Unveiling the Novel Dual Specificity Protein Kinases in Bacillus anthracis

Background: Characterization of Ser/Thr protein kinases present in Bacillus anthracis genome. Results: Dual specificity protein kinases were identified, of which one is similar to the eukaryotic DYRK superfamily. Conclusion: B. anthracis has lost key tyrosine kinases and gained novel dual specificity kinases. Significance: Reporting the first prokaryotic enzyme similar to DYRKs shows that this class of enzymes is not restricted to eukaryotes. Dual specificity protein kinases (DSPKs) are unique enzymes that can execute multiple functions in the cell, which are otherwise performed exclusively by serine/threonine and tyrosine protein kinases. In this study, we have characterized the protein kinases Bas2152 (PrkD) and Bas2037 (PrkG) from Bacillus anthracis. Transcriptional analyses of these kinases showed that they are expressed in all phases of growth. In a serendipitous discovery, both kinases were found to be DSPKs. PrkD was found to be similar to the eukaryotic dual specificity Tyr phosphorylation-regulated kinase class of dual specificity kinases, which autophosphorylates on Ser, Thr, and Tyr residues and phosphorylates Ser and Thr residues on substrates. PrkG was found to be a bona fide dual specificity protein kinase that mediates autophosphorylation and substrate phosphorylation on Ser, Thr, and Tyr residues. The sites of phosphorylation in both of the kinases were identified through mass spectrometry. Phosphorylation on Tyr residues regulates the kinase activity of PrkD and PrkG. PrpC, the only known Ser/Thr protein phosphatase, was also found to possess dual specificity. Genistein, a known Tyr kinase inhibitor, was found to inhibit the activities of PrkD and PrkG and affect the growth of B. anthracis cells, indicating a possible role of these kinases in cell growth and development. In addition, the glycolytic enzyme pyruvate kinase was found to be phosphorylated by PrkD on Ser and Thr residues but not by PrkG. Thus, this study provides the first evidence of DSPKs in B. anthracis that belong to different classes and have different modes of regulation.


EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-E. coli strain DH5␣ (Novagen) was used for cloning and BL21 (DE3) (Stratagene) for the expression of recombinant proteins. E. coli cells were grown and maintained with constant shaking (220 rpm) at 37°C in LB broth supplemented with 100 g/ml ampicillin when needed. B. anthracis Sterne strain was grown in LB broth at 37°C with shaking at 220 rpm. For solid medium, LB-agar was used for both E. coli and B. anthracis, containing the appropriate antibiotic.
RNA Isolation from B. anthracis and Quantitative Real-time PCR-RNA was isolated from B. anthracis Sterne strain cells grown to early log (A 600 ϭ 0.2-0.3), midlog (A 600 ϭ 0.8 -1.0), late log (A 600 ϭ 1.5-1.7), and stationary phases (A 600 Ͼ2.2, ϳ30 h), using the protocols described previously with some modifications (38 -40). To isolate RNA, cells were centrifuged at 6,000 ϫ g for 15 min and resuspended in 400 l of hot lysis buffer (50 mM Tris (pH 8.0), 1 mM EDTA, and 1% SDS) kept at 65°C followed by incubation with lysozyme (0.25 mg/ml) at 37°C for 10 min. Lysis was carried out by transferring the sus-pension to 1.5-ml vials containing 400 l of acidified glass beads (425-600 nm; Sigma) and 500 l of phenol kept at 65°C. This was followed by incubation at 65°C for 15 min with vortexing for 30 s after every 5 min. The upper aqueous phase containing RNA was then separated by centrifugation at 9,500 ϫ g for 15 min at 4°C, to which equal volumes of TRIzol (Invitrogen) and 100 l of chloroform/ml of TRIzol was added, mixed well, and centrifuged at 9,500 ϫ g for 15 min at 4°C. RNA from the upper aqueous phase was then precipitated with LiCl 2 (0.5 M final) and three volumes of ice-cold isopropyl alcohol followed by incubation at Ϫ80°C for 2 h and centrifugation at 16,000 ϫ g for 20 min at 4°C. The pellet was then washed with 70% ethanol, and the air-dried pellet was dissolved in 20 l of nuclease-free water and stored at Ϫ80°C.
Before performing cDNA synthesis, RNA was treated with DNase (Ambion) according to the manufacturer's protocol to remove the traces of genomic DNA. cDNA was made from 400 ng of RNA of each phase, according to the protocol provided by the supplier (Applied Biosystems), which was then used for checking the expression of the gene with gene-specific primers. For quantitative real-time PCR, a standard curve was prepared using serial dilutions of the kinase clone for the corresponding kinase gene in different copy numbers (copy numbers 10 1 , 10 2 , 10 3 , 10 4 , 10 5 , and 10 6 ). Real-time PCR was performed using SYBR Green master mix according to the manufacturer's instructions. For the standard curve, reactions were run in triplicates, under the same conditions and using the same primers as that for the kinase genes. For expression analysis, 1 l of cDNA (each phase) was used in a 15-l PCR, in duplicates, together with no template controls. The fusA gene, which encodes translation elongation factor G, was used as an external control (41). All primers were sequence-specific for each gene analyzed, with PCR products between 70 and 100 bp.
Gene Manipulation-For cloning of PrkC c (bas3713, the cytosolic region of amino acids 1-337), PrkD (bas2152), PrkG (bas2037), PrpC (bas3714), and BasPyk (bas4492), the genes were PCR-amplified from B. anthracis genomic DNA using gene-specific forward and reverse primers. The resulting PCR products were then digested and cloned into the BamHI and XhoI sites of pProEx-HTc (Invitrogen) and/or pGEX-5X-3 (GE Healthcare) vectors previously digested with the same enzymes. The clones were confirmed with restriction digestion and DNA sequencing (TCGA, New Delhi, India). The details of primers and plasmids are provided in supplemental Table S1.
To generate essential lysine mutants and phosphorylation site derivatives of both kinases, site-directed mutagenesis was carried out using the QuikChange XL site-directed mutagenesis kit (Stratagene) as per the manufacturer's instructions, using HTc-PrkD and HTc-PrkG as templates. The clones were confirmed with DNA sequencing (TCGA).
Protein Expression and Purification-The recombinant plasmids were transformed and overexpressed in E. coli BL-21 (DE3) cells. The proteins were purified by Ni 2ϩ -NTA or glutathione-Sepharose affinity columns (Qiagen) as described earlier (42). The purified proteins were visualized by 10 -15% SDS-PAGE, and the concentrations were estimated by a Bradford assay (Bio-Rad).
Immunoblotting Analysis-To detect the phosphorylated residues of PrkD and PrkG, proteins were resolved by SDS-PAGE and transferred onto a nitrocellulose membrane. Blots were then blocked with 3% bovine serum albumin in phosphate-buffered saline containing 0.1% Tween 20 (PBST) overnight at 4°C. This was followed by incubation with primary antibodies anti-Tyr(P) (Upstate Biotechnology) and anti-Ser(P)/Thr(P) (Invitrogen) at 1:10,000 dilution and goat antirabbit IgG secondary antibodies (Bangalore Genei) (1:10,000) for 1 h each at room temperature. The blots were developed using the SuperSignal West Pico Chemiluminescent Substrate kit (Pierce) according to the manufacturer's instructions. Autophosphorylated PrkC was kept as a negative control for the ␣-Tyr(P) blot because it has been reported to be phosphorylated on Ser and Thr residues (30). Phosphorylated myelin basic protein (MyBP) phosphorylated by the Tyr kinase Abl (New England Biolabs) was used as a positive control.
Generation of Polyclonal Antibodies against PrkG in Rabbits and Verification of PrkG Expression-Purified His 6 -PrkG protein (400 g) was emulsified in Freund's complete adjuvant (1:1 ratio) and injected into rabbits. Subsequently, injections of 400 g each of His 6 -PrkG in 1 ml of Freund's incomplete adjuvant were given three times at 21-day intervals. Fourteen days after the final injection, animals were bled, and titers of anti-His 6 -PrkG were determined by ELISA as described previously by Koul et al. (43). The molecular weight of PrkG was verified using 40 g of B. anthracis whole cell lysate. The lysate was resolved by SDS-PAGE along with suitable positive (His 6 -PrkG (1 g)) and negative (glutathione S-transferase (2 g)) controls. Proteins were transferred onto nitrocellulose membrane (Bio-Rad), and a standard procedure for immunoblotting was followed using 1:20,000 dilutions of primary and secondary antibodies, as described under "Immunoblotting Analysis." In Vitro Kinase and Phosphatase Assays-In vitro phosphorylation assays of 25 pmol of PrkC c , PrkD, and PrkG or their derivatives were carried out in kinase buffer (20 mM HEPES (pH 7.2), 1 mM DTT, 5 mM MgCl 2 , and 5 mM MnCl 2 for all of the kinases) containing 2 Ci of [␥-32 P]ATP (Board of Radiation and Isotope Technology, Hyderabad, India) followed by incubation at 25°C for 20 min. The ionic requirement was estimated by in vitro kinase assays with 25 pmol of kinase in 20 mM HEPES (pH 7.2) and 1 mM DTT, together with either metal ion (MgCl 2 /MnCl 2 ) and 2 Ci of [␥-32 P]ATP. Nucleotide specificity assays of PrkD and PrkG were performed similarly by using 2 Ci of [␥-32 P]ATP or [␥-32 P]GTP (Board of Radiation and Isotope Technology). For substrate phosphorylation, 5 g of MyBP and 2-5 g of BasPyk were used. Proteins were separated by 10 -15% SDS-PAGE and analyzed by a PhosphorImager (FLA 2000, Fuji).
In vitro dephosphorylation assays were carried out by adding PrpC (ϳ35 pmol) to the reactions after kinase assays, followed by additional incubation for different times up to 30 min at 37°C, as also described by Sajid et al. (44). Reactions were terminated by 5ϫ SDS sample buffer followed by boiling at 100°C for 5 min. Proteins were separated by 10 -15% SDS-PAGE and analyzed by a PhosphorImager.
To check the phosphatase specificity, non-radioactive P i release assays were performed. The sequence of phosphopep-tides (Millipore) that were used in the assay was as follows: KRpTIRR (catalog no. 12-219), RRApSVA (catalog no. 12-220), and RRLIEDAEpYAARG (catalog no. 12-217). The assays were performed according to the instructions provided in the manual Ser/Thr phosphatase assay system by Promega.
Kinase Inhibition Assays-For inhibition assays, the specified inhibitor (Calbiochem and Sigma) was added in the concentration gradient in the kinase buffer. In the resultant mixture, PrkD/PrkG (25 pmol) was added, and further in vitro kinase assays were carried out as discussed before. The control reactions contained DMSO (solvent in which inhibitors were dissolved).
Construction of B. anthracis mcsB Deletion Strain (Bas⌬mcsB)-The Cre-loxP genetic modification method was used to introduce precise genetic knockouts into the B. anthracis gene encoding McsB as previously discussed in detail by Pomerantsev et al. (45). Precisely, the pSC vector was used to produce a deletion of region (800 bases deleted of a total of 1062 nucleotides of mcsB) in twostep recombination using the primer pair sequence provided in supplemental Table S1. Plasmid pCrePAS was used for elimination of DNA regions containing a spectinomycin resistance cassette located between two similarly oriented loxP sites. The deletion of the mcsB gene in the Bas⌬mcsB strain was confirmed by PCR and DNA sequencing.
B. anthracis Growth Inhibition-B. anthracis primary culture was inoculated from a single colony on an LB agar plate and grown until reaching an A 600 of 1.0. Secondary cultures were then inoculated from the primary culture (0.1%). To study the effect of genistein, cultures were grown in the presence of 20 and 100 M, in duplicate. Absorbance was measured at 600 nm and plotted against time.
Phosphoamino Acid (PAA) Analysis-For this analysis, 100 pmol of autophosphorylated 32 P-PrkD and 32 P-PrkG was separated by SDS-PAGE after an in vitro kinase assay and electroblotted onto Immobilon PVDF membrane (Millipore). PAA analysis by two-dimensional thin layer electrophoresis (2D-TLE) was performed as described previously (46). After 2D-TLE, the plate was dried and sprayed with 1.0% ninhydrin. The plate was kept at 55°C for 15 min, and three spots corresponding to Ser(P), Thr(P), and Tyr(P) were observed. The plate was kept for exposure under x-ray for 3-4 days. The autoradiogram was analyzed by overlapping the corresponding TLC plate to identify the radiolabeled Ser(P)/Thr(P)/Tyr(P) residues. Substrates 32 P-MyBP (phosphorylated by PrkD or PrkG) and 32 P-BasPyk (phosphorylated by PrkD) were also similarly analyzed.
Identification of Phosphorylation Sites-For protein analysis by mass spectrometry, a Coomassie-stained protein band was cut from a polyacrylamide gel and digested with trypsin as described previously (47). Proteolytic fragments were analyzed by LC-MS/MS with electrospray ionization using an Applied Biosystems 5600 Triple-TOF mass spectrometer (AB-Sciex, Toronto, Canada). Interpretation of spectra was performed manually with the aid of Protein Prospector software (University of California, San Francisco).
Enzymatic Assay of BasPyk and Determination of Kinetic Parameters-Pyruvate kinase activity was determined at 37°C by the coupled lactate dehydrogenase assay as described previously (50, 51) except for some modifications. The reaction mixture (100 l) contained 100 mM Tris-HCl buffer (pH 7.4), 10 mM MgCl 2 , 1 mM NADH (Sigma), 20 mM ADP (Sigma), pH 7.2, 1-8 mM phosphoenolpyruvate (Sigma) and 15 g of lactate dehydrogenase (Sigma). The reaction was started by the addition of 10 nM enzyme. The reaction was performed for 10 min, and absorbance was measured at 340 nm (Bio-Rad Microplate reader 680 XR). K m and V max values were determined by nonlinear regression analyses (curve fit) using GraphPad Prism software. Data (specific activity) from three independent experiments were plotted, and results are presented as average values Ϯ S.D.

RESULTS
In Silico Analysis of B. anthracis STPKs-To identify Ser/Thr protein kinases, a BLASTp search (with default settings) was performed in the B. anthracis Sterne strain non-redundant protein sequence database (NCBI, National Institutes of Health) using the sequence of BASPrkC. We identified two genes, bas2152 (named PrkD) and bas2037 (named PrkG), which apparently encode Ser/Thr protein kinases. Additionally, two other STPK-encoding genes were also identified as bas2464 and bas2527 by pkinase software (pfam). Sequence analysis also revealed that B. anthracis has five putative STPKencoding genes, whereas B. subtilis has four such putative STPK-encoding genes (supplemental Table S2).
Although PrkC and its homologs are known to be activated through C-terminal sensor PASTA domains, PrkD and PrkG seem to only possess N-terminal kinase domains, as suggested by SMART domain analysis (supplemental Fig. S1). Alignment of catalytic domain sequences of both PrkD and PrkG with PrkC revealed that all 12 signature motifs, which are characteristic of their eukaryotic counterparts, are more conserved in PrkD in comparison with PrkG (Table 1 and supplemental Fig. S2). The main differences were found in conserved catalytic Arg and Asp residues, which divide these kinases into RD and non-RD classes (52,53). PrkD is an RD kinase, and therefore phosphorylation of the activation loop residues governs its activity. PrkG belongs to the non-RD kinase class with no activation loop. Other key differences between the two kinases are shown in Table 1.
To identify homologs of these kinases in other Bacillus species, a BLASTp search was performed using the protein sequences of PrkD and PrkG. The results indicated that PrkD homologs are conserved in the Bacillus species (Fig. 1A), whereas PrkG is unique to the pathogenic Bacillus cereus group ( Fig. 1B and Table 1). PrkG homologs in the B. cereus group have tetratricopeptide repeat (TPR) domains at the C terminus in addition to the kinase domain. Therefore, in order to understand the differences in domain organization, flanking genes of PrkG were also compared with their homologs. It was found that the kinase domain (Bas2037) and TPR domain (Bas2038) exist as two separate proteins in most B. anthracis strains sequenced so far (B. anthracis strains A1055, A0442, A0465, CNEVA-9066, and Kruger B being the only exceptions) but not in other B. cereus group species. A nonsense mutation in most B. anthracis strains resulted in a stop codon (Fig. 1B), and therefore a truncated form of the kinase is expressed in the bacterium, as confirmed by immunoblotting with polyclonal antibodies against PrkG (supplemental Fig. S3).
Expression of B. anthracis Kinases during Different Growth Phases-The presence of multiple kinase-encoding ORFs in the B. anthracis genome intrigued us to define their transcriptional pattern and to determine the general expression profiles. We quantified the transcript number of each of the three kinase genes by quantitative real-time PCR and found variation in expression of these regulatory enzymes in various stages of growth (Fig. 1C). The bas3713 (prkC) was consistently expressed in early log (A 600 ϭ 0.2-0.3), midlog (A 600 ϭ 0.8 -1.0), and late log (A 600 ϭ 1.5-1.7) phases, whereas the expression decreased in stationary phase (A 600 Ͼ 2.2, ϳ30 h). The expression of bas2152 (prkD) was lower than that of bas3713 in all growth phases, except in stationary phase, where it was approximately 3 times more abundant than bas3713. Transcripts of bas2037 were also present in all four stages, and expression was comparable with bas3713, with the exception of the midlog phase, where bas2037 was almost 2-fold more abundant than bas3713. Copy number was calculated by constructing a standard curve using a known plasmid with the same gene as the standard. Translational enzyme elongation factor G (fusA) was used as an external control, whose expression was higher than all kinases in all growth phases except for stationary phase (Fig. 1C). This expression profile of fusA corroborates with earlier reports (41,54), where it has been used as an external control and a similar expression profile has been shown.
These data indicate that all of the kinases are expressed during the growth cycle, albeit at different levels.
Biochemical Characterization of PrkD and PrkG-Signal transduction proteins, such as kinases, play an important role in the adaptation of living organisms to changing environments (2). Therefore, it is crucial to elucidate their regulation. In order to characterize these kinases, the genes coding PrkD and PrkG were cloned, overexpressed, and purified from E. coli as His 6 tag fusion proteins. To analyze their nucleotide specificity and ionic requirements, in vitro kinase assays were performed with either [␥-32 P]ATP or [␥-32 P]GTP in the presence of different divalent ions. Proteins were resolved by SDS-PAGE and then analyzed on an autoradiogram for comparison of phosphorylation signal by quantification with ImageGauge software (Fuji). Both kinases were found to be ATP-dependent and required Mg 2ϩ or Mn 2ϩ ions for activity ( Fig. 2A). Under the conditions tested, both kinases were found to be maximally active at 25°C ( Fig. 2B and supplemental Fig. S4A). Therefore, all further assays were performed at 25°C in the presence of Mg 2ϩ and Mn 2ϩ . Furthermore, time-dependent phosphorylation assays were performed to calculate the activation time for these kinases. Although both PrkD and PrkG were able to autophosphorylate within 1 min, optimal phosphorylation was achieved within 5 min ( Fig. 2C and supplemental Fig. S4B).
Identification of Autophosphorylation Sites in PrkD and PrkG-Most STPKs exhibit autophosphorylation at multiple sites, the loss of which may impair the kinase activity. Therefore, PAA analysis of PrkD and PrkG was performed after an in vitro kinase assay by 2D-TLE to identify the phosphorylated residue(s). Interestingly, both kinases were found to be phosphor- ylated on Tyr residues in addition to Ser and Thr (Figs. 3A and 4A). This was further confirmed by immunoblotting with ␣-Tyr(P) antibodies (supplemental Fig. S5) as well as ␣-Ser(P) and ␣-Thr(P) antibodies (data not shown).
To prove this distinctive nature of PrkD and PrkG, it was pertinent to determine the sites of phosphorylation. Mass spectrometry analysis of PrkD identified several phosphorylated Ser/Thr residues and three Tyr residues (Fig. 3B). Of these, one  In vitro kinase assays were performed followed by autoradiography and analyzed by ImageGauge software. Phosphorylation of wild-type PrkD was taken as 100% and used to calculate the relative phosphorylation. Each experiment was repeated three times with fresh purification lots of the kinase and its derivatives. The corresponding SDS-PAGE and autoradiogram are shown in supplemental Fig. S6, A and B. phosphorylated residue (Tyr 182 ) and two probable phosphoresidues (Thr 180 and Thr 181 ) were present in the activation loop of the PrkD catalytic domain. A multiple sequence alignment of the activation loop of PrkD and its homologs revealed that Tyr 182 is conserved only in the pathogenic B. cereus group (Fig.  1A). Therefore, to study the impact of Tyr phosphorylation on kinase activation and to gain a better understanding of its effect on the phosphorylation of PrkD, all three residues were mutated to generate single mutants PrkD T180A , PrkD T181A , and PrkD Y182F as well as the PrkD T181A/Y182F double mutant. Equal amounts of PrkD and its mutants were used in the in vitro kinase assays with [␥-32 P]ATP, resolved by SDS-PAGE, and then analyzed on an autoradiogram for loss of phosphorylation signal by quantification with ImageGauge software. The analysis indicated that maximum loss in activity was observed in the PrkD Y182F mutant (Fig. 3C and supplemental Fig. S6A), suggesting that this site is central for PrkD activation. The PrkD T181A mutant also exhibited a loss in signal, and the importance of Thr 181 and Tyr 182 was also confirmed by generating the double mutant PrkD T181A/Y182F , which showed ϳ70% loss in phosphosignal. These results indicate that the two major phosphorylation sites in the PrkD activation loop are Thr 181 and Tyr 182 . Interestingly, PrkD T180A showed hyperphosphoryla-tion, which indicates that it may not be phosphorylated and may only have an accessory role in controlling the phosphorylation within the activation loop. The importance of Tyr 182 was reflected by minor loss of kinase activity between the PrkD T180A/T181A mutant and PrkD. Furthermore, to find the effect of Ser 162 , the only phosphorylated Ser residue present adjacent to the activation loop, we also generated PrkD S162A and PrkD S162A/Y182F mutants. Both mutants showed loss in autophosphorylation activity compared with PrkD, with more prominent losses exhibited by PrkD S162A/Y182F (Fig. 3D and supplemental Fig. S6B) Together, these results suggest that PrkD is a dual specificity kinase.
Mass spectrometry analysis of PrkG also verified it to be a dual specificity kinase, which autophosphorylates at Ser, Thr, and Tyr residues. In PrkG, the phosphorylation sites are scattered throughout the protein as Ser 108 , Ser 196 , Thr 245 , Tyr 262 , and Tyr 349 (Fig. 4B). Because there is no activation loop present in this kinase, the contribution of all of these sites to PrkG activation was individually assessed. Site-directed mutagenesis was performed to create PrkG S108A , PrkG S196A , PrkG T245A , PrkG Y262F , and PrkG Y349F mutants, and the kinase activity of these mutants was compared with native PrkG. As shown in Fig. 4C and supplemental Fig. S6C, PrkG Y349F showed the maximum loss of phosphorylation compared with the other mutants. For the other mutants, minor losses were observed, with the exception of PrkG S108A , which was hyperphosphorylated compared with native PrkG. These results suggest that Tyr 349 is the major residue that regulates autophosphorylation of PrkG.
Substrate Phosphorylation Mechanisms of PrkD and PrkG-An understanding of the substrate phosphorylation mechanisms of these enzymes was necessary for further investigation of their nature. Phosphorylation of the nonspecific kinase substrate MyBP demonstrated that PrkD and PrkG are capable of substrate phosphorylation (Figs. 5A and 6A). Based on the sequence homology with Mycobacterium tuberculosis PknB and B. anthracis PrkC, the invariant Lys residues in subdomain II, essential for catalytic activity, were identified as Lys 53 of PrkD and Lys 57 of PrkG (29, 42) (supplemental Fig. S2). Mutations of PrkD K53M and PrkG K57M resulted in complete loss of autokinase activity and phosphorylation on MyBP (Figs. 5A and 6A). MyBP was found to be advantageous as a common substrate and was further employed to study the role of PrkD-and PrkG-mediated phosphorylation of substrates.
PAA analysis of PrkD-dependent phosphorylation on MyBP showed that only Ser and Thr residues were phosphorylated, but no signal was observed that corresponded to Tyr(P) (Fig.  5B). This feature strikingly resembles the DYRK class. To check the effect of PrkD mutations on substrate phosphorylation activity of PrkD, we compared the phosphorylation of MyBP substrate. We found that PrkD Ser 162 , Thr 181 , and Tyr 182 residues were required for efficient substrate phosphorylation, with Tyr 182 having the strongest effect between the three residues (Fig. 5, C and D, and supplemental Fig. S6, D and E).
To understand the role of Tyr phosphorylation in PrkG, the status of substrate phosphorylation was studied. PAA analysis demonstrated that MyBP was phosphorylated on Ser, Thr, and Tyr residues by PrkG (Fig. 6B). The substrate phosphorylation  AUGUST 3, 2012 • VOLUME 287 • NUMBER 32

JOURNAL OF BIOLOGICAL CHEMISTRY 26755
ability of the PrkG phosphorylation site mutants was also assessed, and interestingly, the PrkG T245A and PrkG Y349F mutants had the greatest impairment in phosphorylation of MyBP ( Fig. 6C and supplemental Fig. S6F). The role of Thr 245 and Tyr 349 was further verified by mutating multiple phosphorylation sites in PrkG in combination (data not shown). This  shows that the mechanism of autophosphorylation and the substrate phosphorylation in PrkG is distinct from that of PrkD and involves Thr residues in addition to Tyr residues. Therefore, PrkG is the first identified dual specificity kinase of the Bacillus species that can phosphorylate itself and its substrate on Ser, Thr, and Tyr residues.
Inhibition of PrkD and PrkG by Specific Inhibitors and the Effect on Cell Growth-Understanding the regulation of PrkD and PrkG by specific inhibitors can help in the development of novel drugs, which can affect the growth of B. anthracis by interfering with physiological function. We used several kinase inhibitors for these assays on the basis of their potency against specific eukaryotic STPK classes. Staurosporine, which acts against PKC (55), KN93 against Ca ϩ2 /CaM kinase II (56), PKR inhibitor against RNA-dependent protein kinases (57), and genistein (5,7-dihydroxy-3-(4-hydroxyphenyl) chromen-4one) against Tyr protein kinases (58) were used for these assays. Staurosporine only inhibited PrkD, and KN93 showed slight inhibition of PrkG (data not shown). Both kinases were active in the presence of the PKR inhibitor (data not shown). Under the given conditions, only genistein inhibited the kinase activity of both kinases (Fig. 7, A and B, and supplemental Fig. S7, A and  B). Genistein is known to inhibit only Tyr kinases (58), which supports our previous results showing that PrkD and PrkG are regulated by Tyr phosphorylation.
To analyze the functional relevance of PrkD and PrkG inhibition by genistein and its effect on B. anthracis, we performed growth curve assays in the presence of genistein. The growth of B. anthracis cultures was inhibited by genistein at 20 and 100 M concentrations (Fig. 7C). Because McsB is probably the only known Tyr kinase reported in B. anthracis, the observed effect could be due to concerted inhibition of McsB together with PrkD and PrkG. We subsequently used the B. anthracis strain, which lacks McsB (Bas⌬mcsB) and found similar inhibitory effects by genistein on the growth of this mutant strain (supplemental Fig. S7C).
Role of PrpC (BA-Stp1) in Dephosphorylation of PrkD and PrkG-All cells maintain a check on protein kinases by expressing regulators, such as protein phosphatases. PrpC and its homologs have been shown to dephosphorylate and consequently regulate the activity of their cognate kinase PrkC (BA-Stk1) in both B. subtilis and B. anthracis (29,59). Because autophosphorylation is necessary for kinase activity, PrpC-dependent dephosphorylation of the kinase can hinder downstream signaling. To determine whether PrkD and PrkG can also be inactivated by PrpC, autophosphorylated kinases were incubated with PrpC in a time-dependent manner. PrkC, a known substrate of PrpC, was used as positive control (data not shown). Over the experimental time period, PrkG was completely dephosphorylated by PrpC, whereas the dephosphorylation of PrkD remained incomplete ( Fig. 7D and supplemental  Fig. S8, A and B). The dephosphorylation of Tyr-phosphorylated kinases indicates that PrpC, which is a Ser/Thr phosphatase, may also possess Tyr phosphatase activity. To explore this hypothesis, peptides phosphorylated at Ser, Thr, and Tyr residues were incubated individually with both His 6 -tagged and GST-tagged PrpC. To our surprise, both His-tagged PrpC and GST-tagged PrpC dephosphorylated Tyr(P) peptides in addition to Ser(P) and Thr(P) peptides (supplemental Fig. S8C). A comparison between these phosphopeptides indicated that PrpC displays a preference for Ser(P) residues but is also able to dephosphorylate Thr(P) and Tyr(P) residues.
Identification of Pyruvate Kinase as the PrkD Substrate-To prove the DYRK-like activity of PrkD, it was necessary to show that PrkD phosphorylates its substrates on Ser/Thr and not on Tyr residue(s). To identify the substrates, we searched the homologs of B. subtilis phosphoproteins in B. anthracis (suppplemental Table S3) and chose Bas4492 (BasPyk) among the predicted phosphoproteins. BasPyk is involved in the last and critical step of glycolysis, which could affect glucose metabolism. Additionally, the homolog of BasPyk in M. tuberculosis, MtbPykA, was previously shown to be phosphorylated by PknJ (60). Therefore, we incubated recombinant BasPyk with PrkD in an in vitro kinase assay and found that it was phosphorylated (Fig. 8A). Importantly, no phosphorylation was observed on BasPyk by PrkD K53M . The specificity of this observation was further shown by the lack of phosphorylation of BasPyk by PrkC and PrkG in an in vitro kinase assay (Fig. 8B). The Ser(P) and Thr(P) sites of PrkD were identified on BasPyk by PAA analysis (Fig. 8C). Mass spectrometry analysis revealed that BasPyk phosphorylation occurs on one Thr (Thr 9 ) and two Ser residues (Ser 105 and Ser 401 ) (Fig. 8D). We also identified multiple additional substrates (Spo0M (Bas2153) and Bas2056) of PrkD in vitro and verified that phosphorylation did not occur on Tyr residues by immunoblotting with anti-Tyr(P) antibodies (data not shown).
The importance of PrkD residues, including Tyr 182 , in the substrate phosphorylation mechanism was also verified on BasPyk. Loss of Ser 162 and Tyr 182 in PrkD resulted in a significant loss in BasPyk phosphorylation, which reaffirmed the role of Tyr 182 in the kinase function (Fig. 8, E and F, and supplemental Fig. S9, A and B).
We could not identify any substrate phosphorylated by PrkG among multiple probable substrates (Ef-Tu, Ef-G, SsbA, Bas4487, and Bas1176) using the same strategy. PknG of Corynebacterium glutamicum is known to regulate glutamine metabolism by phosphorylating OdhI (61). Initial quantitative expression profiling data indicated that PrkG expression decreases in the presence of high concentrations of glutamine in the medium. However, experiments attempting to validate the OdhI homolog (Bas1177) as an in vitro substrate of PrkG failed. Therefore, additional focused genetic studies on PrkG will dissect its role in B. anthracis physiology.
PrkD-dependent Phosphorylation of BasPyk in Surrogate Host E. coli and Effect of Phosphorylation on Its Activity-PrkD-dependent phosphorylation on BasPyk was verified in surrogate host E. coli by co-expressing PrkD and BasPyk. The BasPyk was expressed in the presence of either PrkD or PrkD K53M . Phosphorylation status of purified forms was further verified by ProQ Diamond staining, and protein amounts were validated by SYPRO Ruby protein gel stain. PrkD phosphorylated BasPyk (Pyk-P), whereas its inactive mutant, PrkD K53M , was unable to phosphorylate BasPyk (Pyk-UP) when expressed in E. coli (Fig. 9A). The activities of Pyk-P and Pyk-UP were further compared. Both of the proteins showed similar K m values, but Pyk-UP exhibited higher specific activity (ϳ25%) as compared with Pyk-P ( Table 2, Fig. 9B, and supplemental Fig. S9C).

DISCUSSION
In this report, we have identified two dual specificity kinases, PrkD and PrkG, which are encoded in the genome of B. anthracis. Independent of its "select agent" status (62), B. anthracis has been projected as an outstanding model organism for studies of microbial ecology, evolution, cell development, and hostpathogen interactions (63)(64)(65). Based on the findings of this study, this medically important system can be used as a model to study signal transduction. The advantage of B. anthracis as a model organism is the presence of dual specificity kinases and a phosphatase together with the availability of expression analysis systems.
Quantitative real-time PCR analyses of PrkD and PrkG revealed the temporal expression profile of these kinases in various growth phases compared with PrkC. Both PrkD and PrkG are expressed in all phases of growth, indicating their possible role in the overall physiology of B. anthracis. The PrkD and PrkG expression levels were even higher than PrkC in the stationary phase. The expression of the three kinases at different phases alludes to the possible cross-talk between these two kinases and PrkC. Although cross-talk between STPKs has not been explored in detail, two co-expressing STPKs of M. tuberculosis, PknA and PknB, have been shown to undergo interdependent phosphorylation (66).
Biochemical characterization revealed the optimum conditions required for activation of PrkD and PrkG. Importantly, Mn 2ϩ is required for their activation, which is also thought to be a preferred cofactor for Tyr kinases and dual specificity kinases (67,68). To understand the regulation of these kinases, the sites of autophosphorylation were identified. Tyr phosphorylation of the two kinases was first observed by PAA analysis and immunoblotting, which was later verified by mass spectrometry. In PrkD, all of the phosphorylation sites were within the kinase domain except for Tyr 18 . The four regulatory phosphorylation sites were found in or adjacent to the activation loop, which includes Tyr 182 . Phosphorylation of one or more residues in the activation loop permits the kinase to refold and position for substrate binding (53). Electrostatic interactions among phosphorylated residues in the loop govern these conformational changes (53). The contribution of Ser 162 , Thr 180 , Thr 181 , and Tyr 182 was evaluated to elucidate the PrkD autophosphorylation mechanism. Whereas the loss of Thr 180 led to enhanced phosphorylation, the loss of Ser 162 , Thr 181 and Tyr 182 resulted in reduced autophosphorylation. Between PrkD-Thr 181 , PrkD-Ser 162 , and PrkD-Tyr 182 , Tyr 182 was found to be central to PrkD activation and overall kinase activity. Interestingly, kinase activity of another DSPK, PknD of C. pneumoniae, is also dependent on its activation loop residues (25). The rationale of designating Bas2152 as PrkD was its similarity to PknD, both being RD kinases.
The non-RD kinases of both pathogen and host have been implicated in pathogenesis, although RD kinases statistically outnumber them (69,70). One such example is PknG of M. tuberculosis, which is also a non-RD kinase with C-terminal TPR motifs (71) and resembles the architecture of PrkG homologs from the B. cereus group. Based on these previous findings, we named Bas2037 as PrkG in B. anthracis. PknG has been shown to be phosphorylated on redox-sensing Trx (thioredoxin) motifs (71). These motifs are absent in PrkG, indicating that it must have some alternate way of activation. The only other characterized bacterial non-RD kinase is PknG of C. glutamicum, which is very similar to its M. tuberculosis homolog. In PrkG activation, multiple sites are involved, and the loss of a single site did not yield a complete loss of kinase activity. The loss of Ser 196 , Thr 245 , Tyr 262 , and Tyr 349 had a negative impact,  AUGUST 3, 2012 • VOLUME 287 • NUMBER 32 whereas loss of Ser 108 had a positive effect on kinase activity of PrkG. Our results also suggest that PrkG adopts a unique activation mechanism because Tyr 349 is located outside of the kinase domain. PrkG is the first bacterial non-RD kinase that has been shown to possess dual specificity and is a bona fide DSPK.

Dual Specificity Protein Kinases of B. anthracis
The importance of phosphorylated residues for PrkD and PrkG kinase activity was confirmed by their effect on the phosphorylation of MyBP and BasPyk substrates. PrkD phosphorylated both proteins only on Ser and Thr residues, whereas PrkG phosphorylated MyBP on Ser, Thr, and Tyr residues as identified by the PAA analysis. Phosphorylation of BasPyk was further verified by mass spectrometry of the substrate. In PrkD, three residues were found to be important for substrate phosphorylation. The first, Thr 181 , was found to be important but not necessary for autophosphorylation but was found to be necessary for substrate phosphorylation. The other two, more critical, residues in PrkD activation were Ser 162 and Tyr 182 , the loss of which is insurmountable for autophosphorylation and substrate phosphorylation.
As previously discussed, PrkD homologs exist in all Bacillus species, but Tyr 182 is absent in all species except for the B. cereus group. The Tyr phosphorylation activity of PrkD was restricted to autophosphorylation. On the basis of these results, we report PrkD as a protein kinase similar to the DYRK class of kinases, which is an undiscovered enzyme class of the bacterial world. In eukaryotes, the DYRK class of kinases is known to regulate critical cellular functions, such as controlling the cell cycle, cytokinesis, cell differentiation, and brain development (34,35,(72)(73)(74). Eukaryotic DYRKs are known to target the RPX(S/T)P motif with a preference for Pro at the Pϩ1 position and Arg at the PϪ3 position (75). In PrkD, Arg 15 , Arg 178 , and Pro 246 may assist in the phosphorylation of Thr 18 , Thr 181 , and Thr 245 , respectively. Importantly, these residues are also conserved in B. cereus group homologs and absent in the B. subtilis homolog (supplemental Fig. S10). Additionally, the absence of characteristic DYRK motifs, such as the DYRK homology box, N-terminal autophosphorylation accessory region, and motif rich in Pro, Glu, Ser, and Thr residues (PEST) in PrkD, indicates that although they behave similarly, their regulatory mechanism may not be identical. Although we have observed profound similarities between PrkD and the DYRK class of kinases, further structural evidence is required to identify all variations of this enzyme. PrkG activation seems to be quite complex and is dependent on multiple residues. The two critical residues Tyr 349 and Thr 245 are located so distant from each other that further structural evidence is required to fully understand the activation and substrate binding mechanisms.
The present study also provides evidence of the dual specific nature of PrpC, which is the only Ser/Thr protein phosphatase in B. anthracis. In bacteria, Ser/Thr phosphatases are known to inactivate their cognate kinases, and thus their role is directly proportional to the importance of the kinase with which they are involved. PrpC was previously shown to be important for B. anthracis and B. subtilis as a partner of PrkC (29,59). Our results demonstrate that PrpC can also dephosphorylate the dual specificity protein kinases and may have a broader role beyond PrkC in B. anthracis. Interestingly, the Ser/Thr phosphatase PrpZ from Salmonella enterica also showed a similar dual specific nature with Tyr(P) peptides, but its protein substrates remain unknown (76).
The inhibition of PrkD and PrkG by genistein was used to study the physiological significance of these enzymes. B. anthracis vegetative cell growth was inhibited in the presence of 20 M genistein, which became more significant at 100 M. Similar effects on B. anthracis growth were also reported for genistein by Hong et al. (77). Furthermore, identification of BasPyk as a substrate of PrkD provides initial insight of the role of this kinase in glucose metabolism.
Identification of BasPyk as a substrate of PrkD in the in vitro conditions as well as in the surrogate host E. coli provides initial insights into the role of this kinase. The phosphorylated form of BasPyk was purified by co-expressing with the kinase. Compared with its unphosphorylated form, the phosphorylated BasPyk showed a reduced rate of product formation (V max ) and specific activity. The decrease in BasPyk specific activity may influence the phosphoenolpyruvate catabolism. In addition, pyruvate being an important metabolite, any variation in its concentration may affect several metabolic pathways.  In conclusion, we found that PrkD is conserved throughout the Bacillus species, but only the homologs present in the B. cereus group have a conserved Tyr residue in the activation loop. PrkG is present only in the B. cereus group and is expressed with a kinase domain lacking the TPR region at the C terminus in most of the B. anthracis strains sequenced so far. We hypothesize that specialization of B. anthracis as a warm blooded animal pathogen may have reduced the range of environmental stimuli to which it was exposed. This may have led to the loss of some Tyr kinases, which ultimately resulted in the evolution of dual specificity enzymes.