doi:10.1016/j.jmb.2005.05.049
J. Mol. Biol. (2005) 350, 953–963
Proteomic Identification of M. tuberculosis Protein Kinase Substrates: PknB Recruits GarA, a FHA Domain-containing Protein, Through Activation Loop-mediated Interactions A. Villarino1†, R. Duran2†, A. Wehenkel1, P. Fernandez1, P. England1 P. Brodin3, S. T. Cole3, U. Zimny-Arndt4, P. R. Jungblut4, C. Cerven˜ansky2 and P. M. Alzari1* 1
Unite´ de Biochimie Structurale (URA 2185 CNRS), Institut Pasteur, Paris, France 2
Laboratorio de Bioquı´mica Analı´tica, Instituto de Investigaciones Biolo´gicas Clemente Estable and Facultad de Ciencias, Montevideo Uruguay 3
Unite´ de Ge´ne´tique Mole´culaire Bacte´rienne Institut Pasteur, Paris, France 4
Max-Planck Institut for Infection Biology, Berlin Germany
Genes for functional Ser/Thr protein kinases (STPKs) are ubiquitous in prokaryotic genomes, but little is known about their physiological substrates and their actual involvement in bacterial signal transduction pathways. We report here the identification of GarA (Rv1827), a Forkheadassociated (FHA) domain-containing protein, as a putative physiological substrate of PknB, an essential Ser/Thr protein kinase from Mycobacterium tuberculosis. Using a global proteomic approach, GarA was found to be the best detectable substrate of the PknB catalytic domain in non-denatured whole-cell protein extracts from M. tuberculosis and the saprophyte Mycobacterium smegmatis. Enzymological and binding studies of the recombinant proteins demonstrate that docking interactions between the activation loop of PknB and the C-terminal FHA domain of GarA are required to enable efficient phosphorylation at a single N-terminal threonine residue, Thr22, of the substrate. The predicted amino acid sequence of the garA gene, including both the N-terminal phosphorylation motif and the FHA domain, is strongly conserved in mycobacteria and other related actinomycetes, suggesting a functional role of GarA in putative STPK-mediated signal transduction pathways. The ensuing model of PknB–GarA interactions suggests a substrate recruitment mechanism that might apply to other mycobacterial kinases bearing multiple phosphorylation sites in their activation loops. q 2005 Elsevier Ltd. All rights reserved.
*Corresponding author
Keywords: Ser/Thr protein kinase; GarA (Rv1827); Forkhead-associated domain; activation loop; substrate docking
Introduction Microbial genomics has had a strong impact on our perception of reversible phosphorylation mechanisms in bacteria. Genome-wide studies have now confirmed that Ser, Thr and Tyr protein kinases † A.V. & R.D. contributed equally to this work. Abbreviations used: STPK, Ser/Thr protein kinase; MBP, myelin basic protein; FHA, Forkhead-associated; WCE, whole-cell protein extracts; CF, cell filtrate; MS, mass spectrometry; PKA, protein kinase A; PSD, postsource decay. E-mail address of the corresponding author:
[email protected]
and phosphatases, once thought to be largely restricted to eukaryotes, are found widely in bacterial proteomes as well, and may play important roles in cell signaling.1 The genome of Mycobacterium tuberculosis includes genes encoding one phospho-Ser/Thr phosphatase (pstP), two phospho-Tyr phosphatases (ptpA, ptpB) and as many as 11 Ser/Thr protein kinases (pknA to pknL). Except for PknG and PknK, which are soluble proteins, all others Ser/Thr protein kinases (STPKs) are predicted to be transmembrane “receptor-like” proteins.2 Secreted PtpB appears to be required for survival of the bacillus in guinea pigs,3 and secretion of PknG within macrophage phagosomes promotes intracellular survival of the parasite.4
0022-2836/$ - see front matter q 2005 Elsevier Ltd. All rights reserved.
954 However, except for these roles in TB pathogenesis, the functions of “eukaryotic-like” protein kinases and phosphatases in mycobacterial signal transduction pathways remain largely to be elucidated. Only three out of the 11 M. tuberculosis STPKs (i.e. PknA, PknB and PknG) appear to be required to sustain mycobacterial growth,5 suggesting a possible involvement of these enzymes in essential signal transduction pathways. Inactivation of the pknG gene by allelic exchange resulted in decreased viability of M. tuberculosis both in vitro and in mice.6 The two other genes, pknA and pknB, are located in a single operon that also contains the genes pstP,2 encoding the only Ser/Thr protein phosphatase, as well as rodA and pbpA, two genes that code for morphogenic proteins involved in peptidoglycan synthesis during cell growth.7 This operon is highly conserved in mycobacterial and other related genomes, including Mycobacterium leprae,8 where extensive gene decay resulted in the loss of all other kinases (except for pknG and pknL) and phosphatase functional genes. In previous work, we have shown that PknB is regulated by autophosphorylation and dephosphorylation by the Ser/Thr protein phosphatase PstP,9 as observed for other bacteria where the genes encoding the kinase–phosphatase pair are genetically linked.10–12 In PknB, two threonine residues (Thr171 and Thr173) from the activation loop were identified as the only sites that were fully autophosphorylated in the entire catalytic domain (residues 1–279). Either phosphatase-treated PknB or Thr-to-Ala point mutants decreased the autophosphorylation activity significantly as well as the kinase activity towards the surrogate substrate myelin basic protein (MBP).9 The crystal structure of PknB in complex with nucleotide (the first for a bacterial STPK) revealed a conserved protein fold and catalytic machinery as seen for its eukaryotic homologs.13,14 However, a disordered activation loop in the two independent crystal structures of PknB suggests an induced-fit mode of binding for the, so far unknown, physiological substrate(s). A major obstacle in our understanding of protein kinase biology in prokaryotes is the identification of physiologically relevant kinase substrates. As a first step towards the discovery of de novo protein substrates of mycobacterial kinases, we report here the identification of GarA, a Forkheadassociated (FHA) domain-containing protein, as a putative physiological substrate of PknB using a global proteomic approach. GarA is selectively phosphorylated by PknB in whole-cell protein extracts (WCE) and cell filtrates (CF) of M. tuberculosis and Mycobacterium smegmatis. We further investigated protein kinase–substrate interactions by mass spectrometry, enzymological and binding studies of wild-type and mutant proteins. These studies led us to propose a molecular mechanism for GarA substrate recruitment by mycobacterial protein kinase PknB, which involves specific interactions between the kinase activation loop and the substrate FHA domain.
Substrate Recruitment by M. tuberculosis PknB
Results Identification of de novo PknB substrates from whole-cell protein extracts The catalytic domain of PknB (PknB1–279) is active on the surrogate substrate MBP, but nothing is known about its physiological substrate(s).9 To gain some insight into putative endogenous protein substrates, we studied the ability of PknB1–279 to phosphorylate native proteins in WCE representing the soluble protein fraction of M. tuberculosis strain H37Rv. Several phosphorylated bands were observed in 1D SDS-PAGE (Figure 1(a) and (b)). However, only one band appeared fully phosphorylated when the time of reaction was decreased to two minutes and when the concentration of PknB1–279 was decreased from 2.5 mM to 0.25 mM (Figure 1(a)), suggesting that this band might contain a specific PknB substrate. No kinase activity was observed when the mycobacterial protein extract was incubated without addition of PknB 1–279, likely because nine out of the 11 mycobacterial STPKs are membrane-bound proteins and are not represented in the soluble protein extract used in our experiments. To resolve and identify the putative protein substrate(s) detected on the 1D SDS-PAGE, the reaction mixture was analyzed by 2D PAGE, autoradiography and mass spectrometry (MS). Except for the spots assigned to PknB1–279 itself (autophosphorylation), only a single cluster of five phosphorylated spots, corresponding to protein(s) of relatively low abundance in the WCE, was observed in the low molecular mass (Mr) and acidic isoelectric point (pI) regions (Figure 1(c) and (d)). As a simple control of specificity, we used mouse protein kinase A (PKA) instead of PknB1–279 in the kinase assay. Although PKA was active on many mycobacterial proteins, it was unable to phosphorylate the cluster of spots containing the PknB substrate(s) (Figure 1(b)). MS analysis (Mascot score 118, sequence coverage 67%) and proteome database† comparisons indicated that all the spots in the cluster corresponded to the same protein, Rv1827 or GarA, whose theoretical Mr (17.3 kDa) and pI (4.4) are close to the observed values. The fact that a single protein focuses in more than one spot is not unexpected. This kind of microheterogeneity has been observed frequently in 2D PAGE of mycobacterial and non-mycobacterial proteins,15–18 and may be due to differences in a number of post-translational modifications (deamination, oxidation of sidechains, etc.) other than phosphorylation itself.19 A truncated form of GarA, CFP-17, has been identified in the CF of M. tuberculosis H37Rv.20 Both CFP-17 and whole-length recombinant GarA were found to induce protective immunity to TB in animal models.20,21 Prompted by these observations, we used the same proteomic approach as † http://www.mpiib-berlin.mpg.de/2D-PAGE
955
Substrate Recruitment by M. tuberculosis PknB
above to examine the CF fraction of M. tuberculosis and we again identified full-length GarA (but not the truncated form) as the only substrate of PknB1–279 detectable by autoradiography (data not shown). GarA is a 162 residue protein that includes an FHA domain in its C-terminal region. FHA domains are small protein modules that mediate protein– protein interactions via pThr recognition, and are involved frequently as mediators of protein–protein interactions in STPK-dependent signal transduction pathways.22 Orthologues of garA, with predicted sequence identities of more than 70% are found in mycobacteria (Mycobacterium bovis, M. leprae, Mycobacterium avium subsp paratuberculosis, M. smegmatis), related corynebacteria (Corynebacterium efficiens, Corynebacterium glutamicum, Corynebacterium diphtheriae), Rhodococcus fascians and Nocardia farcinica, all members of the actinomycetale order. Multiple sequence alignment of these proteins reveals a strong conservation of the C-terminal FHA domain as well as a sequence motif in the N-terminal region with no assigned role (Figure 2). Of particular interest is M. smegmatis GarA (84% sequence identity with the M. tuberculosis protein), since this protein has been described as a putative regulator of glycogen accumulation during the exponential phase of cell growth.23 To verify whether M. tuberculosis PknB (which has 92% sequence identity in its catalytic domain with the M. smegmatis enzyme) can phosphorylate M. smegmatis GarA, we used the same proteomic approach as before. In both M. smegmatis WCE and CF proteins, we found a cluster of spots in 2D PAGE similar to that observed for M. tuberculosis GarA (data not shown) and, for the CF fraction, identified them as GarA by MS. Altogether, these experiments show that GarA is the optimal detectable substrate of PknB among all soluble proteins present in the WCE and CF fractions of both M. tuberculosis and M. smegmatis. Characterization and phosphorylation of recombinant GarA
Figure 1. Identification of PknB 1–279 substrates. (a) PhosphoImager quantification of protein 33P incorporation observed in 1D SDS-PAGE of phosphorylated proteins in M. tuberculosis WCE after incubation with PknB1–279 (red, 30 minutes at 2.5 mM; blue, two minutes at 2.5 mM; green, two minutes at 0.25 mM). A single band (labeled S) appears fully phosphorylated under all experimental conditions; the peak corresponding to PknB1–279 autophosphorylation (labeled A) is indicated. (b) Incubation of WCE proteins with PknB1–279 (thick line) or mouse PKA (thin line). (c) Silver-stained 2D PAGE of M. tuberculosis WCE incubated with PknB1–279. The boxed region includes the only cluster of spots (besides PknB1–279 itself) with detectable phosphorylation, including five isoforms with a pI between 4.0 and 4.3. (d) Enlarged view of the boxed region in (c) after incubation with PknB1–279. Both the silver-stained gel (left) and the autoradiograph (right) are shown. The circled spots correspond to GarA (Rv1827), which is not phosphorylated by PKA.
The full-length protein from M. tuberculosis (GarA1–162) and a truncated form corresponding to CFP-17 (GarA31–162) were produced in recombinant form and purified to homogeneity (Figure 3(a)). The two proteins displayed very similar sedimentation coefficients (w1.3 S) in analytical ultracentrifugation experiments; these values correspond to a mass of about 15 kDa and suggest that both forms of GarA exist as monomers in solution. On the other hand, the two proteins displayed a markedly different migration pattern on native gels (data not shown), which can probably be explained by the acidic character of the N-terminal region (eight Glu/Asp in 30 residues). PknB1–279 was found to phosphorylate full-length GarA, as indicated by the incorporation of 33P and the electrophoretic shift of GarA protein (Figure 3(b)). After two minutes of kinase reaction, 50% of GarA was phosphorylated (2.3 fmol of [33P]GarA
956
Substrate Recruitment by M. tuberculosis PknB
Figure 2. Multiple sequence alignment of close homologues of M. tuberculosis GarA having more than 70% sequence identity. The amino acid sequence and secondary structure elements of human Ki67 FHA domain (PDB code 1r21)33 are shown below the alignment. The single phosphorylation site of GarA is indicated by a red star, and FHA residues critical for pThr recognition are indicated by blue stars (Mt, M. tuberculosis; Ma, M. avium sp. paratuberculosis; Ml, M. leprae; Ms, M. smegmatis; Rf, Rhodococcus fascians; Nf, Nocardia farcinica; Cg, Corynebacterium glutamicum; Ce, C. efficiens; Cd, C. diptheriae).
per pmol of PknB1–279). In contrast, PknB1–279 was inactive on the truncated protein GarA31–162, suggesting that the phosphorylation site was located in the N-terminal region of the protein. Under the same experimental conditions (0.25 mM PknB1–279, enzyme to substrate molar ratio of 1:100) the activity of PknB1–279 on the surrogate substrate MBP was undetectable. The apparent kinetic parameters of the phosphorylation of GarA by PknB1–279, determined at three different substrate concentrations (10–30 mM), are VmaxZ0.04 mM/s and K m (ATP)Z2.2 mM. Though these values should be taken with caution (since the protein substrate was used at subsaturating concentrations), they fall within the range of values reported for other kinases and are thus consistent with a possible interaction between the two proteins in vivo.24 Next, both forms of recombinant GarA were incubated with PknB1–279 and ATP, and analysed by mass spectrometry. As expected, the molecular mass of GarA31–162 was unchanged after incubation with PknB1–279, whereas that of GarA1–162 showed a mass shift corresponding to the incorporation of a single phosphate group (Figure 3(c)). The acceptor residue in GarA was further identified by mass spectrometry and peptide sequencing studies. Comparison of mass spectra from trypsin and EndoGluC digestions before and after PknB1–279 phosphorylation revealed in each case a single monophosphorylated peptide, corresponding to residues 1–26 (plus one residue from the His-tag) after digestion with trypsin, residues 21–32 after digestion with EndoGluC and residues 21–26 after
double digestion (see Materials and Methods). In all cases, peptide phosphorylation was confirmed by enzymatic treatment with alkaline phosphatase and post-source decay (PSD) MS fragmentation experiments. Finally, direct protein sequencing of the purified phospho-peptides led to the identification of Thr22 in the conserved motif 20-ETTSVFRAD-28 (Figure 2) as the single phosphate-acceptor residue in GarA. Consistent results were obtained with native GarA from the M. tuberculosis CF preincubated with PknB1–279, where we observed the incorporation of one phosphate group to a single peptide (residues 11–26) including Thr22 (average theoretical MHCZ1894.8, measured m/zZ1894.3, detected in MS linear mode measurements). PknB–GarA interactions Somewhat surprisingly, PknB1–279 was active on GarA and the surrogate substrate MBP, but not on purified tryptic fragments from these proteins that included the corresponding phosphorylation sites, suggesting that PknB activity requires entire protein substrates. Since FHA domains are known to specifically bind pThr residues,22 and the catalytic domain of PknB has itself up to four phosphorylated residues in its activation loop,14 with two of them (Thr171 and Thr173) fully phosphorylated,9 we asked whether the FHA domain of GarA could mediate kinase–substrate interactions. Indeed, PknB1–279 was shown to form a stable complex with GarA in the absence of ATP (Figure 4(a)), as indicated by the disappearance of the band corresponding to GarA when it is incubated with
Substrate Recruitment by M. tuberculosis PknB
Figure 3. Characterization and PknB phosphorylation of whole-length GarA (GarA1–162) and a truncated form lacking the first 30 residues (GarA31–162). (a) SDS-PAGE of the two proteins after purification by metal-affinity and gel-filtration chromatography. (b) PknB1–279 kinase assay. The enzyme:substrate mixtures (0.25 mM PknB1–279, molar ratio 1:100) were incubated at 30 8C for the specified time(s). Proteins were subjected to 1D PAGE, stained with Coomassie brilliant blue, dried and exposed to a PhosphoImager. Both the stained gel (left) and the autoradiograph (right) are shown. (c) Whole-protein MALDI spectra of the two proteins before and after (inset) the kinase assay. (d) A representation of GarA showing the N-terminal phospho-acceptor residue and the C-terminal FHA domain.
PknB1–279. Sedimentation velocity studies of the GarA–PknB1–279 complex showed that its sedimentation coefficient was higher than those of each of the two proteins analyzed separately. Using the Svedberg equation and fitted values of s and D, the
957 expected mass of the sZ2.7 S species is about 45 kDa, corresponding closely to a 1:1 stoichiometry for the interaction. To further characterize kinase–substrate interactions, surface plasmon resonance studies were carried out with the two recombinant forms of GarA (GarA1–162 and GarA31–162) immobilized on the sensor chip. These experiments confirmed the interaction of full-length GarA with PknB1–279 and, more importantly, demonstrated that the truncated form of GarA (i.e. lacking the phosphorylation motif) interacts with the kinase in a way similar to that of the full-length protein (Figure 4(b)), suggesting strongly that the kinase active-site cleft is not involved in this interaction. Furthermore, the interaction was abolished when PknB1–279 was treated beforehand with the cognate phosphatase PstP (Figure 4(b)), or when the double PknB1–279 mutant (T171A/T173A) was used instead of wildtype PknB1–279 (Figure 4(c)). The above observations demonstrate unambiguously that the FHA domain of GarA is interacting directly with pThr residue(s) from the kinase activation loop. Furthermore, PstPtreated PknB1–279 and the double Thr-to-Ala PknB mutant were found to have undetectable or very low activity against GarA1–162 (at least 100 times lower than the wild-type PknB1–279 activity at two minutes of reaction) (Figure 4(d)), indicating a direct correlation between FHA binding and PknB activity. We next asked whether the two fully phosphorylated Thr residues of PknB1–279 were necessary for complex formation and activity. The point mutants T171A and T173A of PknB1–279, which are still phosphorylated in the non-mutated Thr,9 interacted with GarA1–162, albeit less efficiently than wild-type PknB1–279 (Figure 4(c)), and retained a partial activity on GarA1–162 (two to three times lower than the activity of wild-type PknB1–279 after two minutes of reaction, see Figure 4(d)). Altogether, the above results demonstrate that a phosphorylated activation loop (including the two pThr residues) is required for both optimal PknB1–279–GarA interaction and maximal kinase activity. In addition to PknB, we have observed recently that three other M. tuberculosis kinases (PknD, PknE and PknF) have multiple pThr residues in their activation loops (Figure 5(a); and our unpublished results). Indeed, similar results have been observed for B. subtilis PrkC,25 suggesting that multiple phosphorylation of the activation loop might be a general feature of bacterial STPKs. We therefore investigated whether GarA could be a substrate for PknD, PknE and PknF. All three enzymes were able to phosphorylate GarA to some extent (Figure 5(b)), thus emphasizing the importance of the docking interaction between the substrate FHA domain and the kinase activation loop. While the three kinases displayed a lower activity than PknB1–279 (probably modulated by sequence-specific features), it is important to remark that the activity of the four protein kinases on the common surrogate substrate MBP was undetectable under the experimental conditions used to produce Figure 5(b). For
958
Substrate Recruitment by M. tuberculosis PknB
Figure 4. PknB1–279–GarA interactions. (a) Formation of a complex between PknB1–279 and GarA in the absence of ATP as seen in native gels (left) and sedimentation coefficient distribution (right: GarA, dotted line; PknB1–279, broken line; complex, continuous line). The same enzyme and substrate concentration was used in all cases (90 mM for sedimentation studies). PknB1–279 migrates as two bands in native gels, corresponding to partial monophosphorylation at its N-terminal His-tag.9 (b) Real-time interactions of PknB with immobilized GarA31–162 (top) and immobilized GarA1–162 (bottom) assayed by surface plasmon resonance. Proteins injected were phosphorylated PknB1–279 (thick line) and dephosphorylated PknB1–279 (thin line). (c) SPR assays of immobilized GarA31–162 with PknB1–279 (purple, wild-type; red, T171A; green, T173A; blue, T171A/T173A). (d) Kinase assay of PknB1–279 (wildtype and Thr-to-Ala mutants) on GarA (enzyme:substrate ratio of 1:100). Proteins were subjected to 1D PAGE, stained with Coomassie brilliant blue, dried and exposed to a PhosphoImager. Both the stained gel (left, labeled ST) and the autoradiograph (right, labeled AR) are shown. Total activities (in fmol [33P]GarA per pmol of PknB) for two minutes and 30 minutes of reaction are the following: wildtype PknB, 2.28 and 4.65; PknBT171A, 0.73 and 4.53; PknBT173A, 1.02 and 5.26; and PknB2TA, 0.04 and 0.15 (all experimental errors within 15%).
comparison purposes, the activity of PknB1–279 on MBP (enzyme to substrate molar ratio 1:60, two minutes of reaction) was 0.02 fmol of [33P]MBP per pmol of PknB, i.e. 150 times lower than the PknB activity for GarA at a lower enzyme to substrate molar ratio (1:100). A molecular model of PknB–GarA interaction The above results can be rationalized in terms of a model of PknB substrate recruitment, in which the activation loop provides a secondary docking site for FHA-mediated binding, linking substrate recruitment to enzyme activation (Figure 6). Direct FHA recognition of the pThr residue(s) in the PknB activation loop has been demonstrated unambiguously by the experiments described above (Figure 4), since the catalytic domain of PknB1–279 used in these studies has only two fully phosphorylated Thr residues in the activation loop,9 and the double Thr-to-Ala substitution (or preincubation of PknB1–279 with phosphatase)
completely abolished GarA recognition. The existence of multiple phosphorylation sites in the kinase activation loop appears to be important for optimal substrate binding and kinase activity, probably because phosphate groups may be required to bind both the FHA domain (substrate recruitment) as well as the conserved Arg cluster including Arg137 from the catalytic loop (enzyme activation). The proposed interaction model is consistent with, and may account for, two other previous observations: (a) the PknB activation loop is disordered in two independent crystal structures of kinase–nucleotide complexes,13,14 suggesting strongly that further substrate-docking interactions (such as those involving the FHA domain of GarA) may be required to stabilize the active conformation (i.e. through the interaction of the activation loop phosphate group(s) with the conserved Arg residue in the catalytic loop); and (b) PknB1–279 is unable to phosphorylate tryptic peptides derived from the protein substrates GarA or MBP that include the corresponding phosphorylation sites (probably
959
Substrate Recruitment by M. tuberculosis PknB
activity was critically dependent on FHA-mediated kinase recognition.26
Discussion
Figure 5. Different M. tuberculosis kinases are able to phosphorylate GarA. (a) Sequence comparison of the activation loops of four protein kinases, with autophosphorylated residues indicated in green. (b) Autoradiography of kinase assays using GarA as a substrate (0.25 mM kinase, enzyme to substrate molar ratio of 1:100). The activity of PknD, PknE and PknF on GarA at 30 minutes of reaction were, five, 32 and two times lower than the PknB activity, respectively.
because the activation loop is not able to achieve its active conformation). Furthermore, it is possible that mycobacterial kinases PknD, PknE and PknF could phosphorylate GarA (Figure 5(b)) using a similar mechanism, because the interaction between the FHA domain and the activation loop seems to contribute most of the enzyme–substrate binding energy (Figure 4(b) and (c)) and all the above kinases have two or more pThr residues in their activation loops (Figure 5(a)). This might be the case also for two other FHA-containing proteins (Rv1747 and EmbR), which were reported to be substrates in vitro of mycobacterial kinases PknF and PknH, respectively.26,27 Indeed, those authors suggested that recruitment and phosphorylation of Rv1747 could depend on the interaction between its FHA domains and the phosphorylated form of PknF,27 and, in agreement with this hypothesis, demonstrated by mutagenesis studies that PknH
Figure 6. Proposed mechanism for the interactions between PknB and GarA, in which phosphorylation of the activation loop is required for both substrate recruitment and enzyme activation.
In this study, we have identified GarA, an FHA domain-containing protein, as the optimal detectable substrate of PknB1–279 in a non-denatured WCE from mycobacteria, using a proteomic approach that combined 2D PAGE, autoradiography and MS identification. This type of proteome-wide approach represents a useful tool for unbiased de novo substrate identification, as illustrated here with PknB, an essential receptor-like transmembrane kinase for which no hints were previously available about its putative substrates, and similar work with the eukaryotic mitogen-activated protein kinase MKK6.28 Although “optimal” substrate does not necessarily imply “physiological” substrate (and further work is undoubtedly required to validate this hypothesis), in the case of the GarA–PknB1–279 interaction a few lines of evidence lend support to its possible physiological relevance in mycobacteria: (a) the optimal detectable kinase substrate identified in the WCE contains an FHA domain, which is a protein module known to bind specifically pThr residues and involved frequently in eukaryotic STPK-dependent signal transduction pathways;22 (b) two other mycobacterial kinases, PknF and PknH, were reported to phosphorylate FHA domain-containing proteins; 26,27 (c) the presence of a single phosphorylation site in the N-terminal region of GarA (which contains several additional Ser or Thr residues outside the FHA domain) argues for a specific PknB–GarA interaction. Such specificity is absent, for instance, from PknB-mediated phosphorylation of the surrogate substrate MBP, in which the kinase transfers phosphate groups to at least five different Thr or Ser residues (our unpublished results); (d) the amino acid sequences corresponding to the phosphorylation motif and the FHA domain of GarA are largely conserved in mycobacteria and related actinomycetes (Figure 2), in agreement with a similar conservation pattern of the operon including the pknB gene; and (e) M. smegmatis GarA has been proposed to act as a STPK-dependent regulator of glycogen degradation during cell growth.23 The molecular model of PknB–GarA interactions proposed here evokes the eukaryotic STPKs that rely on direct docking interactions with their substrates, using sites distinct from the active-site cleft, to achieve specificity.29 In some of these cases, the interaction with the docking site is also mediated by FHA domains, as in the essential protein kinase Rad53 from budding yeast, which recruits in this way its downstream substrate Dun1 upon DNA damage.30 However, PknB1–279 differs from the above STPKs, in that the FHA domain binds the kinase activation loop, suggesting that
960 multiple phosphorylation sites in this loop may fulfil roles other than enhancement of the catalytic activity. In this respect, the PknB–GarA interaction is more reminiscent of the SH2-mediated recruitment of the adaptor protein APS to the activated insulin receptor,31 where kinase–substrate interactions occur exclusively via the phosphorylated activation loop. Oriented peptide library screenings were carried out for a series of FHA domain-containing proteins, including M. tuberculosis GarA.32 These studies demonstrated the absence of well-defined sequence consensus motifs for most FHA domains, except the common requirement of a pThr residue for binding. Indeed, various crystal and solution structures of FHA–peptide complexes have further confirmed that pThr recognition plays a critical role in complex formation and suggested that FHA domains may have evolved to bind the pThr-X-X-X motif in an extended conformation, mostly through backbone interactions,33 thus accounting for their low peptide sequence specificity. The M. tuberculosis genome contains at least seven genes encoding proteins with one or more FHA domains, and at least four (but probably more) Ser/Thr protein kinases have two or more pThr residues in their activation loops. The proposed model of protein kinase–FHA substrate interactions (Figure 6) might therefore be relevant, besides PknB–GarA, to other cases such as the observed interactions between PknF and FHA-containing Rv1747,26 or PknH and FHA-containing EmbR.27 However, it also raises the question about the specificity of these interactions, given the usual promiscuity of protein kinase activity in vitro.34 As a matter of fact, GarA is a substrate of four different kinases (Figure 5), and one of these enzymes (PknF) is active also on Rv1747. Unfortunately, our understanding of protein kinase/substrate interactions in mycobacteria (and prokaryotes in general) remains very superficial, and just a few target substrates for bacterial STPKs have been identified so far, most of them due to the presence of their genes in the close vicinity of cognate protein kinase genes.26,27,35,36 Here, we report a proteome-wide approach to identify de novo protein substrates of an essential M. tuberculosis protein kinase and put forward a model of activation loop-mediated PknB substrate recruitment that may have important biological implications in cell signaling. Further work is required to investigate whether GarA phosphorylation in vivo may be related directly to its function as regulator of glycogen degradation during cell growth,23 and, more generally, to ascertain the precise role of STPKs in bacterial physiology and their potential value as new targets for therapeutic intervention against tuberculosis.
Materials and Methods Mycobacterial cell growth and protein extraction Strains M. smegmatis mc2 155 and M. tuberculosis H37Rv
Substrate Recruitment by M. tuberculosis PknB
were grown in 20 ml of Sauton liquid medium and 0.05% (v/v) Tween-80 at 37 8C for two days and for six days, respectively. Cultures were harvested by centrifugation. The culture filtrate (CF) was recovered after filtration (0.22 mm pore size) and concentrated to 1 mg/ml. The pellet was washed twice and suspended in 20 mM Tris– HCl (pH 7.5), with EDTA-free protease inhibitors (Roche). Cells were broken by shaking with 106 mm acid washedglass beads and centrifuged to obtain the supernatant fraction (WCE) at 10 mg/ml. Protein kinase assays The wild-type and point mutants T171A and T173A of the PknB catalytic domain (PknB1–279, lacking both the juxtamembrane region and the extracellular domain) were produced as described,9 and mouse PKA was purchased from Biolabs. The enzyme preparation of PknB1–279 used in the present work was autophosphorylated during protein expression in Escherichia coli, and further incubation of the purified enzyme with ATP did not change the phosphorylation pattern significantly. As previously reported,9 two Thr residues from the activation loops (Thr171 and Thr173) were the only fully phosphorylated residues in the entire catalytic domain, and their phosphate groups are both removed after treatment with the cognate mycobacterial Ser/Thr phosphatase, PstP. Mycobacterial protein extracts were incubated with 0.25–2.5 mM PknB1–279 and 0.1 mM ATP spiked with [g-33P]ATP (0.4 mCi/ml of reaction) in the kinase buffer (50 mM Hepes (pH 7), 1 mM DTT, 0.01% (w/v) Brij35, 2 mM MnCl2) containing 6% (v/v) glycerol at 30 8C for different lengths of time. The reaction was stopped by adding EDTA (15 mM final concentration) and the proteins resolved in 1D or 2D PAGE. The 1D gels were dried and analyzed with a PhosphorImager (Molecular Dynamics). Mycobacterial protein extracts were also incubated without kinase to verify possible endogenous phosphorylation. Two-dimensional electrophoresis The 2D PAGE of M. tuberculosis WCE was performed in 23 cm!30 cm gels as described.37 For analytical experiments, gels were loaded with 30 mg of protein sample, previously desalted with the 2-D Clean-Up Kit (Amersham), silver-stained and autoradiographed to detect phosphorylated spots. Preparative gels were loaded with 300 mg of protein sample and stained with Coomassie brilliant blue. Kinase-treated M. tuberculosis CF and M. smegmatis CF/WCE (5 mg or 30 mg) were loaded onto pH 3–6 IPG strips (Biorad) using a Protean IEF Cell. After isoelectric focussing, the IPG strips were equilibrated for 15 minutes in buffer A (50 mM Tris–HCl (pH 8.8), 6 M urea, 30% glycerol, 2% (w/v) SDS) C1% (w/v) DTT followed by 15 minutes in buffer AC4% (w/v) iodoacetamide, and then loaded onto an SDS/12% (w/v) polyacrylamide gel (1 mm thick). Electrophoresis was performed at a constant current of 18 mA, and the gels were silver-stained for visual inspection or stained with Coomassie brilliant blue for spot identification by mass spectrometry. Production of recombinant GarA The coding sequences of GarA31–162 and GarA1–162 were amplified by PCR from cosmid MTCY1A11, using appropriate primers with NdeI and BamHI restriction
961
Substrate Recruitment by M. tuberculosis PknB
sites. Digested PCR products were ligated with phage T4 DNA ligase (Biolabs) in a pET28 expression vector, and the constructs verified by DNA sequencing. Transformed E. coli BL21 (DE3 pLys) cells were grown at 37 8C in LB medium with 30 mg mlK1 kanamycin. The recombinant proteins were purified to homogeneity by metal-affinity (Ni-column) and gel-filtration (Superdex 75) chromatography. Analytical ultracentrifugation Sedimentation velocity experiments were carried out at 20 8C and 48,000 rpm using the Proteomelab XLI analytical ultracentrifuge (Beckman-Coulter) and absorption optics (280 nm). Samples were equilibrated beforehand into analysis buffer (25 mM Hepes (pH 8), 100 mM NaCl, 2 mM b-mercaptoethanol, 5% glycerol). The buffer parameters were: density 1.00591 g mlK1, viscosity 0.01044 P (1 PZ10K1 Pa s). The extinction coefficients of PknB1–279 and GarA were 14,900 MK1 cmK1 and 1490 MK1 cmK1, respectively, and their partial specific volumes (as calculated with SEDNTRP) 0.73 ml gK1 and 0.72 ml gK1, respectively. Runs were performed for four hours, during which time 100 scans were taken. Sedimentation velocity data were fit to finite element solutions of the Lamm equation using a model of continuous c(s) distribution with the program SEDFIT.38 Mass spectrometry Mass spectra were acquired in linear or reflector modes on a Voyager DE-PRO MALDI TOF system (Applied Biosystems). Sample preparation (HPLC procedures, enzymatic dephosphorylation, and proteolytic digestions both in-gel and in-solution) and MS measurements (calibration procedures and matrix selection) were as described.9,39 Peptide masses determined in reflector mode were employed for protein identification using the MASCOT program. Phosphopeptides were identified by mass measurement in linear mode and confirmed by PSD-MS analysis. A tryptic peptide corresponding to GarA sequence 1–26 plus the C-terminal His-tag Ser residue, either native or with an oxidized methionine residue, was observed only in the control sample, whereas a mass signal corresponding to the oxidized peptide sequence plus a single phosphate group was observed after PknB phosphorylation (average theoretical MHCZ3141.3, measured value 3141.1). Upon digestion with EndoGluC, a mass signal corresponding to peptide 21–32 present in the control sample disappeared after phosphorylation and a new mass signal 80 Da higher became apparent (average MHCZ1453.4). For the double digestion with trypsin and EndoGluC, a similar mass shift of 80 Da before and after phosphorylation was observed for peptide 21–26 (TTSVFR). For mass spectrometry, kinase assays were performed for 30 minutes at 35 8C in 50 mM Hepes (pH 7.0), 1 mM DTT, 2 mM MnCl2 120 mM ATP and 10 mM GarA (enzyme to substrate molar ratio 1:5). To identify the phosphorylation site unambiguously, purified phospho-peptide 21–28 was subjected to Edman degradation with an Applied Biosystems model 494. Surface plasmon resonance assays Experiments were performed at 25 8C in kinase buffer without glycerol on a Biacore 2000 instrument. GarA1–162 and GarA31–162 were immobilized on two independent
flow-cells of a CM5 sensorchip, using the Amine Coupling Kit (Biacore), to a level of 4000 resonance units (RU) or 9250 RU respectively. Eight different concentrations of wild-type PknB1–279, the single PknB1–279 mutants (T171A) and (T173A), the double Thr-to-Ala PknB1–279 mutant, and PstP-treated PknB1–279 (20 nM– 20 mM) were injected onto the surfaces at a flow-rate of 20 ml minK1. The association profiles were doublesubtracted from those obtained by excluding each of the ligands in turn, and analyzed using the Biaevaluation 3.1 software (Biacore AB).
Acknowledgements We thank Ida Rosenkrands for the gift of CFP-17 and Monika Schmid, Jacques d’Alayer and Ce´dric Fiez-Vandal for technical help. This work was supported by the European Commission (X-TB, contract QLK2-CT-2001-02018), the Institut Pasteur (PTR-46, GPH-5), the Ministe`re de la Recherche, France (contract 01-B-0095) and DINACYT, Uruguay (PDT29/160).
References 1. Kennelly, P. J. (2002). Protein kinases and protein phosphatases in prokaryotes: a genomic perspective. FEMS Microbiol. Letters, 206, 1–8. 2. Cole, S. T., Brosch, R., Parkhill, J., Garnier, T., Churcher, C., Harris, D. et al. (1998). Deciphering the biology of Mycobacterium tuberculosis from the complete genome sequence. Nature, 393, 537–544. 3. Singh, R., Rao, V., Shakila, H., Gupta, R., Khera, A., Dhar, N. et al. (2003). Disruption of mptpB impairs the ability of Mycobacterium tuberculosis to survive in guinea pigs. Mol. Microbiol. 50, 751–762. 4. Walburger, A., Koul, A., Ferrari, G., Nguyen, L., Prescianotto-Baschong, C., Huygen, K. et al. (2004). Protein kinase G from pathogenic mycobacteria promotes survival within macrophages. Science, 304, 1800–1804. 5. Sassetti, C. M., Boyd, D. H. & Rubin, E. J. (2003). Genes required for mycobacterial growth defined by high density mutagenesis. Mol. Microbiol. 48, 77–84. 6. Cowley, S., Ko, M., Pick, N., Chow, R., Downing, K. J., Gordhan, B. G. et al. (2004). The Mycobacterium tuberculosis protein serine/threonine kinase PknG is linked to cellular glutamate/glutamine levels and is important for growth in vivo. Mol. Microbiol. 52, 1691–1702. 7. Matsuhashi, M. (1994). Utilization of lipid-linked precursors and the formation of peptidoglycan in the process of cell growth and division. In Bacterial Cell Wall (Ghuysen, J. M. & Hakenbeck, R., eds), pp. 55–71, Elsevier, Amsterdam. 8. Cole, S. T., Eiglmeier, K., Parkhill, J., James, K. D., Thomson, N. R., Wheeler, P. R. et al. (2001). Massive gene decay in the leprosy bacillus. Nature, 409, 1007–1011. 9. Boitel, B., Ortiz-Lombardia, M., Duran, R., Pompeo, F., Cole, S. T., Cerven˜ansky, C. & Alzari, P. M. (2003). PknB kinase activity is regulated by phosphorylation in two Thr residues and dephosphorylation by
962
10.
11.
12.
13.
14.
15.
16.
17.
18.
19. 20.
21.
22. 23.
PstP, the cognate phospho-Ser/Thr phosphatase in Mycobacterium tuberculosis. Mol. Microbiol. 49, 1493–1508. Obuchowski, M., Madec, E., Delattre, D., Boel, G., Iwanicki, A., Foulger, D. & Seror, S. J. (2000). Characterization of PrpC from Bacillus subtilis, a member of the PPM phosphatase family. J. Bacteriol. 182, 5634–5638. Rajagopal, L., Clancy, A. & Rubens, C. E. (2003). A eukaryotic type serine/threonine kinase and phosphatase in Streptococcus agalactiae reversibly phosphorylate an inorganic pyrophosphatase and affect growth, cell segregation, and virulence. J. Biol. Chem. 278, 14429–14441. Nova´kova´, L., Saskova´, L., Pallova´, P., Janecek, J., Novotna´, J., Ulrych, A. et al. (2005). Characterization of a eukaryotic type serine/threonine protein kinase and protein phosphatase of Streptococcus pneumoniae and identification of kinase substrates. FEBS J. 272, 1243–1254. Ortiz-Lombardia, M., Pompes, F., Boitel, B. & Alzari, P. M. (2003). Crystal structure of the catalytic domain of the PknB serine/threonine kinase from Mycobacterium tuberculosis. J. Biol. Chem. 278, 13094–13100. Young, T. A., Delagoutte, B., Endrizzi, J. A., Falick, A. M. & Alber, T. (2003). Structure of Mycobacterium tuberculosis PknB supports a universal activation mechanism for Ser/Thr protein kinases. Nature Struct. Biol. 10, 168–174. Weldingh, K., Rosenkrands, I., Jacobsen, S., Rasmussen, P. B., Elhay, M. J. & Andersen, P. (1998). Two-dimensional electrophoresis for analysis of Mycobacterium tuberculosis culture filtrate and purification and characterization of six novel proteins. Infect. Immun. 66, 3492–3500. Harboe, M., Nagai, S., Patarroyo, M. E., Torres, M. L., Ramirez, C. & Cruz, N. (1986). Properties of proteins MPB64, MPB70, and MPB80 of Mycobacterium bovis BCG. Infect. Immun. 52, 293–302. Bini, L., Sanchez-Campillo, M., Santucci, A., Magi, B., Marzocchi, B., Comanducci, M. et al. (1996). Mapping of Chlamydia trachomatis proteins by immobilinepolyacrylamide two-dimensional electrophoresis: spot identification by N-terminal sequencing and immunoblotting. Electrophoresis, 17, 185–190. Sonnenberg, M. G. & Belisle, J. T. (1997). Definition of Mycobacterium tuberculosis culture filtrate proteins by two-dimensional polyacrylamide gel electrophoresis, N-terminal amino acid sequencing, and electrospray mass spectrometry. Infect. Immun. 65, 4515–4524. Mann, M. & Jensen, O. N. (2004). Proteomic analysis of post-translational modifications. Nature Biotechnol. 21, 255–261. Weldingh, K., Rosenkrands, I., Jacobsen, S., Rasmussen, P. B., Elhay, M. J. & Andersen, P. (1998). Two dimensional electrophoresis for analysis of Mycobacterium tuberculosis culture filtrate and purification and characterization of six novel proteins. Infect. Immun. 66, 3492–3500. Weldingh, K. & Andersen, P. (1999). Immunological evaluation of novel Mycobacterium tuberculosis culture filtrate proteins. FEMS Immunol. Med. Microbiol. 23, 159–164. Durocher, D. & Jackson, S. P. (2002). The FHA domain. FEBS Letters, 513, 58–66. Belanger, A. E. & Hatfull, G. F. (1999). Exponentialphase glycogen recycling is essential for growth of Mycobacterium smegmatis. J. Bacteriol. 181, 6670–6678.
Substrate Recruitment by M. tuberculosis PknB
24. al-Obeidi, F. A., Wu, J. J. & Lam, K. S. (1998). Protein tyrosine kinases: structure, substrate specificity, and drug discovery. Biopolymers, 47, 197–223. 25. Madec, E., Stensballe, A., Kjellstrom, S., Cladiere, L., Obuchowski, M., Jensen, O. N. & Seror, S. J. (2003). Mass spectrometry and site-directed mutagenesis identify several autophosphorylated residues required for the activity of PrkC, a Ser/Thr kinase from Bacillus subtilis. J. Mol. Biol. 330, 459–472. 26. Molle, V., Kremer, L., Girard-Blanc, C., Besra, G. S., Cozzone, A. J. & Prost, J. F. (2003). An FHA phosphoprotein recognition domain mediates protein EmbR phosphorylation by PknH, a Ser/Thr protein kinase from Mycobacterium tuberculosis. Biochemistry, 42, 15300–15309. 27. Molle, V., Soulat, D., Jault, J. M., Grangeasse, C., Cozzone, A. J. & Prost, J. F. (2004). Two FHA domains on an ABC transporter, Rv1747, mediate its phosphorylation by PknF, a Ser/Thr protein kinase from Mycobacterium tuberculosis. FEMS Microbiol. Letters, 234, 215–223. 28. Fernando, P., Deng, W., Pekalska, B., Derepentigny, Y., Kothary, R., Kelly, J. F. & Megeney, L. A. (2005). Active kinase proteome screening reveals novel signal complexity in cardiomyopathy. Mol. Cell Proteomics, 4, 673–682. 29. Biondi, R. M. & Nebreda, A. R. (2003). Signalling specificity of Ser/Thr protein kinases through docking-site-mediated interactions. Biochem. J. 372, 1–13. 30. Lee, S. J., Schwartz, M. F., Duong, J. K. & Stern, D. F. (2003). Rad53 phosphorylation site clusters are important for Rad53 regulation and signaling. Mol. Cell. Biol. 23, 6300–6314. 31. Hu, J., Liu, J., Ghirlando, R., Saltiel, A. R. & Hubbard, S. R. (2003). Structural basis for recruitment of the adaptor protein APS to the activated insulin receptor. Mol. Cell. 12, 1379–1389. 32. Durocher, D., Taylor, I. A., Sarbassova, D., Haire, L. F., Westcott, S. L., Jackson, S. P. et al. (2000). The molecular basis of FHA domain: phosphopeptide binding specificity and implications for phosphodependent signaling mechanisms. Mol. Cell. 6, 1169–1182. 33. Li, H., Byeon, I. J., Ju, Y. & Tsai, M. D. (2004). Structure of human Ki67 FHA domain and its binding to a phosphoprotein fragment from hNIFK reveal unique recognition sites and new views to the structural basis of FHA domain functions. J. Mol. Biol. 335, 371–381. 34. Berwick, D. C. & Tavare´, J. M. (2004). Identifying protein kinase substrates: hunting for the organgrinder’s monkeys. Trends Biochem. Sci. 29, 227–232. 35. Matsumoto, A., Hong, S. K., Ishizuka, H., Horinouchi, S. & Beppu, T. (1994). Phosphorylation of the AfsR protein involved in secondary metabolism in Streptomyces species by a eukaryotic-type protein kinase. Gene, 146, 47–56. 36. Nariya, H. & Inouye, S. (2002). Activation of 6-phosphofructokinase via phosphorylation by Pkn4, a protein Ser/Thr kinase of Myxococcus xanthus. Mol. Microbiol. 46, 1353–1366. 37. Jungblut, P. R., Schaible, U. E., Mollenkopf, H. J., Zimny-Arndt, U., Raupach, B., Mattow, J. et al. (1999). Comparative proteome analysis of Mycobacterium tuberculosis and M. bovis BCG strains: towards functional genomics of microbial pathogens. Mol. Microbiol. 33, 1103–1117. 38. Schuck, P. (2000). Size distribution analysis of
Substrate Recruitment by M. tuberculosis PknB
macromolecules by sedimentation velocity ultracentrifugation and Lamm equation modeling. Biophys. J. 78, 1606–1619. 39. Parodi-Talice, A., Duran, R., Arrambide, N., Prieto, V.,
963 Pin˜eyro, M. D., Pristch, O. et al. (2004). Abstract proteome analysis of the causative agent of Chagas disease: Trypanosoma cruzi. Int. J. Parasitol. 34, 881–886.
Edited by I. B. Holland (Received 14 March 2005; received in revised form 20 May 2005; accepted 20 May 2005)
