The Pleckstrin Homology Domain of Protein Kinase D Interacts Preferentially with the η Isoform of Protein Kinase C*

The results presented here demonstrate that protein kinase D (PKD) and PKCη transiently coexpressed in COS-7 cells form complexes that can be immunoprecipitated from cell lysates using specific antisera to PKD or PKCη. The presence of PKCη in PKD immune complexes was initially detected by in vitro kinase assays which reveal the presence of an 80-kDa phosphorylated band in addition to the 110-kDa band corresponding to autophosphorylated PKD. The association between PKD and PKCη was further verified by Western blot analysis and peptide phosphorylation assays that exploited the distinct substrate specificity between PKCs and PKD. By the same criteria, PKD formed complexes only very weakly with PKCε, and did not bind PKCζ. When PKCη was coexpressed with PKD mutants containing either complete or partial deletions of the PH domain, both PKCη immunoreactivity and PKC activity in PKD immunoprecipitates were sharply reduced. In contrast, deletion of an amino-terminal portion of the molecule, either cysteine-rich region, or the entire cysteine-rich domain did not interfere with the association of PKD with PKCη. Furthermore, a glutathione S-transferase-PKDPH fusion protein bound preferentially to PKCη. These results indicate that the PKD PH domain can discriminate between closely related structures of a single enzyme family, e.g. novel PKCs ε and η, thereby revealing a previously undetected degree of specificity among protein-protein interactions mediated by PH domains.

PKD with PKC in COS-7 cells leads to the formation of a stable PKD⅐PKC complex. Strikingly, we found very little evidence of complex formation between PKD and the PKC⑀ isoform despite its close similarity to PKC, and no evidence for a stable interaction between PKD and PKC. Our results also demonstrate that the PH domain is critical for stable PKD⅐ PKC complex formation, thus indicating that these domains can mediate highly selective protein-protein interactions.

EXPERIMENTAL PROCEDURES
Cell Culture and Transfections-COS-7 cells were maintained by subculture in 10-cm tissue culture plates, every 3-4 days in Dulbecco's modified Eagle's medium supplemented with 10% fetal bovine serum at 37°C in a humidified atmosphere containing 10% CO 2 . For experimental dishes, confluent cells were subcultured at a density of 6 ϫ 10 4 cells/ml in 6-or 10-cm dishes on the day prior to transfections. All transfections and co-transfections were carried out with equivalent amounts of DNA (6 g/6-cm dish, 12 g/10-cm dish), using vector pcDNA3 as the control DNA added to single transfections. Transfections were carried out in Opti-MEM (Life Technologies, Inc.) using Lipofectin (Life Technologies, Inc.) at 10 l/6-cm dish or 20 l/10-cm dish, added to cells in a final volume of 2.5 ml/6-cm dish or 5 ml/10-cm dish, following formation of DNA-Lipofectin complexes according to the protocol provided by the manufacturer. Cells were allowed to take up complexes in the absence of fetal bovine serum for 5-6 h or overnight, then fetal bovine serum (10% final concentration) in Opti-MEM was added to the dishes to yield a final volume of 5 ml/6-cm dish or 10 ml/10-cm dish. Cells were used for experiments after a further 48 -72 h of incubation.
Preparation of PKD PH Domain Fusion Protein-The cDNA sequence spanning the entire PKD PH domain (aa 418 -567) was amplified by PCR from wild-type PKD using specific oligonucleotide primers (forward primer, 5Ј-GATGGATCCGTGAAGCACACGAAGCGGAGG-3Ј; reverse primer, 5Ј-GCGGAATTCAGAAATATCTTTGTGTGAGTTGGA-3Ј) containing restriction sites for BamHI and EcoRI, respectively (underlined). The resulting PCR product was subcloned as a BamHI-EcoRI fragment into the vector pGEX4T3 (Pharmacia Biotech Inc.) to generate the bacterial expression construct pGEX-GST-PKDPH. The 42-kDa GST-PKD PH domain fusion protein (GST-PKDPH) was expressed in Escherichia coli, purified on glutathione-agarose beads, eluted with 25 mM reduced glutathione, dialyzed against phosphate-buffered saline, and stored at Ϫ20°C in 40% glycerol. Purity and concentration of the recombinant protein were assessed by SDS-PAGE and Coomassie Brilliant Blue staining.
Assays of PKC⅐GST-PKDPH Fusion Protein Binding in Vitro-COS-7 cells transiently transfected with the different PKC isoforms were lysed 72 h after transfection by removal of growth medium from cells on ice and addition of lysis buffer (50 mM Tris-HCl, pH 7.6, 2 mM EGTA, 2 mM EDTA, 1 mM dithiothreitol, 10 g/ml aprotinin, 100 g/ml leupeptin, 1 mM AEBSF (Pefabloc), and 1% Triton X-100), and the resulting extracts were combined with either GST (control) or GST-PKDPH fusion proteins preadsorbed onto glutathione-agarose beads. After 2 h at 4°C, the complexes were washed 8 times with lysis buffer, and bound proteins were extracted with SDS-PAGE sample buffer and subjected to SDS-PAGE and Western analysis using isoform-specific antisera to detect associated PKC, as described in figure legends.
Immunoprecipitations-COS-7 cells transfected with wild-type or mutant PKD or co-transfected together with different PKC isoforms were lysed as described above. Small amounts (typically 1/10) of these total lysates were saved and combined with equal volumes of SDS-PAGE sample buffer (1 M Tris-HCl, pH 6.8, 6% SDS, 0.5 M EDTA, 4% 2-mercaptoethanol, 10% glycerol) for Western blot analysis. PKD was immunoprecipitated at 4°C for 3 h with either the PA-1 antiserum (1:100 dilution) raised against the synthetic peptide EEREMKALS-ERVSIL that corresponds to the predicted COOH-terminal region of PKD, as described previously (12), or a 1:200 dilution of a commercial antiserum (PKD C-20, Santa Cruz Biotechnologies), which also recognizes the COOH-terminal region of PKD. PKCs were immunoprecipitated using respective PKC antisera at 1:100 dilution. Immune complexes were recovered using protein A coupled to agarose.
In Vitro Kinase Assays-Immune complexes were washed twice with lysis buffer, then twice with kinase buffer consisting of 30 mM Tris-HCl, pH 7.4, 10 mM MgCl 2 , 1 mM dithiothreitol. Autophosphorylation reactions were initiated by combining 20 l of immune complexes with 10 l of a phosphorylation mixture containing [␥-32 P]ATP (3 Ci/reaction diluted with cold ATP to give a final concentration of 100 M) in kinase buffer. Reactions were transferred to a water bath at 30°C for 10 min, then terminated by addition of 1 ml of ice-cold kinase buffer and removed to an ice bucket. Immune complexes were recovered by centrifugation, and the proteins were extracted for SDS-PAGE analysis by addition of SDS-PAGE sample buffer. Dried SDS-PAGE gels were subjected to autoradiography to visualize radiolabeled protein bands.
For assays of exogenous substrate phosphorylation, immune complexes were processed as for autophosphorylation reactions, then substrates (either syntide-2 or ⑀-peptide at final concentrations 2.5 or 1.75 mg/ml, respectively) were added in the presence of [␥-32 P]ATP (2 Ci/ reaction diluted with cold ATP to give a final concentration of 100 M) in kinase buffer (final reaction volume, 30 l), and transferred to a water bath at 30°C for 10 min. Reactions were terminated by adding 100 l of 75 mM H 3 PO 4 , and 75 l of the mixed supernatant was spotted to Whatman P-81 phosphocellulose paper. Papers were washed thoroughly in 75 mM H 3 PO 4 , dried, and radioactivity incorporated into peptides was determined by detection of Cerenkov radiation in a scintillation counter.
Western Blot Analysis-For Western blot analysis, immune complexes and proteins associated with glutathione-agarose/GST fusion protein complexes were washed as for in vitro kinase reactions, then extracted by boiling in SDS-PAGE sample buffer. Samples of cell lysates were directly solubilized by boiling in SDS-PAGE sample buffer. Following SDS-PAGE on 8% gels, proteins were transferred to Immobilon-P membranes (Millipore), as described previously (21) and blocked by overnight incubation with 5% non-fat dried milk in phosphate-buffered saline, pH 7.2. Membranes were incubated at room temperature for 2 h with antisera specifically recognizing either PKD or the different PKC isoforms, at a dilution of 1 g/ml, in phosphatebuffered saline containing 3% non-fat dried milk. Immunoreactive bands were visualized using either horseradish peroxidase-conjugated anti-rabbit IgG and subsequent enhanced chemiluminescence detection or 125 I-labeled protein A followed by autoradiography.
Materials-[␥-32 P]ATP (6000 Ci/mmol) was from Amersham International (United Kingdom). Protein A-agarose and AEBSF (Pefabloc) were from Boehringer Mannheim (UK). Antisera (PKD C-20, PKC C-20, PKC⑀ C-15, PKC C-15, and PKC␣ C-15) used in Western blot analysis were from Santa Cruz Biotechnologies, Palo Alto, CA. PKC standard proteins were from Calbiochem. Opti-MEM and Lipofectin were from Life Technologies, Inc. Glutathione-Sepharose was from Pharmacia Biotech. Syntide-2 peptide and immunizing peptide EE-REMKALSERVSIL corresponding to the PKD COOH terminus were synthesized at the Imperial Cancer Research Fund. ⑀-Peptide was from Alexis Biochemicals. All other reagents were from standard suppliers or as described in the text and were the highest grade commercially available.

An 80-kDa Phosphoprotein Associates with PKD in Cells
Coexpressing PKD and PKC-COS-7 cells transiently transfected with pcDNA3-PKD either alone or together with PKC⑀, PKC, or PKC expression constructs encoding constitutively active PKC mutants were incubated with or without 200 nM PDBu for 10 min and lysed. PKD was immunoprecipitated from the extracts and the resulting immunoprecipitates were sub-jected to in vitro kinase assays, followed by SDS-PAGE analysis and autoradiography to detect the prominent 110-kDa band corresponding to autophosphorylated PKD.
In agreement with previous results (12, 19 -21), PKD isolated from cells transfected with PKD alone displayed low catalytic activity and PDBu treatment of these cells resulted in isolation of persistently activated PKD. Furthermore, PKD isolated from cells co-transfected with PKD together with active mutant PKCs ⑀ or was in persistently activated form in the absence of PDBu treatments of these cells (Fig. 1A). Strikingly, when PKD immunoprecipitates from PKD/PKC cotransfected cells were assayed, an additional phosphorylated band corresponding to an apparent molecular mass of approximately 80 kDa was also apparent in the autoradiograms (Fig.  1A). This band did not appear when PKD was coexpressed with the active PKC ⑀ or mutant forms despite their level of overexpression shown by Western blot analysis (Fig. 2), and despite the fact that the PKC⑀ mutant induced PKD activation to essentially the same degree as did the PKC mutant (Fig.  1A). This latter result also indicates that the presence in PKD immunoprecipitates of the phosphoprotein giving rise to the 80-kDa band resulting from coexpression with PKC is not uniquely required for the persistent activation of PKD. Expression of PKC on its own followed by immunoprecipitation with the PA-1 antiserum and in vitro kinase assays did not result in the appearance of the 80-kDa band (Fig. 1A). Thus, the immunoprecipitation and subsequent labeling of the 80-kDa band required PKD.
Initially, we considered whether this band might be the result of either nonspecific immunoprecipitation by the PA-1 antiserum or long-term overexpression of the active mutant PKC. To test this, we used a commercially available antiserum (rather than PA-1) to perform PKD immunoprecipitations from cells coexpressing PKD together with the wild-type PKC, followed by in vitro kinase assays. As shown in Fig. 1B, this assay also led to the isolation of immune complexes containing 110-and 80-kDa phosphoproteins. Thus, it was possible to substitute either the primary antibody used in the immunoprecipitation or the wild-type for the active mutant PKC and still retain both bands. In addition, both autophosphorylated PKD and the coprecipitated 80-kDa phosphoprotein band were eliminated when the immunoprecipitation reactions were carried out in the presence of the immunizing peptide used to generate the PA-1 antiserum (Fig. 1B).
To test whether the 80-kDa band represented a proteolytic fragment of autophosphorylated PKD, we co-transfected an expression construct, pcDNA3-PKD/K618M, which encodes a kinase-deficient mutant of PKD, together with PKC, and performed immunoprecipitations with the PA-1 antiserum followed by in vitro kinase assays. As shown in Fig. 1C, this assay again resulted in the appearance of the 80-kDa band. Importantly, whereas the intensity of this band was similar to that seen when the wild-type PKD was used for co-transfection, the intensity of the 110-kDa band was drastically reduced in comparison with that generated by the co-transfection with wild- type PKD 2 (Fig. 1C, left). These data also indicate that the kinase activity of PKD is not required for the association with the 80-kDa phosphoprotein.
PKC Specifically Associates with PKD-Previous studies indicated that ⑀-peptide, a peptide based on the pseudosubstrate domain of PKC⑀, is a substrate for all PKCs (24), but is a poor substrate for PKD (11,12,15). In agreement with these studies, anti-PKD immunoprecipitates from lysates of cells transfected with PKD (or the kinase-deficient mutant, PKD/ K618M) did not phosphorylate ⑀-peptide, even when PKD was activated within cells by PDBu treatments (Fig. 1C, right). Surprisingly, PKD immunoprecipitates from lysates of cells co-transfected with PKD (either wild-type or PKD/K618M) and PKC (either constitutive active or wild-type) contained activity which strongly phosphorylated ⑀-peptide. Since this activity was associated with both catalytically active and inactive forms of PKD, and PKD does not phosphorylate the peptide, we conclude that the activity was due to an associated protein kinase (e.g. contributed by the 80-kDa phosphoprotein) whose activity did not require the catalytic activity of PKD.
Since PKC has an apparent molecular mass of approximately 80 kDa in SDS-PAGE, and since the 80-kDa phosphoprotein had appeared only when PKD was coexpressed together with mutant or wild-type PKC, it seemed plausible that coexpression of PKD and PKC resulted in the formation of a PKD⅐PKC complex that persists during immunoprecipitation of PKD. To test this hypothesis directly, we used Western blot analysis to examine whether PKC was present in PKD immunoprecipitates. In view of the results indicating that the 80-kDa band is not detected in PKD immunoprecipitates from lysates of cells co-transfected with PKD together with either PKC or PKC⑀ (Fig. 1A), we also performed Western blot analysis to assess whether these isoforms were present in PKD immunoprecipitates. Since PKC␣ is abundantly expressed in COS-7 cells, we also tested for the presence of this isoform in PKD immunoprecipitates. Lysates of cells either transfected with PKD or co-transfected with PKD and PKCs ⑀, , or were either examined directly by Western blot analysis using polyclonal antibodies that specifically recognize PKCs ␣, , ⑀, or or subjected to immunoprecipitation reactions with PA-1 antiserum followed by Western analysis. As shown in Fig. 2, D and F, anti-PKC immunoblotting of PKD immunoprecipitates indicated that both wild-type and mutant PKC associated with PKD. Similar analysis did not detect the presence of PKCs ␣, , or ⑀ (either wild-type or constitutively active) in PKD immunoprecipitates from the corresponding cell lysates (Fig. 2, A-C and E). However, longer exposure of the autoradiograms did reveal a faint band of anti-PKC⑀ immunoreactivity (not shown), indicating that PKC⑀ may also associate with PKD but to a much lesser degree.
Although the experiments in Figs. 1 and 2 indicate that PKC was present in PKD immunoprecipitates from cells cotransfected with these two proteins, these results were obtained using antibodies directed against PKD. To substantiate further the existence of a PKD-PKC interaction, we also performed immunoprecipitations using an antibody directed against the opposite partner in the association, i.e. PKC, followed by Western analysis to detect the interacting PKD protein. As illustrated in Fig. 2G, in vitro kinase assays revealed the presence of an autophosphorylated band in the position corresponding to PKD (110 kDa) that depended on the presence of co-transfected PKD. Furthermore, Western blot analysis revealed the presence of immunoreactive PKD when both proteins were co-transfected. Thus, the results shown in Fig. 2G demonstrate the presence of PKD in PKC immunoprecipitates from lysates of cells co-transfected with PKD and PKC.
Phosphorylation of PKC and PKD Substrates by PKD Immunoprecipitates-To further address the specificity of the PKD-PKC interaction, we measured phosphorylation of exogenous substrates by PKD immunoprecipitates from cells expressing each PKD-PKC combination. In initial experiments, we examined phosphorylation of ⑀-peptide by endogenously expressed and transfected PKC isoforms in anti-PKC immunoprecipi- tates. Results shown in Fig. 3A demonstrate that PKC or PKC⑀ immunoprecipitates from mock-transfected COS-7 cells contained low levels of ⑀-peptide phosphorylation activity that was dramatically increased by transfection of these isoforms. PKC immunoprecipitates from mock-transfected cells contained similar activity to the immunoprecipitates from PKCor PKC⑀-transfected cells. These results demonstrate the ability of each PKC tested to phosphorylate ⑀-peptide.
For assays of PKD immunoprecipitates we used both ⑀-peptide, a poor substrate for PKD (Fig. 1), and syntide-2 (25,26), a synthetic peptide previously demonstrated to be an excellent substrate for PKD (11), thereby exploiting the distinct substrate specificity of PKCs and PKD. In this way, we measured both PKD activation and the retention of PKC activity bound to PKD in the same immune complexes. In agreement with pre-vious results, when the PA-1 antiserum was used to immunoprecipitate PKD from cells overexpressing only PKD, syntide-2 assays revealed a low basal activity that was dramatically increased upon PDBu stimulation of cells (Fig. 3B, upper  graph). Again, these immunoprecipitates did not phosphorylate ⑀-peptide to any significant extent, even after the cells had been stimulated with PDBu (Fig. 3B, lower graph). In contrast with these results, when PKD was immunoprecipitated from cells co-transfected with PKD and either mutant or wild-type PKC, both syntide-2 and ⑀-peptide were strongly phosphorylated, indicating the presence of both PKD and PKC enzyme activities. These results extend those of Fig. 1 by providing evidence that PKC activity, and not that of PKC or PKC⑀, is preferentially associated with PKD. Thus, phosphorylation of the PKC⑀ peptide, an excellent PKC substrate, correlates with the appearance of the 80-kDa band in the in vitro kinase assays (Fig. 1) and with the detection of immunoreactive PKC in PKD immunoprecipitates (Fig. 2).
Interestingly, in cells co-transfected with PKD and PKC⑀, phosphorylation of syntide-2 was also nearly maximal even in the absence of PDBu stimulation even though phosphorylation of ⑀-peptide was only slightly above control levels. This result demonstrated that, in agreement with results in Figs. 1 and 2, PKC⑀ had activated PKD during coexpression but is retained in the PKD immune complexes only slightly, implying that the involvement of PKCs in PKD activation is mediated by a transient event rather than, or in addition to, the PKD-PKC association itself. As expected, immunoprecipitates from cells cotransfected with PKD and PKC phosphorylated syntide-2 in a manner indistinguishable from that from cells expressing only PKD, and did not phosphorylate ⑀-peptide. These results confirm that PKD preferentially forms complexes with PKC rather than with PKC⑀, and does not form complexes with PKC.
The PKD PH Domain Is Required for Formation of a PKD⅐PKC Complex-Recent reports have shown that PH domains within the Bruton's tyrosine kinase and the serinethreonine kinase PKB/Akt can mediate association of these proteins with multiple isoforms of PKC (27,28). In contrast, our results demonstrated that PKD preferentially associates with PKC. To examine whether the PH domain of PKD could mediate this specific association of PKD with PKC, we cotransfected COS-7 cells with wild-type PKC together with expression constructs encoding either intact PKD, PKD lacking the entire PH domain, or PKD with the PH domain truncated by partial deletions (15). PKD immunoprecipitates from these cells were subjected to Western blot analysis and phosphorylation of ⑀-peptide to assess the presence of PKC in the immune complexes (Fig. 4, A and B). In agreement with data shown in Fig. 2, wild-type PKC was co-immunoprecipitated with intact PKD (Fig. 4B). Similarly, these immunoprecipitates contained PKC activity, as revealed by ⑀ peptide assays (Fig.  4A). In contrast, when PKC was coexpressed with any of the PKD mutants containing either complete or partial deletions of the PH domain, both PKC activity (Fig. 4A) and PKC immunoreactivity (Fig. 4B) in PKD immunoprecipitates were sharply reduced. Control Western blots demonstrated that PKC was expressed in each transfection in similar amounts, as was each of the wild-type or mutant PKD proteins (Fig. 4B).
In order to determine the specificity of inhibition of PKD-PKC interaction by deletion of the PKD PH domain, we also examined the effect of other deletion mutants of PKD on the PKD-PKC interaction. COS-7 cells co-transfected with PKC together with PKD mutants lacking either a portion of the amino terminus containing the hydrophobic sequence of PKD, the first or second cysteine-rich regions, or the entire tandem cysteine-rich domain (22) were lysed and immunoprecipitated either with the PA-1 antiserum or with the PKC antiserum. As shown in Fig. 4C, Western analysis of each of these reciprocal immunoprecipitates revealed the presence of the opposite binding partner in the association. Similar to the results shown in Fig. 4, A and B, deletion of the PH domain prevented the detection of both binding partners in the respective immunoprecipitates (Fig. 4C). Taken together, these data indicate that the PH domain of PKD is necessary for efficient complex formation with PKC. However, it appears that the PH domain is not the only determinant of binding to PKD, as some PKC immunoreactivity and activity was associated with immunoprecipitates of the truncated PKD mutants (Fig. 4, A and B).
The PH Domain of PKD Selectively Binds to PKC-To further examine the possibility that the PH domain of PKD interacts preferentially with PKC, we incubated lysates of cells transfected with activated PKC or or wild-type PKC or ⑀ with a fusion protein, GST-PKDPH, which contains the fulllength PH domain (residues 429 -557) of PKD. The recovered fusion protein was extracted and subjected to Western blot analysis to detect associated PKCs. As shown in Fig. 5A, PKC (activated and wild-type) proteins were retained by their abil- were co-transfected with wild-type PKC together with PKD deletion mutants PKD⌬NH 2 lacking an amino-terminal portion containing the transmembrane region (⌬NH 2 ), PKD⌬PH (⌬PH), PKD-⌬Cys1 lacking the first cysteine-rich region (⌬Cys1), PKD-⌬Cys2 lacking the second cysteine-rich region (⌬Cys2), or PKD⌬CRD lacking the entire cysteinerich domain (⌬CRD). (In the experiment shown, a control Western blot of cell lysates indicated that the PKD⌬CRD protein was expressed to somewhat higher levels than the other PKD truncation mutants, thereby accounting for the somewhat higher intensity of this band in the blot of PKC immunoprecipitates.) At 72 h post-transfection, these cells were lysed and the lysates subjected to immunoprecipitations with PKC or PKD antibodies, as indicated. Immunoprecipitates were washed and analyzed by SDS-PAGE prior to Western analysis. PKD immunoprecipitates were probed using the anti-PKC antiserum, and PKC immunoprecipitates were probed using the anti-PKD antiserum, as indicated. The arrow indicates the expected position for PKD⌬PH protein. A, GST or GST-PKD PH (1 g each) was incubated with lysates from cells transfected with active mutant PKC (*), wild type PKC⑀ (⑀), wild-type PKC (), or active mutant PKC (*) prior to further processing. B, upper panels, different amounts of standard PKC ⑀ and proteins, as indicated, were subjected to Western blot analysis using the isoform-specific antisera. Middle panels, lysates were prepared from cells transfected with either PKC or PKC⑀ and subjected, in parallel, to Western blot analysis using the indicated amounts of cell lysates (diluted 1:1 with 2 ϫ SDS-PAGE gel loading buffer), as indicated. Lower panels, GST (10 g) or different amounts of GST-PKDPH (100 ng, 300 ng, 1 g, 3 g, 10 g and 15 g) were preadsorbed to glutathioneagarose beads, combined with lysates from cells transfected with either wild-type PKC or PKC⑀, as indicated, and then incubated and further processed as in A. Results shown are representative of at least three independent experiments. ity to associate with the GST-PKDPH fusion protein. In contrast, much less PKC⑀ was recovered from parallel cell lysates with this fusion protein. Control Western blot analysis of cell lysates confirmed that PKC⑀ was indeed overexpressed in these cells, allowing us again to infer that there was a preferential binding of PKD to PKC over PKC⑀. To address this issue more quantitatively, we performed Western blot analysis using different amounts of transfected cell lysates and purified protein standards (PKC⑀ and PKC) to determine the absolute amounts of each PKC isoform produced in cell lysates and recovered by the fusion protein. As determined from data shown in Fig. 5B, each PKC isoform was present in the cell lysates in similar amounts (approximately 1.5 g/ml for each PKC in the initial lysates). We then examined the binding of GST-PKDPH to PKC⑀ and PKC in cell lysates as a function of fusion protein concentration. These experiments demonstrated that the fusion protein bound to PKC at a concentration as low as 0.1 g/ml. In contrast, very little PKC⑀ was bound to the fusion protein even at a 100-fold higher concentration (Fig. 5B). DISCUSSION Our previous studies have shown that PKD is activated in vivo by treatment with biologically active phorbol esters or multiple agonists via a PKC-dependent pathway (19 -21). The results presented here demonstrate the formation of a complex between PKC and PKD and thus, identify a new aspect of the relationship between these two enzymes. Although further studies will be required to address the exact physiological role of this complex, the association of PKD and PKC could play a part in the regulation of the activity and/or subcellular localization of these enzymes. Indeed, recent findings from Cantley and co-workers (29) indicated that transfected PKD/PKC formed complexes with endogenous phosphatidylinositol 4-kinase and phosphatidylinositol 4-phosphate 5-kinase enzymes present in COS-7 cells. Importantly, truncation mutants of PKD/PKC lacking a portion of the molecule between the NH 2terminal hydrophobic region and the PH domain failed to retain the binding to the lipid kinases (29). In view of the results presented here, we conclude that different domains of the regulatory region of the PKD molecule are involved in mediating interactions with different proteins. Together, these results suggest the attractive possibility that PKD/PKC may act as a scaffold protein, through its different domains, promoting the assembly of signaling enzymes.
From our results, it is clear that the PH domain of PKD mediates a major portion of the binding to PKC, other (even larger) deletions of the PKD molecule being without effect. Recently, two important signaling proteins, PKB/Akt and BTK, have been shown to interact via their PH domains with multiple isoforms of PKC (27,28,30). Interestingly, there are important differences between these studies with BTK and PKB/ Akt and the findings presented here with PKD. 1) The studies with BTK and PKB/Akt indicated that the PH domain of these protein kinases interacts with multiple isoforms of PKC. In contrast, our results show that PKD associates preferentially with PKC. In fact, the fusion protein containing only the PKD PH domain was sufficient to isolate the wild-type PKC from cell lysates, and in this "pull-out" experiment, the PH domain mirrored the results seen with the intact PKD, i.e. it exhibited a striking preference for binding PKC over the closely related PKC⑀, and did not interact with PKC. 2) The binding of BTK PH domain to PKC⑀ requires a region within the C1 regulatory domain in the vicinity of the pseudosubstrate domain of this enzyme (31). We find that PKD associates with an active PKC mutant with a deletion from amino acids 155 to 171 within its pseudosubstrate domain (as well as with wild-type PKC). Therefore, the interaction of the PKD PH domain with PKC does not require the pseudosubstrate portion of the molecule.
Recently, a model for the binding of PH domains to proteins has been presented (32) in which a candidate sequence (HIKX 8 E), identified by sequence homology analysis, was proposed to act as a target for the PH domains. However, the putative binding sequence HIKX 8 E is present in both PKC and PKC, and consequently is unlikely to play a critical role in determining the differential binding of the PH domain of PKD to different PKCs found in the present study.
In conclusion, our findings demonstrate that coexpression of PKD with PKC leads to the formation of a stable PKD⅐PKC complex. Strikingly, we found very little evidence of complex formation between PKD and the PKC⑀ isoform despite its close similarity to PKC, and no evidence for a stable interaction between PKD and PKC. Our results also demonstrate that the PH domain is critical for stable PKD⅐PKC complex formation. We conclude that the PKD PH domain can discriminate between closely related structures of a single enzyme family, e.g. novel PKCs ⑀ and , thereby revealing a previously undetected degree of specificity among protein-protein interactions mediated by PH domains.