CORE Metadata, citation and similar papers at core.ac.uk

Provided by Elsevier - Publisher Connector

ORIGINAL ARTICLE 10.1111/j.1469-0691.2004.01034.x

Single-nucleotide polymorphism-based differentiation and drug resistance detection in from isolates or directly from sputum C. Arnold1, L. Westland1,2, G. Mowat3, A. Underwood1, J. Magee3 and S. Gharbia1

1Genomics, Proteomics and Bioinformatics Unit, Centre for Infections, Health Protection Agency, London, UK, 2Hogeschool van Utrecht, Faculteit Natuur & Techniek, Institute of Life Sciences and Chemistry, Utrecht, The Netherlands and 3Regional Centre for Mycobacteriology, Health Protection Agency Newcastle Laboratory, Newcastle, UK

ABSTRACT The rapid technique of pyrosequencing was used to examine 123 samples (in the form of DNA extracts and inactivated sputum) of Mycobacterium spp. Of 99 Mycobacterium tuberculosis samples investigated for single-nucleotide polymorphisms (SNPs), 68% of isoniazid-resistant isolates analysed had an AGC fi ACC mutation in katG at codon 315, resulting in the Ser fi Thr substitution associated previously with isoniazid resistance. Of the rifampicin-resistant isolates, 92% showed SNPs in rpoB at codons 516, 531 or 526. Inactivated sputum samples and DNA extracts could both be analysed by pyrosequencing, and the method was able to differentiate rapidly between the closely related species of the M. tuberculosis complex (M. tuberculosis, , , Mycobacterium canetti and ), except between M. tuberculosis, M. canetti and one of two M. africanum strains. This low-cost, high-throughput technique could be used as a rapid screen for drug resistance and as a replacement for some of the time-consuming tests used currently for species identification. Keywords Antibiotic susceptibility testing, identification, Mycobacterium tuberculosis complex, pyrosequencing, resistance Original Submission: 5 April 2004; Revised Submission: 2 September 2004; Accepted: 11 September 2004 Clin Microbiol Infect 2005; 11: 122–130

amikacin, and fluoroquinolones such as ofloxacin INTRODUCTION and ciprofloxacin, are used as second-line agents. It has been estimated that one-third of the global Drug resistance in M. tuberculosis occurs fol- population is infected with Mycobacterium tuber- lowing spontaneous mutation in target genes culosis, with c. 9 million cases of active tubercu- such as katG, rpoB and inhA [3]. The mechanisms losis (TB) and > 2 million deaths occurring each and the dominant mutations resulting in the drug year [1,2]. These numbers are increasing, as is the resistance profiles have been well-documented number of individuals infected with strains resist- [3–5]. Isoniazid resistance is often associated with ant to first-line anti-TB drugs. In the UK, the mutation in a gene encoding catalase-peroxidase current first-line treatment is a 2-month course of (katG), most frequently a Ser315Thr mutation [6]. isoniazid, rifampicin and pyrazinamide, often Rifampicin inhibits transcription by binding to with the addition of ethambutol, followed by the RNA polymerase b-subunit, encoded by the isoniazid and rifampicin for a further 4–6 months. rpoB gene [7], with resistance resulting from Aminoglycosides such as streptomycin and mutations in the rifampicin resistance-determin- ing region of rpoB, an 81-bp region encoding amino-acids 507–533. The most common (70%) Corresponding author and reprint requests: C. Arnold, mutations are at codons 526 and 531, and Genomics, Proteomics and Bioinformatics Unit, Health Pro- tection Agency, 61 Colindale Avenue, London NW9 5HT, UK are associated with high levels of resist- E-mail: [email protected] ance (> 100 mg ⁄ L). Less frequent mutations,

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases Arnold et al. Single-nucleotide polymorphisms in M. tuberculosis 123 particularly in codon 516, are associated with members of the genus Mycobacterium, and spe- lower levels of resistance. The fluoroquinolones cifically the M. tuberculosis complex (MTBC), were considered to be alternative drugs, as they using a composite assay of these three genes. prevent transcription and cell division by sup- As public health intervention strategies and pressing DNA supercoiling [8]. DNA gyrase, transmission models may also benefit from the encoded by gyrAB, catalyses negative supercoil- interpretation of data in the context of phyloge- ing of DNA. Mutations in the short quinolone netically relevant genetic frameworks (geno- resistance-determining region and at codon 94 of types) based on SNPs [14,15], a pyrosequencing gyrA overcome the DNA gyrase-blocking action assay to determine genotype was also developed. of fluoroquinolones [8]. These three composite high-throughput assays, A rapid screening system for these mutations designed to be used on sputum samples and among isolates of M. tuberculosis is essential. A cultures in the early stages of growth to deter- sequence-based method, rather than a DNA mine resistance profile, species and genotype, migration or a hybridisation approach, is ideal were applied to three groups of M. tuberculosis for the development of a fast, accurate and high- isolates to determine their utility in public health throughput system, as this will aid in the identi- microbiology. fication of alternative mutations in the regions analysed, as well as those characterised previ- MATERIALS AND METHODS ously. The present study investigated the possi- bility of detecting the most common mutations Samples by pyrosequencing, a short-read (c. 30–50 bp) In total, 123 samples of mycobacteria were studied, of which 98 sequencing technique based on the quantitative were M. tuberculosis nucleic acid extracts. detection of pyrophosphate released following Group 1 comprised 72 M. tuberculosis DNA extracts from three northwest London hospitals [16]. Seven (9.7%) of these nucleotide incorporation into a growing DNA extracted isolates were isoniazid-resistant (Table 1). There chain. The technology has also been used previ- were six epidemiologically related groups of isolates (100 ously for single-nucleotide polymorphism (SNP) and 203; 113 and 128; 129 and 264; 139 and 140; 232, 251, detection [9]. 252 and 253; 1085 and 1086). Group 2 comprised 20 M. tuberculosis DNA extracts and The slow-growing nature of M. tuberculosis seven inactivated sputum samples containing M. tuberculosis means that several weeks can elapse before the from northeast England (Newcastle) (Table 1). Fourteen of the susceptibility profile of an isolate is known and extracted isolates were isoniazid- or isoniazid- and rifampicin- appropriately guided anti-TB chemotherapy can resistant. The sputum samples, containing > 90 bacilli ⁄ field, were inactivated by boiling in 1 M NaOH for 10 min following be put in place. Although genotypic testing cannot dithiothreitol liquefaction (Sputasol; Oxoid, Basingstoke, UK). wholly replace phenotypic methods for the detec- Four of these samples yielded isoniazid-resistant M. tubercu- tion of antibiotic resistance, as not all mechanisms losis isolates. A further four sputum samples contained of action (and therefore specific mutations) are mycobacterial species other than those causing TB, i.e., known for the different drugs used, a molecular M. avium complex, M. malmoense and M. chelonae, which were isoniazid-resistant. screen capable of use on primary clinical samples Group 3 comprised 24 DNA extracts from 11 different to detect most mutations associated with resist- Mycobacterium spp. (Table 2) obtained from K. Kremer ance could be invaluable. This study therefore (National Institute of Public Health and the Environment, assessed pyrosequencing as a high-throughput, Bilthoven, The Netherlands) [17]. cost-effective genotypic screen for M. tuberculosis isoniazid resistance in which common mutations Primer design associated with resistance were detected in spu- Amplification and pyrosequencing primers were designed by tum samples and early cultures. visual comparison and confirmed with the use of on-line Speciation of mycobacteria has been attempted software (http://www.pyrosequencing.com). The sequence of M. tuberculosis H37Rv, ATCC 27294, was used as an in-silico previously by sequencing all or part of the 16S template. Pyrosequencing primers were designed to detect rRNA gene [10], but this approach has failed to resistance or species-associated SNPs (Fig. 1). Two of the delineate this highly conserved genus. Other regions analysed (katG 463 and gyrA 95) were control regions regions of the genome that are useful for myco- that were not associated with drug resistance, but were associated with genotype assignation of M. tuberculosis [14], bacterial speciation include rpoB [11,12] and gyrB namely: CTG ⁄ ACC = genotype I; CGG ⁄ ACC = genotype II; [13]. Therefore, the present study also assessed and CGG ⁄ AGC = genotype III (Table 1). pyrosequencing as a rapid test for identifying

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 122–130 124 Clinical Microbiology and Infection, Volume 11 Number 2, February 2005

Table 1. Strain information for Mycobacterium tuberculosis isolates in groups 1 and 2: resistance profile, single-nucleotide polymorphism profile and genotype

katG 315 rpoB 516 rpoB 526 rpoB 531 gyrA 94 Identifier Resistance (WT = AGC) (WT = GAC) (WT = CAC) (WT = TCG) (WT = GAC) katG 463 gyrA 95 Genogroup

Northwest London isolates (DNA extracts) WT profile 1 104, 124, 125, 126, WT AGC GAC CAC TCG GAC CTG ACC I 129, 131, 138, 142, 145, 149, 154, 157, 200, 208, 209, 212, 234, 241, 264, 267, 280, 282, 284, 285, 286, 982, 1010, 1023, 1026, 1053 WT profile 2 98, 100, 105, 106, WT AGC GAC CAC TCG GAC CGG ACC II 113, 114, 121, 122, 128, 136, 139, 140, 156, 203, 205, 218, 222, 223, 231, 237, 263, 269, 1002, 1022, 1030, 1054, 1062, CDC1551 WT profile 3 127, 229, 276, 1024, WT AGC GAC CAC TCG GAC CGG AGC III H37Rv WT profile 4 226, 1043 WT AGC GAC CAC TCG GAC Weak ACC I ⁄ II Resistant profile 1 137, 232, 251, 252, INH ACC GAC CAC TCG GAC CTG ACC I 253, 1085, 1086 Newcastle isolates (DNA extracts) WT profile 1 16, 17 WT AGC GAC CAC TCG GAC CTG ACC I WT profile 2 12, 13 WT AGC GAC CAC TCG GAC CGG ACC II WT profile 3 14, 15 WT AGC GAC CAC TCG GAC CGG AGC III Resistant profiles 1INH⁄ RIF ACC GAC CAC TTG GAC CTG ACC I 8 INH AGC GAC CAC TCG GAC CTG ACC I 9 INH AGC GAC CAC TCG GAC CGG AGC III 10 INH ⁄ RIF ⁄ EMB ⁄ PZA ACC GCC CAC TCG GAC CTG ACC I 11 INH ⁄ RIF ⁄ EMB AGC GAC TCC Weak GAC CGG ACC II 22 INH ⁄ RIF ⁄ EMB ACC GAC CAC TTG GAC CTG ACC I 23 INH ⁄ RIF ⁄ PZA AGC GAC CAC TTG GAC CGG ACC II 24 INH ⁄ RIF ⁄ EMB AGC GAC TCC Weak GAC CGG ACC II 25 INH ⁄ RIF ACC GAC GAC TCG GAC CGG ACC II 26 INH ⁄ RIF ⁄ EMB ACC GAC TAC TCG GAC CTG ACC I 27 INH ⁄ RIF ⁄ EMB ACC GAC GAC TCG GAC CGG ACC II 31 INH ⁄ RIF ⁄ EMB AGC GAC CAC TCG GAC CGG ACC II 33 INH ⁄ RIF ⁄ EMB ACC GAC CAC TTG GAC CTG ACC I 35 INH ⁄ RIF ACC TAC CAC TCG GAC CTG ACC I Newcastle isolates (inactivated sputum) 36 INH AGC GAC CAC TCG ND CTG ND I 37 INH AGC GAC CAC TCG ND CTG ND I 38 INH ACC GAC CAC TCG ND CGG ND II ⁄ III 39 INH ACC GAC CAC TCG ND CTG ND I 40 WT AGC ND ND ND ND CGG ND II ⁄ III 41 WT AGC ND ND ND ND CTG ND I 42 WT AGC GAC CAC TCG ND CGG ND II ⁄ III

WT, wild-type; INH, isoniazid; RIF, rifampicin; EMB, ethambutol; PZA, pyrazinamide; ND not done.

The rifampicin resistance detection assay and the rnpB PCR species identification assay were designed by M. Krabbe (Pyrosequencing AB, Uppsala, Sweden). The rpoB species PCR was used to generate products from the rpoB, katG, gyrA, identification assay was designed to generate c. 100-bp gyrB and rnpB genes in a final volume of 50 lL with c. 50 ng products by means of an alignment of sequences reported by of extracted DNA or 1 lL of inactivated sputum, and 25 lL Lee et al. [11]. of PCR Ready Mix (Sigma-Aldrich, Poole, UK) containing The gyrB MTBC species identification assay targeted the 1.5 U of Taq polymerase, 10 mM Tris-HCl, 50 mM KCl, MTBC-associated SNPs identified by Niemann et al. [13] and 1.5 mM MgCl2, gelatine 0.001% v ⁄ v, 0.2 mM dNTPs and Kasai et al. [18], and was designed to distinguish M. tubercu- stabilisers, and 10 pmol each of the forward and reverse losis, M. bovis, M. africanum and M. microti. primers (one of which was biotinylated; Fig. 1). PCR involved

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 122–130 Arnold et al. Single-nucleotide polymorphisms in M. tuberculosis 125

Table 2. Identification of Mycobacte- gyrB SNP gyrB SNP gyrB SNP gyrB SNP rium spp. in group 3 on the basis rnpB ID rpoB ID position position position position of single-nucleotide polymorphism Identifier Species assay assay 675 756 1410 1450 and short-sequence data analysis 1 M. tuberculosis (H37Rv) MTBC MTBC C G C G 2 M. bovis BCG MTBC MTBC C A T T 3 M. canettii MTBC MTBC C G C G 4 M. canettii MTBC MTBC C G C G 5 M. africanum MTBC MTBC C G C T 6 M. africanum MTBC MTBC C G C G 7 M. microti MTBC MTBC T G C T 8 M. microti MTBC MTBC T G C T 9 M. bovis MTBC MTBC C A T T 10 M. bovis MTBC MTBC C A T T 11 M. avium M. avium M. avium CGCG 12 M. avium M. avium M. avium CGCG 13 M. kansasii M. kansasii M. kansasii CGCG 14 M. kansasii M. kansasii M. kansasii CGCG 15 M. gordonae Unclear M. gordonae CGCT 16 M. gordonae Unclear M. gordonae CGCT 17 M. smegmatis M. smegmatis M. smegmatis CGCG 18 M. smegmatis M. smegmatis M. smegmatis CGCG 19 M. xenopi M. xenopi M. xenopi C No data C No data 20 M. xenopi M. xenopi M. xenopi C No data C No data 21 M. tuberculosis MTBC MTBC C G C G 22 M. tuberculosis MTBC MTBC C G C G 23 M. tuberculosis MTBC MTBC C G C G 24 M. tuberculosis MTBC MTBC C G C G

MTBC, Mycobacterium tuberculosis complex. denaturation at 94C for 2 min, followed by 35 cycles of 94C resulting in a Ser fi Thr substitution (Fig. 2), was for 30 s, 60C for 30 s and 72C for 1 min, followed by a final identified in all of the isoniazid-resistant extracts extension at 72C for 10 min. Amplification products were checked on a 96-well Ready-To-Run agarose gel (Amersham (Table 1). Two groups (n =4; n = 2) of these Biosciences, Chalfont St Giles, UK). When PCR performed samples were related epidemiologically. In group directly on sputum samples resulted in low yields, 1 lLof 2, the same mutation was identified in ten of the 18 the initial product was re-amplified to generate sufficient isoniazid-resistant samples (Table 1). Mutations product for sequencing. associated with resistance at gyrA codon 94 were not detected, although PCR products and pyrose- Pyrosequencing quence data were generated from all samples, and Pyrosequencing was performed with the SQA kit (Pyrosequ- all replicates gave identical pyrosequence data. In encing AB) and the pyrosequencing primers shown in Fig. 1. group 2, the rpoB positions analysed showed Briefly, 20 lL of biotinylated PCR product was bound to streptavidin-coated sepharose beads and denatured. After mutations, compared to the wild-type, in 11 of washing, the products were transferred to a 96-well microtitre the 12 extracts of isoniazid- and rifampicin-resist- plate and 20 qmol of pyrosequencing primer was annealed in a ant isolates. Point mutations were identified at 45-lL reaction volume. Once the reaction cartridge was loaded codon 516 in two samples (10 and 35), at codon 526 with dNTPs, enzyme and substrate from the kit, the cart- in five samples (11, 24–27), and at codon 531 in four ridge and reaction plate were placed in the instrument (Pyrosequencing AB) for analysis. Depending on the number samples (1, 22, 23 and 33) (Table 1). No mutations of nucleotide additions (each nucleotide addition cycle were identified at gyrA codon 95. takes 1 min to complete), 96 reactions were completed in PCR products were not amplified from the four c. 10–15 min. inactivated sputum samples containing Mycobac- terium spp. other than those causing TB. How- RESULTS ever, in group 2, a katG codon 315 mutation was identified in two of the four inactivated sputum Resistance mutations samples containing isoniazid-resistant MTBC The most frequent mutations associated with (determined by resistance ratio testing [19]) isoniazid, rifampicin and fluoroquinolone resist- (Table 1). ance were detected with the primers described in Fig. 1. PCR products were generated from all 99 Genotyping M. tuberculosis samples tested (both extracted nucleic acid and sputum samples). In group 1, an SNPs at katG codon 463 and gyrA codon 95 are not AGC fi ACC mutation in katG at codon 315, associated with resistance, but were used to

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 122–130 126 Clinical Microbiology and Infection, Volume 11 Number 2, February 2005

Fig. 1. Location of single-nucleotide polymorphisms (SNP) and pyrosequencing primers in PCR amplicons. SNPs of interest are shown in bold, and pyrosequencing primers are shown in italics. Products generated with a biotinylated forward primer (i.e., sequenced in the opposite direction) are marked with asterisks. separate genotypes I–III. The results are shown in assays could separate members of the MTBC. Table 1. However, the gyrB SNP assays correctly differen- tiated between the various species, except between M. tuberculosis, M. canetti and one of Identification the two M. africanum strains (Table 2). For the DNA sequences of all rpoB and rnpB fragments MTBC only, SNP position 675 was always C generated were compared with sequences gener- except for M. microti (T), position 756 was always ated previously for mycobacteria [11,12]. Both the G except for M. bovis (A), position 1410 was rnpB and the rpoB short (25–30 bp) sequence always C except for M. bovis (T), and position 1450 assays correctly identified all of the mycobacterial was always T except for M. tuberculosis, M. canetti species tested, with the exception of M. gordonae, and M. africanum type II (G). A schematic diagram which failed to produce a clean sequence with the showing the identification algorithm used is given rnpB assay (Table 2). As expected, neither of these in Fig. 3.

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 122–130 Arnold et al. Single-nucleotide polymorphisms in M. tuberculosis 127

effect’, whereby resistant TB strains gain addi- tional resistance to further anti-TB drugs. Correct treatment with first-line drugs usually prevents development of drug resistance in most patients, but it aggravates drug resistance in patients harbouring resistant strains, so that previous anti-TB treatment is a risk factor for MDR [20]. Antibiotic susceptibility testing is carried out on isolates from all new TB cases in the UK. The katG gene is the most commonly targeted region involved in isoniazid resistance, with mutations at codon 315 occurring in 30–90% of isoniazid- resistant isolates, depending on geographical distribution [7,21–23]. Recently, real-time molecular assays have been developed to detect mutations associated with Fig. 2. (A) A pyrogram of the wild type for katG Ser315. antibiotic resistance in M. tuberculosis. These The sequence reads GCGGCATC. (B) A pyrogram of a mutation conferring isoniazid resistance (AGC fi ACC) include the use of specific hybridisation probes at katG 315. The sequence reads CCGGCATC. (N.B. The for use with the LightCycler, a line probe assay first nucleotide in this codon (A) is the last base in the and PCR restriction fragment length polymorph- pyrosequencing primer). ism analysis [24–26]. The real-time PCR methods are fast, but DNA is extracted from cultured rpoB ID clinical isolates, which adds to the cost and time of the assay, and data interpretation can be subjective. Line probe assays are also carried out MOTT MTBC on DNA extracted from cultures, and the assay can take 1–2 days to complete [25]. PCR restric- tion fragment length polymorphism analysis gyrB ID1 gyrB ID2 requires DNA extraction, amplification, a 4-h restriction endonuclease digestion period and polyacrylamide gel electrophoresis. In M. tuberculosis samples from the UK, the SNP 675 SNP 756 SNP 1410 SNP 1450 C>T G>A C>T T>G AGC fi ACC mutation in katG at codon 315, =M. microti = M. bovis = M. bovis = M. tuberculosis or resulting in a Ser fi Thr substitution, was detec- M. canetti or ted by pyrosequencing in 68% of isoniazid-resist- M. africanum type II ant isolates. Among the 12 isoniazid- and Fig. 3. Schematic algorithm for the Mycobacterium tubercu- rifampicin-resistant isolates analysed, point muta- losis pyrosequencing identification assay. tions at codons 516, 531 or 526 were identified in 11 (92%) cases. Importantly, it was possible to sequence PCR products amplified from all spu- DISCUSSION tum samples containing M. tuberculosis, although In the UK, molecular tests for multidrug resist- samples containing Mycobacterium spp. other than ance (MDR) are carried out only following those causing TB were not amplified. The latter consideration of epidemiological factors such as finding may be a result of a lower bacterial load or nationality or close contact with an individual a reduced specificity of primers. Moreover, the known to be infected with multidrug-resistant TB. mutations conferring isoniazid resistance in atyp- Nationally, < 1% of isolates are multidrug-resis- ical mycobacteria are not necessarily related to tant [20], although isoniazid resistance is more those in M. tuberculosis. The ability to use clinical common (6–10%). Inappropriate prescription of samples without the need for expensive and time- isoniazid to patients infected with an isoniazid- consuming DNA extraction procedures would resistant strain can cause liver damage and may enable this method to be used more widely than is drive mutation to MDR by the so-called ‘amplifier possible at present. As the whole assay is

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 122–130 128 Clinical Microbiology and Infection, Volume 11 Number 2, February 2005 conducted in a high-throughput, 96-well format, advantages of identifying isoniazid resistance hundreds of isolates could be analysed in < 6 h. easily and cheaply in a single day, rather than The longest step in the process is the PCR (a step weeks or months, for the individual patient and in required in all molecular detection assays de- the wider context of preventing MDR are clear, scribed so far), while the pyrosequencing techni- and antibiotic resistance screening is recommen- que itself, including the PCR clean-up, takes c. ded for all new M. tuberculosis infections. 20 min for 96 samples and costs c. 2.70 euro for Phenotypic tests are available for the identifi- each SNP analysed. cation of Mycobacterium spp., but are time-consu- The pyrosequencing method produced clear ming and may lack discriminatory power [11]. and objective sequence data. These assays are Several molecular tests have been developed, based on short-read sequences, designed to give mostly based on 16S rRNA sequence analysis. control sequence data following or preceding the However 16S rRNA sequencing fails to differen- SNPs of interest. This objectivity is extremely tiate species belonging to the MTBC. Both of the useful for applications in medical microbiology, short-sequence assays described here, based on as the technique does not therefore require the rnpB and rpoB, identified the different Mycobacte- skilled interpretation of bands or peaks required rium spp. tested, although some sequence data with other methods. from the rnpB assay were unreadable, presumably In the PCR restriction fragment length poly- because of a lack of complementarity of the morphism study by Leung et al. [26], it was primers with genomic DNA from some of the observed that a particular SNP position in katG more diverse species, or possibly because of at codon 463, thought previously to be associated sequence-specific looping of the single-stranded with isoniazid resistance [27], was found primar- DNA template. The use of single-stranded bind- ily in isolates from the south China region. Of 375 ing protein or the addition of NNN (where N is (102 isoniazid-resistant and 273 isoniazid-suscept- any base) to the 5¢-end of the non-biotinylated ible) isolates, 317 (85%) had Leu463, compared PCR primer during its synthesis may counteract with the major wild-type Arg463 associated with this effect. isolates from the western world. This particular The SNPs used in this assay to differentiate SNP is associated with genotype I, thought to be members of the MTBC were first identified by evolutionarily old, and was described first by Kasai et al. [18] and described in more detail by Sreevatsan et al. [14] and in more detail by Niemann et al. [13]. These SNPs were able to Gutacker et al. [15]. Sreevatsan et al. observed differentiate between the members of the MBTC, that clusters of cases were caused by genotypic except for the three most closely related species, groups I and II, but not group III organisms, as namely M. tuberculosis, M. canetti and M. africanum. the newly emergent group III organisms were The present assay discriminated between one of associated with sporadic cases, suggesting that the M. africanum isolates and M. tuberculosis. It is the pathogen was evolving towards a state of likely that the M. africanum isolate identical (G) to reduced transmissibility or virulence. The study M. tuberculosis at SNP position 1450 belongs to type in northwest London [16] also revealed epidem- II, whereas the other M. africanum isolate tested iologically related groups of isolates, all of which belongs to type I. Niemann et al. [13] were able to were either genotype I or II, supporting the discriminate between the two M. africanum types hypothesis that genotypes I and II were associated by analysing a fifth SNP at gyrB position 1311. with increased transmissibility or virulence. How- Although the differentiation of M. tuberculosis, ever, in a separate study of related infections in M. canetti and M. africanum type II was not achieved the northeast of England, a genotype III cluster with pyrosequencing, and therefore remains based was identified in four unrelated clusters (data not on phenotypic characteristics or the use of methods shown). such as deletion analysis and amplified fragment The ability to perform the pyrosequencing length polymorphism analysis [28,29], it is likely assay on inactivated sputum samples has consid- that other SNPs that can be used to separate these erable advantages, including the fact that the closely related species will be identified. Indeed, speed of this high-throughput assay means that Goh et al. [30] have already published data indica- genotypic resistance screening of all newly diag- ting the use of a particular SNP to differentiate nosed infections in the UK is achievable. The M. canetti from other members of the MBTC.

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 122–130 Arnold et al. Single-nucleotide polymorphisms in M. tuberculosis 129

In summary, the present study succeeded in 11. Lee H, Bang H-E, Bai GH-J, Cho S-N. Novel polymorphic developing and piloting rapid, high-throughput, region of the rpoB gene containing Mycobacterium species- specific sequences and its use in identification of myco- short-sequence and SNP-based assays to speciate . J Clin Microbiol 2003; 41: 2213–2218. and screen for drug resistance in this important 12. Herrmann B, Pettersson B, Everett KD, Mikkelsen NE, human pathogen. These low-cost assays can be Kirsebom LA. Characterization of the rnpB gene and used to complement current techniques, and may RNase P RNA in the order Chlamydiales. Int J Syst Evol be used to replace some of the time-consuming 2000; 50: 149–158. 13. Niemann S, Harmsen D, Rusch-Gerdes S, Richter E. Dif- biochemical tests used currently for identification, ferentiation of clinical Mycobacterium tuberculosis complex and also as a rapid screen for the detection of drug isolates by gyrB DNA sequence polymorphism analysis. J resistance in M. tuberculosis infections in the UK. Clin Microbiol 2000; 38: 3231–3234. The ability to obtain sequence data directly from 14. Sreevatsan S, Pan X, Stockbauer KE et al. Restricted gene polymorphism in the Mycobacterium tuberculosis complex sputum samples containing M. tuberculosis sug- indicates evolutionarily recent global dissemination. Proc gests that the assays described here could indicate Natl Acad Sci USA 1997; 94: 9869–9874. species or drug resistance much more rapidly than 15. Gutacker MM, Smoot JC, Lux Migliaccio CA et al. Gen- methods used currently, and therefore that they ome-wide analysis of synonymous single nucleotide could have a considerable impact on public health polymorphisms in Mycobacterium tuberculosis complex organisms: resolution of genetic relationships among clo- intervention strategies. sely related microbial strains. Genetics 2002; 162: 1533– 1543. 16. Kumar D, Saunders N, Watson JM et al. Clusters of new ACKNOWLEDGEMENTS tuberculosis cases in north-west London: a survey from three hospitals based on IS6110 RFLP typing. J Infect 2000; The authors would like to acknowledge M. Krabbe for design 40: 132–137. of the rnpB identification assay and Pyrosequencing AB for 17. Kremer K, Frothingham R, Haas WH et al. Comparison software modification. of methods based on different molecular epidemiologi- cal markers for typing of Mycobacterium tuberculosis REFERENCES complex strains: interlaboratory study of discriminatory power and reproducibility. J Clin Microbiol 1999; 37: 1. Chan ED, Iseman MD. Current medical treatment for 2607–2618. tuberculosis. BMJ 2002; 325: 1282–1286. 18. Kasai H, Ezaki T, Harayama S. Differentiation of phylo- 2. World Health Organization. Tuberculosis fact sheet number genetically related slowly growing mycobacteria by their 104. Geneva: WHO, 2002. gyrB sequences. J Clin Microbiol 2000; 38: 301–308. 3. Riska PF, Jacobs WR, Alland D. Molecular determinants 19. Heifets LB, Cangelosi GA. Drug susceptibility testing of drug resistance in tuberculosis. Int J Tuberc 2000; 4: of Mycobacterium tuberculosis: a neglected problem at S4–S10. the turn of the century. Int J Tuberc Lung Dis 1999; 7: 4. Ramaswamy SV, Reich R, Dou S-J, Jasperse L, Pan X, 564–581. Wanger A. Single nucleotide polymorphisms in genes 20. Fodor T, Vadasz I, Lorinczi I. Drug-resistant tuberculosis associated with isoniazid resistance in Mycobacterium in Budapest. Int J Tuberc Lung Dis 1998; 2: 732–735. tuberculosis. Antimicrob Agents Chemother 2003; 47: 1241– 21. Piatek AS, Telenti A, Murray MR et al. Genotypic analysis 1250. of Mycobacterium tuberculosis in two distinct populations 5. Somoskovi A, Parsons LM, Salfinger M. The molecular using molecular beacons: implications for rapid suscepti- basis of resistance to isoniazid, rifampin and pyrazina- bility testing. Antimicrob Agents Chemother 2000; 44: 103– mide in Mycobacterium tuberculosis. Respir Res 2001; 2: 164– 110. 168. 22. Haas WH, Schilke K, Brand J et al. Molecular analysis of 6. Zhang Y, Heym B, Allen B, Young D, Cole S. The catalase- katG gene mutations in strains of Mycobacterium tubercu- peroxidase gene and isoniazid resistance of Mycobacterium losis complex from Africa. Antimicrob Agents Chemother tuberculosis. Nature 1992; 358: 591–593. 2000; 41: 1601–1603. 7. Ramaswamy SV, Musser JM. Molecular genetic basis of 23. Van Soolingen D, de Haas PEW, Van Doorn HE, Kuijper E, antimicrobial agent resistance in Mycobacterium tuberculo- Rinder H, Borgdorff MW. Mutations at amino acid posi- sis: 1998 update. Tuberc Lung Dis 1998; 79: 3–29. tion 315 of the katG gene are associated with high-level 8. Van Doorn HR, Kuijper EJ, Van Ende A et al. The sus- resistance to isoniazid, other drug resistance, and suc- ceptibility of Mycobacterium tuberculosis to isoniazid and cessful transmission of Mycobacterium tuberculosis in the the Arg > Leu mutation at codon 463 are not associated. Netherlands. J Infect Dis 2000; 182: 1788–1790. J Clin Microbiol 2001; 39: 1591–1594. 24. Rindi L, Bianchi L, Tortoli E, Lari N, Bonanni D, Garzelli 9. Ronaghi I, Nygren M, Lundeberg J, Nygren P. Analyses of C. A real-time PCR assay for detection of isoniazid secondary structures in DNA by pyrosequencing. Anal resistance in Mycobacterium tuberculosis clinical isolates. Biochem 1999; 267: 65–71. J Microbiol Meth 2003; 55: 797–800. 10. Rogall T, Flohr T, Bottger E. Differentiation of mycobac- 25. Johansen IS, Lundgren B, Sosnovskaja A, Thomsen VO. terial species by direct sequencing of amplified DNA. Direct detection of multi-drug resistant Mycobacterium J Gen Microbiol 1990; 136: 1915–1920. tuberculosis in clinical specimens in low- and high-incidence

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 122–130 130 Clinical Microbiology and Infection, Volume 11 Number 2, February 2005

countries by line probe assay. J Clin Microbiol 2003; 41: the basis of genomic deletions. J Clin Microbiol 2003; 41: 4454–4456. 1637–1650. 26. Leung E, Kam K-M, Chiu A et al. Detection of katG 29. Van den Braak N, Simons G, Gorkink R et al. A new high- Ser315Thr substitution in respiratory specimens from pa- throughput AFLP approach for identification of new tient with isoniazid-resistant Mycobacterium tuberculosis genetic polymorphism in the genome of the clonal using PCR-RFLP. J Med Microbiol 2003; 52: 999–1003. microorganism Mycobacterium tuberculosis. J Microbiol Meth 27. Cockerill FR, Uhl JR, Temesgen Z et al. Rapid identifica- 2004; 56: 49–62. tion of a point mutation of the Mycobacterium tuberculosis 30. Goh KS, Legrand E, Sola C, Rastogi N. Rapid differen- catalase-peroxidase (katG) gene associated with isoniazid tiation of ‘Mycobacterium canettii’ from other Mycobacte- resistance. J Infect Dis 1995; 171: 240–245. rium tuberculosis complex organisms by PCR-restriction 28. Huard RC, de Oliveira Lazzarini LC, Butler WR, Van analysis of the hsp65 gene. J Clin Microbiol 2001; 39: Soolingen D, Ho JL. PCR-based method to differentiate the 3705–3708. subspecies of the Mycobacterium tuberculosis complex on

2004 Copyright by the European Society of Clinical Microbiology and Infectious Diseases, CMI, 11, 122–130