Rapid protein kinase D translocation in response to G protein-coupled receptor activation. Dependence on protein kinase C.

Protein kinase D (PKD)/protein kinase C (PKC) mu is a serine/threonine protein kinase that can be activated by physiological stimuli like growth factors, antigen-receptor engagement and G protein-coupled receptor (GPCR) agonists via a phosphorylation-dependent mechanism that requires PKC activity. In order to investigate the dynamic mechanisms associated with GPCR signaling, the intracellular translocation of a green fluorescent protein-tagged PKD was analyzed by real-time visualization in fibroblasts and epithelial cells stimulated with bombesin, a GPCR agonist. We found that bombesin induced a rapidly reversible plasma membrane translocation of green fluorescent protein-tagged PKD, an event that can be divided into two distinct mechanistic steps. The first step, which is exclusively mediated by the cysteine-rich domain in the N terminus of PKD, involved its translocation from the cytosol to the plasma membrane. The second step, i.e. the rapid reverse translocation of PKD from the plasma membrane to the cytosol, required its catalytic domain and surprisingly PKC activity. These findings provide evidence for a novel mechanism by which PKC coordinates the translocation and activation of PKD in response to bombesin-induced GPCR activation.

mediated activation of ␤ isoforms of phospholipase C (4) to produce two second messengers: inositol 1,4,5-trisphosphate, which mobilizes Ca 2ϩ from internal stores; and diacylglycerol, which activates PKC (5,6). There are multiple related PKC isoforms (7)(8)(9), which can be classified into three distinct subgroups on the basis of structural and regulatory differences: the conventional PKCs (␣, ␤ I , ␤ II , and ␥), which are regulated by calcium, diacylglycerol (DAG), and phospholipids; the novel PKCs (␦, ⑀, , and ), which are regulated by DAG and phospholipids; and the atypical PKCs ( and ), regulation of which is less characterized, but that have been proposed to be regulated by D-3 phosphoinositides (10). The DAG-regulated PKC isoforms all bind phorbol esters and are major cellular targets for this class of tumor promoter (11). It is increasingly recognized that each isoform has specific functions in vivo and that the biological activity of individual PKC isoforms is intimately regulated by their subcellular localization (12). However, the mechanisms by which PKC-mediated signals in the plasma membrane are propagated to critical downstream cytosolic targets remain largely undefined.
Protein kinase D (PKD)/protein kinase C is a serine/threonine protein kinase with structural, enzymological, and regulatory properties different from other PKC family members (13,14). The most salient features of PKD structure include the presence of a catalytic domain distantly related to Ca 2ϩ -regulated kinases, a pleckstrin homology (PH) domain that regulates its enzymatic activity, and a highly hydrophobic stretch of amino acids in its N-terminal region (15)(16)(17). The N-terminal region also contains a cysteine-rich domain (CRD) comprising a tandem repeat of cysteine-rich, zinc finger-like motifs, which confers high affinity binding to phorbol esters and plays a negative role in the regulation of catalytic activity of PKD (13,14,18,19). PKD can be activated in intact cells by pharmacological agents including biologically active phorbol esters and cell permeant DAGs as well as by physiological stimuli including GPCR agonists, growth factors, and antigen-receptor engagement (1-3, 20 -25). In all cases, PKD activation has been shown to be mediated by a PKC-dependent signal transduction pathway that involves the phosphorylation of Ser 744 and Ser 748 within the activation loop of the catalytic domain of PKD (1,23,24,26,51).
In unstimulated cells, PKD has been localized to the cytosol and to several intracellular compartments including Golgi and mitochondria (20,(27)(28)(29)(30)(31). Recently, it has been demonstrated that treatment of COS-7 cells and lymphocytes with phorbol 12,13-dibutyrate induces a striking and persistent transloca-* The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact. tion of PKD from the cytosol to the plasma membrane. Antigenreceptor engagement also caused rapid PKD translocation to the plasma membrane but in contrast to the response to phorbol 12,13-dibutyrate, this PKD redistribution was transient, with PKD returning to the cytosol within minutes of antigen engagement (20,31). These studies demonstrated that the translocation of PKD to the plasma membrane is mediated by its CRD, but the mechanism responsible for its reverse translocation from the plasma membrane to the cytosol has not been defined. Furthermore, very little is known regarding the intracellular distribution of PKD in cells stimulated by neuropeptide agonists that act through GPCRs to induce G␣ q -mediated activation of phospholipase C.
In the present study, we examined the temporal and spatial distribution of PKD using real-time visualization of fluorescence-tagged PKD in Swiss 3T3 fibroblasts and MDCK epithelial cells stimulated with bombesin, an agonist that activates PKD through a well characterized G q -coupled heptahelical receptor (32). We found that bombesin induces a striking and transient PKD translocation to the plasma membrane of fibroblasts and epithelial cells. Our results demonstrate that the intracellular redistribution of PKD in response to bombesin stimulation can be separated in two distinct mechanistic steps. The first step corresponds to the CRD-mediated plasma membrane translocation of PKD, whereas the second one corresponds to its rapid reverse translocation from the plasma membrane to the cytosol. This last step requires the catalytic domain of PKD and PKC-mediated phosphorylation of Ser 744 and Ser 748 within the activation loop of PKD. Thus, PKC is not only required for the phosphorylation of PKD at the activation loop Ser 744 and Ser 748 (see Ref. 51) but also promotes the rapid reverse translocation of PKD. These findings provide evidence for a novel mechanism coordinating the translocation and activation of PKD in response to bombesin-induced GPCR activation in fibroblasts and epithelial cells.

EXPERIMENTAL PROCEDURES
cDNA Constructs-Vectors encoding chimeric fusion proteins between green fluorescent protein (GFP) and PKD wild-type and mutant PKD⌬PH, PKD⌬CRD, and PKDP287G have been described previously (31). The single point mutant S916A was generated by PCR site-directed mutagenesis employing as template an EcoRI fragment isolated from pcDNA3-PKD (21), the set of primers GGTAAGCTAGCGAATCA-GAGGATGGCGACACGCTCACT//TGAGAAAACGCTACAGTGTGGA, and Pfu Turbo DNA polymerase (Stratagene, La Jolla, CA). The PCR product was digested with NheI, purified, and subcloned into pGFP-PKD previously digested with NheI. The pGFP-PKD-S744A/S748A vector was constructed by subcloning the HpaI/NheI cDNA fragment, isolated from pcDNA3-PKD-S744A/S748A (26), into pGFP-PKD previously digested with HpaI/NheI. The pGFP-PKD-S744A/S748A/S916A vector was constructed by subcloning the same PCR-generated NheI fragment used to construct pGFP-PKD-S916A into the NheI site of pGFP-PKD-S744A/S748A. NcoI digestion of pGFP-PKD and religation of the purified product encompassing its CRD was employed to construct the pGFP-PKD-CRD vector. In the resulting vector, five new amino acid residues and a stop codon (Glu-Thr-Cys-Trp-Arg-Stop) were introduced after the Met at position 365 of PKD, immediately after its CRD. The pGFP-PKD-⌬Cat vector was constructed by digesting pGFP-PKD with BamHI/NheI, Klenow fill-in, and ligation of the product encompassing the catalytic domain of PKD. In the resulting vector, one new amino acid residue and a stop codon (Cys-Stop) were introduced after the Gly at position 558 of PKD, immediately after its PH domain. The pPKD-RFP vector, encoding a chimeric fusion protein between the red fluorescent protein (RFP) from Discosoma sp. (33) and PKD wild-type, was constructed following this procedure. First, an EcoRI fragment isolated from pcDNA3-PKD (21) was subcloned into pGem3Z (Promega Corp., Madison, WI) resulting in the pGPKD vector. PCR site-directed mutagenesis using as template an EcoRI fragment isolated from pcDNA3-PKD (21), the set of primers TTCCGTCGACCCGGGGAGGG-GATGGAATGCGAGGATGCTGAC//TGAGAAAACGCTACAGTGTG-GA, and Pfu Turbo DNA polymerase was employed to replace the stop codon in PKD for the amino acid residue Ala and to simultaneously introduce a SalI restriction site. The PCR product was digested with NheI/SalI, purified, and subcloned into pGPKD digested with NheI/Sa-lI. The resulting vector, pGPKD-Stop919Ala was digested with EcoRI/SalI and the purified fragment containing the coding sequence for PKD without its stop codon was subcloned into pDsRed1-N1 (CLONTECH Laboratories, Inc., Palo Alto, CA) previously digested with EcoRI/SalI. All the constructs generated were confirmed by DNA sequence analysis. A schematic representation of all these constructs is shown in Fig. 1A. The products of expression of the different constructs were confirmed by Western blot using antibodies against GFP, PKD/PKC (C20) (Santa Cruz Biotechnology, Inc., Santa Cruz, CA) or RFP (CLONTECH Laboratories, Inc.). Dr. J. Battey (Laboratory of Molecular Biology, NIDCD, National Institutes of Health, Bethesda, MD) kindly provided the vector bNR-pCD2 containing the cDNA encoding the bombesin/gastrin-releasing peptide receptor.
Cell Culture and Transfections-Stock cultures of Swiss 3T3 and Madin-Darby canine kidney cells (MDCK) were maintained at 37°C in Dulbecco's modified Eagle's medium, supplemented with 10% fetal bovine serum in a humidified atmosphere containing 10% CO 2 and 90% air. For experimental purposes, cells were plated onto 15-mm no. Real-time Cell Imaging-In order to maintain a constant temperature of 37°C during the experimental procedures, cells grown in the 15-mm glass coverslips were mounted in a RC-25 perfusion chamber (Warner Instrument Corp., Hamden, CT) and perfused with medium preheated at 37°C by a TC-344B chamber system heater controller (Warner Instrument Corp.). The medium was supplemented with 10 mM HEPES, pH 7.2. The microscope used was a Zeiss epifluorescent Axioskop with a Zeiss Achroplan 40ϫ/0.75-watt water immersion objective (Carl Zeiss Inc., Jena, Germany). Images were captured as uncompressed 24-bit TIFF files with a SPOT cooled (-12°C) single CCD color digital camera (three-pass method) driven by SPOT version 2.1 software (Diagnostic Instruments, Inc., Sterling Heights, MI). GFP fluorescence was observed with a HI Q filter set for fluorescein isothiocyanate (Chroma Technology, Brattleboro, VT). For the experiments employing real-time imaging, 50 cells were analyzed for each experiment and each experiment was performed in quadruplicate. The selected single cell displayed in the appropriate figures was representative of 85% of the population of positive cells.
For video presentation, cell images were captured during 10 min (exposure time 400 ms) every 15 s as a time series of 24-bit uncompressed TIFF files. Selected images were cropped using Adobe Photoshop version 5.5 (Adobe Systems, Inc., San Jose, CA), assembled with the public domain software GifBuilder version 0.5 (Y. Piguet), and saved as Quicktime (version 4) files (Apple Corp., Cupertino, CA).
Quantitative analysis of the relative plasma membrane fluorescence intensity was performed on images of live cells obtained with a Leica TCS-SP upright laser-scanning confocal microscope (Leica, Heidelberg, Germany) using a 63 ϫ 1.2-watt HCX PL APO water immersion objective under 488-nm argon excitation and 500 -550-nm emission. During image acquisition, a constant temperature of 37°C was maintained following the methodology described above. Sequential images before and after bombesin stimulation were captured as uncompressed 8-bit TIFF files using Leica TCS-NT software (version 1.6.587). Quantification was performed using the public domain NIH Image program, version 1.62, and Adobe Photoshop, version 5.5 (Adobe Systems Inc.). The relative change in plasma membrane fluorescence intensity in the cell midsection, before and after bombesin stimulation, was calculated for each condition in four Swiss 3T3 cells by line intensity profiles across the cell plasma membrane.
Indirect Immunofluorescence-Swiss 3T3 cells cultured as described above in Lab-Tek chamber slides (Nalge Nunc International, Naperville, IL) were fixed for 15 min at 25°C in 10% buffered formalin phosphate and permeabilized with 0.2% Triton X-100 in phosphatebuffered saline (PBS) for 5 min at 25°C. The fixed cells were first incubated during 18 h at 25°C in blocking buffer (PBS, 1% gelatin, 0.05% Tween 20) (BB) and then stained at 37°C for 60 min with a rabbit polyclonal anti-PKD/PKC antibody diluted in BB. Subsequently, the cells were washed with PBS, 0.05% Tween 20 (WF) at 25°C and costained at 37°C for 60 min with a mouse monoclonal directed against a Golgi-associated protein (58-kDa Golgi protein) (34). After the cells were thorough washed with WF, they were stained at 37°C for 60 min with fluorescein-conjugated goat-anti rabbit diluted in BB, washed again with WF, and incubated at 37°C for 60 min with Texas Redconjugated goat anti-mouse immunoglobulins. Finally the cells were washed with WF and the samples mounted with a Gelvatol-glycerol solution containing 2.5% 1,4-diazobicyclo-[2.2.2]octane. The samples were examined with a Zeiss epifluorescent Axioskop with a Zeiss Plan-Apochromat 40ϫ/1.0 or 63ϫ/1.40 oil immersion objectives. Images were captured as uncompressed 24-bit TIFF files with a SPOT cooled (Ϫ12°C) single CCD color digital camera (three pass method) driven by SPOT version 2.1 software (Diagnostic Instruments, Inc.). Fluorescein or Texas Red signals were observed with HI Q filter sets for fluorescein isothiocyanate or rhodamine/tetramethylrhodamine B isothiocyanate, respectively (Chroma Technology). The cell images displayed in the figures are representative of more than 100 cells.
Cell Fractionation-Cytosolic and membrane fractions of Swiss 3T3 cells were prepared according to the method of McKenzie (35). Briefly, the cells were resuspended in ice-cold 10 mM Tris-HCl, 0.1 mM EDTA, pH 7.5 (TE buffer), containing protease inhibitors (leupeptin (10 g/ml) and phenylmethylsulfonyl fluoride (5 mM)) and homogenized by 20 strokes in a glass Potter-Elvehjem type homogenizer. The homogenate was centrifuged at 500 ϫ g for 15 min to pellet nuclei and unbroken cells, leaving a membrane-containing supernatant. Plasma membranes were collected by centrifugation of the supernatant for 30 min at 40,000 ϫ g. The plasma membrane-containing pellet was washed with TE buffer and re-centrifuged for 30 min at 40,000 ϫ g. Proteins in the resulting supernatant fractions were pooled and precipitated with icecold acetone. Pellets corresponding to the cytosolic and plasma membrane fraction were resuspended in 2ϫ SDS-PAGE and subsequently analyzed by SDS-PAGE and Western blot analysis.
Western Blot Analysis-Protein samples resuspended in 2ϫ SDS-PAGE sample buffer were resolved under denaturing conditions by 10% SDS-PAGE and transferred to Immobilon-P transfer membranes (Millipore Corp., Bedford, MA) using standard procedures. The membranes were incubated with primary rabbit polyclonal antibody against GFP (Santa Cruz Biotechnology, Palo Alto, CA) or a phosphospecific antisera that recognizes PKD phosphorylated at Ser 916 (36) and alkaline phosphatase-conjugated anti-rabbit secondary antibody. Signals were detected employing a chemifluorescent substrate (Vistra Systems, Amersham Pharmacia Biotech) and a Storm 840 PhosphorImager (Molecular Dynamics, Sunnyvale, CA). Quantification of the detected signals was done with ImageQuant software, version 1.2 (Molecular Dynamics).
Materials-pS916 antisera was produced by the central antibody production facility of Imperial Cancer Research Fund, London, United Kingdom. The anti-PKD (clone C20) antibody and its corresponding blocking peptide sc-639P as well as the anti-GFP antibodies were obtained from Santa Cruz Biotechnology, Inc., Santa Cruz, CA. Alkaline phosphatase-conjugated anti-rabbit was obtained from Tropix-PE Applied Biosystems (Bedford, MA). Ro 81-3220 was obtained from Calbiochem-Novabiochem Corp., San Diego, CA). GF I (GF 109203X), bombesin, fluorescein-conjugated goat anti-rabbit immunoglobulins, and the mouse monoclonal antibody against the Golgi-associated 58-kDa protein (clone 58k-9) were obtained from Sigma. Texas Red-conjugated goat anti-mouse immunoglobulins were obtained from Molecular Probes (Eugene, OR). All the other reagents were the highest grade commercially available.

Bombesin Stimulates GFP-PKD Activation in Swiss 3T3 Fibroblasts and MDCK Epithelial Cells-Previously we reported
that the fusion of a GFP tag to PKD did not produce any detectable effect on its basal catalytic activity, phorbol ester binding, and phorbol ester-mediated PKD kinase activation (20). In addition, the inherent fluorescence of GFP allowed us to visualize the translocation of GFP-PKD in live cells (20). In order to verify that GFP-PKD displays biological properties similar to that of native PKD in GPCR-stimulated cells, we examined whether GFP-PKD is rapidly activated in response to bombesin receptor activation in fibroblasts and epithelial cells.
Swiss 3T3 fibroblasts, a model system extensively used to elucidate signaling by endogenously expressed GPCRs (37), were transiently transfected with GFP-PKD. MDCK cells, an epithelial cell line, which does not express endogenous bombesin receptors, were cotransfected with plasmids encoding GFP-PKD and bombesin GPCR. Cell cultures were then stimulated with 10 nM bombesin for various times (0 -60 min). Active PKD autophosphorylates on a C-terminal serine residue, Ser 916 , and antibodies that recognize Ser 916 -phosphorylated PKD can be used to monitor the activity of PKD (36). As shown in Fig. 1B, the pS916 antiserum only weakly recognized GFP-PKD from unstimulated cells, but reacted strongly with active GFP-PKD isolated from either Swiss 3T3 or MDCK cells up to 60 min after stimulation with bombesin. Western blot analysis with anti-GFP antibodies verified the expression and equal loading of GFP-PKD. The results presented in Fig. 1B indicate that activation of either endogenously or exogenously expressed bombesin GPCR induces a rapid and persistent increase in the catalytic activity of GFP-PKD within fibroblasts and epithelial cells.
Intracellular Distribution of Fluorescence-tagged and Endogenous PKD in Swiss 3T3 Cells-Previous reports indicated that PKD partially localizes to the Golgi compartment as well as to the cytosol, mitochondria and plasma membrane in different cell types (20,(27)(28)(29)(30)(31). We also detected that in Swiss 3T3, in addition to its cytosolic localization, some GFP-PKD fluorescence was more pronounced at the perinuclear area, consistent with a partial localization of this protein to the Golgi compartment. Interestingly, we also found that when Swiss 3T3 cells were shifted from 37°C to 20°C, GFP-PKD accumulated within 5 min in a single area adjacent to the nucleus (Fig. 2, A-C), consistent with Golgi localization. Similarly, when Swiss 3T3 expressing GFP-PKD were preincubated at 20°C and then shifted to 37°C, the fluorescent signal previously detected at one prominent spot in the perinuclear area redistributed to the cytosol within 7 min (Fig. 2, D and E). These results indicate that GFP-PKD can associate with the Golgi and imply that its dissociation from this compartment is a temperature-sensitive process. We did not detect any substantial redistribution of GFP-PKD to the Golgi compartment after bombesin stimulation in Swiss 3T3. These findings, in addition to those already published (20,(27)(28)(29)(30)(31), suggest that the preferential localization of PKD to different cellular compartments may be dependent on the cell type and experimental conditions.
To rule out possible interference with the normal intracellular distribution of PKD due to the presence of the fluorescent tag at its N terminus, we analyzed the localization of a chimeric protein between the RFP from Discosoma sp. fused to the C terminus of PKD. No difference was detected at 37°C in the distribution of PKD-RFP compared with GFP-PKD in Swiss 3T3 cells (Fig. 2F). Further support for these observations were obtained by indirect immunofluorescence analysis of the distribution of endogenous PKD and the Golgi-associated 58-kDa protein in Swiss 3T3 cells. As illustrated by representative cells displayed in Fig. 2 (G and I), the distribution of endogenous PKD at 37°C was very similar to that detected with either GFP-PKD or PKD-RFP at the same temperature (Fig. 2, A, E, and F). The inclusion of the immunizing peptide encompassing the C terminus of PKD completely prevented the staining of the endogenous PKD (data not shown). We detected, in ϳ20% of the analyzed cells, an increased PKD signal in a single area adjacent to the nucleus which colocalized with the signal corresponding to the Golgi-associated 58-kDa protein, an specific Golgi marker (34,38). However, the amount and distribution of PKD in the remaining cells was very homogenous, through the cytosol, further demonstrating that the distribution of the fluorescence-tagged PKDs was comparable to the endogenously expressed PKD (Fig. 2, compare I and J with G and H). Subsequent experiments examining agonist induced GFP-PKD re-distribution were carried out at 37°C.
Activation of Bombesin-GPCR Induces Transient Translocation of GFP-PKD from the Cytosol to the Plasma Membrane of Fibroblasts and Epithelial Cells-Next, we analyzed the effect of bombesin stimulation on the intracellular distribution of GFP-PKD expressed in either Swiss 3T3 cells or MDCK cells. As illustrated by the image presented in Fig. 3, GFP-PKD expressed in unstimulated Swiss 3T3 cells was distributed throughout the cytosol and excluded from the nucleus with very little fluorescent signal localized to the plasma membrane.
Real time imaging revealed that bombesin stimulation of Swiss 3T3 cells induced a rapid translocation of GFP-PKD to the plasma membrane, causing a localized fluorescence in the plasma membrane at the cell periphery and partially masking the cell nuclei (Fig. 3A). Translocation of GFP-PKD occurred within 2 min and reached a maximum between 3 and 5 min. The association of GFP-PKD with the plasma membrane was transient. The reverse translocation of GFP-PKD from the plasma membrane to the cytosol was virtually complete within 10 min of bombesin stimulation (Fig. 3A). A similar time course of GFP-PKD translocation to the plasma membrane and reverse translocation to the cytosol in bombesin-stimulated Swiss 3T3 fibroblast was also observed and quantified by laser confocal microscopy (Fig. 3B) and subsequently confirmed by biochemical fractionation, as described below (Fig. 9).
Because the cell surface distribution of membrane proteins in fibroblasts and epithelial cells can be different, we also analyzed the translocation of GFP-PKD in response to bombesin in MDCK epithelial cells cotransfected with plasmids encoding GFP-PKD and bombesin receptor. MDCK cells are one of the best studied epithelial cell model systems (39). The distribution of GFP-PKD in unstimulated MDCK cells was almost identical to that in Swiss 3T3 cells. Specifically, GFP-PKD was homogenously distributed throughout the cytosol and excluded from the nucleus in the majority of the MDCK cells (Fig. 3A). Stimulation of MDCK cells with 10 nM bombesin promoted a striking translocation of GFP-PKD to the plasma membrane detected as a localized fluorescence at the cell periphery, which was evident within 2 min and reached a maximum between 3 and 5 min (Fig. 3A). Redistribution of GFP-PKD to the plasma membrane was also evident from the partial masking of the cell nuclei by the diffuse fluorescence signal at the cell surface. A prominent localization of GFP-PKD was also detected at intercellular plasma membrane junctions.
Similar to the data obtained with Swiss 3T3 cells, bombesininduced GFP-PKD translocation to the plasma membrane of MDCK cells was transient. The fluorescent signal previously detected at the plasma membrane returned to the cytosol within 10 min of bombesin stimulation (Fig. 3A). The remarkable dynamics of the transient translocation of GFP-PKD in MDCK cells in response to activation of the bombesin GPCR can be observed as a movie (see panel a of the supplemental materials, available in the on-line version of the journal). In contrast, the distribution of GFP (detected in the nuclei as well as throughout the cytosol of both cell types) was not affected by bombesin stimulation in either Swiss 3T3 fibroblasts or MDCK cells (Fig. 3A). The precise timing of GFP-PKD translocation varied from cell to cell, but consistent maximal plasma membrane localization of GFP-PKD was detected within 3-5 min after bombesin stimulation. GFP-PKD then dissociated from the plasma membrane, returning to the cytosol completely within 8 -10 min of bombesin stimulation. The morphology of non-transfected Swiss 3T3 or MDCK cells was indistinguishable from GFP-or GFP-PKD-transfected ones (data not shown). The results presented in Fig. 3

demonstrate that agonist stimulation of the bombesin GPCR induces a striking and transient translocation of GFP-PKD to the plasma membrane in both Swiss 3T3 fibroblasts and MDCK epithelial cells. The Cysteine-rich Domain of PKD Is Sufficient to Mediate the Plasma Membrane Translocation of GFP-PKD in Response to
GPCR Activation-In order to identify the domain(s) of PKD responsible for its translocation to the plasma membrane in response to GPCR activation, we examined the redistribution of a set of GFP-PKD mutants (see Fig. 1A). The N-terminal regulatory region of PKD contains a DAG-binding CRD comprising a tandem repeat of cysteine-rich motifs and a PH domain (13). To determine the contribution of the CRD and the PH domain to bombesin-induced PKD translocation, different GFP-tagged PKD mutants were expressed transiently in either Swiss 3T3 or MDCK cells, and the translocation of these molecules was monitored by real-time imaging of live cells after bombesin stimulation.
The GFP-PKD-⌬CRD mutant contains a deletion of the entire CRD domain, whereas the GFP-PKD P287G mutant contains a proline to glycine substitution within the second cysteine-rich motif of the CRD. In both cases, binding of phorbol esters/DAG to PKD is prevented (19). As shown in Figs. 4 and 5, GFP-PKD-⌬CRD and GFP-PKD-P287G were distributed evenly throughout the cytosol of unstimulated Swiss 3T3 and MDCK cells and excluded from the nucleus, a distribution indistinguishable from that of wild-type GFP-PKD. Following bombesin GPCR stimulation, wild-type GFP-PKD transiently translocated from the cytosol to the plasma membrane, as observed previously. In contrast, bombesin was unable to induce any detectable plasma membrane translocation of either GFP-PKD-⌬CRD or GFP-PKD-P287G (Figs. 4 and 5). These results demonstrated that the bombesin-induced translocation of PKD to the plasma membrane of either fibroblast or epithelial cells was dependent on the integrity of the CRD, the DAG/ phorbol ester-binding domain of PKD.
The analysis of agonist-induced membrane translocation of classic PKC␤II demonstrated that this process is mediated by the cooperative interaction of C1 and C2 region (40). PKD does not posses a C2 region but contains a PH domain, located between the CRD and the catalytic domain of PKD. This do-main could be involved in the plasma membrane translocation of PKD since PH domains are modular protein motifs that mediate protein-protein as well as protein-lipid interactions (41,42). Real-time analysis of the distribution of a fusion protein consisting of GFP and a PH domain-deleted PKD (see Fig.  1A) showed that despite a partial nuclear localization of GFP-PKD-⌬PH, bombesin stimulation caused its translocation to the plasma membrane of Swiss 3T3 and MDCK cells (data not shown). We conclude that the PH domain of PKD, in contrast to the CRD, does not contribute to PKD translocation induced by bombesin GPCR activation.
In order to determine whether the CRD is sufficient to mediate bombesin-induced PKD recruitment to the plasma membrane, we analyzed the distribution of fusion proteins consisting of GFP and PKD lacking its catalytic domain (GFP-PKD-⌬Cat) and GFP and the CRD of PKD (GFP-PKD-CRD) in unstimulated and bombesin-stimulated cells. As illustrated in Figs. 4 and 5, the deletion of the entire catalytic domain of PKD did not interfere with bombesin-induced plasma membrane translocation of the resulting fusion protein GFP-PKD-⌬Cat, despite its partial nuclear localization. The results in Figs. 4 and 5 also show that bombesin stimulation promoted the translocation to the plasma membrane of GFP-PKD-CRD in both Swiss 3T3 and MDCK cells, demonstrating that the CRD is necessary and sufficient to mediate the translocation of PKD to the plasma membrane in response to GPCR activation. A summary of the translocation properties of the different mutants analyzed is shown in Fig. 1A.
PKC Kinase Activity Inhibition Prevents the Rapid Reverse Translocation of GFP-PKD from the Plasma Membrane to the Cytosol after GPCR Activation-The analysis of the distribution of the PKD fusion proteins lacking catalytic domains revealed an important difference between these proteins and GFP-PKD. In contrast to GFP-PKD, GFP-PKD-⌬Cat and GFP-PKD-CRD remained associated with the plasma membrane for 25-30 min after bombesin stimulation (data not shown), suggesting that the rapid reverse translocation of the fluorescencelabeled PKD required its catalytic domain. We hypothesize that the activation of the catalytic domain of PKD plays a major role in promoting its rapid reverse translocation from the plasma membrane to the cytosol. We examined this hypoth- FIG. 4. Plasma membrane translocation of GFP-PKD mutants in Swiss 3T3 cells in response to bombesin stimulation. Swiss 3T3 cells were transfected with constructs encoding a set of different PKD mutants fused to GFP and imaged in real time 18 h after transfection, at 37°C. Cells expressing the GFP proteins were visualized with an epifluorescence microscope and representative images captured immediately before and at the indicated times after 10 nM bombesin stimulation. Bar, 10 m. esis using pharmacological and mutational approaches that prevent PKD activation in response to bombesin GPCR stimulation.
Bombesin-and G q -induced PKD activation is mediated by a PKC-dependent signal transduction pathway (1,23,24,26,32). These studies have placed PKD downstream of classical/novel PKC enzymes in a hierarchical signaling cascade regulated by DAG. To examine the role of PKCs in the regulation of GFP-PKD translocation by the bombesin GPCR, Swiss 3T3 and MDCK cells were preincubated with either Ro 31-8220 or GF I, which inhibit PKCs and prevent PKD activation (1), and then the distribution of GFP-PKD in these cells was monitored after bombesin stimulation. Treatment with either Ro 31-8220 or GF I did not affect the plasma membrane translocation of GFP-PKD, indicating that classical/novel PKCs are not necessary for PKD translocation. This conclusion is consistent with the notion that PKD translocation is mediated by the CRD. However, the inhibitors of PKC activity dramatically delayed the plasma membrane dissociation of GFP-PKD (Fig. 6), indicating that the kinase activity of PKC is necessary for the rapid reverse translocation of GFP-PKD. These results were confirmed by biochemical fractionation as described below (Fig. 9).
Mutations of Ser 744 and Ser 748 , within the Activation Loop of PKD, Prevent the Rapid Reverse Translocation of GFP-PKD from the Plasma Membrane to the Cytosol after GPCR Activation-As shown in the accompanying paper (51), bombesin stimulation of either Swiss 3T3 or MDCK cells induces rapid phosphorylation of Ser 744 and Ser 748 , within the activation loop of PKD (1,23,24,26). Here we examined whether the activation loop phosphorylation is also required for rapid PKD translocation. If phosphorylation of Ser 744 and Ser 748 were necessary for the plasma membrane dissociation of PKD, their substitution by non-phosphorylatable amino acids should impair the dissociation of the mutated protein. In order to test this hypothesis, we replaced Ser 744 and Ser 748 with alanine and examined the distribution properties of the resulting fusion protein GFP-PKD-S744A/S748A in Swiss 3T3 and MDCK cells after bombesin stimulation.
The distribution of GFP-PKD-S744A/S748A in non-stimulated cells was identical to that of GFP-PKD. After bombesin stimulation, the translocation and association of GFP-PKD-S744A/S748A to the plasma membrane of both cell types did not show any significant difference, as compared with GFP-PKD. Thus, substitution of Ser 744 and Ser 748 for alanine did not prevent the plasma membrane translocation of GFP-PKDS744A/S748A (Figs. 7 and 8). In addition, mutation of Ser 916 , an autophosphorylation site in the C terminus of PKD (36), did not prevent the plasma membrane translocation of GFP-PKD-S916A (Figs. 7 and 8). In addition, simultaneous substitution in PKD of the serine residues 744, 748, and 916 for alanine had no effect on the plasma membrane translocation or GFP-PKDS744A/S748A/S916A after bombesin stimulation of Swiss 3T3 and MDCK cells (data not shown).
The salient feature of the results shown in Figs. 7 and 8 is that GFP-PKD-S744A/S748A did not return from the plasma membrane to the cytosol even after 10 min of bombesin stimulation. This persistent residence of GFP-PKD-S744A/S748A in the plasma membrane over time can also be observed as a movie (see panel b of the supplemental materials, available on-line). Furthermore, quantitative analysis by laser confocal microscopy of over time relative fluorescence intensity in the plasma membrane of Swiss 3T3 cells expressing GFP-PKD-S744A/S748A also showed that this protein did not return to the cytosol after 10 min of bombesin stimulation (Fig. 7B). A significant reverse translocation of GFP-PKD-S744A/S748A was seen 25-30 min later than GFP-PKD (data not shown). In contrast, GFP-PKD-S916A translocated to the plasma membrane and reverse translocated to the cytosol as rapidly as GFP-PKD following bombesin stimulation (Figs. 7 and 8). Thus, either pharmacological inhibition of PKC or mutation of Ser 744 and Ser 748 to alanine strikingly prevented the rapid reverse translocation of fluorescence-tagged PKD after bombesin stimulation in both Swiss 3T3 fibroblasts and MDCK epithelial cells.
Biochemical fractionation of Swiss 3T3 expressing GFP-PKD or GFP-PKD-S744A/S748A after bombesin stimulation was employed to further confirm a role of PKC and Ser 744 and Ser 748 in the reverse translocation of PKD. Western blot analysis of the soluble (cytosolic) and particulate (membrane) fractions, using anti-GFP antibody, showed a significant increase of GFP-PKD in the particulate fraction within 1 min of bombesin stimulation (Fig. 9). After 5 min of bombesin stimulation, the distribution of GFP-PKD was indistinguishable from that in the unstimulated cultures. Thus, biochemical fractionation confirm our evidence from real-time imaging that bombesin stimulation induces transient redistribution of GFP-PKD in Swiss 3T3 cells.
Pretreatment of the cells with the PKC inhibitor Ro 31-8220  7. A, plasma membrane association of GFP-PKD mutants in Swiss 3T3 cells in response to bombesin stimulation. Swiss 3T3 cells were transfected with constructs encoding different PKD mutants fused to GFP and imaged in real time 18 h after transfection, at 37°C. Cells expressing the GFP proteins were visualized with an epifluorescence microscope and representative images captured immediately before and at the indicated times after 10 nM bombesin stimulation. Bar, 10 m. Selected cell regions were magnified for comparison (insets). B, quantitative analysis of plasma membrane relative fluorescence. Swiss 3T3 cells were transfected with a construct encoding GFP-PKD-S744A/S748A and imaged in real time by confocal laser scanning fluorescence microscopy 18 h after transfection, at 37°C. The relative change in plasma membrane fluorescence intensity in the cell midsection was calculated, for each condition, by line intensity profiles across the cell plasma membrane. The data represent the means Ϯ standard deviation summarized from four different Swiss 3T3 cells expressing GFP-PKD-S744A/S748A before and after 10 nM bombesin stimulation. RMFI, relative plasma membrane fluorescence intensity. did not affect the membrane translocation of GFP-PKD after bombesin stimulation, as revealed by its detection in the particulate fraction (Fig. 9). However, preincubation of Swiss 3T3 cells with Ro 31-8220 impaired the reverse translocation of GFP-PKD from the particulate to the soluble fraction (Fig. 9). A similar impairment in reverse translocation was observed with GFP-PKD-S744A/S748A. This fusion protein translocated from the soluble to the particulate fraction of bombesin-stimulated Swiss 3T3 within 3 min of stimulation. However, GFP-PKD-S744A/S748A, in contrast to GFP-PKD, was impaired in its reverse translocation to the cytosol after bombesin stimulation (Fig. 9).
Although we cannot rule out that other post-translational modifications of PKD may also be involved in its plasma membrane dissociation, our results demonstrate, for the first time, that the kinase activity of PKC and the phosphorylation of Ser 744 and Ser 748 are key events in the rapid intracellular translocation of PKD.

DISCUSSION
In the present study, we used real-time visualization as well as biochemical fractionation of GFP-PKD chimeras to demonstrate that agonist activation of the bombesin GPCR induces a redistribution of PKD to the plasma membrane of fibroblast and epithelial cells. PKD reverse translocated from to the plasma membrane to the cytosol within minutes of bombesin stimulation. In order to elucidate the mechanism(s) involved in bombesin-induced PKD translocation, we analyzed the redistribution of a variety of PKD mutants. This analysis revealed that PKD translocation could be divided into at least two entirely different mechanistic steps: 1) PKD translocation from the cytosol to the plasma membrane and 2) reverse translocation from the plasma membrane to the cytosol.
The first step, bombesin-stimulated plasma membrane PKD translocation, is completely prevented by deletion of the CRD and severely impaired by a single amino acid substitution in the second cysteine-rich motif of the CRD (P287G), in agreement with results indicating that this motif is responsible for mediating DAG binding to PKD. Furthermore, a fusion protein consisting of GFP and the CRD of PKD rapidly translocated to the plasma membrane in response to bombesin stimulation in both fibroblasts and epithelial cells. We conclude that the CRD is both necessary and sufficient to mediate the first phase of PKD translocation in response to GPCR activation. These results are consistent with previous findings using GFP-PKD in phorbol ester-stimulated COS-7 cells and antigen-stimulated T and B lymphocytes (20,31).
The mechanism(s) underlying the second step of PKD trafficking, namely PKD reverse translocation, has not been examined in previous studies. It is known that bombesin binding to its cognate GPCR promotes G␣ q -mediated activation of ␤ isoforms of phospholipase C to produce the second messengers inositol 1,4,5-trisphosphate and DAG, which can also be gen- FIG. 8. Plasma membrane association of GFP-PKD mutants in MDCK cells in response to bombesin stimulation. MDCK cells were transfected with constructs encoding different PKD mutants fused to GFP and imaged in real time 18 h after transfection, at 37°C. Cells expressing the GFP fusion proteins were visualized with an epifluorescence microscope and representative images captured immediately before and at the indicated times after 10 nM bombesin stimulation. Bar, 10 m. Selected cell regions were magnified for comparison (insets).
FIG. 9. Biochemical fractionation analysis: Effect of PKC inhibition or mutations in the activation loop of PKD on the plasma membrane association of GFP-PKD fusion proteins after bombesin stimulation. Swiss 3T3 cells transfected with constructs encoding GFP-PKD and GFP-PKD-S744A/S748A were fractionated into soluble (cytosolic) (S) and particulate (membrane) (P) fractions (see "Experimental Procedures") immediately before and at the indicated times after 10 nM bombesin stimulation. Top panel, Swiss 3T3 cells transfected with pGFP-PKD were fractionated 18 h after transfection as indicated above. Middle panel, Swiss 3T3 cells transfected with pGFP-PKD were incubated for 30 min with 2.5 M Ro 31-8220, an inhibitor of PKC, 18 h after transfection and fractionated as indicated above after bombesin stimulation. Bottom panel, Swiss 3T3 cells transfected with pGFP-PKD-S744A/S748A were fractionated 18 h after transfection as indicated above after bombesin stimulation. Proteins in the cytosolic and membrane fractions were resolved in 10% SDS-PAGE and transferred to Immobilon-P transfer membranes. The GFP fusion proteins on the membranes were decorated with anti-GFP as primary antibody and alkaline phosphatase-conjugated anti-rabbit as secondary antibody. Signals were detected with a chemifluorescent substrate and a phosphorimager and quantified with ImageQuant software. The data are expressed as a percentage of the total GFP-PKD detected protein in arbitrary units (100% equals soluble plus particulate fractions). erated by other routes (4 -6). Although some reports indicated that DAG accumulation can be sustained for 30 min in Swiss 3T3 cell stimulated with bombesin (43), it is generally thought that DAG accumulation is transient as a consequence of rapid DAG turnover in the cell. Thus, a simple model to explain the reverse translocation of PKD is that the plasma membrane pool of DAG, which binds the CRD of PKD, is removed thus enabling the return of PKD to the cytosol.
Several lines of evidence produced in this study indicate that this model is not sufficient to explain the rapid reverse translocation of PKD. For example, a fusion protein of GFP and PKD lacking its catalytic domain translocated efficiently to the plasma membrane in response to bombesin but did not return to the cytosol with the same kinetics as GFP-PKD. Furthermore, a fusion protein of GFP with the CRD of PKD (GFP-PKD-CRD), which can be regarded as a DAG fluorescent label indicator (44), moved to the plasma membrane after bombesin stimulation but only returned to the cytosol 25-30 min later than GFP-PKD. These findings indicated, for the first time, that the rapid reverse translocation of PKD to the cytosol required its catalytic domain.
We produced two independent lines of evidence to support the notion that the rapid reverse translocation of PKD is coupled to PKC-mediated PKD activation. Specifically, pharmacological inhibition of PKC or mutation to alanine of Ser 744 and Ser 748 , which are located in the activation loop of PKD, strikingly retarded the reverse translocation of fluorescence-tagged PKD after bombesin stimulation in both Swiss 3T3 and MDCK cells. The results presented in the accompanying paper (51) demonstrate that bombesin induces rapid phosphorylation of Ser 744 and Ser 748 in both Swiss 3T3 fibroblast and MDCK epithelial cells, as indicated by using novel antibodies directed against the phosphorylated state of these residues. We conclude that PKCs not only play a critical role in PKD activation but also are required to trigger the rapid reverse translocation of PKD.
Recent studies demonstrated that translocation of Aplysia PKC as well as vertebrate PKC␤II is regulated by autophosphorylation (40, 45). Although these studies also concluded that phosphorylation promotes plasma membrane dissociation of these kinases, there are important differences between our results with PKD and these findings with PKCs. Although PKC autophosphorylation is thought to occur during protein maturation (46), PKD phosphorylation is acutely induced through a PKC-dependent pathway (1,23,24). Furthermore, phosphorylation-mediated plasma membrane dissociation of PKCs has been proposed to act as a feedback mechanism that leads to signal termination (40), whereas PKD phosphorylation leads to its activation (26).
Previous reports demonstrated that PKD/PKC can be found associated to PLC␥ within the activated B cell receptor complex and to Bruton's tyrosine kinase in addition to regulate the function of the epidermal growth factor receptor and the Na ϩ /H ϩ exchanger, all of which can be found at the plasma membrane (25,(47)(48)(49). It is conceivable that the plasma membrane translocation and simultaneous activation of PKD play a role in the regulation of the above mentioned proteins or another unknown targets. Furthermore, we hypothesized that the translocation of PKD to the plasma membrane is required for PKC-mediated phosphorylation at the activation loop (as demonstrated in the accompanying article (Ref. 51)), leading to a sustained PKD activation.
An important role of PKCs in PKD activation has been documented in a variety of cell types stimulated by many stimuli including neuropeptides, bioactive lipids, growth factors, antigens, constitutively activated G␣ q , and oxidative stress (1-3, 20 -25, 32, 36, 50). In all these cases, the activity of PKD is not longer dependent on membrane-delimited second messengers; therefore, PKD can be active in the cytosol. Indeed, our own results indicate that GFP-PKD remains active long after it returned to the cytosol in bombesinstimulated fibroblasts and epithelial cells (see Fig. 1B). An attractive hypothesis is that PKCs coordinate PKD activation with rapid reverse translocation thus providing a novel mechanism that ensures that activated PKD moves efficiently away from the plasma membrane into the interior of the cell where it propagates DAG-PKC signals initiated at the cell surface.