The ada operon of Mycobacterium tuberculosis encodes two DNA methyltransferases for inducible repair of DNA alkylation damage

DNA Repair 10 (2011) 595–602

Contents lists available at ScienceDirect

DNA Repair journal homepage: www.elsevier.com/locate/dnarepair

The ada operon of Mycobacterium tuberculosis encodes two DNA methyltransferases for inducible repair of DNA alkylation damage Mingyi Yang a,b,c , Randi M. Aamodt a,c , Bjørn Dalhus a,b,c , Seetha Balasingham a,c , Ina Helle b,c , Pernille Andersen b,c , Tone Tønjum a,c , Ingrun Alseth a,c , Torbjørn Rognes a,c,d , Magnar Bjørås a,b,c,∗ a

Department of Microbiology, Oslo University Hospital, Rikshospitalet, PO Box 4950 Nydalen, NO-0424 Oslo, Norway Institute of Clinical Biochemistry, University of Oslo, Rikshospitalet, PO Box 4950 Nydalen, NO-0424 Oslo, Norway c Centre for Molecular Biology and Neuroscience, Oslo University Hospital, Rikshospitalet, PO Box 4950 Nydalen, NO-0424 Oslo, Norway d Department of Informatics, University of Oslo, PO Box 1080 Blindern, NO-0316 Oslo, Norway b

a r t i c l e

i n f o

Article history: Received 2 December 2010 Received in revised form 1 March 2011 Accepted 15 March 2011 Available online 12 May 2011 Keywords: M. tuberculosis DNA damage Alkylation Ada AlkA Methyltransferase DNA glycosylase

a b s t r a c t The ada operon of Mycobacterium tuberculosis, which encodes a composite protein of AdaA and AlkA and a separate AdaB/Ogt protein, was characterized. M. tuberculosis treated with N-methyl-N -nitro-Nnitrosoguanidine induced transcription of the adaA-alkA and adaB genes, suggesting that M. tuberculosis mount an inducible response to methylating agents. Survival assays of the methyltransferase defective Escherichia coli mutant KT233 (ada ogt), showed that expression of the adaB gene rescued the alkylation sensitivity. Further, adaB but not adaA-alkA complemented the hypermutator phenotype of KT233. Purified AdaA-AlkA and AdaB possessed methyltransferase activity. These data suggested that AdaB counteract the cytotoxic and mutagenic effect of O6 -methylguanine, while AdaA-AlkA most likely transfers methyl groups from innocuous methylphosphotriesters. AdaA-AlkA did not possess alkylbase DNA glycosylase activity nor rescue the alkylation sensitivity of the E. coli mutant BK2118 (tag alkA). We propose that AdaA-AlkA is a positive regulator of the adaptive response in M. tuberculosis. It thus appears that the ada operon of M. tuberculosis suppresses the mutagenic effect of alkylation but not the cytotoxic effect of lesions such as 3-methylpurines. Collectively, these data indicate that M. tuberculosis hypermutator strains with defective adaptive response genes might sustain robustness to cytotoxic alkylation DNA damage and confer a selective advantage contributing to host adaptation. © 2011 Elsevier B.V. All rights reserved.

1. Introduction Methylating agents comprise a major class of DNA damaging agents that are found both endogenously and in the environment. Cells have evolved multiple DNA repair mechanisms to counteract the cytotoxic and mutagenic effect of methylation on DNA [1]. 3-Methylpurines (3mA and 3mG) and 1-methyladenine (1mA) are major cytotoxic lesions repaired by the base excision repair pathway (BER) and oxidative demethylation, respectively, while O6 -methylguanine (O6 -mG) is a major mutagenic lesion removed by methyltransferases (for review) [2,3]. Many bacteria counteract the deleterious effect of environmental exposure to alkylating agents by an inducible response termed

∗ Corresponding author at: Department of Microbiology, Oslo University Hospital, Rikshospitalet, PO Box 4950 Nydalen, NO-0424 Oslo, Norway. Tel.: +47 23074060; fax: +47 23074061. E-mail addresses: [email protected], [email protected] (M. Bjørås). 1568-7864/$ – see front matter © 2011 Elsevier B.V. All rights reserved. doi:10.1016/j.dnarep.2011.03.007

the adaptive response or Ada response [3]. In Escherichia coli four genes are induced by the adaptive response, ada, alkA, alkB and aidB, in which the methyltransferase Ada acts as a positive regulator of the operon. The 3mA DNA glycosylase AlkA is induced 10 fold when cells are exposed to sublethal doses of alkylating agents. A second constitutively expressed 3mA DNA glycosylase Tag is the major alkylbase DNA glycosylase activity in E. coli under normal growth conditions. The AlkB protein is an Fe (II)-dependent dioxygenase that repairs lesions such as 1mA and 3mC by oxidative demethylation [4,5]. AidB is a DNA binding protein predicted to catalyze direct repair of alkylated DNA [3,6]. E. coli Ada is composed of two major domains: the N-terminal AdaA fold and the C-terminal AdaB fold. Methyl groups of the mutagenic and cytotoxic bases O6 -mG and O4 -methylthymine (O4 -mT) are transferred to the Cys321 residue residing in the 19 kDa AdaB domain [7]. The 20 kDa AdaA domain removes the methyl group of innocuous methylphosphotriesters (MPT) in DNA by transfer to its Cys38 residue. The methylation of Cys38 converts the Ada protein into a transcriptional activator with specific DNA binding affinity to genes containing the ada operator sequence in their promoters,

596

M. Yang et al. / DNA Repair 10 (2011) 595–602

including the ada gene itself. Unmethylated Ada protein inhibits transcriptional activation mediated by the methylated form [8]. Consequently, MPTs act as molecular sensors for fluctuating levels of DNA alkylation in bacteria [9]. Additionally, E. coli harbors a second constitutively expressed O6 -mG DNA methyltransferase, Ogt, which is homologous to the AdaB domain [10]. The adaptive response is conserved among many bacterial species, in which protein homologues have been identified from sequenced genomes. Notably, the domains of Ada and AlkA proteins exist in diverse combinations in different prokaryotes. Furthermore, the genomic organization of genes involved in the adaptive response differs between various species. In Mycobacterium tuberculosis, a domain fusion of AdaA with AlkA constitutes an operon with the ogt gene, here termed adaA-alkA and adaB, respectively. Searches in genome databases reveal that the AdaA-AlkA fusion seems to be as common as AdaA with AdaB in bacteria [3]. M. tuberculosis is probably the most widespread human pathogen and tuberculosis is the cause of more deaths in adults worldwide than any other infectious disease, with the highest incidence found in developing countries. This pathogen is a member of the M. tuberculosis complex, which groups the genetically highly conserved sub-species Mycobacterium bovis, Mycobacterium africanum, Mycobacterium canettii, Mycobacterium bovis BCG, Mycobacterium caprae, Mycobacterium pinnipedii and Mycobacterium microti. As a facultative intracellular pathogen, M. tuberculosis survives and replicates inside human macrophages. Accordingly, it is subjected to a hostile environment in which alkylating stress and reactive oxygen- and nitrogen radicals are produced that can induce deleterious effects, including DNA damage. Previous studies have suggested that sequence variations in several strains of M. tuberculosis have led to transient mutator phenotypes and host adaptation [4,11,12]. In this work we have characterized the ada operon of M. tuberculosis by biochemical analysis and functional complementation assays of E. coli mutants. Activity assays with purified recombinant proteins demonstrated that both AdaB and AdaA-AlkA possess DNA methyltransferase activity, whereas no 3mA DNA glycosylase activity could be detected. Furthermore, AdaB suppressed the hypermutator phenotype of the E. coli ada ogt mutant while AdaA-AlkA showed no effect. These results suggest that AdaB counteract the mutagenic effect of O6 -mG, whereas AdaA-AlkA possesses a methyltransferase activity that removes methyl groups from innocuous MPTs.

level of the housekeeping gene sigA was used as endogenous control to normalize the relative expression level of target genes. A standard curve of Ct value was generated with a serial dilution of sigA cDNA to confirm that the real-time PCR reactions were run in the linear range. Primer pairs (sequence details in Supplementary data Table S1) for qPCR were designed with the software Primer Express provided by Applied Biosystems.

2.3. E. coli strains, plasmids and DNA constructs E. coli ER2566 (New England Biolabs) was used for cloning and overexpression of M. tuberculosis AdaA-AlkA and AdaB. The E. coli strains AB1157 (wild type), BK2118 (tag alkA) [14] and KT233 (ada ogt) [15] were used in survival and mutagenesis experiments. The adaA-alkA and adaB genes were amplified by PCR using M. tuberculosis genomic DNA as template. In addition, the adaA and alkA domains of the adaA-alkA gene were amplified separately. All PCR fragments were cloned into the BamHI and PstI sites of pUC18 (New England Biolabs). The resulting plasmids were termed pUCadaA, pUC-alkA, pUC-adaB and pUC-adaA-alkA. Next, the ORFs of adaA-alkA and adaB were inserted into the BamHI and NdeI sites of pET28b (Novagen), which were termed pET-adaA-alkA and pETadaB. Primers used for cloning are listed in Supplementary data Table S1.

2.4. Survival and mutagenesis Overnight cultures were diluted 1000 fold in fresh LB medium with ampicillin (100 ␮g/ml) and grown until OD600 was 0.5. Isopropyl-␤-d-thiogalactopyranoside (IPTG) at a final concentration of 0.1 mM was added to the cultures and the cells were further grown until OD600 was1.0. The cells were washed in M9 buffer and subsequently incubated for 15 min or 30 min at 37 ◦ C in M9 buffer with various concentrations of MNNG (Aldrich) or methyl methanesulfonate (MMS, Sigma). Finally, cells were washed with M9 buffer and spread on LB plates with ampicillin (100 ␮g/ml) or rifampicin (50 ␮g/ml) at appropriate dilutions. Surviving colonies were counted after incubation at 37 ◦ C for one or two days.

2.5. Protein expression and purification 2. Materials and methods 2.1. Mycobacterial strains, growth condition, MNNG treatment and RNA isolation M. tuberculosis H37Rv was grown in Middlebrook 7H9 medium with ADC and 0.05% Tween 80 at 37 ◦ C while shaking until OD600 was 0.5. The cultures were divided into two equal volumes, of which one of the cultures was treated with 3 ␮M N-methyl-N -nitro-Nnitrosoguanidine (MNNG) for 60 min. The total RNA was isolated using RNeasy spin columns (Qiagen) as described in Olsen et al. [13]. 2.2. Quantitative RT-PCR The cDNA was synthesized from total RNA using the Highcapacity cDNA reverse transcription kit from Applied Biosystems (Foster City, CA) according to the manufacture’s protocol. The quantitative PCR reaction was performed using the StepOnePlusTM instrument (Applied Biosystems, Foster City, CA) with the Power SYBR® Green PCR Master Mix kit (Applied Biosystems, Warrington, UK). Each cDNA sample was analyzed in triplicate. The expression

Ten liters of E. coli ER2566 carrying plasmid pET-adaB or pETadaA-alkA were grown at 37 ◦ C in LB-kanamycin (50 ␮g/ml) until OD600 was 0.5. Next, IPTG was added to the cells at a final concentration of 0.25 mM and incubation was continued at 18 ◦ C over night. The cells were collected, washed once in cold water, resuspended in 150 ml extract buffer (300 mM NaCl, 50 mM Na2 HPO4 , pH 7.5) and run through the French Pressure Cell Press (SLM Aminco) at 12000 PSIG twice. The cell-free protein extract obtained after centrifugation (15,000 × g for 30 min at 4 ◦ C) was loaded on a 3 ml nickel agarose column (Qiagen) equilibrated with extract buffer. The column was washed with 10 ml extract buffer supplemented with 50 mM imidazole and subsequently eluted with extract buffer supplemented with 300 mM imidazole. The protein samples were subjected to SDS-PAGE and the protein bands corresponding to the expected size of AdaB and AdaA-AlkA were verified by mass spectrometry. Bradford protein assay (Bio-Rad) was used to measure protein concentration. To ensure that the purified recombinant protein was not contaminated with endogenous methyltransferases or alkylbase DNA glycosylase, we followed the same purification protocol with extracts prepared from ER2566 carrying the empty vector pET28b. No methyltransferase or alkyl DNA glycosylase activity could be detected in the fractions from the Ni-column.

M. Yang et al. / DNA Repair 10 (2011) 595–602

597

Fig. 1. (A) Schematic representation of chromosomal organization of the adaptive response genes in E. coli and the structural homologues in M. tuberculosis. The adaA, adaB and alkA genes are differently composed in E. coli and M. tuberculosis. Whereas adaA is fused with alkA in M. tuberculosis, adaA is fused to adaB in E. coli. The arrows indicate promoter regions. (B) Predicted methyltransferase functions of M. tuberculosis AdaA-AlkA (upper panel) and AdaB (bottom panel) based on homology to E. coli Ada and AdaB. The AdaA-AlkA protein is predicted to transfer methyl groups from methylphospotriesters of the DNA backbone to a conserved cysteine residue (Cys34) of the AdaA domain, suggesting that AdaA-AlkA is converted to a positive transcriptional regulator of the adaptive response to DNA alkylation damage in M. tuberculosis. The AdaB protein is predicted to transfer methyl groups from O6 -mG in DNA to its conserved Cys 126. dR = deoxyribose.

2.6. DNA methyltransferase assay

2.8. Homology modeling of M. tuberculosis AlkA

N-[3 H] methyl-N -nitrosourea (MNU; 1.5 Ci mmol−1 ) was used to prepare alkylated calf thymus DNA (6000 d.p.m. ␮g−1 DNA) [16]. The methyltransferase activity was assayed in a reaction buffer containing 90 mM Tris, 90 mM Borate, 2 mM EDTA (TBE buffer) at 37 ◦ C for 60 min. The reaction mixture contained 1.5 ␮g MNU treated calf thymus DNA and protein as indicated in a total volume of 3 ml. Proteins were precipitated by incubating with 100 ␮l of 10 mg/ml BSA (in TBE buffer) and 1 ml of 4 M perchloric acid (HClO4 ) for 30 min at 70 ◦ C, and centrifuged at 4000 rpm for 20 min. The pellet was washed with 4 ml of 1 M HClO4 and resolved in 0.4 ml of 0.1 M HCl. The samples were transferred into tubes containing ULTIMA GOLDTM MV scintillation liquid (Packard BioScience B. V., Netherlands), and the radioactivity was measured in a scintillation counter (Liquid Scintillation Analyzer, TRI-CARB 2900TR).

The model of the C-terminal AlkA domain of M. tuberculosis AdaA-AlkA was built using the Swiss-Model homology-modeling server [17], and is based on the crystal structure of E. coli AlkA without DNA (pdb-code 1mpg). M. tuberculosis AdaA-AlkA has a few inserts compared with E. coli AlkA (Supplementary data Fig. S1B), but they are all located at surface-exposed loops distant from the DNA binding groove.

2.7. Alkylbase DNA glycosylase assay The DNA glycosylase activity was assayed in a reaction buffer containing 50 mM MOPS pH 7.5, 1.0 mM EDTA, 5% glycerol, 1.0 mM DTT at 37 ◦ C for 30 min. The reaction mixtures contained 0.3 ␮g MNU treated calf thymus DNA (same substrate as for the DNA methyltransferase assay) and protein as indicated in a total volume of 50 ␮l. The reaction was stopped by adding 73 ␮l stop buffer (0.41 M NaAc, 0.027% carrier-DNA, 0.82 mg/ml BSA) and 350 ␮l 100% ethanol and kept at −80 ◦ C for 30 min. DNA was precipitated by centrifugation at 13,000 rpm for 15 min at 4 ◦ C. 350 ␮l of the supernatant was transferred to UITIMA GOLDTM MV scintillation liquid, and the radioactivity was measured in a scintillation counter (Liquid Scintillation Analyzer, TRI-CARB 2900TR).

3. Results 3.1. Organization of the ada operon in M. tuberculosis To defend against fluctuating concentrations of alkylating agents many bacteria mount an inducible response that enhances transcription of several genes protecting against alkylating stress. In E. coli, induced alkylation resistance is acquired by the expression of four genes, ada, alkA, alkB and aidB (Fig. 1A). Components homologous to the Ada and AlkA proteins in E. coli are fused in different combinations in other species. The evolutionary fusion of AdaA and AlkA appears to be as common as AdaA with AdaB. By example, the N-terminal domain of Ada (AdaA) is fused with AlkA in M. tuberculosis (Fig. 1A). Further, the ORF of AdaA-AlkA forms an operon that also includes a putative AdaB methyltransferase. The adaA-alkA and adaB (annotated as ogt in the genome sequence) [18] gene sequences overlap by four basepairs. Characteristic sequence signatures and residues are conserved in both AdaA-AlkA and AdaB. This includes the zinc finger of AdaA, the active site thiol of AdaA (Cys38 in E. coli, Cys34 in M. tuberculosis) and AdaB (Cys321 in E. coli, Cys126 in M. tuberculosis), the helix-hairpin-helix motifs (HhH) of AdaB and AlkA, the arginine finger of AdaB as well as the catalytic aspartic acid (Asp238 in E. coli, Asp441 in M. tuberculosis) of AlkA (Supplementary Fig. S1A–C). The predicted methyltransferase

598

M. Yang et al. / DNA Repair 10 (2011) 595–602

Relative mRNA level

40

*

time quantitative PCR on RNA samples collected before and after treatment with MNNG. The results showed that adaA-alkA and adaB expressions were 16 and 27 times upregulated, respectively, in response to MNNG, whereas alkB and aidB were moderately upregulated, 2.6 and 1.6 times, respectively (Fig. 2). It thus appears that an adaptive response for inducible repair of DNA alkylation damage at least include adaA-alkA and adaB of the ada operon in M. tuberculosis.

Control MNNG

30

* 20

10

*

*

alkB

aidB

0 adaA-alkA

adaB

Fig. 2. Adaptive response for inducible repair of DNA alkylation damage in M. tuberculosis. The cells were grown with or without (control) 3 ␮M MNNG for 60 min, the total RNA were isolated and the transcription levels for adaA-alkA, adaB, alkB and aidB were measured by quantitative real-time PCR. The housekeeping gene sigA was used as endogenous control to normalize the relative expression level of target genes. The data are presented as mean of six measurements from three independent experiments with standard deviation. *P < 0.01.

activities of M. tuberculosis AdaA-AlkA and AdaB are illustrated in Fig. 1B, in which Ada-AlkA and AdaB transfer methyl groups from MPT and O6 -mG in DNA, respectively. 3.2. Ada response for inducible repair of DNA alkylation damage in M. tuberculosis Regulation of putative alkylation resistance genes, such as adaAalkA, adaB, alkB and aidB, in M. tuberculosis were examined by real

3.3. Functional complementation analysis of the E. coli mutants BK2118 alkA tag and KT233 ada ogt with the M. tuberculosis ada operon To elucidate the characteristics of the ada operon of M. tuberculosis, plasmid vectors (pUC18) containing adaA-alkA or adaB genes were transformed into E. coli mutants lacking the alkylbase DNA glycosylases Tag and AlkA (BK2118) or methyltransferases Ada and Ogt(KT233) for functional analysis. Expression of adaB suppressed the MNNG sensitivity of the KT233 mutant (Fig. 3A and B), whereas adaA-alkA expression showed no effect on survival (Fig. 3A). This data suggested that AdaB possesses O6 -mG transferase activity. Moreover, the mutation frequency, measured as rifampicin resistance, of KT233 expressing AdaB was reduced below the wild type level (Fig. 3C), supporting the survival data. In contrast, expression of AdaA-AlkA showed no change in the mutation frequency of KT233, indicating that the putative AdaA domain of AdaA-AlkA is not involved in removal of the mutagenic O6 mG lesion. Surprisingly, survival experiments with BK2118 cells expressing AdaA-AlkA or the AlkA domain alone plated on media containing the alkylating agent MMS showed no suppression of

Fig. 3. Functional complementation of the methyltransferase deficient E. coli mutant KT233 (ada ogt) with M. tuberculosis adaA-alkA and adaB. (A) Wildtype E. coli carrying pUC18 and KT233 cells carrying pUC18, pUC18-adaA-alkA or pUC18-adaB were treated with or without (control) 680 ␮M MNNG for 30 min at 37 ◦ C. Serial dilutions (1–10−4 ) were spotted on LB plates and incubated at 37 ◦ C. Spot formation was visualized after one day of incubation. (B) E. coli wild type carrying pUC18 and KT233 mutant cells carrying pUC18 or pUC18-adaB were exposed to increasing concentrations of MNNG for 15 min, spread on LB plates and incubated at 37 ◦ C. Surviving colonies were counted after two days of incubation. Each point is the mean three independent experiments with standard deviation. (C) E. coli wild type carrying pUC18 and KT233 mutant cells carrying pUC18, pUC18-adaA-alkA or pUC18-adaB were treated with 7 ␮M MNNG for 15 min at 37 ◦ C, serially diluted and plated on LB plates with or without 50 ␮g/ml rifampicin. The number of rifampicin resistant mutants were counted after two days of incubation and related to the total number of cells. The data are presented as mean of three independent experiments with standard deviation. *P < 0.01 when compared to the KT233 strain carrying pUC18.

Wild type/pUC18

BK2118/pUC18

BK2118/pUC-alkA

BK2118/pUC-adaA-alkA

Released methylated bases (fmol)

M. Yang et al. / DNA Repair 10 (2011) 595–602

Survival (%)

100 10 1 0.1 0.01

599

100

75

50

25

0 AdaA-AlkA

0.001 0.5

1

1.5

MMS (µM) Fig. 4. Functional complementation of the DNA glycosylase deficient E. coli mutant BK2118 (tag alkA) with M. tuberculosis adaA-alkA. E. coli wildtype cells carrying pUC18 and BK2118 carrying pUC18-adaA-alkA, pUC-alkA or pUC18 were treated with increasing concentrations of MMS for 15 min at 37 ◦ C. Surviving colonies were counted after two days of incubation. The data are presented as mean of three independent experiments with standard deviation.

alkylation sensitivity (Fig. 4), indicating that the AlkA domain of AdaA-AlkA is not involved in repair of cytotoxic N-alkylated bases such as 3mA and 3mG. 3.4. Analysis of methyltransferase and 3-methyladenine DNA glycosylase activity

Transferred methyl (fmol)

Recombinant M. tuberculosis AdaA-AlkA and AdaB proteins fused with N-terminal His tags were purified to homogeneity from E. coli. To examine DNA methyltransferase activity and removal of alkylated bases, DNA treated with [3 H]-labeled MNU was incubated with protein extracts or purified enzymes. The methyltransferase

40

A

Fig. 6. Alkylbase DNA glycosylase activity of recombinant AdaA-AlkA and AdaB. Each protein sample (100 ng AdaA-AlkA or AdaB) was incubated with [3 H]-MNU treated DNA at 37 ◦ C for 30 min, and the DNA was precipitated with ethanol. The radioactivity of the supernatant (released bases) was measured by a liquid scintillation counter. The DNA glycosylase AlkD from Bacillus cereus was included as positive control. The data are presented as mean of three measurements with standard deviation. *P < 0.01.

activity was monitored as the amount of label present in the protein fraction. Methyltransferase assays with protein extracts prepared from cells overexpressing AdaA-AlkA and AdaB showed significantly higher transfer of methyl groups as compared to the control extract (only vector) (Fig. 5A). Next, methyltransferase assays with increasing amounts of purified enzymes demonstrated that methyl residues were transferred to both AdaA-AlkA and AdaB (Fig. 5B and C). Finally, we tested 3mA DNA glycosylase activity of the purified proteins; however, no excision of alkylated bases could be detected in AdaA-AlkA or AdaB proteins (Fig. 6). In order to exclude that the N-terminal His-tag interfere with 3mA DNA glycosylase activity, we tested activity in extracts from BK2118 carrying plas-

*

30

*

20 10 0 Vector

pET-adaA-alkA

25

B

20 15 10 5 0 0.0

pET-adaB

1.0

E. coli extract Transferred methyl (fmol)

AlkD

Enzyme

2

Transferred methyl (fmol)

0

AdaB

2.0

3.0

AdaA-AlkA (pmol)

30

C

25 20 15 10 5 0 0.0

1.0

2.0

3.0

AdaB (pmol) Fig. 5. DNA methyltransferase activity of recombinant AdaA-AlkA and AdaB. The protein samples were incubated with [3 H]-MNU treated DNA at 37 ◦ C for 60 min, precipitated with perchlorid acid and resolved in HCl. The methyltransferase activity was measured by liquid scintillation counting of the label in the protein fraction. (A) Methyltransferase activity of cell-free extract (10 ␮g) prepared from E. coli KT233 carrying pUC18, pUC18-adaA-alkA or pUC18-adaB. (B) and (C) Methyltransferase activity of purified recombinant AdaA-AlkA and AdaB at increasing concentrations. The data are presented as mean of three measurements with standard deviation. *P < 0.01.

600

M. Yang et al. / DNA Repair 10 (2011) 595–602

Fig. 7. Modeling the base binding pocket in M. tuberculosis AdaA-AlkA and comparison with E. coli 3-metyl adenine DNA glycosylases I (AlkA) and II (Tag). (A) Crystal structure of active site in E. coli AlkA with abasic DNA (AP-site). Modeling of 3mA in this pocket suggests that the positively charged, alkylated base is stacking face-to-face with Trp272 and edge-on with Tyr222. (B) Model of the corresponding pocket in M. tuberculosis AdaA-AlkA. The edge-on stacking residue Tyr222 in E. coli AlkA is replaced by Val425. Several other aromatic residues in AlkA are replaced by smaller amino acid residues in M. tuberculosis AdaA-AlkA: Tyr273 by Thr480, Trp14 by Gly214, Phe18 by His218 and Tyr239 by Leu442. These sequence changes could result in a more flexible nucleotide binding pocket with less affinity for rigid and planar nucleotides. (C) Crystal structure of E. coli Tag in complex with abasic DNA and free 3mA illustrating aromatic stacking as a mechanism of recognition of alkylated bases in DNA [28].

mids pUC18 or pUC18-Ada-AlkA. The pUC18-Ada-AlkA construct expresses the native protein, however, no 3mA DNA glycosylase activity could be detected (data not shown). In view of the survival and mutagenesis data together with the results of the enzyme activity assays, it appears that AdaA-AlkA and AdaB transfer methyl groups from MPT’s and O6 -mG, respectively. However, direct evidence that Ada-AlkA transfer MPT’s is lacking. None of the proteins showed alkylbase DNA glycosylase activity. 3.5. Comparative modeling of AdaA-AlkA In order to understand the molecular basis for the observed absence of 3mA glycosylase activity of M. tuberculosis AdaA-AlkA, a homology model of the AlkA-like domain was built using the canonical E. coli AlkA as a template. The C-terminal AlkA domain of M. tuberculosis AdaA-AlkA shares ∼32% sequence identity with E. coli AlkA (Supplementary Fig. S1B). A detailed comparison of the model of M. tuberculosis AdaA-AlkA with E. coli AlkA reveal several sequence discrepancies in residues close to the alkylbase recognition pocket (Fig. 7A and B). N-alkylated bases bear a partial positive charge, and it has been suggested that a key element in the specific recognition of alkylated bases over native bases involve favourable stacking between the positively charged alkylated base and electron rich, aromatic protein residues. Structures of the E. coli 3mA DNA glycosylases AlkA and Tag clearly demonstrate this (Fig. 7A and C). The residues Trp272 and Tyr222 in AlkA are thought to be directly involved in such stacking [19], whereas in Tag, the free 3mA stack with Trp46 (Fig. 7C). Specifically, our model of AdaAAlkA shows that Tyr222 in AlkA is replaced by Val425, which probably reduces the affinity toward charged purine substrates. Furthermore, compared with the tight stacking of several aromatic residues in the second and third coordination sphere around the pocket in AlkA, AdaA-AlkA contains small amino acid chains that could destabilize or reduce the rigidity of the base pocket. Although the catalytic residue Asp 441 is conserved in the AlkA domain it is less likely that this residue could obtain a favourable position for catalysis in lack of a proper recognition pocket. This computer model supports that the AlkA domain of AdaA-AlkA has lost the specificity for alkylated bases but retained the ability to bind DNA. 4. Discussion Previous works have suggested that sequence variations in the adaptive response genes of M. tuberculosis lead to transient hypermutators, which favors host adaptation [12,20]. Here, we have

characterized the ada operon of M. tuberculosis, which encodes a composite protein of AdaA and AlkA, and a separate AdaB protein. Exposure of M. tuberculosis to MNNG strongly increased transcription of adaA-alkA and adaB, demonstrating an inducible response to methylating agents. Functional complementation assays of E. coli ada ogt cells showed that only adaB reduced the spontaneous mutation frequency and rescued the MNNG sensitivity of the methyltransferase deficient mutant. Purified AdaA-AlkA and AdaB possessed methyltransferase activity on methylated DNA but no 3mA DNA glycosylase activity. These data suggested that the AdaB protein mediates a methyltransferase activity counteracting the cytotoxic and mutagenic effect of O6 -mG. Further we propose that AdaA-AlkA transfers methyl groups from innocuous methylphosphotriesters although direct evidence are lacking. It thus appears that the AlkA domain of AdaA-AlkA has no role in base removal of cytotoxic alkylation lesions such as 3-methylpurines but may have evolved a structural function in DNA recognition. Our data suggest that the main role of the adaptive response in M. tuberculosis is to suppress O6 -mG alkylation induced mutagenesis rather than removal of cytotoxic DNA lesions such as 3mA and 3mG. To our knowledge this is the first functional characterization of the AdaA-AlkA fusion protein family. Surprisingly, AdaA-AlkA possesses no alkylbase DNA glycosylase activity despite the amino acid conservation with the AlkA family. However, our structural predictions indicate that critical residues of the base recognition pocket are changed to residues repealing the specificity of the AlkA domain to alkylated bases but still retaining the ability to bind DNA. Consequently, we may speculate whether the AlkA domain of AdaA-alkA has evolved a function required for DNA binding and scanning the genome for methylphosphotriesters and operator sequences of the adaptive response genes. However, it remains to provide experimental evidence that AdaA-AlkA has a potential role as a regulator of the adaptive response in M. tuberculosis. Notably, the AdaA domain of AdaA-AlkA lacks the C-terminal HhH motif, which is present in AdaA proteins without a fusion partner such as the AdaA protein of B. subtilis [21]. However, the AlkA domain of AdaA-AlkA contain a HhH motif and several structural studies of DNA glycosylases and transferases demonstrate that the HhH motif is required for DNA binding (review by Dalhus et al., 2009) [2]. Therefore, we suggest that the entire AlkA fold of the AdaA-AlkA fusion protein has functionally replaced the C-terminal HhH motif of the AdaA fold. In addition to the AlkA domain of AdaA-AlkA, the M. tuberculosis genome encodes two putative alkylbase DNA glycosylases, Tag and Aag. The 3mA-DNA glycosylase Tag, which was first discovered in E. coli, is widespread in prokaryotes but not

M. Yang et al. / DNA Repair 10 (2011) 595–602

present in mammalian cells [22,23]. Aag was first characterized in mammalian cells [24], however, Aamodt et al. showed that Bacillus subtilis encodes a functional Aag DNA glycosylase [25]. Thus, it appears that M. tuberculosis harbors alternative DNA glycosylases, other than AlkA, to remove cytotoxic alkylated bases such as 3-methylpurines. Our work demonstrates that MNNG strongly upregulates expression of adaptive response genes such as adaA-alkA and adaB in M. tuberculosis. A previous study showed that the alkylating agent streptozotocin induced O6 -mG transferase activity in M. smegmatis, supporting our findings [26]. However, M. bovis showed no regulation of O6 -mG transferase activity by streptozotocin [26]. Interestingly, the sequence of the M. bovis adaA-alkA gene contains a stop codon at the end of the adaA sequence, separating the gene into two open reading frames for AdaA and AlkA. We may speculate if a functional AdaA-AlkA protein requires the AlkA domain to induce the adaptive response to alkylation damage in mycobacteria. To investigate this hypothesis it would be tempting to introduce a stop codon into the M. tuberculosis ada-alkA gene and test if this destroys induction of the methylation inducible genes. M. tuberculosis is expected to sustain significant levels of nitrosative stress that form potent DNA alkylating compounds during the course of an infection. Durbach et al. constructed a M. tuberculosis mutant strain lacking the ada operon, including adaAalkA and adaB, to address the consequences of alkylation damage [20]. The ada mutant showed a 100-fold increase in mutation frequency as compared to wild-type cells after treatment with MNNG, supporting that AdaB possesses a methyltransferase activity counteracting the mutagenic effect of O6 -mG. Moreover, the ada mutant displayed hypersensitivity to MNNG comparable to the hypersensitivity of E. coli KT233 (ada ogt) [20], demonstrating the cytotoxic effect of O6 -mG. Further, this result is in agreement with our biochemical data demonstrating that AdaA-AlkA of the Ada operon in M. tuberculosis possesses no DNA glycosylase activity toward cytotoxic alkylating products such as 3mA and 3mG. To examine whether the same M. tuberculosis mutant would be manifested in a growth phenotype in vivo, mice were infected with wild type and mutant strains and bacillary loads were counted in various organs [20]. However, no differences in bacillary loads were found in any organs of the mice infected with the mutant strain as compared to the wild-type control, indicating that permeation of nitrosative stress to the level of cytotoxic alkylated DNA base lesions is not significant in adaptive response-defective M. tuberculoses strains in vivo. In sum, those data and the present work strongly suggest that the adaptive response to alkylation damage is restricted to suppress mutagenesis in M. tuberculosis. Consequently, these findings indicate that pathogenic M. tuberculosis strains with a defective adaptive response to alkylation may confer a selective advantage contributing to adaptation to the host and environmental changes. In this context, a transient hypermutator phenotype that sustains the robustness to cytotoxic alkylation DNA lesions can secure cellular fitness. Notably, Ebrahimi-Rad et al. reported that nine out of 55 W-Beijing strains had a characteristic mutation on codon 37 (Arg to Leu) of the adaB gene [12]. The relative contribution of W-Beijing genotype strains to the current worldwide TB epidemic is increasing, suggesting that the adaptive response genes of the ada operon are important targets for further investigations [27]. Further, Nouvel et al. sequenced the ada operon in 55 multi-drug resistant (MDR) Central African Republic (CAR) strains and identified three strains with a mutation in adaB (codon15, Thr to Ser) and 11 strains with mutations in adaA-alkA (codon 12, Ile to Val; codon 79, Trp to AMBER; codon 337, Thr to Asn) [11]. Moreover, the same study report that 34 out of 194 non-MDR CAR strains carry adaAalkA mutations. In conclusion, our data underscore the importance of the adaptive response in mycobacterial species and suggest that

601

the methyltransferases are essential to suppress alkylation induced mutagenesis in M. tuberculosis. Funding This work was supported by grants from The Research Council of Norway to MB and the European Commission to TT. Conflict of interest The authors declare that there are no conflicts of interest. Appendix A. Supplementary data Supplementary data associated with this article can be found, in the online version, at doi:10.1016/j.dnarep.2011.03.007. References [1] B. Sedgwick, P.A. Bates, J. Paik, S.C. Jacobs, T. Lindahl, Repair of alkylated DNA: recent advances, DNA Repair (Amst.) 6 (2007) 429–442. [2] B. Dalhus, J.K. Laerdahl, P.H. Backe, M. Bjoras, DNA base repair—recognition and initiation of catalysis, FEMS Microbiol. Rev. 33 (2009) 1044–1078. [3] B. Sedgwick, T. Lindahl, Recent progress on the Ada response for inducible repair of DNA alkylation damage, Oncogene 21 (2002) 8886–8894. [4] P.O. Falnes, R.F. Johansen, E. Seeberg, AlkB-mediated oxidative demethylation reverses DNA damage in Escherichia coli, Nature 419 (2002) 178–182. [5] S.C. Trewick, T.F. Henshaw, R.P. Hausinger, T. Lindahl, B. Sedgwick, Oxidative demethylation by Escherichia coli AlkB directly reverts DNA base damage, Nature 419 (2002) 174–178. [6] M.S. Rohankhedkar, S.B. Mulrooney, W.J. Wedemeyer, R.P. Hausinger, The AidB component of the Escherichia coli adaptive response to alkylating agents is a flavin-containing, DNA-binding protein, J. Bacteriol. 188 (2006) 223–230. [7] B. Demple, B. Sedgwick, P. Robins, N. Totty, M.D. Waterfield, T. Lindahl, Active site and complete sequence of the suicidal methyltransferase that counters alkylation mutagenesis, Proc. Natl. Acad. Sci. U.S.A. 82 (1985) 2688–2692. [8] B.M. Saget, G.C. Walker, The Ada protein acts as both a positive and a negative modulator of Escherichia coli’s response to methylating agents, Proc. Natl. Acad. Sci. U.S.A. 91 (1994) 9730–9734. [9] T. Lindahl, B. Sedgwick, M. Sekiguchi, Y. Nakabeppu, Regulation and expression of the adaptive response to alkylating agents, Annu. Rev. Biochem. 57 (1988) 133–157. [10] M.C. Wilkinson, P.M. Potter, L. Cawkwell, P. Georgiadis, D. Patel, P.F. Swann, G.P. Margison, Purification of the E. coli ogt gene product to homogeneity and its rate of action on O6-methylguanine, O6-ethylguanine and O4methylthymine in dodecadeoxyribonucleotides, Nucleic Acids Res. 17 (1989) 8475–8484. [11] L.X. Nouvel, V.T. Dos, E. Kassa-Kelembho, J. Rauzier, B. Gicquel, A non-sense mutation in the putative anti-mutator gene ada/alkA of Mycobacterium tuberculosis and M. bovis isolates suggests convergent evolution, BMC Microbiol. 7 (2007) 39. [12] M. Ebrahimi-Rad, P. Bifani, C. Martin, K. Kremer, S. Samper, J. Rauzier, B. Kreiswirth, J. Blazquez, M. Jouan, S.D. van, B. Gicquel, Mutations in putative mutator genes of Mycobacterium tuberculosis strains of the W-Beijing family, Emerg. Infect. Dis. 9 (2003) 838–845. [13] I. Olsen, S.V. Balasingham, T. Davidsen, E. Debebe, E.A. Rodland, S.D. van, K. Kremer, I. Alseth, T. Tonjum, Characterization of the major formamidopyrimidine-DNA glycosylase homolog in Mycobacterium tuberculosis and its linkage to variable tandem repeats, FEMS Immunol. Med. Microbiol. 56 (2009) 151–161. [14] N.D. Clarke, M. Kvaal, E. Seeberg, Cloning of Escherichia coli genes encoding 3-methyladenine DNA glycosylases I and II 8, Mol. Gen. Genet. 197 (1984) 368–372. [15] M.M. Leclere, M. Nishioka, T. Yuasa, S. Fujiwara, M. Takagi, T. Imanaka, The O6methylguanine-DNA methyltransferase from the hyperthermophilic archaeon Pyrococcus sp. KOD1: a thermostable repair enzyme 1, Mol. Gen. Genet. 258 (1998) 69–77. [16] S. Riazuddin, T. Lindahl, Properties of 3-methyladenine-DNA glycosylase from Escherichia coli, Biochemistry 17 (1978) 2110–2118. [17] T. Schwede, J. Kopp, N. Guex, M.C. Peitsch, SWISS-MODEL: an automated protein homology-modeling server, Nucleic Acids Res. 31 (2003) 3381–3385. [18] S.T. Cole, R. Brosch, J. Parkhill, T. Garnier, C. Churcher, D. Harris, S.V. Gordon, K. Eiglmeier, S. Gas, C.E. Barry III, F. Tekaia, K. Badcock, D. Basham, D. Brown, T. Chillingworth, R. Connor, R. Davies, K. Devlin, T. Feltwell, S. Gentles, N. Hamlin, S. Holroyd, T. Hornsby, K. Jagels, A. Krogh, J. McLean, S. Moule, L. Murphy, K. Oliver, J. Osborne, M.A. Quail, M.A. Rajandream, J. Rogers, S. Rutter, K. Seeger,

602

[19]

[20]

[21] [22] [23]

M. Yang et al. / DNA Repair 10 (2011) 595–602 J. Skelton, R. Squares, S. Squares, J.E. Sulston, K. Taylor, S. Whitehead, B.G. Barrell, Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence, Nature 393 (1998) 537–544. T. Hollis, Y. Ichikawa, T. Ellenberger, DNA bending and a flip-out mechanism for base excision by the helix-hairpin-helix DNA glycosylase, Escherichia coli AlkA, EMBO J. 19 (2000) 758–766. S.I. Durbach, B. Springer, E.E. Machowski, R.J. North, K.G. Papavinasasundaram, M.J. Colston, E.C. Bottger, V. Mizrahi, DNA alkylation damage as a sensor of nitrosative stress in Mycobacterium tuberculosis, Infect. Immun. 71 (2003) 997–1000. F. Morohoshi, K. Hayashi, N. Munakata, Bacillus subtilis ada operon encodes two DNA alkyltransferases, Nucleic Acids Res. 18 (1990) 5473–5480. T. Lindahl, New class of enzymes acting on damaged DNA, Nature 259 (1976) 64–66. G. Evensen, E. Seeberg, Adaptation to alkylation resistance involves the induction of a DNA glycosylase, Nature 296 (1982) 773–775.

[24] L. Samson, B. Derfler, M. Boosalis, K. Call, Cloning and characterization of a 3methyladenine DNA glycosylase cDNA from human cells whose gene maps to chromosome 16, Proc. Natl. Acad. Sci. U.S.A. 88 (1991) 9127–9131. [25] R.M. Aamodt, P.O. Falnes, R.F. Johansen, E. Seeberg, M. Bjoras, The Bacillus subtilis counterpart of the mammalian 3-methyladenine DNA glycosylase has hypoxanthine and 1,N6-ethenoadenine as preferred substrates, J. Biol. Chem. 279 (2004) 13601–13606. [26] I. Wiid, R. Grundlingh, W. Bourn, G. Bradley, A. Harington, E.G. Hoal-van Helden, H.P. van., O(6)-alkylguanine-DNA alkyltransferase DNA repair in mycobacteria: pathogenic and non-pathogenic species differ, Tuberculosis (Edinb.) 82 (2002) 45–53. [27] J.R. Glynn, J. Whiteley, P.J. Bifani, K. Kremer, S.D. van, Worldwide occurrence of Beijing/W strains of Mycobacterium tuberculosis: a systematic review, Emerg. Infect. Dis. 8 (2002) 843–849. [28] A.H. Metz, T. Hollis, B.F. Eichman, DNA damage recognition and repair by 3methyladenine DNA glycosylase I (TAG), EMBO J. 26 (2007) 2411–2420.