DnaE2 Polymerase Contributes to In Vivo Survival and the Emergence of Drug Resistance in Mycobacterium tuberculosis

Cell, Vol. 113, 183–193, April 18, 2003, Copyright 2003 by Cell Press

DnaE2 Polymerase Contributes to In Vivo Survival and the Emergence of Drug Resistance in Mycobacterium tuberculosis Helena I.M. Boshoff,1,2 Michael B. Reed,1 Clifton E. Barry III,1,* and Valerie Mizrahi2,* 1 Tuberculosis Research Section Laboratory of Host Defenses National Institute of Allergy and Infectious Diseases National Institutes of Health Twinbrook II 12441 Parklawn Drive Rockville, Maryland 20852 2 MRC/NHLS/WITS Molecular Mycobacteriology Research Unit School of Pathology University of the Witwatersrand and National Health Laboratory Service Johannesburg South Africa

Summary The presence of multiple copies of the major replicative DNA polymerase (DnaE) in some organisms, including important pathogens and symbionts, has remained an unresolved enigma. We postulated that one copy might participate in error-prone DNA repair synthesis. We found that UV irradiation of Mycobacterium tuberculosis results in increased mutation frequency in the surviving fraction. We identified dnaE2 as a gene that is upregulated in vitro by several DNA damaging agents, as well as during infection of mice. Loss of this protein reduces both survival of the bacillus after UV irradiation and the virulence of the organism in mice. Our data suggest that DnaE2, and not a member of the Y family of error-prone DNA polymerases, is the primary mediator of survival through inducible mutagenesis and can contribute directly to the emergence of drug resistance in vivo. These results may indicate a potential new target for therapeutic intervention. Introduction Mycobacterium tuberculosis (MTb) is responsible for the largest number of deaths attributable to a single infectious organism, and fully one third of the world’s population is estimated to be infected with this pathogen (Dye et al., 2002). Control strategies for tuberculosis rely heavily on chemotherapy. As such, the development of drug resistance is a serious threat to any attempt to control this disease (Ramaswamy and Musser, 1998). Understanding the mechanisms by which this organism develops drug resistance would provide valuable insight that might lead to interventions designed to avert this emerging problem. Pathogenic Mycobacteria are exposed to many adverse conditions during their parasitism of the human *Correspondence: [email protected] (C.E.B.); mizrahiv@pathology. wits.ac.za (V.M.)

host. These conditions include such factors as nutritional deprivation, hypoxia, reactive oxygen or nitrogen species, and possibly other variable human bactericidal systems produced during immune surveillance, in addition to external factors such as anti-tubercular drug administration (Russell, 2001). Genetic mutation is one means of generating individual strains with enhanced fitness under conditions that threaten the continued survival of the bacterial population. Modulation of mutation rate by the stress-inducible bacterial SOS response facilitates the generation of mutants during adverse conditions and is thought to occur through the induction of low fidelity, or “mutator,” polymerases (Goodman, 2002). Error-prone DNA replication or inaccurate repair of base-pair mismatches may contribute directly to the unexpectedly high rate of acquisition of drug resistance during infection with pathogens such as Pseudomonas sp. (Bjorkholm et al., 2001; Oliver et al., 2000). The emergence of multidrug resistant strains of MTb in TB patients appears to be more common than would be expected when considering the rarity of individual drug resistance mutations observed in vitro. While multiple mutations occur at a rate corresponding to the product of the individual mutation frequencies in vitro and should therefore be exceedingly unusual, multiple drug resistance occurs in between 1%–3% of the total tuberculosis isolates (Espinal et al., 2000). This suggests that in vitro mutation frequencies may not accurately reflect the mutation rate during parasitism of the human host. In vitro studies have indicated that the mutation rate of MTb when cultured under normal growth conditions is within the range of that reported for most other organisms (Boshoff et al., 2001; Mizrahi et al., 2000 and references therein), but the effect of environmental stress on the mutational dynamics of this organism has not been investigated. The induction of the SOS response as a result of DNA damage during infection has been shown to increase the frequency of both targeted and untargeted mutagenesis in intracellular pathogens capable of SOS mutagenesis (Schlosser-Silverman et al., 2000). The SOS response has been characterized in both MTb and Mycobacterium smegmatis (MSm) and, as in other organisms, RecA and LexA appear to regulate this response in Mycobacteria (Davis et al., 2002a; Durbach et al., 1997; Brooks et al., 2001). Although several genes have been identified that are under LexA control, LexA/ RecA-independent regulatory mechanisms may also exist for other damage-inducible genes in Mycobacteria (Davis et al., 2002a; Durbach et al., 1997; Brooks et al., 2001). In E. coli, stress-induced increases in mutational frequency are thought to result from the action of PolB and two newly discovered, Y family DNA polymerases encoded by umuC and dinP, which form part of the SOS regulon and are capable of mutagenic translesion synthesis (TLS) (Napolitano et al., 2000; Maor-Shoshani et al., 2000; Pham et al., 2001). Surprisingly, however, the MTb homologs of dinP and dinX were not upregulated after DNA damage by mitomycin C (MMC) treatment (Brooks et al., 2001). Although the main replicative DNA polymerase (Pol III) of E. coli can contribute to TLS

Cell 184

Figure 1. Effect of UV Irradiation on Survival and Mutation Frequency of MSm and MTb (A) Dilutions of MSm strains were plated in triplicate on LA and irradiated at different energy fluences. Diamond, WT (mc2155); circle, mc2155(pOmsmE2); square, dnaE2::aph; triangle, dnaE2::aph attB::dnaE2. (B) Log-phase MSm cells were irradiated and allowed to recover for different times before plating on Rif plates. (C) MTb strains were irradiated and allowed to recover for different times before plating on Rif plates. Diamond, WT H37Rv; square, dnaE2::hyg; triangle, dnaE2::hyg attB::dnaE2. The results shown are from one representative experiment.

in the absence of a proofreading function, the physiological relevance of this pathway remains to be established, and upregulation of the gene encoding dnaE was likewise not reported following DNA damage (Borden et al., 2002). MTb contains two apparently functionally redundant replicative DNA polymerases, designated DnaE1 and DnaE2. One possible explanation for this redundancy is that error-prone DNA repair may be the major function of one of these enzymes. To gain insight into mutational dynamics in Mycobacteria, we investigated whether MSm and MTb were capable of induced mutagenesis in response to DNA damage, explored the genetic basis for this phenomenon, and established the relevance of this process to in vivo growth and the development of drug resistance.

lead to an observed increase in mutation frequency, MTb and MSm were exposed to hydrogen peroxide (H2O2) for 24 hr before plating on Rif. This oxidizing treatment also resulted in mutation induction (data not shown). To determine whether the increase in mutation frequency after DNA damage in MSm was mediated by the SOS response and therefore recA-dependent, UVinduced mutagenesis was assessed in a recA mutant (Frischkorn et al., 1998). In contrast to the induced mutagenesis observed in wild-type (WT) MSm, UV treatment did not result in an increased recovery of Rif resistant mutants in the recA mutant (Table 1), confirming that this process was recA-dependent.

Results

Identification of DNA Damage-Inducible Genes by Whole-Genome Expression Profiling Since the complete genome sequence is available for Mtb strain H37Rv (http://genolist.pasteur.fr/Tuberculist), whole-genome transcriptional profiling was performed by microarray analysis after exposure of this organism to various forms of DNA damage in an attempt to identify the gene(s) responsible for induced mutagenesis. Expression of 158, 157, and 197 genes was significantly affected by DNA damage using UV irradiation, mitomycin C (MMC), and H2O2 treatment, respectively. Seventyeight genes were regulated in common in response to all three agents (Table 2). The predicted functions of many of these induced genes were consistent with DNA repair, metabolism, or genome plasticity and several were counterparts or homologs of genes induced by DNA damage in E. coli (Courcelle et al., 2001; Khil and Camerini-Otero, 2002). One notable observation was the widespread induction of transposases, resolvases, and insertion elements by DNA damage. Members of the 13E12 repeat family have recently been reported to be induced by MMC treatment (Davis et al., 2002a). None of the DNA-damaging treatments resulted in induction of the dinP and/or dinX genes. The only recognizable DNA polymerase-encoding gene upregulated following UV irradiation was dnaE2.

DNA Damage Results in a recA-Dependent Increase in Mutation Frequency To explore whether a damage-inducible mutagenesis system was operable in Mycobacteria, we used UV irradiation to induce DNA damage, a technique that has been studied in considerable detail in organisms such as E. coli (Rangarajan et al., 2002; Courcelle et al., 2001). Direct exposure of Mycobacteria to UV light is highly lethal (Figure 1A), but within the surviving bacterial population there was a 20- to 50-fold increase in the frequency of appearance of Rifampicin (Rif)-resistant mutants in both MSm and MTb (Table 1 and Figures 1B and 1C). The number of mutants obtained was found to be dependent on the time allowed for recovery following UV exposure, and maximal numbers of mutants were obtained after 3–4 (MSm) and 24–36 (MTb) hr of growth, respectively (Table 1 and Figures 1B and 1C). Similarly, a 16-fold increase in streptomycin (SM)-resistant mutants of MSm was also obtained 90 min after DNA damage. The types of DNA damage that MTb is predicted to sustain during infection are difficult to predict (Boshoff et al., 2001) and are unlikely to be comprehensively represented by the lesions created by UV irradiation. To assess whether other forms of DNA damage could also

dnaE2 and Inducible Mutagenesis in M. tuberculosis 185

Table 1. Mutation Frequencies of Mycobacterial Strains before and after UV Irradiation Strain M. smegmatis mc2155 SMR5 dnaE2::aph recA::hyg dnaE2::aph attB::dnaE2 mc2155 (pOmsmE2) M. tuberculosis H37Rv dnaE2::hyg dnaE2::hyg attB::dnaE2 a

SM Mutation frequency ⫻ 108 Untrtd 3.0 ⫾ 1

UV (1.5h)a 40 ⫾ 10

3.5 ⫾ 2

3.7 ⫾ 1.5

6 ⫾ 2.5 2⫾1

40 ⫾ 20 37 ⫾ 19

Rif Mutation frequency ⫻ 108 Untrtd 2.0 ⫾ 1 7⫾2 7⫾1 2.5 ⫾ 1.5

UV (1h) 4.5 ⫾ 1

UV (1.5h) 20 ⫾ 10 30 ⫾ 10 2⫾1

4.5 ⫾ 2 5⫾2 4⫾1

UV (3h) 80 ⫾ 15

UV (12h)

8 ⫾ 0.5 3 ⫾ 1.7

100 ⫾ 20 4⫾1 120 ⫾ 50

Bacterial cells were recovered for the indicated amount of time following UV treatment.

The Kinetics of DNA Damage-Induced Expression of dnaE2 The observed induction of dnaE2 by microarray hybridization was confirmed by real-time, quantitative RT-PCR (QRT-PCR) analysis using the recA gene as a damageinducible control (Figure 2). Expression of recA in response to DNA damage closely mimicked that reported by others (Brooks et al., 2001), and dnaE2 expression closely paralleled that of recA in both MSm and MTb. Although dnaE2 and recA were induced by H2O2 (data

not shown), the absolute induction levels were highly variable between experiments, probably due to the cytotoxicity of this treatment. In the recA mutant strain of MSm, dnaE2 levels were unaltered over the time course of the experiment (Figure 2C), confirming that dnaE2 is solely LexA-regulated. Finally, QRT-PCR analysis was used to confirm that the other replicative polymerase, dnaE1, was not induced by any of the DNA-damaging treatments as suggested by the microarray experiments (data not shown).

Table 2. Top DNA Damage Induced Genes Involved in DNA Modificationa Rv no.

UV

MMC

H2O2

Gene product

Rv0336, Rv0515, Rv2100, Rv1702c Rv2015c, Rv1765c, Rv1128c, Rv3466, Rv1588 Rv0393, Rv1587c Rv0094c Rv3827, Rv0606, Rv0829, Rv3386 Rv2978c, Rv3776, Rv2791c, Rv0922, Rv1764, Rv0796, Rv3475, Rv2479c Rv3187, Rv3185, Rv1756c, Rv2649, Rv1369, Rv2355, Rv2167c, Rv3326 Rv2792c, Rv2979c, Rv0605, Rv3828c Rv3074 Rv2737c Rv2736c Rv3296 Rv3585 Rv0058 Rv1317c Rv1277 Rv3202c Rv1638c Rv2024 Rv2593c Rv2594c Rv1633 Rv3394c Rv3856c Rv3198c Rv1379 Rv3370c Rv2720 Rv2902c Rv1316c Rv0054

⫹ ⫹

⫹ ⫺

⫹ ⫹

13E12 repeat family (HNH endonuclease and putative transposase domains) 13E12 repeat family (HNH endonuclease and putative transposase domains)

⫹

⫺

⫹ ⫹

⫹ ⫺

⫺ ⫹ ⫹ ⫹

13E12 repeat family (HNH nuclease and/or putative transposase domain) 13E12 repeat family (contains transposase domain) Transposases or transposase fragments Transposases

⫺

⫺

⫹

Transposases

⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺

⫺ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫹ ⫺ ⫺ ⫺ ⫺

⫹ ⫹ ⫹ ⫹

Resolvases Contains HNH nuclease domain recA, RecA recX, Regulatory protein lhr, Probable ATP-dependent helicase radA, DNA repair dnaB, Replicative DNA helicase alkA, 3-methyladenine DNA glycosidase II Possible DNA repair exonuclease (SbcD family) UvrD/Rep DNA helicase domain uvrA, Excinuclease ABC subunit A Contains DEAD-like helicase domain ruvA, Holliday junction binding protein ruvC, Holliday junction resolvase uvrB, Excinuclease ABC subunit B Contains IMS family domain involved in UV protection DNA polymerase X family and phosphoesterase family domains uvrD2, DNA helicase pyrR, Pyrimidine biosynthesis regulatory protein dnaE2, DNA polymerase III ␣ chain lexA, LexA repressor rnhB, RNase HII ogt, Methylated⫺DNA⫺protein⫺cysteine methyltransferase ssb, Single⫺stranded DNA binding protein

⫹ ⫹ ⫹ ⫹ ⫹ ⫹ ⫺ ⫺ ⫹ ⫹

⫺ ⫺

⫺ ⫺

⫺ ⫹ ⫹

⫺ ⫹ ⫹

a Changes in mRNA levels in response to exposure to UV irradiation, MMC or H2O2 were monitored using DNA microarrays. Genes were included if their expression levels were increased 2-fold or more in at least 4 microarray experiments in response to the DNA-damaging treatment relative to an untreated control. Genes are annotated as described in the Tuberculist database (http://genolist.pasteur.fr/Tuberculist).

Cell 186

Figure 2. Analysis of Gene Expression Levels after DNA Damage Kinetics of induction of dnaE2 (filled triangle) and recA (square) in MTb H37Rv by UV (A) and MMC treatment (B). MTb gene expression levels were normalized relative to the gnd gene with normalization relative to the sigA gene giving similar results. Kinetics of the transcriptional response of MSm mc2155 dnaE2 (filled triangle), recA (square), and dnaE2 of the MSm recA::hyg mutant (open triangle) after (C) UV irradiation, and (D) MMC treatment. M. smegmatis gene expression levels were normalized relative to sigA.

In MTb, expression of recA remained elevated even 72 hr after UV damage, presumably due to the persistence of cross-linked DNA lesions. MSm recA levels similarly remained elevated 6 hr after irradiation (data not shown). Peak expression of dnaE2 and recA occurred within 60 min of UV irradiation in MSm (Figure 2C and data not shown). For MMC treatment, peak expression of these genes occurred 90 min after initiation of DNA damage, the longer time for induction possibly reflecting the time required for the drug to penetrate the cell wall prior to inducing a mutagenic effect. MMC was present during the duration of the time over which RNA was isolated from MSm so that gene expression levels remained upregulated throughout the treatment period. In contrast, MTb was treated for 90 min followed by removal of the drug. In this experiment the level of recA and dnaE2 mRNA expression decreased 36 hr after the initial addition of the drug. Peak expression levels of dnaE2 and recA were observed 20–36 hr post-treatment in MTb (Figures 2A and 2B). DNA Damage-Induced Mutagenesis Is Mediated Primarily by DnaE2 Allelic exchange mutants of MSm and MTb in which the dnaE2 gene was disrupted by an antibiotic resistance cassette were constructed. Loss of dnaE2 function had no discernable effect on in vitro growth kinetics in either species when assessed in pure culture or in cocultures seeded with a 100-fold excess of either WT or mutant strain (data not shown). However, damage-induced en-

hancement of the mutation frequency to Rif and/or SMresistance was completely lost in both MSm and MTb mutants (Table 1, Figure 1C, and data not shown). The damage-induced mutator phenotype could be restored in response to both H2O2 and UV treatment in the MSm dnaE2::aph and MTb dnaE2::hyg mutants by complementation with a copy of the dnaE2 gene integrated at the attB site, confirming that induced mutagenesis was strictly dependent upon the dnaE2 gene product (Table 1 and data not shown). A mutation rate estimation based on the Luria-Delbru¨ck fluctuation assay confirmed that the number of mutations per cell per generation of the MTb WT, dnaE2::hyg, and dnaE2::hyg attB::dnaE2 strains were similar under normal in vitro growth conditions (2–3 ⫻10⫺10 mutations per cell per generation). The apparent mutation rate of the WT and complemented strains were enhanced by UV irradiation as expected but application of such protocols to pulsed UV irradiation treatment is not straightforward. The mutation rate of the dnaE2 knockout mutant was unaffected by DNA damage after UV irradiation (Supplemental Data available at http://www.cell.com/cgi/content/full/113/2/ 183/DC1) confirming its role in DNA repair. MSm overexpressing dnaE2 from the multicopy plasmid pOmsmE2 did not show increased mutation frequencies under noninduced conditions (Table 1). QRTPCR analysis confirmed that dnaE2 expression levels were increased 10- to 15-fold in this strain (data not shown) compared to the parental WT (data not shown). Similarly, induction of DNA damage by UV irradiation enhanced dnaE2 levels 40- to 45-fold within 60 min in this strain but this failed to increase the observed mutation frequency (Table 1). The MTb and MSm dnaE2 mutant strains were also hypersensitive to UV radiation (Figure 1A and data not shown). The UV sensitivity of the MSm mutant was restored to WT levels by complementation with a functional copy of dnaE2. This phenotype parallels the damage hypersensitivity of Bacillus subtilis yqjH (Sung et al., 2003) and E. coli ⌬umuDC and ⌬umuDC ⌬polB mutants (Rangarajan et al., 1999). DnaE2 Is a Probable Translesion Polymerase Genotypic analysis of Rif-resistant MTb mutants resulting from exposure to UV irradiation revealed that there were significant DnaE2-dependent differences in the types of mutations occurring in the rpoB gene of MTb and MTb dnaE2::hyg. Following UV irradiation, the WT strain gave rise to a high proportion of mutants (35.6%) containing a double CC→TT transition resulting in the simultaneous change of two codons (ACCCAC→ACTTAC), only one of which results in an amino acid substitution (Table 3). This mutation presumably reflects error-prone TLS across a cyclobutane pyrimidine dimer or a pyrimidine-pyrimidone (6-4) photoproduct. This characteristic mutation was never observed in the MTb dnaE2::hyg strain or in non-UV generated MTb Rif-resistant mutants. The TCG→TTG mutation that results in a Ser531Leu substitution is the most common substitution observed in clinical and in vitro generated Rif-resistant isolates (Morlock et al., 2000). In our hands this was also the most frequent mutation observed in spontaneous mutants arising from the untreated WT strain. However, in

dnaE2 and Inducible Mutagenesis in M. tuberculosis 187

Table 3. rpoB Mutations in Rif-Resistant Mutants Derived from MTb* WT

dnaE2::hyg

Mutation

Substitution

UV (45)

⫺UV (30)

⫹UV (46)

⫺UV (30)

CAC→TAC CCAC→TTAC TCG→TTG GCC→GTC CAC→CGC TCG→TTG CAG del. CAC→GAC CAC→CGC CAG→CTG AAC→AAA TTCATG ins TCG→CAG TCG→TGG

His526→Tyr ThrHis526→ThrTyr Ser531→Leu Ala500→Val His526→Arg Ser522→Leu Gln516 del. His526→Asp His526→Pro Gln516→Leu Asn519→Lys PheMet after Met515 Ser531→Gln Ser531→Trp

40% 35.6% 22.2% 2.2% 0% 0% 0% 0% 0% 0% 0% 0% 0% 0%

20% 0% 40% 0% 13.3% 10% 0% 6.7% 0% 0% 0% 3.3% 3.3% 3.3%

17.4% 0% 32.6% 0% 15.2% 19.6% 2.2% 4.3% 4.3% 2.2% 2.2% 0% 0% 0%

16.7% 0% 56.7% 0% 0% 3.3% 3.3% 13.3% 0% 0% 0% 0% 0% 6.7%

* The number of mutants recovered from UV-treated samples (⫹) or from untreated controls (⫺) is given in brackets. The % killing during irradiation of the liquid culture was ca. 1% for WT and 5⫺10% for dnaE2::hyg.

mutants recovered from UV irradiated cells, the predominant mutations were the CAC→TAC (His526Tyr), followed by the double transition mutation followed by the Ser531Leu mutation. Interestingly, all of the Rif-resistant mutants recovered from UV irradiated WT MTb contained C→T transitions in their rpoB genes. The mutational spectrum observed was, however, dependent upon the nature of the DNA insult used, with H2O2 treatment giving different predominant mutations than UV treatment (data not shown). dnaE2 Plays a Role in In Vivo Persistence and in the Emergence of Drug Resistance To examine the role of dnaE2 during host infection, QRTPCR was used to analyze RNA prepared from bacilli isolated from the lungs and spleens of aerosol infected mice 3-, 6- and 8-weeks following infection. Expression levels could not be reliably assessed at earlier stages of the infection as sufficient quantities of mycobacterial RNA could not be recovered from the relatively low numbers of bacilli present during the early, exponential growth phase. Isocitrate lyase, encoded by the icl gene, has been suggested to play a specific role in persistence of MTb infections and its expression has been shown to be upregulated during prolonged murine infection (McKinney et al., 2000; J. McKinney, personal communication). We therefore simultaneously evaluated icl and dnaE2 expression levels normalized relative to the sigA gene by quantitative real-time PCR. icl was found to be upregulated by a factor of 15 ⫾ 5-fold within a similar time frame as dnaE2, which was upregulated 6 ⫾ 3-fold in bacilli recovered from 3-, 6-, and 8-week infected mouse organs (compared to in vitro grown cultures, n ⫽ 2 mice per time point, 6 mice total). However, given the limited number of mice used per time point, the time course of upregulation of gene expression in the lungs of infected mice could not be accurately determined. Growth and persistence of the dnaE2 knockout strain of MTb was also assessed following infection of mice with 10–100 bacilli by aerosol. Initial growth rates and peak counts of bacteria prior to the control of proliferation by the acquired immune response (Flynn and Chan,

2001) were identical for both strains in the lungs of these mice (Figure 3A) and visible histopathology appeared very similar (data not shown). Nine months following infection, colony counts from the lungs of mice infected with the dnaE2 mutant were 10-fold lower (7.1 ⫾ 0.2 ⫻ 105) than colony counts from lungs of mice infected with the parental strain (1 ⫾ 0.3 ⫻ 107). More dramatically, the median survival time of dnaE2::hyg infected mice was 384 days compared to 222 days for WT infected mice (Figure 3B). In a separate infection, D2B6F1 mice, which succumb more quickly to MTb infection, were infected with WT H37Rv, dnaE2::hyg, and dnaE2::hyg attB::dnaE2 strains. Equivalent mortality (20%) was observed in the WT and complemented strains at 150 days while no mortality was observed in the dnaE2 mutant (data not shown). To evaluate the role of DnaE2 in the development of drug resistance, 180 C57BL/6 mice were infected over three separate experiments with either WT, dnaE2 knockout, or the dnaE2-complemented knockout strain and subjected to either high dose or low dose daily Rif treatment. Treatment was initiated when bacterial growth in the organs was controlled by the immune response. Enumeration of Rif-resistant bacteria sug-

Figure 3. Virulence of MTb Strains in C57BL/6 Mice (A) Lung CFU counts determined at different times post-infection for the WT and dnaE2::hyg strains, filled and open bars, respectively (n ⫽ 4 to 5 for each time point). (B) Time-to-death analysis in mice infected with WT (filled square) and dnaE2::hyg (open square) MTb strains. Ten mice were used for time-to-death studies.

Cell 188

Table 4. Role of dnaE2 in the Evolution of Drug Resistance In Vivo Lung CFU ⫻ 104

No. of RifR mutants

Rifampicin Treatment Protocol

WT

dnaE2::hyg

dnaE2::hyg attB::dnaE2

1

930 ⫾ 150 575 ⫾ 35 3.5 ⫾ 1.5

340 ⫾ 100 419 ⫾ 50 5.7 ⫾ 2

ND ND 4.4 ⫾ 2

25 mg/kg 10 mg/kg 4 10 mg/kg 3

WT

dnaE2::hyg

dnaE2::hyg attB::dnaE2

2

1 0 0

ND ND 1

8 18 5 10

ND, not determined; 1, two treatment cycles, homogenates pooled in groups of 10 and plated six weeks after termination of treatment; 2, obtained from three groups of mice (4, 3 and 1 mutants, respectively); 3, one treatment cycle with pooled lung homogenates plated 1 week after treatment; 4, one treatment cycle with individual lung homogenates plated immediately after termination of treatment; 5, all from one mouse.

gested that drug resistance emerged more frequently in the WT and complemented knockout strains than in the dnaE2 knockout mutant (Table 4). Sequence analysis of the rpoB Rif-resistance determining region of the 18 mutants recovered from WT infected mice from the first low-dose treatment experiment indicated that these were comprised of 4 different alleles (His526Tyr, Ala500Val, Gln513Arg, and no mutation) suggesting that these were the result of at least 4 independent mutational events. The reduction in the in vivo mutation frequency of the dnaE2 mutant inferred from these experiments was consistent with that observed for the reduction in induced mutagenesis in vitro. Discussion MTb maintains an often decades long subclinical infection in humans and may have acquired specific adaptations that enable this organism to maintain unusual genetic stability and flexibility in the absence of active replication. The extremely low rate of synonymous mutation in the absence of genetic selection has been a noted feature of this organism and contrasts with the inexplicably high frequency of emergence of drug resistance in vivo (Ramaswamy and Musser, 1998; Espinal et al., 2000). The underlying molecular mechanisms by which Mycobacteria induce genetic variation may be substantially different than those employed in many other bacteria. For example, Mycobacteria apparently lack the ability to induce the Y family of error prone DNA polymerases following DNA damage which has been used to suggest that this organism might be incapable of damage-induced mutagenesis (Brooks et al., 2001). This study clearly demonstrates that MTb (and MSm) elevates mutational levels in response to DNA damage but does so by a mechanism that does not involve the induction of known error-prone polymerases. To understand the molecular basis for the elevation of mutation rates in Mycobacteria challenged by DNA damage, we explored the global transcriptional responses of Mtb to DNA damage by UV, MMC, and H2O2. These experiments revealed both similarities and striking differences between the profiles of genes induced in MTb in response to various forms of DNA damage when compared to gram-negative organisms such as E. coli. The key regulator of the SOS response, RecA, was strongly induced in response to all three forms of DNA damage, corroborating previous work demonstrating that the Mycobacterial RecA was functionally similar

to its E. coli counterpart (Durbach et al., 1997; Frischkorn et al., 1998). As in E. coli (Courcelle et al., 2001; Khil and Camerini-Otero, 2002), genes encoding proteins involved in resolution of Holliday junctions and a subset of genes associated with nucleotide excision repair were induced by UV irradiation. The ssb gene was upregulated by MMC and H2O2 but not by UV treatment, suggesting that the type of lesion produced has an influence on transcription of lexA-independent repair systems (Davis et al., 2002a, 2002b). In E. coli, ssb is induced upon UV treatment while it is downregulated by MMC-induced damage but also appears to be regulated independently of lexA (Courcelle et al., 2001; Khil and Camerini-Otero, 2002). Many of the genes induced by DNA damage in E. coli (mutT, dut, nth, nfo, mfd, recN, cho, recF, recR, gyrA, gyrB, umuC, umuD, dinP) were not induced by DNA damage in MTb. In contrast, some genes (alkA, ogt, radA, lhr, dnaB, sbcD) were uniquely damage-induced in MTb. Finally, some induced genes of unknown function (Rv3202, Rv2024, and Rv3394c) lack obvious counterparts in E. coli but nonetheless contain helicase or IMS domains, linking them functionally with DNA repair. We did not observe induction of the mycobacterial dinP or dinX genes under any of these conditions, in accordance with previous reports (Brooks et al., 2001). Thus, members of the Y family of error-prone DNA polymerases in MTb are apparently not part of the stressinduced SOS regulon. Similarly, Rv2823c, a member of a novel family of predicted DNA polymerases implicated in DNA repair (Makarova et al., 2002), was not induced after DNA damage. However, one of the two putative replicative polymerases, dnaE2, was strongly upregulated in response to all forms of DNA damage suggesting that this gene might encode the polymerase activity responsible for damage-induced mutagenesis in Mycobacteria. This was confirmed by the loss of this phenotype following disruption of dnaE2 in both MSm and MTb. Inducible mutagenesis could be restored by complementation with a functional dnaE2, proving that the observed phenotype was not due to downstream polar effects caused by the dnaE2::aph or dnaE2::hyg disruptions. In both organisms, induction of dnaE2 was greatest after DNA damage by UV irradiation, which may reflect more extensive DNA distortion caused by this treatment. The dnaE2 gene encodes a homolog of the gramnegative ␣ subunit of the polymerase III holoenzyme that is responsible for the catalytic activity for DNA repli-

dnaE2 and Inducible Mutagenesis in M. tuberculosis 189

cation. In eubacteria, family C polymerases are essential for replicative DNA synthesis (Ito and Braithwaite, 1991). The class I family C DNA polymerase (dnaE) has been shown to be essential for DNA replication in all eubacteria, with the exception of cyanobacteria where removal of an intein from the precursor class III polypeptide reconstitutes the DnaE-type enzyme (Huang and Ito, 1999). Most gram-positive bacteria also require a class II family C polymerase (polC) for DNA replication (class II and class I polymerases synthesize the leading and lagging strands of the chromosome, respectively) (Dervyn et al., 2001; Inoue et al., 2001). However, the presence of multiple dnaE genes in several organisms, including Mycobacteria other than M. leprae (Mizrahi et al., 2000), Streptomyces spp., Agrobacterium tumefaciens, Xanthomonas spp., Sinorhizobium meliloti, Mesorhizobium loti, and Pseudomonas aeruginosa has been unexplained. It is intriguing that organisms with multiple dnaE genes all contain one or more dinP homologs. The level of sequence homology of the MSm and MTb dnaE1 genes with the single M. leprae homolog of dnaE strongly suggested that dnaE1 and not dnaE2 encoded the major replicative DNA polymerase (Mizrahi et al., 2000). Consistent with this prediction, we have been unable to inactivate the dnaE1 genes of MTb and MSm (unpublished results). dnaE2, however, does not appear to be essential for chromosomal replication since it can be inactivated in both organisms. In E. coli, TLS is primarily performed by the UmuD’2C (pol V) complex, although pol IV and pol II can also perform certain types of TLS (Napolitano et al., 2000). The activity of pol V is stimulated by the ␤ clamp, suggesting that pol V (Goodman, 2002 and references therein) may function with components of the holoenzyme complex in vivo. DinB has also been shown to interact with the ␤ clamp (Goodman, 2002). Based on the high degree of sequence similarity between MTb DnaE2 and E. coli DnaE that spans the entire protein, it is probable that DnaE2 also functions in association with the other subunits of the replicative complex, although the consensus ␤ clamp pentapeptide binding motif (Dalrymple et al., 2001) appears to be absent from this polymerase. Intriguingly, this motif is either absent or contains a nonconsensus proline residue in the additional DnaE enzymes of organisms with multiple copies of this gene, suggesting differences in association with the sliding clamp with respect to DnaE1. The observation that overexpression of dnaE2 did not result in an increased spontaneous mutation frequency suggests that DnaE2 either requires additional subunits for activity that are coinduced in response to DNA damage or that the affinity of DnaE2 for DNA and for subunits of the holoenzyme complex is much lower than that of DnaE1 in the absence of a replication-blocking lesion. Although there was no evidence for upregulation of genes encoding proteins that could interact with the replicative complex, the replicative helicase DnaB was also induced by DNA damage. There have been no reports of direct interaction between DnaB and DnaE, but interactions between DnaB and components of the holoenzyme complex could play a role in coupling primosome and replisome activities (Gao and McHenry 2001). To our knowledge, this is the first time that a family C polymerase has been implicated in TLS in the presence of a WT

copy of the gene encoding the proofreading subunit (dnaQ) of the replicative holoenzyme. Microarray analysis indicated that the levels of dnaQ were unaltered under the various forms of DNA damage. The spectrum of damage-induced mutations analyzed in this study was consistent with the hypothesis that DnaE2 induces mutation via its ability to function as a TLS, but this analysis was limited to UV-induced rpoB mutations that are nonlethal and confer Rif resistance. Since the mutations that confer Sm resistance are also mainly restricted to base substitutions (Ramaswamy and Musser, 1998), it remains to be established whether DnaE2 can introduce other mutations such as frameshifts. The role of TLS is both to generate mutations that may be beneficial for survival as well as to enhance survival after DNA damage. UV hypersensitivity of both MSm and MTb dnaE2 mutants and the depressed rate of emergence of drug resistance are two phenotypic sequellae of loss of DnaE2 function. The environmentdependent mutator phenotype conferred by genes such as dinP and umuDC may also promote survival in response to stressful conditions by globally altering mutation and repair rates following their induction (Metzgar and Wills, 2000). The apparent upregulation of MTb dnaE2 in a mouse infection model suggests that this effect may be important during infection. Indeed, the dnaE2 mutant of MTb was severely attenuated in longterm murine infections, suggesting that continued repair of DNA damage (and possibly generation of mutations) during infection is essential for survival of the bacilli. The induction of dnaE2 during host infection further raises the possibility that DnaE2 contributes directly to mutation in the face of DNA damage caused by reactive oxygen and nitrogen intermediates generated by the immune response. Thus, DnaE2 may contribute to the rise of mutants that are better adjusted for survival under bottlenecks encountered during infection and/or therapy. Since drug resistance in MTb is chromosomally encoded (Ramaswamy and Musser 1998), we propose that dnaE2-mediated mutagenesis may play a particularly important role in the evolution of drug resistance in this organism. Our preliminary studies support this model, since drug resistance during therapy emerged more frequently in WT MTb than in the DnaE2 knockout. These studies not only establish DnaE2 as the major mediator of DNA damage-induced mutagenesis in Mycobacteria but also suggest that ongoing DNA repair is an essential process during persistent infection with MTb and provide a new potential target for preventing the emergence of drug resistance. Experimental Procedures Growth of Strains and Generation of dnaE2 Mutants The strains, plasmids, and oligonucleotides used in this study are described in Table 5. Growth of mycobacterial strains and generation of mutants were performed as previously described (Karunakaran and Davies, 2000; Boshoff and Mizrahi, 2000; Parish and Stoker, 2000) with antibiotics used at the following concentrations for MTb and MSm, respectively: Rif, 2 and 200 ␮g/ml; kanamycin (Km), 25 and 50 ␮g/ml, and hygromycin (Hyg) and streptomycin (Sm) at 50 ␮g/ml for both mycobacterial species. The sequence of the M. smegmatis dnaE2 gene was obtained from the unfinished genome sequence of M. smegmatis (http://www.tigr.org/tdb/mdb/ mdbinprogress.html). In the MSm dnaE2::aph mutant, a 1.09 kb

Cell 190

Table 5. List of Strains, Plasmids and Oligonucleotides Used in this Study Strains

Description

H37Rv dnaE2::hyg dnaE2::hyg attB::dnaE2 mc2155 dnaE2::aph dnaE2::aph attB::dnaE2 mc2155(pOmsmE2) SMR5 recA::hyg

Laboratory strain (ATCC 27294) of M. tuberculosis H37Rv dnaE2 knockout mutant of M. tuberculosis H37Rv dnaE2::hyg complemented with dnaE2 at attB site High-frequency transformation mutant of M. smegmatis ATCC 706 dnaE2 knockout mutant of mc2155 dnaE2::aph complemented with dnaE2 at attB site mc2155 carrying pOmsmE2 SM resistant strain of M. smegmatis recA knockout mutant of SMR5

This work This work Snapper et al., 1990 This work This work This work Frischkorn et al., 1998 Frischkorn et al., 1998

Plasmids p2NIL pGEM3Z(+)f pHINT pMV206 pOLYG pGOAL17 p2NtbE2 p2NtbE2h p2NtbE2KO pMVtbE2 pGmsmE2 pGmsmE2k pGmsmE2KO pHmsmE2 pOmsmE2

Cloning vector, KmR Cloning vector, ApR E. coli-Mycobacterium integrating shuttle vector, HygR E. coli-Mycobacterium integrating shuttle vector, KmR E. coli-Mycobacterium multicopy shuttle vector, HygR Plasmid carrying sacB and lacZ genes as a PacI cassette, ApR 5.39 kb Asp718 fragment from H37Rv containing dnaE2 gene cloned into p2NIL, KmR p2NtbE2 carrying hyg gene in SacI site of dnaE2, KmR Knockout vector - p2NtbE2h with PacI cassette from pGOAL17 cloned into PacI site, KmR Complementing construct with the 5.39 kb Asp718 fragment from p2NtbE2 cloned into pMV206 pGEM3Z(⫹)f carrying 7.167 kb Asp700-BglII M. smegmatis DNA fragment containing dnaE2, ApR Constructed by replacing the NcoI-PstI fragment in pGmsmE2 (spanning from 784-1874 bp from the start codon of dnaE2) with aph, KmR Knockout vector - pGmsmE2k carrying PacI cassette from pGOAL17 in the XbaI site, KmR Complementing construct with the ScaI-XbaI fragment from pGmsmE2 cloned in pHINT pOLYG carrying the ScaI-XbaI fragment from pGmsmE2

Oligonucleotides

Forward primer (5⬘-3⬘)

Reverse primer (5⬘-3⬘)

Probe (5⬘-3⬘)

M. M. M. M. M. M. M. M. M. M. M. M. M.

aatgaccggcgcgct ccggtggaatgggcg

cgcggagctggttgatg aatttcaccaagccgattgc gagcatgcccagcccgag tgtccacacgtcggcg tcgtcttcatcccagacgaaa gtccttgcgtgcttgacgca gctgtccagatcgccattg atcaccaccgtgggaacatc gctcgtcgacgtcggagg gcggagctggttgatgaaga cagcgaggcgaagctcag ttgccgggcttgcctt aagtagccgtcaccgtcgaa

ataattcgggcaccacggcgatgt cagcgtcctgcaatgggacaaaga

†

tuberculosis recA tuberculosis dnaE2 tuberculosis dnaE2† tuberculosis dnaE1 tuberculosis sigA tuberculosis sigA† tuberculosis gnd tuberculosis icl tuberculosis icl† smegmatis recA smegmatis dnaE2 smegmatis sigA smegmatis dnaE1

cgtgtgactccaccttgttga gacgaggagatcgctgaacc gtccacaacggcatcgagta catccgcactttgacgtctg atgaccggcgcgttga cgaacgtcacggcatcac cccaccgggaattcgtaag cgctgctgtgtgtgcagac

Parish & Stoker, 2000 Promega O’Gaora et al., 1997 Mdluli et al., 1996 O’Gaora et al., 1997 Parish & Stoker, 2000 This work This work This work This work This work This work This work This work This work

cgccgaacgggtgcagtcct ccgaaaaggacaaggcctccgg tccgacatgcagctcatcggtga tcggctcgcggccgatgt caactcgggcaccaccgcg cgccaatttcggtttcccggaaa cgaaagggtgtacgtggcagcgac caagaccctctcggatcccacgc

Used for first strand cDNA synthesis on mycobacterial RNA isolated from mouse lungs and spleen.

NcoI–PstI fragment of dnaE2 was removed and replaced by a kanamycin resistance cassette. The MTb ortholog was disrupted by insertion of a hyg marker into the SacI site of the gene. UV sensitivity of strains was determined by plating dilutions (in triplicate) on solid medium and irradiation of open plates in a Stratalinker 1800 (Stratagene, 254nm) at 0–70 mJ/cm2. Mutation frequencies were determined by UV irradiation as described above in 6 ml volumes of liquid medium containing cells harvested from log-phase cultures (OD650nm ⬇ 0.6) with subsequent recovery in the original volume of growth medium before plating at different times postirradiation. Cells were maintained in log-phase by dilution with growth medium. Cells were plated onto SM- or Rif-containing plates (ⵑ108 cells/plate) while viable counts were assessed by plating, in triplicate, serial dilutions from two independent samplings, on antibiotic-free medium. To monitor mutation frequencies of MTb and MSm strains after oxidative DNA damage, log-phase cultures were treated with 4 mM H2O2 for 1–24 hr before plating as above. All mutation frequency assessments were performed using at least

four independent biological replicates with a total of at least 3 ⫻ 109 cells plated on antibiotic-containing plates in each case to eliminate possible jackpot effects. Estimation of Mutation Rates A Luria-Delbru¨ck fluctuation assay was set up using 80 1-ml cultures each containing approximately 5000 cells/ml in medium containing 15% sterile filtered (0.2 ␮m) culture supernatant from a log-phase H37Rv culture. Cell suspensions were passed through a 5 ␮m filter before dilution to ensure a single cell suspension. Some cultures were irradiated at 20 mJ/cm2 before dilution. Cultures were incubated for 2 weeks at 37⬚C with daily agitation before plating the total volume on 2 ␮g/ml Rif plates. Some cultures were used for CFU determinations. The mutation rate was estimated by the method of the mean (Luria and Delbru¨ck, 1943) and reported as number of mutations per culture/2(number of cells per culture) (Rosche and Foster, 2000). The raw data giving the number of mutants recovered from each culture and the calculated number of mutations per cul-

dnaE2 and Inducible Mutagenesis in M. tuberculosis 191

ture is supplied in the Supplemental Data (available at http:// www.cell.com/cgi/content/full/113/2/183/DC1). RNA Purification Mycobacterial cultures (200 ml) were grown to an OD650 of 0.3–0.4 and were treated either with MMC at 0.2 ␮g/ml or H2O2 at 4 mM, harvested, and resuspended in the same original volume of medium after 90 min treatment. For microarray experiments, the DNA damaging agents were not removed during the treatment period. For DNA damage by UV, cells were harvested, resuspended in 6 ml medium in a 78 cm2 dish and irradiated as above at 40 mJ/cm2 before being resuspended in 200 ml of medium and returned to 37⬚C. RNA was extracted as described by Sherman et al. (2001) and subjected to an additional round of DNase I treatment for quantitative RNA analysis using the Ambion DNA-free kit. Mycobacterial RNA was extracted from infected mouse lungs by homogenization of the organs in Trizol (Pro 300D homogenizer, Pro Scientific), harvesting of the Mycobacteria by centrifugation, and subsequent RNA extraction as above.

mg/kg Rif suspended in 0.1% agar after which organs of 5 mice were plated to monitor efficiency of bacterial clearance. The treated mice were allowed to recover for 4 weeks, after which a second 2-week cycle of rifampicin treatment was initiated and lungs subsequently plated as above. Six weeks after the second cycle of gavage, lung homogenates were pooled in groups of 10 and plated on Rif (1 ␮g/ml) to monitor emergence of drug resistance. Serial dilutions were plated to assess organ loads. In a second round of infections, mice were infected as above with 130–150 CFU of MTb and treated for 2 weeks by daily gavage with low-dose Rif (10 mg/kg). One week after treatment, 20 mice per group were sacrificed and pooled lung homogenates were plated as above. In a third round of infection, mice were infected as above with 100 CFU of MTb H37Rv, dnaE2::hyg, or the dnaE2::hyg attB::dnaE2 strains and treated 10 weeks after infection with low-dose Rif as above. Mice were sacrificed and lung homogenates of individual mice plated as above. Acknowledgments

Microarray Analysis of Gene Expression Microarrays were prepared at the Microarray Research Facility (NIAID/ NIH) using the MTb oligonucleotides purchased from Operon Technologies and printed according to the procedures outlined on the Stanford University website (http://cmgm.stanford.edu/pbrown/ protocols/index.html). Fluorescently labeled cDNA was prepared from 4 ␮g of RNA and all steps in preparation and hybridization of microarrays carried out as described (Wilson et al., 1999; Schoolnik et al., 2001). For microarray analysis, the mAdb Tool Gateway software provided by the CIT (http://nciarray.nci.nih.gov/) was employed. For each time point, hybridizations were performed at least in duplicate using RNA extracted from two independent experiments. Spots were normalized using the 50th percentile (median) and only included in the analysis if the values for both channels were at least one standard deviation above the average value of the negative control spots that contained randomized hexamer oligonucleotides. Signals were calculated as the mean of the feature intensity minus the median of the feature background. Spots were omitted from further analysis if the calculated signal to background ratios in both were less than two. However, all previous selection criteria were ignored if the calculated signal in one channel was at least three times higher than the value of the negative control spots plus one standard deviation. Any discrepancy between the induction levels observed by microarray analysis and real-time PCR could be ascribed to the fact that the signal of spots (e.g., dnaE2, lexA) was marginally stronger than the average value of the negative control spots plus one standard deviation. In some cases, the value was lower for several arrays, in which case the spot was excluded from analysis. The array data have been deposited in the Gene Expression Omnibus at NCBI (GEO: http://www.ncbi.nlm.nih.gov/geo) with GEO Platform accession number GPL280 and sample accession numbers GSM5272–GSM5299.

This work was partly funded by the Medical Research Council of South Africa, the National Research Foundation, and the NHLS. V.M. was also supported by an International Research Scholar’s grant from the Howard Hughes Medical Institute. We are most grateful to Erik Bo¨ttger for providing us with the recA mutant strain of M. smegmatis, which was instrumental in providing experimental evidence for an SOS-dependent error-prone repair system in mycobacteria. We thank Digby Warner, Edith Machowski, Stephanie Dawes, Steven Durbach, Mike Wilson, and staff at the Microarray Research facility for helpful discussions, and Juliano Timm and John McKinney for sharing their method for RNA extraction from mouse tissues. We thank the Institute for Genome Research for allowing public access to the M. smegmatis and M. avium genome sequence information.

Real-Time, Quantitative RT-PCR Assay First-strand cDNA synthesis was performed as described above and real-time quantitative PCR was performed on the ABI Prism 7700 sequence detection system using the PE Applied Biosystems Taqman PCR core reagent kit. Taqman primer and probe sets are described in Table 5. For MTb, RNA levels were normalized to the gnd or sigA genes (Brooks et al., 1999; Manganelli et al., 1999), and for MSm to the corresponding sigA gene. Gene expression was determined using two independent induction experiments using a control untreated cell sample for each time point. Each RNA sample was independently reverse transcribed twice and tested in triplicate or quadruplicate for each probe-primer set.

Boshoff, H.I., and Mizrahi, V. (2000). Expression of Mycobacterium smegmatis pyrazinamidase in Mycobacterium tuberculosis confers hypersensitivity to pyrazinamide and related amides. J. Bacteriol. 182, 5479–5485.

Infection of Mice Six-to-eight-week old female C57BL/6 mice (Taconic) were infected with approximately 50 CFU of MTb via the aerosol route and infection monitored by plating of lung homogenates from 4–5 mice at different times post-infection (Manca et al., 2001). Time-to-death analysis was performed on 10 remaining mice. Mice were euthanized when they became moribund. Six weeks after aerosol infection, 50 mice from each group were treated for 2 weeks by daily gavage with 25

Received: September 16, 2002 Revised: March 24, 2003 Accepted: April 1, 2003 Published: April 17, 2003 References Bjorkholm, B., Sjolund, M., Falk, P.G., Berg, O.G., Engstrand, L., and Andersson, D.I. (2001). Mutation frequency and biological cost of antibiotic resistance in Helicobacter pylori. Proc. Natl. Acad. Sci. USA 98, 14607–14612. Borden, A., O’Grady, P.I., Vandewiele, D., Fernandez de Henestrosa, A.R., Lawrence, C.W., and Woodgate, R. (2002). Escherichia coli DNA polymerase III can replicate efficiently past a T-T cis-syn cyclobutane dimer if DNA polymerase V and the 3⬘ to 5⬘ exonuclease proofreading function encoded by dnaQ are inactivated. J. Bacteriol. 184, 2674–2681.

Boshoff, H.I.M., Durbach, S.I., and Mizrahi, V. (2001). DNA metabolism in Mycobacterium tuberculosis: implications for drug resistance and strain variability. Scand. J. Infect. Dis. 33, 101–105. Brooks, P.C., Movahedzadeh, F., and Davis, E.O. (2001). Identification of some DNA damage-inducible genes of Mycobacterium tuberculosis: apparent lack of correlation with LexA binding. J. Bacteriol. 183, 4459–4467. Courcelle, J., Khodursky, A., Peter, B., Brown, P.O., and Hanawalt, P.C. (2001). Comparative gene expression profiles following UV exposure in wild-type and SOS-deficient Escherichia coli. Genetics 158, 41–64. Dalrymple, B.P., Kongsuwan, K., Wijffels, G., Dixon, N.E., and Jennings, P.A. (2001). A universal protein-protein interaction motif in the eubacterial DNA replication and repair systems. Proc. Natl. Acad. Sci. USA 98, 11627–11632. Davis, E.O., Dullaghan, E.M., and Rand, L. (2002a). Definition of the

Cell 192

mycobacterial SOS box and use to identify LexA-regulated genes in Mycobacterium tuberculosis. J. Bacteriol. 184, 3287–3295.

a mechanistic basis for SOS untargeted mutagenesis. Proc. Natl. Acad. Sci. USA 97, 565–570.

Davis, E.O., Springer, B., Gopaul, K.K., Papavinasasundaram, K.G., Sander, P., and Bottger, E.C. (2002b). DNA damage induction of recA in Mycobacterium tuberculosis independently of RecA and LexA. Mol. Microbiol. 46, 791–800.

McKinney, J.D., Honer zu Bentrup, K., Munoz-Elias, E.J., Miczak, A., Chen, B., Chan, W.T., Swenson, D., Sacchettini, J.C., Jacobs, W.R. Jr., and Russell, D.G. (2000). Persistence of Mycobacterium tuberculosis in macrophages and mice requires the glyoxylate shunt enzyme isocitrate lyase. Nature 406, 735–738.

Dervyn, E., Suski, C., Daniel, R., Bruand, C., Chapuis, J., Errington, J., Janniere, L., and Ehrlich, S.D. (2001). Two essential DNA polymerases at the bacterial replication fork. Science 294, 1716–1719. Durbach, S.I., Andersen, S.J., and Mizrahi, V. (1997). SOS induction in Mycobacteria: analysis of the DNA-binding activity of a LexA-like repressor and its role in DNA damage induction of the recA gene from Mycobacterium smegmatis. Mol. Microbiol. 26, 643–653.

Mdluli, K., Sherman, D.R., Hickey, M.J., Kreiswirth, B.N., Morris, S., Stover, C.K., and Barry, C.E., 3rd. (1996). Biochemical and genetic data suggest that InhA is not the primary target for activated isoniazid in Mycobacterium tuberculosis. J. Infect. Dis. 174, 1085–1090. Metzgar, D., and Wills, C. (2000). Evidence for the adaptive evolution of mutation rates. Cell 101, 581–584.

Dye, C., Williams, B.G., Espinal, M.A., and Raviglione, M.C. (2002). Erasing the world’s slow stain: strategies to beat multidrug-resistant tuberculosis. Science 295, 2042–2046.

Mizrahi, V., Dawes, S.S., and Rubin, H. (2000). DNA Replication. In Molecular Genetics of Mycobacteria, G.F. Hatfull and W.R. Jacobs, Jr., eds. (Washington, D.C.: ASM Press), pp. 159–172.

Espinal, M.A., Kim, S.J., Suarez, P.G., Kam, K.M., Khomenko, A.G., Migliori, G.B., Baez, J., Kochi, A., Dye, C., and Raviglione, M.C. (2000). Standard short-course chemotherapy for drug-resistant tuberculosis: treatment outcomes in 6 countries. JAMA 283, 2537– 2545.

Morlock, G.P., Plitkaytis, B.B., and Crawford, J.T. (2000). Characterization of spontaneous, in vitro-selected, rifampin-resistant mutants of Mycobacterium tuberculosis strain H37Rv. Antimicrob. Agents Chemother. 44, 3298–3301.

Flynn, J.L., and Chan, J. (2001). Immunology of tuberculosis. Annu. Rev. Immunol. 19, 93–129.

Napolitano, R., Janel-Bintz, R., Wagner, J., and Fuchs, R.P. (2000). All three SOS-inducible DNA polymerases (Pol II, Pol IV and Pol V) are involved in induced mutagenesis. EMBO J. 19, 6259–6265.

Frischkorn, K., Sander, P., Scholz, M., Teschner, K., Prammananan, T., and Bo¨ttger, E.C. (1998). Investigation of Mycobacterial recA function: protein introns in the RecA of pathogenic Mycobacteria do not affect competency for homologous recombination. Mol. Microbiol. 29, 1203–1214.

O’Gaora, P., Barnin, S., Hayward, C., Filley, E., Rook, G., Young, D., and Thole, J. (1997). Mycobacteria as immunogens: development of expression vectors for use in multiple mycobacterial species. Med. Princ. Prac. 6, 91–96.

Gao, D., and McHenry, S. (2001). ␶ binds and organizes Escherichia coli replication proteins through distinct domains. Domain IV, located within the unique C terminus of ␶, binds the replication fork helicase, DnaB. J. Biol. Chem. 276, 4441–4446. Goodman, M. (2002). Error-prone repair DNA polymerases in prokaryotes and eukaryotes. Annu. Rev. Biochem. 71, 17–50. Huang, Y.P., and Ito, J. (1999). DNA polymerase C of the thermophilic bacterium Thermus aquaticus: classification and phylogenetic analysis of the family C DNA polymerases. J. Mol. Evol. 48, 756–769. Inoue, R., Kaito, C., Tanabe, M., Kamura, K., Akimitsu, N., and Sekimizu, K. (2001). Genetic identification of two distinct DNA polymerases, DnaE and PolC, that are essential for chromosomal DNA replication in Staphylococcus aureus. Mol. Genet. Genomics 266, 564–571. Ito, J., and Braithwaite, D.K. (1991). Compilation and alignment of DNA polymerase sequences. Nucleic Acids Res. 19, 4045–4057. Karunakaran, P., and Davies, J. (2000). Genetic antagonism and hypermutability in Mycobacterium smegmatis. J. Bacteriol. 182, 3331–3335. Khil, P.P., and Camerini-Otero, R.D. (2002). Over 1000 genes are involved in the DNA damage response of Escherichia coli. Mol. Microbiol. 44, 89–105.

Oliver, A., Canton, R., Campo, P., Baquero, F., and Blazquez, J. (2000). High frequency of hypermutable Pseudomonas aeruginosa in cystic fibrosis lung infection. Science 288, 1251–1254. Parish, T., and Stoker, N.G. (2000). Use of a flexible cassette method to generate a double unmarked Mycobacterium tuberculosis tlyA plcABC mutant by gene replacement. Microbiology 146, 1969–1975. Pham, P., Rangarajan, S., Woodgate, R., and Goodman, M.F. (2001). Roles of DNA polymerases V and II in SOS-induced error-prone and error-free repair in Escherichia coli. Proc. Natl. Acad. Sci. USA 98, 8350–8354. Ramaswamy, S., and Musser, J.M. (1998). Molecular genetic basis of antimicrobial agent resistance in Mycobacterium tuberculosis: 1998 update. Tuber. Lung Dis. 79, 3–29. Rangarajan, S., Woodgate, R., and Goodman, M.F. (1999). A phenotype for enigmatic DNA polymerase II: a pivotal role for pol II in replication restart in UV-irradiated Escherichia coli. Proc. Natl. Acad. Sci. USA 96, 9224–9229. Rangarajan, S., Woodgate, R., and Goodman, M.F. (2002). Replication restart in UV-irradiated Escherichia coli involving pols II, III, V, PriA, RecA and RecFOR proteins. Mol. Microbiol. 43, 617–628. Rosche, W.A., and Foster, P.L. (2000). Determining mutation rates in bacterial populations. Methods 20, 4–17.

Luria, S.E., and Delbru¨ck, M. (1943). Mutations of bacteria from virus sensitivity to virus resistance. Genetics 28, 491–511.

Russell, D.G. (2001). Mycobacterium tuberculosis: here today, and here tomorrow. Nat. Rev. Mol. Cell Biol. 2, 569–577.

Makarova, K.S., Aravind, L., Grishin, N.V., Rogozin, I.B., and Koonin, E.V. (2002). A DNA repair system specific for thermophilic Archaea and bacteria predicted by genomic context analysis. Nucleic Acids Res. 30, 482–496.

Schlosser-Silverman, E., Elgrably-Weiss, M., Rosenshine, I., Kohen, R., and Altuvia, S. (2000). Characterization of Escherichia coli DNA lesions generated within J774 macrophages. J. Bacteriol. 182, 5225– 5230.

Manca, C., Tsenova, L., Bergtold, A., Freeman, S., Tovey, M., Musser, J.M., Barry, C.E., 3rd, Freedman, V.H., and Kaplan, G. (2001). Virulence of a Mycobacterium tuberculosis clinical isolate in mice is determined by failure to induce Th1 type immunity and is associated with induction of IFN-␣/␤. Proc. Natl. Acad. Sci. USA 98, 5752– 5757.

Schoolnik, G.K., Voskuil, M.I., Schnappinger, D., Yildiz, F.H., Meibom, K., Dolganov, N.A., Wilson, M.A., and Chong, K.H. (2001). Whole genome DNA microarray expression analysis of biofilm development by Vibrio cholerae O1 E1 Tor. Meth. Enzymol. 336, 3–18.

Manganelli, R., Dubnau, E., Tyagi, S., Kramer, F.R., and Smith, I. (1999). The Mycobacterium tuberculosis ECF ␴ factor ␴E: role in global gene expression and survival in macrophages. Mol. Microbiol. 31, 715–724. Maor-Shoshani, A., Reuven, N.B., Tomer, G., and Livneh, Z. (2000). Highly mutagenic replication by DNA polymerase V (UmuC) provides

Sherman, D.R., Voskuil, M., Schnappinger, D., Liao, R., Harrell, M.I., and Schoolnik, G.K. (2001). Regulation of the Mycobacterium tuberculosis hypoxic response gene encoding ␣-crystallin. Proc. Natl. Acad. Sci. USA 98, 7534–7539. Snapper, S.B., Melton, R.E., Kieser, T., Mustafa, S., and Jacobs, W.R., Jr. (1990). Isolation and characterization of efficient plasmid transformation mutants of Mycobacterium smegmatis. Mol. Microbiol. 4, 1911–1919.

dnaE2 and Inducible Mutagenesis in M. tuberculosis 193

Sung, H., Yeamans, G., Ross, C.A., and Yasbin, R.E. (2003). Roles of YqjH and YqjW, homologs of the Escherichia coli UmuC/DinB or Y superfamily of DNA polymerases, in stationary-phase mutagenesis and UV-induced mutagenesis of Bacillus subtilis. J. Bacteriol. 185, 2153–2160. Wilson, M., DeRisi, J., Kristensen, H.H., Imboden, P., Rane, S., Brown, P.O., and Schoolnik, G.K. (1999). Exploring drug-induced alterations in gene expression in Mycobacterium tuberculosis by microarray hybridization. Proc. Natl. Acad. Sci. USA 96, 12833– 12838. Accession Numbers The array data discussed in this paper have been deposited in the Gene Expression Omnibus at NCBI (GEO: http://www.ncbi.nlm.nih. gov/geo) with GEO Platform accession number GPL280 and sample accession numbers GSM5272–GSM5299.