another


because of the insufficient resolution of the marker
BC01

160.4
505.1
100
4

104

Groups obtained from BinClass Classification.

chosen. The topologies yielded by the more resolutive


panC and fAFLP analyses are slightly different, thus we
Bina class

conclude that theses topologies are more representative


than the conserved rrs gene for the closely related
Strains excluded from the study.

groups II, III, IV and V.


Only three strains have been clustered to group VII
Average Shannon code length

among over 400 strains, themselves selected among


thousand of strains around the world. This attests of the
scarceness of group VII strains. Their scarceness can be
Average distortion

due to (i) a declining population or to (ii) an uncommon


Total of strains
AFLP groupc

specific niche that allows isolated multiplication of strains.


For the following experiments, we thus considered these
DU01
DU02
DU03
DU04
DU05
DU06
DU07
DU08
DU09
DU10

DU12
DU11

three independent strains as representative of group VII in


b.

d.

spite of this low number.


a.

c.

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
B. cereus ecological diversification 855
a
Table 2. Range of growth temperature recorded for each genetic group.

% strains with positive growth at:


Phylogenetic
group 4°C 5°C 7°C 8°C 10°C 15°C 20°C 35°C 37°C 40°C 43°C 45°C 50°C 55°C

VII 0 0 0 0 0 0 100 100 100 100 100 100 100 0


III 0 0 0 0 0 100 100 100 100 100 100 85 0 0
IV 0 0 0 0 100 100 100 100 100 100 83 58 0 0
I 0 0 0 0 75 100 100 100 100 100 25 0 0 0
V 0 0 0 14 100 100 100 100 100 100 0 0 0 0
II 0 0 73 87 100 100 100 100 100 100 0 0 0 0
VI 0 40 100 100 100 100 100 100 86 0 0 0 0 0

a. Strains affiliated to groups II to VII were significantly (P < 0.05) associated to a specific range of growth (R): group VII to R01 (20–50°C), group
III to R02 (15–45°C) and R03 (15–43°C), group IV to R05 (10–45°C) and R06 (10–43°C), group V to R07 (10–40°C), group II to R08 (8–40°C)
and R09 (7–40°C), and group VI to R10 (7–37°C), R11 (5–37°C) and R12 (5–35°C). Strains of group I exhibited more variable R, globally not
significantly different from those of groups IV or from those of group V, ranging from R04 (15- 40°C) to R07.

Ranges of growth temperatures cantly (P < 0.05) distributed in thermal niches 2 (inter-
mediate temperatures) and thermal niches 3. Thus, all
In our high throughput screening, psychrotrophs (strains
phylogenetic populations are able to colonize foods, but
able to grow at temperatures ⱕ 7°C) were restrictively
they are preferentially distributed according to the fit
recovered in two phylogenetic groups out of seven. In
between their thermal properties and the thermal condi-
group VI, 100% of strains were able to grow at 7°C (Fig. 1)
tions to which food or their transformation is associated.
and 40% at 5°C, whereas in genetic group II 76% of
This selective association probably explains the fact that,
strains were able to grow at 7°C and 0% at 5°C. Interest-
in a previous work, the B. cereus population of the warm
ingly, these psychrotrophs belong to two distant phyloge-
parts of the cheese dairy process was found to be differ-
netic groups (II and VI).
ent from that of cold part of the process (Pirttijarvi et al.,
In our second experiment, strains affiliated to groups II
1998). Similar results were also recorded for B. cereus
to VII differed significantly in growth range characteristics
genotypes on a purée processing line (Guinebretiere and
(P < 0.05, see Table 2), whereas strains of group I had
Nguyen-The, 2003).
similar growth range characteristics (R04 to R07) as those
In the different thermal niches and particularly in
from groups IV (R05, R06) or from group V (R07). The
thermal niches 1 and 2 (soils, rivers, plants, insects, . . .),
genetic group VII had the particularity of being composed
the moderate thermotolerant strains of group VII probably
of strains able to grow between 20°C to 50°C and should
are in competition with the more adapted strains of groups
thus be considered as a moderate thermotolerant group.
I to VI, which could explain their scarceness. Their pres-
In all cases, while the breadth of temperature range did
ence in dehydrated foods may be explained by the selec-
not significantly change, we observed that more psychro-
tive effect of high heat treatments during processing and
tolerant groups have lost the ability to grow at higher
by the absence of competitive bacterial growth in the
temperatures. Enhanced psychrotolerance may have
finished product due to low aw.
come at the cost of increased thermal specialization for
some phylogenetic groups such as the psychrotolerant
groups VI and II.
Distribution of thermal cspA and rrs genotypes
The psychrotolerant cspA genotype as defined by Francis
Thermal groups and habitats
and colleagues (1998) was found only in the genetic
Habitats represent different ‘thermal niches’ depending on group VI with 100% cspA signature-positive strains
the thermal conditions they offer to bacteria (Table 3). (Fig. 1). Thus, the ‘psychrotolerant’ cspA signature must
Strains of groups VI, II, V (growth profiles R07 to R12, see be considered a specific characteristic of genetic group
Table 2) were significantly (P < 0.05) distributed in thermal VI. The three strains of the minor fAFLP DU11 group (see
niches 1 (temperatures with high fluctuations frequently Table 1) that were found close to group VI possessed the
covering low temperatures (ⱕ 8°C), whereas strains of mesophilic cspA genotype.
groups III and VII (growth profiles R01 to R03) were The psychrotolerant rrs signature defined by Pruss and
significantly (P < 0.05) associated with thermal niches 3 colleagues (1999b) was detected in all groups, except
(warm temperatures: mammals, dehydrated/starched group VII. We observed a differential distribution accord-
foods). Strains of groups IV (growth profiles R05 to R06) ing to the genetic group (Fig. 4B). The psychrotolerant
seemed to form intermediate broad population signifi- groups II and VI and the intermediate group V contained

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
856 M.-H. Guinebretière et al.

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
B. cereus ecological diversification 857
Fig. 2. Phylogenetic relationships between bacterial strains of the B. cereus group, based on panC sequences analysis. Strains selected
for this analysis were representative of the fAFLP genetic groups II, III, IV, V and VI. The branching pattern was generated by
Neighbour-Joining method (Saitou and Nei, 1987). Numbers on branches are bootstrap values above 50%. Bar, 0.01 nucleotide substitution
per site. (1) B. anthracis CEB 94–0040; (2) B. weihenstephanensis WSBC 10204T = lMG 18989T; (3) B. mycoides CIP 103472T = lMG 7128T;
(4) B. thuringiensis CIP 53.137T = lMG 7138T; (5) B. cereus ATCC 14579T = lMG 6923T. The corresponding genetic groups and their subgroups
to which the strains belong are mentioned on branches and on the right side of the figure respectively.

a high number of strains carrying this rrs signature in temperature. They stipulate that the ability of a strain to
abundance (subtypes P, P/m, P/M, Fig. 4B). In contrast, grow at low or high temperature may be influenced by the
groups III and I contained a low number of strains with this ratio of these two structures. This is in coherence with our
signature and in low amount (p/M). Group IV contained a results and this may explain why extension of the lower
low part of its strains (18%) with a significant amount of rrs thermal limit in the B. cereus group was accompanied by
signature (subtype P/M), in coherence with the fact that a trade-off with performance at higher thermal limits.
this population is distributed among diverse thermal
niches (2, 3 and in a low proportion 1).
Position of the current phenotypic species
Pruss and colleagues (1999b) hypothesized that this
adaptive mutation allows two different states for the rrs All the mesophilic rhizoidal colony-forming strains
RNA molecule of the B. cereus group and that this may (received as B. mycoides or B. pseudomycoides) were
lead to two different structure of the 30S subunit, one of clustered in phylogenetic group I, whereas all the psy-
these structures allowing translation initiation at low chrotolerant rhizoidal colony-forming strains (received
temperature, and the other one allowing it at higher with an identification as B. mycoides) were found in

Fig. 3. Phylogeny of the B. cereus Group based on rrs and ITS1 genes sequence analysis, and species of the B. cereus group associated to
each group. The Phylogenetic tree was generated by the Neighbour-Joining method (Saitou and Nei, 1987). Numbers on branch nodes are
bootstrap values above 50%. The bar indicates 0.005 nucleotide substitution per site. Highly similar topologies were obtained using the
Maximum-Parsimony (Kluge and Farris, 1969) method and Maximum-Likelihood method (Felsenstein, 1981). (P) indicates the Psychrotolerant
group; (I) the Intermediate group with moderate psychrololerance characteristics; (M) the Mesophilic group and (MT) the moderate
thermotolerant group. Species associated to each genetic group are mentioned on the right hand of the figure. For group VI, B. cereus VI
strains were undistinguishable from B. weihenstephanensis strains and thus were considered as B. weihenstephanensis.

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
858 M.-H. Guinebretière et al.
Table 3. Distribution of strains according to the origin and according to the genetic group (in percentage of the total number of strains in each
group).a

Thermal Group I Group II Group III Group IV Group V Group VI Group VII
Origin nicheb (28)c (33) (96) (101) (17) (143) (3)

% from natural environment (soil, water, air, plants) 1, 2 40 40 12 29 41 35


% from Animals:
Mammals, feces 3 9 23 2
Insects 2 1 16 1
% from the Dairy Industry:
Dairy environment 1, 2 28 3 1 22
Raw milk 1, 2 14 7 1 8
% from foods:
Chilled pasteurized milk 1 7 3 4 6 21
Chilled pasteurized vegetables 1 1 2 5
Cooked foods with vegetables as major ingredient 3 6 2 4 6
Cooked foods with meat as major ingredient 3 6 8 7 6
Milk products (cheese, cream, ice cream . . .) 2, 1 3 3 2 1
Diverse cooked foods 2, 3 3 1 1
Raw vegetables 1, 2 18 1 2 6 2
Dehydrated foods (starch or proteins) 3 11 8 70
Herbes and spices 3 3 30
Starched foods (rice, pasta . . .) 3 3 11 10 6
Unknown source 11 9 7 16 29 4
Food poisoning foods 0 21 33 22 12 0 50

a. Absence of numbers indicates: Nb. = 0.


b. 1: highly fluctuating temperatures frequently encompassing low temperatures; 2: intermediate temperatures; 3: warm temperatures
(see Experimental procedures)
c. Parenthetical numbers indicate the total number of strains in each group.

genetic group VI (Fig. 1). This topology can be explained (VI) (Fig. 1). In phylogenetic group VI these two species did
by the fact that most strains received as B. mycoides not form separate entities and had similar psychrotoler-
(including reference strains) were identified before the ance properties. Group VI also contained 20% of strains
description of the B. pseudomycoides species. The mor- received as B. cereus or B. thuringiensis (Fig. 1). It
phological similarities observed between the two species therefore appears that psychrotolerant organisms of the
and difficulties in differentiating the two species with B. cereus Group, despite their restrictive phylogenetic
simple methods (only some levels in 12:0 iso and 13:0 position in groups II and VI, may be B. weihenstephanen-
anteiso fatty acids have been proposed for differentiating sis, B. mycoides strains as well as B. cereus and B. thur-
between the two species (Nakamura, 1998) may have ingiensis strains. Bacillus cereus strains clustered into
favoured the non-revision of strain identity. In our study, group VI are not distinguishable from B. weihenstephan-
we clearly show that the B. pseudomycoides species cor- ensis strains by any known characters and should be
responds to phylogenetic group I (Fig. 1), a separate phy- considered as B. weihenstephanensis based on the
logenetic group composed of mesophilic rhizoidal colony- current identification keys.
forming strains that can be easily distinguished from The 12 strains of B. anthracis were recovered in
the psychrotolerant rhizoidal colony-forming strains genetic group III, and belonged to a well-delimited and
(B. mycoides ‘sensu stricto’, group VI) by the absence of homogenous phylogenetic subgroup BC08. Interestingly,
‘psychrotolerant’ cspA signature. in contrast to the monomorphic character usually attrib-
Bacillus cereus and B. thuringiensis species were uted to B. anthracis species, the phylogenetic subgroup
highly polyphyletic, being components of the phylogenetic BC08 also contained 11 strains received as B. cereus
groups II, III, IV, V and VI (Fig. 1, Table S1). These strains and five strains received as B. thuringiensis
species thus cover the entire spectrum of growth tem- strains.
peratures observed in the B. cereus Group. When they
belonged to the same genetic group, they differed by the
Position of the food pathogens
presence of the ‘cry-carrying’ plasmid.
By contrast, B. mycoides ‘sensu stricto’ (psychrotolerant Because of their ubiquitous nature, B. cereus Group
rhizoidal colony-forming strains) and B. weihenstephanen- strains may be readily recovered in foods. In this study,
sis were clustered together in the same phylogenetic group the food poisoning strains (49 strains connected to

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
B. cereus ecological diversification 859

A Fig. 4. Presence of the ‘psychrotolerant’ rrs


sequence signature in strains and distribution
according to the genetic group.
A. Illustration of the five patterns (P, P/m,
P/M, p/M, M) obtained for strains during the
experiment and of the three resulting
genotypes (P, psychrotrophic; I, intermediate;
M, mesophilic). P: two bands (351 bp, 122 bp)
resulting from the SspI digestion of amplicons
473bp containing 100% rrs copies with the sequence
351bp signature; P/m: three bands (473 bp, 351 bp,
122 bp) resulting from the SspI digestion of
amplicons containing a majority of rrs copies
122bp with the sequence signature; P/M: three
bands (473 bp, 351 bp, 122 bp) resulting from
the SspI digestion of amplicons containing
around 50% of rrs copies with the sequence
P P/m P/M p/M M signature; p/M: three bands (473 bp, 351 bp,
122 bp) resulting from the SspI digestion of
amplicons containing a minority of rrs copies
with the sequence signature; M: one band
(473 bp) resulting from amplicons that does
P I M not contain rrs copies with the sequence
signature.
B B. Percentage of strains with ‘psychrotolerant’
rrs sequence signature according to the
100% genetic group (I to VII). Number of tested
strains: 22 for group I, 32 for group II, 27 for
90% group III, 22 for group IV, 17 for group V, 143
80% for group VI and 2 for group VII. Results for
the subtypes P, P/m, P/M, p/M, M are detailed
70% in the table in abscissa of the figure.
60%
% 50% P
40% I
30% M
20%
10%
0%
VII I III IV V II VI Group
P 0 0 0 0 0 10 87
P
P/m 0 0 0 0 12 16 10

I P/M 0 0 0 18 47 50 3

p/M 0 14 37 41 35 12 0
M
M 100 86 63 41 6 12 0

diarrhoeal syndrome and 17 strains connected to emetic the strains of the phylogenetic group VI, including
syndrome) were positioned across genetic groups II, III, B. weihenstephanensis strains and psychrotolerant rhiz-
IV, V and VII (Fig. 1). The diarrhoeal strains were oidal strains (B. mycoides ‘sensu stricto’ ), did not grow
present in all these five groups, whereas the emetic at 37°C or grew slowly compared with the strains of
strains were only distributed in the mesophilic group III other phylogenetic groups. This temperature constraint
inside subgroup BC05. It is noteworthy that among the and other stresses, including that implied by the intesti-
143 strains that were clustered in the psychrotolerant nal bacteria, greatly decrease the risk of intoxication by
group VI, none have been connected with food poison- such strains. This also may explain why food poisoning
ings. Stenfors and colleagues (2002) showed that most strains are statistically not associated to this group. By
B. weihenstephanensis strains are not cytotoxic on vero contrast, the psychrotolerant strains of genetic group II
cells and grow only with difficulty at 40°C (a temperature can be pathogenic, such as the strains NVH 0861/00,
close to 37°C). In the conditions that were used here, RIVM BC 120, RIVM BC 126.

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
860 M.-H. Guinebretière et al.

Discussion well correlated with the higher evolutionary status of


group VI for the psychrotolerant signature and with its
Psychrotolerance and evolution of the B. cereus group
higher level of adaptation to low temperatures. A similar
The phylogenetic branching order of the different thermal correlation between the mutation rate and the level of
lineages provides information on the direction of pheno- adaptation has been previously shown for the pathogenic
typic change during cladogenesis. The most thermotoler- Escherichia coli (Matic et al., 1997).
ant lineage is basal (groups VII), while strains with lower In conclusion, the global evolution of the B. cereus
thermotolerance or greater psychrotolerance branch later Group seems to be strongly determined by ecological
along the phylogeny (Figs 1 and 3). This indicates a tran- adaptations; the extension of ecological tolerance limits,
sition from a moderately thermotolerant status (group VII) in particular temperature tolerance, may have been an
to a mesophilic status (group I). This was followed by a important mechanism by which strains of the different
shift in growth ranges for the most recently appeared groups (I to VII) adapted to novel environments and his-
groups (II, III, IV, V and VI), resulting in specialization for torically by which these groups have diverged. This rich
groups II and VI (psychrotrophic groups). This strongly adaptive faculty guarantees development of new evolu-
supports the hypothesis of a multiemergence of psychro- tionary lines and long-term maintenance of the B. cereus
tolerance in the B. cereus Group. The differential pres- Group sensu lato.
ence of the rrs signature in all groups shows that all
groups (I to VI) appear to be genetically concerned by this
Practical applications for detecting B. cereus Group
evolution but the relative quantity of this signature accu-
strains in foods
mulated in strains indicates that, in this respect, the
groups do not evolve at the same rate. The present work provides an important complement to
We may speculate that the emergence of more cold- the existing phylogenetic topologies (Helgason et al.,
adapted populations corresponds to the colonization of 2004; Hill et al., 2004; Priest et al., 2004; Sorokin et al.,
new or different environments for which B. cereus organ- 2006) through additional data concerning the B. wei-
isms had to adapt. According to the index reference of henstephanensis, B. mycoides and B. pseudomycoides
Ochman and Wilson (1987), group VII should have species, and through the discovery of a global ecotypic
diverged 125 millions years ago. One hypothesis is that structure. These results have important practical conse-
this group was initially associated to a rare or restricted/ quences. A particularly original result is the existence
isolated thermal niche that supported moderately high of two very different groups of psychrotrophs (groups II
temperatures (up to 50°C) such as hot springs, volcanic and VI). Food manufacturers need fast and effective
warm soil, etc. . . . The transition to a mesophilic status methods of detection for psychrotrophs to predict the
may correspond to the colonization of lower thermal shelf-life of perishable products more accurately and to
niches such as diverse soils, rivers and animals (insects, reduce the incidence of food poisoning by strains of the
mammals). An argument for this is the major genome B. cereus Group. Detection of psychrotolerant signatures
expansions observed for the sequenced strains of groups on the rrs gene (Pruss et al., 1999b) and cspA gene
III, IV and VI (Lapidus et al., 2007), similar to the signifi- (Francis et al., 1998) have been previously proposed.
cant modification of genome size that had previously been Our study shows that the ‘psychrotolerant’ rrs signature
associated to a modification of habitat for Frankia sp. can be regarded as a general indication of a strain’s
strains (Normand et al., 2007). Furthermore, the associa- adaptive potential, but it is not sufficient to accurately
tion between the thermal niches and the genetic groups, identify group II or VI. Furthermore, the ‘psychrotolerant’
strongly correlates the stepwise pattern of diversification cspA signature cannot be defined as a marker for all
observed for the clade made of groups II to VI with the psychrotrophs but just for group VI strains. Finally, the
appearance of diverse biological niches and thus with health risk is not the same for the psychrotolerant strains
different ‘thermal’ niche histories for these genetic groups. depending on their genetic affiliation to group II or group
The thermal niches associated to the mesophilic and VI. The problem therefore arises of how to detect all
psychrotolerant groups may act as microgeographical psychrotrophs and how to discriminate the two types of
separations, preventing competition between each type of psychrotrophs. Because the group II strains are more
population, so that they are able to independently evolve prone to induce food poisoning, a method complemen-
at different rates for psychrotolerance. According to the tary to that proposed by Francis and colleagues (1998)
rrs-ITS1 sequence phylogeny, the psychrotolerant group is needed to allow the detection of all psychrotrophs in
VI that appeared nearly at the same time as the clade food.
made up of groups II to V, seems to be characterized by The highly polyphyletic character of B. cereus sensu
a higher mutation rate. Although the mutation rate data stricto and B. thuringiensis will result in highly different
was recorded on a limited number of strains, this seems characters for strains inside each species. We thus

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
B. cereus ecological diversification 861

propose to add the phylogenetic number (I to VII) to the Screening of psychrotolerant strains and rhizoidal
species name when identifying strains: B. cereus II, III, IV, colony-forming strains
V and VII and B. thuringiensis II, III, IV, V and VI. Tmin and
Psychrotolerant strains were detected by testing growth at
Tmax, as well as rhizoidal morphology of colony and cspA
7°C and at 5°C. Given the high number of strains (425
signature provide a first intention approach for differenti- strains), the experiment was conducted as a rapid screen-
ating strains of the different phylogenetic groups I to VII. ing test. The test was performed on J-agar Petri dishes
The origin of food ingredients and the temperature (Claus and Berkeley, 1986) in two independent experi-
applied during processing of foods and storage will have ments. A colony from a fresh culture (24 h) on J-agar was
a differential selective effect on the bacterial strains of the picked up and inoculated by streaking on a J-agar plate.
B. cereus Group, tightly related to their specific thermal Growth was conducted in incubators at 30°C ⫾ 0.5 for
48 h (control of viability), at 7.1°C ⫾ 0.4 for 21 days and
characteristics. Then for consumers, the health hazards
at 5.1°C ⫾ 0.5 for 21 days. The type strains WSBC
do not concern all ingested strains with the same degree 10204T = LMG 18989T and ATCC 14579T = LMG 6923T were
of risk. Strains of the different phylogenetic groups will not used as positive and negative controls respectively. For
have the same potential to develop under intestinal con- each experiment and each incubation condition, two J-agar
ditions, the strains of phylogenetic groups VII, III and IV Petri dishes per strain were inoculated. Strains were con-
probably being the best performers. sidered as positive for growth when colonies appeared after
Currently, the public health authorities do a clear dis- the above incubation times.
The appearance of rhizoidal colony-forming strains on
tinction between B. cereus sensu stricto (recognized as
J-agar after a 48 h culture at 30°C was also monitored for
a potential food poisoning agent of risk group 2) and each strain.
the species B. thuringiensis, B. weihenstephanensis,
B. mycoides and B. pseudomycoides (not recognized as
potential pathogenic agents). As suggested above, the Fluorescent Amplified Fragment Length
risk seems to be rather associated to the phylogenetic Polymorphism analysis
group. For example a B. cereus II or a B. thuringiensis Total DNA was extracted for each strain as described previ-
VI will not have the same food poisoning potential as, ously (Guinebretiere and Nguyen-The, 2003). Digestion and
respectively, a B. cereus VI or a B. thuringiensis III. ligation steps were carried out according to the protocols
Therefore, we recommend that risk groups, instead of of Ticknor and colleagues (2001) with slight modifications.
being defined by reference to the current phenotypic The selective polymerase chain reaction (PCR), capillary
electrophoresis and numerical analyses were performed as
species, be defined by reference to the seven major
described previously (Thompson et al., 2001).
phylogenetic groups established here. It is certain also The fAFLP banding patterns were converted into vector
that some additional studies are needed to quantitatively format. To find a near-optimal discretization method, a
characterize the pathogenic potential conferred by each number of alternative discretization algorithms were evalu-
group. ated in order to preserve maximal information content of the
original fingerprint patterns (Dawyndt et al., 2005). The most
conservative vector representation resulted from a discreti-
Experimental procedures
zation by the sliding window algorithm introduced by Austin
Bacterial strains and their origins and colleagues (2004), with a position tolerance e of 0.7%
and a resolution d of 0.1%, producing a binary vector repre-
We randomly selected 425 independent strains from different sentation of length 994.
laboratories and international collections in Europe and the To obtain an unbiased view of the relationships between
USA (see Table S1). These strains covered a broad spectrum the fAFLP profiles, we compared the outcomes of two differ-
of different geographic origins, substrate origins, RAPD ent mathematical approaches for classifying the genomic
profiles, serotypes and virulence characteristics. To get a fingerprint patterns. The first classification resulted from a
complete coverage of the B. cereus group, strains from hierarchical clustering, using the Dice similarity coefficient
all six currently defined species were studied. Also, strains and applying the UPGMA (Sneath and Sokal, 1973). Bootstrap
representative of species (B. cereus ATCC 14579T, values reflecting the robustness of a classification were
B. thuringiensis CIP 53137T, B. mycoides DSM 2048T, obtained from 1000 replicates (Bionumerics 3.5 software).
B. pseudomycoides DSM 12442T, B. weihenstephanensis The second classification resulted from a non-hierarchical
WSBC 10204T and B. anthracis CEB 94–0040) were clustering. It optimized a given information theoretic
included. Given their known monomorphic characteristics, we expression. Indeed, the binary vector representation also
restricted the number of B. anthracis and emetic strains to made it possible to apply a minimization of stochastic com-
represent the maximum diversity observed for each entity plexity (Gyllenberg et al., 1998) for classification of the fAFLP
according to tandem repeat polymorphism (Le Fleche et al., fingerprint patterns. Minimization of stochastic complexity
2001) or RAPD pattern (Ehling-Schulz et al., 2005). was performed using the BinClass software package (Gyllen-
In order to determine the position of food poisoning strains, berg et al., 2001). The robustness of the classification is
we included 49 strains connected to diarrhoeal syndrome and reflected by a high level of overall congruence between the
17 strains connected to emetic syndrome (see Table S1). two methodologies (Dawyndt et al., 2005).

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
862 M.-H. Guinebretière et al.

The centroid of a class is, by definition, the vector giving Targeting of thermal DNA signatures for the
the frequencies of one’s for the different attributes. Rounding discrimination of psychrotolerant genotypes
off each component of the centroid to the nearest binary
value (0 or 1) gives the hypothetical mean organism (HMO) A total of 252 strains were tested for the cspA signature
(Sneath and Sokal, 1973). As a measure of the heterogeneity sequence defined by Francis and colleagues (1998), includ-
of a class, we chose its distortion, which is the average ing all psychrotolerant strains (by reference to the screening
number of bits by which the members of the class differ from test) as well as a subset of mesophilic representatives of
the HMO. An alternative quantifier for describing the hetero- each mesophilic fAFLP groups (20% minimum). For each of
geneity of a class is the Shannon code length (Cover and these bacterial strains, the cspA signature was detected by a
Thomas, 1991), which we also included. triplex PCR assay as previously described (Francis et al.,
1998), with minor modifications (use of Goldstar DNA poly-
merase (Eurogentec), 50 ng of DNA per reaction and a 15 ml
panC gene sequence analysis final volume).
A total of 246 strains, including all the psychrotolerant
The panC gene was chosen from a pilot experiment on mul- strains (by reference to the screening test) and mesophilic
tiple locus sequencing of B. cereus strains (Candelon et al., representatives of each mesophilic fAFLP group strains (20%
2004). This gene was sequenced for 67 selected strains, minimum), were tested for the rrs ribosomal DNA signature
representing AFLP groups. Polymerase chain reaction ampli- defined by Pruss and colleagues (1999b). The rrs ribosomal
fication and sequencing procedures were carried out as pre- DNA signature was characterized for each strain using the
viously described (Candelon et al., 2004). PCR-RFLP protocol described by Pruss and colleagues
The consensus sequences were compared using the (1999b), with minor modifications (use of Goldstar DNA poly-
multiple alignment program CLUSTAL ¥ 1.8 (Thompson et al., merase (Eurogentec) and a 15 ml final volume). The resulting
1997). A phylogenetic tree was constructed using the PCR product representing a partial DNA fragment of rrs gene
Neighbour-Joining (NJ) method (Saitou and Nei, 1987), and (473 bp) was submitted to digestion by SspI restriction
NJplot software (Perriere and Gouy, 1996) was used to gen- enzyme. A 10 ml aliquot of the digestion mixture (containing
erate a graphic representation of the resulting tree. Bootstrap 5 U SspI, 2 mg BSA and SspI buffer) was added directly to the
estimates (Felsenstein, 1985) were obtained from 1000 15 ml of PCR product and incubated at 37°C for 4 h. The
replicates. resulting DNA fragments were separated on 2% agarose gel.
The ‘psychrotolerant’ rrs signature was detected by the pres-
ence of two DNA fragments (351 bp and 122 bp) after SspI
rrs gene and Intergenic transcribed spacer rrs – rrl digestion (Pruss et al., 1999b), whereas in the absence of
genes (ITS1) sequence analysis the targeted restriction site, only one band of 473 bp was
present. As the relative number of rrs copies with this signa-
The rrs and ITS1 DNA genes of 12 representative strains,
ture correlates with the relative intensity of the characteristic
corresponding to the strains nearest to the HMO of each BC
351 bp band as compared with the 473 bp band (Pruss et al.,
class, were amplified by PCR, using the primers described by
1999b), we established three main thermal genotypes
Normand and colleagues (1996). The PCR reactions and
(M = mesophilic genotype, I = intermediate genotype,
amplifications were performed in a final volume of 90 ml,
P = psychrotolerant genotype) and five thermal subtypes
following the conditions described previously (Guinebretiere
(P, P/m, P/M, p/M, M) (see Fig. 4A).
et al., 2001).
The amplification products were purified with a Spin PCR
purification kit (ROCHE diagnostic, Mannheim, Germany) Growth temperature ranges
and were sequenced by MWG (Martinsried, Germany). Five
primers were used in the sequencing reaction to obtain com- A subset of 75 strains representative of each fAFLP group
plete rrs gene and ITS1 sequence (about 1800 bp): S6, S10, and thermal genotype (strains that have similar psychrotoler-
S15, S17 and FGPL 132 (Normand et al., 1996; Guinebre- ant signatures and growth properties) was tested for growth
tiere et al., 2001). ability at extreme temperature. Growth was determined using
To determine the phylogeny of the representative strains nutrient agar and J-agar slants, as described in the Bergey’s
for which the entire rrs-ITS1 was sequenced, phylogenetic Manual of Systematic Bacteriology (Claus and Berkeley,
trees were constructed by (i) the NJ method (Saitou and Nei, 1986). Systematic growth tests at 7°C, 8°C, 37°C and 43°C
1987) using the two-parameter substitution rate method of were included, in addition to the conventional tests at 5°C
Kimura (1980) (ii) the Maximum-Parsimony method (MP) intervals described in the Bergey’s Manual, by reference to
(Kluge and Farris, 1969) and (iii) the Maximum-Likelihood the upper limit (7°C) for defining a psychrotolerant strain
method (ML) (Felsenstein, 1981). Bootstrap estimates (Francis et al., 1998), the standard temperature used by
(Felsenstein, 1985) were obtained from 1000 replicates for several Food Safety Agencies in European countries (8°C),
both the NJ and MP methods. The different analyses were the temperature of human body (37°C), and a taxonomic
implemented using the Phylo-win software (Perriere and criterion (43°C) used for B. cereus Group (Lechner et al.,
Gouy, 1996). Graphic representation of the resulting trees 1998). For the tests at low temperatures, bath temperatures
was obtained using the NJPlot software (Perriere and Gouy, were stabilized at 7, 8, 10, 15 or 20°C in an RTE-111 Re-
1996). A test of comparison of mutation rates was conducted frigerated Bath/Circulator (Neslab Instruments, Newington,
using the RRT procedure (Robinson-Rechavi and Huchon, USA), while for the higher temperature tests, bath tempera-
2000). tures were stabilized at 35, 37, 40, 43, 45, 50 or 55°C in a

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
B. cereus ecological diversification 863
PolyStat 55 heating circulating bath (Bioblock Scientific). versity of Ghent, the Mission des Relations Internationales of
Temperature fluctuation was not greater than ⫾0.2°C for all INRA (Paris, France) and the CEPIA and MICA Departments
tested temperatures. Tubes were immersed in the water of INRA. We are grateful to the Bijzonder Onderzoeksfonds
baths at the appropriate temperatures until equilibrated, and (BOF) of Ghent, the BCCMTM/LMG Bacteria Collection
then inoculated. Cultures growths were examined after 3 and (Ghent) and the FAPESP (Brazil) for their contribution.
5 days for the high temperature tests and every week up to Strains of interest were kindly provided by the USDA
21 days for the low temperature tests. (USA), the DSM (Germany) and the ‘Centre du Bouchet’
The type strains of B. weihenstephanensis (WSBC (France).
10204T), B. licheniformis (CIP 52.71T) and B. circulans
(LMG 6926T) were used as negative or positive controls for
the extreme conditions of growth (7°C, 8°C, 50°C, 55°C).
References

Statistics Anderson, K.G. (1997) Food issues. In Food Industries


Manual. Ranken, M.D., Kill, R.C., and Baker, C.G.J. (eds).
For comparison of the eco-physiological characteristics of London, UK: Springer, pp. 603–617.
strains, we established different key variables. Growth range Austin, B., Dawyndt, P., Gyllenberg, M., Koski, T., Lund, T.,
(R) of each strain was defined by qualitative data correspond- Swings, J., and Thompson, F.L. (2004) Sliding window
ing to the association of its minimal growth temperature (Tmin) discretization: a new method for multiple band matching of
and maximum growth temperature (Tmax). Twelve different bacterial genotyping fingerprints. Bull Math Biol 66: 1575–
values in total were recorded for R among our strains sample: 1596.
R01 (20–50°C), R02 (15–45°C), R03 (15–43°C), R04 Becker, H., Schaller, G., Von Wiese, W., and Terplan, G.
(15–40°C), R05 (10–45°C), R06 (10–43°C), R07 (10–40°C), (1994) Bacillus cereus in infant foods and dried milk
R08 (8–40°C), R09 (7–40°C), R10 (7–37°C), R11 (5–37°C), products. Int J Food Microbiol 23: 1–15.
R12 (5–35°C). Concerning habitats of strains, we categorized Candelon, B., Guilloux, K., Ehrlich, S.D., and Sorokin, A.
the origin of strains according to the thermal niche to which (2004) Two distinct types of rRNA operons in the Bacillus
they were associated: The thermal niche #1 corresponded to cereus group. Microbiology 150: 601–611.
highly fluctuating temperatures frequently covering low tem- Carlin, F., Guinebretiere, M.H., Choma, C., Pasqualini, R.,
peratures (ⱕ 8°C); it included soils/air/rivers/plant surface of Braconnier, A., and Nguyen-The, C. (2000) Spore-forming
Northern Europe, cold surfaces of dairy equipment, pasteur- bacteria in commercial cooked, pasteurised and chilled
ized and refrigerated foods (milk, vegetables . . .), raw veg- vegetable purees. Food Microbiol 17: 153–165.
etables conserved at low temperature. The thermal niche #2 Claus, D., and Berkeley, R.C.W. (1986) Genus Bacillus
corresponded to intermediate temperatures with low fluctua- Cohn 1872, 174AL. In Bergey’s Manual of Systematic
tions; it included insects, Asian or pacific soils, indoor air, Bacteriology. Sneath, P.H.A., Mair, N.S., Sharpe, M.E.,
transformed or raw foods not refrigerated and not heat- and Holt, J.G. (eds). Baltimore, MD, USA: Williams &
treated (milk products such as cheese, vegetables), surfaces Wilkins, pp. 1105–1139.
of equipment not associated with refrigeration, cakes. The Cover, T.M., and Thomas, J.A. (1991) Elements of Informa-
thermal niche #3 corresponded to warm temperatures, tion Theory. New York, NY, USA: John Wiley & Sons.
including mammals, faeces, dehydrated foods (starch, pro- Dawyndt, P., Thompson, F.L., Austin, B., Swings, J., Koski,
teins, herbs and spices), cooked foods (warm meals, rice, T., and Gyllenberg, M. (2005) Application of sliding-window
pasta . . .). discretization and minimization of stochastic complexity for
Tests of independence and particularly the significance per the analysis of fAFLP genotyping fingerprint patterns of
case (Fisher’s exact test) were carried out using the software Vibrionaceae. Int J Syst Evol Microbiol 55: 57–66.
package XLSTAT 2007.7 to determine to what extent the Drobniewski, F.A. (1993) Bacillus cereus and related
different categories of the above variables were significantly species. Clin Microbiol Rev 6: 324–338.
associated to one or the other of the different genetic groups. Ehling-Schulz, M., Svensson, B., Guinebretiere, M.H., Lind-
back, T., Andersson, M., Schulz, A., et al. (2005) Emetic
toxin formation of Bacillus cereus is restricted to a single
Nucleotide sequence accession numbers evolutionary lineage of closely related strains. Microbiology
151: 183–197.
Nucleotide sequence data have been submitted to the EMBL/
Fagerlund, A., Brillard, J., Fürst, R., Guinebretiere, M.H., and
GenBank databases under Accession Nos AM747044 to
Granum, P.E. (2007) Toxin production in a rare and geneti-
AM747097, DQ301053, DQ301054, DQ301062, DQ301066,
cally remote cluster of strains of the Bacillus cereus group.
DQ301068, DQ301078, DQ301089, DQ301090, DQ301137,
BCM Microbiology 2007, 7: 43, doi:10.1186/1471-2180-
DQ301426, DQ301433, AY224725 for the partial panC gene
7-43
sequences and AM747220 to AM747234 for the rrs-ITS1
Felsenstein, J. (1981) Evolutionary trees from DNA
sequences.
sequences: a maximum likelihood approach. J Mol Evol
17: 368–376.
Acknowledgements Felsenstein, J. (1985) Confidence limits on phylogenies: an
approach using the bootstrap. Evolution 39: 783–791.
This work was supported by the European Commission Francis, K.P., Mayr, R., von Stetten, F., Stewart, G., and
(QLK1-CT-2001–00854) and by specific funds from the Uni- Scherer, S. (1998) Discrimination of psychrotrophic and

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
864 M.-H. Guinebretière et al.
mesophilic strains of the Bacillus cereus group by PCR F., Ramisse, V., et al. (2001) A tandem repeats database
targeting of major cold shock protein genes. Appl Environ for bacterial genomes: application to the genotyping of
Microbiol 64: 3525–3529. Yersinia pestis and Bacillus anthracis. BioMed Central
Guinebretiere, M.H., and Nguyen-The, C. (2003) Sources of Microbiol 1: 2. [WWW document]. URL http://www.
Bacillus cereus contamination in a pasteurized zucchini biomedcentral.com/1471-2180/1471/1472
puree processing line, differentiated by two PCR-based Lechner, S., Mayr, R., Francis, K.P., Pruss, B.M., Kaplan, T.,
methods. FEMS Microbiol Ecol 43: 207–215. Wiessner-Gunkel, E., et al. (1998) Bacillus weihen-
Guinebretiere, M.H., and Sanchis, V. (2003) Bacillus cereus stephanensis sp. nov. is a new psychrotolerant species of
sensu lato. Bull Soc Fr Microbiol 18: 95–103. the Bacillus cereus group. Int J Syst Bacteriol 48: 1373–
Guinebretiere, M.H., Berge, O., Normand, P., Morris, C., 1382.
Carlin, F., and Nguyen-The, C. (2001) Identification of bac- Matic, Y., Radman, M., Taddei, F., Picard, B., Doit, C.,
teria in pasteurized zucchini purees stored at different Bingen, E., et al. (1997) Highly variable mutation rates in
temperatures and comparison with those found in other commensal and pathogenic Escherichia coli. Science
pasteurized vegetable purees. Appl Environ Microbiol 277: 1833–1834.
67: 4520–4530. Nakamura, L.K. (1998) Bacillus pseudomycoides sp. nov. Int
Gyllenberg, H.G., Gyllenberg, M., Koski, T., and Lund, T. J Syst Bacteriol 48: 1031–1035.
(1998) Stochastic complexity as a taxonomic tool. Comput Normand, P., Ponsonnet, C., Nesme, X., Neyra, M., and
Methods Programs Biomed 56: 11–22. Simonet, P. (1996) ITS analysis of prokaryotes. In Molecu-
Gyllenberg, M., Koski, T., and Lund, T. (2001) BinClass: a lar Microbiology Ecology Manual. Akkermans, A.D.L., van
software package for classifying binary vectors User’s Elsas, J.D. & De Bruijn, F.J. (eds). Dordrecht, The Nether-
guide. TUCS Technical Report No. 411, June 2001, 58 lands: Kluwer Academic publishers, pp. 1–12.
pages. Normand, P., Lapierre, P., Tisa, L.S., Gogarten, J.P., Alloisio,
Helgason, E., Caugant, D.A., Olsen, I., and Kolsto, A.B. N., Bagnarol, E., et al. (2007) Genome characteristics of
(2000) Genetic structure of population of Bacillus cereus facultatively symbiotic Frankia. sp. strains reflect host
and B. thuringiensis isolates associated with periodontitis range and host plant biogeography. Genome Res 17:
and other human infections. J Clin Microbiol 38: 1615– 7–15.
1622. Ochman, H., and Wilson, A.C. (1987) Evolution in bacteria:
Helgason, E., Tourasse, N.J., Meisal, R., Caugant, D.A., and evidence for an universal substitution rate in cellular
Kolsto, A.B. (2004) Multilocus sequence typing scheme for genomes. J Mol Evol 26: 74–86.
bacteria of the Bacillus cereus group. Appl Environ Micro- Perriere, G., and Gouy, M. (1996) WWW-Query: an on-line
biol 70: 191–201. retrieval system for biological sequence banks. Biochimie
Hill, K.K., Ticknor, L.O., Okinaka, R.T., Asay, M., Blair, H., 78: 364–369.
Bliss, K.A., et al. (2004) Fluorescent amplified fragment Pirttijarvi, T.S.M., Ahonen, L.M., Maunuksela, L.M., and
length polymorphism analysis of Bacillus anthracis, Salkinoja-Salonen, M.S. (1998) Bacillus cereus in a whey
Bacillus cereus, and Bacillus thuringiensis isolates. Appl process. Int J Food Microbiol 44: 31–41.
Environ Microbiol 70: 1068–1080. Pirttijarvi, T.S.M., Andersson, M.A., and Salkinoja-Salonen,
Holmes, J.R., Plunkett, T., Pate, P., and Roper, W.L. (1981) M.S. (2000) Properties of Bacillus cereus and other bacilli
Emetic food poisoning caused by Bacillus cereus. Arch contaminating biomaterial-based industrial processes. Int J
Intern Med 141: 766–767. Food Microbiol 60: 231–239.
Keim, P., Kalif, A., Schupp, J., Hill, K., Travis, S.E., Rich- Priest, F.G., Barker, M., Baillie, L.W.J., Holmes, E.C., and
mond, K., et al. (1997) Molecular evolution and diversity in Maiden, M.C.J. (2004) Population structure and evolution
Bacillus anthracis as detected by amplified fragment length of the Bacillus cereus group. J Bacteriol 186: 7959–7970.
polymorphism markers. J Bacteriol 179: 818–824. Pruss, B.M., Dietrich, R., Nibler, B., Martlbauer, E., and
Kimura, M. (1980) A simple method for estimating evolution- Scherer, S. (1999a) The hemolytic enterotoxin HBL is
ary rates of base substitutions through comparative studies broadly distributed among species of the Bacillus cereus
of nucleotide sequences. J Mol Evol 16: 111–120. group. Appl Environ Microbiol 65: 5436–5442.
Kluge, A.G., and Farris, J.S. (1969) Quantitative phyletics Pruss, B.M., Francis, K.P., von Stetten, F., and Scherer, S.
and the evolution of anurans. Syst Zool 18: 1–32. (1999b) Correlation of 16S ribosomal DNA signature
Kramer, J.M., and Gilbert, R.J. (1989) Bacillus cereus and sequences with temperature-dependent growth rates of
other Bacillus species. In Food Borne Baterial Pathogens. mesophilic and psychrotolerant strains of the Bacillus
Doyle, M.P. (ed.). New York, NY, USA: Marcel Dekker, cereus group. J Bacteriol 181: 2624–2630.
pp. 21–70. Robinson-Rechavi, M., and Huchon, D. (2000) RRTree:
Lapidus, A., Goltsman, E., Auger, S., Galleron, N., Ségurens, relative-rate tests between groups of sequences on a
B., Land, M.L., et al. (2007) Extending the Bacillus cereus phylogenetic tree. Bioinformatics 16: 296–297.
group genomics to putative food-borne pathogens of differ- Saitou, N., and Nei, M. (1987) The neighbor-joining method:
ent toxicity. Chemico-Biol Int 2007, doi:10.1016/j.cbi.2007. a new method for reconstructing phylogenetic trees. Mol
03.003. Biol Evol 4: 406–425.
Larsen, H.D., and Jorgensen, K. (1997) The occurrence of Schnepf, E., Crickmore, N., Van Rie, J., Lereclus, D., Baum,
Bacillus cereus. Danish pasteurized milk. Int J Food Micro- J., Feitelson, J., et al. (1998) Bacillus thuringiensis and its
biol 34: 179–186. pesticidal crystal proteins. Microbiol Mol Biol Rev 62: 775–
Le Fleche, P., Hauck, Y., Onteniente, L., Prieur, A., Denoeud, 806.

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865
B. cereus ecological diversification 865
Shinagawa, K. (1990) Analytical methods for Bacillus cereus F.M. (1997) Isolation and characterisation of Bacillus
and other Bacillus. Int J Food Microbiol 10: 125–141. cereus from pasteurised milk in household refrigerators in
Sneath, P.H.A., and Sokal, R.R. (1973) Numerical Tax- The Netherlands. Int J Food Microbiol 34: 307–318.
onomy: The Principles and Practices of Numerical Thompson, J.D., Gibson, T.J., Plewniak, F., Jeanmougin, F.,
Classification. San Francisco, CA, USA: Freeman, W. H. and Higgins, D.G. (1997) The CLUSTAL–windows interface:
Sorokin, A., Candelon, B., Guilloux, K., Galleron, N., flexible strategies for multiple sequence alignment aided by
Wackerow-Kouzova, N.W.-K., Ehrlich, S.D., et al. (2006) quality analysis tools. Nucleic Acids Res 25: 4876–4882.
Multiple-locus sequence typing analysis of Bacillus cereus Thompson, F.L., Hoste, B., Vandemeulebroecke, K., and
and Bacillus thuringiensis reveals separate clustering and Swings, J. (2001) Genomic diversity amongst Vibrio iso-
a distinct population structure of psychrotrophic strains. lates from different sources determined by fluorescent
Appl Environ Microbiol 72: 1569–1578. amplified fragment length polymorphism. Syst Appl Micro-
Stenfors, L.P., and Granum, P.E. (2001) Psychrotolerant biol 24: 520–538.
species from the Bacillus cereus group are not necessarily Ticknor, L.O., Kolsto, A.B., Hill, K.K., Keim, P., Laker, M.T.,
Bacillus weihenstephanensis. FEMS Microbiol Lett 197: Tonks, M., and Jackson, P.J. (2001) Fluorescent amplified
223–228. fragment length polymorphism analysis of Norwegian
Stenfors, L.P., Mayr, R., Scherer, S., and Granum, P.E. Bacillus cereus and Bacillus thuringiensis soil isolates.
(2002) Pathogenic potential of fifty Bacillus weihen- Appl Environ Microbiol 67: 4863–4873.
stephanensis strains. FEMS Microbiol Lett 215: 47–51.
von Stetten, F., Mayr, R., and Scherer, S. (1999) Climatic
influence on mesophilic Bacillus cereus and psychrotoler- Supplementary material
ant Bacillus weihenstephanensis populations in tropical,
The following supplementary material is available for this
temperate and alpine soil. Environ Microbiol 1: 503–515.
article online:
Svensson, B., Eneroth, A., Brendehaug, J., Molin, G., and
Table S1. Strains from this study and the genetic groups to
Christiansson, A. (2000) Involvement of a pasteurizer in
which they belong.
the contamination of milk by Bacillus cereus in a commer-
cial dairy plant. J Dairy Res 67: 455–460. This material is available as part of the online article from
Te Giffel, M.C., Beumer, R.R., Granum, P.E., and Rombouts, http://www.blackwell-synergy.com

© 2007 The Authors


Journal compilation © 2007 Society for Applied Microbiology and Blackwell Publishing Ltd, Environmental Microbiology, 10, 851–865

You might also like