Chlamydophila pneumoniae PknD Exhibits Dual Amino Acid Specificity and Phosphorylates Cpn0712, a Putative Type III Secretion YscD Homolog

Abstract

Chlamydophila pneumoniae is an obligate intracellular bacterium that causes bronchitis, pharyngitis, and pneumonia and may be involved in atherogenesis and Alzheimer's disease. Genome sequencing has identified three eukaryote-type serine/threonine protein kinases, Pkn1, Pkn5, and PknD, that may be important signaling molecules in Chlamydia. Full-length PknD was cloned and expressed as a histidine-tagged protein in Escherichia coli. Differential centrifugation followed by sodium carbonate treatment of E. coli membranes demonstrated that His-PknD is an integral membrane protein. Fusions of overlapping PknD fragments to alkaline phosphatase revealed that PknD contains a single transmembrane domain and that the kinase domain is in the cytoplasm. To facilitate solubility, the kinase domain was cloned and expressed as a glutathione S-transferase (GST) fusion protein in E. coli. Purified GST-PknD kinase domain autophosphorylated, and catalytic mutants (K33G, D156G, and K33G-D156G mutants) and activation loop mutants (T185A and T193A) were inactive. PknD phosphorylated recombinant Cpn0712, a type III secretion YscD homolog that has two forkhead-associated domains. Thin-layer chromatography revealed that the PknD kinase domain autophosphorylated on threonine and tyrosine and phosphorylated the FHA-2 domain of Cpn0712 on serine and tyrosine. To our knowledge, this is the first demonstration of a bacterial protein kinase with amino acid specificity for both serine/threonine and tyrosine residues and this is the first study to show phosphorylation of a predicted type III secretion structural protein.

Chlamydophila pneumoniae is a gram-negative, obligate intracellular bacterium that is spread between humans via aerosols. It has been implicated in up to 10% of cases of community-acquired pneumonia (15), causes bronchitis, pharyngitis, and sinusitis, and exacerbates asthma and atherosclerosis (11, 17). C. pneumoniae may also play a role in multiple sclerosis and Alzheimer's disease as it has been detected within neurons, microglia, and astrocytes of afflicted patients (2, 7, 30). Seroepidemiological studies have indicated that 50 to 75% of the adult population carries antibodies to C. pneumoniae (9, 22).

Recent sequencing of the Chlamydia genome (14, 31) identified a repertoire of signaling molecules, including the eukaryote-type serine/threonine protein kinases (STPKs) Pkn1, Pkn5, and PknD, that may be important in molecular processes underlying virulence. Members of the STPK family typically catalyze the phosphorylation of serine and/or threonine side chain hydroxyl groups with phosphate derived from ATP, whereas enzymes belonging to the tyrosine kinase family phosphorylate tyrosine residues. We have previously demonstrated the temporally regulated transcription of Pkn1, Pkn5, and PknD in C. pneumoniae and demonstrated the kinase activity of recombinant Pkn1 (19). Verma and Maurelli recently demonstrated that PknD and Pkn1 from Chlamydia trachomatis autophosphorylate on serine and threonine residues and that Pkn1 can phosphorylate recombinant IncG (32), which are important observations that may have implications in the interaction of IncG with the eukaryotic protein 14-3-3β (28). Similarly, Koo and Stephens demonstrated that the histidine kinase CtcB is part of a functional two-component system in C. trachomatis (16) that may regulate transcription. The chlamydiae therefore harbor active protein kinases, but further work is required to identify the downstream substrates and determine the roles of substrate phosphorylation in chlamydial biology.

Recently, forkhead-associated (FHA) domain-containing proteins have been shown to be substrates of bacterial STPKs (12, 24). Interestingly, the chlamydiae contain a protein (Cpn0712 in C. pneumoniae) that has two putative FHA domains (FHA-1 and FHA-2) and therefore may serve as a substrate of one or more of the chlamydial STPKs. A phosphorylated threonine residue within the kinase activation loop may serve as the binding site for one or both FHA domains of Cpn0712. Given that PknD of C. trachomatis was shown to autophosphorylate on threonine residues (32), we undertook biochemical characterization of PknD from C. pneumoniae and investigated whether Cpn0712 could serve as a PknD substrate.

We report here that PknD is an integral membrane protein with a single transmembrane domain that orients the kinase domain in the cytoplasm. We show that the kinase domain is active independent of the rest of the molecule and exhibits increased activity in the presence of manganese. We demonstrate that PknD phosphorylates both FHA domains of Cpn0712. We show that PknD phosphorylates both serine/threonine and tyrosine, demonstrating for the first time the existence of an STPK with dual specificity in a prokaryote.

MATERIALS AND METHODS

Construction of expression plasmids and PknD-AP fusion constructs.

Genomic DNA was isolated from C. pneumoniae CWL029 (VR1310; ATCC; GenBank accession no. AE001363) using a Sigma GenElute kit. AttB-containing primers (Gateway; Invitrogen) specific for the complete Cpn0095 open reading frame (PknD; amino acids 1 to 932) or only the PknD kinase domain (PknD KD; amino acids 1 to 293) were used to amplify the DNA with flanking attB sites. PknD KD mutants (PknD KD[K33G], PknD KD[D156G], PknD KD[K33G; D156G], PknD KD[T185A], and PknD KD[T193A]) were created using overlapping PCR products derived from PknD KD combined with primers containing the desired point mutation(s). The complete PknD open reading frame, PknD KD, and PknD KD mutants were cloned into pDONR₂₀₁ (Gateway; Invitrogen) and subcloned into pDEST₁₅ or pDEST₁₇ to generate the expression vectors pEX₁₇PknD_, pEX₁₅KD, pEX₁₅KD[K33G], pEX₁₅KD[D156G], pEX₁₅KD[K33G; D156G], pEX₁₅KD[T185A], and pEX₁₅KD[T193A]. AttB-containing primers were also used to clone Cpn0712 (amino acids 1 to 845) and fragments of Cpn0712 encompassing the FHA domains (FHA-1, amino acids 1 to 150; FHA-2, amino acids 398 to 547) into expression vectors pDEST₁₅ and pDEST₁₇ as described above. FHA-2 point mutants S441A and N464A were created in pEX₁₅FHA-2 using overlapping PCR as described above. The topology of PknD was investigated by creating PknD-alkaline phosphatase (AP) fusion proteins. The AP gene (phoA) was amplified without a signal sequence (amino acids 27 to 471) from Escherichia coli JM109 and cloned into pUC9 to create p9P. Seven fragments spanning PknD (f1 to f7) were amplified from C. pneumoniae CWL029 genomic DNA and cloned in frame and upstream of phoA in p9P to generate clones p9Pf1 to p9Pf7. All constructs were verified by sequencing.

Sodium carbonate extraction of E. coli membranes.

E. coli cells expressing His-PknD from pEX₁₇PknD (see above) were resuspended in 1/100 of the original culture volume in Tris-HCl (pH 7.5)-150 mM NaCl containing 1× complete EDTA-free protease inhibitors (Roche) and sonicated three times for 10 s on ice (setting 5; Fisher model 100 sonic dismembrator). Lysates were centrifuged at 20,000 × g for 20 min at 4°C to remove intact cells and insoluble debris. The pellets were discarded, and the supernatants were centrifuged at 100,000 × g for 2 h at 4°C. The 100,000-×-g membrane pellets were resuspended in ice-cold 0.1 M sodium carbonate at pH 11.5 or in 0.1 M NaOH at pH 11.5, incubated on ice for 30 min, and centrifuged through sucrose cushions at 140,000 × g for 1 h. Western blotting followed by enhanced chemiluminescence (ECL) detection with mouse monoclonal antibody to His₆ (Amersham) and horseradish peroxidase (HRP)-conjugated goat anti-mouse immunoglobulin G (IgG) and Coomassie blue staining was used to track the partitioning of His-PknD.

AP assays.

PknD-AP fusion proteins p9Pf1 to p9Pf7 were expressed in E. coli strain Rosetta(pLysS), and AP assays were carried out as described previously (20). Briefly, 1 mM isopropyl-β-d-thiogalactopyranoside (IPTG) was used to induce fusion protein expression in Rosetta(pLysS) for 1 h before the cells were washed with ice-cold 1 M Tris-HCl (pH 8.0). E. coli cells were then solubilized with 0.005% (wt/vol) sodium dodecyl sulfate (SDS) and 5% (vol/vol) chloroform, and AP activity was assayed using 0.03% (wt/vol) para-nitrophenyl phosphate as the substrate. AP activity was measured by determining the optical density at 420 nm (OD₄₂₀) and was calculated using the following equation (4): [OD₄₂₀ − (1.75 × OD₅₅₀)] × 1,000/OD₆₀₀ × volume of cells × time, where the units for the volume of cells and time are milliliters and minutes, respectively. Detection of fusion proteins on Western blots was accomplished using ECL with mouse monoclonal antibody to AP (Invitrogen) at a 1:500 dilution and HRP-conjugated goat anti-mouse IgG at a 1:4,000 dilution.

Production of recombinant protein.

E. coli Rosetta(pLysS) or BL21(DE3) was transformed with protein expression vectors (see above) and plated on Luria-Bertani (LB) medium plates containing 100 μg/ml ampicillin and 34 μg/ml chloramphenicol. Three medium-size colonies from each plate were pooled in 5 ml LB broth containing the appropriate antibiotics and grown overnight at 37°C. Overnight cultures were inoculated into 750 ml LB broth containing antibiotics, incubated at 37°C at 270 rpm until the OD₆₀₀ was 0.4 to 0.6, and then cooled on ice to 20°C. Production of recombinant protein was initiated by addition of 0.2 mM IPTG, and cultures were incubated at room temperature (23°C) at 270 rpm for 2 h. Cultures were then centrifuged at 6,000 × g for 20 min, and the pellets were washed with 200 ml of ice-cold mtPBS (140 mM NaCl, 16 mM Na₂HPO₄, 4 mM NaH₂PO₄). The pellets were resuspended either in 10 ml of ice-cold mtPBS for purification on glutathione agarose or in nickel A buffer (20 mM Tris-HCl [pH 7.0], 500 mM KCl, 10 mM imidazole, 10% [vol/vol] glycerol, 0.2% [vol/vol] β-mercaptoethanol, 0.03% [vol/vol] lauryldimethylamine oxide) for purification on nickel-nitrilotriacetic acid agarose, both containing 1× complete EDTA-free protease inhibitors, and frozen overnight at −20°C.

Purification of recombinant protein.

Culture suspensions were thawed on ice, Triton X-100 was added to a concentration of 1% (vol/vol) (for mtPBS suspensions only), and the preparations were incubated on ice for an additional 30 min. DNase I and RNase A were then added at concentrations of 15 and 20 μg/ml, respectively, and cultures were periodically mixed on ice for 30 min. Suspensions were sonicated (as described above), and insoluble material was removed by centrifugation at 30,000 × g for 20 min at 4°C. Supernatants were filtered through 0.45-μm acrodisc filters (Pall Corporation) onto 250 μl glutathione-agarose (Sigma) or nickel-nitrilotriacetic acid agarose (QIAGEN) beads and rotated on a rocking platform for 4 to 16 h at 4°C. The beads were captured by passing the supernatants through Econocolumns (Bio-Rad), and the resin beds were washed twice with 7 ml mtPBS containing 1% (vol/vol) Triton X-100 or nickel A buffer and then washed once in buffer without detergent. Recombinant protein was eluted with 7 ml of 10 mM reduced glutathione in 50 mM Tris-HCl (pH 9.5) or with nickel B buffer (20 mM Tris-HCl [pH 7.0], 500 mM KCl, 300 mM imidazole, 10% [vol/vol] glycerol, 0.2% [vol/vol] β-mercaptoethanol, 0.03% [vol/vol] lauryldimethylamine oxide) and dialyzed twice into 1.5 liters mtPBS containing 0.5 mM dithiothreitol at 4°C. Dialyzed protein was concentrated to 300 μl using Amicon Ultra centrifugation filters and quantified by comparison to bovine serum albumin following SDS-polyacrylamide gel electrophoresis (PAGE) and Coomassie blue staining. The absolute amounts of soluble recombinant proteins obtained from 750-ml E. coli cultures ranged from 20 to 500 μg. Kinase activity was retained for at least 1 month by diluting the concentrated protein solutions 1:1 with 100 mM HEPES (pH 7.4) containing 50% (vol/vol) glycerol and 1 mM dithiothreitol and storing them at 4°C.

In vitro kinase assays.

Aliquots containing 50 to 200 ng of glutathione S-transferase (GST)-PknD KD or GST-PknD KD point mutants (see above) were incubated for 2.5 h at 34°C in 20-μl reaction mixtures containing 25 mM HEPES (pH 7.3), 20 μM ATP, 5 mM MnCl₂, 1× complete EDTA-free protease inhibitors, 20 mM β-glycerophosphate, and 10 μCi [γ-³²P]ATP. The substrates were tested using 1 μg and included His-Cpn0712, His-FHA-2, GST-FHA-1, GST-FHA-2, GST-FHA-2[S441A], and GST-FHA-2[N464A]. Kinase assays were terminated with 5 μl of 5× SDS loading buffer, and the mixtures were heated to 95°C for 5 min prior to SDS-PAGE and blotting onto a polyvinylidene difluoride (PVDF) membrane. Membranes were exposed to Kodak X-OMAT(XR) film for 30 min to 2 h at −80°C and developed to visualize phosphorylated proteins. ECL using mouse monoclonal antibody to GST (Sigma) or His₆ (Amersham) at a 1:10,000 dilution followed by HRP-conjugated goat anti-mouse IgG at a 1:4,000 dilution was used to visualize the protein load.

Phosphoamino acid analysis.

Thin-layer chromatography (TLC) was used to identify phosphoamino acids on PknD KD and on Cpn0712 FHA-2. Briefly, autophosphorylated GST-PknD KD and phosphorylated His-FHA-2 from in vitro kinase assays were resolved by SDS-PAGE and blotted onto PVDF membranes. Proteins were visualized by autoradiography and excised from PVDF membranes, and ascending two-dimensional TLC was carried out essentially as described previously (10). Briefly, phosphorylated proteins were acid hydrolyzed and spotted onto glass-backed cellulose plates along with the phosphoamino acid standards P-Ser, P-Thr, and P-Tyr. The liquid phase used in the first dimension was ethanol-glacial acetic acid-double-distilled H₂O (1:1:1), and TLC was performed for 3.5 h; the liquid phase used in the second dimension was isobutanol-formic acid-double-distilled H₂O (8:3:4), and TLC was performed for 2 h. Plates were sprayed with ninhydrin to locate phosphoamino acid standards and exposed to Kodax X-OMAT(XR) film at −80°C, and the film was developed to identify phosphorylated amino acids.

RESULTS AND DISCUSSION

PknD is an integral membrane protein and the kinase domain is located in the cytoplasm.

Differential centrifugation of E. coli lysates containing His-PknD followed by sodium carbonate and sodium hydroxide extraction of the membrane fraction was used in order to determine the localization of recombinant PknD in E. coli. Sodium carbonate and sodium hydroxide liberate proteins peripherally associated with the membrane or trapped in microvesicles, while integral membrane proteins are retained in the lipid bilayer. Western blotting of the 100,000-×-g E. coli membrane fraction followed by ECL analysis yielded a faint band representing His-PknD (Fig. 1A, lane 1). PknD was retained and enriched in the membrane fraction following extraction with either sodium carbonate (Fig. 1A, lane 3) or sodium hydroxide (data not shown), indicating that PknD is an integral membrane protein when it is expressed in E. coli.

FIG. 1.

PknD is an integral membrane protein with the N-terminal kinase domain located in the cytoplasm and the C-terminal domain located in the periplasm. (A) Enrichment of PknD in E. coli membranes treated with sodium carbonate indicates that PknD is an integral membrane protein. Recombinant PknD (His-PknD) was expressed in E. coli, and differential centrifugation followed by Coomassie blue staining (top) and ECL analysis (bottom) was used to track partitioning of PknD between soluble (lane 2) and membrane (lane 3) fractions after extraction of the 100,000 × g E. coli membrane pellet (lane 1) with sodium carbonate. (B) Seven overlapping fragments of PknD (f1 to f7) were fused upstream of AP and expressed in E. coli, and phosphatase assays were used to determine the location of the transmembrane and kinase domains. The first and last amino acids of each fragment are indicated. (C) ECL detection and phosphatase activity of the PknD-AP fusion proteins produced in E. coli. The AP activity of each fusion protein is indicated below the Western blot. Fusions with fragments f1, f2, and f5 displayed basal levels of AP activity (0 to 1 U) (data not shown), as did fusions with fragments f3 and f6. Fusions with fragments f4 and f7 exhibited high levels of AP activity (1,035 and 45 U), indicating localization of AP in the periplasm and therefore demonstrating that amino acids R608 to I633 of PknD encompass a transmembrane domain with an inside-outside orientation, placing the kinase domain in the cytoplasm.

In order to elucidate the topology of PknD in the E. coli membrane, overlapping fragments of PknD (f1 to f7) (Fig. 1B) were cloned in frame and upstream of AP, expressed in Rosetta(pLysS), and assayed for phosphatase activity (Fig. 1C). The presence of a membrane-spanning domain or signal sequence within the cloned upstream fragment results in periplasmic localization of AP and high phosphatase activity (20). Of the seven PknD-AP hybrid proteins examined (p9Pf1 to p9Pf7), only two (p9Pf4 and p9Pf7) exhibited high phosphatase activity (Fig. 1C). The activity of protein p9Pf4 was 1,000-fold higher than the activity of p9Pf3 (1,035 versus 1 U), and the activity of protein p9Pf7 was 45 U, whereas p9Pf6 was inactive. The PknD peptide fragments of p9Pf4 and p9Pf7 are therefore capable of localizing AP to the periplasm and share amino acids L609 to I633, a region predicted by the in silico program TMpred to encompass a transmembrane domain (L609 to T626). Therefore, the high phosphatase activity of these fusion proteins confirms that the region of PknD from L609 to I633 encompasses a transmembrane domain and that it has an inside-outside orientation. The remaining fusion proteins (p9Pf1, p9Pf2, and p9Pf5) were inactive, and p9Pf1 and p9Pf2 were not detected by Western blotting (data not shown). Cytoplasmic AP fusion proteins are frequently unstable (27); thus, the inability to detect p9Pf1 and p9Pf2 supports the hypothesis that they are cytoplasmic and that PknD peptide fragments f1 and f2 lack export sequences and transmembrane domains. Together, the data support the hypothesis that there is a single transmembrane domain at L609 to T626 with the N-terminal kinase domain of PknD located in the cytoplasm of E. coli.

PknD KD exhibits protein kinase activity and single amino acid substitutions abrogate catalytic activity.

Annotation of PknD from C. pneumoniae strain CWL029 places it in the eukaryote-type protein kinase superfamily based on homology to Hanks domains (13), 12 regions containing variable lengths of highly conserved amino acids that are essential for kinase activity and together form the kinase domain (Fig. 2B). The kinase domain of PknD was cloned as a GST fusion protein (GST-PknD KD) to facilitate production of soluble protein that could be purified and characterized. Incubation of purified GST-PknD KD with [γ-³²P]ATP in an in vitro kinase assay led to the appearance of a phosphorylated protein that was detectable by anti-GST immunoblotting (Fig. 3A and B, lane 1). Therefore, PknD KD is catalytically active and autophosphorylation occurs in the absence of the full-length molecule.

FIG. 2.

(A) Domain organization of PknD and Cpn0712. PknD contains a N-terminal kinase domain and a predicted transmembrane domain (TMD). Cpn0712 is a YscD homolog (E value, 1.2e⁻⁴³) with a predicted transmembrane domain, a phospholipid binding domain (BON), and two FHA domains. Amino acid numbers are indicated. (B) Sequence alignment of the Hanks domains of PknD of C. pneumoniae (PknD Ch. pn.) and C. trachomatis (PknD C. tra.), human ERK2, and bovine cAPK-α. Prototype Hanks domains are based on cAPK-α, and canonical residues are indicated by bold type.

FIG. 3.

Autophosphorylation of PknD is dependent on specific ATP/cation binding and activation loop amino acid residues and is optimal with manganese cations. (A) Fifty to 200 ng of purified GST-PknD KD (lane 1) and mutants GST-PknD KD[K33G] (lane 2), GST-PknD KD[D156G] (lane 3), and GST-PknD KD[K33G; D156G] (lane 4) was incubated with 10 μCi [γ-³²P]ATP in an in vitro kinase assay. Autophosphorylation of PknD KD was detected using autoradiography (top), and immunoblotting with anti-GST antibody revealed the protein load (bottom). Substitution of glycine for lysine (K33G), aspartate (D156G), or both (K33G and D156G) abolished kinase activity, thus confirming that PknD is a Hanks-type protein kinase. (B) One hundred to 200 ng of purified GST-PknD KD (lane 1), GST-PknD KD[T185A] (lane 2), or GST-PknD KD[T193A] (lane 3) was used in an in vitro kinase assay as described above for panel A. Replacing activation loop threonines T185 and T193 with alanine abolished PknD kinase activity (top). Immunoblotting with anti-GST antibody revealed the protein load (bottom). (C) PknD kinase activity is stimulated in the presence of manganese. GST-PknD KD was incubated with increasing concentrations (1, 5, and 20 mM) of the divalent cations magnesium, manganese, and calcium. Autophosphorylation occurred to a greater extent in the presence of manganese than in the presence of magnesium or calcium (top). Optimal autophosphorylation was observed with 5 mM manganese. ECL was used to reveal the protein load (bottom).

Divalent cations are required for protein kinase activity, but their influence on activity varies with the concentration and the type of ion used. We therefore supplemented the kinase reaction buffers with 1, 5, or 20 mM magnesium, manganese, or calcium in order to determine optimal conditions for GST-PknD KD kinase activity. GST-PknD KD exhibited the highest activity in the presence of 5 mM manganese (Fig. 3C). Manganese was therefore used as a cofactor in all subsequent kinase assays.

Invariant amino acids found within Hanks domains of protein kinases are typically essential for catalytic activity. A critical lysine in Hanks domain II binds ATP through a salt bridge, and an aspartate residue in Hanks domain VII coordinates the cationic cofactor (18); mutation of either residue destroys kinase activity. In order to investigate the importance of conserved Hanks residues for PknD enzymatic activity, we constructed K33G (with a mutation located in Hanks domain II) (Fig. 2B) and D156G (with a mutation located in Hanks domain VII) point mutants, and as well as a K33G-D156G mutant, in GST-PknD KD and assayed for kinase activity. Point mutants GST-PknD KD[K33G] and GST-PknD KD[D156G] and double mutant GST-PknD KD[K33G; D156G] were incapable of ³²P incorporation (Fig. 3A), indicating that K33 and D156 are essential for activity and demonstrating that PknD is a Hanks-type protein kinase.

A region of amino acids located between Hanks domains VII and VIII is referred to as the activation loop and contains a variable number of threonines that are phosphorylated in order to promote conformational changes that influence kinase activity (1). In order to determine if activation loop threonines are essential for PknD catalytic activity, we replaced single threonine residues with alanine in GST-PknD KD and assayed for kinase activity. Both GST-PknD KD[T185A] and GST-PknD KD[T193A] were incapable of autophosphorylation (Fig. 3B), demonstrating that both threonines in the PknD activation loop are required for activation of PknD.

YscD homolog Cpn0712 is phosphorylated by PknD.

Cpn0712, a YscD homolog (1.2e⁻⁴³) (www.tigr.org) that contains two FHA domains (Fig. 2A), was cloned and expressed as an N-terminal histidine-tagged protein in E. coli. His-Cpn0712 localized to the 100,000-×-g membrane fraction and was found to be an integral membrane protein after treatment with sodium carbonate and sodium hydroxide (data not shown). In order to determine if Cpn0712 could be phosphorylated by PknD, E. coli membranes enriched with His-Cpn0712 were incubated with and without GST-PknD KD in an in vitro kinase assay. His-Cpn0712 was phosphorylated in the presence, but not in the absence, of GST-PknD KD (Fig. 4A), indicating that Cpn0712 is a substrate of PknD.

FIG. 4.

PknD phosphorylates Cpn0712 and the FHA-1 and FHA-2 domains of Cpn0712. (A) GST-PknD KD was incubated with and without His-Cpn0712 in an in vitro kinase assay. Phosphorylation of Cpn0712 by GST-PknD KD was visualized by autoradiography (top). ECL using anti-His antibody was used to reveal the His-Cpn0712 protein load (bottom). (B) GST-FHA-1 was incubated without (lane 1) and with (lane 2) GST-PknD KD in an in vitro kinase assay. Phosphorylation of FHA-1 by GST-PknD KD was visualized by autoradiography (top). ECL using anti-GST antibody was used to show that there were equivalent amounts of GST-FHA-1 in the lanes. (C) GST-FHA-2 was incubated alone (lane 1), with GST-PknD KD (lane 2), with GST-PknD KD[T185A] (lane 3), or with GST-PknD KD[T193A] (lane 4) and incubated in an in vitro kinase assay mixture. Phosphorylation of FHA-2 by GST-PknD KD was visualized by autoradiography (top). ECL using anti-GST antibody was used to show that there were equivalent amounts of protein in the lanes. An asterisk indicates the location of GST-PknD KD in panels A and B and the location of the kinase and mutants in panel C.

FHA-1 and FHA-2 domains of C. pneumoniae Cpn0712 are phosphorylated by PknD.

In order to determine if PknD phosphorylates the FHA domains of Cpn0712, we cloned the FHA-1 and FHA-2 domains of Cpn0712 as GST fusion proteins and tested them as substrates for GST-PknD KD in an in vitro kinase assay. Both FHA-1 and FHA-2 were phosphorylated in the presence, but not in the absence, of GST-PknD KD (Fig. 4B and C), indicating that FHA-1 and FHA-2 are substrates of PknD. FHA-2 without a GST tag (His-FHA-2) was also phosphorylated by PknD (data not shown), demonstrating that the targets of phosphorylation are the FHA domains and not the GST moieties. In order to determine if PknD activation loop mutants could phosphorylate FHA-2, GST-FHA-2 was incubated with GST-PknD KD[T185A] and GST-PknD KD[T193A] in an in vitro kinase assay. Both point mutants were incapable of phosphorylating FHA-2 (Fig. 4C), indicating that the activation loop threonines are required for substrate phosphorylation. Interestingly, Mycobacterium tuberculosis STPK substrates containing FHA domains have been shown to utilize conserved amino acid residues within the FHA domains to mediate binding to, and phosphorylation by, kinase domains (23). Molle et al. recently demonstrated that R312, S326, and N348 within the FHA domain of EmbR are required for phosphorylation of EmbR by PknH (23). In order to determine if the Cpn0712 FHA-2 domain utilized similar conserved residues to mediate phosphorylation by PknD, we created point mutants GST-FHA-2[S441A] and GST-FHA-2[N464A] and tested them as substrates for PknD. The GST-FHA-2 point mutants were phosphorylated by GST-PknD KD to the same level as GST-FHA-2 (data not shown), suggesting that FHA-2 conserved amino acids S441 and N464 play only a minor role, if any, in mediating the interaction with PknD. Alternatively, other conserved or nonconserved amino acids within the FHA-2 domain may be important in mediating interactions with PknD; it has recently been demonstrated that the ligand specificity of the FHA domains of Rad53 is determined by nonconserved amino acid residues (35). Despite our inability to indicate a role for FHA-2 S441 or N464 in binding to PknD, our results clearly show that Cpn0712 is indeed a substrate of PknD.

PknD phosphorylates serine/threonine and tyrosine residues.

Two-dimensional TLC was used to identify the amino acids phosphorylated by PknD KD. Autophosphorylated GST-PknD KD contained phosphothreonine and phosphotyrosine (Fig. 5A). PknD clearly phosphorylated serine and tyrosine residues on His-FHA-2 (Fig. 5B) (the migration of phosphoamino acid standards is shown in Fig. 5C). Attempts to corroborate the tyrosine specificity of PknD by immunoblotting with antiphosphotyrosine monoclonal antibodies 4G10 and Y-100 were not successful. ECL is not as sensitive as radioactivity and, given that the FHA-2 domain has only two tyrosine residues, may not be able to detect low levels of phosphotyrosine. Alternatively, amino acids flanking a phosphotyrosine(s) on GST-PknD KD and His-FHA-2 may inhibit antibody binding.

FIG. 5.

PknD phosphorylates serine/threonine and tyrosine residues. (A) Autophosphorylated GST-PknD KD was acid hydrolyzed, and two-dimensional TLC was used to separate phosphoamino acids. Autoradiography revealed that PknD KD autophosphorylated predominantly on threonine (³²P-thr) and also on tyrosine (³²P-tyr). The oval indicates the location of the phosphoserine (P-ser) standard. (B) His-FHA-2 phosphorylated by GST-PknD KD was acid hydrolyzed, and two-dimensional TLC was used to separate phosphoamino acids. Autoradiography revealed phosphorylation on serine (³²P-ser) and tyrosine residues. The oval indicates the location of the phosphothreonine (P-thr) standard. (C) Two-dimensional TLC separation of the phosphoamino acid standards phosphoserine (P-ser), phosphothreonine (P-thr), and phosphotyrosine (P-tyr) as visualized with ninhydrin staining. The results are representative results from three independent experiments.

PknD autophosphorylates on threonine and tyrosine residues, suggesting that PknD may represent a unique class of bacterial STPKs that exhibit dual amino acid specificity. Interestingly, kinases with dual specificity have already been identified in plants (25), and multiple lines of evidence point to the presence of kinases with dual specificity in bacteria. Streptomyces griseus contains a single-stranded DNA binding protein that is tyrosine phosphorylated (21), but current methods of genome sequence annotation have not identified tyrosine kinases in the streptomycetes, suggesting the presence of unconventional kinase activity, perhaps catalyzed by STPKs, in these bacteria. Additionally, the DivL protein from Caulobacter crescentus is a predicted histidine kinase that displays unprecedented kinase activity by autophosphorylating on tyrosine (34). There are also examples of STPKs with dual specificity in eukaryotes. Recombinant ERK2 purified from E. coli lysates autophosphorylates on threonine and tyrosine residues (29), and insulin-stimulated ERK1 from eukaryotic cell cultures autophosphorylates on serine, threonine, and tyrosine (26). In Chlamydia, Verma and Maurelli did not detect phosphorylated tyrosine residues on PknD from C. trachomatis serovar L2 using one-dimensional separation of phosphoamino acids, nor did they determine the localization of PknD (32). Autophosphorylation may be different for PknD in vivo since this protein is anchored in the membrane, and we do not know if conformational limitations affect which residues are phosphorylated. Classical biochemical approaches are required to determine the location of specific residues on PknD that are phosphorylated in C. pneumoniae.

In summary, our data indicate that C. pneumoniae PknD utilizes a single transmembrane domain to anchor itself in the bacterial inner membrane and positions the N-terminal kinase domain in the cytoplasm. The FHA domain-containing protein Cpn0712, a type III secretion YscD homolog, is phosphorylated in vitro by the PknD kinase domain. PknD autophosphorylated on threonine and tyrosine and phosphorylated the FHA-2 domain of Cpn0712 on serine and tyrosine; to our knowledge, this is the first example of a prokaryotic STPK that exhibits dual amino acid specificity. The biological significance of tyrosine phosphorylation by PknD of C. pneumoniae is currently unknown, but it is noteworthy that infection of host cells with Chlamydia results in tyrosine phosphorylation of host proteins (3, 8, 33) despite the absence of predicted tyrosine kinases in Chlamydia. Additionally, Chlamydia secretes the actin-recruiting protein Tarp into the host cytosol, and Tarp is believed to be phosphorylated by a host cell kinase (5, 6). Tarp, PknD, and Cpn0712 are the only three chlamydial proteins that have been shown to be phosphorylated on tyrosine.