Casein kinase I associates with members of the centaurin-alpha family of phosphatidylinositol 3,4,5-trisphosphate-binding proteins.

Mammalian casein kinases I (CKI) belong to a family of serine/threonine protein kinases involved in diverse cellular processes including cell cycle progression, membrane trafficking, circadian rhythms, and Wnt signaling. Here we show that CKIalpha co-purifies with centaurin-alpha(1) in brain and that they interact in vitro and form a complex in cells. In addition, we show that the association is direct and occurs through the kinase domain of CKI within a loop comprising residues 217-233. These residues are well conserved in all members of the CKI family, and we show that centaurin-alpha(1) associates in vitro with all mammalian CKI isoforms. To date, CKIalpha represents the first protein partner identified for centaurin-alpha(1). However, our data suggest that centaurin-alpha(1) is not a substrate for CKIalpha and has no effect on CKIalpha activity. Centaurin-alpha(1) has been identified as a phosphatidylinositol 3,4,5-trisphosphate-binding protein. Centaurin-alpha(1) contains a cysteine-rich domain that is shared by members of a newly identified family of ADP-ribosylation factor guanosine trisphosphatase-activating proteins. These proteins are involved in membrane trafficking and actin cytoskeleton rearrangement, thus supporting a role for CKIalpha in these biological events.

Yeast CKI isoforms are involved in DNA repair (2)(3)(4). Recently, many reports (5)(6)(7)(8)(9)(10)(11)(12) indicate that they also play a role in cytokinesis and in vesicle trafficking especially in endocytosis. The functions of the mammalian isoforms are less well understood, but based on high homology with their yeast counterparts, they may have similar biological functions. CKI⑀ and CKI␦ play a role in the regulation of p53 (13,14). CKI⑀ has also been implicated in circadian rhythms in Drosophila (15,16) and in development by transducing the Wnt pathway (17,18). CKI␥ might play a role in cytokinesis and/or in membrane trafficking (19). CKI␣ has been shown to play a role in cell cycle progression (20) and in membrane trafficking (21,22). Recently, CKIs have been shown to be implicated in regulating the nucleocytoplasmic localization of some substrates (23,24).
Several substrates, including nuclear and cytosolic proteins and membrane receptors, have been reported to be phosphorylated at least in vitro by a CKI activity (reviewed in Ref. 1). CKI isoforms are thought to be constitutively active and second messenger-independent. However, it has been shown that CKI␦ and CKI⑀ are regulated by autophosphorylation (25)(26)(27)(28). CKI␣ is also autophosphorylated, but whether this has an effect on its activity is not well defined. CKI␣ is negatively regulated by PtdIns(4,5)P 2 (21). Moreover, CKI isoforms have been reported to phosphorylate some of their substrates only if they were previously phosphorylated by another kinase two or three residues carboxyl-terminal to the CKI phosphorylation site. In this way, the effect of CKI is dependent on other kinases. CKI␣ is present in cells in different spliced forms (1,29) exhibiting different substrate specificities and differences in their protein-protein interactions.
Although the yeast CKI isoforms have been well characterized, the functions of the mammalian CKI isoforms are much less known. Therefore, the identification of mammalian CKI substrates and CKI-binding proteins should help to clarify their cellular function(s). CKI␣ interacts with NF-AT4 (23), the paired helical filaments (30), G-protein-coupled receptors (31), and the AP-3 complex (22). CKI␣ also forms a complex with certain splicing factors but these interactions may be indirect (32).
In the present study, we have shown that CKI␣ interacts with centaurin-␣ 1 . Centaurin-␣ 1 is a PtdIns(3,4,5)P 3 -binding protein containing two PH domains (33)(34)(35) and a zinc finger motif similar to the one found in a newly identified family of ADP-ribosylation factor (ARF) guanosine trisphosphatase-activating proteins (GAP) (reviewed in Refs. 36 -39). The yeast protein that shows the highest homology to centaurin-␣ 1 , Gcs1, also contains a zinc finger motif that confers its ARF-GAP activity (40). Members of this family are involved in membrane trafficking and in actin cytoskeleton rearrangement. Our results suggest that CKI␣ plays a role in membrane trafficking and/or actin cytoskeleton rearrangement, thus confirming previous reports (21,22).

MATERIALS AND METHODS
cDNA Cloning-The cDNA corresponding to rabbit muscle CKI␣ was originally cloned in the pET 3 vector (Novagen) (41). CKI␣ cDNA from this clone was amplified by the polymerase chain reaction (PCR) using two oligonucleotides (5Ј-gggcatatggcgagcagcag-cggctccaagg-3Ј) and (5Јgggggatccttagaaacctgtgggggtttgggc-3Ј) in order to create a 5Ј NdeI site and a 3Ј BamHI site (both are underlined in sequences). Amplified cDNA was inserted in a pET-16b vector (Novagen) at NdeI/BamHI restriction sites to express CKI␣ as a histidine-tagged protein. The CKI␣ cDNA was inserted in a pcDNA3 vector (Invitrogen) from the original clone after being amplified by PCR using two oligonucleotides (5Ј-gggggatccgccaccatggcctacccctacgacgtgcccgactacgcccccgcgagcagcagcggctcc-3Ј) and (5Ј-gcggcggccgcttagaaacctgtgggggtttgggc-3Ј) in order to create a 5Ј BamHI site and a 3Ј NotI site (both are underlined in sequences). The Kozak sequence is shown in italics and the HAtagged sequence in bold. CKI␣ mutants were generated by PCR and cloned in the pSP72 vector (Promega) at the EcoRI and BamHI sites downstream of the T7 promoter. The oligonucleotides (the EcoRI and BamHI sites are underlined) used are 5Ј-ggggaattcgccaccatgtacaaactggtacggaagatcgg-3Ј; 5Ј-ggggaattcgccaccatgtacagagacaacagaacaaggc-3Ј; 5Ј-ggggaattcgccaccatgagcatcaatgcacatcttgg-3Ј; 5Ј-ggggaattcgccaccatgac-cagcctgccgtggcaagg3Ј; 5Ј-ggggaattcgccaccatgtacgagaagattagcgaaaagaagatgtcc-3Ј; 5Ј-gggggatccttaggtcctgaaaaggatgcgg-3Ј; 5Ј-gggggatccttagaaacctgtgggggtttgggc-3Ј. The construction of GST-centaurin-␣ 1 has been described previously (42). Centaurin-␣ 1 cDNA from that vector was recovered after digestion with EcoRI and XbaI and subcloned in a FLAG-cmv2 vector. HA-tagged CKI␣(D136N) in pcDNA3 is a gift from F. McKeon (23). FLAG-centaurin-␣ is from Trevor R. Jackson.
Identification of Centaurin-␣ 1 by Mass Spectrometry-After the final chromatography step during the purification of CKI␣ as a 14-3-3 kinase from brain (Sephacryl S-100 gel exclusion), fractions containing this kinase activity were pooled and loaded on 12.5% SDS-PAGE (43). Gels were stained for 5-10 min and then destained for the minimum time. The 45-kDa band was excised and subjected to trypsin (L-1-tosylamido-2-phenylethyl chloromethyl ketone-treated, Worthington) digestion. The extracts were then dried in a Speedvac vacuum centrifuge and made up to the injection volume with water for on-line liquid chromatography mass spectrometry. Electrospray mass spectrometry of in-gel digested protein was carried out as described (43), and the peptide map data were used in the PeptideSearch program.
Recombinant Protein Purification-Escherichia coli carrying GSTcentaurin-␣ 1 , GST-centaurin-␣, or histidine-tagged CKI␣ plasmid were grown overnight at 37°C in Liquid Broth medium containing 50 g/ml ampicillin and were diluted the following day (1/10) in the same medium. Culture was then continued until the absorbance (600 nm) of the bacterial growth reached 0.6. Expression of the tagged proteins was induced with 0.5 mM isopropyl ␤-D-thiogalactopyranoside for 3-5 h at 25°C. The fusion proteins were purified by affinity chromatography on glutathione-Sepharose beads or nickel columns (Amersham Pharmacia Biotech). Proteins were further purified on a MonoS column using the AKTA purifier (Amersham Pharmacia Biotech) and stored in 20 mM Tris, pH 7.4, 1 mM EDTA, 1 mM EGTA, 100 mM NaCl, and 50% glycerol at Ϫ20°C. Recombinant 14-3-3 was purified as described previously (43).
In Vitro Transcription and Translation-CKI␣ and 14-3-3 subcloned in pcDNA3 were expressed in vitro using a T7 TNT-coupled transcription/translation reticulocyte lysate assay (Promega Corp., Madison, WI). The reactions (50 l) were performed following the manufacturer's instructions using [ 35 S]methionine (Amersham Pharmacia Biotech) for 90 min at 30°C. The reactions were then diluted 4-fold with binding buffer (containing 0.1 or 1% Nonidet P-40 as indicated in figure legends) and split in two for incubation (as indicated in the figure legends) with 1 or 10 g of GST, GST-centaurin-␣, or GST-centaurin-␣ 1 at 30°C for 15 min. Glutathione-Sepharose beads and binding buffer (300 l) were added to the reactions and incubated at room temperature for an additional 1 h. The beads were washed 5 times with 1 ml of binding buffer and electrophoresed on SDS-PAGE. After staining/ destaining, the gels were incubated with Amplify TM (Amersham Pharmacia Biotech), dried, and exposed to film.
Cell Culture and Transfection-COS-7 cells were obtained from the European Collection of Cell Cultures. They were maintained in Dulbecco's modified Eagle's medium (Sigma) supplemented with 10% fetal calf FIG. 1. Co-purification of a 45-kDa protein with CKI␣. CKI␣ was purified from pig brain after four chromatography steps (43). The fractions containing CKI activity were electrophoresed on 10% SDS-PAGE (lane 2), and the gel was stained with Coomassie Blue. CKI␣ was previously identified by mass spectrometry (43). A total of seven other proteins co-purified with CKI␣ (indicated by arrows), suggesting that they may form protein complex(es). The major co-purifying band migrates at 45 kDa. Molecular mass standards (lane 1) are shown in kDa.
Co-immunoprecipitation-Cells were lysed with 1 ml of lysis buffer (25 mM Tris, pH 7.5, 137 mM NaCl, 1% Nonidet P-40, 10% glycerol, 1 mM DTT) containing a mixture of protease inhibitors (Roche Molecular Biochemicals), 1 mM sodium fluoride, 1 mM sodium orthovanadate, and 10 mM sodium pyrophosphate. Lysates were pre-cleared with Pansorbin cells (Roche Molecular Biochemicals) and centrifuged for 20 min at 15,000 ϫ g at 4°C. Mouse anti-FLAG M2 antibodies (Sigma) were added to the lysates for 2 h. Protein-A/G coupled to Sepharose (Amersham Pharmacia Biotech) was then added for an additional 1 h of incubation. The beads were washed 4 times with 1 ml of lysis buffer, and the proteins associated with the beads were resolved on 10% SDS-PAGE. Proteins were transferred onto nitrocellulose (Bio-Rad), and the presence of HA-CKI␣ was detected by Western blotting with a rat anti-HA (Roche Molecular Biochemicals) antibody and ECL detection (Amersham Pharmacia Biotech).
Affinity Chromatography with the CKI Peptide-Two rat brains were homogenized in 20 ml of lysis buffer (50 mM Tris, pH 7.5, 100 mM NaCl, 10% glycerol, 1 mM EDTA, 1 mM EGTA, 2 mM DTT) containing a mixture of protease inhibitors (Roche Molecular Biochemicals). Nonidet P-40 was then added to a final concentration of 0.1%, and the mixture was incubated at 4°C for 2 h with constant agitation and subsequently clarified by centrifugation at 15,000 ϫ g for 30 min, followed by ultracentrifugation at 100,000 ϫ g for 1 h. The resulting high speed supernatant was loaded onto a 1-ml Sulfo-Link (Pierce) column to which 1 mg of a peptide corresponding to residues 214 -233 (C-214 FNRTSLPWQGL-KAATKKQKY 233 ) of CKI␣ was coupled (according to the manufacturer's instructions). A cysteine was added at the amino terminus to allow efficient coupling to Sulfo-Link. Brain extract was also loaded onto a control column. The columns were washed with 50 ml of lysis buffer containing 0.1% Nonidet P-40 and with 50 ml of phosphate-buffered saline. Bound proteins were eluted with 10 ml of 50 mM Tris, pH 7.5, 1 M NaCl, 10% glycerol, 1 mM EDTA, and 2 mM DTT. Eluted fractions and the last 5-ml washes were concentrated on Centricon-10 (Amicon) and analyzed by immunoblotting. The presence of 14-3-3 proteins was detected using a rabbit polyclonal 14-3-3 Pan antibody (KK 1106) from our laboratory. Centaurin-␣ 1 was detected using a rabbit polyclonal anti-  1) was excised and subjected to trypsin digestion and electrospray mass spectrometry. Nine peptides were recovered (shown in boxes). By using the "Pep-tideSearch" program, they all matched with centaurin-␣ 1 and eight of them with centaurin-␣. Both aligned sequences are from rat (33,45). Note that the first peptide is not identical in sequence to the one that we found (from pig brain), but for easier interpretation of the figure, we aligned the sequence of centaurin-␣ and -␣ 1 from the same species (rat). The 45-kDa protein was purified from pig brain, and the actual sequence of the peptide is VPPFYYRPSASDCQLLR (the different amino acids are underlined). This peptide is identical in the pig centaurin-␣ 1 (34). B, a schematic representation of centaurin-␣ and centaurin-␣ 1 . Both these proteins contain a zinc finger motif (Zinc) related to the one found in ARF-GAP proteins. Centaurin-␣ differs from centaurin-␣ 1 by the presence of only one PH domain (PH) rather than two and by a 43-amino acid extension at the carboxyl-terminal end (the differences are shown by dashed boxes). The mass spectrometric analysis was able to differentiate between these two proteins. Indeed, one peptide (KFVLTER) is only present in centaurin-␣ 1 (in its first PH domain) and not in centaurin-␣. body (J49) that was raised against an amino-terminal peptide of centaurin-␣ 1 (33).

A Protein of 45-kDa Co-purifies with CKI␣ in Brain-We
have previously reported that 14-3-3 was phosphorylated on a novel site (44) and identified the kinase as CKI␣ after four conventional chromatography steps (43). A total of seven proteins with molecular masses ranging from 25 to 80 kDa were shown to co-purify with CKI␣ (Fig. 1), and we postulated that they may form protein complex(es) in brain. A protein migrating at 45 kDa represented the most abundant co-purifying protein as judged by Coomassie Blue staining (Fig. 1). The putative association between CKI␣ and the 45-kDa protein would appear to be of high affinity because the complex was observed after elution with 0.5 M NaCl during the two cationic exchange chromatography steps and after elution with 0.6 -0.8 M NaCl from the Affi-Gel blue column (the chromatography steps are described in Ref. 43). Experiments were performed to identify the 45-kDa protein and to elucidate whether it associates with CKI␣.
Identification of the 45-kDa Protein as Centaurin-␣ 1 , a PtdIns(3,4,5)P 3 -binding Protein-The 45-kDa protein band was subjected to in-gel trypsin digestion, and the mass of each peptide was measured by electrospray mass spectrometry. Analysis of the peptide mass map using the "PeptideSearch" program identified the 45-kDa protein as a PtdIns(3,4,5)P 3 binding protein. We identified nine peptides that matched with centaurin-␣ 1 and eight peptides that matched with centaurin-␣ ( Fig. 2A). Centaurin-␣ 1 shares high homology to centaurin-␣. Centaurin-␣ contains an extended carboxyl-terminal tail and only one PH domain compared with centaurin-␣ 1 (Fig. 2B). Therefore, peptide mass analysis differentiated these two proteins and identified the 45-kDa protein unequivocally as centaurin-␣ 1 (42). Centaurin-␣ 1 has been given different names in the literature including p42 IP4 (34) and PIP3BP (35). However, for clarity in the rest of this paper, we refer to it as centaurin-␣ 1 . CKI␣ Associates Physically with Centaurin-␣ 1 -We tested the potential interaction between CKI␣ and centaurin-␣ 1 using purified recombinant proteins. We show that CKI␣ associates with GST-centaurin-␣ 1 and not with GST (Fig. 3A). In addition, CKI␣ also binds directly to centaurin-␣ (Fig. 3A). FIG. 3. CKI␣ associates physically with centaurin-␣ and centaurin-␣ 1 . A, GST, GST-centaurin-␣ (GST-␣), and GST-centaurin-␣ 1 (GST-␣ 1 ) were incubated for 2 h at 4°C with purified histidine-tagged CKI␣. Glutathione-Sepharose beads were then added. Beads were washed and subjected to an in vitro kinase assay using a CKI-specific phosphopeptide substrate. The presence of 32 P incorporated into the peptide was quantified by liquid scintillation counting and represents CKI␣ activity associated with the beads. B, CKI␣ was expressed and 35 S-labeled using a T7 TNT-coupled transcription/translation reticulocyte cell-free system and incubated with 1 g of GST, GST-centaurin-␣ 1 (GST-␣ 1 ) or GST-centaurin-␣ (GST-␣) for 15 min at 30°C in the presence of 0.1% Nonidet P-40. Glutathione-Sepharose beads were added and incubated at room temperature for an additional 1 h. Beads were washed, and samples were analyzed by 10% SDS-PAGE followed by autoradiography (panel Pull down assay). An aliquot of the lysate from the in vitro transcription/ translation assay was loaded on the gel (panel Input). C, CKI␣ and 14-3-3 were synthesized as described above (panel Input). They were incubated with 10 g of GST or GST-centaurin-␣ 1 (GST-␣ 1 ) for 15 min at 30°C in the presence of 1% Nonidet P-40. Beads were washed, and samples were analyzed by 10% SDS-PAGE followed by autoradiography (panel Pull down assay).
Therefore, two distinct experiments showed that CKI␣ interacts physically with both centaurins. The levels of CKI␣ that associated with centaurin-␣ 1 remained the same in the presence of 0.1-1% Nonidet P-40 (data not shown).
In order to verify the specificity of this interaction, [ 35 S]methionine-labeled CKI␣ and 14-3-3 were incubated with either GST or GST-centaurin-␣ 1 (Fig. 3C). By using higher stringency (1% detergent), centaurin-␣ 1 indeed associates with CKI␣ but not with 14-3-3 even though the latter is expressed at much higher levels than CKI␣ (Fig. 3C, panel Input), thus showing the specificity of the interaction between centaurin-␣ 1 and CKI␣. In conclusion, centaurin-␣ and centaurin-␣ 1 associate specifically and directly with CKI␣. As rat centaurin-␣ (except its unique carboxyl-terminal tail) shares 94% identity with rat centaurin-␣ 1 (Fig. 2, A and B), one could imagine that they bind to CKI␣ via a common domain, thus eliminating the importance of the first PH domain of centaurin-␣ 1 and the carboxylterminal domain of centaurin-␣.
CKI␣ Associates in Cells with Centaurin-␣ and Centaurin-␣ 1 -We tested whether CKI␣ associates in cells with centaurin-␣ 1 . For that purpose, we transiently transfected different cell lines in order to check the expression of ectopically expressed CKI␣. However, the expression of CKI␣ was not detectable in cell lysates of PC12 and NIH3T3 cells, was low in HEK 293 cells, and only really detectable in COS-7 cells (data not shown). Therefore, we have used COS-7 cells for the transient transfection experiments as the other cell lines were not suitable for co-immunoprecipitation experiments due to the low expression level of CKI␣. COS-7 cells were co-transfected with HA-tagged CKI␣ and with FLAG-tagged centaurin-␣ or centaurin-␣ 1 . FLAG-tagged proteins were immunoprecipitated, and CKI␣ in the immunoprecipitates was revealed by Western blot using HA antibodies. This experiment shows that CKI␣ is pulled down in centaurin-␣ (Fig. 4, 3rd lane) and centaurin-␣ 1 (Fig. 4, 4th lane) immunoprecipitates and not in the controls (Fig. 4, 1st, 2nd and 5th lanes). Therefore, the experiment demonstrates that CKI␣ associates with both centaurins in COS-7 cells.
Centaurins Are Not Phosphorylated by CKI␣-We tested whether CKI␣ phosphorylates centaurin-␣ and centaurin-␣ 1 in vitro using recombinant proteins. Fig. 5A shows that GSTcentaurin-␣ and GST-centaurin-␣ 1 were not substrates for CKI␣ in vitro. As a positive control, 14-3-3 , a CKI substrate (43), was shown to be phosphorylated (Fig. 5A, lane 14-3-3 ). In addition, recombinant centaurin-␣ 1 is not phosphorylated by CKI␣ indicating that the GST tag itself does not confer conformational restraint and has an effect on the result (data not shown). However, CKI isoforms have been shown to phosphorylate some substrates only if they have been previously phosphorylated by another kinase two or three residues carboxylterminal to the CKI site. Therefore, our in vitro results do not completely eliminate the possibility that centaurins are not substrates for CKI␣. To investigate this possibility, FLAGtagged centaurin-␣ and centaurin-␣ 1 immunoprecipitated from COS-7 cells were subjected to an in vitro kinase assay using purified histidine-tagged CKI␣. However, immunoprecipitated proteins were not phosphorylated by purified CKI␣ (data not shown). Therefore, we conclude from our experiments that centaurin-␣ and centaurin-␣ 1 either purified or immunoprecipitated from COS-7 cells are not substrates for CKI␣. FIG. 5. Centaurins are not phosphorylated by CKI␣ and have no effect on CKI␣ kinase activity. A, 1 g of purified GST, GSTcentaurin-␣ (GST-␣), and GST-centaurin-␣ 1 (GST-␣ 1 ) were assessed for in vitro phosphorylation by purified histidine-tagged CKI␣ as described under "Materials and Methods." Reactions were stopped and electrophoresed on 10% SDS-PAGE. The gel was dried and autoradiographed. 14-3-3 was phosphorylated by CKI␣ as a positive control (lane 14-3-3). B, 0.1, 1, or 10 g of recombinant centaurin-␣ 1 or buffer (b) was preincubated for 15 min with CKI␣. The reaction was initiated with the addition ("ϩ") of the CKI substrate 14-3-3 . No substrate was added in ("Ϫ") as a control in case some degraded forms of centaurin-␣ 1 would be phosphorylated by CKI␣. C, 0, 0.1, 1, or 10 g of recombinant centaurin-␣ 1 or the corresponding volume of buffer (b) was incubated with CKI␣ as described above. The reaction was initiated with the addition of GST-14-3-3 or phosvitin.
Centaurin-␣ 1 Interacts with Residues 217-233 of CKI␣-Mammalian CKI␣ belongs to a family of seven isotypes that show a high degree of homology in their kinase domains and have variable amino-and carboxyl-terminal tails. Therefore, it would be interesting to identify the centaurin-␣ 1 -binding site in CKI␣ in order to elucidate whether centaurin-␣ 1 associates in a region specific to CKI␣ or one that is also present in other CKI isotypes. To address this question, we constructed a set of CKI␣ deletion mutants (Fig. 6A). These mutants were synthesized and labeled with [ 35 S]methionine in a cell-free coupled transcription/translation system (Fig. 6B) and incubated with GST or GST-centaurin-␣ 1 (Fig. 6C). The different mutants were expressed in similar amounts apart from mutants 217-325 and 233-325 that were less well synthesized (Fig. 6B). The results show that the amino-and carboxyl-terminal domains (mutants 17-325 and 17-287) of CKI␣ are not required for the association. In addition, centaurin-␣ 1 binds with high efficiency to all mutants apart from mutant 233-325. Although the levels of expression of mutants 217-325 and 233-325 in cell lysates were the same, mutant 217-325 bound to centaurin-␣ 1 whereas mutant 233-325 did not (Fig. 6, C and D), demonstrating that the binding occurs between residues 217 and 233 within the kinase domain of CKI␣. These residues belong to a loop between helices E and F of CKI␣ that has been suggested to be the target region for protein-protein interactions (46). In order to confirm that residues 217-233 represent the site of interaction with centaurin-␣ 1 , a brain extract was loaded onto a column containing this peptide or a control column. Centaurin-␣ 1 was found to associate to the peptide column (Fig. 6E). This association is highly specific as we did not detect the presence of 14-3-3 proteins that represent 1% of total brain proteins.
In conclusion, our results mapped biochemically the centaurin-␣ 1 -interacting site on a single loop of CKI␣ that had been proposed previously from the three-dimensional x-ray structure to be the site of interaction for CKI partners.
Centaurin-␣ 1 Associates with all Members of the CKI Family-Residues 217-233 of CKI␣ are well conserved in all CKI isoforms as indicated in Fig. 7A. Therefore, it was interesting to determine whether other CKI isoforms were able to associate with centaurin-␣ 1 . To that purpose, all mammalian CKI isoforms were synthesized and labeled with [ 35 S]methionine in a cell-free coupled transcription/translation system and incubated with GST or GST-centaurin-␣ 1 (Fig. 7B). All mammalian CKI isoforms were capable of associating with GST-centaurin-␣ 1 and not with GST alone (Fig. 7B). Interestingly, a FIG. 6. Centaurin-␣ 1 associates with the residues 217-233 of CKI␣. A, schematic representation of a series of deletion mutants of CKI␣ (WT, wild type). Residues 17-287 constitute the kinase domain. Residues 164 -189 represent a loop containing a kinesin homology domain (1). Residues 217-233 are found in a loop that has been proposed to be responsible for the interaction of CKI␣ with other proteins (46). The calculated molecular masses of wild type and deletion mutants of CKI␣ are shown in kDa. B, wild type and mutant CKI␣ were synthesized using an in vitro transcription/translation assay, and an aliquot of the reaction (0.5% input) was analyzed on 15% SDS-PAGE. C, the rest of the reaction was incubated with 10 g of GST or GST-centaurin-␣ 1 using the same protocol as described in Fig. 3B. The binding of CKI␣ was determined by analysis on 15% SDS-PAGE and autoradiography. D, a longer exposure of the experiment shown above is presented for the mutants 217-325 and 233-325 because of their lower expression compared with the other mutants, thus providing means of a better interpretation of the results. E, brain extracts were loaded onto a 1-ml peptide affinity column to which a peptide corresponding to residues 214 -233 (C-214 FNRTSLPWQGLKAAT-KKQKY 233 ) of CKI␣ was coupled (CKI peptide) or a control column (control). The columns were washed, and bound proteins were eluted with buffer containing 1 M NaCl. Eluted fractions (elut) and the last washes (wash) were subjected to SDS-PAGE and analyzed by immunoblotting using centaurin-␣ 1 or 14-3-3 antibodies. A brain extract was also analyzed as a positive control for the antibodies (brain lysate). mutant of CKI␦ deleted of its carboxyl-terminal domain (CKI␦ ⌬317) binds as well as wild type CKI␦ to centaurin-␣ 1 (Fig. 7B). This result is in agreement with data from Fig. 6C that show that the kinase domain of CKI␣ is sufficient for its association to centaurin-␣ 1 . These data have led us to strongly believe that centaurin-␣ 1 interacts with residues 217-233 of CKI␣ that are well conserved in all CKI isoforms. DISCUSSION In this report we have identified centaurin-␣ 1 as a novel CKI␣ partner based on the following evidence: (a) they copurified from brain after elution from four chromatography steps; (b) centaurin-␣ 1 associates in vitro with CKI␣ indicating that the binding is direct; (c) the binding is specific as centaurin-␣ 1 does not interact in vitro with 14-3-3 under the same conditions; (d) they form a protein complex in COS-7 cells as shown in immunoprecipitation experiments; (e) centaurin-␣ 1 interacts with residues 217-233 of CKI␣ using deletion mutants of CKI␣; and (f) centaurin-␣ 1 elutes from a peptide affinity chromatography column containing residues 214 -233 of CKI␣. CKI isoforms are characterized by a conserved core kinase domain and by variable amino-and carboxyl-terminal tails. We report here that centaurin-␣ 1 interacts with the kinase domain and not with the unique tails of CKI␣. Moreover, a mutant of CKI␦ deleted of its carboxyl-terminal domain binds to centaurin-␣ 1 as well as does CKI␦, suggesting that the kinase domain of CKI␦ represents the centaurin-binding site. In addition, the site of interaction within the kinase domain (residues 217-233) is present in a loop between two helices which has been proposed to represent an interaction domain for CKI targets (46). The residues within that loop are well conserved among the CKI family. Indeed, we have shown that all mammalian CKI isoforms are able to associate with centaurin-␣ 1 in vitro. This suggests that the same loop, present in all CKI isoforms, is responsible for the interaction with centaurin-␣ 1 .
Centaurin-␣ 1 belongs to a newly identified family of ARF-GAP proteins (reviewed in Refs. 36 -39). Members of this family share a cysteine-rich GAP domain and contain several other domains such as PH domains, SH3 domains, and ankyrin repeats. These proteins are involved in vesicle trafficking and in actin cytoskeleton rearrangement. Therefore, our data support a role for CKI␣ in these biological events, in agreement with previous reports (21,22). Indeed, CKI␣ interacts with and phosphorylates the clathrin adaptor AP-3 (22), which is involved in endocytosis. It is interesting to note that a genetic interaction between yeast CKI and AP-3 was identified previously (8). Moreover, CKI␣ has been found to co-localize in neurones with synaptic vesicle markers and phosphorylates some vesicle synaptic associated proteins (21). Interestingly, centaurin-␣ 1 has been shown to associate with presynaptic vesicular structures (53). An actin-associated protein kinase shown to be a member of the CKI family phosphorylates actin in vitro (54). The molecular mass of the kinase (37 kDa) suggests that it could be CKI␣, and we have shown that recombinant CKI␣ indeed phosphorylates actin. 2 In addition, the protein DAH (Discontinuous Actin Hexagon) that interacts with the actin cytoskeleton has been shown to be phosphorylated by CKI in vitro (55).
Members of the ARF-GAP family contain several domains for protein-protein interactions, and they have been shown to associate with a number of different proteins. This suggests that ARF-GAP proteins may act as scaffold proteins in addition to their function as GAP proteins. Whether other ARF-GAP proteins interact with CKI is not known. The ARF-GAP proteins Git1 and Git2 have been reported to regulate the internalization of some G-protein-coupled receptors (56 -58). CKI␣ has been shown to interact with and phosphorylate these G-protein-coupled receptors (31,59). In addition, most of the identified ARF-GAP proteins are involved in the Pak signaling pathway (reviewed in Ref. 36). Intriguingly, CKI␥2 has been found to interact with the adaptor molecule Nck (60) in a complex with Pak1 (61), thus raising the possibility that CKI may associate with other ARF-GAP proteins. Therefore, it would be important to investigate whether ARF-GAP proteins interact with CKI isoforms.
Gcs1, the budding yeast homologue of centaurin-␣ 1 , also contains a cysteine-rich domain that is necessary for its ARF-GAP activity (40). As yet, no ARF-GAP activity has been reported for centaurin-␣ 1 , but it is able of rescuing a ⌬Gcs1 strain mutant indicating that centaurin-␣ 1 and Gcs1 may have similar function(s) (62). Gcs1 has been shown to be necessary for 2 T. Dubois, S. K. Maciver, and A. Aitken, unpublished data.

FIG. 7.
Centaurin-␣ 1 associates with all members of the CKI family. A, sequence alignment of residues 217-233 of CKI␣ with those corresponding to the other CKI isoforms (␤-⑀). B, CKI isoforms (␣, ␦, ␦ ⌬317, ⑀, ␥1, ␥2, and ␥3) were synthesized as described in Fig. 3B. They were incubated with 10 g of GST or GST-centaurin-␣ 1 (GST-␣ 1 ) as described in Fig. 3C in the presence of 1% Nonidet P-40 and 0.1% bovine serum albumin. Beads were washed, and samples were analyzed by 12.5% SDS-PAGE followed by autoradiography (pull down assay). Aliquots representing 2% of total lysate were analyzed as a control (lysate 2%). the resumption of cell proliferation from stationary phase (63) and is involved in endocytosis (7). Gcs1 also plays a role in actin cytoskeleton regulation in vivo and binds to actin in vitro (64). As vesicle trafficking is closely associated to actin organization in yeast, Gcs1 may link vesicle trafficking and the actin cytoskeleton (64). Yeast CKIs (Yck1 and Yck2) were shown to suppress the Gcs1 blockage effects on cell proliferation and endocytosis (7). The membrane association of Yck2 was necessary for this effect (7). Yck1/2 is involved in cytokinesis, in bud development (5,6), and regulation of the actin cytoskeleton as yck ts mutants fail to depolarize the actin cytoskeleton during mitosis (5). Another link between Gcs1 and CKI is the ankyrin repeat protein Akr1p. Gcs1 has been shown to interact with Akr1p in yeast two-hybrid experiments (65). Akr1p and Yck1/2 regulate yeast endocytosis, and Akr1p regulates the plasma membrane localization of Yck1/2 (11). These authors proposed that the Yck1/2 membrane localization may involve other proteins such as Gcs1 (65).
Our data suggest that CKI␣ does not phosphorylate centaurin-␣ and centaurin-␣ 1 . In addition centaurin-␣ 1 has no effect on CKI activity. Therefore, what is the functional relevance of the interaction between CKI␣ and these PtdIns(3,4,5)P 3 -binding proteins? As CKI␣ does not contain a lipid binding domain, it may associate with membranes through interaction with other proteins. Centaurin-␣ 1 may represent one of these proteins, as has been proposed for its yeast counterpart (see above and Ref. 65). CKI␣ may also represent a downstream target for centaurin-␣ 1 as suggested by the results in budding yeast showing that CKIs suppress Gcs1 mutant phenotypes.
In conclusion, we have shown an interaction between CKI␣ and centaurin-␣ 1 , a member of the ARF-GAP protein family that is involved in membrane trafficking and actin cytoskeleton regulation. Our present results are in agreement with data reported previously suggesting a role for CKI in membrane trafficking and/or regulation of the actin cytoskeleton. Our findings are further supported by evidence of a genetic link between CKI and Gcs1 in budding yeast.