discrimination of bacillus anthracis and closely related microorganisms by analysis of 16s and 23s...

24
Chemico-Biological Interactions 171 (2008) 212–235 Available online at www.sciencedirect.com Discrimination of Bacillus anthracis and closely related microorganisms by analysis of 16S and 23S rRNA with oligonucleotide microarray Sergei G. Bavykin a,, Vladimir M. Mikhailovich b , Vladimir M. Zakharyev b , Yuri P. Lysov b , John J. Kelly c , Oleg S. Alferov a , Igor M. Gavin a , Alexander V. Kukhtin a,1 , Joany Jackman d,2 , David A. Stahl e , Darrell Chandler a,1 , Andrei D. Mirzabekov a,b a Center for Environmental and Security Science and Technology, Argonne National Laboratory, Argonne, IL 60439, USA b Engelhardt Institute of Molecular Biology, Moscow 117984, Russia c Department of Biology, Loyola University Chicago, Chicago, IL 60626, USA d Georgetown University, School of Medicine, Department of Biochemistry and Molecular Biology, Washington, DC 70001, USA e Department of Civil and Environmental Engineering, University of Washington, Seattle, WA 98195-2700, USA Available online 12 September 2007 Abstract Analysis of 16S rRNA sequences is a commonly used method for the identification and discrimination of microorganisms. However, the high similarity of 16S and 23S rRNA sequences of Bacillus cereus group organisms (up to 99–100%) and repeatedly failed attempts to develop molecular typing systems that would use DNA sequences to discriminate between species within this group have resulted in several suggestions to consider B. cereus and B. thuringiensis, or these two species together with B. anthracis, as one species. Recently, we divided the B. cereus group into seven subgroups, Anthracis, Cereus A and B, Thuringiensis A and B, and Mycoides A and B, based on 16S rRNA, 23S rRNA and gyrB gene sequences and identified subgroup-specific makers in each of these three genes. Here we for the first time demonstrated discrimination of these seven subgroups, including subgroup Anthracis, with a 3D gel element microarray of oligonucleotide probes targeting 16S and 23S rRNA markers. This is the first microarray enabled identification of B. anthracis and discrimination of these seven subgroups in pure cell cultures and in environmental samples using rRNA sequences. The microarray bearing perfect match/mismatch (p/mm) probe pairs was specific enough to discriminate single nucleotide polymorphisms (SNPs) and was able to identify targeted organisms in 5min. We also demonstrated the ability of the microarray to determine subgroup affiliations for B. cereus group isolates without rRNA sequencing. Correlation of these seven subgroups with groupings based on multilocus sequence typing (MLST), fluorescent amplified fragment length polymorphism analysis (AFLP) and multilocus enzyme electrophoresis (MME) analysis of a wide spectrum of different genes, and the demonstration of subgroup-specific differences in toxin profiles, psychrotolerance, and the ability to harbor some plasmids, suggest that these seven subgroups are not based solely on neutral genomic polymorphisms, but instead reflect differences in both the genotypes and phenotypes of the B. cereus group organisms. © 2007 Elsevier Ireland Ltd. All rights reserved. Keywords: Bacillus anthracis; 16S rRNA; 23S rRNA; Microarray; SNP; Environmental sample Corresponding author. Tel.: +1 630 252 3980. E-mail address: [email protected] (S.G. Bavykin). 1 Current address: Akonni Biosystems, Frederick, MD 21701, USA. 2 Current address: Applied Physics Laboratory, Johns Hopkins University, Laurel, MD 20723, USA. 0009-2797/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.cbi.2007.09.002

Upload: independent

Post on 27-Feb-2023

1 views

Category:

Documents


0 download

TRANSCRIPT

Chemico-Biological Interactions 171 (2008) 212–235

Available online at www.sciencedirect.com

Discrimination of Bacillus anthracis and closely relatedmicroorganisms by analysis of 16S and 23S rRNA with

oligonucleotide microarray

Sergei G. Bavykin a,∗, Vladimir M. Mikhailovich b, Vladimir M. Zakharyev b,Yuri P. Lysov b, John J. Kelly c, Oleg S. Alferov a, Igor M. Gavin a,

Alexander V. Kukhtin a,1, Joany Jackman d,2, David A. Stahl e,Darrell Chandler a,1, Andrei D. Mirzabekov a,b

a Center for Environmental and Security Science and Technology, Argonne National Laboratory, Argonne, IL 60439, USAb Engelhardt Institute of Molecular Biology, Moscow 117984, Russia

c Department of Biology, Loyola University Chicago, Chicago, IL 60626, USAd Georgetown University, School of Medicine, Department of Biochemistry and Molecular Biology, Washington, DC 70001, USA

e Department of Civil and Environmental Engineering, University of Washington, Seattle, WA 98195-2700, USA

Available online 12 September 2007

Abstract

Analysis of 16S rRNA sequences is a commonly used method for the identification and discrimination of microorganisms.However, the high similarity of 16S and 23S rRNA sequences of Bacillus cereus group organisms (up to 99–100%) and repeatedlyfailed attempts to develop molecular typing systems that would use DNA sequences to discriminate between species within thisgroup have resulted in several suggestions to consider B. cereus and B. thuringiensis, or these two species together with B. anthracis,as one species. Recently, we divided the B. cereus group into seven subgroups, Anthracis, Cereus A and B, Thuringiensis A and B,and Mycoides A and B, based on 16S rRNA, 23S rRNA and gyrB gene sequences and identified subgroup-specific makers in each ofthese three genes. Here we for the first time demonstrated discrimination of these seven subgroups, including subgroup Anthracis,with a 3D gel element microarray of oligonucleotide probes targeting 16S and 23S rRNA markers. This is the first microarrayenabled identification of B. anthracis and discrimination of these seven subgroups in pure cell cultures and in environmentalsamples using rRNA sequences. The microarray bearing perfect match/mismatch (p/mm) probe pairs was specific enough todiscriminate single nucleotide polymorphisms (SNPs) and was able to identify targeted organisms in 5 min. We also demonstratedthe ability of the microarray to determine subgroup affiliations for B. cereus group isolates without rRNA sequencing. Correlation

of these seven subgroups with groupings based on multilocus sequence typing (MLST), fluorescent amplified fragment lengthpolymorphism analysis (AFLP) and multilocus enzyme electrophoresis (MME) analysis of a wide spectrum of different genes, andthe demonstration of subgroup-specific differences in toxin profiles, psychrotolerance, and the ability to harbor some plasmids,

tral ge

suggest that these seven subgroups are not based solely on neu the genotypes and phenotypes of the B. cereus group organisms.© 2007 Elsevier Ireland Ltd. All rights reserved.

Keywords: Bacillus anthracis; 16S rRNA; 23S rRNA; Microarray; SNP; Env

∗ Corresponding author. Tel.: +1 630 252 3980.E-mail address: [email protected] (S.G. Bavykin).

1 Current address: Akonni Biosystems, Frederick, MD 21701, USA.2 Current address: Applied Physics Laboratory, Johns Hopkins University,

0009-2797/$ – see front matter © 2007 Elsevier Ireland Ltd. All rights reservdoi:10.1016/j.cbi.2007.09.002

nomic polymorphisms, but instead reflect differences in both

ironmental sample

Laurel, MD 20723, USA.

ed.

logical

1

igsmswcedsahoiT

mscbaptbbts(faBfad

ceradbctohecrp

S.G. Bavykin et al. / Chemico-Bio

. Introduction

Bacillus anthracis, the causative agent of the highlynfectious disease anthrax, belongs to the Bacillus cereusroup, which also contains six other closely relatedpecies: B. cereus, Bacillus thuringiensis, Bacillusycoides, Bacillus pseudomycoides, Bacillus weihen-

tephanensis and Bacillus medusa [1–6]. B. anthracis,hich in recent years has become known as a biologi-

al weapon and terrorist tool, can be isolated from thenvironment. For example, B. anthracis is periodicallyetected in sewer water and in soil, among animals andtaff in agricultural settings and in zoos, and epizooticsmong bison are rather common in Canada. While thereas been a general decrease in the number of B. anthracisutbreaks in the U.S., Canada and Europe, anthrax stills a significant problem in West Africa, Spain, Greece,urkey, Albania, Romania and Central Asia [7,8].

Related organisms in the B. cereus group are com-on in samples of food, water or soil [2,9–15], and

ome of the non-anthrax B. cereus group bacteria canause human infections and illness [2,10,16] which cane rather severe [17–19]. For example, B. cereus isubiquitous soil bacterium and opportunistic human

athogen, with six recognized types of clinical infec-ions and disease: (i) local infections, particularly ofurns, traumatic or post-surgical wounds, and eye; (ii)acterimia and septicemia; (iii) central nervous sys-em infections, including meningitis, abscesses, andhunt associated infections; (iv) respiratory infections;v) endosub-circuititis and perisub-circuititis; and (vi)ood poisoning, characterized by toxin-induced emeticnd diarrheagenic syndromes (2). In the food industry,. cereus and B. weihenstephanensis are important as

ood spoilage organisms, easily contaminating raw foodsnd processing equipment, including equipment used inairies and paper mills [2,9–11,14,16,19–22].

B. thuringiensis is widely used as a biological insecti-ide, accounting for more than 90% of the biopesticidesmployed worldwide [23,24]. Despite a long safetyecord, however, there is increasing public concernbout the potential pathogenicity of B. thuringiensisue to sporadic but sometimes severe infections causedy this species, including diarrhea, food poisoning,orneal ulcer, cellulites, and burn/war wound infec-ions. Like B. cereus, B. thuringiensis produces a varietyf virulence factors, which include phospholipase C,emolysin and enterotoxins [23,25–34], and diarrheal

nterotoxin production was detected in some commer-ial B. thuringiensis-based insecticides [35]. For theseeasons, many crop customers require agricultural com-anies to control for the presence of B. thuringiensis

Interactions 171 (2008) 212–235 213

in their products. There is also evidence that specificstrains of B. cereus and B. thuringiensis are associatedwith periodontitis and other oral infections [36].

B. mycoides and B. pseudomicoides are ubiquitous insoil and water, and there are toxigenic strains within eachof these species [37–43]. Recent findings indicated thatB. mycoides, like B. thuringiensis, may produce pestici-dal metabolites [44]. New genomic data indicated that B.medusa is a variant of B. thuringiensis, suggesting thatB. medusa may possess or acquire pathogenic potential.Therefore, detection and surveillance of all microorgan-isms comprising the B. cereus group is of significantinterest in many aspects of worldwide public healthmicrobiology, from confirming acts of bioterrorism todisease surveillance and outbreak prevention.

Besides traditional microbiological techniques, avariety of molecular biological methods have beendescribed for the selective identification of B. anthracis.Most of them use specific genes or proteins as targets forrecognition and discrimination from related microor-ganisms [45–48]. Genomic methods target unique genesequences located on plasmids or in bacterial chromo-somes. These methods include: microarray detectionof genomic markers [33,49–51]; direct testing ofplasmid pXO1 DNA with specific probes [52]; real-timemultiplex PCR strand specific detection of CAP-C geneon the pXO2 plasmid [53]; ribotyping of 16S/23S rRNAgenes [54–56] and their internal transcribed spacers(16S–23S rRNA ITS) [49,57,58]; detecting variationsin gyrA–gyrB intergenic spacer regions [57] and inrpoB gene [59]; identification markers in gyrA and gyrBgenes [60,61] and in other chromosomal sequences[62]; analysis of putative genome-integrated prophages[63]; detection of single nucleotide and repetitiveelement polymorphisms [64–66]; AFLP analysis ofplasmid and chromosomal sequences [67–69]; MMEof chromosomal genes [70], PCR-ELISA [71]; and onchip PCR amplification of anthrax toxin genes [72].Utilization of genomic markers located in single-copygenes for identification of B. anthracis and discrimina-tion of it from related microorganisms is complicatedby horizontal gene transfer (HGT), which affects upto 32% of bacterial genomes [73–75], and by othermechanisms of genome reorganization [76]. Toxingenes are the subject of HGT particularly often [77].On the other hand, discriminative identification of B.cereus group members other then B. anthracis usinggenomic markers was impossible until very recently

[18,58,69,70,78] due to the lack of a genomic-basedclassification of this group. Currently, rRNA genesequences are commonly used in microbial systemat-ics [79,80]. Bacterial genomes usually contain 10–15

logical

214 S.G. Bavykin et al. / Chemico-Bio

copies of rRNA genes, which are involved in translation,a fundamental cell function, and which have coevolvedwith ribosomal proteins. These features would seem tomake rRNA genes the least likely of all genes to expe-rience interspecies HGT [81]. In addition, GenBank(http://www.ncbi.nlm.nih.gov/Genbank/index.html)and Ribosomal Database Project II, release 9 (RDPII, r. 9) (http://rdp.cme.msu.edu/index.jsp) currentlycontain more than 300,000 bacterial rRNA sequencesincluding several hundred belonging to B. cereus grouporganisms. However, it has been considered impossibleto use rRNA gene sequences for B. cereus groupclassification and discrimination due to the high levelof rRNA homology (more than 95%) among B. cereusgroup organisms [78,82–84].

Recently, we reported that the B. cereus group maybe divided into seven subgroups (Anthracis, Cereus Aand B, Thuringiensis A and B, and Mycoides A andB) based on 16S rRNA, 23S rRNA and gyrB genesequences and each subgroup may be characterizedwith specific “signatures” (genomic markers) [85]. Wealso demonstrated that the currently defined species B.cereus, B. thuringiensis and B. mycoides actually rep-resent mixtures of genetically distinct bacterial groups[85]. Here we have utilized rRNA markers in a 3D gelelement oligonucleotide microarray to demonstrate forthe first time that B. anthracis may be unambiguouslydiscriminated from the most closely related organisms(subgroups Cereus A and B), as well as from bacte-ria comprising other, more divergent subgroups of theB. cereus group. This rRNA-based microarray was alsocapable of identification and discrimination of all sevensubgroups within the B. cereus group.

2. Materials and methods

2.1. Bacterial strains

Seven reference strains belonging to the B. cereusgroup, B. anthracis str. Ames, B. anthracis str. Sterne, B.thuringiensis str. B8, B. cereus str. NCTC9620, B. cereusstr. T, B. thuringiensis str. 4Q281, and B. mycoides str.ATCC10206 were obtained as a generous gift from Dr.John Ezzell, U.S. Army Medical Research Institute ofInfection Diseases, Fort Detrick, Frederick, MD, USA.Four other reference organisms, B. cereus NCTC11143,B. cereus ATCC43881, B. mycoides ATCC6462 and B.subtilis B-459, were obtained from the National Collec-

tion of Type Cultures and Pathogenic Fungi (NCTC),London, UK, and American Type Culture Collection(ATCC), Rockville, MD, respectively. Three other non-reference organisms, B. cereus NCTC3329, B. cereus

Interactions 171 (2008) 212–235

HER1414 and B. mycoides ATCC23258 were alsoobtained from NCTC and ATCC. B. thuringiensis B8,related to subgroup Cereus A, apparently represents amisclassification, as it does not produce protein crys-tals or contain any cry genes [85] that are a fundamentalcharacteristics of B. thuringiensis.

2.2. Oligonucleotide microarray design andconstruction

DNA microarrays were constructed with 11 pairsof oligonucleotide probes (ps1 through ps20, ps27 andps28) (Table 1) targeting 16S rRNA sequence “sig-natures” (Fig. 1A) and four pairs of oligonucleotideprobes (ps21 through ps24 and ps29 through ps32)(Table 2) targeting 23S rRNA sequence “signatures”(Fig. 1B) using 3D gel element microarray platform[86]. Probes ps25 (BCG1) and ps26 (BSG1) targetingthe B. cereus group and the B. subtilis group, respec-tively, were selected previously [87]. Each probe was20 nucleotides in length. Oligonucleotides were synthe-sized on an automatic DNA/RNA synthesizer (AppliedBiosystems 394) using standard phosphoramide chem-istry. 5′-Amino-Modifier C6 (Glen Research, Sterling,VA) was linked to the 5′-end of the oligonucleotides.A micromatrix containing 100 �m × 100 �m × 20 �mpolyacrylamide gel pads fixed on a glass slide andspaced 100 �m from each other was manufactured byphotopolymerization [88], and activated as describedearlier [89]. Six nanoliters of individual 1 mM amino-oligonucleotide solutions was applied to each gel padcontaining aldehyde groups [88,90] according to the pro-cedure described earlier [89].

2.3. Sample preparation and hybridization withmicroarray

Total RNA was isolated from frozen cell pel-lets by bead beating followed by phenol–chloroformextraction as previously described [91]. Isolated RNAwas fragmented and labeled with direct magnesium-sodium periodate method and excess of the Lissaminerhodamine B ethylenediamine (LissRhod) (MolecularProbes, Eugene, OR) was removed with butanol treat-ment as described earlier [92].

Samples of labeled and fragmented nucleic acids fromwhole cells were obtained as described earlier [87,92]with some modifications. Fifty microliters of silica sus-

pension containing 40 �l of dry silica was applied to a25-mm-long sterile disposable centrifuge filter unit con-taining an Anopore membrane filter (aluminum dioxide)with a diameter of 6.5 mm and a pore size of 0.2 �m

S.G. Bavykin et al. / Chemico-Biological Interactions 171 (2008) 212–235 215

Fig. 1. Positions of subgroup-specific sequence differences in the 16S rRNA (A) and 23S rRNA (B) genes of seven B. cereus subgroups andcorresponding sequences of reference microorganisms. The 16S rRNA and 23S rRNA sequences of B. anthracis Sterne (GenBank accession nos.A n usedt consens

((d(wHtwlm(

F176321 and AF267877, respectively) in 5′ to 3′ orientation have beeo the consensus sequence. Numbers indicate distance from 5′-end ofequences in (A). R = G or A; Y = T or C.

VectaSpin Micro Centrifuge Filter, catalog # 6830-021)Whatman, Fairfield, NJ) and centrifuged in an Eppen-orf 5415C micro-centrifuge for 30 s at 14,000 rpm∼14,000 × g). Silica in the column was washed onceith 500 �l of diethylpyrocarbonate-treated (DEPC)2O (pipetting of silica was followed by centrifuga-

ion at the same conditions). About (2–5) × 108 cells

ere pretreated in 25 �l of 100 mg/ml freshly diluted

ysozyme at 37 ◦C for 5 min, diluted with 550 �l of aixture (9:4) of L buffer (4.5 M guanidine thyocianate

GuSCN) and 10 mM EDTA pH 8.0) and B buffer

as the consensus sequences. Vertical lines denote nucleotides identicalsus rRNA sequence. Principle of pair probe’s design is shown below

without MgCl2 (4 M GuSCN, 135 mM Tris–HCl pH6.4, 3.5% (w/v) Triton X-100 and 17.5 mM EDTA)(9), suspended carefully to lyse the cells by pipetting,amended with 7 �l 5 M MgCl2, and applied to the sil-ica mini-column (in general pre-treatment is necessaryfor gram-positive cells only). All manipulations withsilica mini-column, except the labeling-fragmentation

step, were carried out at room temperature. The columnwas washed with 500 �l of L–B buffer (4.34 M GuSCN,46 mM Tris–HCl pH 6.4, 65 mM EDTA, 1% (w/v) TritonX-100 (twice), 500 �l of 75% ethanol (twice) and 500 �l

216 S.G. Bavykin et al. / Chemico-BiologicalTa

ble

1D

egre

eof

mat

chin

gbe

twee

nol

igon

ucle

otid

epr

obes

onm

icro

arra

yan

dth

e16

SrR

NA

sequ

ence

sof

refe

renc

em

icro

orga

nism

sfr

omth

eB

.cer

eus

grou

p

Prob

e’s

targ

et

allo

ther

Thu

rBA

nthr

;Cer

A,B

Thu

rA,B

allo

ther

Myc

BA

nthr

;Cer

A;M

ycB

allo

ther

allo

ther

Myc

A

(ps1

/ps2

)a

(77)

b(p

s3/p

s4)a

(90,

92)b

(ps5

/ps6

)a

(90,

92)b

(ps7

/ps8

)a

(192

)9(p

s9/p

s10)

a

(189

,200

)b(p

s11/

ps12

)a

(189

,200

,208

)b(p

s13/

ps14

)a

(200

,208

)b(p

s15/

ps16

)a

(208

)b(p

s19/

ps20

)a

(103

6,10

45)b

(ps1

7/ps

18)a

(101

5)b

(ps2

7/ps

28)a

(133

)b

B.a

nthr

acis

Am

es(A

nthr

)+

/−+

/−2

+/−

2+

/−+

/−2

+/−

3+

/−2

+/−

+/−

2+

/−+

/−B

.ant

hrac

isSt

erne

(Ant

hr)

+/−

+/−

2+

/−2

+/−

+/−

2+

/−3

+/−

2+

/−+

/−2

+/−

+/−

B.c

ereu

sN

CT

C11

143

(Cer

A)

+/−

+/−

2+

/−2

+/−

+/−

2+

/−3

+/−

2+

/−+

/−2

+/−

+/−

B.t

huri

ngie

nsis

B8

(Cer

A)

+/−

+/−

2+

/−2

+/−

+/−

2+

/−3

+/−

2+

/−+

/−2

+/−

+/−

B.c

ereu

sT

(Cer

B)

+/−

+/−

2+

/−2

+/−

+/−

2+

/−3

+/−

2+

/−+

/−2

−/+

+/−

B.c

ereu

sN

CT

C96

20(C

erB

)+

/−+

/−2

+/−

2+

/−+

/−2

+/−

3+

/−2

+/−

+/−

2−/

++

/−B

.cer

eus

AT

CC

4388

1(T

hur

A)

+/−

+/−

2+

/−2

−/+

−/−3

−/−4

+/−

2+

/−+

/−2

−/+

+/−

B.t

huri

ngie

nsis

4Q28

1(T

hur

B)

−/+

−2/+

−2/+

−/+

−/−3

−/−4

+/−

2+

/−+

/−2

−/+

+/−

B.m

ycoi

des

AT

CC

6462

(Myc

A)

+/−

+/−

2+

/−2

−3/ −

2−3

/−5

−2/−

5−/

−3+

/−+

/−2

−2/−

−/+

B.m

ycoi

des

AT

CC

1020

6(M

ycB

)+

/−+

/−2

+/−

2−2

/−3

−2/+

−3/+

−2/+

−/+

−2/+

+/−

+/−

“+”,

deno

tes

perf

ectm

atch

ing;

“−”,

“−2”

,and

“−3”

,den

ote

one,

two

and

thre

em

ism

atch

es,r

espe

ctiv

ely;

Ant

hr,C

erA

,Cer

B,T

hur

A,T

hur

B,M

ycA

and

Myc

B,c

orre

spon

dto

subg

roup

sA

nthr

acis

,Cer

eus

A,C

ereu

sB

,T

huri

ngie

nsis

A,T

huri

ngie

nsis

B,M

ycoi

des

Aan

dM

ycoi

des

B;“

allo

ther

”,de

pict

s“a

llot

her

subg

roup

s”.

aPr

obe’

sna

me.

bPo

sitio

nsof

targ

eted

subg

roup

-spe

cific

diff

eren

ces

on16

SrR

NA

(Fig

.1A

).

Interactions 171 (2008) 212–235

of 100% ethanol (twice)). The silica column containingnucleic acids was sealed at the bottom with a cap from a1.5 ml micro-centrifuge tube and preheated in a sand bathat 95 ◦C for 3 min to remove traces of ethanol. To pre-pare labeling cocktail, 1220 �l 25 mM sodium phosphate(pH 7.0) was applied into an ampoule containing 1 mgof lissamine–rhodamine B ethylenediamine (LissRhod)(Molecular Probes, Inc., Eugene, OR). The dye wascarefully suspended and then centrifuged at 14,000 × gfor 30 s to precipitate insoluble components of the dyemixture. Seven hundred and thirty-two microliters ofsupernatant were mixed with 30 �l of 150 mM 1,10-phenanthroline-HCl (OP) (Fluka, Ronkonkoma, NY)and 30 �l of 15 mM CuSO4, and 132 �l of labeling cock-tail was preheated for 30 s at 95 ◦C. Eighteen microlitersof 33 mM H2O2 (Sigma–Aldrich, Milwaukee, WI, cat#:H1009) was added to the cocktail immediately beforeapplication of 150 �l of hot cocktail mixed with H2O2 tothe preheated mini-column. The mini-column was sealedat the top to prevent evaporation and was incubated in thesand bath for 10 min at 95 ◦C. The reaction was stoppedby removing the column from the sand bath and adding20 �l of freshly prepared stop solution (2.6 M sodiumacetate, pH 5.2 and 70 mM EDTA) and 540 �l of cold100% ethanol. After 5-min incubation at room temper-ature, nucleic acids were precipitated on the column bycentrifugation for 30 s at 14,000 × g. Excess fluorescentlabel was removed by successive washing of the columnwith 500 �l of 75% ethanol (twice) and 100% ethanol(once). For elution of labeled product, the silica in thecolumn was suspended with 45–60 �l of 10 mM sodiumcarbonate (pH 8.5), the column was sealed from the top,placed into a micro-centrifuge tube, incubated for 2 minat 95 ◦C and centrifuged for 30 s at 14,000 × g. Elutionwas performed twice. Yield of labeled nucleic acids wasabout 20–50 �g.

Thirty-five microliters of hybridization solution con-taining 1–3 �g fragmented and labeled RNA, 1 Mguanidine thiocyanate (GuSCN), 5 mM EDTA, and40 mM HEPES (pH 7.5) was passed through a 0.22 �mfilter to remove particulates, then heated at 95 ◦C for3 min and placed on ice. Thirty microliters of thehybridization solution was added to a hybridizationchamber (Grace Biolabs, Bend, OR), and the hybridiza-tion chamber was affixed to a microarray. The microarraywas allowed to hybridize for an appropriate time at roomtemperature in the dark. After hybridization, the cham-ber and hybridization solution were removed from the

microarray, and the microarray was washed twice for 10 seach with 100 �l washing buffer (0.9 M NaCl, 50 mMsodium phosphate (pH 7.0), 6 mM EDTA, and 1% Tween20).

logical

iaRtpcwfcCi

ftslatwssfptt(cu“ri1

peTmvi

A

x

3

3

se

S.G. Bavykin et al. / Chemico-Bio

After hybridization the microarrays were analyzedn washing buffer with a custom made wide-field-high-perture fluorescence microscope (GOI, St. Petersburg,ussia) equipped with a cooled CCD camera (Prince-

on Instruments, Acton, MA), a thermal table, and XYositioners [93,94]. Exposition time during the imageapturing was 0.5–2.0 s. Alternatively, microarrays wereashed for 2–3 s in H2O to remove washing buffer

rom the microarray surface and dried with dust-freeompressed gas. Dry microarrays were analyzed withCD camera equipped portable reader (Aurora Photon-

cs, Lake Barrington, IL) with image capturing time 10 s.All hybridization experiments were performed two to

our times. The signal intensity of each gel element onhe microarray was calculated with MicroChip Imageroftware (Aurora Photonics, Lake Barrington, IL) as fol-ows: the software set up an inner frame to define therea of the gel element and a larger outer frame coveringhe area immediately surrounding the gel element whichas defined as the background for that gel element. The

ignal intensity of the gel element was calculated byubtracting the signal intensity per pixel in the outerrame (i.e. the background) from the signal intensity perixel in the inner frame (i.e. the gel element) and mul-iplying the difference by the number of pixels withinhe inner frame. The average (a) and average deviationAd) for p/mm ratios were calculated between repli-ate hybridizations. Average deviation was calculatedsing equation Ad = (

∑|x − a|)/n, where “x” is value,a”, denotes average and “n” indicates the number ofeplicate experiments. In our experiments summarizedn Tables 3–5, Ad for p/mm signal ratios did not exceed0% of the average, unless specifically indicated.

For identification of single nucleotide polymor-hisms, signal ratios were calculated based onxperimental data and compared to theoretical values.he regression coefficient was calculated using the for-ula y = Ax + k, where x, y and k represent theoretical

alue, experimental value and error, respectively. Errors random deviation from linear rule. Based on data table,

was calculated as A =(

N∑i=1

xiyi

)/

(N∑

i=1

x2i

), where

i and yi are values and N is the size of data table.

. Results

.1. Design of subgroup-specific probes

Analysis of the 16S rRNA, 23S rRNA and gyrB geneequences revealed subgroup-specific “signatures” inach of these three genes. Each “signature” represented

Interactions 171 (2008) 212–235 217

a set of single- or double-nucleotide variations (differ-ences) in gene sequence and differentiated one subgroupfrom all others [85]. We also found that discriminationof subgroups within the B. cereus group using rRNAsequences would require single base mismatch discrim-ination (Fig. 1). The microarray contained pairs of probesincluding one, two or sometimes three subgroup-specificsingle- or double-nucleotide differences. For example,pair ps1/ps2 was selected for identification of 16S rRNAsequence difference A/G (position 77) that is unique forsubgroup Thuringiensis B (Fig. 1A). In this pair probeps2 targeted nucleotide G that is specific for Thuringien-sis B, and probe ps1 recognized nucleotide A that ispresent in all six other subgroups (Table 1 and Fig. 1A).Another pair, ps3/ps4, targeted two subgroup-specificvariations, T/C (position 90) and T/A (position 92).Probe ps4 identified 16S rRNA sequences containingC at position 90 and A at position 92, and probe ps3corresponded to T at position 90 and T at position 92(Table 1 and Fig. 1A). In a similar manner, pair ps11/ps12targeted three subgroup-specific differences, A/C (posi-tion 189), T/G (position 200) and G/C (position 208)(Table 1 and Fig. 1A). In addition, some redundancywas included in the probe set: pair ps5/ps6 targeted thesame two subgroup-specific differences as pair ps3/ps4,but the targeted sequence for ps5/ps6 was shifted severalnucleotides in comparison with pair ps3/ps4.

The microarray (Fig. 2) included three pairs of probesfor identification of bacteria from subgroup Thuringien-sis B (ps1/ps2, ps3/ps4 and ps5/ps6), and one pair,ps7/ps8, for identification of organisms from subgroupsThuringiensis A and B, Anthracis, and Cereus A andB. However, ps7/ps8 did not identify organisms fromsubgroups Mycoides A and B (Table 1). Among fivepairs for identification of Mycoides B subgroup bac-teria, pair (ps9/ps10) targeted variations A/C (position189) and T/G (position 200); pair ps13/ps14 identi-fied variations T/G (position 200) and G/C (position208); pair ps15/ps16 recognized variation G/C (posi-tion 208), and pair ps19/ps20 detected variations T/C(position 1036) and A/G (position 1045) (Table 1 andFig. 1A). Finally, pair ps17/ps18, targeting variationC/A at position 1015 discriminated subgroups Anthracis,Cereus A, and Mycoides B from subgroups Cereus Band Thuringiensis A and B (Table 1 and Fig. 1A). Thelast pair, ps27/ps28, recognized microorganisms of sub-group Mycoides A, discriminating them from organismsfrom all other subgroups, using A/G (position 133) as

a target (Table 1, Fig. 1A). The combined applicationof these probes provided a basis for distinguishing sub-group Anthracis from all other subgroups except CereusA.

218 S.G. Bavykin et al. / Chemico-Biological Interactions 171 (2008) 212–235

Fig. 2. Identification of B. cereus subgroup reference microorganisms with 16S rRNA oligonucleotide microarray. Bulk RNA from referencemicroorganisms was isolated, fluorescently labeled with LissRhod, and hybridized with a microarray bearing 20 b oligonucleotides. Positions ofthe probes and targeted subgroups (in rectangles) are shown in the upper right corner. Reference organisms and their subgroups (in parentheses) areindicated under each hybridization image captured with stationary microscope. For more details about probe design and probe abbreviations, seeTable 1.

S.G. Bavykin et al. / Chemico-Biological Interactions 171 (2008) 212–235 219

Table 2Degree of matching between oligonucleotide probes on microarray and the 23S rRNA sequences of reference microorganisms from the B. cereusgroup

Probe’s target

Anth;Thur A;Myc A,BCer A,Cer B

Anthr;Cer BCer A;Thur A

Myc BAnthr;Cer A,B;Thur B;Myc A

Thur B;Myc AAnthr;Cer A,B;Myc B

(ps21/ps22)a (1559)b (ps23/ps24)a

(insertion 1218–1219)b(ps29/ps30)a (346, 347)b (ps31/ps32)a (174)b

B. anthracis Ames/Sterne (Anthr) +/− +/− −2/+ −/+B. cereus NCTC11143 (Cer A) −/+ −/+ −2/+ −/+B. thuringiensis B8 (Cer A) −/+ −/+ −2/+ −/+B. cereus NCTC9620 (Cer B) −/+ +/− −2/+ −/+B. thuringiensis 4Q281 (Thur B) N/A N/A −2/+ +/−B. mycoides ATCC6462 (Myc A) +/− N/A −2/+ +/−B. mycoides ATCC10206 (Myc B) +/− N/A +/−2 −/+

“+”, denotes perfect matching; “−”, “−2”, and “−3”, denote one, two and three mismatches, respectively; Anthr, Cer A, Cer B, Thur A, Thur B,M reus B,

Fig. 1B

sisgdf(asfampsnpsaa

3r

fsaNACsB

yc A and Myc B, correspond to subgroups Anthracis, Cereus A, Cea Probe’s name.b Positions of targeted subgroup-specific differences on 23S rRNA (

Most bacteria of subgroup Cereus A revealed strain-pecific sequence variations [85], although some isolatesn this subgroup had 16S rRNA sequences identical toubgroup Anthracis (Fig. 1A). Organisms within sub-roup Cereus A do not have any subgroup-specificifferences in their 16S rRNA (Fig. 1A), but do dif-er from subgroup Anthracis in 23S rRNA sequenceFig. 1B). Therefore, we designed probe pairs ps21/ps22nd ps23/ps24 (Table 2) for the discrimination ofubgroups Anthracis and Cereus A using sequence dif-erences located, respectively, at position 1559 (G/A)nd insertion G at positions 1218–1219 in the 23S rRNAolecule (Table 2 and Fig. 1B). As an another exam-

le of using 23S rRNA subgroup-specific variations forubgroup identification, we also demonstrated discrimi-ation of subgroups Mycoides A and Mycoides B withrobe pairs ps29/ps30 and ps31/ps32 that recognize 23Subgroup-specific substitutions AG/GA at positions 346nd 347, and T/A at position 174, respectively (Table 2nd Fig. 1B).

.2. Identification strategy for subgroups andeference microorganisms

To demonstrate the ability of the microarray to dif-erentiate all seven subgroups of B. cereus group, weelected ten reference organisms, B. anthracis Amesnd B. anthracis Sterne (subgroup Anthracis), B. cereusCTC11143 and B. thuringiensis B8 (subgroup Cereus

), B. cereus T and B. cereus NCTC9620 (subgroupereus B), B. cereus ATCC43881 (subgroup Thuringien-

is A), B. thuringiensis 4Q281 (subgroup Thuringiensis), B. mycoides ATCC6462 (subgroup Mycoides A),

Thuringiensis A, Thuringiensis B, Mycoides A and Mycoides B.

).

and B. mycoides ATCC10206 (subgroup Mycoides B),whose subgroup affiliations were previously determinedin accordance with their 16S rRNA and 23S rRNA genesequences [85]. Our objectives were to determine if wecould discriminate organisms from these closely relatedsubgroups, to determine if we could achieve discrimina-tion of bacteria whose rRNA sequences differed by onlyone base, and to determine if we could discriminate B.anthracis from all other closely related organisms in theB. cereus group. For these purposes we fabricated 16SrRNA microarrays for unambiguous identification ofsubgroups Cereus B, Thuringiensis A, Thuringiensis B,Mycoides A, and Mycoides B. This microarray may alsoidentify subgroups Anthracis and Cereus A, and discrim-inate them from five above-mentioned subgroups, butmay not discriminate subgroups Anthracis and CereusA each from other (Fig. 2 and Table 1). Therefore,our 16S rRNA targeted microarray contained “subgroupMycoides B-specific” probe Set #1, “Thuringiensis B-specific” probe Set #3, “Thuringiensis A and B-specific”probe ps8, “Mycoides A-specific” probe ps28 and probepair ps17/ps18 that discriminates subgroups Cereus B,Mycoides A, Thuringiensis A and B from subgroupsAnthracis, Cereus A and Mycoides B, respectively (mapof microarray in Fig. 2).

3.2.1. Identification of subgroups and referencemicroorganisms with 16S rRNA gene sequences3.2.1.1. Identification of subgroup Mycoides B. The

16S rRNA of B. mycoides ATCC10206, related to sub-group Mycoides B [85] (Fig. 1A), forms perfect duplexeswith probes ps10, ps12, ps14, ps16 and ps20 (Set #1,“Mycoides B-specific”), with probes ps1, ps3 and ps5

220 S.G. Bavykin et al. / Chemico-Biological

Tabl

e3

Prob

esi

gnal

ratio

inse

lect

edpa

irs

for

Cer

eus

Aan

dT

huri

ngie

nsis

Bis

olat

esan

dth

eir

mix

ture

s

Prob

epa

irs

B.t

huri

ngie

nsis

B8

(Cer

eus

A)

B.t

huri

ngie

nsis

4Q28

1(T

huri

ngie

nsis

B)

B.t

huri

ngie

nsis

B8:

B.

thur

ingi

ensi

s4Q

281

(1:2

)B

.thu

ring

iens

isB

8:B

.th

urin

gien

sis

4Q28

1(1

:5)

Exp

erim

enta

lT

heor

etic

alE

xper

imen

tal/t

heor

etic

alE

xper

imen

tal

The

oret

ical

Exp

erim

enta

l/the

oret

ical

ps3/

ps4

4.24

50.

653

0.90

51.

216

0.74

40.

863

0.94

10.

917

ps5/

ps6

1.86

00.

182

0.42

60.

412

1.03

30.

306

0.29

81.

027

ps7/

ps8

2.81

00.

270

0.44

20.

470

0.94

10.

351

0.36

80.

954

ps17

/ps1

86.

697

0.25

60.

621

0.61

11.

016

0.41

40.

427

0.97

0

Interactions 171 (2008) 212–235

(Set #4), and with probes ps17 and ps27. At the sametime, this organism includes mismatches with probesps9, ps11, ps13, ps15 and ps19 (Set #2), ps2, ps4 andps6 (Set #3) and with probes ps18 and ps28 (Table 1,Figs. 1A and 2). For this reason, Mycoides B isolateshybridized with the microarray revealed fluorescent sig-nals from Set #1, Set #4, and probes 17 and 27 that werehigher than the signals from Set #2, Set #3, and probes18 and 28 (Fig. 2). Pair ps7/p8 recognized B. mycoidesATCC10206 as a “not Thuringiensis A, B” organismrevealing signal from ps7 stronger than ps8 one, butbecause both probes have one additional mismatch withthis target (Table 1), pair ps7/ps8 signals were too weakand were seen only with significant overexposure (notshown).

3.2.1.2. Identification of subgroup Thuringiensis B. B.thuringiensis 4Q281, affiliated to subgroup Thuringien-sis B [85], forms a perfect match with all probes from Set#3 (“Thuringiensis B-specific”), with probes ps13, ps15and ps19 from Set #2, and with probes ps8 (“Thuringien-sis A, B-specific”), ps18 and ps27 (Fig. 2). Each probein pairs ps9/ps10 and ps11/ps12 included one additionalmismatch with this isolate (Table 1), but also recognizedthis target as a “not Mycoides B” organism and, despitethe additional mismatch, produced pronounced signals(Fig. 2).

3.2.1.3. Identification of subgroup Thuringiensis A.The combination of probe pair ps7/ps8 and probes ps1through ps6 (Set #4 and “Thuringiensis B-specific”Set #3) (Table 1, Fig. 2) unambiguously identifiedsubgroup Thuringiensis A and discriminated it from sub-group Thuringiensis B. Hybridization of 16S rRNA ofB. cereus ATCC43881, which is affiliated with sub-group Thuringiensis A [85] revealed perfect matcheswith probe ps8 (specific to subgroups Thuringiensis Aand B), with probes ps1, ps3 and ps5 (Set #4, “notThuringiensis B”) and with probe ps27 (Fig. 2). There-fore, hybridization of 16S rRNA from bacteria of the B.cereus group with probes from region 77-92 (Fig. 1A,Table 1) was used to discriminate microorganisms ofsubgroup Thuringiensis B from bacteria from all othersubgroups, and it was especially important for the dis-crimination between subgroups Thuringiensis A and B.

3.2.1.4. Identification of subgroups Anthracis andCereus A. B. anthracis Ames (subgroup Anthracis)

and B. cereus NCTC11143 (subgroup Cereus A) [85](Fig. 1A) form perfect matches with probe Set #2 and Set#4, with probes ps7, ps17 and ps27 (Fig. 2 and Table 1).Probe ps17 is specific for subgroups Anthracis, Cereus

S.G. Bavykin et al. / Chemico-Biological Interactions 171 (2008) 212–235 221

Table 5Influence of environmental admixtures on probe signal ratio

Probe pairs K1 pure cellsa (p/mm) K2 environmental samples (p/mm) K1/K2

ps3/ps4 3.100 2.302 1.347ps5/ps6 2.719 1.960 1.387ps14/ps13 7.404 6.424 1.153ps17/ps18 3.462 1.925 1.798ps27/ps28 4.088 3.111 1.314p

perimeo

Araa(mca(AApm(

3tBf#pg

3m

TP

P

pppppp

s31/ps32 1.974

a Logarithmic B. mycoides ATCC10206 cells were used for these exf 3 �g labeled sample with microarray.

, and Mycoides B, forming perfect duplexes with 16SRNA from B. anthracis Ames, B. cereus NCTC11143nd B. mycoides ATCC10206 (subgroup Mycoides B),nd mismatches with all other reference microorganismsFig. 2 and Table 1). In contrast, probe ps18 contains aismatch (at position 1015) for B. anthracis Ames, B.

ereus NCTC11143 and B. mycoides ATCC10206, and isperfect match with all other references microorganisms

Table 1 and Fig. 1A). Discrimination of B. anthracismes and B. cereus NCTC11143 from B. mycoidesTCC10206 was based on a “perfect” signal for probes17 (compare with ps18) in combination with “mis-atch” signals for probe Set #1 (“Mycoides B-specific”)

Fig. 2 and Table 1).

.2.1.5. Identification of subgroup Cereus B. Identifica-ion of B. cereus T, which is related to subgroup Cereus

[85], was established based on perfect match signalsor probes ps7, ps18, ps27, and for probe Sets #2 and4 (Fig. 2 and Table 1). Specifically, pair ps17/ps18rovided differentiation of subgroup Cereus B from sub-

roups Anthracis and Cereus A (Fig. 2 and Table 1).

.2.1.6. Identification of subgroup Mycoides A. B.ycoides ATCC6462 forms perfect duplexes with probes

able 4robe signal ratio for B. mycoides ATCC10206 sample hybridization with the

robe pairs Degree of probe matching Hybridization

5 min

2.5 �ga

s3/ps4 +/−2 1.53s5/ps6 +/−2 1.37s14/ps13 +/−2 2.80s17/ps18 +/− 1.52s27/ps28 +/− 1.80s31/ps32 +/− 1.49

+”, denotes perfect matching; “−”, or “−2”, denote one or two mismatches.a Amount of labeled sample.

1.862 1.060

nts. Images were captured with portable reader after 1 h hybridization

ps28 (“Mycoides A-specific”), ps1, ps3, ps5 (Set #4),ps15 and ps19 (Fig. 2 and Table 1). Each probe in pairsps7/ps8, ps9/ps10, ps11/ps12, ps13/ps14 and ps17/ps18forms additional mismatches with this isolate (Table 1)and for this reason revealed weak signals or no signalsafter hybridization (Fig. 2).

3.2.2. Identification of subgroups and referencemicroorganisms with 23S rRNA gene sequences

Although subgroups Anthracis and Cereus A couldnot be differentiated with 16S rRNA sequences (Fig. 2),utilization of subgroup-specific variations in 23S rRNAgenes helped to resolve this problem (Fig. 3A andTable 2). Here we also demonstrated differentiation ofisolates from subgroups Mycoides A and Mycoides Bas an example of the usefulness of 23S rRNA variationsfor identification of different subgroups (Fig. 3B andTable 2).

3.2.2.1. Differentiation of subgroup Anthracis from sub-group Cereus A. Organisms that belong to subgroup

Cereus A contain 16S rRNA sequences that are identi-cal to subgroup Anthracis or that differ by strain-specificsequence variation only [85] (Fig. 1A). Thus, we used23S rRNA sequences to differentiate subgroups Cereus

microarray

time

20 min 18 h

5 �ga 2.5 �ga 5 �ga 1 �ga

2.94 2.77 4.62 13.401.82 1.61 1.96 3.983.66 3.95 5.07 15.402.06 1.97 2.51 10.002.16 2.27 2.68 4.091.70 1.43 1.62 2.54

Images were captured with stationary microscope.

222 S.G. Bavykin et al. / Chemico-Biological Interactions 171 (2008) 212–235

Fig. 3. Discrimination of subgroups Anthracis and Cereus A (A), or Mycoides A and Mycoides B (B) isolates with 23S rRNA probes. Image capturewas carried out with stationary microscope for B. anthracis Sterne, B. thuringiensis B8 and B. cereus HER1414, and with portable reader for all

of the i

other isolates (A and B). Probe signal ratio is shown on the right sidesee Table 2.

A and Anthracis. The 23S rRNA sequences of B.thuringiensis B8 and B. cereus NCTC11143 differ from

B. anthracis Ames at three sites, Y/C (594), insertion G(1218–1219) and G/A (1559) (Fig. 1B). We designed twopairs of probes, ps21/ps22 and ps23/ps24 to target site1559 and insertion 1218–1219, respectively. Probes ps21

mages. For more details about probe design and probe abbreviations,

and ps23 formed a perfect duplex with the 23S rRNAof subgroup Anthracis but not subgroup Cereus A 23S

rRNA (Table 2 and Fig. 1B). Probes ps22 and ps24 pro-vided complementary information; they have a mismatchwith subgroup Anthracis 23S rRNA and are complemen-tary with Cereus A 23S rRNA (Table 2 and Fig. 1B).

logical

rfAaBsantpgBdHapraiBr1C(hByA

3ammfapytp

3p

s–bgtt

S.G. Bavykin et al. / Chemico-Bio

To validate the discriminative feasibility of 23SRNA probes ps21/ps22 and ps23/ps24, we used themor testing of reference microorganisms from subgroupnthracis (B. anthracis Ames and B. anthracis Sterne)

nd subgroup Cereus A (B. cereus NCTC11143 and. thuringiensis B8), as well as for determination ofubgroup affiliation for organisms B. cereus HER1414nd B. cereus NCTC3329, whose rRNA genes wereot sequenced. We analyzed hybridization signals fromhese two probe pairs only when 16S rRNA targetedrobes revealed a pattern typical for organisms of sub-roups Anthracis and Cereus A. B. anthracis Ames and. cereus NCTC11143 revealed this pattern (Fig. 2), asid B. anthracis Sterne, B. thuringiensis B8, B. cereusER1414 and B. cereus NCTC3329 (Fig. 3A, ps18/ps17

re shown only). 23S rRNA probe pairs ps21/ps22 ands23/ps24 successfully discriminated reference isolateselated to subgroup Cereus A (B. cereus NCTC11143nd B. thuringiensis B8) from reference microorgan-sms of subgroup Anthracis (B. anthracis Ames and. anthracis Sterne), demonstrating hybridization signal

atios for these two pairs 0.9, 0.4 (ps21/ps22, Cereus A),.8, 2.2 (ps21/ps22, Anthracis), and 0.4, 0.4 (ps23/ps24,ereus A), 1.7, 1.7 (ps23/ps24, Anthracis), respectively

Fig. 3A). The same two pairs, ps21/ps22 and ps23/ps24,elped to identify isolates B. cereus HER1414 and. cereus NCTC3329, whose rRNA genes have notet been sequenced, as belonging to subgroup Cereus(Fig. 3A).

.2.2.2. Differentiation of subgroups Mycoides And Mycoides B. Two reference microorganisms B.ycoides ATCC6462 (subgroup Mycoides A) and B.ycoides ATCC10206 (subgroup Mycoides B) were dif-

erentiated effectively with 16S rRNA probes (Fig. 2),s was B. mycoides ATCC23258 (Fig. 3B, probe pairs27/ps28 is shown only), whose rRNA genes have notet been sequenced. Here we demonstrated that thesehree isolates may be also differentiated with 23S rRNArobes (Fig. 3B, Table 2).

.3. Identification of single nucleotideolymorphism sites

Single nucleotide polymorphism or SNP is a DNAequence variation occurring when a single nucleotideA, T, C, or G – in the genome differs between mem-

ers of species or between different copies of multicopyenes. We hypothesized that SNP sites could be iden-ified in multicopied rRNA genes with our microarrayechnology.

Interactions 171 (2008) 212–235 223

3.3.1. SNP in the sites of subgroup-specificdifferences in the isolates with sequenced rRNA

Complete sequencing of the genome of several iso-lates of B. anthracis and B. cereus indicated thatorganisms from the B. cereus group may contain at least10 copies of the rRNA operon [78]. Comparative anal-ysis revealed that a SNP is rather rare event in rRNAsequences in general, but, nevertheless, may be foundeven at some sites of subgroup-specific differences,for example at positions 594 and 1559 of 23S rRNA(Fig. 1B) in B. anthracis AMES, B. cereus ATCC10987and B. cereus ATCC14579 [85]. Our recent sequencing[85] indicated that B. cereus T, B. thuringiensis 4Q281and B. cereus NCTC9620 also have a SNP in their 23SrRNA genes at site 1559 with G:A ratios equal to 1.5:1,1:1 and 1:3.5, respectively (Fig. 4). At the same time,B. anthracis Ames and B. mycoides ATCC10206 haveG at this site, and B. thuringiensis B8 has A in thisposition [85]. We demonstrated the possibility of rec-ognizing polymorphisms in this site by using a pair ofprobes, ps21/ps22 (Table 2). For these probes the perfect-mismatch (p/mm) ratios for B. cereus T, B. thuringiensis4Q281 and B. cereus NCTC9620 were intermediate (1.0,0.72 and 0.64, respectively) between B. anthracis Ames,or B. mycoides 10206 (1.8 and 1.5, respectively) and B.thuringiensis B8 (0.45) (Fig. 4).

3.3.2. Identification and quantification of“polymorphic” sites in mixtures

To demonstrate the ability of our microarray toidentify SNP sites in individual bacterial strains we per-formed hybridizations of mixtures of rRNA samplesisolated from pure cultures and compared them withhybridization images of homogeneous rRNA samplesfrom the same isolates. For these purposes we used the16S rRNA of B. thuringiensis B8 (subgroup Cereus A)and B. thuringiensis 4Q281 (subgroup Thuringiensis B)and a slightly modified 16S rRNA microarray (Fig. 5).B. thuringiensis B8 has a 16S rRNA sequence that iscompletely identical to B. anthracis and differs fromB. thuringiensis 4Q281 by five subgroup-specific dif-ferences (Fig. 1A). Both organisms shared hybridizationsignals with probe Sets #1 and #2, but generated differ-ent signals for probe pairs ps3/ps4, ps5/ps6, ps7/ps8 andps17/ps18. B. thuringiensis B8 forms perfect duplexeswith probes ps3, ps5, ps7 and ps17, but B. thuringiensis4Q281 forms perfect duplxes with probes ps4, ps6, ps8and ps18 (Fig. 5, Tables 1 and 3). Therefore, we selected

probe pairs ps3/ps4, ps5/ps6, ps7/ps8 and ps17/ps18as reference pairs for testing of the microarray for itsability to differentiate quantitatively these two organ-isms in mixtures to imitate polymorphism estimation

224 S.G. Bavykin et al. / Chemico-Biological Interactions 171 (2008) 212–235

roorgancapture

Fig. 4. Identification of single-base polymorphisms in reference miccereus group bacteria to probes targeting the 23S rRNA. Images were

in sites 90, 92, 192, 1015 of 16S rRNA (Fig. 1A).We also performed this experiment with mixtures of B.anthracis AMES/B. cereus NCTC9620 and B. anthracisAMES/B. thuringiensis 4Q281 (not shown). Probe pairsignal ratios for mixtures were calculated using actualsignal intensities from microarray hybridizations (exper-imental ratios) and using predicted intensities based ondata from homogeneous sample hybridizations and theproportions of individual isolates in the mixtures (theo-retical ratios) (Table 3). Results for all above mentionedmixtures revealed good correlation between experimen-tal and theoretical data. For probe pairs with stronghybridization signals (ps5/ps6, ps7/ps8, ps17/ps18) theregression coefficient for experimental and theoreticaldata was 0.96, and the differences between experimen-tal and theoretical ratios did not exceed 10% of theexperimental ratio (Table 3). However, this differencemay be more pronounced, for pairs like ps3/ps4, whereboth probes produce signals of low intensity (Fig. 5F,Table 3).

3.4. Influence of hybridization conditions onmismatch recognition

The time required for hybridization and the amountof labeled sample are important parameters in microbial

diagnostics. We estimated changes in perfect-mismatchratios of hybridization signals for selected probe pairswith the amount of B. mycoides ATCC10206 labeledRNA varying from 1 to 5 �g and the hybridization

isms using hybridization of fluorescently labeled total RNA from B.d with stationary microscope. R = G or A.

time varying from 5 min to 18 h (Table 4). Images werecaptured with stationary microscope. For this experi-ments we selected both 16S and 23S rRNA probe pairscontaining one (ps17/ps18, ps27/ps28 and ps31/ps32)or two mismatches (ps3/ps4, ps5/ps6 and ps14/ps13)(Tables 1, 2 and 4). Results indicated that p/m ratiois significantly decreased with decreased hybridiza-tion time and amount of labeled sample (Table 4).The best recognition (i.e. highest p/m ratios) wasobtained after 18 h microarray hybridization with 1 �gof labeled sample. However, all six selected pairsdemonstrated correct perfect-mismatch recognitions andtherefore revealed correct bacteria recognition even after5 min of hybridization at room temperature and withlabeled sample value as small as 2.5 �g. The p/mmratio was minimal (1.37) for pair ps5/ps6 (Table 4),but nonetheless provided correct recognition assum-ing that average deviation in this experiment did notexceed 10% of p/mm ratio values between differentmicroarrays.

3.5. Influence of environmental admixtures on p/mmsignal ratio

We tested the influence of environmental con-taminants on microarray hybridization. To maximize

inhibition activity, we did not use pre-filtration to removeorganic debris present in the collected samples. Instead,10 ml of ground water samples collected from the bot-tom of wells in Waterfall Forest Preserve (Darien, IL)

S.G. Bavykin et al. / Chemico-Biological Interactions 171 (2008) 212–235 225

Fig. 5. Identification of “polymorphic” sites imitated in mixtures. Samples of B. thuringiensis B8 and B. thuringiensis 4Q281 labeled total RNAwere hybridized with microarray separately (A and B) or mixed in proportions 1:2 (C), or 1:5 (D). Positions of the probes and targeted subgroups (inrectangles) are shown on scheme (E). Images were captured with stationary microscope. For probe signal ratio in pairs selected for SNP quantization,see Table 3. Correlation between actual and predicted data is indicated on chart (F). Values of theoretical (x) vs. experimental (y) signal ratios forfour selected pairs obtained for two different mixtures, 1:2 and 1:5 (Table 3) determine each data point (blue rhombuses) position on the chart.Approximation line is described by equation y = Ax + k, where A and k are regression coefficient and error, respectively. One of the data point obtainedfor pair ps3/ps4 (rhombus in upper right corner) was not representative due to low hybridization signals. Calculations performed without these data( The erra able 3),

wcwhvTmsi3mpc1r

ps3/ps4, for mixture 1:2) revealed volumes of A to be equal to 0.96.nd theoretical ratios did not exceed 10% of the experimental ratio (T

ere mixed with 3.52 × 108 B. mycoides ATCC10206ells and precipitated by centrifugation. The pellets,hich included microbial cells together with debris,ad an average volume of 100 �l which exceeded theolume of pure bacterial cells approximately 20-fold.hese “environmental” samples and control pure B.ycoides ATCC10206 cells were used for microarray

ample preparation in accordance with our standard Sil-ca mini-column labeling-fragmentation procedure and�g of labeled samples were hybridized for 1 h with theicroarray. Hybridization images were captured with

ortable reader. Results indicated that environmentalontamination decreased p/mm pair ratios only slightly,.3-fold on average (Table 5), and did not affect bacterialecognition.

or values for all other data, where differences between experimentalwere 0.04 or less.

3.6. Cross-hybridization of selected probes withnon-target bacteria

Subgroup-specific differences in rRNA sequences forbacteria from the B. cereus group are few and they arelocalized. This creates difficulties in the selection ofprobes specific to individual subgroups of bacteria. Forexample, the sequences of 2 of the 20 probes on themicroarray (ps17 and ps20) (Table 1) selected for iden-tification of B. cereus group reference microorganismsalso match the 16S rRNA sequences of a number of bac-

teria that belong to other groups of the genus Bacillus.The use of probes targeting the entire B. cereus groupand for some other groups may resolve this problem. Todemonstrate this we used probes ps25 and ps26, respec-

226 S.G. Bavykin et al. / Chemico-Biological

Fig. 6. Identification of microbial groups. Microarray containingprobes ps25 and ps26 targeting 16S rRNA of the B. cereus groupand the B. subtilis group, respectively, was hybridized with labeled

bulk RNA of the reference microorganisms. Probe signal ratios areshown in the far right column. Images were captured with stationarymicroscope.

tively. Probe ps25 matches all known sequences of B.cereus microorganisms and probe ps26 targets the B.subtilis group (B. subtilis, B. amiloliquifaciens, B. lenti-morbus, B. popilliae, and B. atrophaeus). Hybridizationanalysis indicated that these probes effectively differ-entiated B. subtilis B-459 from B. anthracis AMES, B.cereus T, B. mycoides ATCC10206, and B. thuringiensis4Q281 (Fig. 6). Thus, the B. cereus group probes canbe used as an internal check to validate probes ps17 andps20.

4. Discussion

4.1. Molecular typing of B. cereus group as a toolfor identification and discrimination of B. anthracisand related microorganisms

A wide spectrum of DNA typing systems devel-oped recently have not been successful in discriminatingbetween B. cereus and B. anthracis, or B. cereus andB. thuringiensis, or B. cereus, B. weihenstephanen-

sis and B. mycoides [18,58,69,70,78]. These systemsincluded 16S and 23S rRNA genes, 16S–23S rRNAITS and gyrB gene analysis [43,49,55,58,61,95–101],MEE [36,70,102–104], MLST [78,104–109] and AFLP

Interactions 171 (2008) 212–235

[18,62,67,69,97] among others. These findings resultedin several suggestions to consider B. cereus and B.thuringiensis [102], or these two bacteria together withB. anthracis [36,58,67,70], as one species.

Our recent findings [85] indicated that the B. cereusgroup could be divided into several genomic groupsbased on 16S and 23S rRNA sequences, and that genomicgroup Anthracis, which unifies all strains of B. anthracis,could be unambiguously discriminated from other mem-bers of the B. cereus group based on these sequences.This grouping was confirmed by analysis of gyrB genesequences and indicated that the genomogroups definedby rRNA and gyrB sequences are inconsistent with theseven currently defined species [85]. We found thatthree of the largest and most widely distributed B.cereus group species (B. cereus, B. thuringiensis and B.mycoides), which had been defined based on morpho-physiological criteria [9], actually represent mixtures ofmicroorganisms related to several genetically distinctbacterial genomogroups [85]. We called such groups“subgroups” and at present time have identified seven ofthem: Anthracis, Cereus A and B, Thuringiensis A and B,and Mycoides A and B [85]. We also performed molec-ular typing and identified subgroup-specific genomicmarkers in 16S rRNA, 23S rRNA and gyrB sequences[85]. Our grouping results confirm AFLP-based find-ings that B. cereus and B. thuringiensis species reveal ahigh degree of polymorphism, suggesting that these twospecies may be polyphyletic [69,97]. Our findings as wellas many others mentioned above, demonstrated that cur-rent taxonomy, which divides the B. cereus group intoseven species (B. anthracis, B. cereus, B. thuringiensis,B. mycoides, B. pseudomycoides, B. weihenstephanensisand B. medusa [1–6]) needs serious corrections.

The B. cereus group was divided into seven subgroupsbased on sequences of three different evolutionarily con-served genes (16S rRNA, 23S rRNA, and gyrB) whichwere analyzed by two different methods: identificationof “subgroup-specific signatures” and phylogenetic treeconstruction. Phylogenetic analysis was performed foreach of these three genes separately with four differ-ent methods of phylogenetic tree construction: minimumevolution, neighbor joining, UPGMA, and maximumparsimony [85]. Therefore, we demonstrated the exis-tence of these seven subgroups by several genomictests, which independently confirmed the presence ofseven new genomic taxons within the B. cereus group.Several studies by other laboratories have produced phy-

logenetic trees for B. cereus group organisms usinga variety of target genes and phylogenetic tools. Forexample, B. cereus group organisms have been com-pared based on 16S rRNA sequences [97], 16S–23S

S.G. Bavykin et al. / Chemico-Biological

Table 6Subgroup affiliationa of type strains and reference microorganisms thatare common in the cited papers

Subgroups Strains

Anthracis All B. anthracis strainsAnthracis–Cereus A B. cereus G9241b

Cereus A B. cereus ATCC10987c; B. cereusWSBC10030; B. thuringiensis97-27b; B. cereus E33Lb; B. cereusNCTC11143c; B. cereus IMSNU12078

Cereus B B. cereus ATCC14579c (DSM31T)d;B. cereus ATCC11778c (ATCC9634;DMS345); B. thuringiensis HD1e

Thuringiensis A B. thuringiensis ATCC33679f; B.cereus ACTC43881

Thuringiensis B B. thuringiensis ATCC10792c

(IMSNU 1208; DSM2046T); B.cereus AH527; B. thuringiensisKCTC 1509c

Mycoides A All B. weihenstephanensis strains; B.cereus AH521c; B. cereusWSBC10204T; B. mycoidesATCC6462c (DSM2048T)

Mycoides B All B. pseudomycoides strains

a For more detailed information about these strains, see Refs.[85,111].

b Affiliation of these isolates to correspondent subgroups wereobtained from analysis of rRNA and gyrB gene sequences extractedfrom whole genome sequences of these bacteria (AAEK00000000;AE017355; NC 006274). B. cereus G9241 represents intermediateform between subgroups Anthracis and Cereus A (S. Bavykin, personalcommunication).

c Sequences of these isolates perfectly match with sequences of theirsubgroups.

d Synonyms of strain title are shown in parenthesis.e Relation of the isolate was established in accordance with its gyrB

gene sequence only, GeneBank accession number AF136390.f Isolate contains substitution C/Y (192) in its 16S rRNA sequence

tCt

rIt[suoi(s[bsb

hat makes it to be intermediate form between Thuringiensis A andereus B [85]. Finally this dilema may be solved after sequencing of

heir 23S rRNA and gyrB genes.

RNA ITS sequences [49,98], and gyrB sequences [61].n addition, AFLP-based phylogenetic tree construc-ion [18], MLST based on six chromosomal genes106], seven housekeeping gene fragments [107], andeveral other targets [17,78,99,104,105,109] have beensed to compare B. cereus group organisms. Whenur results were compared to these studies by look-ng at specific strains that were analyzed in commonsee Table 6), we found good agreement between oureven subgroups and clades found in all of these studies

17,18,61,78,97–99,104–110]. For example, an MLST-ased phylogenetic tree constructed for the concatenatedequences of six chromosomal genes produced four mainranches bearing clades: A; B-II, III, IV; C-I, II and D-

Interactions 171 (2008) 212–235 227

I, II [106], which contained isolates found in four ofour subgroups Anthracis (clade A), Cereus A (cladesB-II, B-III and B-IV), Cereus B (clades C-I and C-II),and Thuringiensis B (clades D-I and D-II), respectively.Similarly, an MLST-based tree generated using concate-nated sequences of seven housekeeping gene fragments[107] produced clades “Anthracis”, “Cereus I”, “Tolwor-thi”, “Kurstaki” and “Others”, which included organismsfrom our subgroups Anthracis, Cereus A, Cereus B,Thuringiensis B and Mycoides A, respectively. In thesame manner, phylogenetic grouping based on distanceanalysis of 16S–23S rRNA ITS sequences [49] revealedgroups I, V, VI and VII, which correspond to our sub-groups Anthracis/Cereus A, Thuringiensis B, MycoidesA and Mycoides B, respectively, as well as a phyloge-netic tree branch bearing groups II, III and IV whichcontain isolates related to subgroup Cereus B.

Our seven subgroups also correspond to some phe-notypic properties of B. cereus group organisms. Forexample, organisms from different subgroups differ intoxin profiles: B. cereus BCRC 17039 (synonym of B.cereus NCTC11143, affiliated to subgroup Cereus A)showed toxin pattern VII; B. cereus BCRC 10603 (syn-onym of B. cereus ATCC14579, subgroup Cereus B)demonstrated toxin pattern I; and B. mycoides BCRC10604 (synonym of B. mycoides ATCC6462, subgroupMycoides A) demonstrated toxin pattern IV [112].Similarly, several lines of evidence suggest that signif-icant phenotypic differences exist between subgroupsMycoides A and Mycoides B. First, von Stetten etal. [100] developed PCR primers that could discrim-inate psychrotolerant and mesophilic strains of the B.cereus group based on 16S rDNA sequences. Compari-son of their primer sequences with the subgroup specificmutations identified in our previous study [85] revealedthat the psychrotolerant primers matched sequencesfrom subgroup Mycoides A and the mesophilic primersmatched sequences from subgroup Mycoides B. Wealso found that the last subgroup, Mycoides B, includedB. pseudomicoides [85], which recently was describedas a new species [5]. B. pseudomycoides revealedmesophilic phenotype and genetically was rather farfrom psychrotolerant strains of B. mycoides and B. wei-henstephanensis [5]. In addition, Hendriksen et al. [43]found distinct toxin profiles in strains identified by PCRas psychrotolerant (90 isolates of B. weihenstephanensisand 86 strains of B. mycoides-like B. weihenstephanen-sis) and strains identified by PCR as mesophilic (10

strains of B. mycoides). These data indicate that sub-groups Mycoides A and Mycoides B differ in toxinprofiles, and support our previous suggestion that psy-chrotolerance is an essential and unique feature of

logical

228 S.G. Bavykin et al. / Chemico-Bio

subgroup Mycoides A, as well as that B. weihenstepha-nensis represents one of the B. mycoides strains affiliatedto subgroup Mycoides A [85]. The B. cereus subgroupsmay also differ in the presence of certain types of plas-mids. For example, extensive analysis of B. cereus groupisolates with sequenced rRNA genes indicated that plas-mids harboring cry genes have never been found insubgroups Anthracis, Mycoides A and Mycoides B [85],but may be present in all of the other four subgroups(S. Bavykin, unpublished). The fact that our seven sub-groups, which are based on rRNA and gyrB sequences,correlate well with groupings based on AFLP, MME andMLST analysis of a wide spectrum of different genes,and the demonstration of subgroup-specific differencesin toxin profiles, psychrotolerance, and the ability to har-bor some plasmids, suggest that these seven subgroupsare not based solely on neutral genomic polymorphisms,but instead reflect differences in both the genotypes andphenotypes of the B. cereus group organisms. Furtherinvestigations that we are performing now will indicatethe final number of subgroups and will group them ingenomospecies, in accordance with genetic principles[79].

Unique chromosomal sequences, as well as anthraxtoxin genes, are the well established genetic markersused in genomic methods of B. anthracis identificationand discrimination from related organisms of B. cereusgroup [33,49–53,57,59–62,64–66]. However, recently anumber of publications revealed the presence of plas-mids with high levels of homology (up to 99.6%) toB. anthracis plasmids pXO1 and pXO2 in a number ofnon-anthrax isolates from the B. cereus group, such asB. cereus ATCC10987, B. cereus F4810/72, B. cereusG9241, B. thuringiensis HD73, B. thuringiensis 97-27,B. cereus D-17, B. cereus ATCC43881, B. thuringien-sis ATCC33679, as well in several isolates from outsidethe B. cereus group, such as B. circulans CBD118and Bacillus sp. CBD119 (related to B. luciferences)[17,78,105,113–117]. Therefore, traditional targets usedfor B. anthracis identification must be supplementedwith new genomic markers. From this point of view, mul-ticopied rRNA genes, as standardized targets commonlyused for most bacteria classification and discrimina-tion, represent a highly useful target for emergent B.cereus organism detection. In addition, the presence of20,000–50,000 ribosomes in one bacterial cell [118]makes rRNA a valuable target for B. cereus organismidentification without preliminary PCR amplification.

The goal of this study was to confirm the existenceand demonstrate the practical usefulness of previouslydeveloped genetic markers for identification and dis-crimination of all seven subgroups of the B. cereus group.

Interactions 171 (2008) 212–235

4.2. Discrimination of subgroups in B. cereus groupwith rRNA probes

A set of 16S/23S rRNA targeted oligonucleotideprobes was designed in the current study (Figs. 1 and 2,Tables 1 and 2) and these probes were immobilizedwithin an oligonucleotide microarray. This microar-ray enabled us for the first time to differentiate allseven subgroups in the B. cereus group from eachother (Figs. 2 and 3), including subgroup Anthracisthat contains all B. anthracis isolates whose rRNA wassequenced [85] (http://www.ncbi.nlm.nih.gov/Genbank/index.html; http://rdp.cme.msu.edu/index.jsp). We suc-cessfully tested our microarray with a set of previouslyidentified [85] reference strains representing all sevensubgroups. Using our microarray we also identified sub-group affiliation for B. cereus HER1414 and B. cereusNCTC3329 (Cereus A) (Fig. 3A) and B. mycoidesATCC23258 (Mycoides A) (Fig. 3B), whose rRNAgenes have not been sequenced yet.

Previous studies have shown that 16S rRNAsequences of organisms from subgroups Anthracis andCereus A, e.g., B. anthracis Sterne and B. cereusNCTC11143, have 99.9–100% similarity and do not con-tain subgroup-specific differences [83,85]. However, wedemonstrated that these organisms, as well as other iso-lates that belong to Anthracis and Cereus A subgroup,may be differentiated using subgroup-specific signa-tures located in their 23S rRNA sequences (Fig. 3A andTable 2).

4.3. Perfect match–mismatch discrimination

This study also indicated that our 3D gel-basedoligonucleotide microarray format [86] was able to dis-criminate microbial strains which differed by only onebase in their rRNA molecules based on hybridization sig-nal ratios between pairs of perfect match and mismatchedprobes. For example, using pairs ps7/ps8 and ps17/ps18we discriminated B. cereus NCTC11143 (Cereus A)from B. cereus T (Cereus B) and B. cereus T from B.cereus ATCC43881 (Thuringiensis A) (Fig. 2), strainsthat differ from each other in their 16S rRNA genesequences by only one substitution (A/C (1015) and T/C(192), respectively) (Fig. 1A and Table 1). Successfulone base mismatch discrimination was also demon-strated for 16S rRNA targeted probe pairs ps1/ps2,ps15/ps16, ps27/ps28 (Table 1 and Fig. 2) and for

pairs ps21/ps22, ps23/ps24 and ps31/ps32 related to23S rRNA (Table 2 and Fig. 3). The ability of ourtechnology to discriminate single base differences sug-gests that organisms affiliated to the same subgroup,

logical

b[t

4m

wdrsstswoittacmmthfttpspraInsUhpwmtnheicebmit

S.G. Bavykin et al. / Chemico-Bio

ut differing by one nucleotide strain-specific mutation85], may be also differentiated with our 3D microarrayechnology.

.3.1. Efficacy of perfect–mismatch pairedicroarray for emergent detectionOne of the obstacles for nucleic acid detection

ith oligonucleotide microarrays is the considerableifference between A-T and G-C pair stability thatesults in sequence-dependent intensity of fluorescentignals from different oligonucleotide probes of theame length. Due to this sequence dependent hybridiza-ion stability, mismatch-forming duplexes can be moretable than perfect duplexes of the same length butith different sequences. Therefore, perfectly matchedligonucleotide probes may produce signals of lowerntensity than other probes containing a mismatch withhe target. Sequence-dependent hybridization impactshe time required for efficient hybridization of all probes,nd as a result most microarray hybridization proto-ols hybridize for at least several hours [119]. However,icroarrays constructed with p/mm pairs in which theismatch is located in the middle of the probe effec-

ively overcome the problem of sequence dependentybridization stability because within a p/mm pair a per-ectly matched sequence should always produce a duplexhat is more stable than one containing mismatch. Forhose cases in which both members of the p/mm probeair generate signals of equal intensity, these probes canimply be eliminated from further consideration. The/mm pair strategy has been widely used in microar-ay technology for one base mismatch discriminationnd as a control for nonspecific hybridization [120,121].n our case, we used p/mm probe pairs to discrimi-ate single- or double-nucleotide variations in rRNAequences and differentiate one subgroup from all others.se of p/mm pairs permitted us to dramatically decreaseybridization time to as little as 5 min while still enablingerfect–mismatch discrimination (Table 4). In addition,e have recently developed a free radical based Silicaini-column method for microarray sample preparation

hat requires only 1 min for labeling-fragmentation ofucleic acids [92]. Therefore, our methods can produceighly purified end-labeled nucleic acids samples withasy-controlled fragment length, in roughly 15 min start-ng from whole bacterial cells. In addition, these methodsan be utilized for both DNA and RNA and should beasily automated. Therefore, combination of the 3D gel-

ased p/mm paired microarray with free radical-basedethod of sample preparation may be effectively used

n development of detectors for emergent bacteria iden-ification.

Interactions 171 (2008) 212–235 229

4.3.2. SNP sites quantificationWe have demonstrated above that our microarray

technology enabled us to discriminate single nucleotidevariations in rRNA sequences. At the same time, vari-ation of p/mm ratios in some sites of subgroup-specificdifferences in different organisms of the same subgroupsuggested that several copies of rRNA genes (total num-ber of rRNA operons may vary from 13 to15 in B. cereusgroup [78]) may contain SNPs. Our recent analysis of thecomplete genome sequences of B. cereus group isolatesconfirmed this. SNPs were found at positions 594 and1559 of 23S rRNA in several isolates [85]. In currentstudy we demonstrated in vivo (Fig. 4) and in artificialmixtures (Fig. 5 and Table 3) that our microarray plat-form was able to identify the presence of SNPs whensingle nucleotide substitutions were present in at leastone of five copies, with an accuracy of p/mm ratio notsmaller then 10% (Table 3). Moreover, based on the datapresented in Fig. 5 it is easy to hypothesize that it wouldbe possible to identify individual species within mixedcommunities using a database of hybridization imagesof individual B. cereus group organisms.

4.3.3. Bacterial identification in environmentalsamples

Analysis of nucleic acids in environmental samplesgenerally requires removal of organic matter (e.g., humicacids), debris, etc., because many environmental con-taminants can inhibit reactions employed for analysis.Membrane filtration with a pre-filter that accumulatesinsoluble environmental contaminants helps but does notcompletely prevent PCR inhibition due to co-elution ofcompounds such as humic substances, metal ions, etc.,that are concentrated on the filter along with targetedbacteria [122]. Our successful testing of our microar-ray with environmental samples without pre-filtrationin presence of organic contaminants (Table 5), whichexceeded volume of pure bacterial cells approximately20-fold, indicated that even at this extreme condition ourmicroarray technology in combination with our previ-ously developed free-radical based silica mini-columnsample preparation method [87,92], worked properlyand has a significant potential for direct analysis of envi-ronmental samples without preliminary cell culturing orPCR amplification.

4.4. Hierarchical approach in microarrayconstruction

In the current study we demonstrated that our microar-ray unambiguously discriminated B. anthracis from allother members of the B. cereus group based on 16S/23S

230 S.G. Bavykin et al. / Chemico-Biological Interactions 171 (2008) 212–235

ted problse posinalysis

Fig. 7. Two strategies for microbial identification. Hierarchically nesand analysis strategy (A and B). The hierarchical strategy reduces fapredecessor nodes in the phylogenetic tree, resulting in a “branch” areporting.

rRNA sequences, and we demonstrated discrimination ofseven subgroups within the B. cereus group (Anthracis,Cereus A and B, Thuringiensis A and B, and MycoidesA and B) that had previously been identified only the-oretically based on “subgroup-specific signatures” [85].Subgroup discrimination was achieved using probe pairsselected in accordance with these subgroup-specificmarkers. However, the limited number of subgroup-specific differences between organisms within the B.cereus group made it difficult to select subgroup-specificprobes that did not match any organisms from closelyrelated, non-B. cereus group species. To address this dif-ficulty, the microarray included probes ps25 and ps26,which were specific for the B. cereus group and the B.

subtilis group, respectively, and which successfully dis-criminated organisms from these two groups (Fig. 6).The combination of group-specific probes ps25 and ps26with probe pairs specific for the B. cereus subgroups rep-

e design and analysis strategy (C and D) and endpoint probe designtives due to concordance of positive microarray signals (red dots) at(D) instead of an “endpoint” analysis (B) for data interpretation and

resents a hierarchical probe design strategy that offerssignificant advantages over an “endpoint” strategy thatfocuses on designing specific probes for each individ-ual target. The “endpoint” approach requires the designof probes that are absolutely unique for each individualtarget (Fig. 7A). The requirement for absolutely uniqueprobes is labor-consuming and necessitates regular ren-ovation of probe sets in response to newly describedorganisms. The “endpoint” strategy also generates only“endpoint” information (Fig. 7B). Alternatively, “hierar-chical” probe design (Fig. 7C) allows one to concentrateprobe selection efforts on the key nodes and, in practice,produces more reliable identification assuming perfectsignals in a chain of nodes as a major requirement of cor-

rect group identification (Fig. 7D). The “hierarchical”approach also does not necessitate frequent re-designof probes characterizing each node, but in fact callsfor doing this rather rarely (in response to considerable

logical

cphp(pbsmsngctcpummtpiomgriaua

A

CSraWuap

0RAg

R

S.G. Bavykin et al. / Chemico-Bio

hanges in structure of phylogenetic tree only) in com-arison with the “endpoint” approach. Therefore, theierarchical approach we have taken combining probeairs differentiating closely related species/isolates“determinative” probe set) with specific (non-paired)robes targeting particular nodes on the target-relatedranch of the phylogenetic tree (“phylogenetic” probeet) should significantly improve the reliability oficroarray-based microbial identification. Recently, we

uccessfully tested a hierarchically organized “phyloge-etic” probe set targeting “all life”, eubacteria, B. subtilisroup and B. cereus group cells with human, Escherichiaoli, B. subtilis and B. thuringiensis targets [87]. Fromhis point of view, further development of hierarchicallyomplicated bacterial phylogenetic trees, where eacharticular branch consists of dozens of nodes, will leads to development of highly reliable microarray-basedicrobial diagnostics. Therefore, combination of highultiplicity of rRNA targets in bacterial cells with simul-

aneous using of “phylogenetic” and “discriminative”robe sets provide highly sensitive and highly reliabledentification of B. anthracis and related B. cereus grouprganisms. The developed subgroup-specific markersay be useful for B. anthracis and other B. cereus

roup pathogens identification in terroristic and envi-onmental events, food outbreaks and bacterial controln agriculture, where B. thuringiensis strains are useds a bio-pesticide, as well as in basic studies of molec-lar mechanisms of toxin actions and their distributionmong B. cereus group isolates.

cknowledgements

We express our gratitude to Gennadiy Yershov; Borishernov; Julia Golova; Eduard Timofeev, Irina Taran,ergey Surzhikov and Ann Gemmell for microarray fab-ication and Alexey Chernyi for helpful consultations innalysis of rRNA sequences and hybridization images.e are also grateful to John Ezzell, who kindly provided

s with reference strains of microorganisms, Wentso Liund Jodi Flax for their help in the early stages of thisroject.

This research was supported by NIH grants 5R01AI59517 and 5R43AI 069627. Defense Advancedesearch Project Agency under Interagency AgreementO-E428 and by the Russian Human Genome Programrant 5/2000.

eferences

[1] P.H.A. Sneath, Endospore-forming Gram-positive rods andcocci, in: J.G. Holt (Ed.), Bergey’s Manual of Systematic Bac-

Interactions 171 (2008) 212–235 231

teriology, The Williams & Wilkins Co., Baltimore, MD, 1986,pp. 1104–1139.

[2] F.A. Drobniewski, Bacillus cereus and related species, Clin.Microbiol. Rev. 6 (1993) 324–338.

[3] P.C.B. Turnbull, Definitive identification of Bacillusanthracis—a review, J. Appl. Microbiol. 87 (1999) 237–240.

[4] S. Lechner, R. Mayr, K.P. Fransis, B.M. Pruss, T. Kaplan,E. Wiebner-Gunkel, G.S. Stewart, A.B. Scherer, Bacillus wei-henstephanensis sp. nov. is a new psychrotolerant species ofthe Bacillus cereus group, Int. J. Syst. Bacteriol. 48 (1998)1373–1382.

[5] L.K. Nakamura, Bacillus pseudomycoides sp. nov., Int. Syst.Bacteriol. 48 (1998) 1031–1034.

[6] M. Delaporte, Description de Bacillus medusa n.sp., C. R. Acad.Sci. Paris 269 (Serie D) (1969) 1129–1131.

[7] M. Hugh-Jones, 1996–97 global anthrax report, J. Appl. Micro-biol. 87 (1999) 189–191.

[8] D.C. Dragon, B.T. Elkin, J.S. Nishi, T.R. Ellsworth, A review ofanthrax in Canada and implications for research on the diseasein northern bison, J. Appl. Microbiol. 87 (1999) 208–213.

[9] P.E. Granum, Bacillus cereus, in: M.P. Doyle, et al. (Eds.),Food Microbiology: Fundamentals and Frontiers, 2nd ed., ASMPress, Washington, DC, 2001, pp. 373–381.

[10] K.T. Rajkowski, J.L. Smith, Update: food poisoning and otherdiseases induced by Bacillus cereus, in: Y.H. Hui, M.D. Pierson,J.R. Gorham (Eds.), Foodborne Disease Handbook, vol. I, 2nded., Marcel Dekker, Inc., New York, 2001, pp. 61–76.

[11] J.L. Schoeni, M.C.L. Wong, Bacillus cereus food poisoning andits toxins, J. Food Prot. 68 (2005) 636–648.

[12] A. Sorokin, B. Kandelon, K. Guilloux, N. Galleron, N.Wackerow-Kouzova, S.D. Ehrlich, D. Bourguet, V. Sanchis,Multiple-locus sequence typing analysis of Bacillus cereus andBacillus thuringiensis reveals separate clustering and a distinctpopulation structure of psychrotrophic strains, Appl. Environ.Microbiol. 72 (2006) 1569–1578.

[13] P.H. Damgaard, Natural occurrence and dispersal of Bacillusthuringiensis in the environment, in: J.-F. Charles, A. Delecluse,C. Nielsen-Le Roux (Eds.), Entomapathogenic Bacteria: FromLaboratory to Field Application, Kluwer Academic Publishers,Dordrecht, 2000, pp. 23–40.

[14] L.P. Stenfors, R. Mayr, S. Scherer, P.E. Granum, Pathogenicpotential of fifty Bacillus weihenstephanansis strains, FEMSMicrobiol. Lett. 215 (2002) 47–51.

[15] N.B. Hendriksen, B.M. Hansen, J.F. Johansen, Occurrence andpathogenic potential of Bacillus cereus group bacteria in a sandyloam, Antonie Van Leeuwenhoek 89 (2006) 239–249.

[16] A. Kotiranta, K. Lounatmaa, M. Haapasalo, Epidemiology andpathogenesis of Bacillus cereus infections, Microbes Infect. 2(2000) 189–198.

[17] A.R. Hoffmaster, J. Ravel, D.A. Rasko, G.D. Chapman, M.D.Chute, C.K. Marston, B.K. De, C.T. Sacchi, C. Fitzgerald, L.W.Mayer, M.C. Maiden, F.G. Priest, M. Barker, L. Jiang, R.Z. Cer,J. Rilstone, S.N. Peterson, R.S. Weyant, D.R. Galloway, T.D.Read, T. Popovic, C.M. Fraser, Identification of anthrax toxingenes in a Bacillus cereus associated with an illness resemblinginhalation anthrax, Proc. Natl. Acad. Sci. U.S.A. 101 (2004)8449–8454.

[18] C.S. Han, G. Xie, J.F. Challacombe, M.R. Altherr, S.S. Bhotika,N. Brown, D. Bruce, C.S. Campbell, M.L. Campbell, J. Chen,O. Chertkov, C. Cleland, M. Dimitrijevic, N.A. Doggett, J.J.Fawcett, T. Glavina, L.A. Goodwin, L.D. Green, K.K. Hill, P.Hitchcock, P.J. Jackson, P. Keim, A.R. Kewalramani, J. Long-

logical

232 S.G. Bavykin et al. / Chemico-Bio

mire, S. Lucas, S. Malfatti, K. McMurry, L.J. Meincke, M.Misra, B.L. Moseman, M. Mundt, A.C. Munk, R.T. Okinaka,B. Parson-Quintana, L.P. Reilly, P. Richardson, D.L. Robin-son, E. Rubin, E. Saunders, R. Tapia, J.G. Tesmer, N. Thayer,L.S. Thompson, H. Tice, L.O. Ticknor, P.L. Wills, T.S. Brettin,P. Gilna, Pathogenomic sequence analysis of Bacillus cereusand Bacillus thuringiensis isolates closely related to Bacillusanthracis, J. Bacteriol. 188 (2006) 3382–3390.

[19] A. Lapidus, E. Goltsman, S. Auger, N. Galleron, B. Segurens,C. Dossat, M.L. Land, V. Broussole, J. Brillard, M.-H. Guine-bretiere, V. Sanchis, C. Nguyen-The, D. Lereclus, P. Richardson,P. Winker, J. Weissenbach, S.D. Ehrlich, A. Sorokin, Extend-ing the Bacillus cereus group genomics to putative food-bornepathogens of different toxicity, Chem. Biol. Interact. 171 (2008)236–249.

[20] M. Ehling-Schultz, M. Fricker, S. Scherer, Bacillus cereus, thecausative agent of emetic type of food-borne illness, Mol. Nutr.Food Res. 48 (2004) 479–487.

[21] J.L. McKillip, Prevalence and expression of enterotoxins inBacillus cereus and other Bacillus spp., a literature review,Antonie Van Leeuwenhoek 77 (2000) 393–399.

[22] T.S. Pirttijarvi, M.A. Andersson, A.C. Scoging, M.S. Salkinoja-Salonen, System., Evaluation of methods for recognising strainsof the Bacillus cereus group with food poisoning potentialamong industrial and environmental contaminants, Syst. Appl.Microbiol. 22 (1999) 133–144.

[23] B. Lambert, M. Peferoen, Insecticidal promise of Bacillusthuringiensis. Facts and mysteries about a successful biopes-ticide, Bioscience 42 (1992) 112–122.

[24] E. Schnepf, N. Crickmore, J. Van Rie, D. Lereclus, J. Baum,J. Feitelson, D.R. Zeigler, D.H. Dean, Bacillus thuringiensisand its pesticidal crystal proteins, Microbiol. Mol. Biol. Rev. 62(1998) 775–806.

[25] H.D. Burges, Bacillus thuringiensis in pest control, PesticideOutlook 12 (2001) 90–98.

[26] B.M. Hansen, S. Salamitou, Virulence of Bacillus thuringien-sis, in: J.-F. Charles, A. Delecluse, C. Nielsen-Le Roux (Eds.),Entomapathogenic Bacteria: From Laboratory to Field Applica-tion, Kluwer Academic Publishers, Dordrecht, 2000, pp. 41–64.

[27] A. Klier, Virulence of Bacillus thuringiensis, in: J.-F. Charles, A.Delecluse, C. Nielsen-Le Roux (Eds.), Entomapathogenic Bac-teria: From Laboratory to Field Application, Kluwer AcademicPublishers, Dordrecht, 2000, pp. 485–504.

[28] M.E. Whalon, B.A. Wingerd, Bt: Mode of action and use, Arch.Insect. Biochem. Physiol. 54 (2003) 200–211.

[29] A.M.G. Rivera, P.E. Granum, F.G. Priest, Common occurrenceof enterotoxin genes and enterotoxicity in Bacillus thuringien-sis, FEMS. Microbiol. Lett. 190 (2000) 151–155.

[30] P.H. Damgaard, P.E. Granum, J. Bresciani, M.V. Torre-grossa, J. Eilenberg, L. Valentino, Characterization of Bacillusthuringiensis isolated from infections in burn wounds, FEMSImmunol. Med. Microbiol. 18 (1997) 47–53.

[31] M. Perani, A.H. Bishop, A. Vaid, Prevalence of �-exotoxin,diarrhoeal toxin and specific �-endotoxin in natural isolates ofBacillus thuringienses, FEMS Microb. Lett. 160 (1998) 55–60.

[32] D.M. Guttmann, D.J. Ellar, Phenotypic and genotypic compari-son of 23 strains from the Bacillus cereus complex for a selection

of known and putative B. thuringiensis virulence factors, FEMSMicrob. Lett. 188 (2000) 7–13.

[33] N. Sergeev, M. Distler, M. Vargas, V. Chizhikov, K.E. Herold, A.Rasooly, Microarray analysis of Bacillus cereus group virulencefactors, J. Microbiol. Methods 65 (2006) 488–502.

Interactions 171 (2008) 212–235

[34] S.G. Jackson, R.B. Goodbrand, R. Ahmed, S. Kasatiya, Bacilluscereus and Bacillus thuringiensis isolated in a gastroenteri-tis outbreak investigation, Lett. Appl. Microbiol. 21 (1995)103–105.

[35] P.H. Damgaard, Diarrhoeal enterotoxin production by strainsof Bacillus thuringiensis isolated from commercial Bacillusthuringiensis-based insecticides, FEMS Immunol. Med. Micro-biol. 12 (1995) 242–250.

[36] E. Helgason, D.A. Caugant, I.K. Olsen, A.B. Kolsto, Geneticstructure of population of Bacillus cereus and B. thuringiensisisolates associated with periodontitis and other human infec-tions, J. Clin. Microbiol. 38 (2000) 1615–1622.

[37] V. Mantynen, K. Lindstrom, A rapid PCR-based DNA testfor enterotoxic Bacillus cereus, Appl. Environ. Microbiol. 64(1998) 1634–1639.

[38] B.M. Pruss, R. Dietrich, B. Nibler, E. Martlbauer, S. Scherer,The hemolytic enterotoxin HBL is broadly distributed amongspecies of the Bacillus cereus group, Appl. Environ. Microbiol.65 (1999) 5436–5442.

[39] S.H. Beattie, A.G. Williams, Detection of toxigenic strains ofBacillus cereus and other Bacillus spp. with an improved cyto-toxicity assay, Lett. Appl. Microbiol. 28 (1999) 221–225.

[40] Y.M. Hsieh, S.J. Sheu, Y.L. Chen, H.Y. Tsen, Enterotoxigenicprofiles and polymerase chain reaction detection of Bacilluscereus group cells and B. cereus strains from foods and food-borne outbreaks, J. Appl. Microbiol. 87 (1999) 481–490.

[41] H.-Y. Tsen, M.L. Chen, Y.M. Hsieh, S.J. Sheu, Y.L. Chen, Bacil-lus cereus group strains, their hemolysin BL activity, and theirdetection in foods using a 16S RNA and hemolysin BL gene-targeted multiplex polymerase chain reaction system, J. FoodProt. 63 (2000) 1496–1502.

[42] I.-C. Yang, D.Y.-C. Shih, T.-P. Huang, Y.-P. Hunag, J.-Y. Wang,T.-M. Pan, Establishment of a novel multiplex PCR assay anddetection of toxigenic strains of the species in the Bacillus cereusgroup, J. Food Prot. 68 (2005) 2123–2130.

[43] N.B. Hendriksen, B.M. Hansen, J.E. Johansen, Occurrence andpathogenic potential of Bacillus cereus group bacteria in sandyloam, Antonie Van Leeuwenhoek 89 (2006) 239–249.

[44] S.D. Heins, D.S. Manker, D.R. Jimenez, P.G. Marrone, Bacil-lus mycoides strain for controlling corn rootworms, US Patent5,906,818 (1999).

[45] D.V. Lim, J.M. Simpson, E.A. Kearns, M.F. Kramer, Current anddeveloping technologies for monitoring agents of bioterrorismand biowarfare, Clin. Microbiol. Rev. 18 (2005) 583–607.

[46] E.G. Dib, S.A. Dib, D.A. Korkmaz, N.K. Mobarakai, J.B.Glaser, Nonhemolytic, nonmotile Gram-positive rods indicativeof Bacillus anthracis, Emerg. Infect. Dis. 9 (2003) 1013–1015.

[47] S.R. Klee, H. Nattermann, S. Becker, M. Urban-Schriefer, D.Jacob, B. Appel, Evaluation of different methods to discriminateBacillus anthracis from other bacteria of the Bacillus cereusgroup, J. Appl. Microbiol. 100 (2006) 673–681.

[48] S.M. Levine, Y.-W. Tang, Z. Pei, Recent advances in the rapiddetection of Bacillus anthracis, Rev. Med. Microbiol. 16 (2005)125–133.

[49] U. Nubel, P.M. Schmidt, E. Reiss, F. Bier, W. Beyer, D.Naumann, Oligonucleotide microarray based on intergenic tran-scribed spacer in ribosomal DNA, FEMS Microbiol. Lett. 240

(2004) 215–223.

[50] D. Volokhov, A. Pomerantzev, V. Kivovich, A. Rasooly, V.Chizhikov, Identification of Bacillus anthracis by multiporemicroarray hybridization, Diagn. Microbiol. Infect. Dis. 49(2004) 163–171.

logical

S.G. Bavykin et al. / Chemico-Bio

[51] J.E. Burton, O.J. Oshota, N.J. Silman, Differential identificationof Bacillus anthracis from environmental Bacillus species usingmicroarray analysis, J. Appl. Microbiol. 101 (2006) 754–763.

[52] R.A. Hutson, C.J. Duggleby, J.R. Lowe, R.J. Manchee, P.C.B.Turnbull, The development and assessment of DNA andoligonucleotide probes for the specific detection of Bacillusanthracis, J. Appl. Bacteriol. 75 (1993) 463–472.

[53] M.A. Lee, G. Brightwell, D. Leslie, H. Bird, A. Hamilton,Fluorescent detection techniques for real-time multiplex strandspecific detection of Bacillus anthracis using rapid PCR, J.Appl. Microbiol. 87 (1999) 218–223.

[54] G. Patra, P. Sylvestre, V. Ramisse, J. Therasse, J.-L. Guedson,DNA fingerprinting of Bacillus anthracis, Salisbury MedicalBulletin, Special Supplement No. 87 (1996) 59.

[55] I. Henderson, Fingerprinting Bacillus anthracis strains, Sal-isbury Medical Bulletin, Special Supplement No. 87 (1996)55–58.

[56] Y.-H. Shangkuan, J.-F. Yang, H.-C. Lin, M.-F. Shaio, Com-parison of PCR-RFLP, ribotyping and ERIC-PCR for typingBacillus anthracis and Bacillus cereus strains, J. Appl. Micro-biol. 89 (2000) 452–462.

[57] L.J. Harrell, G.L. Andersen, K.H. Wilson, Genetic variabilityof Bacillus anthracis and related species, J. Clin. Microbiol. 33(1995) 1847–1850.

[58] D. Daffonchio, A. Cherif, S. Borin, Homoduplex and het-eroduplex polymorphisms of the amplified ribosomal 16S–23Sinternal transcribed spacers describe genetic relationships in the“Bacillus cereus Group”, Appl. Environ. Microbiol. 66 (2000)5460–5468.

[59] K.S. Ko, J.-M. Kim, J.-W. Kim, B.Y. Jung, W. Kim, I.J. Kim, Y.-H. Kook, Identification of Bacillus anthracis by rpoB sequenceanalysis and multiplex PCR, J. Clin. Microbiol. 41 (2003)2908–2914.

[60] W. Hurtle, E. Bode, D.A. Kulesh, R.S. Kaplan, J. Garrison, D.Bridge, M. House, M.S. Frye, B. Loveless, D. Norwood, Detec-tion of the Bacillus anthracis gyrA gene by using a minor groovebinder probe, J. Clin. Microbiol. 42 (2004) 179–185.

[61] M.T. La Duc, M. Satomi, N. Agata, K. Venkateswaran, gyrBas a phylogenetic discriminator for members of the Bacillusanthracis group, J. Microbiol. Methods 56 (2004) 383–394.

[62] L. Radnedge, P.G. Agron, K.K. Hill, P. Jackson, L.O. Ticknor,P. Keim, G.L. Anderson, Genome differences that distinguishBacillus anthracis from Bacillus cereus and Bacillus thuringien-sis, Appl. Environ. Microbiol. 69 (2003) 2755–2764.

[63] S. Sozhamannan, M.D. Chute, F.D. McAfee, D.E. Fouts, A.Akmal, D.R. Galloway, A. Meteczun, L.W. Baillie, T.D. Read,The Bacillus anthracis chromosome contains four conserved,excision-proficient, putative profages, BMC Microbiol. 6 (2006)34.

[64] P. Keim, M.N. Van Ert, T. Pearson, A.J. Vogler, L.Y. Huyih,D.M. Wagner, Anthrax molecular epidemiology and forensics:using the appropriate marker for different evolutionary scale,Infect. Genet. Evol. 4 (2004) 205–213.

[65] M.N. Van Ert, W.R. Easterday, L.Y. Huynh, R.T. Okinaka, M.E.Hung-Jones, J. Ravel, S.R. Zaneki, T. Pearson, T.S. Simonson,J.M. U’Ren, S.M. Kachur, R.R. Leadem-Dougherty, S.D. Rho-ton, G. Zinser, J. Farlow, P.R. Coker, K.L. Smith, B. Wang, L.J.

Kenefic, C.M. Fraser-Liggett, D.M. Vagner, P. Keim, Globalgenetic population structure of Bacillus anthracis, PLoS ONE5 (2007) e46.

[66] A. Cherif, S. Borin, A. Rizi, H. Quzari, A. Boudabous,D. Daffonchio, Characterization of a repetitive element

Interactions 171 (2008) 212–235 233

polymorphism–polymerase chain reaction chromosomalmarker that discriminates Bacillus anthracis from relatedspecies, J. Appl. Microbiol. 63 (2002) 450–462.

[67] P. Kiem, A. Kalif, J. Schupp, K. Hill, S.E. Travis, K. Richmond,D.M. Adair, M. Hugh-Jones, C.R. Kuske, P. Jackson, Molecu-lar evolution and diversity in Bacillus anthracis as detected byamplified fragment length polymorphism markers, J. Bacteriol.179 (1997) 818–824.

[68] P.J. Jackson, K.K. Hill, M.T. Laker, L.O. Ticknor, P. Keim,Genetic comparison of Bacillus anthracis and its close relativesusing amplified fragment length polymorphism and poly-merase chain reaction analysis, J. Appl. Microbiol. 87 (1999)263–269.

[69] K.K. Hill, L.O. Ticknor, R.T. Okinaka, M. Asay, H. Blair,K.A. Bliss, M. Laker, P.E. Pardington, A.P. Richardson, M.Tonks, D.J. Beecher, J.D. Kemp, A.-B. Kolsto, A.C.L. Wong,P. Keim, P.J. Jackson, Fluorescent amplified fragment lengthpolymorphism analysis of Bacillus anthracis, Bacillus cereusand Bacillus thuringiensis isolates, Appl. Environ. Microbiol.70 (2004) 1068–1080.

[70] E. Helgason, O.A. Økstad, D.A. Caugant, H.A. Johansen, A.Fouet, M. Mock, I. Hegna, A.-B. Kolsto, Bacillus anthracis,Bacillus cereus, and Bacillus thuringiensis—one species on thebasis of genetic evidence, Appl. Environ. Microbiol. 66 (2000)2627–2630.

[71] W. Beyer, S. Pocivalsek, R. Bohm, Polymerase chainreaction-ELISA to detect Bacillus anthracis from soilsamples—limitations of present published primers, J. Appl.Microbiol. 87 (1999) 229–236.

[72] B.N. Strizhkov, A.L. Drobyshev, V.M. Mikhailovich, A.D.Mirzabekov, PCR amplification on a microarray of gel-immobilized oligonucleotides: detection of bacterial toxin- anddrug-resistant genes and their mutations, BioTechniques 29(2000) 844–857.

[73] E.V. Koonin, K.S. Makarova, L. Aravind, Horizontal gene trans-fer in procariotes: quantification and classification, Annu. Rev.Microbiol. 55 (2001) 709–742.

[74] R. Jain, M.C. Rivera, J.E. Moore, J.A. Lake, Horizontal genetransfer in microbial genome evolution, Theor. Popul. Biol. 61(2002) 489–495.

[75] J.G. Lawrence, Gene transfer in bacteria: speciation withoutspecies? Theor. Popul. Biol. 61 (2002) 449–460.

[76] G. Santoyo, D. Romero, Gene conversion and concerted evo-lution in bacterial genomes, FEMS Micobiol. Rev. 29 (2005)169–183.

[77] J. Hacker, J. Kaper, Pathogenicity islands and the evolution ofmicrobes, Annu. Rev. Microbiol. 54 (2000) 641–679.

[78] D.A. Rasko, M.R. Altherr, C.S. Han, J. Ravel, Genomics ofthe Bacillus cereus group of organisms, FEMS Micob. Rev. 29(2005) 303–329.

[79] D.J. Brenner, J.T. Staley, N.R. Krieg, Classification of prokary-otic organisms and the concept of bacterial speciation, in: G.M.Garrity (Ed.), Bergey’s Manual of Systematic Bacteriology, 2nded., vol. 2, The Proteobacteria; Part A Introductory Essays,Springer, New York, NY, 2005, pp. 27–31.

[80] G.M. Garrity, J.A. Bell, T. Lilburn, The revised road mapto the Manual, in: G.M. Garrity (Ed.), Bergey’s Manual of

Systematic Bacteriology, 2nd ed., vol. 2, The Proteobacteria;Part A Introductory Essays, Springer, New York, NY, 2005,pp. 159–187.

[81] C.R. Woese, Interpreting the universal phylogenetic tree, Proc.Natl. Acad. Sci. U.S.A. 97 (2000) 8392–8396.

logical

234 S.G. Bavykin et al. / Chemico-Bio

[82] C. Ash, M.D. Collins, Comparative analysis of 23S ribosomalRNA gene sequences of Bacillus anthracis and emetic Bacilluscereus determined by PCR-direct sequencing, FEMS Microbiol.Lett. 94 (1992) 75–80.

[83] C. Ash, J.A.E. Farrow, M. Dorsch, E. Stackebrandt, M.D.Collins, Comparative analysis of Bacillus anthracis, Bacilluscereus, and related species on the basis of reverse transcrip-tase sequencing of 16S rRNA, Int. J. Syst. Bacteriol. 41 (1991)343–346.

[84] C. Ash, J.A.E. Farrow, W. Wallbanks, M.D. Collins, Phy-logenetic heterogeneity of the genus Bacillus revealedby comparative analysis of small-subunit-ribosomal RNAsequences, Lett. Appl. Microbiol. 13 (1991) 202–206.

[85] S.G. Bavykin, Y.P. Lysov, V. Zakhariev, J.J. Kelly, J. Jackman,D.A. Stahl, A. Cherni, Use of 16S rRNA, 23S rRNA, and gyrBgene sequence analysis to determine phylogenetic relationshipsof Bacillus cereus group microorganisms, J. Clin. Microbiol. 42(2004) 3711–3730.

[86] J. Zlatanova, A.D. Mirzabekov, Gel immobilized microarraysof nucleic acids and proteins, Methods Mol. Biol. 170 (2001)17–38.

[87] S.G. Bavykin, J.P. Akowski, V.M. Zakhariev, V.E. Barsky, A.N.Perov, A.D. Mirzabekov, Portable system for microbial sam-ple preparation and oligonucleotide microarray analysis, Appl.Environ. Microbiol. 67 (2001) 922–928.

[88] D. Guschin, G. Yershov, A. Zaslavsky, A. Gemmell, V. Shick, D.Proudnikov, P. Arenkov, Mirzabekov, Manual manufacturing ofoligonucleotide, DNA, and protein microarrays, Anal. Biochem.250 (1997) 203–211.

[89] D. Proudnikov, E. Timofeev, Mirzabekov, Immobilization ofDNA in polyacrylamide gel for the manufacture of DNA andDNA-oligonucleotide microarrays, Anal. Biochem. 259 (1998)34–41.

[90] E.N. Timofeev, S.V. Kochetkova, A.D. Mirzabekov, V.L. Floren-tiev, Regioselective immobilization of short oligonucleotides toacrylic copolymer gels, Nucl. Acids Res. 24 (1996) 3142–3148.

[91] D.A. Stahl, B. Flesher, H.R. Mansfield, L. Montgomery, Useof phylogenetically based hybridization probes for studies ofruminal microbial ecology, Appl. Environ. Microbiol. 54 (1988)1079–1084.

[92] J.J. Kelly, B.K. Chernov, I. Tovstanovsky, A.D. Mirzabekov,S.G. Bavykin, Radical generating coordination complexes asa tool for rapid and effective fluorescent labeling and frag-mentation of DNA or RNA for microarray hybridization, Anal.Biochem. 311 (2002) 103–118.

[93] I.Ya. Barsky, A.P. Grammatin, A.V. Ivanov, E.Ya. Kreindlin,E.Yu. Kotova, V.E. Barsky, A.D. Mirzabekov, Wide-field lumi-nescence microscopes for analyzing biological microarrays, J.Opt. Technol. 65 (1998) 938–941.

[94] G. Yershov, V. Barsky, A. Belgovskiy, Eu. Kirillov, E. Kreindlin,I. Ivanov, S. Parinov, D. Guschin, A. Drobyshev, S. Dubiley,A. Mirzabekov, DNA analysis and diagnostics on oligonu-cleotide microarrays, Proc. Natl. Acad. Sci. U.S.A. 93 (1996)4913–4918.

[95] M.L. Chen, H.Y. Tsen, Discrimination of Bacillus cereus andBacillus thuringiensis with 16S rRNA and gyrB gene basedPCR primers and sequencing of their annealing sites, J. Appl.

Microbiol. 92 (2002) 912–919.

[96] S. Yamada, E. Ohashi, N. Agata, K. Venkateswaran, Cloningand nucleotide sequence analysis of gyrB of Bacillus cereus, B.thuringiensis, B. mycoides, and B. cereus in rice, Appl. Environ.Microbiol. 65 (1999) 1483–1490.

Interactions 171 (2008) 212–235

[97] L.O. Ticknor, A.B. Kolsto, K.K. Hill, P. Keim, M.T. Laker, M.Tonks, P.J. Jackson, Fluorescent amplified fragment length poly-morphism analysis of Norwegian Bacillus cereus and Bacillusthuringiensis soil isolates, Appl. Environ. Microbiol. 67 (2001)4863–4873.

[98] A. Cherif, S. Borin, A. Rizzi, H. Ouzari, A. Boundabous, D.Daffonchio, Bacillus anthracis diverges from related clades ofthe Bacillus cereus group in 16–23S ribosomal DNA inter-genic transcribed spacer containing tRNA genes, Appl. Environ.Microbiol. 69 (2003) 33–40.

[99] D. Daffonchio, N. Raddadi, M. Merabishvili, A. Cherif, L. Car-magnola, L. Brusetti, A. Rizi, N. Chanishvili, P. Visca, R. Sharp,S. Borin, Strategy for identification of Bacillus cereus and Bacil-lus thuringiensis strains closely related to Bacillus anthracis,Appl. Environ. Microbiol. 72 (2006) 1295–1301.

[100] F. von Stetten, K.P. Francis, S. Lechner, K. Neuhaus, S. Scherer,Rapid discrimination of psychrotolerant and mesophilic strainsof the Bacillus cereus group by PCR targeting of the 16S rDNA,J. Microbiol. Methods 34 (1998) 99–106.

[101] D. Wunschel, K.F. Fox, G.E. Black, A. Fox, Discriminationamong the Bacillus cereus group, in comparison to B. subtilis,by structural carbohydrate profiles and ribosomal RNA spacerregion PCR, Syst. Appl. Microbiol. 17 (1994) 625–635.

[102] C.R. Carlson, D.A. Caugant, A.B. Kolsto, Genotypic diversityamong Bacillus cereus and Bacillus thuringiensis strains, Appl.Environ. Microbiol. 60 (1994) 1719–1725.

[103] E. Helgason, D.A. Caugant, M.M. Lecadet, Y. Chen, J.Mahillon, A. Lovgren, I. Hegna, K. Kvaløy, A.B. Kolstø,Genetic diversity of Bacillus cereus/B. thuringiensis isolatesfrom natural sources, Curr. Microbiol. 37 (1998) 80–87.

[104] E. Helgason, N.J. Tourasse, R. Meisal, D.A. Caugant, A.-B.Kolsto, Multilocus sequence typing scheme for bacteria of theBacillus cereus group, Appl. Environ. Microbiol. 70 (2004)191–201.

[105] D.A. Rasko, J. Ravel, O.A. Okstad, E. Helgason, R.Z. Cer, L.Jiang, K.A. Shores, D.E. Fouts, N.J. Tourasse, S.V. Angiuoli,J. Kolonay, W.C. Nelson, A.B. Kolsto, C.M. Fraser, T.D. Read,The genome sequence of Bacillus cereus ATCC 10987 revealsmetabolic adaptations and a large plasmid related to Bacillusanthracis pXO1, Nucl. Acids Res. 32 (2004) 977–988.

[106] K.S. Ko, J.-W. Kim, J.-M. Kim, W. Kim, S. Chung, I.J. Kim,Y.-H. Kook, Population structure of the Bacillus cereus group asdeterminated by sequence analysis of six Housekeeping genesand the plcR gene, Infect. Immun. 72 (2004) 5253–5261.

[107] F.G. Priest, M. Barker, L.W.J. Baillie, E.C. Holmes, M.C.J.Maiden, Population structure and evolution of the Bacilluscereus group, J. Bacteriol. 186 (2004) 7959–7970.

[108] M. Vassileva, K. Torii, M. Oshimoto, A. Okamoto, N. Agata,K. Yamada, T. Hasegawa, M. Ohta, Phylogenetic analysis ofBacillus cereus isolates from severe systemic infections usingmultilocus sequence typing scheme, Microbiol. Immunol. 50(2006) 743–749.

[109] N.J. Tourasse, E. Helgason, O.A. Okstad, I.K. Hegna, A.-B.Kolsto, The Bacillus cereus group: novel aspects of populationstructure and genome dynamics, J. Appl. Microbiol. 101 (2006)579–593.

[110] M. Ehling-Schultz, B. Svensson, M.-H. Guinebretiere, T. Lind-

back, M. Andersson, A. Schulz, M. Fricker, A. Christiansson,P.E. Granum, E. Martlbauer, C. Nguyen-The, M. Salkinoja-Salonen, S. Scherer, Emetic toxin formation of Bacillus cereusis restricted to a single evolutionary lineage of closely relatedstrains, Microbiology 151 (2005) 183–197.

logical

S.G. Bavykin et al. / Chemico-Bio

[111] S.G. Bavykin, Y.P. Lysov, V. Zakhariev, J.J. Kelly, J. Jackman,D.A. Stahl, A. Cherni, J. Clin. Microbiol. 44 (2006) 2676 (Erra-tum).

[112] I.-C. Yang, D.Y.-C. Shih, T.-P. Huang, Y.-P. Huang, J.-Y. Wang,T.-M. Pan, Establishment of a novel multiplex PCR assay anddetection of toxigenic strains of the species in the Bacillus cereusgroup, J. Food Prot. 68 (2005) 2123–2130.

[113] C. Berry, S. O’Neil, E. Ben-Dov, A.F. Jones, L. Murphy, M.A.Quail, M.T. Holden, D. Harris, A. Zaritsky, J. Parkhill, Completesequence and organization of pBtoxis, the toxin-coding plas-mid of Bacillus thuringiensis subsp. israelensis, Appl. Environ.Microbiol. 68 (2002) 5082–5095.

[114] G.A. Van der Auwera, L. Andrup, J. Mahillon, Conjugativeplasmid pAW63 brings new insights into the genesis of theBacillus anthracis virulence plasmid pXO2 and of the Bacillusthuringiensis plasmid pBT9727, BMC Genom. 6 (2005) 103.

[115] M. Ehling-Schultz, M. Fricker, H. Grallert, P. Rieck, M. Wag-ner, S. Scherer, Cereulide synthetase gene cluster from emeticBacillus cereus: structure and location on a mega virulence plas-mid related to Bacillus anthracis toxin pXOI, BMC Microbiol.6 (2006) 20.

[116] V.A. Luna, D.S. King, K.K. Peak, F. Reeves, L. Heberlein-Larson, W. Veguilla, L. Heller, K.E. Duncan, A.C. Cannons,

P. Amuso, J. Cattani, Bacillus anthracis virulent plasmid pX02genes found in large plasmids of two other Bacillus species, J.Clin. Microbiol. 44 (2006) 2367–2377.

[117] J. Pannucci, R.T. Okinaka, R. Sabin, C.R. Kuske, Bacil-lus anthracis pXO1 plasmid sequence conservation among

Interactions 171 (2008) 212–235 235

closely related bacterial species, J. Bacteriol. 184 (2002)134–141.

[118] H. Bremer, P.P. Dennis, Modulation of chemical composi-tion and other parameters of the cell by growth rate, in:F.C. Neidhardt, J.L. Ingraham, K.B. Low, B. Magasanik,M. Schaechter, H.E. Umbarger (Eds.), Escherichia coli andSalmonella typhimurium: Cellular and Molecular Biology, vol.2, American Society for Microbiology, Washington, DC, 1987,pp. 1527–1542.

[119] J. Vitko Jr., D.R. Franz, M. Alper, P.D.E. Biggins, L.D. Brandt,C. Bruckner-Lia, H.A. Burge, R. Ediger, M.A. Hollis, L.L.Laughlin, R.P. Mariella Jr., A.R. McFarland, R.P. Schaudies,Nucleic acid sequence-based identification for detect-to-warnapplications, in: Sensor Systems for Biological Agent Attacks,The National Academies Press, Washington, DC, 2005, pp.84–104.

[120] R. Mei, E. Hubbell, S. Bekiranov, M. Mittmann, F.C. Chris-tians, M.M. Shen, G. Lu, J. Fang, W.M. Liu, T. Ryder, P.Kaplan, D. Kulp, T.A. Webster, Probe selection for high-densityoligonucleotide arrays, Proc. Natl. Acad. Sci. U.S.A. 100 (2003)11237–11242.

[121] E.L. Brodie, T.Z. DeSantis, J.P.M. Parker, I.X. Zubietta, Y.M.Piceno, G.L. Andersen, Urban aerosol harbor diverse and

dynamic bacteria populations, Proc. Natl. Acad. Sci. U.S.A. 104(2007) 299–304.

[122] K.S. Reynolds, C.P. Gerba, I.L. Pepper, Rapid PCR-based mon-itoring of infectious enteroviruses in drinking water, Water Sci.Technol. 35 (11–12) (1997) 423–427.