Regulation of Ergothioneine Biosynthesis and Its Effect on Mycobacterium tuberculosis Growth and Infectivity*

Background: Mycobacterium tuberculosis synthesizes ergothioneine, a sulfur-containing molecule with unknown function. Results: egtD encodes for a histidine methyltransferase that is essential for ergothioneine biosynthesis and is negatively regulated through M. tuberculosis serine/threonine protein kinase D. Conclusion: M. tuberculosis modulates intracellular ergothioneine levels in response to starvation. Significance: Mechanisms by which M. tuberculosis senses and adapts to nutrient starvation is essential for understanding persistence and disease latency. Ergothioneine (EGT) is synthesized in mycobacteria, but limited knowledge exists regarding its synthesis, physiological role, and regulation. We have identified Rv3701c from Mycobacterium tuberculosis to encode for EgtD, a required histidine methyltransferase that catalyzes first biosynthesis step in EGT biosynthesis. EgtD was found to be phosphorylated by the serine/threonine protein kinase PknD. PknD phosphorylates EgtD both in vitro and in a cell-based system on Thr213. The phosphomimetic (T213E) but not the phosphoablative (T213A) mutant of EgtD failed to restore EGT synthesis in a ΔegtD mutant. The findings together with observed elevated levels of EGT in a pknD transposon mutant during in vitro growth suggests that EgtD phosphorylation by PknD negatively regulates EGT biosynthesis. We further showed that EGT is required in a nutrient-starved model of persistence and is needed for long term infection of murine macrophages.

Ergothioneine (EGT) is synthesized in mycobacteria, but limited knowledge exists regarding its synthesis, physiological role, and regulation. We have identified Rv3701c from Mycobacterium tuberculosis to encode for EgtD, a required histidine methyltransferase that catalyzes first biosynthesis step in EGT biosynthesis. EgtD was found to be phosphorylated by the serine/ threonine protein kinase PknD. PknD phosphorylates EgtD both in vitro and in a cell-based system on Thr 213 . The phosphomimetic (T213E) but not the phosphoablative (T213A) mutant of EgtD failed to restore EGT synthesis in a ⌬egtD mutant. The findings together with observed elevated levels of EGT in a pknD transposon mutant during in vitro growth suggests that EgtD phosphorylation by PknD negatively regulates EGT biosynthesis. We further showed that EGT is required in a nutrientstarved model of persistence and is needed for long term infection of murine macrophages.
Mycobacterium tuberculosis, the causative agent of tuberculosis, senses and adapts its physiology to ensure survival within the host, enabling it to overcome oxidative and nitrosative challenges associated with intracellular infection. Xenobiotics, including free radicals, produced by the host to cope with infection; antibiotics; and general bacterial respiration are countered by an intricate detoxification system utilized by the bacterium that is composed of (a) enzymes such as catalase, superoxide dismutase, and alkyl hydroperoxidase; (b) truncated hemoglobins; (c) oxidoreductases; and (d) redox coupling systems (1). The most common defense mechanism against reac-tive oxygen and nitrogen species damage in eukaryotes and Gram-negative bacteria relies on glutathione, a low molecular weight thiol. Mycobacteria, like most Gram-positive bacteria, do not produce glutathione; rather, they synthesize mycothiol (MSH) 2 at millimolar levels, making it the most abundant low molecular weight thiol in these species (2). Similarly to glutathione, MSH also serves as an antioxidant that is important in maintaining a highly reducing environment inside the cell (3). Several studies have demonstrated the role of MSH in detoxifying reactive species by either (i) donating a reducing equivalent (4) or (ii) forming an S-conjugate composed of MSH and the respective agent (5). Consistent with these findings, mycobacterial species deficient in MSH are found to have varying but increased sensitivity to H 2 O 2 (3,4,6), nitric oxide (NO) (7), and other redox cycling agents (3,4,8,9). Despite the importance of MSH in protecting mycobacteria against oxidative and nitrosative stressors, M. tuberculosis strains lacking MSH remain viable in vivo, suggesting a compensatory mechanism (10).
Fahey and co-workers (11) showed that Mycobacterium smegmatis mutants devoid of MSH displayed a marked elevation in another sulfur-containing low molecular weight compound known as ergothioneine (EGT). EGT exists primarily as a thione under physiological conditions and differs significantly from other cysteine-containing thiols as it possesses a lower redox couple value (E 0 Ј ϭ Ϫ0.06 V) (12). Although a poorer reductant, EGT is described as an effective antioxidant through quenching of singlet oxygen, scavenging of hydroxyl radicals, and inhibition of heavy metal-catalyzed reactions (for a review, see Ref. 13), and its depletion in mammalian cells leads to augmented oxidative damage and cell death in the presence of exogenous stressors (14). Despite the presence of EGT in mammalian cells, biosynthesis is limited to a subset of organisms, which includes actinobacteria, cyanobacteria, and specific fungi and yeast (15)(16)(17)(18). Mammals obtain EGT from dietary sources and concentrate it in tissues exposed to high levels of oxidative stress via a specific transporter (19).
Among microorganisms, EGT biosynthesis follows a similar pathway, that is the conversion of histidine to EGT through the intermediate hercynine. The genes encoding EGT biosynthesis in M. smegmatis have been identified (20), paving the way to study the role EGT in M. tuberculosis physiology and pathogenesis. Still, it remains to be discovered as to which physiological conditions trigger M. tuberculosis to up-regulate EGT biosynthesis in vivo. As observed in Fig. 1, EGT biosynthesis is an energetically costly process for mycobacteria. The pathway follows five enzymatic steps that consume the amino acids histidine and cysteine and cofactors such as ATP, S-adenosylmethionine (AdoMet), and pyridoxal phosphate. Therefore, EGT production is likely to be tightly regulated to avoid metabolic stress under unfavorable environmental conditions. Mycobacterial adaptation to varying conditions is dependent on both the classical bacterial two-component systems comprising histidine kinases and response regulators and the eukaryotic-like Ser/Thr protein kinases (STPKs) (21). Signaling through STPKs has recently emerged as a key regulatory mechanism in M. tuberculosis, playing roles in the transport of metabolites (22), cell division (23), and virulence (24). Several metabolic pathways such as mycolic acid (25)(26)(27)(28), glutamine (29), phthiocerol dimycocerosates (30), and glucan (31) biosyntheses have been shown to be regulated by M. tuberculosis STPK phosphorylation.
In the present study, using combined biochemical and genetic analyses, we showed that EGT biosynthesis is dependent on EgtD, a histidine methyltransferase catalyzing the first reaction step in EGT biosynthesis. We further demonstrated that EgtD is under phosphorylation control by the STPK PknD, leading to increased up-regulation of EGT biosynthesis during starvation and enhancing the survival of M. tuberculosis in an in vitro model of persistence.
Cloning and Protein Expression and Purification-All genes were amplified by PCR from H37Rv genomic DNA using standard methods for cloning. Recombinant plasmids were further transformed into E. coli BL21(DE3) cells for protein expression. Strains harboring the genes to produce recombinant protein were used to inoculate LB broth from an overnight culture (1:100), and the cells were induced with 1-thio-␤-D-galactopyranoside once an A 600 of 0.6 -0.8 was reached. Protein purification was carried out on either nickel-nitrilotriacetic acid resin (Qiagen) or glutathione-agarose (Sigma-Aldrich) columns according to the manufacturers' supplied guidelines. Site-directed mutagenesis was performed following the Stratagene QuikChange protocol, and parental pMV261-egtD and pET28-egtD were used as a template.
EgtD in Vitro Methylation Activity-A reaction containing 10 mM histidine, 4 mM AdoMet, 1 mM Mg(OAc) 2 , 5 mM NaCl, 20 g of EgtD, and 5 g of S-adenosylhomocysteine hydrolase was prepared as described (20) and incubated overnight at 37°C. Detection of methylated histidine products in the reaction was carried out by ESI-MS at the University of Victoria Genome British Columbia Proteomics Centre as detailed below. EgtD substrate specificity was analyzed using a continuous enzymecoupled SAM510 TM methyltransferase assay (G-Biosciences) according to the manufacturer's protocol. Reactions were initiated immediately following the addition of a 1 mM concentration of each tested substrate and monitored at 510 nm for 30 min at room temperature.
Construction and Complementation of the M. tuberculosis ⌬egtD Mutant-The ⌬egtD mutant strain was constructed via allelic exchange using the conditionally replicating mycobacteriophage phAE159 as described previously (32). To construct the egtD knock-out phage, flanking regions comprising 1000-bp upstream and downstream regions of the egtD gene were amplified from H37Rv genomic DNA. The up-and downstream flanking regions of egtD were cloned into the p0004S cosmid prior to ligation of this recombinant cosmid with phAE159. The ligated DNA was packaged into phage with Gigapack III Gold packaging extract (Stratagene) and E. coli HB101 cells that were previously grown in MgSO 4 and maltose overnight. Colonies were selected for growth on LB plates containing 150 g/ml hygromycin, and phage DNA was extracted and electroporated into M. smegmatis mc 2 155. Transformation plates were incubated at 30°C for 3 days. Plaques were picked to prepare high titer phage stocks (10 9 pfu/ml) in M. smegmatis. Phages were transduced into M. tuberculosis H37Rv and plated on Middlebrook 7H10 supplemented with OADC and hygromycin (50 g/ml). After 4 weeks, hygromycin-resistant colonies appeared and were cultured for analysis by PCR and Southern hybridization to identify clones in which allelic exchange had occurred within the Rv3701c gene.
Southern Blot Hybridization-Southern blotting was performed using the digoxigenin hybridization system (Roche Applied Science). Chromosomal DNA (12 g) from both the H37Rv wild-type and Rv3701c-null mutant strains was digested with AflIII. Digested DNA was resolved on a 1% agarose gel prior to its transfer to a nylon membrane via capillary method overnight. Hybridization was performed at 68°C overnight with a digoxigenin-11-dUTP-labeled probe. Anti-digoxigenin antibodies were used to detect the probe hybridized to its DNA target.
EGT Extraction from M. tuberculosis-M. tuberculosis strains were grown to their desired A 600 , and 4 ml of cells were harvested by centrifugation. Cells were washed twice in the same volume of double distilled H 2 O. Following washing, the cells were resuspended in 2 ml of 70% acetonitrile with 25 ng/ml internal standard 1-methyl-4-phenylpyridinium ion. Cells were disrupted with the MagNAlyser (Roche Applied Science) and 0.1-mm silica beads (BioSpec) at a speed of 7000 rpm for 60-s intervals followed by 2 min of rest at Ϫ20°C (repeated four times). The extract was then filter-sterilized using 0.22-m nylon polypropylene Spin-X centrifuge tubes prior to exiting the biosafety containment level 3 laboratory for further analysis by ESI LC-MS/MS.
ESI LC-MS/MS Analysis of EGT-EGT was quantified using an Agilent Technologies 1200 binary HPLC system coupled to an AB Sciex 5500 Q-Trap triple quadrupole mass spectrometer. Separation was performed on a Zorbax HILIC Plus column (Agilent Technologies; 100 ϫ 2.1 mm, 3.5-m particle size). Acetonitrile (76%) and water (24%), both containing 0.1% formic acid, were used as the mobile phase at a flow rate of 200 l/min. The peak area of EGT was measured using Analyst 1.5.2 (AB Sciex) and normalized by the weighted contribution of the peak areas of the 1-methyl-4-phenylpyridinium ion internal standard. Identification of EGT was based on its theoretical m/z value, MS/MS fragmentation data, and its retention time, which was verified by analyzing a pure standard (Oxis International Inc.). Calibration curves were generated through a series of EGT standard additions to the sample. Regression coefficients of each calibration curve were all greater than 0.99.
Protein-Protein Interaction Assay-Protein-protein interactions were investigated using the mycobacterial protein fragment complementation assay as described previously (33). M. tuberculosis EgtD and STPKs (PknA, PknB, PknD, and PknK) were amplified by PCR and cloned into pUAB100 (expressing murine dihydrofolate enzyme fragments F1 and F2) and pUAB200 (expressing murine dihydrofolate fragment F3), respectively. EgtD was co-transformed with each of the four kinases into M. smegmatis mc 2 155, and co-transformants were selected for on 7H11/kanamycin/hygromycin plates. Co-transformants were replated on 7H11/kanamycin/hygromycin plates supplemented with 0 and 10 g/ml trimethoprim and analyzed for growth over 4 -5 days.
In Vitro Kinase Assay-An in vitro phosphorylation screen was performed as described previously (34) using 1 g of EgtD in 20 l of the assay buffer (20 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 5 mM MnCl 2 , 1 mM DTT) and varying concentrations of kinase (0.1-1 g) to obtain optimal autophosphorylation activity. Kinases used for screening in this assay were PknA, PknB, PknD, PknE, PknF, PknG, PknH, and PknK. Reactions were commenced by the addition of 10 Ci of [␥-32 P]ATP (PerkinElmer Life Sciences; 3000 Ci/mmol) and incubated at room temperature (23°C) for 30 min. Following the incubation period, reactions were arrested using SDS sample loading buffer and heated at 95°C for proteins bands. EgtD dose-dependent phosphorylation kinetics were performed as described above with minor changes. First, the kinase was left to autophosphorylate for 20 min prior to the addition of EgtD. Cold ATP (100 M) was spiked into 10 Ci of [␥-32 P] ATP prior to the addition of ATP to the reactions. The reaction kinetics were monitored by excising the bands corresponding to EgtD and subjecting them to scintillation counting (Beckman Coulter LS 6500). Kinetic parameters were calculated using Prism Software (Graph-Pad 6.04). EgtD phosphorylation sites were analyzed using LC-MS/MS (phosphopeptide analysis) as described (24) with 1 mM non-radiolabeled ATP.
Cell-based Phosphorylation-egtD was cloned into pGEX-4T3, and pknD was cloned into pET-28. Both plasmids were co-transformed into E. coli BL21. Co-transformants were selected on LB plates containing ampicillin and kanamycin. Cultures were induced with 1 mM 1-thio-␤-D-galactopyranoside and further grown for 16 h at 25°C. Both proteins were purified as described above and resolved by SDS-PAGE to ensure adequate expression of both proteins in the culture. Approximately 20 g of recombinant protein was then subjected to phosphopeptide analysis by LC-MS/MS to determine egtD phosphorylation sites in a cell-based system.
TLC Analysis of Phosphorylated EgtD Activity-Phosphorylated EgtD was obtained through the in vitro kinase assay outlined above. The reaction varied slightly in that 8 M EgtD, 10 M non-radiolabeled ATP, and 0.4 M kinase were used. Reactions were incubated at room temperature for 1.5 h to obtain the maximum yield of phosphorylated EgtD. Next, 15 mM histidine, 2.5 mM NaCl, 500 M Mg(OAc) 2 , and 10 M S-adenosylhomocysteine nucleosidase were added to the 8 M phosphorylated EgtD. The methylation reaction was initiated upon addition of 10 mol of S-[methyl-14 C]adenosyl-L-methionine (PerkinElmer Life Sciences; 60 mCi/mmol). Ten-microliter samples were taken from the reaction at various time points, stopped with 1 l of 1% trifluoroacetic acid, and stored at Ϫ20°C until use. Samples were subjected to TLC using PEI Cellulose F plates (EMD Millipore, Darmstadt, Germany) and developed in butanol/water/acetic acid (60:25:15, v/v). The separation of radiolabeled AdoMet and methylated histidine was visualized using PhosphorImager SI (GE Healthcare) following 7 days of exposure. Ninhydrin (0.03%) was used to detect histidine on the plate. The spots corresponding to histidine were cut from the plate and subjected to scintillation counting (Beckman Coulter LS 6500) to quantify the formation of methylated histidine over time.
Macrophage Infection-Murine J774A.1 macrophages were purchased from the American Type Culture Collection (ATCC catalogue number TIB-67) and were stored and prepared according to the manufacturer's guidelines. Macrophages were prepared by seeding onto a 24-well plate at a density of 2.5 ϫ 10 5 cells/well in culture medium (Dulbecco's modified Eagle's medium, high glucose supplemented with 1% glutamine, 10% fetal bovine serum, 1% HEPES, 1% non-essential amino acids). Cells were left overnight. The following day, J774A.1 cells were infected with exponentially growing M. tuberculosis (A 600 ϭ 0.5) at a multiplicity of infection of 5:1. J774A.1 cells were incubated with M. tuberculosis for 3 h at 37°C in 5% CO 2 . Wells were next washed three times and resuspended in culture medium containing 100 g/ml gentamicin to remove any remaining extracellular M. tuberculosis. For cfu counting, cells were washed three times with J774A.1 culture medium, and the macrophages were lysed using 0.025% SDS at the selected time points postinfection. Serial dilutions of the lysate were plated onto Middlebrook 7H10 agar medium supplemented with OADC and the appropriate antibiotics. Colonies were counted after incubation for 3 weeks at 37°C.
Starvation Studies-Mycobacterial cultures were grown with shaking in Middlebrook 7H9, 0.2% (v/v) glycerol, 10% OADC, 0.05% tyloxapol to an A 600 of 0.8. Cells were washed twice with PBS and then resuspended in PBS prior to leaving the cultures to stand at 37°C in sealed bottles (35). M. tuberculosis viability during starvation was determined by counting cfu from triplicate cultures over a 4-week period. Serial dilutions of the cells were performed and followed by plating onto Middlebrook 7H10 agar medium supplemented with OADC and the appropriate antibiotics. Colonies were counted after incubation for 3 weeks at 37°C. The extraction and quantification of intracellular EGT levels were performed at various time points throughout starvation as described in the methods above.

EGT Biosynthesis Pathway in M. tuberculosis-Previously,
the EGT biosynthetic pathway was identified in M. smegmatis and was described to consist of five clustered genes (20). These genes encode for ␥-glutamylcysteine ligase (egtA), a formylglycine-like enzyme (egtB), a glutamine amidotransferase (egtC), a histidine methyltransferase (egtD), and lastly a pyridoxal 5-phosphate-binding protein (egtE). Using the NCBI Basic Local Alignment Search Tool, we identified the open reading frames Rv3700c--Rv3704c (Table 1) to encode for EGT biosynthesis in M. tuberculosis.
Rv3701c and Rv3704c are predicted to encode EgtD and EgtA (Fig. 1B), respectively, and commit the necessary amino acids to the pathway, suggesting an optimal site for post-translational modification. Transposon site hybridization studies identified Rv3701c, but not Rv3704c, to be essential for growth in animal models and murine macrophages (36,37). Therefore, the requirement for Rv3701c during infection may not only implicate EGT in the pathogenesis of M. tuberculosis but may also represent a critical point in the pathway responsible for orchestrating EGT biosynthesis in response to changes in the environment of the bacilli.
Rv3701c Encodes for egtD in EGT Biosynthesis-We cloned Rv3701c from M. tuberculosis H37Rv for expression in E. coli and assayed the purified recombinant protein. Using the SAM510 assay, the methylation activity of EgtD was assessed in the presence of each of the proteinogenic amino acids with the exception of cysteine due to its interference with the assay (38). The consumption of AdoMet in the reaction was continuously monitored at 1-min intervals over a 30-min time period. No change in absorbance was observed, suggesting that histidine was the only amino acid that could undergo methylation (data not shown). We further explored substrate specificity using the same methods and ruled out methylation activity with histamine, imidazole, and the histidine derivatives 1-methyl-L-histidine, 3-methyl-L-histidine, and ␣-methyl-DL-histidine. These findings validate EgtD as a methyltransferase with high specificity for the amino acid histidine.
The methyltransferase assay we used for the enzyme characterization does not identify the number of methyl groups added to a substrate but solely indicates whether methylation is in fact occurring. Thus, we performed ESI-MS analysis on a methylation reaction containing EgtD and histidine in an attempt to isolate hercynine (␣-N,N,N-trimethylhistidine) variants. Fig.  2A illustrates that EgtD produces three methylation products, mono-, di-, and trimethylated histidine. As observed in Fig. 2B, when the same reaction was prepared in the absence of EgtD, histidine was not transformed into any of the expected methylated products. The dependence of reaction velocity on histidine concentration was measured in the presence of 5.3 M EgtD using the SAM510 assay at various concentrations of histidine (0 -15 mM). Experimental data were best fitted using the Michaelis-Menten equation (R 2 ϭ 0.99) and the V max (114.5 Ϯ 1.8 M/min⅐mg) and K m (422.0 Ϯ 37.4 M). Based on these values, we further calculated the k cat (1.3 ϫ 10 Ϫ2 s Ϫ1 ) and egtD Is Essential for EGT Biosynthesis but Not Needed for Growth-To study the necessity and role of EgtD in EGT biosynthesis, a M. tuberculosis ⌬egtD null mutant was constructed in M. tuberculosis H37Rv. We replaced the genomic Rv3701c open reading frame by specialized transduction with a hygromycin resistance cassette (Fig. 3A) and verified the knock-out strain formation by Southern blotting and PCR (Fig. 3, B and C, respectively). Intracellular EGT levels were quantified in the parental wild-type M. tuberculosis, the Rv3701c-null mutant, and a complemented strain in which the Rv3701c mutant was transformed with a plasmid containing the Rv3701c gene (pMV261:⌬egtD). EGT could not be detected in extracts prepared from the ⌬egtD mutant, whereas both the wild-type and  SEPTEMBER 18, 2015 • VOLUME 290 • NUMBER 38

Regulation of Ergothioneine Biosynthesis in M. tuberculosis
complemented strains synthesized considerable amounts of EGT (Fig. 3D). Through this experiment, we demonstrated that EgtD encoded by Rv3701c is required for EGT biosynthesis in M. tuberculosis. EgtD Is a Substrate of M. tuberculosis Ser/Thr Kinases-M. tuberculosis uses a number of signal transduction systems as a means for adaptive gene expression and regulation of metabolic processes in response to external stimuli (39). Screening of a set of M. tuberculosis STPKs for their ability to phosphorylate purified EgtD identified PknA, PknB, PknD, and PknK as kinases capable of phosphorylating EgtD (Fig. 4A). No radioactive bands were observed in the presence of the other STPKs or when EgtD was incubated in the absence of kinase. These findings indicate that EgtD is an in vitro substrate of several of M. tuberculosis STPKs, suggesting that EGT biosynthesis could be under phosphorylation control in mycobacteria.
To test whether EgtD interacts with the four identified kinases in vivo, a cell-based interaction assay using the mycobacterial protein fragment complementation assay was performed in M. smegmatis (33).The mycobacterial protein fragment complementation assay involves the reassembly of complementary fragments F1 and F2 (expressed by pUAB100) and fragment F3 (expressed by pUAB200) of murine dihydrofolate reductase enzyme, conferring resistance to trimethoprim. As illustrated in Fig. 4B, growth of M. smegmatis co-transformed with pUAB100-egtD and pUAB200 containing pknB, pknD, or pknK was present. No growth was observed in the strain co-expressing EgtD and PknA, suggesting that these two proteins do not interact under in vivo growth conditions. The interaction between EgtD and the kinases was weaker than the positive control, CFP-10 and ESAT-6, which is expected due to the transient nature of kinase-substrate interactions. Therefore, the mycobacterial protein fragment complementation assay provided evidence that EgtD is potentially a substrate for the M. tuberculosis STPKs PknB, PknD, and PknK in mycobacteria.
PknD Preferentially Phosphorylates EgtD in Vitro-It is apparent from Fig. 4A that PknD possesses the greatest phosphorylation capacity toward EgtD in comparison with PknB and PknK. To verify this, we analyzed the phosphorylation kinetics of EgtD. We confirmed that PknD possesses the greatest k cat /K m and lowest K m for the methyltransferase ( Table 2). As these findings suggest the PknD interaction to have the greatest relevance, we therefore focused on the effect of PknD phosphorylation of EgtD for the remainder of the study.
Phosphorylation Negatively Regulates EgtD Methyltransferase Activity-Phosphorylation of a protein introduces a negative charge on the targeted amino acid(s) that can ultimately affect protein activity. To investigate the effect of phosphorylation on the methylation activity of EgtD, we developed an assay designed to monitor the transfer of methyl-14 C from AdoMet to histidine. Phosphorylated and unphosphorylated EgtD was prepared using the in vitro kinase assay in the presence or absence of PknD. Following 2 h of incubation at room temperature, the kinase assays containing EgtD were added to the methylation assay containing S-[methyl-14 C]adenosyl-L-methionine and histidine. Samples were taken at various time points from the reaction and separated by TLC (Fig. 5A). The formation of [methyl-14 C]histidine was visualized by autoradiography (upper), whereas total histidine was observed on TLC plates developed with ninhydrin (lower). As expected, the combined kinase and methylation assay lacking EgtD did not form [methyl-14 C]histidine after 2 h. However, methylation activity of phosphorylated EgtD was visibly slower than that of the unphosphorylated enzyme. Histidine spots were then excised from the TLC plate, and the cellulose stationary phase was added to scintillation fluid for quantification. Counts per minute (cpm) phosphorylated and non-phosphorylated EgtD were plotted as a function of time for both enzyme sets (Fig. 5B). An approximate 20% reduction in the activity of phosphorylated EgtD was observed compared with non-phosphorylated EgtD. Although ATP concentrations were in excess in these reactions, it remains likely that a mixed population of phosphorylated and non-phosphorylated protein existed, potentially underestimat- ing the effect of phosphorylation on the methylation activity of EgtD. Nonetheless, the rate of formation of [methyl-14 C]histidine was significantly slower for phosphorylated EgtD, suggesting that PknD may negatively regulate EgtD in M. tuberculosis.
Thr 213 Is the Major Phosphorylation Site of EgtD-Mass spectrometry was used to identify the nature and location of the phosphorylation site(s) on M. tuberculosis EgtD as performed previously (24). MS/MS analysis of purified recombinant EgtD incubated in the presence of PknD identified phosphorylation on the trypsin-digested 205 AYDDPGGVTAQFNR 218 peptide that was located on Thr 213 in the C terminus of the protein (Fig.  6A). No autophosphorylation was observed in the negative control containing EgtD incubated with ATP in the absence of PknD.
The phosphopeptide identified by mass spectrometry was validated by substituting EgtD Thr 213 with alanine through site-directed mutagenesis to prevent phosphorylation. The autoradiogram in Fig. 6B shows that T213A site-directed mutagenesis resulted in abrogation of EgtD phosphorylation compared with intact EgtD. The above findings further illustrate that EgtD Thr 213 is the major phosphorylation site for PknD in vitro.
We next wanted to further validate the phosphorylation site of EgtD in a cell-based system. Due to potential interference of other mycobacterial STPKs, we decided to study the interaction between EgtD and PknD using E. coli as a heterologous host because it lacks any known STPKs. The active kinase domain of PknD and full-length recombinant wild-type EgtD were subcloned into pET-28a and pGEX-4T3, respectively, to achieve compatible expression conditions. Purified EgtD expressed in the presence and absence of PknD inside E. coli was subjected to LC-MS/MS analysis to identify the phosphorylation sites. As seen previously in vitro, EgtD co-expressed with PknD was monophosphorylated on peptide 205 AYDDPGGVTAQFNR 218 (Fig. 6C), but no phosphorylation was observed in the absence of the kinase in E. coli. From these results, we concluded that Thr 213 is the major phosphorylation site of EgtD both in vitro and in an in vivo cell-based system.
M. tuberculosis EgtD Phosphomimetic Mutant Synthesizes Lower EGT Levels-Introduction of a negative charge through the substitution of acidic residues such as Asp or Glu has been shown previously to mimic phosphorylation of a protein with regard to functional activity (25, 27, 28, 31). We wanted to further determine the effect of phosphorylation on EGT levels in M. tuberculosis and thus constructed EgtD variant strains with constitutively altered activities. The ⌬egtD mutant was transformed with pMV261 derivatives allowing constitutive expression of different egtD alleles under the control of the hsp60 promoter: egtD_WT, phosphomimetic egtD_T213E, and phosphoablative egtD_T213A. As shown in Fig. 7A, EGT levels in the EgtD_T213A-overexpressing strain were comparable with those in EgtD_WT with a minor reduction likely due to the nature of the amino acid substitution. In contrast, overexpression of EgtD_T213E was accompanied by a significant decrease in EGT levels in comparison with EgtD_T213A and EgtD_WT, providing stronger evidence that EgtD phosphorylation negatively regulates EGT biosynthesis in M. tuberculosis.
To further confirm the effect of EgtD phosphorylation on EGT biosynthesis, we investigated EGT levels in the CDC1551 pknD:Tn mutant (Fig. 7B). Because PknD is expressed during mid-log phase (40) and we observed phosphorylation to negatively regulate EgtD methylation activity and EGT biosynthesis, we hypothesized that the pknD mutant would exhibit higher levels of EGT than wild-type M. tuberculosis in culture. Intracellular EGT was extracted from CDC1551 wild type and pknD:Tn mutant at an A 600 of 0.5. As expected, the pknD:Tn mutant had significantly higher levels of EGT than wild type, indicating that in the absence of PknD regulation the enzymatic activity of EgtD was enhanced.
Intracellular EGT Levels Increase during Late Logarithmic Phase-We explored a potential role for EGT in M. tuberculosis growth. In vitro, MSH was shown to be essential for M. tuberculosis growth in the absence of catalase (10,41). As EGT was previously suggested to have overlapping functions with MSH in mycobacteria (42), we tested whether the ⌬egtD mutant possesses similar growth defects under in vitro growth conditions. No differences in growth were observed in rolling or standing cultures containing Middlebrook 7H9 supplemented with OADC or albumin, dextrose, and sodium chloride for the wildtype, ⌬egtD, and ⌬egtD complemented strains after 14 days (data not shown). However, when we quantified intracellular EGT levels during the various stages of H37Rv wild-type growth, we observed EGT to begin accumulating during late logarithmic phase (Fig. 8A). These findings suggest that M. tuberculosis may be involved in conditions involving growthlimiting factors such as nutrient limitation, endogenous waste production, or low pH.
EGT Is Required for M. tuberculosis Survival during Starvation-To further understand the relevance of M. tuberculosis regulation of EGT biosynthesis, we further sought out conditions where intracellular EGT levels may be regulated. Microarray studies analyzing the global adaptation of M. tuberculosis to nutrient starvation discovered pknD to be down-regulated after 4 h (43). As we described PknD to be a negative regulator of M. tuberculosis EgtD, we therefore hypothesized that EGT levels will increase during starvation.
First, to identify whether EGT plays a role in later stages of starvation, we grew wild-type H37Rv in PBS containing 0.05% tyloxapol and extracted intracellular EGT from cells at five time points over a 6-week period. After 1 week, we observed an increase in intracellular EGT levels that was maintained over 6   (Fig. 8B). Because intracellular EGT levels are elevated during long term starvation, it was important for us to determine whether EGT is required for the survival of M. tuberculosis under these conditions. We starved our cultures in PBS containing 0.05% tyloxapol and monitored bacterial survival by counting cfu for up to 4 weeks. The H37Rv wild-type and the ⌬egtD complemented strains did not show a significant loss in viability over this time period; however, the ⌬egtD mutant cfu were reduced by a magnitude greater than one log after 3 weeks (Fig. 9A). From these findings, we conclude that M. tuberculosis requires EGT for its survival during long term starvation. Growth of M. tuberculosis ⌬egtD Is Attenuated in Murine J774A.1 Macrophages-As EGT has been shown to scavenge a number of reactive oxygen and nitrogen species in vitro, we wanted to assess whether EGT was able to protect M. tuberculosis from bactericidal micromolecules generated by the macrophage. Controversy still exists as to whether human macrophages produce adequate levels of NO. Because the expression of functional inducible NO synthase in mouse macrophages has been clearly demonstrated, we decided to study the role of EGT in J774A.1 macrophages (44). Independent infection of the J774A.1 cells showed an approximate half-log reduction in cfu at 120 h in the ⌬egtD mutant (Fig. 9B). The minor impact EgtD has on the survival of M. tuberculosis inside the macrophage suggests that EGT is probably needed to a greater extent at later stages of infection.

Discussion
The inherent energy cost of synthesizing EGT in mycobacteria led us to believe that biosynthesis is likely regulated as it would not be energetically economic for the cell to continuously canalize numerous metabolites into the pathway. The identification of Rv3701c as part of the EGT biosynthetic cluster in mycobacteria and its implication in pathogenesis prompted us to further characterize its function in M. tuberculosis. Using a combination of genetic and biochemical assays, we confirmed the role of Rv3701c in M. tuberculosis EGT biosynthesis and annotated the enzyme as EgtD because of its activity as a histidine methyltransferase and involvement in the production of hercynine. Partially methylated histidine products were also formed in the EgtD methylation reaction. In earlier work, Reinhold et al. (45) also observed the conversion of histidine to ␣-N-methyl-L-histidine and ␣-N,N-dimethyl-L-histidine derivatives and hercynine in the presence of a cell-free extract of Neurospora crassa mycelium. These methylated intermediates are the preferential substrates of EgtD in N. crassa (45) and M. smegmatis (20), indicating that EgtD methylates histidine in a stepwise reaction due to an increased affinity for the methylated intermediates (41). The specificity of EgtD for these methylated derivatives likely acts as a regulatory mechanism, limiting the quantity of histidine taken up for EGT biosynthesis.
We have found that similarly to M. smegmatis EgtD is essential in the production of EGT in M. tuberculosis (42). These findings along with the commitment of histidine to the pathway suggested EgtD to be a likely candidate for regulation in EGT biosynthesis. We identified EgtD to act as a substrate of PknD and negatively regulate EGT biosynthesis in M. tuberculosis. The fact that Thr 213 was demonstrated to be the site of phosphorylation in vitro and inside E. coli combined with the negative effect of phosphorylation on EgtD activity and EGT biosynthesis points to Thr 213 as a critical residue in catalysis. Through the crystallization of M. smegmatis EgtD, Vit et al. (41) demonstrated Thr 213 to be responsible for binding the imidazole ring of histidine through a water-mediated hydrogen bond. Therefore, the introduction of a phosphate at this residue impedes the interaction between the methyltransferase and its substrate, resulting in the observed loss in EgtD activity.
The decline (20%) in the catalytic efficiency observed when EgtD was phosphorylated in vitro (Fig. 5) varied from the levels of EGT synthesized from the genetic experiments (Fig. 7) typical to differences between in vitro and in vivo experiments. Enzymatic reactions proceed with greater efficiency in their natural environments, which likely describes the enhanced suppression of EGT production in the pknD:Tn mutant (35%) in comparison with our in vitro experiment (20%). Furthermore, our in vitro assay monitors the formation of methylated histidine and not the synthesis of ergothioneine (which was measured in the EgtD phosphomutant strains). We suggest that phosphorylation of EgtD inhibits the di-or trimethylation of histidine, preventing EGT biosynthesis from proceeding.
To gain a greater appreciation for the regulation of EGT biosynthesis in M. tuberculosis and to clarify its role in pathogenesis, we assessed the survival of the ⌬egtD mutant in murine macrophages. During infection, the mutant was only slightly attenuated (half-log) at 120 h, an unexpected result because of the reputation of EGT as an antioxidant. Perhaps MSH provides a compensatory effect in M. tuberculosis in the absence of EGT ex vivo; however, current work does not show increased levels of MSH in the absence of EGT (11,42). The pknD:Tn mutant showed no difference in viability when infecting J774A.1 macrophages (13), further supporting the notion that EGT levels have a limited effect on the survival of M. tuberculosis in murine macrophages (43). Additional studies examining the survival of M. tuberculosis ⌬egtD in gp91 PhoxϪ/Ϫ and NOS2 Ϫ/Ϫ macrophages are needed to attribute a clearer role to EGT protecting the bacilli against host-generated reactive oxygen and nitrogen species.
We did not observe any growth defect between the ⌬egtD mutant and the wild-type and complemented strains when grown in the presence or absence of catalase. These findings differ from those observed with MSH where catalase is essential when growing M. tuberculosis in vitro (10). As a result, we believe that EGT does not provide protection against peroxides during in vitro growth of M. tuberculosis. Studies examining the survival of M. smegmatis ⌬egtD when challenged with cumene hydroperoxide or tert-butyl hydroperoxide found the mutant to be similar to wild type (42), which additionally suggests little to no involvement of EGT in the elimination of peroxides in mycobacteria.
Additional growth studies identified intracellular EGT levels to be correlated with M. tuberculosis growth. Specifically, a substantial rise was recorded when M. tuberculosis reached late logarithmic phase, and an even greater amount of EGT accumulated during stationary phase. Elevated levels of intracellular EGT were also observed during the stationary phase of M. smegmatis by Ta et al. (11), but the reverse was shown by Sao Emani et al. (42). Stationary phase is characterized by a number of growthlimiting factors, one of which is nutrient deprivation. An earlier study using nutrient starvation in M. tuberculosis identified pknD to be significantly down-regulated 4 h following starvation (35). Because EgtD is negatively regulated by PknD, we expected to observe increased levels of intracellular EGT during starvation. Despite being metabolically costly to the cell, M. tuberculosis maintained elevated intracellular EGT levels over 5 weeks, suggesting that EGT is needed during long term starvation. It is possible that the discrepancy in EGT levels between our starved (weeks 1-6; ϳ300 ng/10 8 cells) and stationary  phase (A 600 ϭ 2.0; 750 ng/10 8 cells) cultures were the result of M. tuberculosis sensing the gradual depletion of nutrients under non-chemostatic culture conditions, providing the bacilli with the opportunity to accumulate more EGT before entering stationary phase. We further assessed the contribution of EGT to the survival of M. tuberculosis during nutrient starvation and found the ⌬egtD mutant to have over a 1-log reduction in viability at weeks 3 and 4 in comparison with the wildtype strain, confirming the role of EGT in long term starvation.
Earlier EGT was shown to accumulate during nutrient-limited growth of Schizosaccharomyces pombe under a wide range of glucose concentrations (0 -111 mM) (46). EGT was also found to accumulate 1 h following nitrogen starvation in S. pombe (47). The doubling time of S. pombe is substantially shorter compared with M. tuberculosis, and thus it is plausible that EGT is required by the cell sooner and is synthesized more quickly than we observed in M. tuberculosis. Interestingly, these low glucose and nitrogen conditions in S. pombe are characterized by the cells transitioning from a dividing to a quiescent state (46,47).
The fungi Colletotrichum graminicola and N. crassa were found to contain 17 and 5 times the amount of EGT in their conidial fractions than mycelium, respectively (48). The exact function of accumulating EGT in the conidia of ascomycetous fungi is unknown; however, EGT was shown to play a role in conidial longevity and germination (48,49). Analysis of the ability of endogenous EGT to protect N. crassa against the toxic effects of menadione and cupric sulfate showed no effect on mycelial growth or spore germination (48). EGT also did not offer any protection against DNA damage when conidia were exposed to 254 nm UV light (49). The only antioxidant property EGT exerted in N. crassa was that against exogenous peroxide where the germination of the EGT-minus mutant was significantly more sensitive to tert-butyl hydroperoxide than the wild type (48). However, the quiescent S. pombe deletion and overexpression EGT mutants did not show any sensitivity or resistance to hydrogen peroxide or tert-butyl hydroperoxide (18). Based on the above findings that EGT provides minimal protection in microorganisms against oxidative stress and in conjunction with its chemical properties (poor reducing power and resistance to auto-oxidation), we postulate that the primary function of EGT in M. tuberculosis is as not as an antioxidant.
As seen with other sulfur-containing compounds, the primary function of EGT could be in biosynthesis or energy metabolism. For example, EGT was recently observed to be involved in C8 sugar transfer and activation in the biosynthesis of the antibiotic lincomycin A in Streptomyces lincolnensis (50). In this reaction, EGT served as a carrier during the first glycosylation step to channel the lincosamine unit for further condensation with a methylated derivative of 4-propyl-L-proline. Interestingly, the reaction does not consume but recycles EGT through a thiol exchange with MSH, which may explain the low abundance of this molecule in cells. In addition to glycosylation reactions, EGT was also found to be involved in the biosynthesis of two bohemamine-type pyrrolizidine alkaloids (spithioneines A and B) synthesized by the marine bacterium Streptomyces spinoverrucosus. During spithioneine A and B synthesis, EGT is incorporated directly to form a polyketide, which is thought to result from an atom of EGT carrying out a nucleophilic attack on the epoxide of the bohemamine (51).The association of EGT with glycosylation reactions and polyketide synthesis suggests its involvement in enzymatic reactions. Although unable to synthesize lincomycin A, M. tuberculosis encodes a number of polyketide synthases and performs a number glycosylation reactions that are primarily involved in the synthesis of cell wall lipids. Interestingly, under starvation conditions and in a murine model of chronic infection, sulfolipid synthesis is upregulated, whereas the majority of cell wall synthesis genes are repressed (35,52). The direct role sulfolipids play in M. tuberculosis virulence remains unclear; however, these lipids have been identified to negatively regulate the intracellular survival of M. tuberculosis in a species-specific manner and mediate the susceptibility of M. tuberculosis to human cationic antimicrobial peptides (53).
A number of studies have implicated PknD in regulating the adaptation of M. tuberculosis to extracellular stressors through cell wall remodeling. During cell wall synthesis, PknD positively regulates the transport of phthiocerol dimycocerosate (54). Later work also described one of the substrates of PknD, osmosensory protein A (OprA), to be up-regulated in the presence of increasing extracellular osmolarity and to enable adaptation through modifying peptidoglycan thickness (55, 56). Several other enzymes involved in the synthesis of cell wall components are also substrates of PknD, including malonyl-CoA:acyl carrier protein transacylase (FabD) and the ␤-ketoacyl-acyl carrier protein synthases KasA and KasB (39).
It is evident that EGT plays a role in the long term survival and non-replicating/dormant state of bacteria, fungi, and yeast, but whether its mechanism involves cell wall synthesis or long term energy storage or whether it acts as a cytoprotectant remains to be discovered. In tuberculosis, both the phagosome and granuloma are thought to be sites of nutrient deprivation during M. tuberculosis infection (57,58). Starvation and hostile conditions in the host are suspected to act as external triggers that terminate growth and render the bacilli phenotypically resistant to drugs (59). This state of non-replicating persistence not only makes latent disease difficult to eradicate, but it is now accepted that this subset of bacteria is responsible for prolonged treatment in active tuberculosis cases. Therefore, it is of relevance to clarify the role of EGT as an antioxidant and further investigate enzymatic reactions related to metabolism that may involve EGT in M. tuberculosis. Studies are currently underway in our laboratory to address such questions and to further understand the contribution of EGT to M. tuberculosis physiology and pathogenesis.