A DNA microarray facilitates the diagnosis of Bacillus anthracis in environmental samples


Ludwig Eduard Hoelzle, Winterthurerstr. 270, CH-8057 Zürich, Switzerland. E-mail: lhoezle@vetbakt.uzh.ch


Aims:  In order to improve the diagnosis of Bacillus anthracis in environmental samples, we established a DNA microarray based on the ArrayTube technology of Clondiag.

Methods and Results:  Total DNA of a bacterial colony is randomly biotinylated and hybridized to the array. The probes on the array target the virulence genes, the genomic marker gene rpoB, as well as the selective 16S rDNA sequence regions of B. anthracis, of the Bacillus cereus group and of Bacillus subtilis. Eight B. anthracis reference strains were tested and correctly identified. Among the analysed environmental Bacillus isolates, no virulent B. anthracis strain was detected.

Conclusions:  This array clearly differentiates B. anthracis from members of the B. cereus group and other Bacillus species in environmental samples by chromosomal (rpoB) and plasmid markers. Additionally, recognition of B. cereus strains harbouring the toxin genes or atypical B. anthracis strains that have lost the virulence plasmids is feasible.

Significance and Impact of the Study:  The array is applicable to the complex diagnostics for B. anthracis detection in environmental samples. Because of low costs, high security and easy handling, the microarray is applicable to routine diagnostics.


The rapid identification of Bacillus anthracis in complex substrates such as dust and powder samples is a challenging diagnostic task and important for routine laboratories. Bacillus anthracis is difficult to differentiate phenotypically from members of the Bacillus cereus group by culture methods (Turnbull 1999). Further, B. anthracis has a nearly identical 16S rDNA sequence to B. cereus. Therefore, a molecular identification of B. anthracis can be achieved by PCRs targeting the plasmid-encoded virulence genes and specific chromosomal DNA sequences such as Ba813, vrrA and rpoB (Andersen et al. 1996; Jackson et al. 1997; Ramisse et al. 1999; Qi et al. 2001; Patra et al. 2002). This requires time-consuming tests susceptible to cross-contamination. Recently, a variety of molecular assays, e.g. real-time PCR, multiplex PCR and a combination of multiplex PCR with a DNA microarray, were developed to improve the diagnostics (Ellerbrok et al. 2002; Wang et al. 2004; Bavykin et al. 2008).

The objective of the study was to develop a DNA microarray to differentiate B. anthracis from environmental Bacillus isolates obtained from samples released as ‘white powders’ with regard to potential bioterrorism threats. Aforementioned microarrays are based on fluorescence-labelled PCR products. As the labelling reagents are expensive and the development of the techniques is time consuming, such systems are not ideal for routine use in diagnostic laboratories. In this study, the ArrayTube™ (AT) platform (Clondiag, Jena, Germany) and biotinylated whole bacterial DNA were used to diagnose B. anthracis. This platform is rapid and uses inexpensive reagents. Because of the hybridization with whole bacterial DNA instead of PCR products, the cross-contamination risk is minimized. All important genomic and plasmid markers can be analysed in parallel.

Materials and methods

Bacterial strains and growth conditions

Strains used are listed in Table 1. Environmental samples (1 g) were homogenized in 10 ml phosphate buffered saline and pasteurized (65°C, 10 min) to inactivate vegetative cells. Blood-TSA (blood-trypticase-soy-agar; Oxoid, Basel, Switzerland) was used for cultivation. Bacillus anthracis strains were handled under biosafety level 3 conditions.

Table 1.   Strains used in this study. Characterization was performed by PCRs of the virulence genes, 16S gene sequencing and hybridizations to the array
Reference strains or isolates respectivelyAccession number rpoB pag lef cya capAC capBHighest homology with the 16S rRNA gene ofSequence identity in percentResult of ArrayTube hybridizationPhenotypic group
  1. rpoB, gene encoding the subunit β of the RNA polymerase; pag, lef and cya, genes encoding the toxin components protective antigen, lethal factor and oedema factor respectively; capAC and capB, genes encoding the capsule proteins; +, a PCR product could be detected; −, no PCR product could be detected; NP, strain assignment by hybridization signals not predictable; ND, not done; Bold, hybridizations of these strains are shown in Fig. 2b-o.

  2. *Prof. R. Böhm, University of Hohenheim, Germany.

  3. †Prof. J. Frey, Institute of Veterinary Bacteriology, University of Berne, Switzerland.

Bacillus anthracis A15*++++B. anthracis
Bacillus cereus
99·9 99·9B. anthracis; toxin genesND
B. anthracis A73*+++B. anthracis
B. cereus
99·3 99·3B. anthracis; capsule genesND
B. anthracis 12/07/60++++++B. anthracis
B. cereus
B. anthracis; all virulence genesND
B. anthracis 07/11/62++++++B. anthracis
B. cereus
B. anthracis; all virulence genesND
B. anthracis 09/11/61++++++B. anthracis
B. cereus
B. anthracis; all virulence genesND
B. anthracis TV 06/81++++++B. anthracis
B. cereus
B. anthracis; all virulence genesND
B. anthracis SterneGenBank:NC005945++++B. cereus99·2B. anthracis; toxin genesND
B. anthracis A58+B. cereus99·3B. anthracis; no virulence genesND
B. cereus ATCC 14579GenBank:NC004722+B. cereus ATCC 1457999·7B. cereus groupND
Bacillus mycoides ATCC 6462GenBank:AF155956B. anthracis Ames99·0B. cereus groupND
Bacillus thuringiensis ATCC 10792GenBank:AF290545B. cereus
B. thuringiensis
B. cereus groupND
Bacillus subtilis ATCC 6051GenBank:AB042061B. subtilis99·4B. subtilisND
M63/3B. subtilis97·2B. subtilis2
M63/4_2B. subtilis99·6B. subtilis3
M65/1_1Bacillus licheniformis88·9NP5
M71/1_4B. subtilis96·9B. subtilis4
M96/2B. licheniformis97·7NP7
M100+B. anthracis Ames99·1B. cereus group5
M101/1+B. cereus98·1B. anthracis; no virulence genes6
M125/1_1Bacillus clausii61·9genus Bacillus4
M125/2B. subtilis100·0B. subtilis7
M131/1_2B. clausii99·3NP2
M134Bacillus pumilus98·8B. subtilis1
M135/1_4B. clausii95·4NP1
M141/11+B. cereus, B. thuringiensis, B. anthracis96·0B. anthracis; no virulence genes6

Phenotypic characterization of ‘environmental’ bacillus strains

All environmental isolates were checked for the absence of B. anthracis virulence genes by PCR. Gram-staining characteristics (Romeis 1989), haemolysis, motility and penicillin susceptibility were evaluated.

DNA preparation

An inoculation loop was tipped into a bacterial colony, and the attaching cells were lysed at 56°C for 2 h in 400 μl lysis buffer (0·1 mol l−1 Tris–HCl, pH 8·0, 0·05% Tween 20, 0·24 mg ml−1 proteinase K), denatured at 95°C for 10 min and filtrated [0·2 μm (Acrodisc®, Syringe filters; Pall Corporation, Ann Arbor, MI, USA); (Perreten et al. 2005)]. DNA concentrations were determined spectrophotometrically.

PCR and sequencing

16S rDNA, rpoB and B. anthracis virulence genes were amplified by PCR. Oligonucleotides and annealing temperatures are listed in Table 2. PCRs were performed using 100 ng template DNA, the HotStarTaq Master mix (Qiagen, Hombrechtikon, Switzerland) and 0·8 μmol l−1 primers.

Table 2.   Oligonucleotides and annealing temperatures
GeneForward primer sequence (5′–3′)Reverse primer sequence (5′–3′)T, °CReferences
rpoBTTGAAATTTATGAGCGTCTATAAGATTGTTCCTTCTGC45GenBank:NC_003997 forw: 109091–109111, rev: 109303–109320
cyaAGTATTATATCCTTTTCAGTATTATTTTCAATTTCATTATAGGC50GenBank:NC_007322 forw: 122641–122663, rev: 124889–124909
lefATGTAATTAAAAGCTTCCGCTTTTATTTACTAATCAGCTT45GenBank:NC_007322 forw: 150957–150975, rev: 151106–151126
capACAATTTGATTTCCAATTTATCATCATCAGCCCGTATTTATG55GenBank:NC_007323 forw: 53991–54013, rev: 55444–55460

Sequences of the 16S rDNA (MWG-Biotech, Ebersberg, Germany) were aligned with databank entries using fasta and the gcg package software (http://www.bio.uzh.ch/bioc/).

DNA array

Twenty-two probes target the plasmid-encoded virulence genes and five probes cover the rpoB gene of B. anthracis. Fifteen16S rDNA targeting probes were designed on sites with higher variability between the members of the B. cereus group as well as of Bacillus subtilis compared with the corresponding B. anthracis sequence. Target genes of the microarray were selected from the GenBank database or from published primer/probe sequences (Table 3; Fig. 1). Common features are length between 22 and 30 nt, melting temperature between 67 and 69°C and GC content of 35–45%. Probes were custom-spotted by Clondiag onto a glass surface integrated into the AT system. Genetic markers and gene groups are represented by at least two different probes. All probes were spotted twice.

Table 3.   Oligonucleotide probes spotted on the array
IDProbe-nameProbe-sequence (5′–3′) Probe-targetReferenceProbe position
  1. Bold: nucleotide differences from the sequence of B. anthracis Ames Ancestor NC_007530, as indicated in Fig. 1

1capA_1CCAAAACCAGTTGCCAGTGCATTGGCapsule-biosynthesis protein ADang et al. (2001)
2capA_2CATTTACGTGATAATGGTACTGCAATTCCapsule-biosynthesis protein AGenBank:NC_00732354 166–54 139
3capA_3ACGTGATAATGGTACTGCAATTCTTGCapsule-biosynthesis protein AGenBank:NC_00732354 136–54 161
4capA_4GGTACTGCAATTCTTGATGTTGTACCTCapsule-biosynthesis protein AGenBank:NC_00732354 125–54 151
5capB_1GAAGCAGAAGCACTTATTTGTGAATGTACapsule-biosynthesis protein BGenBank:NC_00732356 453–56 480
6capB_2ATATGGATGTTATGGGACCTACACTTGACapsule-biosynthesis protein BGenBank:NC_00732356 338–56 365
7capB_3AAGAGGTTGCAGAAGAGAGAAATACAAACapsule-biosynthesis protein BGenBank:NC_00732356 227–55 254
8capB_4TAGAAGGCTGGTCAACAAGTGAAATTATCapsule-biosynthesis protein BGenBank:NC_00732355 729–55 756
9capC_1CCACGGAATTCAAAAATCTCAAATGGCATCapsule-biosynthesis protein CEllerbrok et al. (2002)
10capC_2GGCAACGCTAATTACAGGTATTTGTTCapsule-biosynthesis protein CGenBank:NC_00732355 326–55 351
11capC_3ATTCCGTGGTATTGGAGTTATTGTTCCapsule-biosynthesis protein CGenBank:NC_00732355 248–55 273
12capC_4GCAAATACAATTCAAAGACAAGGGTTACCapsule-biosynthesis protein CGenBank:NC_00732355 209–55 236
13cya_1TAAATATGAATTGTAGCTGTGTGCCAAGAdenylate-cyclase (oedema factor)GenBank:NC_007322122 376–122 403
14cya_2TACTATTTGCTATATCCTCCTCACAGGCAdenylate-cyclase (oedema factor)GenBank:NC_007322122 663–122 690
15cya_3CACCTGACCATAGAACGGTATTAGAGTTAdenylate-cyclase (oedema factor)GenBank:NC_007322123 356–123 383
16lef_1TTGCATATTATATCGAGCCACAGCATCGTGZinc-endopeptidase (lethal factor)Dang et al. (2001)
17lef_2AATGAGGTACAAGAAGTATTTGCGAAAGzinc-endopeptidase (lethal factor)GenBank:NC_007322151 030–151 057
18lef_3GGTACAAGAAGTATTTGCGAAAGCTTZinc-endopeptidase (lethal factor)GenBank:NC_007322151 027–151 052
19pag_1AAAGGTTACAGGACGGATTGATAAGAATProtective antigenGenBank:NC_007322144 576–144 603
20pag_2CTAGTGAAGTACATGGAAATGCAGAAGTProtective antigenGenBank:NC_007322144 764–144 791
21pag_3GTTCTTTGATATTGGTGGGAGTGTATCTProtective antigenGenBank:NC_007322144 801–144 828
22pag_4AATTGATCATTCACTATCTCTAGCAGGGProtective antigenGenBank:NC_007322144 864–144 891
23rpoB_TTCCAAAGCGCTATGATTTAGCAAATGTRNA-polymerase βQi et al. (2001)
25rpoB_C1ACTTGTGTCTCGTTTCTTCGATCCAAAGCGRNA-polymerase βEllerbrok et al. (2002)
26rpoB_C2TAGGTCGCTACAAGATCAACAAGAAGTRNA-polymerase βGenBank:AE_017334109 190–109 225
27rpoB_T1TAGGTCGCTATAAGATCAACAAGAAGTRNA-polymerase β, mismatchGenBank:AE_017334109 190–109 225
28rrs-low-varTCGTCAGCTCGTGTCGTGAGATGTT16S rDNA low variability regionGenBank:NC_0075301045–1069
29rrs-Banth-high1_2xCGACATCCTCTGACAACCCTAGAGATA16S rDNA high variability regionGenBank:NC_007530976–1001
30rrs-Bthur-high1_ACGACATCCTCTGAAAACCCTAGAGATA16S rDNA high variability regionGenBank:AF_290545986–1011
31rrs-Bsub-high1_CTGACATCCTCTGACAATCCTAGAGATA16S rDNA high variability regionGenBank:AB_0420611002–1027
32rrs-Banth-high2AGGGCTTCTCCTTCGGGAGCAGAGTG16S rDNA high variability regionGenBank:NC_0075301001–1026
33rrs-Bmyc-high2AGGGCTTCCCCTTCGGGGGCAGAGTG16S rDNA high variability regionGenBank:AF_1559561018–1043
34rrs-Bsub-high2AGGACGTCCCCTTCGGGGGCAGAGTG16S rDNA high variability regionGenBank:AB_0420611027–1052
35BsuR1_22AGGTTTTCGGATCGTAAAGCTC16S rDNAGenBank:AB_042061422–443
36BsuR2_22GATCGTAAAGCTCTGTTGTTAG16S rDNAGenBank:AB_042061431–452
37BsuR3a_22GAACAAGTACCGTTCGAATAGG16S rDNAGenBank:AB_042061457–478
38BsuR3b_22GAACAAGTGCCGTTCGAATAGG16S rDNAGenBank:AB_042061457–478
39BsuR4_22TTCGAATAGGGCGGTACCTTGA16S rDNAGenBank:AB_042061469–490
40GneR1_22AGGCCTTCGGGTTGTAAAGTAC16S rDNAGenBank:NC_000913416 596–416 626
41GneR2_22GGTTGTAAAGTACTTTCAGCGG16S rDNAGenBank:NC_000913224 194–224 215
42EcoR3_22GGAAGGGAGTAAAGTTAATACC23S rDNAGenBank:NC_0009134 034 002–4 034 023
43EcoR4_22TTAATACCTTTGCTCATTGACG23S rDNAGenBank:NC_0009134 034 016–4 034 037
44BanR1_22AGGCTTTCGGGTCGTAAAACTC16S rDNAGenBank:NC_00753029 524–29 545
45BanR2_22GGTCGTAAAACTCTGTTGTTAG16S rDNAGenBank:NC_00753029 533–29 554
46BanR3_22GAACAAGTGCTAGTTGAATAAG16S rDNAGenBank:NC_0075309765–9786
47SalR3a_22GGAAGGTGTTGTGGTTAATAAC16S rDNAGenBank:NC_0069054 328 702–4 328 723
48SalR3b_22GGAAGGTGTTGTGGTTAATACC16S rDNA, mismatchGenBank:NC_0069054 328 702–4 328 723
49P13’NH2UnknownInternal hybridization control of Clondiag  
50htag4-Bio wisp2_129_158r BioBiotin   
Figure 1.

 Probe design of the 16S rDNA region. The 16S rRNA gene sequences of the Bacillus cereus group members are nearly homologous. Here, an alignment of these sequences as well as of the corresponding Bacillus subtilis sequence of higher variable regions is shown. *nt positions showing differences between the sequences; Ba Ames Anc: Bacillus anthracis Ames Ancestor NC_007530. Other strains and accession numbers are given in Table 1. The probes are described in Table 3.

DNA biotinylation and microarray hybridization

Genomic DNA was biotinylated and hybridized to the array as described by Perreten et al. (2005). Instead of QMT hybridization buffer, 3× DNA Buffer (Clondiag) was used. Prehybridization (100 μl 1× BSA/1× SSPE; 10 mmol l−1 NaHPO4, 0·18 mol l−1 NaCl, 1 mmol l−1 EDTA, pH 7·4) was performed at 50°C for 15 min.

Results were obtained by reading the microarray in an AT Reader (ATR01; Clondiag) at 25°C taking a picture after 40 min. Analysis was performed by the ionoclust software provided by Clondiag. Hybridization signals were considered to be specific if they appeared in duplicates and if the mean intensity was 0·05 or above.


Phenotypic and genotypic characterization of the Bacillus isolates

A total of 158 environmental samples were submitted to our laboratory for B. anthracis diagnostic. Out of these samples, we obtained 43 Bacillus isolates that were characterized phenotypically. Based on the phenotype, the strains were assigned to seven groups. From each group, two isolates were chosen for array hybridization (group three consisted of only one member).

The 16S rDNA of the 13 environmental Bacillus isolates that were used for hybridization were sequenced (Table 1). Additionally, all isolates were examined using a B. anthracis-specific rpoB PCR (Qi et al. 2001). PCR products were obtained for three of the isolates (M100, M101/1 and M141/11). The PCRs targeting B. anthracis virulence genes were negative for all isolates.

Specificity testing

To evaluate the specificity of the array probes, biotinylated PCR amplicons of the capAC, capB, pag, cya, lef, rpoB and the 16S rRNA gene region of B. anthracis A73 (capsule) and B. anthracis A15 (toxin genes) were hybridized to the AT. Specific hybridization signals were observed in all experiments (Fig. 2a). Hybridization with reference strains (Table 1) revealed the expected probe signals confirming their specificity (Fig. 2b–h). Four B. anthracis field isolates were also hybridized to the array (Fig. 2g). All virulent B. anthracis strains showed positive hybridization signals with spots 1–22. Only probe 18, detecting a region on the lef gene, was negative. Bacillus anthracis strains harbouring only one virulence plasmid could be recognized and differentiated by their hybridization patterns (positive for spots 1–12 and for 13–22 indicating the presence of pXO2 and pXO1 respectively). Strain A58 (cured of both plasmids) differed from B. cereus by its hybridization pattern of the genomic rpoB gene region (spots 23–27).

Figure 2.

 Hybridizations of several DNA samples to the ArrayTube. (a) Biotinylated PCR products of capAC amplified from strain Bacillus anthracis A73, cya, lef and the 16S rRNA gene region amplified from strain B. anthracis A15; (b–n) whole biotinylated DNA content of tested strains: (b) Bacillus cereus; (c) Bacillus thuringiensis; (d) Bacillus subtilis; (e) non-virulent B. anthracis A73 (pX01−; pX02+); (f) non-virulent B. anthracis Sterne (pX01+; pX02−); (g) virulent diagnostic Banthracis isolate 07/11/62; (h) non-virulent B. anthracis A58; (i) M134; (j) M63/3; (k) M63/4_2; (l) M71/1_4; (m) M65/1_1; (n) M141/11 and (o) M96/2.

Hybridization with strains obtained from environmental samples

The discriminatory power of the array for B. anthracis from other members of the genus Bacillus was tested with DNA of strains isolated from environmental samples. Thirteen Bacillus isolates representing seven different groups were tested (Table 1).

All isolates could be differentiated from virulent B. anthracis because no positive hybridization results for any B. anthracis virulence gene probes were detected. Considering the rpoB and 16S rDNA probes, M101/1 and M141/11 could be typed as B. anthracis without plasmid-encoded genes. M63/3, M63/4_2, M71/1_4 and M125/2 were identified as B. subtilis and M100 as B. cereus. This result was confirmed by 16S rDNA sequencing. For the other isolates, no species identification was possible. Figure 2i–o illustrates arrays representing each phenotypic group.


State-of-the-art diagnosis of B. anthracis in environmental samples requires simultaneous detection of chromosomal markers as well as of plasmid-encoded virulence genes. An unambiguous routine diagnosis is time-consuming and sometimes uncertain because Bacillus strains with atypical characteristics have been described recently, i.e. penicillin-resistant B. anthracis strains (Bradaric and Punda-Polic 1992; Lalitha and Thomas 1997) or B. cereus isolates harbouring the virulence plasmids (Hoffmaster et al. 2004). In this study, we developed a microarray for the parallel detection of chromosomal markers (rpoB) and the virulence genes of B. anthracis. Furthermore, 16S rDNA-specific probes were included in the array to distinguish the B. cereus group from other Bacillus species.

Instead of fluorescence labelling of multiple PCR fragments, we biotinylated the entire bacterial DNA. The advantage is that it is not necessary to establish several PCRs or a multiplex PCR and multiple and expensive labelling reactions. This minimizes the required working steps, the total costs as well as the cross-contamination risk in routine application. Using our array, a definite diagnosis is achievable within 12 h.

Specificity of our microarray was proven by hybridization of biotinylated control PCR fragments and whole DNA of well-defined Bacillus strains. Thereby, all reference strains and field isolates of B. anthracis could be identified and unambiguously differentiated from other Bacillus strains. Only probes 18, 45 and 46 did not reveal the expected hybridization signals. However, the application of the whole bacterial DNA instead of specific PCR products as a template allows customizing the probe composition on newly produced microarray batches.

Interestingly, our results indicate that the reference strain B. cereus ATCC 14579 could be distinguished from the B. anthracis strains at the 16S rDNA level. Probe 30, originally designed to differ between Bacillus thuringiensis and B. anthracis, revealed a specific hybridization signal with B. cereus ATCC 14579. No signal was obtained with B. anthracis. It remains to be elucidated whether this is true for other B. cereus strains.

The hybridization results of the environmental Bacillus isolates confirmed the discriminatory power of our array, i.e. all isolates were clearly diagnosed as non-B. anthracis or avirulent B. anthracis. The 16S rDNA sequence of the strain M100 showed the highest identity with the B. anthracis strain Ames. The DNA did not hybridize with probes 24 and 26 on the array. Thus, it was identified as B. cereus; not as B. anthracis. The isolates M101/1 and M141/11 were assigned to B. cereus and the clusters B.  cereus, B. thuringiensis and B. anthracis by 16S rDNA sequencing. By hybridization, both strains could be identified as B. anthracis without virulence genes because of high similarity to the rpoB hybridization pattern of B. anthracis A58. Therefore, the array has the potential to recognize B. anthracis strains cured of the virulence plasmids that would be overseen by 16S rDNA sequencing. Such strains are of epidemiological significance (Bode et al. 2004) and important to be recognized for bio-warfare prevention.

We conclude that our array is a reliable tool for the diagnosis of B. anthracis in environmental samples. It is possible to clearly distinguish all constitutions of B. anthracis (fully virulent, harbouring only one plasmid or without plasmids). Further, B. cereus harbouring the virulence plasmids would be detected. It is applicable in routine diagnostics because of its safety and simple handling. The cross-contamination risk and therefore false-positive results are minimized. Furthermore, the application of the entire bacterial DNA as a template enables the adaption of the array layout to novel findings in B. anthracis research.


This study was funded by the canton of Zurich, Amt für Abfall, Energie und Luft (AWEL), Switzerland. We thank Vincent Perreten, Institute of Veterinary Bacteriology, University of Berne, Switzerland, for the fruitful discussions concerning the biotinylation of the sample DNA. The strains B. anthracis A15 and B. anthracis A73 were kindly provided by Prof. R. Böhm, University of Hohenheim, Germany; the strains B. anthracis Sterne and B. anthracis A58 were obtained from Prof. J. Frey, Institute of Veterinary Bacteriology, University of Berne, Switzerland.