Open Access

Genotyping of Bacillus anthracis strains based on automated capillary 25-loci Multiple Locus Variable-Number Tandem Repeats Analysis

  • Florigio Lista1, 2Email author,
  • Giovanni Faggioni1,
  • Samina Valjevac3, 4,
  • Andrea Ciammaruconi1,
  • Josée Vaissaire5,
  • Claudine le Doujet5,
  • Olivier Gorgé3, 4,
  • Riccardo De Santis1,
  • Alessandra Carattoli6,
  • Alessandra Ciervo6,
  • Antonio Fasanella7,
  • Francesco Orsini1,
  • Raffaele D'Amelio2, 8,
  • Christine Pourcel4,
  • Antonio Cassone6 and
  • Gilles Vergnaud3, 4
BMC Microbiology20066:33

https://doi.org/10.1186/1471-2180-6-33

Received: 14 October 2005

Accepted: 06 April 2006

Published: 06 April 2006

Abstract

Background

The genome of Bacillus anthracis, the etiological agent of anthrax, is highly monomorphic which makes differentiation between strains difficult. A Multiple Locus Variable-number tandem repeats (VNTR) Analysis (MLVA) assay based on 20 markers was previously described. It has considerable discrimination power, reproducibility, and low cost, especially since the markers proposed can be typed by agarose-gel electrophoresis. However in an emergency situation, faster genotyping and access to representative databases is necessary.

Results

Genotyping of B. anthracis reference strains and isolates from France and Italy was done using a 25 loci MLVA assay combining 21 previously described loci and 4 new ones. DNA was amplified in 4 multiplex PCR reactions and the length of the resulting 25 amplicons was estimated by automated capillary electrophoresis. The results were reproducible and the data were consistent with other gel based methods once differences in mobility patterns were taken into account. Some alleles previously unresolved by agarose gel electrophoresis could be resolved by capillary electrophoresis, thus further increasing the assay resolution. One particular locus, Bams30, is the result of a recombination between a 27 bp tandem repeat and a 9 bp tandem repeat. The analysis of the array illustrates the evolution process of tandem repeats.

Conclusion

In a crisis situation of suspected bioterrorism, standardization, speed and accuracy, together with the availability of reference typing data are important issues, as illustrated by the 2001 anthrax letters event. In this report we describe an upgrade of the previously published MLVA method for genotyping of B. anthracis and apply the method to the typing of French and Italian B. anthracis strain collections. The increased number of markers studied compared to reports using only 8 loci greatly improves the discrimination power of the technique. An Italian strain belonging to the B branch was described, and two new branches, D and E, are proposed. Owing to the upgrading achieved here, precise genotyping can now be produced either by automated capillary electrophoresis, or by the more accessible but slower and for some markers slightly less accurate agarose gel methodology.

Background

Bacillus anthracis, a spore forming Gram positive bacteria, is the etiological agent of anthrax, a zoonosis with a worldwide distribution. The disease can be transmitted to humans by contact with infected animals or contaminated animal products. In addition to being an occupational disease, limited to farmers and veterinarians, anthrax has received considerable attention and is one of the most feared microorganism with respect to bioterrorism. In 2001, letters containing spores were mailed causing the death of five persons and several cases of cutaneous anthrax [1]. B. anthracis is a member of the Bacillus cereus group, containing Bacillus cereus, Bacillus thuringiensis, Bacillus mycoides and B. anthracis [2]. B. anthracis is characterized by an extremely low genetic variability, making strains differentiation very challenging [3, 4]. The importance of strain differentiation of biothreat agents including B. anthracis is increasingly recognized as a way for identifying the source of the attack, as illustrated not only by the 2001 events but also, earlier on, by the Sverdlovsk [5] and Tokyo [6] events. Such events together suggested how crucial is the development of microbial forensics for biosecurity. Measures aimed at limiting the risk of deliberate release of dangerous pathogens, would require that isolates kept in different institutions around the world are precisely genotyped, and that the genotype profiles are shared by all countries having accepted to follow these rules, for strain accountability purposes.

Because the molecular methods classically used to differentiate between strains in other species failed to discriminate B. anthracis isolates, new approaches were developed. Using Amplified Fragment Length Polymorphism in 79 B. anthracis isolates, 31 polymorphic chromosomal regions were observed out of 1,000 fragments, and eventually, the most interesting polymorphisms turned out to be the result of tandem repeat variations [7]. Tandem repeats variability was studied most extensively in eukaryotes, and was shown to be associated with different aspects of DNA replication and recombination, including replication slippage and double-strand breaks repair [8]. They do not vary by the same mechanisms which are inducing point mutations, and the mutation rate at some Variable Number of Tandem Repeats (VNTR) loci may be considerably higher than nucleotide substitutions [911] in an as yet not predictable way [12, 13]. The level of intraspecific polymorphism in tandemly repeated sequences varies from one locus to another and needs to be experimentally measured by typing representative strain collections. Approaches based on VNTR analysis (called MLVA for Multiple Locus VNTR Analysis) are now increasingly used to characterize in particular recently emerged, highly monomorphic pathogens including B. anthracis [12, 14], Mycobacterium tuberculosis [15], Yersinia pestis [16]Brucella [17] (see [18, 19] for a review). Multiple alleles can exist in the population for each tandem repeat locus. One advantage of MLVA typing as compared for instance to SNP (Single Nucleotide Polymorphism) typing is that a new isolate can be compared to previously characterised strains in an unbiased way. One weakness is that the phylogenetic value of MLVA clustering must be carefully checked and validated by using independent approaches with less homoplastic markers [20]. In previous reports, a multilocus VNTR analysis (MLVA8) method was proposed for genotyping of B. anthracis strains, using six chromosomal and two plasmid marker loci (vrrA, vrrB1, vrrB2, vrrC1, vrrC2, CG3, pXO1, pXO2 [14]). In the initial work, an automated fluorescent DNA sequencer was used to determine the size of the PCR fragments and this set of markers is now largely employed for B. anthracis genotyping [2126]. However, this MLVA8 assay is not sufficient for molecular forensics approaches, the resolution achieved is too limited. For instance the vast majority of French and Italian strains have been assigned to MLVA8 genotypes (GT) 1 and 3 which differ only at one marker, located on one of the two virulence plasmids [14, 22, 26]. For this reason, additional markers were subsequently developed to produce an MLVA15 panel [11], and the initial clustering nomenclature [14] was revised [20]. Unfortunately none of the corresponding data has been published so far. Le Flèche et al. [12] extended the MLVA8 assay by proposing 14 additional markers (Bams1, 3, 5, 7, 13, 15, 21, 22, 23, 24, 25, 28, 30, 31), which greatly improved the resolution power of the assay. DNA fragment sizing was achieved on standard agarose gels stained by ethidium bromide, a technique utilising widely accessible, and very basic, equipment.

In order to have a more rapid and accurate genotyping system for B. anthracis we propose an automated capillary based method using essentially the panel of loci described by Le Flèche et al [12], with the addition of 4 new markers, and apply the method to collections of strains from Italy and France. We compare the clustering achieved with available epidemiological data for B. anthracis, and introduce two new B. anthracis groups, D and E. The present report deals with the standardization and further improvement of currently available genotyping methods for B. anthracis, the typing of a larger collection of strains, the comparison with available published data, and the enlargement of the publicly accessible databases.

Results

MLVA25 genotyping by capillary electrophoresis

A collection of 160 B. anthracis isolates, comprising strains from Italy and France and a few reference strains were analyzed to generate MLVA typing data. DNA was amplified in 4 multiplex PCR reactions, comprising 25 loci. The loci are 21 markers described previously [12, 14] (Bams7 is not currently amenable to capillary electrophoresis typing because of its allele size range and was not included), and the four new markers Bams34, 44, 51, 53 (Table 1). All twenty-five markers can be amplified in only four PCR reactions by taking into account the allele size range for each locus (Table 2). The labelled amplicons were separated on a CEQ 8000 Beckmann DNA analysis system and the electropherograms were analysed by CEQ Fragment Analysis System software to determine the length of the fragments by reference to a 75–1000 bps size ladder (in red on Figure 1). After testing several primer concentrations and ratios, a balanced level of fluorescence peak intensity was obtained by using between 5 to 10 ng of template DNA per multiplex PCR reaction and the concentration of primers indicated (Table 2).
Table 1

Comparison between MLVA8, MLVA20, MLVA25

MLVA8

MLVA20

MLVA25

pXO1

 

pXO1

pXO2

 

pXO2

vrrA

vrrA

vrrA

vrrB1

vrrB1

vrrB1

vrrB2

vrrB2

vrrB2

vrrC1

vrrC1

vrrC1

vrrC2

vrrC2

vrrC2

CG3

CG3

CG3

 

bams1

bams1

 

bams3

bams3

 

bams5

bams5

 

bams13

bams13

 

bams15

bams15

 

bams21

bams21

 

bams22

bams22

 

bams23

bams23

 

bams24

bams24

 

bams25

bams25

 

bams28

bams28

 

bams30

bams30

 

bams31

bams31

  

bams34

  

bams44

  

bams51

  

bams53

 

bams7

 
Table 2

Composition of the 4 multiplex PCR according to the size range and the different dye label compatibility.

 

Locus

Repeat units

Fragment size range (bps)

Number of alleles

New alleles

HDGI index (total n = 67)*

Primer conc. (μM)

Multiplex A

CG3 a

5

153–158

2

 

0.50

0.30

 

bams44 a

39

339–417

2

 

0.21

0.30

 

bams3 a

15

474–609

7

1

0.51

0.30

 

vrrB 2 c

9

135–171

4

 

0.27

0.15

 

bams5 c

39

229–385

5

2

0.49

0.30

 

bams15 c

9

409–616

10

5

0.62

0.60

 

bams1 b

21

380–485

5

 

0.54

0.15

 

vrrC1 b

9

517–616

4

 

0.42

0.30

Multiplex B

bams13 a

39

337–868

13

4

0.61

0.30

 

vrrB1 b

9

193–256

4

 

0.36

0,10

 

bams28 b

24

373–493

4

1

0.35

0.15

 

vrrC2 b

18

532–604

3

 

0.40

0.10

 

bams53 c

12

212–236

2

 

0.21

0.60

 

bams31 c

9

331–952

8

4

0.58

0.60

Multiplex C

vrrA a

12

290–326

4

 

0.40

0.06

 

bams25 a

15

376–391

2

 

0.11

0.15

 

bams21 a

45

541–676

3

 

0.09

0.30

 

bams34 c

39

238–581

5

 

0.40

0.30

 

bams24 b

42

469–637

6

3

0.27

0.30

Multiplex D

bams51 b

45

358–538

4

 

0.33

0.60

 

bams22 b

36

555–735

5

1

0.34

0.15

 

bams23 c

42

399–651

5

2

0.40

0.30

 

bams30 a

9

268–916

12

4

0.71

0.30

 

pXO1 b

3

123–135

5

 

0.77

0.15

 

pXO2 a

2

132–145

8

 

0.58

0.15

The number of diverse allele and new allele are described for the 25 markers. Primer concentration for each set of marker in the multiplex is also reported.

* HGDI (Hunter-Gaston Diversity Index) calculated as in [36]

The alleles are identified with the number of repeats versus the former nomenclature [12]. Locus a , locus b , locus c are loci amplified by a forward primer respectively labeled with cy5.5, cy5 and WellRed dyeD2.

Figure 1

Representative electropherograms after capillary electrophoresis separation of the fragments amplified by 4 multiplex PCR. B. anthracis alleles are showed in blue, green and black peaks, whereas the red peaks are for the molecular weight size standard by Microzone.

The assay was highly reproducible as tested by performing duplicate typings. The intra-experimental (within the same run) as well as the inter-experimental (between runs) variation of fragment sizing showed variations lower than one base over the whole range of PCR fragments investigated here. The accuracy of the data was determined by comparison of observed PCR fragment size to the exact values calculated by direct sequencing and available genome sequence data [27]. In some instances the amplicon size determined after electrophoresis separation did not match the expected values (Table 3). This is well illustrated for instance with Bams30 (Table 3) for which the offset increases with allele size, i.e. repeat copy number. In any case, the offset is conserved, reproducible, and the differences in calculated allele size are consistent with a 9 bp repeat unit variation.
Table 3

Comparison between product sizes inferred by CEQ Fragment Analysis System software (observed) and the actual sizes obtained by sequencing data coming from direct sequencing of the PCR product (expected) or available in Genbank.

Marker

Allele fragment sizes

Alleles

Note

 

Expected

Observed

Number of repeats

as reported [13]

 

BAMS1 21 bp

380

386

11

1

 
 

401

407

12

2

 
 

422

430

13

3

 
 

443

451

14

4

 
 

485

494

16

5

 

BAMS3 15 bp

474

475

21

1

 
 

519

520

24

2

 
 

549

550

26

3

 
 

564

565

27

4

 
 

579

580

28

5

new allele, previously defined as 5 or 6

 

594

595

29

  
 

609

611

30

6

 

BAMS5

229

231

3

 

new allele

39 bp

307

305

5

1

 
 

332

329

5.5

 

new allele

 

346

343

6

2

 
 

385

382

7

3

 

BAMS13 9 bp

337

339

17

1

 
 

346

348

18

 

new allele

 

364

365

20

2

 
 

373

372

21

 

new allele

 

391

393

24

3

 
 

427

427

27

4

 
 

454

456

30

5

 
 

463

465

31

 

new allele

 

481

483

33

 

new allele

 

778

785

66

7

 
 

814

821

70

8

 
 

841

849

73

9

 
 

868

876

76

10

 

BAMS15 9 bp

409

412

23

 

new allele

 

418

422

24

1

 
 

445

449

27

 

new allele

 

535

541

37

2

 
 

571

579

41

3

 
 

580

588

42

 

new allele, previously defined as 3

 

589

597

43

4

 
 

598

606

44

 

new allele, previously defined as 4 or 5

 

607

616

45

5

 
 

616

627

46

 

new allele

BAMS21 45 bp

541

535

7

1

 
 

631

625

9

2

 
 

676

669

10

3

 

BAMS22 36 bp

555

547

11

1

 
 

627

619

13

2

 
 

663

657

14

 

new allele

 

699

690

15

3

 
 

735

726

16

4

 

BAMS23 42 bp

399

393

5

 

new allele

 

567

559

9

 

new allele

 

609

600

10

1

 
 

639

629

10.5

2

 
 

651

641

11

3

 

BAMS24 42 bp

469

474

8

3

 
 

511

519

9

 

new allele

 

553

562

10

 

new allele

 

582

593

10.5

 

new allele

 

595

603

11

5

 
 

637

645

12

6

 

BAMS25 15 bp

376

380

12

1

 
 

391

395

13

2

 

BAMS28 24 bp

373

378

9

 

new allele

 

397

405

10

1

 
 

469

478

13

2

 
 

493

502

14

3

 

BAMS30 9 bp

268

270

6

1

 
 

367

369

17

2

 
 

493

496

31

3

 
 

673

681

51

4

 
 

691

698

53

5

 
 

727

735

57

6

 
 

754

763

60

7

 
 

835

848

69

 

new allele

 

853

865

71

 

new allele

 

862

874

72

 

new allele

 

889

902

75

8

 
 

916

929

78

 

new allele

BAMS31 9 bp

331

332

15

1

 
 

637

642

49

 

new allele

 

691

697

55

 

new allele

 

700

705

56

 

new allele

 

727

728

59

2

 
 

772

775

64

3

 
 

781

783

65

 

new allele, previously defined as 3

 

952

955

84

4

 

BAMS34

238

235

3

  

39 bp

386

378

7

  
 

425

421

8

  
 

503

498

9

  
 

581

576

11

  

BAMS44 39 bp

339

336

6

  
 

417

413

8

  

BAMS51 45 bp

358

364

6

  
 

448

452

8

  
 

493

501

9

  
 

538

548

10

  

BAMS53 12 bp

212

213

6

  
 

236

237

8

  

vrrA 12 bp

290

289

8

1

 
 

302

301

9

2

 
 

314

314

10

3

 
 

326

325

11

4

 

vrrB1 9 bp

193

194

12

2

 
 

220

221

15

3

 
 

229

229

16

4

 
 

256

256

19

5

 

vrrB2 9 bp

135

137

4

1

 
 

153

153

6

2

 
 

162

162

7

3

 
 

171

170

8

4

 

vrrC1 9 bp

517

512

46

3

 
 

535

530

48

4

 
 

583

574

53

5

 
 

616

611

57

6

 

vrrC2 18 bp

532

528

17

1

 
 

568

565

19

2

 
 

604

601

21

3

 

CG3 5 bp

153

155

1

1

 
 

158

160

2

2

 

pXO1 3 bp

123

128

6

  
 

126

131

7

  
 

129

134

8

  
 

132

137

9

  
 

135

140

10

  

pXO2 2 bp

132

136

5.5

  
 

133

137

6

  
 

135

139

7

  
 

137

141

8

  
 

139

143

9

  
 

141

145

10

  
 

143

147

11

  
 

145

149

12

  
Because Bams30 was initially identified as a 27 bp repeat unit marker [12], we investigated representative Bams30 alleles by sequencing (EMBL accession numbers AM182499 to AM182509) in order to clarify this discrepancy. Sequence data indicates that Bams30 alleles are composed of 27 base-pair units at one end of the array, and 9 bp units at the other end. Table 4A provides the allele codes of Bams30 alleles. Repeat units are not identical, different letters are used to code the different variant units which are observed [28]. Uppercase letters code for 27-base-pair repeat units whereas lowercase letters code for 9 base-pair repeat unit. The correspondence between letters and repeat units is indicated in Table 4B. The two regions are variable, and as a result the locus behaves like a 9 bp repeat unit tandem repeat.
Table 4

Bams30 allele codes

Table 4A

     

Genotype

cluster

Bams30 allele

PCR fragment sizea

observed sizea

allele codingbc

1

A1a

75

889

906 (+17)

ABCD EEFGGHGGG IG GJ abcdefcdghdghdefciefjkgk

42

A3a

75

889

906 (+17)

ABCD EEFGGHGGG IG GJ abcdefcdghdghdefciefjkgk

41

A3a

75

889

906 (+17)

ABCD EEFGGHGGG IG GJ abcdefcdghdghdefciefjkgk

33

A1a

75

889

906 (+17)

ABCD EEFGGHGGG IG GJ abcdefcdghdghdefciefjkgk

55

D

75

889

906 (+17)

ABCD EEFGGHGGG IGLGJ abcdefcdghd efciefjkgk

34

A1a

72

862

874 (+12)

ABCD EEFGGHGGG IG J abcdefcdghdghdefciefjkgk

37

A1a

72

862

874 (+12)

ABCD EEFGGHGGG IG J abcdefcdghdghdefciefjkgk

4

A1a

72

862

874 (+12)

ABCD EEFGGHGGG IG J abcdefcdghdghdefciefjkgk

54

A4

71

853

865 (+12)

ABCD EEFGMHGGG I LGJ ab cdghdghdefciefjkgk

67

C

71

853

ND

ABCD EEFGGHGGGGGI LGJ abcdef ciefjkgk

5

A1a

69

835

848 (+13)

ABC EFGGHGGG IG GJ abcdefcdghdghdefciefjkgk

58

E

60

754

763 (+9)

ABCD EEFGG IG IJ abcdefcdghd efciefjkgk

57

E

60

754

763 (+9)

ABCD EEFGG IG IJ abcdefcdghd efciefjkgk

51

A3b

57

727

735 (+8)

ABCD EEFGGHG J abcdefcdghd efciefjkgk

53

A3b

57

727

ND

ABCD EEFGGHG J abcdefcdghd efciefjkgk

52

A3b

53

691

698 (+7)

ABCD EEFGGHG J abcdefcdghd efjkgk

46

A3b

51

673

680 (+7)

ABCDEEEFG J abcdefcdghd efciefjkgk

48

A3b

51

673

681 (+7)

ABCDEEEFG J abcdefcdghd efciefjkgk

49

A3b

31

493

496 (+3)

ABCD EFGGH lj gk

63

B2

17

367

369 (+2)

AB cdghd efjkgk

64

B2

17

367

369 (+2)

AB cdghd efjkgk

66

B2

17

367

369 (+2)

AB cdghd efjkgk

62

B2

17

367

369 (+2)

AB cdghd efjkgk

59

B1

6

268

270 (+2)

AK

CEB9727

thuring.

   

ABCDEBGIIIGGGIBJ and ff**ff***fgdgfcafgh**abcdfabciefcfgfjka*

Table 4b

     

27 bp repeat unit code

 

9 bp repeat unit code

   

A

CAACAGGAATAACGGGAGCAACCGGTG

a

CAACGGGAG

  

B

CAACAGGAATAACGGGAGCAACAGGTC

b

ACACAGGTC

  

C

CGACGGGAACAACAGGAGCGACCGGTG

c

TAGCAGGAG

  

D

CGACAGGAATAACGGGAGTGACCGGTG

d

CAACGGGCC

  

E

CAACAGGAATAACGGGAGTGACCGGTG

e

CGACAGGAG

  

F

CAACAGGAATAACGGGAGTGACAGGTC

f

CAACAGGTC

  

G

CGACGGGAATAACGGGAGCAACAGGTC

g

CAACAGGAG

  

H

CGGCGGGAATAACGGGAGTGACAGGTC

h

ATACTGGAG

  

I

CGGCGGGAATAACGGGAGCAACAGGTC

i

CAACAGGCC

  

J

CGACGGGAACAACAGGAGTAACCGGTC

j

TAACGGGAG

  

K

CAACAGGAATAACGGGAGCAACAGGTG

k

CAACTGGTG

  

L

CGACGGGAATAACGGGAGTAACCGGTC

l

CGACGGGAA

  

M

CGACGGGAATAACGGGAGTAACAGGTC

    

a the PCR fragm ent size is the exact size, as deduced from sequence data.

The observed size is as m easured by capillary electrophoresis (the offset is indicated, and is proportional to allele size)

b Upercase letters: 27 bp repeat units; Lowercase letters: 9 bp repeat units

c "*" represent 9 base-pairs repeat units with a sequence m otif not observed in the B. anthracis alleles

To better understand the origin of this behaviour, we investigated the structure of this locus in the closely related B. cereus-B. thuringiensis group by taking advantage of available genome sequences. Figure 2 shows the organization of the locus in the genomes of B. anthracis (accession number AE016879), B. cereus ATCC14579 (AE016877), B. cereus G9241 [29], B. cereus ATCC10987 (AE017194), B. cereus E33L (NC_006274), and B. thuringiensis CEB97/027 (AE017355) also called Bt9727. The gene bearing Bams30 in B. anthracis (Ba2450) possess a collagen-like helix repeat and is flanked by two genes Ba2449 and Ba2451. Their homologues in the other strains were depicted using the same code. The organization in B. cereus G9241 is similar to B. anthracis, but in all four other strains, the situation is less simple. In B. cereus ATCC 14579, two genes possessing a collagen-like helix repeat protein are observed, instead of one in B. anthracis: Both genes possess a tandem repeat with similarity to Bams30, encoding the collagen-like repeat. Whereas in ATCC 10987, the two tandem repeats have nine base-pair repeat units, Bt9727-2241 contains a 27 bp repeat unit presumably derived from an ancestral 9 bp unit. Bt9727 is by far the closest B. anthracis neighbor among the 5 sequenced strains [30]. The composite organization of the Bams30 tandem repeat in B. anthracis can be explained by a rearrangement occurring in a Bt9727-type Bams30 locus. Strain "A1055", which is closest to the last common B. anthracis ancestor [20] shows the same rearrangement, which strengthens the validity of Bams30 as an interesting complementary tool for the identification of B. anthracis (the presence of the virulence plasmids, pXO1 and pXO2 is not sufficient since for instance "A1055" is lacking pXO1 [20]) even if the rearrangement is not strictly limited to B. anthracis as shown by the G9241 strain situation.
Figure 2

Organization of Bams30 locus in: (B.ant. Ames), B. anthracis strain Ames; (Bc ATCC 14579), B. cereus 14579; (Bc G9241), B. cereus G9241 and B. thuringiensis 97/027. The position of Bams30 primers is indicated by small black arrows; the primers localized in the NH2 terminus conserved part of the Exs H gene are indicated by small gray arrows. Tandem repeats with similarity to Bams30 are represented with striped boxes. Homologous genes are depicted using the same code.

Correspondance with agarose gel assignments

For reasons including accessibility to the different technologies, it is important to ensure that allele assignments deduced from agarose gel or capillary electrophoresis are compatible, so that the different methods can be used. For this purpose, allele sizes (expressed in basepairs, bps and converted to the number of repeat units, U) previously estimated by agarose gel were re-examined according to the capillary fragment analysis and available sequence data. For Bams3 (15 bp repeat unit) one allele, previously estimated at 609 bps (30 U) by agarose gel analysis, was corrected to 594 bps corresponding to 29 U, by capillary electrophoresis. Bams15 turned out to vary like a 9-mer rather than 18-mer repeat unit and two alleles were resolved (580 bps for 42 9-base pair units (42 U) and 598 bps for 44 U, previously grouped with 571 bps (41 U), 589 bps (43 U) or 607 bps (45 U). For Bams31 (9 bp repeat unit) one allele was resolved by capillary electrophoresis (781 bps for 65 U, previously grouped with the 772 bps for 64 U).

For markers included in the MLVA8 assay [14] which could be typed on agarose (i.e. with the exception of pXO1 and pXO2) the data obtained here were in agreement with results generated by agarose gel electrophoresis [12]. Published MLVA8 [14, 2126] data contains some allele size assignments which do not correspond to the actual size. This is due to clerical errors in the initial report [14], as noticed previously [24] and data for some loci must be converted as described in the material and methods section in order to be fully compatible.

Four new markers have been included. For Bams34, a 39-mer repeat marker, five alleles were described ranging from 238 to 581 bps (ranging from 4 to 13 repeat units). Bams44 (39-mer) and Bams53 (12-mer) show two alleles (respectively 339/417 bps or 8 and 12 repeat units, and 212/236 bps (6 and 8 repeat units)). A third allele is predicted from genome sequence data for the "Western" strain, but was not experimentally observed among the collection investigated here. For Bams51 (45-mer) four alleles were observed ranging from 338 to 538 bps (6 to 10 units).

Data analysis

When the eight markers comprising the MLVA8 assay are used, 30 different genotypes are resolved among the 160 isolates. With only few exceptions, these genotypes correspond to previously published profiles. Most exceptions are minor variations of previous types. The vast majority of French and Italian isolates belong to GT1 and GT3 within cluster A1a [14] in agreement with previous reports [22, 26]. When the whole MLVA25 dataset is used, 67 genotypes are resolved. Isolates which are not resolved in this analysis have the same epidemiological history (when known). The clustering analysis is shown in Figure 3. Although MLVA is not by its very nature a tool with a high phylogenetic value per se, especially when the available strains do not fully cover the species diversity, the clustering achieved is in excellent agreement with the known relationship between the ancestral C type (genotype 67), the B cluster (genotypes 59 to 66, with the two well-separated subgroups, B1 and B2), and the A cluster (represented on one end by the "Ames" strain, initially assigned to A3b and more recently to A2 [20], and on the other end by the "Western" strain, initially assigned to A1a and more recently to A1 [20]). More precisely, the "Vollum" strain (genotype 54; strain "9" in [20]) and the "Australia94" strain (genotype 45; presumably strain "20" in [20]) are in the expected position with respect to the A group.
Figure 3

A Clustering analysis of genotyping data. The 67 genotypes are numbered from 1 to 67 (left-most column). The origin of the strains (when known) is indicated; ND, unknown. "Cluster" refers to the clustering and "group" refers to the genotype (GT) number based upon MLVA8 as proposed in [14]. Genotypes with no group assigned are usually minor variations of known, neighbouring groups. A number of reference strains are indicated in ().

Genotypes 55–56, and 57–58 are loosely connected to the others, and are suggested by clustering analysis as being intermediates between the B cluster and the A cluster. This position is reminiscent of the position occupied by strains "6", "7" and "8" investigated by Pearson et al. [20]. It may be of interest to note that "6" which branched out in an ancestral position compared to the whole "A" cluster in the SNP analysis [20], originates from Zambia, whereas genotypes 57 and 58 in Figure 3 are from Cameroon. We propose to call these very distinct MLVA groups types D and E (Figure 3).

Discussion

Bacillus anthracis genotyping is essential for epidemiological studies and for biosecurity related (including microbial forensics) issues [21]. Current methodologies such as MLVA8 are not sufficient to discriminate two geographically close isolates in natural outbreaks, since for instance most French and Italian isolates will be classified as GT1 or 3 [22, 26]. Molecular characterization needs to be as precise as possible in order to discriminate isolates. The rational of this work was to update the MLVA method for quick (single-day analysis) and accurate discrimination of Bacillus anthracis isolates. The markers are stable, this is indirectly illustrated by the typing of "Sterne" strains from different origins. Genotypes 46 and 48 correspond to respectively two and five "Sterne" strains. Genotype 46 are the CIP (Collection Institut Pasteur) reference strains 7700 and 7702, genotype 48 covers "Sterne" vaccine strains from different countries, and the sequenced genome. The only difference in the two genotypes is for locus Bams28 [see additional file 1].

The number of VNTR loci was increased from 22 to 25 by addition of four new markers (and removal of Bams7, because allele sizes up to 2 kilobases are not adapted to typing on current capillary machines). The overall assay was converted from 25 monoplex PCR reactions, run on agarose gels and checked for size by eye or by using gel image analysis software, to 4 multiplex PCR analysed by automated capillary based fluorescent electrophoresis and internal size standard measurement for allele calculation. For most loci, accurate size estimate is achieved directly by the capillary electrophoresis machine. In other instances however, the sizing proposed by the machine, although quite reproducible, does not correspond to the real size. This could be due to the nature of the gel matrix, or to differences during migration between labelled PCR fragments and the ladder used, but most probably by a slightly biased flanking sequence or repeat unit-specific mobility pattern. Usually, the size estimate is shifted by a constant value (offset) but in some instances, the offset increases with increasing allele size (increasing repeat copy number). For this reason, allele numbering must be done carefully, using a correspondence table as presented here in Table 3. If different equipments, or reagents, are used, the correspondence indicated here must be checked by typing a number of reference strains (such as the widely available "Sterne" vaccine strain), before alleles can be confidently called and data can be merged.

Size measurement from multiplexed, automated PCR products allows for single-day isolate genotyping which is of importance in a crisis situation. It also provides a slightly increased discrimination power as compared to the MLVA20 assay run on agarose gels for the same analysed markers. This difference is essentially due to the lack of resolution of agarose for the larger, 9 bp repeat unit loci, such as Bams30 and for Bams3, Bams13, Bams15 and Bams31 typing, where new alleles were identified. For Bams15 and Bams30, capillary resolution was able to show a 9 bps repeat unit length variation instead of the respectively 18 bps and 27 unit length proposed initially.

The present investigation suggests that a recombination event at the Bams30 locus may have occurred prior to the emergence of B. anthracis. The event directly affects genes coding for constituents of the exosporium. Bc2381 is related to Bcl A, and Bc2382 corresponds to the exosporium H gene exs H [31]. In B. cereus ATCC 10987, B. thuringiensis 97/027 (Bt9727) and B. cereus genome E33L, an additional ORF is located within the region. The Bams30-containing gene in B. anthracis is the result of a recombination between Bt9727-2241 (coding for the Bcl A-like protein) and Bt9727-2244 (exs H). The beginning and the end of the protein sequence coded by Ba2450 are respectively identical to the beginning of Bt9727-2241 and to the end of Bt9727-2244. Interestingly, a similar rearrangement is observed independently in the B. cereus strain G9241 which caused an anthrax like disease in a human patient [29]. As a result of this fusion, the first part of the tandem repeat itself contains a 27 bp repeat unit whereas the second part is a 9 bp repeat tandem repeat. It may be of interest to observe that two of the most polymorphic tandem repeats in B. anthracis, Bams13 and Bams30, are associated with genes coding for components of the exosporium (Bams13 is part of the Bcl A gene [32]).

Conclusion

In this report we propose an improved automated genotyping system of B. anthracis by MLVA, extending the range of VNTR containing markers to 25, which increases the resolution power as compared to previous methods. The MLVA25 assay was used to genotype Italian and French isolates in a very fast and highly reproducible manner. We also better defined previously characterized alleles and describe new allelic variants. MLVA25 is a very accurate and discriminatory genotyping method and could be a promising technique for future epidemiological studies as well as differentiation of deliberated vs. naturally occurring outbreaks. The improved assay and resulting data remains compatible with the initial agarose gel based assay. The genotyping data can be queried on the web page MLVA web-service [33]. This site allows the comparison of local typing data to the data generated in this report.

Methods

Bacterial strains and isolates

The strains and DNA samples are from the collection maintained by the French Ministry of Defence at Centre d'Etude du Bouchet (CEB), from Istituto Superiore di Sanità (ISS) Italian collections and from the Italian Reference Center for Anthrax (ISZ, Foggia). Total genomic DNA was extracted as previously reported [30].

VNTR amplification and analysis

Twenty-five loci were used, 21 from previous studies [12, 14] and 4 new ones identified by comparing B. anthracis genome sequences using the site GPMS, Genomes, Polymorphism and Minisatellites [34] as described by Denoeud et al. [13]. Multiplex PCR reactions were prepared as follows: 5–10 ng DNA were amplified in a final volume of 15 μl containing 1× PCR Roche reaction buffer (10 mM Tris-HCl, 1,5 mM MgCl2, 50 mM KCl pH 8.3), 0.2 mM dNTPs (Amersham-PharmaciaTM), the appropriate concentrations of each primer (one of which was fluorescently labelled) as reported in Table 2, and 1 U Taq polymerase (Roche). The four multiplex PCR amplifications were performed on Peltier Thermal Cycler (PTC) DNA Engine DYAD or PTC200 (MJ Research) with a starting denaturation step at 96°C for 3 min, followed by 36 cycles of denaturation at 95°C for 20 seconds, annealing at 60°C for 30 seconds and extension at 65°C for 2 min. The reactions were terminated by a final incubation at 65°C for 5 min.

The PCR products were diluted 1:5 and 5 μl of the dilution was added to a mix containing 40 μl of Sample Loading Solution (SLS, Beckmann Coulter, Fullerton, CA., USA) and 0.5 μl of MicroSTEP-15a (800) size marker (Microzone Haywards Heath, UK). The samples were separated by electrophoresis in CEQ Separation Gel LPA I (Urea in buffered sieving matrix; Polyacrylamide, Beckman Coulter) on a CEQ 8000 automatic DNA Analysis System (Beckman-Coulter, Fullerton, CA., USA) with the following conditions: denaturation 90°C for 120 sec, inject 2.0 kV for 20 sec, separation 3.0 kV for 180 min. Allele sizes were estimated using CEQ Fragment Analysis System analysis software, by comparing the amplicons to the internal size standard (MicroStep-15a), consisting of fifteen labelled fragments ranging from 75 to 1000 bp.

Bams30 alleles sequencing was done as previously described [30]. The corresponding sequences accession numbers (EMBL) are AM182499-AM182509.

MLVA data analysis

Previously published MLVA8 typing data was recovered from the literature [14, 2126]. Different conventions have been used for allele calling so that some minor corrections must be applied before merging the data. As noticed by Cheung et al. [24], some of the sizes initially indicated [14] do not correspond to the actual size, as can be deduced for instance from published genome sequence data. There is a 1 bp error in the vrr A allele sizes (for instance the allele indicated 313 in [14] is 314 bp). This has no consequences in terms of repeat unit copy number since vrr A has a 12 bp motif. The correction which needs to be applied to vrr C1 (9 bp repeat unit) is a bit more complex and as follows: the 520, 538, 583, 613 and 685 bp alleles must be converted to respectively 517, 535, 580, 616, and 688. The pXO1-aat sizes are underestimated by 3 bp. As a result the correct "Ames" strain MLVA8 code expressed in base-pairs reads 314-229-153-580-532-158-126-141 (or in terms of repeat copy numbers: 10-16-6-53-17-2-7-10) for vrr A, vrr B1, vrr B2, vrr C1, vrr C2, CG3, pXO1-aat and pXO2-at respectively. The "Sterne" strain is 314-229-162-580-532-158-132-135. Since at least one of these two strains is included in the cited publications, it is possible to deduce which conversion(s) needs to be applied to the corresponding data set.

The MLVA25 genotype was deduced from the genbank sequence data for strains "Ames", "Sterne", "Vollum", "Kruger", "Western", "BA1055", "Australia94".

All the data produced was managed using BioNumerics software version 4.5 (Applied Maths, Sint-Marteens-Latem, Belgium). Clustering was done using the categorical similarity coefficient and the Single Linkage Method.

Declarations

Acknowledgements

We thank Dr. Antonio Battisti from the Istituto Zooprofilattico Sperimentale del Lazio e della Toscana (Rome, Italy) for the generous gift of his collection of B. anthracis isolates. Work on the typing and molecular epidemiology of dangerous pathogens is supported by the French and Italian Ministry of Defence. This work was part of the European Biodefence project CEPA13.14 involving biodefence institutions from Sweden, Norway, the Nederlands, Germany, France and Italy. ACia and RDes are partially supported by the "Progetto Antrace" – ISS-Ministero della Salute. Preliminary sequence data for B. anthracis strains was obtained from The Institute for Genomic Research [27] using the Blast querying tool available from the National Center for Biotechnology Information [35].

Authors’ Affiliations

(1)
Army Medical Research Center, Via Santo Stefano Rotondo 4 00184 Rome (Italy)
(2)
Cattedra di Allergologia e Immunologia Clinica, II Facoltà di Medicina, Università di Roma "La Sapienza"
(3)
Division of Analytical Microbiology, Centre d'Etudes du Bouchet
(4)
Institut de Génétique et Microbiologie, Université Paris Sud
(5)
AFSSA/LERPAZ, LNR/CNR associé Laboratoire du Charbon
(6)
Istituto Superiore di Sanità, Viale Regina Elena
(7)
Istituto Zooprofilattico Sperimentale della Puglia e della Basilicata – Anthrax Reference Institute of Italy
(8)
Direzione Generale della Sanità Militare, Via Santo Stefano Rotondo

References

  1. Bush LM, Abrams BH, Beall A, Johnson CC: Index case of fatal inhalational anthrax due to bioterrorism in the United States. N Engl J Med. 2001, 345: 1607-1610. 10.1056/NEJMoa012948.View ArticlePubMedGoogle Scholar
  2. Turnbull PC: Definitive identification of Bacillus anthracis – a review. J Appl Microbiol. 1999, 87: 237-240. 10.1046/j.1365-2672.1999.00876.x.View ArticlePubMedGoogle Scholar
  3. Price LB, Hugh-Jones M, Jackson PJ, Keim P: Genetic diversity in the protective antigen gene of Bacillus anthracis. J Bacteriol. 1999, 181: 2358-2362.PubMed CentralPubMedGoogle Scholar
  4. Jackson PJ, Hill KK, Laker MT, Ticknor LO, Keim P: Genetic comparison of Bacillus anthracis and its close relatives using amplified fragment length polymorphism and polymerase chain reaction analysis. J Appl Microbiol. 1999, 87: 263-269. 10.1046/j.1365-2672.1999.00884.x.View ArticlePubMedGoogle Scholar
  5. Meselson M, Guillemin J, Hugh-Jones M, Langmuir A, Popova I, Shelokov A, Yampolskaya O: The Sverdlovsk anthrax outbreak of 1979. Science. 1994, 266: 1202-1208.View ArticlePubMedGoogle Scholar
  6. Takahashi H, Keim P, Kaufmann AF, Keys C, Smith KL, Taniguchi K, Inouye S, Kurata T: Bacillus anthracis incident, Kameido, Tokyo, 1993. Emerg Infect Dis. 2004, 10: 117-120.PubMed CentralView ArticlePubMedGoogle Scholar
  7. Keim P, Kalif A, Schupp J, Hill K, Travis SE, Richmond K, Adair DM, Hugh-Jones M, Kuske CR, Jackson P: Molecular evolution and diversity in Bacillus anthracis as detected by amplified fragment length polymorphism markers. J Bacteriol. 1997, 179: 818-824.PubMed CentralPubMedGoogle Scholar
  8. Vergnaud G, Denoeud F: Minisatellites: Mutability and Genome Architecture. Genome Res. 2000, 10: 899-907. 10.1101/gr.10.7.899.View ArticlePubMedGoogle Scholar
  9. Vergnaud G, Mariat D, Apiou F, Aurias A, Lathrop M, Lauthier V: The use of synthetic tandem repeats to isolate new VNTR loci: cloning of a human hypermutable sequence. Genomics. 1991, 11: 135-144. 10.1016/0888-7543(91)90110-Z.View ArticlePubMedGoogle Scholar
  10. Moxon ER, Rainey PB, Nowak MA, Lenski RE: Adaptive evolution of highly mutable loci in pathogenic bacteria. Curr Biol. 1994, 4: 24-33. 10.1016/S0960-9822(00)00005-1.View ArticlePubMedGoogle Scholar
  11. Keim P, Van Ert MN, Pearson T, Vogler AJ, Huynh LY, Wagner DM: Anthrax molecular epidemiology and forensics: using the appropriate marker for different evolutionary scales. Infect Genet Evol. 2004, 4: 205-213. 10.1016/j.meegid.2004.02.005.View ArticlePubMedGoogle Scholar
  12. Le Flèche P, Hauck Y, Onteniente L, Prieur A, Denoeud F, Ramisse V, Sylvestre P, Benson G, Ramisse F, Vergnaud G: A tandem repeats database for bacterial genomes: application to the genotyping of Yersinia pestis and Bacillus anthracis. BMC Microbiol. 2001, 1: 2-10.1186/1471-2180-1-2.PubMed CentralView ArticlePubMedGoogle Scholar
  13. Denoeud F, Vergnaud G: Identification of polymorphic tandem repeats by direct comparison of genome sequence from different bacterial strains: a Web-based ressource. BMC Bioinformatics. 2004, 5: 4-10.1186/1471-2105-5-4.PubMed CentralView ArticlePubMedGoogle Scholar
  14. Keim P, Price LB, Klevytska AM, Smith KL, Schupp JM, Okinaka R, Jackson PJ, Hugh-Jones ME: Multiple-Locus Variable-Number Tandem Repeat Analysis Reveals Genetic Relationships within Bacillus anthracis. J Bacteriol. 2000, 182: 2928-2936. 10.1128/JB.182.10.2928-2936.2000.PubMed CentralView ArticlePubMedGoogle Scholar
  15. Le Flèche P, Fabre M, Denoeud F, Koeck JL, Vergnaud G: High resolution, on-line identification of strains from the Mycobacterium tuberculosis complex based on tandem repeat typing. BMC Microbiol. 2002, 2: 37-10.1186/1471-2180-2-37.PubMed CentralView ArticlePubMedGoogle Scholar
  16. Pourcel C, Andre-Mazeaud F, Neubauer H, Ramisse F, Vergnaud G: Tandem repeats analysis for the high resolution phylogenetic analysis of Yersinia pestis. BMC Microbiol. 2004, 4: 22-10.1186/1471-2180-4-22.PubMed CentralView ArticlePubMedGoogle Scholar
  17. Le Fleche P, Jacques I, Grayon M, Al Dahouk S, Bouchon P, Denoeud F, Nockler K, Neubauer H, Guilloteau LA, Vergnaud G: Evaluation and selection of tandem repeat loci for a Brucella MLVA typing assay. BMC Microbiol. 2006, 6: 9-10.1186/1471-2180-6-9.PubMed CentralView ArticlePubMedGoogle Scholar
  18. Lindstedt BA: Multiple-locus variable number tandem repeats analysis for genetic fingerprinting of pathogenic bacteria. Electrophoresis. 2005, 26: 2567-2582. 10.1002/elps.200500096.View ArticlePubMedGoogle Scholar
  19. Springer-Verlag, Vergnaud G, Pourcel C: Multiple Locus VNTR (Variable Number of Tandem Repeat) Analysis (MLVA). Molecular Identification, Systematics and Population Structure of Prokaryotes. Edited by: Stackebrandt E. 2006, Springer-Verlag, 83-104.View ArticleGoogle Scholar
  20. Pearson T, Busch JD, Ravel J, Read TD, Rhoton SD, U'Ren JM, Simonson TS, Kachur SM, Leadem RR, Cardon ML, Van Ert MN, Huynh LY, Fraser CM, Keim P: Phylogenetic discovery bias in Bacillus anthracis using single-nucleotide polymorphisms from whole-genome sequencing. Proc Natl Acad Sci USA. 2004, 101: 13536-13541. 10.1073/pnas.0403844101.PubMed CentralView ArticlePubMedGoogle Scholar
  21. Hoffmaster AR, Fitzgerald CC, Ribot E, Mayer LW, Popovic T: Molecular subtyping of Bacillus anthracis and the 2001 bioterrorism-associated anthrax outbreak, United States. Emerg Infect Dis. 2002, 8: 1111-1116.PubMed CentralView ArticlePubMedGoogle Scholar
  22. Fouet A, Smith KL, Keys C, Vaissaire J, Le Doujet C, Levy M, Mock M, Keim P: Diversity among French Bacillus anthracis isolates. J Clin Microbiol. 2002, 40: 4732-4734. 10.1128/JCM.40.12.4732-4734.2002.PubMed CentralView ArticlePubMedGoogle Scholar
  23. Gierczynski R, Kaluzewski S, Rakin A, Jagielski M, Zasada A, Jakubczak A, Borkowska-Opacka B, Rastawicki W: Intriguing diversity of Bacillus anthracis in eastern Poland – the molecular echoes of the past outbreaks. FEMS Microbiol Lett. 2004, 239: 235-240. 10.1016/j.femsle.2004.08.038.View ArticlePubMedGoogle Scholar
  24. Cheung DT, Kam KM, Hau KL, Au TK, Marston CK, Gee JE, Popovic T, Van Ert MN, Kenefic L, Keim P, Hoffmaster AR: Characterization of a Bacillus anthracis isolate causing a rare case of fatal anthrax in a 2-year-old boy from Hong Kong. J Clin Microbiol. 2005, 43: 1992-1994. 10.1128/JCM.43.4.1992-1994.2005.PubMed CentralView ArticlePubMedGoogle Scholar
  25. Ryu C, Lee K, Hawng HJ, Yoo CK, Seong WK, Oh HB: Molecular characterization of Korean Bacillus anthracis isolates by amplified fragment length polymorphism analysis and multilocus variable-number tandem repeat analysis. Appl Environ Microbiol. 2005, 71: 4664-4671. 10.1128/AEM.71.8.4664-4671.2005.PubMed CentralView ArticlePubMedGoogle Scholar
  26. Fasanella A, Van Ert M, Altamura SA, Garofolo G, Buonavoglia C, Leori G, Huynh L, Zanecki S, Keim P: Molecular diversity of Bacillus anthracis in Italy. J Clin Microbiol. 2005, 43: 3398-3401. 10.1128/JCM.43.7.3398-3401.2005.PubMed CentralView ArticlePubMedGoogle Scholar
  27. The Institute for Genomic Research.http://www.tigr.org/
  28. Vergnaud G: Yersinia pestis genotyping. Emerg Infect Dis. 2005, 11: 1317-1318. author reply 1318–1319PubMed CentralView ArticlePubMedGoogle Scholar
  29. Hoffmaster AR, Ravel J, Rasko DA, Chapman GD, Chute MD, Marston CK, De BK, Sacchi CT, Fitzgerald C, Mayer LW, Maiden MC, Priest FG, Barker M, Jiang L, Cer RZ, Rilstone J, Peterson SN, Weyant RS, Galloway DR, Read TD, Popovic T, Fraser CM: Identification of anthrax toxin genes in a Bacillus cereus associated with an illness resembling inhalation anthrax. Proc Natl Acad Sci USA. 2004, 101: 8449-8454. 10.1073/pnas.0402414101.PubMed CentralView ArticlePubMedGoogle Scholar
  30. Valjevac S, Hilaire V, Lisanti O, Ramisse F, Hernandez E, Cavallo JD, Pourcel C, Vergnaud G: Comparison of minisatellite polymorphisms in the Bacillus cereus complex: a simple assay for large-scale screening and identification of strains most closely related to Bacillus anthracis. Appl Environ Microbiol. 2005, 71: 6613-6623. 10.1128/AEM.71.11.6613-6623.2005.PubMed CentralView ArticlePubMedGoogle Scholar
  31. Todd SJ, Moir AJ, Johnson MJ, Moir A: Genes of Bacillus cereus and Bacillus anthracis encoding proteins of the exosporium. J Bacteriol. 2003, 185: 3373-3378. 10.1128/JB.185.11.3373-3378.2003.PubMed CentralView ArticlePubMedGoogle Scholar
  32. Sylvestre P, Couture-Tosi E, Mock M: Polymorphism in the collagen-like region of the Bacillus anthracis BclA protein leads to variation in exosporium filament length. J Bacteriol. 2003, 185: 1555-1563. 10.1128/JB.185.5.1555-1563.2003.PubMed CentralView ArticlePubMedGoogle Scholar
  33. MLVA web-service.Google Scholar
  34. GPMS, Genomes, Polymorphism and Minisatellites.http://minisatellites.u-psud.fr/
  35. The National Center for Biotechnology Information.http://www.ncbi.nlm.nih.gov/
  36. Hunter PR, Gaston MA: Numerical index of the discriminatory ability of typing systems: an application of Simpson's index of diversity. J Clin Microbiol. 1988, 26: 2465-2466.PubMed CentralPubMedGoogle Scholar

Copyright

© Lista et al; licensee BioMed Central Ltd. 2006

This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Advertisement