Sphingosine-dependent Protein Kinase-1, Directed to 14-3-3, Is Identified as the Kinase Domain of Protein Kinase Cδ*

Some protein kinases are known to be activated by d-erythro-sphingosine (Sph) or N,N-dimethyl-d-erythro-sphingosine (DMS), but not by ceramide, Sph-1-P, other sphingolipids, or phospholipids. Among these, a specific protein kinase that phosphorylates Ser60, Ser59, or Ser58 of 14-3-3β, 14-3-3η, or 14-3-3ζ, respectively, was termed “sphingosine-dependent protein kinase-1” (SDK1) (Megidish, T., Cooper, J., Zhang, L., Fu, H., and Hakomori, S. (1998) J. Biol. Chem. 273, 21834–21845). We have now identified SDK1 as a protein having the C-terminal half kinase domain of protein kinase Cδ (PKCδ) based on the following observations. (i) Large-scale preparation and purification of proteins showing SDK1 activity from rat liver (by six steps of chromatography) gave a final fraction with an enhanced level of an ∼40-kDa protein band. This fraction had SDK1 activity ∼50,000-fold higher than that in the initial extract. (ii) This protein had ∼53% sequence identity to the Ser/Thr kinase domain of PKCδ based on peptide mapping using liquid chromatography/mass spectrometry and liquid chromatography/tandem mass spectrometry data. (iii) A search for amino acid homology based on the BLAST algorithm indicated that the only protein with high homology to the ∼40-kDa band is the kinase domain of PKCδ. The kinase activity of PKCδ did not depend on Sph or DMS; rather, it was inhibited by these sphingoid bases, i.e. PKCδ did not display any SDK1 activity. However, strong SDK1 activity became detectable when PKCδ was incubated with caspase-3, which releases the ∼40-kDa kinase domain. PKCδ and SDK1 showed different lipid requirements and substrate specificity, although both kinase activities were inhibited by common PKC inhibitors. The high susceptibility of SDK1 to Sph and DMS accounts for their important modulatory role in signal transduction.

The chaperone or modulatory protein 14-3-3 has received increasing attention for its interaction with and functional modification of various key molecules involved in signal transduction, transport, cell proliferation, and apoptosis (18). SDK1 specifically phosphorylates 14-3-3␤, 14-3-3, and 14-3-3 only in the presence of Sph and DMS, but not 16 other lipids tested (including Cer and Sph-1-P). The phosphorylation takes place at Ser 60 , Ser 59 , or Ser 58 , located in helix 3, which may interfere with interaction between two molecules of 14-3-3 to form a dimer. Non-phosphorylated 14-3-3 prefers to form a dimer and may function as a chaperone or modulator of major signal transducers such as Raf1, Bad, and Ber. SDK1-dependent phosphorylation may counteract this effect and thereby modulate signal transduction to induce cell proliferation, differentiation, or apoptosis. However, the molecular identity and biochemical properties of SDK1 have not yet been clearly elucidated. Here, we describe the characterization of SDK1 and identify it as the C-terminal half kinase domain of PKC␦, distinguishable from PKC␦ per se.
Procedure for Purification of SDK1 from Rat Liver 150 g of rat liver at one time was thawed, cut into small pieces, and homogenized in a Waring blender (30 s at high speed, three times) in ice in a cold room with 1000 ml of ice-cold homogenization buffer containing 20 mM Tris-HCl, 0.5 mM EDTA, 0.5 mM EGTA, 12 mM NaF, 5% glycerol, 3 g/ml leupeptin, 5 g/ml aprotinin, 3 g/ml chymostatin, 1 mM phenylmethylsulfonyl fluoride, and 1 mM dithiothreitol (DTT) (pH 8.0). The homogenate was centrifuged for 30 min at 4000 rpm in Beckman centrifuge. The supernatant was adjusted to pH 8.0 and ultracentrifuged for 1 h at 100,000 ϫ g. The supernatant was filtered through a 0.5-m filter and used for six steps of chromatographic purification, guided by SDK1 activity.
Step 1: Q-Sepharose Fast Flow Chromatography-The pH of the filtered extract described above was adjusted to 8.0 and applied to a Q-Sepharose Fast Flow XK 26/20 column (Amersham Biosciences) preequilibrated with Buffer A (20 mM Tris-HCl, 1 mM EDTA, 10 mM NaF, and 10 mM DTT (pH 8.0)) as described previously (16). Elution was performed by a combination of stepwise and linear gradient changes of the NaCl concentration (0 -1.0 M) under conditions of 4 ml/min and 8 ml/fraction. 10 l of each fraction was analyzed for SDK1 activity using 14-3-3␤ as substrate (see below). The activity was consistently found between fractions 22 and 31 (see Fig. 1A).
Step 2: Hydrophobic Interaction Chromatography-Fractions showing SDK1 activity from Step 1 were pooled, added with powdered KCl to give a final concentration of 1 M, filtered through a 0.5-m membrane, and applied to a phenyl-Sepharose 6 Fast Flow column (Amersham Biosciences). All SDK1 activity was recovered in the flow-through fraction with increased specific activity (see "Results").
Step 3: Cation-exchange Chromatography on a Heparin-Sepharose Column-The flow-through fraction (15 ml) from Step 2 was subjected to buffer exchange by pass-through a HiPrep 26/10 desalting column (Amersham Biosciences) and eluted with 2 column volumes (100 ml) of 50 mM MES (pH 6.7), 1 mM EDTA, and 10 mM NaF. Proteins were eluted with 15-20 ml of MES buffer prior to elution of KCl. The bufferexchanged sample was applied to a heparin column pre-equilibrated with the same MES buffer and subjected to gradient elution with the same buffer containing increasing NaCl concentrations under conditions of 1.5 ml/min and 4.5 ml/fraction. High SDK1 activity was consistently found between fractions 15 and 21 (see Fig. 1B).
Step 4: Hydroxylapatite Chromatography-The fraction with SDK1 activity resulting from Step 3 was buffer-exchanged to 10 mM KH 2 PO 4 . Elution was performed with increasing KH 2 PO 4 concentrations under conditions of 1.0 ml/min and 2 ml/fraction. Peak SDK1 activity was observed between fractions 10 and 20 (see Fig. 1C).
Step 6: Mono Q Chromatography-The SDK1 active fraction from Step 5 was buffer-exchanged to Buffer A (pH 8.0), placed on a Mono Q HR 5/5 column (1-ml bed volume; Amersham Biosciences) pre-equilibrated with Buffer A, and subjected to gradient elution with 0.15-0.5 M NaCl under conditions of 0.25 ml/min and 0.5 ml/fraction. High SDK1 activity was observed in fractions 17-21; the maximal peak was in fraction 18, 19, or 20 depending on the sample (in one case, shown in Fig. 1E, the peak was in fraction 20). A band with molecular mass of ϳ40 kDa was detected in the SDK1 active fractions. This band, separated by SDS-PAGE, was therefore subjected directly to sequence analysis after in-gel digestion.
Step 7: Size-exclusion Chromatography-The SDK1 active fraction from Step 6 was centrifuged at 10,000 ϫ g for 10 min to eliminate impurities, subjected to gel filtration through a Superdex 200 HR 10/30 column (24-ml bed volume; Amersham Biosciences), and eluted with 50 mM Tris-HCl (pH 7.0) and 0.15 M NaCl under conditions of 500 l/min and 500 l/fraction.

Determination of PKC␦ Activity
PKC␦ activity with the common PKC peptide substrate RFARKGSL-RQKNV (where S denotes the phosphorylation site; Panvera Corp.) was determined by incubating the following mixture (50-l total volume): 2 l of 0.5 M HEPES (pH 7.4), 5 l of 100 mM MgCl 2 , 5 l of 1 mM EGTA, 0.5 l of 100 M peptide substrate, 0.5 Ci of [␥-32 P]ATP, 6.7 l of 750 M ATP, 5 l 10ϫ lipid mixture (2 mg/ml phosphatidylserine (PS), 200 g/ml diacylglycerol, 20 mM HEPES (pH 7.4), and 0.3% Triton X-100), and 5 l containing 25 ng of PKC␦ in 10 mM HEPES (pH 7.4), 5 mM DTT, and 0.01% Triton X-100. The reaction mixture was incubated for 10 min at 30°C. The reaction was stopped by spotting 25 l of this mixture onto phosphocellulose membrane (P-81 paper, Upstate Biotechnology, Inc., Lake Placid, NY). The membranes were washed three to five times with 0.5% phosphoric acid and counted in a scintillation counter. PKC␦ activity was quantified as the amount of 32 P-phosphorylated peptide trapped on the membrane as measured in the scintillation counter.

Determination of SDK1 Activity in Chromatographic Fractions from Rat Liver
This was performed by a modification of the previously described method (16,19), including quantitative comparison of activities using a PhosphorImager (Cyclone storage phosphor screen, PerkinElmer Life Sciences). Briefly, 10 l of sample solution, i.e. aliquot of the chromatographic fraction, was mixed with 3 l of 50 mM Tris (pH 7.5) containing 1 g of 14-3-3␤ (1 l of a 1 g/l solution) and 0.15 l of 10 mM DMS in Me 2 SO (final DMS concentration of 50 M in the incubation mixture). The reaction was initiated by addition of 10 l of ATP/Mg 2ϩ solution (75 M unlabeled ATP and 2.5 Ci of [␥-32 P]ATP in Tris-HCl (pH 7.5), 3 mM DTT, and 45 mM MgAc) and H 2 O to adjust the total volume to 30 l. The reaction mixture was incubated for 15 min at 30°C, and the reaction was stopped by addition of 10 l of 4ϫ concentrated Laemmli sample buffer and heating for 3 min at 100°C. Phosphorylated 14-3-3␤ was separated by 12% SDS-PAGE and transferred to a polyvinylidene difluoride membrane. The membrane was air-dried, and the activity of phosphorylated 14-3-3␤ band was determined by contact of the membrane with the screen of the PhosphorImager. The intensities of each band were quantified by OptiQuant software, subtracted by the value for a control 14 C marker, and compared in terms of digital light units.

Cleavage of PKC␦ by Caspase-3, Determination of SDK1 Activity in the Released Catalytic Domain, and Effect of PKC␦ Inhibitors
PKC␦ was incubated with caspase-3 as described previously (20). Briefly, 200 ng of human recombinant PKC␦ was added to the following mixture with a total reaction volume of 20 l: 2 units of active human recombinant caspase-3 in 25 mM HEPES (pH 7.4), 250 mM sucrose, 1 mM EDTA, and 2.5 mM DTT. Cleavage was confirmed by Western blotting using a polyclonal antibody against the carboxyl terminus of PKC␦.
For SDK1 activity determination of caspase-3-treated PKC␦, 8.5 l of the above mixture (containing 85 ng of caspase-3-treated PKC␦) or of untreated full-length PKC␦ was incubated for 15 min at 30°C with 1 g Controls were carried out in the absence of DMS. The reaction was stopped by addition of Laemmli buffer (4ϫ concentrated) and heating at 95°C for 5 min. 14-3-3 phosphorylation was analyzed by 10 -12% SDS-PAGE. Gels were subjected to autoradiography using an intensifier screen (exposure time of 2 h at Ϫ80°C) or a PhosphorImager. In some reactions, 5-150 M rottlerin (Calbiochem) or 1-10 M GF109203X (Alexis Biochemicals, San Diego, CA) was used as a PKC␦ inhibitor. Other lipids (Sph-1-P, C 2 -Cer, and C 16 -Cer) were also tested to verify SDK1 activity.

Determination of a Partial Amino Acid Sequence of SDK1
One-dimensional SDS-PAGE-500 l of fraction 18, associated with very high SDK1 activity from Step 6 (Mono Q), was precipitated with an equal volume of 20% trichloroacetic acid, centrifuged at 16,000 ϫ g for 30 min, washed with 1 ml of cold ethyl ether, and centrifuged at 16,000 ϫ g for 10 min. Washing with cold ethyl ether was repeated three times to eliminate trichloroacetic acid. The final precipitate was dissolved in 10 l of sample buffer (62.5 mM Tris-HCl (pH 6.8) containing 10% glycerol, 2% SDS, 5% mercaptoethanol, and 0.0025% bromphenol blue). The resulting mixture was loaded onto a gel and separated by one-dimensional SDS-PAGE. The conditions were as follows: gel size of 130 ϫ 130 ϫ 1 mm; 4 and 10% acrylamide containing 2.6% piperazine diacrylamide/acrylamide (Bio-Rad) for the stacking and running gels, respectively; and running buffer consisting of 25 mM Tris and 192 mM glycine (pH 8.45) containing 0.1% SDS. Proteins were concentrated at 5 mA/gel for 70 min in stacking gel and run at 20 mA/gel for 2 h. Gels were stained with 0.1% Coomassie Blue R-350 (Bio-Rad) dissolved in 10% acetic acid and 30% methanol and washed with 7% acetic acid and 30% methanol for 20 min (three times) and with 10% acetic acid for 20 min (three times). The protein profile was analyzed by a Master Scan image analyzer (Scanlytics, Billerica, MA). Molecular masses of bands were calculated based on molecular mass standard proteins (broadrange; Bio-Rad).
In-gel Digestion with Trypsin and Extracted Peptides-The ϳ40-kDa band was subject to in-gel digestion as described previously (21). Briefly, the band was excised manually using a razor blade, placed in an Eppendorf tube, washed with H 2 O (10 min, 37°C, five times), and destained in 100 l of 50% CH 3 CN and 100 mM ammonium carbonate (pH 8.5) for 10 min at 37°C until colorless. The gel was dehydrated in 100 l of CH 3 CN in an Eppendorf tube for 10 min at 37°C and dried in Speed-Vac TM for 5 min. The dried residue was immersed in 50 l of 0.001% trypsin (10 ng/l) in 100 mM ammonium carbonate (pH 8.5) and incubated overnight at 37°C. The original trypsin solution was 20 g/20 l of 50 mM acetic acid. The incubation mixture in the Eppendorf tube was centrifuged, and the residue was extracted with 50% CH 3 CN and 0.1% trifluoroacetic acid and centrifuged again. The residue was further extracted with 15% isopropyl alcohol, 20% formic acid, 25% CH 3 CN, and 40% H 2 O and finally with 80% CH 3 CN. All of the extracts were successively dried in a single Eppendorf tube, and the residue was dissolved in 6 l of ultrapure water. Aliquots (2 l) were used for protein identification and for amino acid sequencing by mass spectrometry (MS).
Peptide Sequencing-Peptide mapping was performed using the API QSTAR pulsar hybrid mass spectrometer system with a micro-liquid chromatograph (Magic 2002, Michrom BioResource, Auburn, CA). Conditions of micro-LC were as follows: Magic C18 column (0.2 mm, inner diameter ϫ 50 mm) and elution with 0.1% formic acid (solvent A) and 0.1% formic acid in 90% CH 3 CN (solvent B) using a program of 3% solvent B for 2 min, gradient at 2.1%/min for 45 min, 100% solvent B for 5 min, and a flow rate of 2.5 l/min. The QSTAR pulsar hybrid mass spectrometer system consists of nanoelectrospray ionization as the ionization source and quadrupole time-of-flight MS. Mass accuracy was Ϯ0.1 mass unit. Conditions of MS were as follows: ion spray voltage of 3.0 -3.8 kV, voltage for the electron multiplier of 2200 V, nitrogen 10 curtain gas for MS and MS/MS analyses, and nitrogen 10 collision gas and collision energy of 20 -55 eV for MS/MS analysis. Conditions of micro-LC were as follows: Magic C18 column (0.2 mm, inner diameter ϫ 50 mm), elution with solvent A and solvent B using a program of 3% solvent B for 2 min, gradient at 2.1%/min for 45 min, 100% solvent B for 5 min, and a flow rate of 2.5 l/min.
To identify protein in the ϳ40-kDa band, peptide mapping of one-third of the in-gel digested product was performed by micro-LC/MS (QSTAR) using the PROWL (ProFound) search engine 2 and the NCBI Database (22,23). The major ion peaks of the total ion chromatogram were further analyzed to obtain the amino acid sequences of the tryptic peptides by LC/MS/MS using the Mascot search engine (22,24,25) 3 under the same conditions (QSTAR with a Magic micro-liquid chromatograph) and the same data base (one-third of the same sample used).
LCQ-DECA XP (ThermoFinnigan, San Jose, CA) with a Magic micro-liquid chromatograph was also used for determination of the amino acid sequences of the tryptic peptides using the MS/MS mode (one-third of the sample used). The conditions of MS were as follows: ion spray voltage of 1.8 kV, transfer tube temperature of 250°C for MS and MS/MS analyses, helium collision gas, and normalized collision energy of 35% for MS/MS analysis. Identification of probable sequences from MS/MS data was performed using the TurboSequest search engine (22,26,27) 4 and the NCBI Database (in house).

Inhibition of the SDK1 Activity of Caspase-3-treated PKC␦ by the 14-3-3 Peptide
Aliquots of the caspase-3-released kinase domain from 85 ng of PKC␦ were incubated (5 min, 30°C) with various concentrations of the 14-3-3 icosapeptide, previously identified as the SDK1-dependent phosphorylation site, or of the unrelated control peptide (17 amino acids) with the sequence described under "Materials." The incubation mixture was assayed for SDK1 activity with 14-3-3␤ as substrate, DMS as activator, and [␥-32 P]ATP, followed by determination of phosphorylated 14-3-3␤ separated by SDS-PAGE as described above.

RESULTS
Purification of Proteins with SDK1 Activity-The results of the chromatography steps (Q-Sepharose Fast Flow, heparin-Sepharose, hydroxylapatite, chromatofocusing, Mono Q, and gel filtration pattern of the fraction obtained by Mono Q) are shown in Fig.  1 (A-F, respectively). The results of purification by phenyl-Sepharose (Step 2) are essentially the same as described previously (16) and therefore are not shown. The SDK1 activity of each fraction at each chromatography step, expressed as digital light units, is also indicated in each panel of Fig. 1. The yield of proteins, their SDK1 activities, specific activities (SDK1 activity/mg of protein), and -fold purification are shown in Table I. An ϳ55,000-fold purification was obtained at the Mono Q step in comparison with the initial extract; specific SDK1 activity was increased to a similar degree. However, a large number of protein bands were still present, separated by SDS-PAGE, even at the Mono Q step. Perhaps some of these proteins were present as complexes and were extremely difficult to separate into components, even after many chromatography steps. Further separation was achieved only by SDS-PAGE, whereby SDK1 activity was lost because of denaturation.
The chromatographic fractions showing the highest SDK1 activity on the Mono Q column (fraction 18, 19, or 20, depending on the batch) showed a band with a molecular mass ϳ40 kDa upon SDS-PAGE, which was absent in other fractions with no or minimal SDK1 activity. Therefore, amino acid sequence analysis was performed for this band as described under "Experimental Procedures." The ϳ40-kDa Protein from a Fraction Containing SDK1 Activity Has the Same Sequence Found in the C-terminal Half Domain of PKC␦-Based on procedures described above, the in-gel digested product of the ϳ40-kDa band was analyzed by micro-LC/MS and MS/MS combined with QSTAR or LCQ-DECA XP, and the probable peptide sequence was studied using three online search engines (PROWL, Mascot, and TurboSequest) as described under "Experimental Procedures." The probable sequences of peptide fragments as determined by these three methods are indicated in Table II. The overlap from the three methods confirms the reality of a common sequence as shown in Table III. The sequence so far determined covered ϳ53% of that of the C-terminal half domain (catalytic domain) of PKC␦.
SDK1 Activity Is Harbored in the Catalytic Domain of PKC␦ Released by Caspase-3-Based on the sequence data in Tables II and III, we searched the NCBI Database with the BLAST algorithm. 5 The only protein with high homology to the sequence in Table II was rat PKC␦ (28), although the homology 2 Available at prowl.rockefeller.edu/cgi-bin.ProFound. 3 Available at www.matrixscience.com. 4 Available at fields.scripps.edu/sequest. 5 Available at www.ncbi.nlm.nih.gov/blast. Step 2 (phenyl-Sepharose column chromatography) is not shown because the pattern was was restricted to the C-terminal half (catalytic domain). Fulllength PKC␦ showed novel PKC␦ activity with the common PKC substrate, requiring diacylglycerol, PS, and Mg 2ϩ , but not Ca 2ϩ . However, the kinase activity of PKC␦ with the common PKC substrate (RFARKGSLRQKNV) was inhibited by Sph or DMS ( Fig. 2A). Strong SDK1 activity, highly dependent on the presence of Sph or DMS, was found when PKC␦ was treated with caspase-3 (Fig. 2B). Full-length PKC␦ did not phosphorylessentially the same as described previously (16).   ate 14-3-3 in the presence of DMS (Fig. 3A, group a), i.e. did not show typical SDK1 activity. In contrast, PKC␦ treated with caspase-3 showed strong SDK1 activity (Fig. 3A, group b), and release of an ϳ40-kDa band was indicated by Western blotting with antibody directed to the C-terminal region of PKC␦ (Fig.  3B). A similar strong SDK1 activity was observed for rat liver ϳ40-kDa SDK1 separated by Mono Q column chromatography without caspase-3 treatment (see below).

SDK1 Activities of the Rat Liver Fraction and of the Caspase-3-released Domain of PKC␦, Their Lipid Requirements, and
Their Susceptibility to PKC␦ Inhibitors and the 14-3-3 Icosapeptide-Over half (53%) of the amino acid sequence of the ϳ40-kDa protein present in SDK1 active fraction 18 (or 19) from the Mono Q column chromatography of rat liver extract (Fig. 1E) aligned identically with the C-terminal half kinase domain of the rat liver PKC␦ protein. The fraction showed strong SDK1 activity, similar to that of the caspase-3-released domain of PKC␦, and both of these activities were greatly enhanced by DMS and inhibited by PKC inhibitors (see below).
Lipid requirements for SDK1 activity were determined using the caspase-3-released domain of PKC␦. The activity was detectable only in the presence of Sph, DMS, or N,N,N-trimethylsphingosine, but not Sph-1-P, C 2 -Cer, or C 18 -Cer, similar to   Table II. The sequence is aligned with 53% of the catalytic domain sequence of PKC␦ (28). Lightface letters indicate probable sequence of the catalytic domain, initiated from the caspase-3-cleavable site (amino acids 1-346, corresponding to amino acids 328 -673 of PKC␦). The ATP-binding region signature is underlined; the Ser/Thr protein kinase active-site signature is double-underlined. reported results for SDK1 of 3T3-A31 cells (16,29). The SDK1 activity of the kinase domain determined in the presence of DMS was strongly inhibited by PS (Fig. 4), in contrast to PKC␦ activity (with PKC substrate), which was enhanced by PS. The contrasting properties of SDK1 and PKC␦ are listed in Table  IV. The SDK1 activities of both the ϳ40-kDa protein present in rat liver Mono Q fraction 18 or 19 (Fig. 5A) and the ϳ40-kDa protein released from PKC␦ by caspase-3 (Fig. 5B) were inhibited by rottlerin (preferential inhibitor of PKC␦) and by GF109203X (general inhibitor of PKC).

NNGTYGKIWE GSNRCRLENF TFQKVLGKGS FGKVLLAELK GKERYFAIKY LKKDVVLIDD
SDK1 Activity of the ϳ40-kDa Kinase Domain of PKC␦ Is Inhibited by the 14-3-3␤ Peptide-SDK1 activity was originally identified to phosphorylate Ser 60 in 14-3-3␤, Ser 59 in 14-3-3, and Ser 58 in 14-3-3 (16). We therefore tested the activity of the kinase domain released from PKC␦ by caspase-3 using an icosapeptide (20-amino acid peptide) that includes the phosphorylation site (S) of 14-3-3, i.e. YKNVVGARRSSWRVISSIEQ. Phosphorylation failed to occur in the peptide under the same conditions used to test the SDK1 activity of 14-3-3. However, the SDK1 activity of the catalytic domain released from PKC␦ was completely inhibited by this peptide at 120 M. In contrast, a control peptide with the unrelated amino acid sequence Ac-YGGSAESS(Aib)KSEASSK(Aib)SA-CONH 2 had no inhibitory effect on SDK1 activity (Fig. 6). DISCUSSION The Ser protein kinase that catalyzes Sph-or DMS-dependent phosphorylation of 14-3-3 at a defined site is termed SDK1 (16). SDK1 has been considered as a modulator of functional membrane proteins through 14-3-3 phosphorylation, which may affect dimer formation; however, the kinase involved has not been identified. We previously tried to perform affinity purification of this kinase employing the 14-3-3 icosapeptide that includes the phosphorylation site, but without success. We then initiated large-scale purification from rat liver as described in this study. The major difficulty was that the number of protein bands sep-arated by SDS-PAGE was not diminished after many steps of chromatography, even though total or specific SDK1 activity increased progressively at each step of purification. This suggests that multiple protein complexes were present and that their components were not separable by the various types of chromatography we employed. The components were separable only by SDS-PAGE, but SDK1 activity was thereby lost. The presence of an ϳ40-kDa band associated with high SDK1 activity and the absence of such band in fractions without SDK1 activity were noticed at the Mono Q chromatography step. Finally, the ϳ40-kDa band was identified as having Ͼ53% of the same amino acid sequence found in the kinase domain of PKC␦. The ϳ40-kDa band was presumed to be SDK1, and this was confirmed by several lines of study as described here. However, there is no indication that the ϳ40-kDa SDK1 protein occurs naturally, based on an NCBI Database search with the BLAST algorithm. Thus, it is unclear at this time whether the ϳ40-kDa SDK1 protein is present independently of PKC␦ or is a cleavage product of PKC␦. Studies by mRNA analysis with possible differential splicing are under way.
We have now identified SDK1 as being closely associated with PKC␦, although PKC␦ per se has no SDK1 activity. Many recent studies suggest that PKC␦ translocated to the mitochondrial membrane may be involved in the apoptotic process, but a definitive mechanism is still under debate (Ref. 30 and references therein). PKC␦ activity requires PS and diacylglycerol and is inhibited by Sph or DMS, in striking contrast to SDK1 activity, which requires Sph or DMS and is inhibited by PS ( Fig. 4 and Table IV). The C-terminal half kinase domain of PKC␦ is now identified as displaying all properties of SDK1, i.e. high dependence on Sph or DMS and no dependence on Cer or Sph-1-P. Sph at low concentration (5 M) showed a maximal effect when low levels of the kinase and 14-3-3 were used with ethanol as vehicle (16). The Sph or DMS concentration required for the maximal effect on SDK1 activity depended on the vehicle used (ethanol, Me 2 SO, or octyl glucose), quantity of the kinase and of the 14-3-3 substrate, and composition of the reagents in the reaction mixture. The effect of these variable factors on SDK1 activity was much reduced when a higher concentration of Sph or DMS (50 M) was used. DMS is a stronger base than Sph, and its enhancement of SDK1 activity was more consistent than that of Sph. Therefore, 50 M DMS was used in the assay system for technical convenience. The enhancing effect of Sph was highly stereospecific because Lthreo-Sph, Cer, and Sph-1-P had no effect on SDK1 activity. Likewise, L-threo-DMS, stearylamine, and hexadecyltrimethylammonium bromide have no effect on SDK1 activity despite their cationic aliphatic structures (17). These findings (high susceptibility to Sph and DMS under certain conditions and  strict structural requirements) indicate that the effect of Sph and DMS on SDK1 is physiological, in analogy to other specific lipid effects on signal transducer molecules.
The substrate specificity of SDK1 for 14-3-3 with a defined peptide sequence is indicated by the inhibition of SDK1 activity by the icosapeptide in the helix 3 domain. The kinase activities of both SDK1 and PKC␦ may share the same ATP-binding site and "Ser/Thr kinase signature" (Table II) because both kinase activities are inhibited by common PKC inhibitors.
Sph has been suggested to be a "second messenger" because its low level is greatly increased upon stimulation of cells, e.g. platelet-derived growth factor and insulin-like growth factor cause activation of ceramidase, which converts Cer to Sph and fatty acid (31,32). An increase in Sph may be due in part to synthesis from serine and palmitoyl-CoA, the classic synthetic system for Sph (33), although de novo synthesis in response to cell stimulation has not been clarified. Treatment of HL-60 cells with phorbol ester increases Sph levels by 3-fold, whereby the cells differentiate into macrophages. Treatment of the same cells with exogenous Sph causes apoptosis (34). DMS is considered to be derived from Sph by methylation through N-methyltransferase (35). The chemical level of cellular Cer is much higher than that of Sph, and the level of DMS is much lower than that of Sph. For example, the quantities of Cer, Sph, Sph-1-P, and DMS in 1 ϫ 10 6 HL-60 cells, determined by LC/MS analysis, were ϳ12.07 g, 0.61 ng, 0.048 ng, and 2.49 pg, respectively (36). The low levels of Sph and DMS and their increase in response to stimulation of cells are consistent with the hypothesis that they function as second messengers.
Sph converted from Cer or synthesized de novo may be shortlived because it is readily converted to Cer or Sph-1-P. A minor FIG. 5. SDK1 activities of the rat liver fraction and of caspase-3-treated PKC␦ and its inhibition by the PKC inhibitors rottlerin and GF109203X. A, the rat liver fraction from the Mono Q step (Fig. 1E) containing the ϳ40-kDa protein was subjected to SDK1 activity assay, and this activity was tested for inhibition by 150 M rottlerin (panel a) and 5 M GF109203X (GF; panel b). A 4-l sample was used for phosphorylation of 1 g of 14-3-3␤ in the absence (Ϫ) or presence (ϩ) of DMS. B, human recombinant PKC␦ and caspase-3-treated PKC␦ were subjected to SDK1 activity assay, and the activity was tested for dose-dependent inhibition by rottlerin (panel a) and GF109203X (panel b). Quantities of the inhibitor and the absence (Ϫ) or presence (ϩ) of DMS are indicated on the abscissa. Quantities of PKC␦ and caspase-3 were the same as described in the legends of Figs. 2B and 4. The bar graphs in A and B show the SDK1 activities in the presence (ϩ; black bars) and absence (Ϫ; white bars) of DMS. In A and B, data from one of two experiments with very similar results are shown.
FIG. 6. Inhibitory effect of the 14-3-3␤ icosapeptide on the SDK1 activity of caspase-3-treated PKC␦. The SDK1 activity of caspase-3-treated PKC␦ (quantities of PKC␦ and caspase-3 as described in the legends of Figs. 2B and 4) was determined in the presence of various concentrations (abscissa) of the 14-3-3␤ icosapeptide (YKNV-VGARRSSWRVISSIEQ) and the control peptide (17 amino acids) as described under "Inhibition of the SDK1 Activity of Caspase-3-treated PKC␦ by the 14-3-3 Peptide." SDK1 activity in the presence of the 14-3-3␤ icosapeptide (black bars) was strongly inhibited, whereas the control peptide (gray bars) had no effect. Values are the means of triplicate experiment; S.D. values are shown by the error bars. quantity of DMS, if synthesized, is inhibitory to Sph kinase (37) and is absolutely stable. Thus, Sph in combination with DMS may be an effective inducer of various kinases, including SDK1 as well as other SDKs (17) whose biochemical properties remain to be elucidated.
The substrates for SDK1 are 14-3-3 protein family members (16,29). How phosphorylation of 14-3-3 proteins by SDK1 contributes to the Sph/DMS-mediated signaling pathway remains to be elucidated. However, identification of SDK1 as the catalytic domain of PKC␦ raises several intriguing possibilities. It may help explain a role for Sph/SDK1 in promoting cell death signaling.
14-3-3 proteins normally form homo-or heterodimers through their N-terminal helical structure, which is required for many of their regulatory roles in cells. For example, an intact dimeric form of 14-3-3 is essential for functional association of 14-3-3 with Raf1 (38,39). Phosphorylation of 14-3-3 in the dimer interface would be expected to interfere with its dimer formation, thus interfering with 14-3-3-mediated functions. Accumulating evidence suggests that 14-3-3 can promote cell survival through its interaction with multiple pro-apoptotic proteins such as Bad and Bax (40,41). For example, proapoptotic Bad complexed with Bcl-2/Bcl-x neutralizes the antiapoptotic Bcl-2/Bcl-x effect, leading to cell death. This neutralization effect is inhibited by binding of 14-3-3 to Bad (40). Thus, 14-3-3 binding to Bad blocks apoptosis. It has been demonstrated that the catalytic domain of PKC␦ or the SDK1 protein is released from its full-length proenzyme under a variety of cell stress or death conditions by activated caspase-3 (42)(43)(44). It is conceivable that the released catalytic domain of PKC␦/ SDK1 is stimulated by Sph/DMS, leading to phosphorylation of 14-3-3 in the dimer interface. We propose that such phosphorylation may interfere with its binding to diverse ligands and neutralizes its role in inhibition of cell death. It is possible that phosphorylation of 14-3-3 by SDK1 may lead to dissociation of 14-3-3 to Bad or Bax, allowing Bad or Bax to induce mitochondrial dysfunction and cell death. It may not be a coincidence that cleaved PKC␦ has been shown to target mitochondria (45,46). Extensive further study is needed to confirm the possibility that inhibition of dimer formation by SDK1 blocks such a pro-survival function of 14-3-3.
It is also significant that we are able to suggest a mechanism that may control the dimerization status of 14-3-3 proteins. The 14-3-3 family is composed of multifunctional regulatory proteins that can interact with Ͼ100 protein partners in cells. A general mode of 14-3-3 binding is controlled by phosphorylation of target proteins at a defined phosphoserine/phosphothreonine motif (18,47,48). However, it remains unclear how the function of 14-3-3 itself is regulated and what determines the target specificity. Reversible regulation of 14-3-3 dimerization through phosphorylation in the dimer interface may provide a mechanism by which 14-3-3 activity is controlled. Identification of SDK1 as a homolog of PKC␦ defines a signaling pathway that may directly control the ability of 14-3-3 to bind and to affect the function of diverse cellular proteins.