|Home | About | Journals | Submit | Contact Us | Français|
Protein kinase D (PKD) isoenzymes regulate the formation of transport carriers from the trans-Golgi network (TGN) that are en route to the plasma membrane. The PKD C1a domain is required for the localization of PKDs at the TGN. However, the precise mechanism of how PKDs are recruited to the TGN is still elusive. Here, we report that ADP-ribosylation factor (ARF1), a small GTPase of the Ras superfamily and a key regulator of secretory traffic, specifically interacts with PKD isoenzymes. ARF1, but not ARF6, binds directly to the second cysteine-rich domain (C1b) of PKD2, and precisely to Pro275 within this domain. Pro275 in PKD2 is not only crucial for the PKD2-ARF1 interaction but also for PKD2 recruitment to and PKD2 function at the TGN, namely, protein transport to the plasma membrane. Our data suggest a novel model in which ARF1 recruits PKD2 to the TGN by binding to Pro275 in its C1b domain followed by anchoring of PKD2 in the TGN membranes via binding of its C1a domain to diacylglycerol. Both processes are critical for PKD2-mediated protein transport.
The protein kinase D (PKD) family of serine/threonine kinases comprises PKD1/PKCμ, PKD2, and PKD3/PKCν (Rykx et al., 2003 ). PKDs are activated either directly via phorbol esters or indirectly by various mechanisms, including G protein-coupled receptors (Jamora et al., 1999 ). These kinases are involved in various fundamental biological processes, including regulation of Golgi structure and function and consequently protein transport from the TGN to the plasma membrane (Rozengurt et al., 2005 ). At the TGN, PKDs are activated by G protein β1γ2 and β3γ2 subunits via the Golgi-associated nPKCη (Díaz Anel and Malhotra, 2005 ) and are required for the shedding of cargo containing vesicles from the TGN (Bard and Malhotra, 2006 ). This might be accomplished by PKD1-induced phosphorylation of phosphatidylinositol-4 kinase IIIβ (PI4KIIIβ) (Hausser et al., 2005 ) and ceramide transfer protein (CERT) (Fugmann et al., 2007 ). Whereas there is substantial data on the function of PKDs at the TGN, the recruitment of these kinases to the TGN is less well understood. PKDs consist of an N-terminal regulatory domain followed by two cysteine-rich zinc finger regions, termed C1a and a C1b domain; a pleckstrin homology domain; and the catalytic domain at the C terminus. Localization of PKD1 at the TGN requires its binding to diacylglycerol (DAG) via its first cysteine-rich domain (Maeda et al., 2001 ; Baron and Malhotra, 2002 ). However, it is currently unclear how PKDs are targeted specifically to DAG at the Golgi. We were interested in the mechanism by which PKDs are recruited to the Golgi compartment. TGN recruitment of PKDs could be due to a particular local concentration of DAG or to the recruitment of PKDs by a specific Golgi resident protein. Because there is little evidence for the first assumption, we addressed the second option.
ADP-ribosylation factors (ARFs) are small GTPases that act as sensors of the lipid environment and transducers of information which results in changes in ARF activity and the consequent assembly of protein structures on the membranes. Based on their sequence homology, the human ARF proteins are classified into three classes: class I (ARF1 and ARF3), class II (ARF4 and ARF5), and class III (ARF6) (Moss and Vaughan, 1998 ). Both class I and II ARFs localize to the Golgi compartment and are thought to play overlapping and redundant roles (Gillingham and Munro, 2007 ). ARF6 is divergent from the other members localizing to the plasma membrane and the endocytic system. ARFs play a central role in the secretory pathway by regulating the membrane association and/or activation of many effector proteins such as coat proteins COPI (coatomer) (Donaldson et al., 1992 ; Palmer et al., 1993 ), clathrin adaptor proteins (Zhu et al., 1998 ; Ooi et al., 1998 ; Donaldson and Jackson, 2000 ; Boehm et al., 2001 ), Golgi-localizing, γ-adaptin ear homology domain, ARF-binding protiens (Dell'Angelica et al., 2000 ), four-phosphate-adaptor proteins (Godi et al., 2004 ), various lipid-modifying enzymes (Brown et al., 1993 ; Honda et al., 1999 ; Godi et al., 1999 ), and other effectors (Boehm et al., 2001 ). ARF6 is a major regulator of endocytosis, cytokinesis, and the organization of the actin cytoskeleton (Donaldson, 2003 ). Because class I and II ARFs localize to the Golgi and ARF1 plays a key role in the recruitment of various complexes of cytosolic proteins, including PKD substrates such as PI4KIIIβ to the Golgi network (Godi et al., 1999 ; Hausser et al., 2005 ; D'Souza-Schorey and Chavrier, 2006 ), we examined a potential role of ARFs in recruiting PKDs to the plasma membrane at the example of PKD2.
In the present study, we show that ARF1 specifically and directly interacts with PKD2. Furthermore, ARF1 and PKDs colocalize at the TGN in immunocytochemistry. The interaction of PKD2 with ARF1 is mediated via its C1b domain and regulated by ARF1 activity. We identify Pro275 within the C1b domain of PKD2 as the crucial amino acid for ARF1 binding, recruitment of PKD2 to the TGN and PKD2 function at the Golgi, i.e., the regulation of protein transport from the TGN to the plasma membrane. Our results provide a novel model in which the localization of PKD2 to the TGN is accomplished by both, the C1a and C1b domain of the kinase. The C1b domain of PKD2 interacts with ARF1 that recruits PKD2 to the TGN where the kinase is anchored by binding to local DAG via its C1a domain.
HeLa and HEK293-T-cells were grown in DMEM with 10% (vol/vol) fetal calf serum, 100 U/ml penicillin, and 100 mg/ml streptomycin in a humidified atmosphere containing 10% CO2 at 37°C. Exponentially growing HeLa cells were transfected with Lipofectamine LTX (Invitrogen, Carlsbad, CA). HEK293-T-cells were grown to 50% confluence and transfected with polyethylenimine (Polysciences, Warrington, PA).
Anti-TGN46 antibody was purchased from Novus Biologicals (Littleton, CO); anti-hemagglutinin (HA) antibody was from Santa Cruz Biotechnology (Santa Cruz, CA); anti-green fluorescent protein (GFP) antibodies were from Roche Diagnostics (Mannheim, Germany) and Santa Cruz Biotechnology; anti-Myc antibodies were from Cell Signaling Technology (Danvers, MA) and Millipore (Billerica, MA); and anti-glutathione transferase (GST) and anti-His antibodies were from Millipore. Anti-PKD2 and anti-PKD3 antibodies were purchased from Bethyl Laboratories (Montgomery, TX). Anti-PKD2 antibody for immunocytochemistry was purchased from Allele Biotech (San Diego, CA). Alexa 488-, Alexa 594-, and Alexa 647-labeled anti-mouse or anti-rabbit immunoglobulin G were purchased from Invitrogen. Brefeldin A (BFA) was purchased from Biomol (Hamburg, Germany). All other reagents were of the highest grade available.
pEGFP-PKD2-WT, D695A, S706/710E, ΔCRD, ΔC1a, ΔC1b, ΔPH, and ΔKD have been described previously (Auer et al., 2005 ). Deletion of amino acids (aa) 1-137 (PKD2-Δ1-137) was performed in pcDNA3-PKD2-WT (Auer et al., 2005 ) with a 5′ sense primer (5′-gccaccttcgaggaattccagat-3′) containing an EcoRI and a 3′ antisense primer (5′-cattgctgggatcctgtgaagaacc-3′) containing a BamHI site. The fragment was cloned into an EcoRI- and BamHI-digested pEGFP-C2 vector (BD Biosciences Clontech, Erembodegem, Belgium). Deletion of aa 323-368 (PKD2-Δ323-368) was performed with pcDNA3-PKD2-WT as a template by a splice-overlap polymerase chain reaction (PCR) strategy using Taq polymerase (Invitrogen). These primers used resulted in the deletion of aa 323-368: 5′FP (5′-acgatgacgataaagaattcgccaccgcc-3′), 3′Δ (5′-aaccctccgcagcgtggtgcagtcattagggacgcgggtgg-3′), 5′Δ (5′-ccacccgcgtccctaatgactgtgcaccacgctgcgggagggtt-3′), and 3′RP (5′-ggcaactagaaggcacagtcgaggctgat-3′). The final PCR product obtained was then subcloned into a pEGFP-C2 vector to yield pEGFP-PKD2-Δ323-368. Site-specific mutations within EGFP-PKD2-WT, EGFP-PKD2-D695A, EGFP-PKD2-ΔC1b, GFP-PKD1, and GFP-PKD3, resulting in single and double aa substitution mutants (EGFP-PKD2-P149G, EGFP-PKD2-P275G, EGFP-PKD2-D695A-P275G, EGFP-PKD2-ΔC1b-S706/710E, GFP-PKD1-P287G, and GFP-PKD3-P282G), were performed by a PCR approach using the QuikChange site-directed mutagenesis system (Stratagene, La Jolla, CA). His-tagged C1b domain was generated by amplification of C1b domain from pEGFP-PKD2-WT using a 5′ sense primer (5′-cgtacggatccgccgcacaccttcctcatccacagc-3′) containing BamHI site and a 3′ antisense primer (5′-cgaatccatggtgcagtcattagggacgcgggtggc-3′) containing NcoI site. The fragment was cloned into BamHI- and NcoI-digested pRSET-B vector (Invitrogen). Myc-PKD2-WT, D695A, and D695A-P275G were generated by digestion of enhanced green fluorescent protein (EGFP)-PKD2 constructs with EcoRI and XhoI and then subcloning into pCMV-Tag 3B vector (Stratagene). GST-tagged ARF1 was generated by amplification of full-length human ARF1 from fetal brain cDNA library by PCR, using a 5′ sense primer (5′-gcggatccgggaacatcttcgccaac-3′) containing BamHI site and a 3′ antisense primer (5′-gcctcgagtcacttctggttccggag-3′) containing XhoI site. The fragment was cloned into BamHI- and XhoI-digested pGEX-6P1 (GE Healthcare, Uppsala, Sweden) vector. GST-ARF1-T31N and GST-ARF1-Q71L were generated by site-directed mutagenesis. GST-ARF1 Δ17-Q71L was generated using GST-ARF1 as template with a 5′ sense primer (5′-ctactggaattcatgcgcatcctcatggtgggcctg-3′) containing EcoRI site and a 3′ antisense primer (5′-cgataactcgagtcacttctggttccggagctgattgg-3′) containing XhoI site. The fragment was cloned into EcoRI- and XhoI-digested pGEX-6P1 vector. GFP-PKD1 and GFP-PKD3 were provided by Dr. Johan van Lint (Katholieke Universiteit Leuven, Leuven, Belgium). ARF1-monomeric red fluorescent protein (mRFP) was a kind gift from Dr. Julie Donaldson (National Institutes of Health, Bethesda, MD). ARF1-Myc was a gift from Dr. Jean Gruenberg (University of Geneva, Geneva, Switzerland). ARF1-T31N-HA, ARF3-T31N-HA, and ARF4-T31N-HA were provided by Dr. Juan S. Bonifacino (National Institutes of Health). ARF5-T31N-HA was provided by Dr. Gwyn Gould (University of Glasgow, Scotland, United Kingdom). pSRα-ARF6-T27N-HA was provided by Dr. Philippe Chavrier (Centre National de la Recherche Scientifique/Institut Curie, Paris, France). GST-ARF6-T27N was generated using pSRα-ARF6-T27N-HA as template with a 5′ sense primer (5′-tccccggaattcatggggaaggtgctatccaaaatc-3′) containing EcoRI site and a 3′ antisense primer (5′-cggccgctcgagctattaagatttgtagttagaggttaac-3′) containing XhoI site. The fragment was cloned into EcoRI- and XhoI-digested pGEX-6P1 vector. GST-ARF6-WT and GST-ARF6-Q67L were generated by site-directed mutagenesis. Signal sequence from human growth hormone fused to horseradish peroxidase (ss-HRP) was a gift from Dr. Frederic Bard (Institute of Molecular and Cell Biology, Proteos, Singapore). pEGFP-Furin was provided by Dr. Gary Thomas (Vollum Institute, Portland, OR). Vesicular stomatitis virus-G protein (VSV-G)-GFP was provided by Dr. Jennifer Lippincott-Schwartz (National Institutes of Health). All these constructs were confirmed by DNA sequence analysis.
Recombinant proteins were produced and purified as described previously (Cohen et al., 2007 ). In brief, Escherichia coli BL21 host strain was transformed with the pGEX-GST-ARF1, pGEX-GST-ARF6, or pRSET-B-His-C1b expression vectors. Single colonies were inoculated in a 50 ml of liquid broth medium with appropriate antibiotics and cultured overnight at 37°C. Overnight cultures were inoculated (2% inoculum) and grown to OD600 of 0.6–0.9 and induced with 1 mM isopropyl β-d-thiogalactoside for 4 h at room temperature. Bacterial cells were pelleted at 4°C, and the pellets were stored at −80°C. Proteins were purified from the bacterial lysates by glutathione-Sepharose 4B beads (for ARF1 and ARF6 constructs) and Nickel-nitrilotriacetic acid agarose (for His-C1b construct) following the manufacturer's instructions.
In vitro binding studies with His-C1b and GST-ARF1 or GST-ARF6 mutants were done as described previously (Cohen et al., 2007 ). Sepharose-bound GST-ARF1 and ARF6 proteins (~20 μg) and the purified His-tagged PKD2-C1b domain (~20 μg) were incubated for 3 h on ice in 0.4 ml of binding buffer (PBS, pH 7.2; 1 mM MgCl2, 1 mM DTT, 0.2% Triton, and 0.1% Tween 20). Beads (50 μl) were washed twice with binding buffer (1 ml), and the resulting protein complexes were analyzed by SDS-polyacrylamide gel electrophoresis (PAGE).
Immunoprecipitations and Western blotting were performed as described previously (von Blume et al., 2007 ). Band intensities were quantified using BioProfil BIO-1D software, version 12.04 (Vilber Lourmat Deutschland GmbH, Eberhardzell, Germany).
HEK293-T-cells expressing various PKD isoforms or mutants were lysed in radioimmunoprecipitation assay buffer (50 mM Tris-HCl, pH 8.0, 1% NP-40, 0.5% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 5 mM EDTA, 10% glycerol, 2.5 mM MgCl2, and protease and phosphatase inhibitor cocktail (Roche Diagnostics), and the extracts were incubated with 50 μg of GST-ARF1 immobilized on glutathione-Sepharose beads for 2 h at 4°C. Beads were washed extensively, resuspended in Laemmli buffer, and analyzed by SDS-PAGE and Western blotting.
HeLa cells were seeded on glass coverslips and transfected the following day. Twenty-four hours after transfection, the cells were fixed with 4% formaldehyde in PBS for 10 min, blocked, and permeabilized with blocking buffer (0.05% saponin and 1% bovine serum albumin in PBS) for 30 min. The coverslips were incubated with primary antibodies diluted in blocking buffer for 1 h at room temperature, washed, incubated with secondary antibodies diluted in blocking buffer for 1 h, and then washed and mounted using FluorSave reagent (Calbiochem, Nottingham, United Kingdom). Imaging was performed with a TCS SPE confocal laser scanning microscope (Leica, Wetzlar, Germany), with 405 (Hoechst), 488 (GFP), 532 (mRFP and Alexa Fluor 594), and 635 nm (Alexa Fluor 647) laser lines and a 63× oil objective or with an LSM 510 confocal laser scanning microscope (Carl Zeiss, Jena, Germany), with 488, 543, and 633 nm laser lines and a 63× oil objective. Images were processed using ImageJ (http://rsbweb.nih.gov/ij/). The fluorescence intensity of either endogenous PKD2 or overexpressed EGFP-PKD2 within a standardized perinuclear region was measured using the quantification tool in ImageJ.
ss-HRP secretion assay was performed as described previously (Bard et al., 2006 ).
HeLa cells were cotransfected with constructs encoding for a temperature-sensitive viral VSV-G-GFP glycoprotein along with either PKD2 wild type (WT) or mutants and cultured at 39.5°C for 16 h. Then, 100 μg/ml cycloheximide was added before 2-h incubation at 20°C. The cells were then shifted to 32°C in the presence of cycloheximide and subsequently fixed at 15, 45, and 90 min, and processed for immunofluorescence microscopy. Imaging was performed with a TCS SPE confocal laser scanning microscope (Leica) with a 63× oil objective. The fluorescence intensity within a standardized perinuclear region was measured using ImageJ (http://rsbweb.nih.gov/ij/).
siGENOME nontargeting small interfering RNA (siRNA) and siRNA oligonucleotides targeting the 5′ untranslated region of PKD2 (5′-cuggguucuagauccgcgguu-3′) and PKD3 (5′-aaggugaaauccucuucguuu-3′) were from Dharmacon RNA Technologies (Lafayette, CO). The day before transfection, HeLa cells were plated to ensure 50% confluence on the day of transfection. Knockdown transfections were performed using 20 nM purified siRNA and HiPerfect (QIAGEN, Hilden, Germany). After 24 h of first round of siRNA transfection, HeLa cells were retransfected with siRNA and later followed by transfection with ss-HRP and EGFP expression vector or EGFP-PKD2 vectors (EGFP-PKD2-WT, EGFP-PKD2-ΔC1b, or EGFP-PKD2-P275G) using Lipofectamine LTX (Invitrogen). ss-HRP activity was measured as described previously (Bard et al., 2006 ).
The statistical significance of the difference between means was determined using the two-tailed Student's t test. Differences were considered significant at p < 0.05.
Because ARFs are prominent Golgi resident proteins and assemble protein complexes at the Golgi, we first we examined whether there was a physical interaction between ARF1 and PKD2. We performed pull-down assays of exogenously expressed GFP-PKD1, EGFP-PKD2, and GFP-PKD3 using GST-ARF1. All three PKD isoforms interacted with GST-ARF1 (Figure 1A). In addition, we performed a pull-down assay with immobilized ARF1 and HeLa cell lysates to detect endogenous PKD2 that is bound to ARF1. As depicted in the Figure 1B, we observed that a significant proportion of endogenous PKD2 (~10% of the input) interacted with immobilized ARF1. This interaction was confirmed by coimmunoprecipitation assays between EGFP-PKD2 and ARF1-Myc expressed in HEK293-T-cells. As shown in Figure 1C, EGFP-PKD2 was detectable in Myc-ARF1 immunoprecipitates, and, vice versa, Myc-ARF1 was found in EGFP-PKD2 immunoprecipitates. The amount of coimmunoprecipitated EGFP-PKD2 and ARF1-Myc was ~3 and 5%, respectively, compared with the immunoprecipitated ARF1-Myc and EGFP-PKD2.
GTP-bound, active ARF1 recruits coat components and various effectors that constitute the critical machinery of the vesicle budding process at the Golgi (D'Souza-Schorey and Chavrier, 2006 ). To determine whether the interaction between PKD2 and ARF was dependent on ARF1 activity, we examined the binding of constitutively active and inactive ARF1 mutants. PKD2 binding to the active ARF1 mutant ARF1-Q71L was twofold higher compared with ARF1-WT or the dominant-negative ARF1-T31N mutant (Figure 1D). A similar, twofold increase in binding of PKD2 to ARF1 was obtained when ARF1-Δ17-Q71L was used, a soluble and fully active mutant form of ARF1 (Boehm et al., 2001 ; (Figure 1D). Thus, PKD2 binds to both active and inactive ARF1 mutants but the amount of PKD2 bound to ARF1 was enhanced when ARF1 existed in active confirmation.
So far, we have established that PKDs and ARF1 interact physically. It was now important to determine whether there was a spatial relationship between PKD2 and ARF1 at the Golgi compartment. Coexpression of EGFP-PKD2 and ARF1-mRFP in HeLa cells revealed that both proteins colocalized at the TGN as determined by costaining with TGN46, a resident enzyme of the TGN and trans-Golgi (Figure 2A). There was also a colocalization of ARF1-mRFP with the other two PKD isoforms, GFP-PKD1 and GFP-PKD3 (Figure 2, B and C). In addition, we observed colocalization of endogenous PKD2 with overexpressed ARF1-mRFP (Figure 2D).
To determine whether ARF proteins could play a role in the recruitment of PKD2 to the TGN, we expressed dominant-negative mutants of ARF1, -3, -4, -5, and -6. Expression of these ARF mutants renders the specific endogenous ARFs inactive, presumably by binding to and sequestering ARF-guanine-nucleotide exchange factors (Dascher and Balch, 1994 ). In the presence of ARF1-T31N, ARF3-T31N, and ARF5-T31N, endogenous PKD2 was largely detectable in the cytosol (Figures 2, F, G, and I, respectively). ARF4-T31N overexpression displayed a moderate effect on the redistribution of endogenous PKD2 from the perinuclear region to the cytosol (Figure 2H). In contrast, upon coexpression of an inactive ARF6-T27N mutant, there was no change in PKD2 localization and PKD2 exhibited the well-known perinuclear localization (Figure 2, J and K). These data were further confirmed by transient overexpression of EGFP-PKD2 and ARF inactive mutants and immunostaining of the TGN compartment (Supplemental Figure S1). These data suggest that the localization of PKD2 to the TGN seems to be mainly regulated by class I and II ARFs. ARF6, the sole member of class III ARFs, seems to have no influence on localization of PKD2 to the TGN.
Having established that PKD2 physically interacts with ARF1, colocalizes with ARF1 at the TGN and that recruitment of PKD2 to the TGN is regulated by class I and II ARFs, we aimed to identify the site in PKD2 that interacts with ARF1. We used various PKD2 deletion mutants, some of which we described previously (Auer et al., 2005 ; Figure 3A). All mutants exhibited a similar level of expression except for PKD2-Δ1-137 (Figure 3B). The interaction of PKD2 with ARF1 was only abolished upon deletion of the cysteine-rich zinc finger domain. Further analysis revealed that a PKD2 mutant lacking only the C1b domain also failed to interact with ARF1, whereas a PKD2 mutant lacking only the C1a domain could still interact with the GTPase (Figure 3C). Thus, the C1b, but not the C1a domain of PKD2 is critical for its interaction with ARF1. These data were further confirmed by an in vitro binding assay using purified recombinant ARF1 and the C1b domain of PKD2. The C1b domain interacted exclusively with ARF1, but not with ARF6 (Figure 3D). Also in vitro, binding of the recombinant PKD2-C1b domain to the active ARF1 mutants ARF-Q71L and ARF1-Δ17-Q71L was increased by 1.5- to 1.75-fold compared with wild-type or dominant-negative ARF1 (Figure 3D). In addition, we observed a similar result when we performed in vitro direct binding assays with the recombinant PKD2-C1b domain and purified GST-ARF1 that was preloaded with either guanosine 5′-O-(3-thio)triphosphate (GTPγS) or guanosine diphosphate (GDP), respectively. PKD2-C1b domain interacted to a similar degree with wild-type ARF1 and GDP-loaded ARF1, but the amount of PKD2-C1b bound to GTPγS-loaded ARF1 was 1.5-fold higher compared with wild-type and GDP-loaded ARF1 (Supplemental Figure S2).
Next, we examined the role of the C1b domain in the recruitment of PKD2 to ARF1 at the TGN. A PKD2 mutant lacking the C1b domain (PKD2-ΔC1b) failed to colocalize with ARF1. Whereas ARF1-mRFP was largely localized in the perinuclear area corresponding to the Golgi, PKD2-ΔC1b was predominantly detectable in the cytoplasm, but not in the perinuclear area (Figure 3E, bottom, and F; data not shown). This demonstrates that the C1b domain of PKD2 is critical for its colocalization with ARF1 and the recruitment of the kinase to the TGN.
Previous data show that the C1a domain of PKD1 is necessary for DAG binding (Maeda et al., 2001 ; Baron and Malhotra, 2002 ) and that this binding is required for the localization of PKD1 at the TGN. Both PKD2-ΔC1a and PKD2-ΔC1b did not localize to the Golgi (Figure 3F). Thus, Golgi targeting of PKD2 requires both, the C1a domain and the C1b domain of PKD2.
PKD2 exhibits reduced catalytic activity when it lacks the C1b domain (Auer et al., 2005 ; Supplemental Figure S3A). Because the catalytic activity of PKD1 is required for its efficient recruitment to the TGN (Maeda et al., 2001 ), we compensated the reduced catalytic activity of PKD2-ΔC1b by introducing phosphomimetic glutamic acid residues at the critical serine residues within the activation loop of the kinase. The PKD2-ΔC1b-S706/710E mutant exhibited an elevated level of catalytic activity comparable to wild-type PKD2 that was well above that of catalytically inactive PKD2 (Supplemental Figure S3A). However, the PKD2-ΔC1b-S706/710E mutant neither interacted (Supplemental Figure S3B) nor colocalized with ARF1 (Supplemental Figure S3C) and did not localize to the TGN (Supplemental Figure S3D), demonstrating that the lack of catalytic activity was not responsible for the lack of Golgi targeting of PKD2-ΔC1b.
Next, we wanted to identify the region within the C1b domain of PKD2 required for the interaction with ARF1. There are critical Pro residues within the zinc finger domains that are conserved in all the three PKD isoforms (Figure 4A). Pro155 within the C1a domain of PKD1 is necessary for the binding of the kinase to DAG and for its localization at the TGN (Maeda et al., 2001 ; Baron and Malhotra, 2002 ). Pro155 and 287 within the C1a and C1b domain of PKD1, respectively, have been described as important residues for plasma membrane localization of PKD1 in lymphocytes (Spitaler et al., 2006 ).
Mutation of the conserved Pro within the C1a domain of PKD2 (EGFP-PKD2-P149G) abolished the perinuclear, TGN localization of the kinase as determined by TGN46 costaining and exhibited a largely cytosolic localization (Figure 4B, top), similar to previous studies using the corresponding PKD1-Pro155G mutant (Maeda et al., 2001 ). Interestingly, the exchange of Pro at position 275 in the C1b domain of PKD2 to Gly also resulted in a predominantly cytosolic localization of this mutant (Figure 4B, bottom). In addition, EGFP-PKD2-P275G failed to colocalize with ARF1 (Figure 4C). The interaction of EGFP-PKD2-P275G with ARF1 was also reduced by 80% compared with wild-type PKD2 in a pull-down assay (Figure 4D). Occasionally, we observed a double band for PKD2-ΔC1b and PKD2-P275G. This is most likely due to C-terminal degradation of the protein. The critical role of Pro275 in the binding of PKD2 to ARF1 was further confirmed in an in vivo coimmunoprecipitation assay. The amount of PKD2 detectable in ARF1-immunoprecipitates was reduced by 80% when the Pro275 was mutated compared with the wild-type PKD2 (Supplemental Figure S4). These data indicate that Pro275 in the C1b domain is critical for binding ARF1 and also critical for recruiting PKD2 to the TGN. The mutation of the corresponding conserved Pro residues in PKD1 and PKD3, Pro287 and Pro282, respectively, also resulted in a complete loss of TGN localization of these PKD isoforms (Figure 4E).
Expression of a catalytically inactive PKD1 or PKD2 leads to tubulation of the TGN and thereby inhibits vesicle shedding and protein transport (Liljedahl et al., 2001 ; Figure 4F, top). A P155G mutation in the C1a domain of kinase dead PKD1 prevents post-Golgi tubulation normally induced by catalytically inactive PKD1 (Maeda et al., 2001 ). Expression of a kinase dead EGFP-PKD2 additionally mutated at Pro275G in the C1b domain of PKD2 (EGFP-PKD2-D695A-P275G) in HeLa cells abolished post-Golgi tubulation normally observed upon expression of the catalytically inactive mutant (Figure 4F, bottom).
Short-term BFA treatment induces tubulation of the early Golgi cisternae, TGN, and the endosomes (Lippincott-Schwartz et al., 1991 ). Under these conditions, PKD1 does not dissociate from the TGN because DAG can still serve as the receptor for PKD1 at the TGN. However, PKD1 redistributes into the BFA-induced tubules positive for TGN46 and furin, a recycling TGN-plasma membrane-endosomal marker (Maeda et al., 2001 ). On short-term (5-min) BFA treatment of HeLa cells, wild-type PKD2 also redistributed into BFA-induced, furin-positive tubules in line with the data obtained with PKD1 (Figure 5A). However, both EGFP-PKD2-ΔC1b and EGFP-PKD2-P275G did not exhibit any redistribution to BFA-induced tubules (Figure 5, B and C). This further proves that these mutants are not recruited to the TGN in the first place and can therefore be not redistributed into furin-positive tubules upon BFA treatment.
Next, we asked whether Pro275 in PKD2 was also critical for the PKD2-mediated transit of proteins from the TGN to the plasma membrane. We used ss-HRP as a marker for vesicular transport. ss-HRP is a secretory marker in which the signal sequence from human growth hormone (hGH) is fused to HRP (Connolly et al., 1994 ). When this construct is expressed in cells, it is synthesized and transported along the secretory pathway. Because it has the characteristic signal sequence, it is exported to the plasma membrane and secreted into the supernatant of cells. The secreted protein can be quantified by measuring the chemiluminescence signal obtained by adding enhanced chemiluminescence substrate (Bard et al., 2006 ). In cells expressing wild-type PKD2, ss-HRP was efficiently secreted into the supernatant (Figure 6A). The amount of ss-HRP secreted was not significantly affected in cells expressing PKD2-ΔC1b or PKD2-P275G (Figure 6A) as these mutants do not localize to the Golgi, do not interact with ARF1, and are unlikely to compete with endogenous PKDs for the interaction with ARF1 at the TGN. Therefore, endogenous PKDs can act normally at the TGN and regulate protein transport. A similar secretory phenotype was observed with the green fluorescent protein-tagged ts045 mutant VSV-G-GFP, a well-characterized exocytic marker (Bergmann, 1989 ). In cells expressing wild-type PKD2, VSV-G was also efficiently transported to the plasma membrane, and the transport of VSV-G was not affected in cells expressing PKD2-ΔC1b or PKD2-P275G (data not shown).
To further determine whether PKD2-ΔC1b and PKD2-P275G had any functional activity at the TGN, we expressed these mutants in HeLa cells that were depleted of endogenous PKDs by specific siRNAs. HeLa cells express predominantly PKD2 and PKD3 (Bossard et al., 2007 ). The siRNA-mediated knockdown of PKD2 and PKD3 in HeLa cells resulted in efficient depletion (70–80%) of both PKD2 and PKD3 protein (Supplemental Figure 5A). After siRNA knockdown of PKD2 and PKD3, HeLa cells were transfected with a plasmid encoding ss-HRP together with an EGFP-expression vector, wild-type PKD2, EGFP-PKD2-ΔC1b, or EGFP-PKD2-P275G. The expression level of the various EGFP-tagged PKD2 constructs was comparable (Supplemental Figure 5B). Knockdown of PKD2 and PKD3 resulted in a substantial decrease of HRP secretion in line with previous data (Bossard et al., 2007 ; (Figure 6B). Expression of wild-type PKD2 in HeLa cells depleted of endogenous PKD2 and PKD3 rescued HRP secretion. In marked contrast, neither expression of EGFP-PKD2-ΔC1b nor of EGFP-PKD2-P275G resulted in a significant increase in secreted HRP and therefore failed to rescue the secretory block imposed by the PKD2/PKD3 knockdown (Figure 6B). These rescue experiments demonstrate the critical role of the C1b domain and particularly Pro275 in PKD2 for TGN targeting of the kinase and subsequently PKD2 function at the TGN, i.e., the regulation of protein transport from the TGN to the plasma membrane.
Having demonstrated that PKD2-ΔC1b and PKD2-P275G cannot rescue the phenotype induced by knockdown of PKDs due to a lack of Golgi localization, these mutants should also modify the effect of kinase dead PKD2. In cells expressing kinase-dead PKD2 (PKD2-D695A) that localizes to the Golgi and can compete with endogenous PKDs, VSV-G accumulated within TGN tubules that were positive for PKD2-D695A, and protein transport was substantially interrupted (Figure 6C), in line with previous data (Liljedahl et al., 2001 , Yeaman et al., 2004 ). Fluorescence quantification confirmed that cells expressing PKD2-D695A exhibited a 1.7-fold increase in VSV-G protein at the Golgi region (Figure 6D). Interestingly, cells expressing a PKD2-D695A-P275G double mutant did not exhibit an increase in VSV-G protein at the Golgi region, indicating that this mutant failed to block VSV-G transport. This shows that the Pro275 mutation prevents Golgi targeting of the kinase dead, dominant-negative–acting kinase and consequently its dominant-negative effect on endogenous PKDs and their role in vesicle shedding and protein transport from the TGN (Figure 6, C and D). This finding further confirmed that Pro275 in the PKD2-C1b domain is crucial for the function of the kinase at the TGN. A similar secretory phenotype was observed with HRP protein secretion. The P275G mutation also reversed the inhibitory effect of PKD2-D695A on HRP secretion (Figure 6E).
PKDs play a major role in regulating protein transport from the TGN to the plasma membrane. There has been a substantial amount of work elucidating the mechanisms by which PKDs regulate vesicle shedding. A local pool of DAG plays a major role in the recruitment of PKD to the TGN (Bard and Malhotra, 2006 ; Bossard et al., 2007 ). Active PKD phosphorylates PI4KIIIβ and CERT, the two important PKD substrates identified so far at the TGN. The PKD-mediated phosphorylation of PI4KIIIβ and CERT is critical in regulating the cross-talk between the membrane lipid biogenesis and protein secretion. This in turn regulates the maintenance of local DAG and PKD tethering to the TGN, which ultimately ensures a controlled vesicular transport process from the TGN (Hausser et al., 2005 ; Bard and Malhotra, 2006 ; Fugmann et al., 2007 ). Previous work demonstrated that the C1a domain of PKDs is crucial for the localization of PKDs at the TGN via binding of DAG (Maeda et al., 2001 ). However, the precise mechanisms how PKDs are recruited to the TGN are as yet less clear.
The small GTPases of the ADP-ribosylation factor family are also master regulators of the structure and function of the Golgi complex. Among the three classes of the ARF family, class I and II were reported to exert their function at the Golgi compartment. Active ARF1 recruits COPI coats which interact with the bona fide cargo proteins and generate functional vesicles that operate in the intra-Golgi and Golgi-endoplasmic reticulum retrograde trafficking zones of the membrane trafficking process (Orci et al., 1993 ). In addition, ARF1 was also shown to be one of the major components of the sorting machinery and is involved in controlling multiple TGN exit pathways (De Matteis and Luini, 2008 ). Many of the effectors and regulators of ARF1 play an important role in the formation and scission of vesicles destined for distinct compartments of the cell.
Here, we demonstrate that PKD2 specifically and directly interacts with ARF1. Binding of PKD2 to ARF1 is affected by the nature of the nucleotide bound to the GTPase and the association of PKD2 with ARF1 is enhanced when the GTPase exists in active confirmation. However, there is also an interaction between PKD2 and inactive ARF1. This is in line with previous reports that described the association of ARF1 with effector proteins such as the HIV Nef protein and the μ subunit of the adaptor protein complex AP-4 that interact with ARF1 independently of the nucleotide status of the GTPase (Boehm et al., 2001 ; Faure et al., 2004 ).
Expression of dominant-negative mutants of ARFs that are locked in the GDP conformation serve as an important tool in studying the effect of different ARF isoforms on the subcellular localization of various effector proteins. We found that the localization of PKD2 at the TGN was regulated by class I and II ARFs, which are known to play indistinguishable roles at the Golgi complex.
ARF1 and PKD2 not only interact in vitro and in vivo but also colocalize at the TGN as demonstrated by immunocytochemistry. In addition, the interaction of PKD2 with ARF1 is specifically mediated by its C1b domain and Pro275 within this domain is the central amino acid required for the ARF1-PKD2 interaction. A PKD2 mutant lacking the C1b domain or exhibiting a P275G exchange not only fails to interact with ARF1 but also does not localize to the Golgi and is largely localized in the cytoplasm. This points to a crucial role for the interaction of PKD2 with ARF1 to target PKD2 to the Golgi compartment. Our study shows that ARF1 functions as an important receptor for PKD in addition to local pool of DAG at the TGN. This also explains the requirement of an additional mechanism to target PKD to the TGN, despite the fact that DAG is present at various cellular locations. In line with this conclusion, our data further demonstrate that the loss of the C1b domain or the P275G exchange abolish the functional activity of PKD2 at the Golgi compartment. Both mutants cannot rescue the block of protein transport from the TGN to the plasma membrane induced by siRNA-mediated knockdown of endogenous PKD2 and -3 in HeLa cells. Furthermore, a lack of the C1b domain or a P275G mutation also abolished the dominant-negative effect of kinase dead PKD2 on protein transport from the TGN.
In conclusion, these data suggest a novel model in which the localization of PKD2 to the TGN requires both, the C1a and C1b domain: ARF1 recruits PKD2 from the cytoplasm (Figure 7a) to the Golgi apparatus via binding of Pro275 in the C1b domain of PKD2. The kinase is then anchored at the TGN by interacting with DAG in the membrane via its C1a domain. Both processes are required to accomplish vesicle shedding (Figure 7b). When the C1b domain is deleted or the critical Pro275 in PKD2 is mutated, ARF1 binding is impaired resulting in cytoplasmic localization of PKD2 and loss of function at the Golgi (Figure 7c). These data provide the first link between the “classical” machinery regulating protein transport at the Golgi compartment, namely, ARF proteins, and PKDs and demonstrate that the direct interaction of both is crucial for efficient protein transport from the TGN to the plasma membrane.
We are thankful to Yuichi Wakana, Sandra Mitrovic, Juan Duran, Josse Van Galen and Kota Saito of the Malhotra lab for useful discussions. We thank Timo Zimmermann and Frank Dolp for support with the TCS SPE confocal microscope and LSM510 confocal laser scanning microscope. G.V.P is a fellow of the GRK1041 and International Graduate School in Molecular Medicine, Ulm. T. S. is supported by Deutsche Forschungsgemeinschaft (SFB 518/B3 and A15 and GRK1041).
This article was published online ahead of print in MBC in Press (http://www.molbiolcell.org/cgi/doi/10.1091/mbc.E09-09-0814) on January 20, 2010.