Forkhead-associated Domain-containing Protein Rv0019c and Polyketide-associated Protein PapA5, from Substrates of Serine/Threonine Protein Kinase PknB to Interacting Proteins of Mycobacterium tuberculosis*

Mycobacterium tuberculosis profoundly exploits protein phosphorylation events carried out by serine/threonine protein kinases (STPKs) for its survival and pathogenicity. Forkhead-associated domains (FHA), the phosphorylation-responsive modules, have emerged as prominent players in STPK mediated signaling. In this study, we demonstrate the association of the previously uncharacterized FHA domain-containing protein Rv0019c with cognate STPK PknB. The consequent phosphorylation of Rv0019c is shown to be dependent on the conserved residues in the Rv0019c FHA domain and activation loop of PknB. Furthermore, by creating deletion mutants we identify Thr36 as the primary phosphorylation site in Rv0019c. During purification of Rv0019c from Escherichia coli, the E. coli protein chloramphenicol acetyltransferase (CAT) specifically and reproducibly copurifies with Rv0019c in a FHA domain-dependent manner. On the basis of structural similarity of E. coli CAT with M. tuberculosis PapA5, a protein involved in phthiocerol dimycocerosate biosynthesis, PapA5 is identified as an interaction partner of Rv0019c. The interaction studies on PapA5, purified as an unphosphorylated protein from E. coli, with Rv0019c deletion mutants reveal that the residues N-terminal to the functional FHA domain of Rv0019c are critical for formation of the Rv0019c-PapA5 complex and thus constitute a previously unidentified phosphoindependent binding motif. Finally, PapA5 is shown to be phosphorylated on threonine residue(s) by PknB, whereas serine/threonine phosphatase Mstp completely reverses the phosphorylation. Thus, our data provides initial clues for a possible regulation of PapA5 and hence the phthiocerol dimycocerosate biosynthesis by PknB, either by direct phosphorylation of PapA5 or indirectly through Rv0019c.

Genes encoding serine/threonine protein kinases (STPKs) 4 once thought to be unique to eukaryotes are now an established integral component of many prokaryotic genomes (1)(2)(3)(4)(5). The coordinated action of STPKs with a cohort of protein modules leads to the formation of numerous signaling cascades essential for cell survival. Various protein modules that recognize phosphopeptide binding sites typically exposed after kinase autophosphorylation have been the subject of active research in the past few years (6). In eukaryotes such an association is best exemplified by the binding of proteins containing Src homology 2 domains with phosphotyrosyl residues of activated receptor tyrosine kinases (7). In parallel with tyrosine kinase signaling, the eukaryotic serine/threonine protein kinases associate with various phosphopeptide binding modules, namely, 14-3-3, WW, Polo-box, and FHA domains (6). Of these FHA domains, the phosphothreonine binding modules are uniquely found in prokaryotes to higher eukaryotes (8).
FHA domains were first identified as a conserved region of ϳ75 amino acids through sequence analysis in a subset of forkhead-type transcription factors (9). Subsequent biochemical and structural studies revealed that the functional FHA domain spans ϳ100 -150 amino acids and flanks beyond the FHA core homology region were required to form a stable 11-stranded ␤ sandwich structure (10 -12). FHA domains share very low sequence homology, with only seven conserved residues in more than 65% of known FHA domains (13). The conserved residues are interspersed in loops connecting ␤ strands and are clustered in close proximity to form an interaction with the phosphothreonine residue and the peptide backbone of the interacting protein (14). Initially identified in 20 proteins from yeast, mammals, and bacteria (9), FHA domains are now known to be a part of 2554 unrelated proteins from bacteria to humans. In eukaryotes, FHA domain-containing proteins play crucial roles in diverse cellular processes such as cell cycle regulation, DNA damage response, protein degradation, vesicular transport, and signal transduction (15).
Although a wealth of information is available on the function of FHA domains in eukaryotes, their role in prokaryotic biology is only beginning to unravel (16). The information acquired in the past few years point toward their roles in secretion pathways, transcription, cellular metabolism, and signal transduction (17)(18)(19)(20)(21). Among prokaryotes, Mycobacterium tuberculosis has emerged as an attractive model system for understanding FHA domain-mediated signaling for two reasons: 1) M. tuberculosis encodes for 11 eukaryotic like STPKs (PknA to PknL, except PknC) and seven FHA domains in six proteins (22), thus offering an experimentally malleable system to understand molecular themes common to prokaryotes and eukaryotes. 2) Recent studies show that STPKs play an important role in M. tuberculosis survival and pathogenicity (23) and FHA domains are known to play a key role in STPK-mediated signaling events.
Biochemical studies carried out on three FHA domain-containing proteins of M. tuberculosis have provided unambiguous evidence for their involvement in STPK-mediated signal transduction pathways crucial for M. tuberculosis virulence. EmbR, a SARP family transcription factor, was shown to act downstream of PknH in a FHA domain-dependent manner to upregulate embCAB arabinosyltransferases, resulting in a higher LAM/LM ratio, an important determinant of mycobacterial virulence (24 -26). Rv1747 (ABC transporter), a component of PknF-mediated signal transduction and the only protein comprised of two FHA domains, was shown to be important for M. tuberculosis growth in macrophages and mice (27). Recent work on GarA, a small protein mainly occupied by the FHA domain, has provided an indirect link to the regulation of the mycobacterial metabolism by STPKs PknB and PknG (28,29). The phosphoindependent interaction of GarA with metabolic enzymes was shown to be blocked by an intramolecular inhibition mechanism triggered by PknG/PknB-mediated GarA phosphorylation (29).
The phosphoindependent interactions of FHA domains, long believed to interact solely with phosphoproteins, is an emerging concept with few examples (29,30). In this study we have attempted to understand the role of the previously uncharacterized FHA domain-containing protein Rv0019c and have shown its association with phosphorylated and unphosphorylated proteins. We have also provided vital clues for a possible involvement of Rv0019c in mycobacterial virulence.

EXPERIMENTAL PROCEDURES
Bacterial Strains and Growth Conditions-Escherichia coli strain DH5␣ (Novagen) was used for cloning, and BL21(DE3) CodonPlus (Stratagene) or 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 medium supplemented with 100 g/ml of ampicillin and/or 25 g/ml of chloramphenicol, when needed.
Cloning, Expression, and Purification of Rv0019c, Its Mutants, and GarA-The rv0019c gene, its fragments coding for 19c⌬T (Rv0019c deletion mutant devoid of N-terminal 1-30 residues), 19cFHA (Rv0019c deletion mutant comprising of FHA domain from residues 52-155), and 19cCHR (Rv0019c deletion mutant encompassing FHA domain core homology region from residues 80 to 155) and the gene encoding for garA were PCR amplified using M. tuberculosis genomic DNA as a template and the forward and reverse primers (Table 1), containing BamHI and XhoI restriction sites, respectively. The amplified products were digested with restriction enzymes, BamHI and XhoI, and ligated into vector pGEX-5X-3 (GE Healthcare), previously digested with the same enzymes to yield the plasmids listed in Table 2. The pGEX-rv0019c-⌬T construct was utilized for generating R87A, S101A, T36A, and T50A mutations. pGEX-rv0019c-⌬T was subjected to site-directed mutagenesis using the QuikChange XL Site-directed Mutagenesis kit (Stratagene) and primer pairs carrying the desired mutations ( Table  1). The plasmids thus generated are listed in Table 2. The integrity of all constructs was confirmed by DNA sequencing. E. coli BL21-CodonPlus or BL21(DE3) cells were transformed with pGEX-5X-3 vector derivatives expressing 19c⌬T, 19cFHA, 19cCHR, R87A, S101A, T36A, T50A, and GarA proteins. For overexpression, recombinant E. coli strains harboring the pGEX-5X-3 derivatives were used to inoculate 1000 ml of LB medium supplemented with chloramphenicol and/or ampicillin followed by incubation at 37°C with shaking (220 rpm) until A 600 reached 0.8. Isopropyl 1-thio-␤-D-galactopyranoside was added to a final concentration of 1 mM and growth was continued for an additional 2 h at 37°C. Cells were harvested and stored at Ϫ80°C. Pellets were thawed on ice and resuspended in cell lysis buffer A (50 mM Tris-Cl, pH 8.0, 300 mM NaCl, 1 mM dithiothreitol, 1 mM EDTA, 1ϫ protease inhibitor mixture (Roche Applied Science)) and lysed by sonication. The cell lysates were centrifuged at 23,430 ϫ g at 4°C for 20 min. The supernatant containing recombinant proteins was collected and incubated with glutathione-Sepharose 4B affinity resin (GE a Restriction sites are underlined and specified in the parentheses after the sequence. Mutagenized codons are shown as bold. b F, forward; R, reverse.

Rv0019c Interacts with PapA5
Sandwich ELISA-ELISA was done using the method described by Lu et al. (32) with minor modifications. Briefly, the GST-tagged proteins were dissolved in coating buffer (0.1 M NaHCO 3 , pH 9.2) at a concentration of 10 g/ml and adsorbed (50 l/well) on the surface of a 96-well ELISA plate (Maxisorb, Nunc) overnight at 4°C. After rinsing the wells five times with wash buffer (phosphate-buffered saline, pH 7.4, 0.05% Tween 20) the reactive sites were blocked with blocking buffer (2% bovine serum albumin in wash buffer) for 2 h at room temperature. The adsorbed proteins were challenged with varying concentrations of soluble His 6 -tagged proteins (50 l/well) dissolved in blocking buffer for 1 h at room temperature. After five washes with wash buffer, the wells were treated with penta-Hishorseradish peroxidase-conjugated monoclonal antibody (Qiagen) at 1:10,000-fold dilution for 1 h at room temperature. Followed by five washes, the chromogenic substrate OPD (0.4 mg/ml o-phenylenediamine dihydrochloride in 0.1 M phosphate/citrate buffer, pH 5) and H 2 O 2 were used to visualize the interaction. After addition of stop solution (2.5 M H 2 SO 4 ) the absorbance was read at 490 nm.
In Vitro Kinase Assay-In vitro kinase assays were performed with 1-2 g of PknB, PknBK40M, PknB-(1-331), or PknB-(1-331) threonine mutants (Thr 171 , Thr 173 , Thr 171/173 , Thr 294 , and Thr 309 ) added separately in 15 l of kinase buffer (20 mM PIPES, pH 7.2, 5 mM MgCl 2 , 5 mM MnCl 2 ) with 2-4 g of the desired protein (19c⌬T,19cFHA,19cCHR, R87A, S101A, T36A, T50A, GST, PapA5). Reactions were started with the addition of 10 Ci of [␥-32 P]ATP (BRIT, Hyderabad, India) and incubated at 25°C for 20 min. The reactions were terminated by the addition of 5ϫ SDS sample buffer and a subsequent boiling for 5 min. Samples were resolved on 10% SDS-PAGE. Gels were dried and visualized using FLA2000 PhosphorImager (Fuji). For visualization of the phosphorylation signal on cleaved proteins, removal of recombinant tags was achieved by addition of proteases (Factor Xa (Novagen) for GSTtagged proteins and TEV for His 6tagged proteins) after the kinase reaction had proceeded for 20 min followed by an additional incubation for 2 h at 20°C. The reactions were stopped using SDS buffer and resolved on 15% SDS-PAGE. Likewise, for dephosphorylation of phosphorylated 19c⌬T and PapA5, reactions were incubated with MstP cat (500 ng) for an additional 0, 1, and 30 min at 25°C, stopped, and resolved on 10% SDS-PAGE. The signals were visualized by autoradiography. Phosphoamino Acid Analysis-19c⌬T and PapA5 were phosphorylated by PknB as mentioned above, separated by SDS-PAGE, and electroblotted onto Immobilon PVDF membrane (Millipore). The phosphorylated 19c⌬T and PapA5 proteins were detected by autoradiography and the bands corresponding to ␥-32 P-labeled substrates were excised and hydrolyzed in 6 M HCl for 1 h at 110°C. The hydrolyzed samples containing liberated phosphoamino acids were lyophilized and redissolved for phosphoamino acid analysis by two-dimensional thin layer electrophoresis as described (33).
Immunoblotting-To determine the phosphorylation status of recombinant CAT and PapA5 by Western blot analysis, the freshly purified proteins were resolved by SDS-PAGE along with positive (PknB for anti-phosphothreonine) and negative (GST) controls and transferred onto nitrocellulose membrane (Bio-Rad). After overnight blocking of membrane with 3% bovine serum albumin in PBST (phosphate-buffered saline, pH 7.2, 0.1% Tween 20), the blot was incubated for 1 h at room temperature with antibodies directed against phosphoserine or phosphothreonine (Invitrogen) dissolved in PBST at 1:10,000 dilution. Followed by five washes the blot was incubated for 1 h at room temperature with anti-rabbit horseradish peroxidase polyclonal antibody (Sigma) dissolved in PBST (1:10,000 dilution). After five washes the blots were developed using ECL Plus kit (GE Healthcare) according to the manufacturer's instructions.

Domain Architecture and Genomic Organization of Rv0019c-
Rv0019c is a small 155-amino acid protein encompassing a predicted N-terminal transmembrane domain/signal peptide and a C-terminal FHA domain occupying the majority of the protein (Fig. 1A). Amino acid sequence alignment of Rv0019c with FHA domains of mycobacterial proteins with known x-ray or NMR structure (34,29) shows the conservation of six of the seven hallmark residues for FHA domains in Rv0019c (Fig. 1B). Of the six FHA domain-containing proteins encoded in the M. tuberculosis genome, four have been characterized and the conserved residues in the FHA domains have been shown to play a crucial role in their interaction with autophosphorylated STPKs (24,35,36). In the M. tuberculosis genome, Rv0019c resides within a cluster of genes encoding for essential STPKs PknA and PknB, morphogenic proteins PbpA and RodA, and the only identified Ser/Thr phosphatase MstP. The orthologs of Rv0019c and PknB show strong conservation and display similar genomic organization in various members of actinomycetales (supplemental Fig. S1). The ubiquitous presence of Rv0019c orthologs along with those of PknB hints at a role of M. tuberculosis Rv0019c in PknB-mediated signaling events through an inevitable interaction between the two proteins mediated by the Rv0019c FHA domain.
In Vitro Interaction of Rv0019c with PknB-To determine whether Rv0019c develops a direct physical contact with PknB, interaction studies were carried out using sandwich ELISA as described under "Experimental Procedures." Recombinant N-terminal GST-tagged or His 6 -tagged Rv0019c was invariably localized to inclusion bodies during purification from E. coli (data not shown). A deletion mutant lacking the predicted N-terminal transmembrane region/signal peptide (1-30 amino acids) was generated to facilitate its purification from soluble fraction. As shown in Fig. 2A, robust interaction was observed between the recombinant GST-tagged deletion product of Rv0019c (19c⌬T) and purified autophosphorylated His 6tagged PknB (0 -50 ng). Notably, PknB is capable of catalyzing autophosphorylation during its expression in E. coli (37) and hence the autophosphorylated purified kinase was used directly for interaction studies. GST alone acted as a negative control, whereas GarA, a characterized FHA domain containing the interaction partner of PknB was included as a positive control ( Fig. 2A). As mentioned above, the FHA domain-containing proteins primarily form contact with their phosphorylated targets through conserved residues in their FHA domains. To ensure that the observed 19c⌬T-PknB interaction was mediated by the FHA domain of 19c⌬T, the conserved residues in its FHA domain core homology region (Arg 87 and Ser 101 ) were mutagenized to alanine by site-directed mutagenesis. When compared with 19c⌬T, FHA domain mutants 19c⌬TR87A (R87A) and 19c⌬TS101A (S101A) displayed a significant reduction in their capacity to interact with PknB (Fig. 2B). Alternatively, the PknB kinase-dead mutant (PknBK40M), deficient of autophosphorylation, also exhibited lower binding affinity with 19c⌬T (Fig. 2C). Because PknBK40M was purified as an unphosphorylated protein from E. coli (Fig. 5E), we expected a total loss of interaction with 19c⌬T, the implication of the basal contact thus observed is discussed in later sections.
In Vitro Phosphorylation of Rv0019c by PknB-An interaction between a kinase and a FHA domain-containing protein does not necessarily lead to the phosphorylation of the latter. Recently Molle et al. (38) demonstrated the lack of PknH-mediated phosphorylation of EmbR2, an FHA domain containing EmbR homolog, albeit an efficient interaction of EmbR2 with phosphorylated peptides derived from PknH. Therefore, the ability of proximally located PknB to act as a "true" Rv0019c kinase was tested. When incubated in the presence of PknB and [␥-32 P]ATP, phosphotransfer was evident on 19c⌬T (Fig. 3A). No incorporation of radiolabel was observed on 19c⌬T when incubated alone or in the presence of PknBK40M thereby validating a PknB-specific phosphorylation of 19c⌬T (Fig. 3A). To ensure that the observed phosphorylation is localized on the native protein and not on the recombinant GST tag, the fusion tag was removed with Factor Xa protease and phosphorylation was confirmed to be specifically localized on the released 19c⌬T protein (data not shown). Furthermore, in agreement with the interaction studies on FHA domain mutants (R87A and S101A) with PknB, both mutants were labeled to a much lesser extent as compared with 19c⌬T under similar reaction conditions (Fig. 3B). As mentioned previously, Rv0019c is encoded within a cluster of genes comprising Ser/Thr phosphatase MstP. Therefore, it was pertinent to believe that 19c⌬T would be targeted by MstP for dephosphorylation. As expected the phosphorylation of 19c⌬T was completely reversed by MstP (Fig. 3C).  DECEMBER 11, 2009 • VOLUME 284 • NUMBER 50

JOURNAL OF BIOLOGICAL CHEMISTRY 34727
Because the 19c⌬T-PknB interaction and the consequent 19c⌬T phosphorylation was primarily dependent on the FHA domain of 19c⌬T and FHA domains are known phosphothreonine interaction modules, we next mutagenized the four phosphorylatable threonine residues (Thr 171 , Thr 173 , Thr 294 , and Thr 309 ) in PknB (39). Thr 171 and Thr 173 residues are situated in the activation loop of PknB, whereas Thr 294 and Thr 309 reside in the juxtamembrane region. For ease of purification, the threonine mutations were created in a smaller construct of PknB encompassing the catalytic domain and the linker juxtamembrane region (1-331 amino acids) yielding a soluble protein (31). When compared with wild type PknB-(1-331), the T171A and T173A mutants displayed weak activity toward 19c⌬T, whereas the phosphotransfer on 19c⌬T by T294A and T309A mutants was comparable with that of wild type PknB-(1-331) (Fig. 3D). These results clearly indicated at the necessity of threonine residues (Thr 171 and Thr 173 ) in the activation loop of PknB for effective phosphorylation of 19c⌬T. The activation loop threonine residues were previously shown to be critical for the phosphorylation of GarA by PknB (40), whereas similar residues in PknH played no role in its interaction with the FHA domain containing protein EmbR (34).
We next tried to analyze the nature of the amino acid labeled by PknB on 19c⌬T and found that 19c⌬T was exclusively phos-phorylated on threonine residue(s) following incubation with PknB in the presence of [␥-32 P]ATP (Fig.  3E). In conclusion, 19c⌬T phosphorylation on threonine residue(s) by PknB was found to be dependent on a fully functional activation loop of PknB and the FHA domain of 19c⌬T.
Identification of the Phosphorylation Site in 19c⌬T-Identification of phosphorylation sites in the substrate aids gained mechanistic insight into the complex signal transduction cascade operational within the cell. To determine the target site(s) of PknB in 19c⌬T, two N-terminal deletion mutants of 19c⌬T were created: one comprised of residue 52-155 constituting the predicted functional FHA domain (19cFHA) and the other encompassing residue 80 -155 constituting the FHA domain core homology region (19cCHR) (Fig. 4A). To determine whether the deletion mutants thus created retain their functional integrity, interaction studies were carried out with His 6 PknB. As shown in Fig. 4B, 19cFHA readily formed a complex with PknB, whereas, 19cCHR and PknB showed no significant physical association. The observation was in agreement with previous studies emphasizing at the role of residues beyond the FHA core homology region in maintaining the structural and functional integrity of the FHA domain (10 -12). 19cFHA was then subjected to the phosphorylation assay with PknB in the presence of [␥-32 P]ATP. Although PknB labeled 19c⌬T, no phosphotransfer was observed on 19cFHA (Fig. 4C). A faint signal on 19cFHA could only be observed after a prolonged exposure (data not shown). Because 19cFHA represented a functional unit with an intact capacity to interact with PknB (Fig. 4B), 19c⌬T-like radiolabeling was expected. The lack of phosphorylation of 19cFHA by PknB clearly pointed out the presence of a major phosphorylation site N-terminal to 19cFHA. Sequence analysis showed the presence of two threonine residues (Thr 36 and Thr 50 ) within the motif 31-51 as the putative target sites of PknB (Fig. 4D). Both threonine residues were substituted to alanine using site-directed mutagenesis. Because point mutations may lead to structural perturbations in a protein, the mutants were subjected to interaction studies with the kinase to ensure that they retained functional integrity. The interaction of T36A and T50A with PknB was comparable with that of the PknB-19c⌬T interaction (data not shown), hence, the mutants were subjected to phosphorylation assays with PknB. As shown in Fig. 4E, a substantial loss of phosphotransfer was observed on T36A with a minor phospholabeling observed only on prolonged exposure. Thus, threonine at position 36 was identified as the major phosphorylation site in 19c⌬T.
Identification of a 19c⌬T Interacting Protein from E. coli-By purification of 19c⌬T and its aforementioned deletion mutants from E. coli, we observed a reproducible and specific co-purification of E. coli proteins corresponding to 50 and 25 kDa with 19c⌬T and 19cFHA (Fig. 5, A and B). Interestingly, similar E. coli proteins failed to co-purify with the functionally inactive deletion mutant 19cCHR, FHA domain mutants R87A and S101A, and GST alone. Therefore it was tempting to speculate that the co-purified proteins were specific interaction partners of 19c⌬T and that co-purification was dependent on the functional FHA domain of 19c⌬T. The co-purified E. coli proteins corresponding to 25 and 50 kDa were identified by mass spectrometry as E. coli CAT. CAT is a bacterial enzyme that confers resistance to antibiotic chloramphenicol (41). Because CAT exists as a trimeric protein (41), the band corresponding to 50 kDa probably represented the SDS stable dimeric subunit and the one at 25 kDa was the dissociated monomer. To ascertain that the observed 19c⌬T FHA domain specific pulldown of CAT was mimicked in vitro, E. coli CAT was expressed and purified as a recombinant His 6 -tagged protein (CAT) and interaction studies were carried out with 19c⌬T and its mutants. As shown in Fig. 5C, CAT displayed a robust interaction with 19c⌬T and 19cFHA, whereas no interaction was evident between CAT and 19cCHR. Furthermore, in agreement with the copurification results, FHA domain mutants R87A and S101A failed to interact with CAT in vitro (Fig. 5D). Thus, like the PknB-19c⌬T interaction, the CAT-19c⌬T interaction was also found to be dependent on the functional FHA domain of 19c⌬T.
Because 19c⌬T and its mutants displayed a strikingly similar interaction pattern with PknB and CAT, we argued that CAT purified from E. coli could possibly be a phosphorylated protein. To test the hypothesis, recombinant CAT was analyzed for its phosphorylation status using anti-phosphothreonine/serine antibodies. As shown in Fig. 5E, CAT was found to be phosphorylated on both threonine and serine residues, thus providing a possible explanation for the observed 19c⌬T-CAT interaction. Because, E. coli CAT was found to be phosphorylated on threonine residue(s) and FHA domains are phosphothreonine binding modules, we next suspected that the co-purification of E. coli CAT with 19c⌬T could be due to the high affinity of a FHA domain for a phosphoprotein. To rule out the possibility of a promiscuous interaction, GST-GarA was purified under similar purification conditions and subjected to interaction studies with CAT. Interestingly, no interaction was observed between purified GarA and CAT (Fig. 5F), providing conclusive evidence for the specificity of the CAT-19c⌬T interaction.
19c⌬T Interacts with PapA5, the Structural Homolog of E. coli CAT in M. tuberculosis-The FHA domain-specific interaction of CAT with 19c⌬T and the tantalizing specificity prompted us to investigate for a sequence homolog of E. coli CAT in M. tuberculosis. To our surprise, no sequence homolog was identified in M. tuberculosis when the E. coli CAT sequence was used as a query to search the M. tuberculosis genome. However, we could find a report on PapA5, a polyketide-associated protein with acetyltransferase activity, sharing considerable structural similarity with E. coli CAT (42). M. tuberculosis PapA5 is involved in the synthesis of virulence enhancing lipids and is a suspected membrane protein (43,44). Although E. coli CAT and M. tuberculosis PapA5 share only 12% sequence identity, the regions of structural similarity shared between the two proteins stretch beyond the acetyltransferase domain and span 127 amino acid residues at the 3.9 Å root mean square deviation (42). To test the possibility of a PapA5-19c⌬T interaction, PapA5 was expressed and purified as a recombinant His 6 -tagged protein and allowed to interact with 19c⌬T. As shown in Fig. 6A, an interaction was apparent between immobilized 19c⌬T and PapA5 (0 -250 ng) in solution, whereas the functionally inactive deletion mutant 19cCHR and GST failed to form a complex with PapA5 (Fig. 6A). The specificity of interaction was further ascertained by incubating GarA with PapA5. Because GarA displayed no affinity for PapA5 (Fig. 6B), the polyketide-associated protein with a structural similarity with E. coli CAT was identified as a specific interaction partner of 19c⌬T.
PapA5-19c⌬T Interaction Is Not Dependent on the Functional FHA Domain of 19c⌬T-Although PapA5 displayed an affinity for 19c⌬T, the complex was formed at a significantly higher concentration of PapA5 (0 -250 ng) when compared with CAT (0 -50 ng) in CAT-19c⌬T interaction under similar reaction conditions. Because, PapA5 was purified as an unphosphorylated protein from E. coli (Fig. 5E), we reasoned that in the absence of phosphorylation and hence the phosphointeraction motif for the FHA domain of 19c⌬T, the 19c⌬T-PapA5 complex would be loosely held by hitherto unidentified interactions. To investigate the nature of these interactions and determine the role of the FHA domain, if any, in the phosphoindependent contact of 19c⌬T with PapA5, we utilized FHA domain mutants R87A and S101A. As shown in Fig. 7A, the 19c⌬T-PapA5 complex was unaffected by the S101A mutation, whereas association was apparently reduced when Arg 87 was mutagenized. The results were in stark contrast from those obtained from 19c⌬T-PknB and 19c⌬T-CAT interaction studies, where the mutation of Arg 87 and Ser 101 residues abrogated the complex formation (see Figs. 2B and 5D). These results prompted us to investigate the role of residues N-terminal of the functional FHA domain in 19c⌬T-PapA5 complex formation. To our surprise, the deletion mutant 19cFHA, encompassing the func-

Rv0019c Interacts with PapA5
tional FHA unit and devoid of residues 31-51, was unable to form a complex with PapA5 (Fig. 7B). The result was rather unexpected because 19cFHA was capable of forming a high affinity complex with PknB and CAT (see Figs. 4B and 5C). From the results obtained from interaction studies on PapA5 with 19c⌬T deletion mutants we concluded that in the absence of phosphorylation on the target protein (PapA5), 19c⌬T interacts primarily through residues upstream of the functional FHA domain , whereas the FHA domain acts cooperatively to hold the complex. The hypothesis was further supported by the observation that the previously observed phosphoindependent basal interaction of 19c⌬T with PknBK40M was substantially affected by deletion of residues from 31 to 51 (Fig. 7C). Additionally, as seen for the 19c⌬T-PapA5 interaction, the mutation of conserved residues in the FHA domain did not completely abrogate the interaction but had a cooperative role to play in PknBK40M-19c⌬T complex formation (Fig. 7D).
PapA5 Is a Substrate of PknB-As shown above, PapA5 and PknBK40M displayed a strikingly similar interaction pattern with 19c⌬T and its mutants. Also, E. coli CAT, the structural homolog of PapA5, was identified as a phosphorylated protein. Therefore, it was pertinent to believe that PapA5 could also be modified by a similar post-translational modification. To test the same, we carried out phosphorylation reactions where PapA5 was incubated with PknB in the presence of [␥-32 P]ATP. As speculated, PapA5 was efficiently phosphorylated by PknB (Fig. 8A) and no incorporation of radiolabel was observed when PapA5 was incubated alone or in the presence of PknBK40M, thereby validating a PknB-specific phosphorylation of PapA5 (Fig. 8A). The recombinant fusion tag was removed with TEV protease and the phospholabel was confirmed to be specifically localized on the released PapA5 protein (data not shown). Furthermore, the released PapA5 was found to be exclusively phosphorylated on threonine residue(s) (Fig. 8B) as analyzed by twodimensional thin layer electrophoresis. Because regulation of signaling proteins by reversible phosphorylation is a common theme in signal transduction, we next questioned whether PknB-mediated PapA5 phosphorylation could be reversed by Ser/Thr phosphatase MstP. As shown in Fig. 8C, MstP completely dephosphorylated PapA5 in a time-dependent manner, thereby reversing the kinase-mediated signaling. Thus, PapA5 was shown to be reversibly modified by the PknB-MstP kinase phosphatase pair.

DISCUSSION
In this report we show that the FHA domain-containing protein Rv0019c interacts with autophosphorylated cognate kinase PknB primarily through the conventional phosphopeptide binding region and is consequentially phosphorylated at its    (42). In the M. tuberculosis genome, the gene encoding for PapA5 is situated in a cluster of genes implicated in the biosynthesis of PDIMs, the virulence enhancing lipids (22). PapA5 catalyzes the final step in PDIM biosynthesis by mediating the direct transfer of mycocerosic acid synthase-bound mycocerosic acids onto the pthiocerols by forming direct physical contact with the mycocerosic acid synthase protein (43). Because no sequence homolog of E. coli CAT was identified in the M. tuberculosis genome, we investigated the possibility of 19c⌬T interaction with PapA5. Our results showed that PapA5 formed a complex with 19c⌬T although at much higher concentrations. Nonetheless, the 19c⌬T-PapA5 interaction was highly specific as the FHA domain-containing protein GarA failed to form any physical contact with PapA5. Surprisingly, PapA5 displayed a fairly unusual interaction pattern with 19c⌬T mutants. The mutation of Arg 87 and Ser 101 , shown to be detrimental for interaction of 19c⌬T with PknB and CAT had a marginal impact on the 19c⌬T-PapA5 interaction. Furthermore, the deletion of residues from 31 to 51 severely affected formation of the 19c⌬T-PapA5 complex. A similar mutant, 19cFHA, formed a high affinity complex with PknB and CAT. We argued that the variable pattern of PapA5 interaction with 19c⌬T mutants could possibly be due to the differential phosphorylation status of PapA5. Although PknB and CAT were purified as phosphorylated proteins, PapA5 was purified as an unphosphorylated protein from E. coli. Phosphoindependent interactions of the FHA domain-containing proteins have been previously reported for Chk2, the human homolog of S. cerevisiae Rad53 (30), and more recently for GarA (29). For Chk2, the phosphoindependent binding region located on the FHA domain was reported to be distinct from the conventional phosphopeptide binding motif, whereas for GarA the FHA surface forming contacts with the unphosphorylated proteins was shown to be similar to the Thr(P) binding surface. Furthermore, the minimal functional FHA domain of GarA-(55-149) was sufficient for its interaction with unphosphorylated target proteins and indeed caused more potent inhibition of target enzymes when compared with the full-length protein. However, the interaction studies on 19c⌬T with PapA5 seem to suggest that 19c⌬T forms phosphoindependent interactions primarily through the motif upstream of the functional FHA domain. The FHA domain plays a minimal role in phosphoindependent contacts of 19c⌬T. As seen for the 19c⌬T-PapA5 interaction, the phosphoindependent interaction of 19c⌬T with PknBK40M, an autophosphorylation deficient mutant of PknB, was also found to depend on residues 31-51, further substantiating our hypothesis. Thus we propose that in the presence of phosphorylation on the target protein, 19c⌬T interacts primarily through the critical residues in the FHA domain, whereas the upstream residues (residues 31-51) act cooperatively to stabilize the complex. In the absence of target protein phosphorylation the upstream residues primarily hold the complex because the interaction of these residues is expected to be independent of the phosphorylation status of the target protein.

Reversible Phosphorylation of PapA5 by the PknB-MstP
Kinase Phosphatase Pair-The results discussed above clearly indicated that PapA5 could be a plausible target of PknB, and rightly so PapA5 was shown to be phosphorylated by PknB. PapA5 was modified exclusively on threonine residue(s) and phosphorylation was completely reversed by the serine/threonine phosphatase MstP. In the report on the crystal structure of PapA5, Buglino et al. (42) proposed that the interaction of PapA5 with its substrates through one of its solvent-exposed channels would require conformational changes. Phosphorylation could be one such modification bringing about the desired conformational changes. Alternatively PapA5 phosphorylation might affect its interaction with other components of PDIM biosynthesis (43) or with Rv0019c. In future, PapA5 phosphorylation and its interaction with Rv0019c if proven in vivo would provide a valuable link to regulation of the PDIM biosynthesis by PknB, thus providing another example of the profound effect of STPK-mediated phosphorylation in M. tuberculosis virulence.