Properties, Regulation, and in Vivo Functions of a Novel Protein Kinase D

Protein kinase D (PKD) isoforms are protein kinase C effectors in signaling cascades controlled by diacylglycerol (DAG). All PKDs are regulated by DAG/phorbol 12-myristate 13-acetate-binding C1 domains and an activation loop (A-loop). To understand how PKD isoforms diversify DAG signaling networks, it is essential to determine redundant and novel properties of their regulatory domains, characterize factors controlling PKD gene expression, and discover their in vivo physiological roles. Studies on a novel PKD, Caenorhabditis elegans DKF-2 (D kinase family-2), addressed these topics. The C1b domain mediates phorbol 12-myristate 13-acetate-induced translocation and activation of DKF-2. However, when DAG is elevated, C1a and C1b contribute equally to targeting/activation of DKF-2. DKF-2 C1 domains do not inhibit catalytic activity; they mediate delivery of DKF-2 to a membrane where protein kinase C phosphorylates Ser925 and Ser929 in the A-loop. This potently stimulates DKF-2 catalytic activity. Phosphorylation of Ser925 alone switches on 70% of maximal kinase activity. Persistent phosphorylation of Ser929 tags DKF-2 for proteasomal degradation; Ser(P)925 plays a minor role in DKF-2 degradation. GATA enhancer sequences govern DKF-2 expression in intestine in vivo. Adult life span increases 40% in animals lacking DKF-2. In thermally stressed wild type animals, the DAF-16 transcription factor is segregated from the nuclei of adult intestinal cells. In contrast, DAF-16 enters adult intestinal nuclei of DKF-2-deficient, thermally stressed animals, where it can trigger gene transcription that protects against various insults. The results suggest a mechanism for increased longevity and show that a PKD links DAG signals to regulation of stress responses and life span.

Current knowledge of PKD regulation and functions is incomplete. One puzzle is that homologous regulatory domains control catalytic activity via dissimilar mechanisms in different PKDs (5,17). For example, the PKD1 PH domain inhibits catalytic activity (18), whereas PH domain mutations in a Caenorhabditis elegans PKD (named DKF-1, for D kinase family 1) extinguish kinase activity (17). Physiological roles of PKDs have been inferred from studies on transfected cells, but it is not known if roles assigned to PKDs are major or minor functions of D kinases in normal differentiated cells. In vivo functions of PKDs have not been analyzed by gene disruption or "knock-in" approaches, and identities of most PKD effectors are unknown. Consequently, we lack understanding of PKD functions in normal, organized tissues and overall animal physiology.
Activated PKDs translocate from plasma membrane to Golgi and mitochondrial membranes and nuclei, where elevated kinase activity persists for minutes to hours. Routing of PKDs to multiple intracellular destinations enables dissemination of DAG signals to an expanded range of substrate-effector proteins. Thus, properties of PKDs apparently enable transiently produced DAG to exert both short and long term temporal and spatial control over cell functions. To fully comprehend the physiological roles of PKDs, it is essential to elucidate mechanisms underlying the generation and maintenance of activated PKDs. Current knowledge of three contributing factors, the regulation of the recruitment and activation of PKDs at the cell surface, control of PKD stability, and determinants of PKD gene expression, is limited.
Central problems in PKD regulation and functions can be addressed in C. elegans. This model organism employs signaling molecules, mechanisms, and pathways that are conserved from nematodes to man (19 -22). Moreover, classical studies on intrinsic biochemical and regulatory properties of C. elegans PKDs can be coupled to incisive in vivo analysis. For instance, techniques for facile gene disruption and targeted cell-specific mRNA/protein expression (23)(24)(25)(26), in concert with molecular and classical genetics, provide powerful tools for illuminating regulation and physiological functions of PKDs in intact animals. Insights acquired from studies on C. elegans PKDs can guide complementary investigations in mammalian systems.
Studies on DKF-1, a C. elegans PKD, confirmed the value of the model system (17,27). DKF-1 has the substrate specificity and structural hallmarks of a typical PKD. However, DKF-1 and PKDs differ as follows: (a) in regulatory properties and functions of their C1 and PH domains and (b) in mechanisms that control A-loop phosphorylation. In addition, DKF-1 is activated by DAG but not by PKCs. Gene disruption revealed that DKF-1 is essential for neuromuscular control of lower body movement (27). These investigations provided novel insights into the activation mechanism, regulatory capabilities, and physiological function of a member of the PKD family.
We now report the characterization of a novel C. elegans PKD, named DKF-2. DKF-2 and mammalian PKD1 have similar structures and substrate specificities. However, regulatory properties and functions of C1 and PH domains are strikingly different in the two kinases. Two phosphoserines in the A-loop differentially control catalytic activity and stability of DKF-2. GATA enhancer sequences regulate selective expression of DKF-2 in intestinal cells in vivo. In adult intestine, DKF-2 negatively regulates a transcription factor that activates stress response genes. This suggests an explanation for the discovery that DKF-2 deficiency significantly extends life span.
DNA and Protein Data Analysis-Analysis of sequence data, sequence comparisons, and data base searches were performed using Blast programs (NCBI Server, National Institutes of Health, Bethesda), the SMART website (EMBL, Heidelberg, Germany), ClustalW, and pattern/motif search programs (European Bioinformatics Institute, Hinxton, UK) and programs supplied by the Swiss Institute of Bioinformatics (Geneva, Switzerland). Information about C. elegans genomic DNA, cDNAs, ESTs and proteins was obtained from Worm-Base (Cold Spring Harbor Laboratory, Cold Spring Harbor, NY).
Preparation of Antibodies Directed against DKF-2-A fragment of DKF-2 cDNA encoding amino acids 1-203 was synthesized via PCR, using primers with NdeI (5Ј) and BamHI (3Ј) restriction sites. Amplified DNA was digested with NdeI and BamHI and cloned into the isopropyl 1-thio-␤-D-galactopyranoside-inducible bacterial expression plasmid pET-14b (Novagen). Vector DNA encodes a His 6 tag peptide that is appended to the N terminus of partial DKF-2 protein. A 1-liter culture of Escherichia coli BL21 was transformed with recombinant plasmid, induced with isopropyl 1-thio-␤-D-galactopyranoside, and disrupted as described previously (28). Approximately 2 mg of DKF-2 fusion protein was purified to near homogeneity by affinity chromatography on nickel-nitrilotriacetic acid-agarose resin (Qiagen). DKF-2 fusion protein was injected into rabbits (0.4-mg initial injection; 0.2 mg for each of five booster injections) at 3-week intervals (Proteintech, Chicago). Antiserum was collected at 3-week intervals. IgGs directed against DKF-2 were affinity-purified by chromatography on Sepharose 4B beads derivatized with DKF-2 fusion protein (1.5 mg of protein/ml beads) and stored as reported previously (29).
Antibodies Directed against PKD1 and Phosphorylated A-loops of PKD1 and DKF-2-Rabbit IgGs directed against Ser(P) 744 in the PKD1 A-loop were purchased from Cell Signaling (Danvers, MA). Di-phosphorylated A-loop peptide (Ser(P) 744 and Ser(P) 748 ) was used as antigen, but the antibodies selectively bind Ser(P) 744 (Ser(P) 925 in DKF-2) and do not ligate Ser(P) 748 (Ser(P) 929 in DKF-2). For details, see Waldron et al. (10) and Fig. 7C under "Results." Rabbit anti-PKD1 IgGs were obtained from Santa Cruz Biotechnology (Santa Cruz, CA). The antibodies were generated by injection of nonphosphorylated C-terminal peptide antigen and are used to detect total endogenous PKD1. These IgGs bind nonphosphorylated PKD1 with high affinity but have reduced affinity when PKD1 is phosphorylated, as described by Rennecke et al. (30).
Expression of DKF-2 in Mammalian Cells-Full-length DKF-2 cDNA was excised from recombinant pBluescript SKII by digestion with XbaI and Asp 718 and cloned into corresponding sites in pcDNA3.1ϩ (Invitrogen). This mammalian expres-sion vector contains a constitutively active cytomegalovirus promoter. Hamster AV-12 or human HEK293 cells were grown and transfected with recombinant plasmid DNAs using Lipofectamine TM (Invitrogen) as described previously (27). Cloned cell lines stably expressing modest amounts of WT and mutant DKF-2 proteins were isolated by selection with 1 mg/ml G418 for 14 days.
Immunoprecipitation and in Vitro Kinase Assays-Cells were lysed in buffer containing 1% Triton X-100 as reported (31). Operations were performed at 4°C. DKF-2, DKF-1, or endogenous PKD1 was isolated by immunoprecipitation with monospecific, affinity-purified antibodies. (Tests for antibody cross reactivity were negative.) After centrifugation (40,000 ϫ g, 30 min) the supernatant solution (0.3 mg of protein) received 1 g of anti-DKF-2 IgG (or anti-DKF-1 IgG or anti-PKD IgG (Santa Cruz Biotechnology)). Samples were incubated for 3 h. Next, 25 l of protein G-Sepharose 4B beads was added, and incubations were extended 1 h. Subsequently, bead-bound immune complexes were washed five times as reported previously (27). Amounts of immunoprecipitated WT and mutant DKF-2 proteins are assessed by Western immunoblotting (e.g. see Fig. 7B, below). All kinase assays for a given set of experimental conditions are then performed with equal amounts of WT and/or mutant DKF-2 proteins. Catalytic activity of DKF-2 was quantified by measuring incorporation of 32 P radioactivity from [␥-32 P]ATP into peptide substrate. Reaction buffer (30 l), which contains 25 mM Tris-HCl, pH 7.4, 30 M Syntide-2 peptide substrate (Calbiochem), 100 M [␥-32 P]ATP (ϳ150 cpm/pmol), 5 mM MgCl 2 , 0.5 mM EGTA, and 1 mM dithiothreitol, was added to immune complexes, and samples were incubated at 30°C for 10 min. Reactions were terminated by adding 5 l of 0.2 M EDTA, pH 8.0. Reaction mixtures were applied to P81 filter papers (Whatman). Filters were washed and air-dried as previously indicated (27). 32 P radioactivity incorporated into Syntide-2 was measured in a scintillation counter.
To test effects of PKC inhibition on DKF-2 catalytic activity, cells were incubated with bisindolylmaleimide I (GF109203X) (Axxora) for 1 h. Cells were then treated with specified concentrations of PMA for 10 min. In vitro kinase assays were performed as described above. Possible direct effects of GF109203X on DKF-2 activity were tested in vitro by adding the inhibitor to the kinase reaction mixture.
Preparation of Cytosol and Total Membrane Proteins in Translocation Assays-Cells were disrupted in hypotonic detergent-free buffer, as stated previously (17,31). Cytosol (supernatant) and total membranes (pellet) were isolated by centrifugation at 120,000 ϫ g. Volumes of cytosol and pellet fractions were adjusted so that equal aliquots represent equal numbers of starting cells (normalization to protein could also be used; cytosol and pellet each contained 50% of total protein). Samples containing 30 g of cytosolic or membrane proteins were applied to each lane in denaturing gels. For Western blot analysis, 14-3-3 (invariant in cytosol) and IGF-1 receptor (invariant in membranes) were monitored as loading controls. Antibodies directed against 14-3-3 and IGF-1 receptor were obtained from Santa Cruz Biotechnology. For in vitro kinase assays, membrane-bound DKF-2 was solubilized with 1% Triton X-100 and immunoprecipitated with affinity-purified IgGs (see above). All DKF-2 protein was solubilized from total membranes (or intact cells) by buffer containing Triton X-100. No DKF-2 was evident in the final detergent-resistant pellet.
Electrophoresis of Proteins and Western Immunoblot Assays-Proteins were size-fractionated by electrophoresis in a denaturing polyacrylamide (8%) gel as reported previously (32). Western blots of size-fractionated proteins were prepared, blocked, incubated with anti-DKF-2 (or other) IgGs (1:1000) and washed, as indicated in previous studies (31,32). Unless noted otherwise, each lane in Western blots received 30 g of protein.
Antigen-antibody complexes were visualized and quantified by using peroxidase-coupled secondary antibodies, an enhanced chemiluminescence procedure, and image analysis software (ImageQuant, GE Healthcare, and ImageJ) as described previously (32) Signals were recorded on x-ray film.
Determination of C. elegans Life Span-C. elegans was grown on agar plates seeded with E. coli OP50 at 20°C. Animal development was synchronized by standard procedures. When the nematodes reached adulthood, 80 -100 animals were placed on fresh plates and were assayed for viability at various intervals. Animals were considered dead when they ceased to move, stopped feeding, and failed to respond to touch with a wire. During the assay nematodes were separated from progeny and transferred to a fresh plate every 24 h. Data were plotted and analyzed with PRISM software (Graphpad).
DAF-16-GFP Translocation Assay-A stably integrated transgene, in which DAF-16 promoter/enhancer DNA drives expression of a DAF-16-GFP fusion protein (33), was crossed into WT and dkf-2(pr3) null genetic backgrounds. Animals were incubated at 33°C for 2 h and then fixed with 4% paraformaldehyde for 15 min. Duplicate batches of WT and mutant animals were maintained at 20°C and assayed in parallel. Animals were mounted, and subcellular location of DAF-16-GFP was determined by fluorescence microscopy as described previously (17,27).
Other Experimental Procedures-Detailed descriptions of PCR amplification and analysis of a partially deleted gene, generation and analysis of transgenic animals, confocal immunofluorescence microscopy, protein determination, and deletion and site-directed mutagenesis are provided in recent papers (17,27).

RESULTS
Identification and Characterization of cDNA Encoding DKF-2-A search of the C. elegans genome (in WormBase) identified two genes encoding predicted PKDs. One gene included exons for DKF-1, a previously characterized PKD (27). A second gene, designated T25E12.4a, was distributed over four cosmids (GenBank TM accession numbers Z82052, Z92967, AL021507, and AL021572, respectively) generated by the C. elegans Genome Project. Conceptual transcription, splicing, and translation (Genefinder software) yielded a predicted, full-length (3.7 kbp) cDNA encoding a novel PKD isoform. cDNAs (yk335b9 and yk417d2) potentially encoding PKD were obtained from the C. elegans EST project (Dr. Y. Kohara, National Institute of Genetics, Mishima, Japan). Only minor portions of DNA inserts were characterized previously. Thus, the recombinant DNAs were fully sequenced. Both clones had a 5Ј-untranslated region (60 bp), an initiator ATG sequence coupled to a long open reading frame (3210 bp), and a 462-bp 3Ј-untranslated region. The open reading frame encodes a protein composed of 1070 amino acids (Fig. 1A), which was named DKF-2. According to C. elegans nomenclature conventions, mRNA, cDNA, and protein are identified with uppercase lettering (DKF-2); the corresponding gene is named with lowercase italics (dkf-2).
Organization of the dkf-2 Gene-Alignment of DKF-2 cDNA sequence with corresponding C. elegans genomic DNA (Wormbase) revealed exon and intron boundaries in the dkf-2 gene. The gene contains 18 exons and spans ϳ40-kb of DNA (supplemental Table SI).
Structure/Function Relationships in DKF-2-Functions for conserved amino acids in Ser/Thr protein kinases have been established by biochemical analysis, determinations of threedimensional structures, and mutagenesis (34 -37). Thus, motifs within the DKF-2 amino acid sequence (M r ϭ 121,000) are associated with functions by analogy. A C-terminal region (residues 768 -1026, Fig. 1A) constitutes the kinase domain. A GXGXXGX 16 K motif (residues 777-799, Fig. 1A) typically mediates binding of ␤and ␥-phosphates of ATP. Lys 799 is essential for catalysis and ATP binding. A DFG motif (residues 914 -916) marks the N-terminal boundary of the A-loop. The Asp 914 carboxyl group binds ␥-phosphate and metal ion in MgATP and stabilizes pentavalent phosphorus in the transition state. An APE tripeptide (residues 938 -940) terminates the A-loop. Glu 940 , in concert with Asp 952 and Arg 1014 , stabilizes the core catalytic domain. HCDLKPEN (residues 891-898) is identified as the DKF-2 "catalytic loop" motif, which positions substrate proteins in an optimal orientation for catalysis (34 -37).
Sequence Conservation and Divergence among DKF-2, DKF-1, and PKDs-Among protein kinases in standard data bases, DKF-2 shared maximal overall sequence identity (45-49%) with mammalian PKDs 1-3 (supplemental Table SII). C. elegans DKF-2 and DKF-1 were more divergent (36% identity), and all other protein kinases were Ͻ30% identical with DKF-2. Sequence conservation was much greater within critical functional modules. For example, kinase domains of DKF-2 and PKDs are ϳ75% identical despite the evolutionary distance between nematodes and mammals. Catalytic and A-loops, which mediate the phosphotransferase reaction, are 100% identical in DKF-2 and human PKDs (supplemental Fig. S1, A and B). PKDs are activated when A-loop serines (Ser 744 and Ser 748 in PKD1, supplemental Fig. S1B) are phosphorylated by PKC (9,10). Thus, Ser 925 and Ser 929 , which occupy corresponding positions in the DKF-2 A-loop, are candidate phospho-acceptors ( Fig. 1A and supplemental Fig. S1B). Conservation of all 27 residues in DKF-2 and PKD A-loops (supplemental Fig. S1B) suggests that this structural module governs catalysis by a mechanism optimized by strong selection pressure. A-loop sequences of DKF-2 and DKF-1 exhibit substantial dissimilarity (supplemental Fig. S1B and Table SII). In DKF-1, QFRKT replaces the SFRRS di-phosphorylation motif in the A-loop. This change is consistent with the discovery that DKF-1 is activated by PKC-independent phosphorylation of a single A-loop Thr (27).
C1 domains of DKF-2 and PKDs share 70 -80% amino acid sequence identity (supplemental Table SII). Moreover, 6 Cys and 2 His, which control C1 folding and DAG/phorbol ester binding activity, are conserved in perfect register (supplemental Fig. S1, C and D). A critical Pro that governs DAG binding affinity in PKCs invariably occupies position 11 in DKF-2 and PKD C1 domains (supplemental Fig. S1, C and D). DKF-2 and PKD PH domains share ϳ45% identity (and 15% conserved amino acid substitutions) (supplemental Table SII). Amino acids essential for PH domain folding and functions (K 643 and W 717 , Fig. 1A) are conserved in DKF-2. Overall, the highly homologous regulatory and catalytic domains of the nematode and mammalian kinases indicate that DKF-2 is the optimal PKD prototype in C. elegans.
Linker regions and N-and C-terminal extensions of DKF-2 and PKDs are not homologous (supplemental Table SII). These segments of the polypeptides may mediate isoform-specific interactions with regulatory or scaffold proteins. The unique N-terminal region of DKF-2 (Table SII) indicates that DKF-2 is a distinct member of the PKD family.
PMA Induces Translocation, Phosphorylation, and Activation of DKF-2-AV-12 cells stably expressing full-length DKF-2 were disrupted in detergent-free buffer. After centrifugation, ϳ85% of DKF-2 protein was recovered in cytosol ( Fig. 2A, upper panel, lanes 1 and 2). Immunofluorescence microscopy confirmed DKF-2 was dispersed in cytoplasm and excluded from nuclei ( Fig. 2B, image 1). The small proportion of DKF-2 associated with the membrane pellet ( Fig. 2A, upper panel, lane 2) was neither concentrated at the periphery nor sufficiently enriched at an organelle to yield a differential fluorescence signal (Fig. 2B, image 1). Thus, the particulate kinase may be distributed among several intracellular locations. Incubation of cells with 300 nM phorbol 12-myristate 13-acetate (PMA) for 10 min induced efficient transfer of DKF-2 from cytoplasm to membranes ( Fig. 2A, upper panel, lanes 3 and 4). Fluorescence microscopy revealed that DKF-2 was selectively enriched at the cell periphery and depleted from the cytoplasm (Fig. 2B, image 2). All DKF-2 pro-  (14-3-3). Loading controls for DKF-1 were similar (not shown). Chemiluminescence signals were recorded on x-ray film. The asterisk indicates that DKF-2 protein increases in apparent molecular weight and becomes heterogeneous (band broadening) when cells are treated with PMA. B, PMA promotes accumulation of DKF-2 at the cell periphery. Cells transfected with a DKF-2 transgene were incubated in the presence or absence of 1 M PMA, as indicated. After 10 min, cells were fixed, permeabilized, and incubated with anti-DKF-2 IgGs and fluorescein isothiocyanate-tagged secondary antibodies as described (17,27). Signals were recorded via confocal immunofluorescence microscopy. TPA, 12-Otetradecanoylphorbol-13-acetate. C presents Western immunoblots containing cytosolic (S) and membrane (P) proteins (30 g/lane) isolated from nontransfected cells (lanes 1-4) and cells stably expressing DKF-2 (lanes 5-8). Cells were treated with 1 M PMA for 10 min as indicated. In the upper panel, lanes 1-4 were incubated with anti-PKD1 IgGs. Note that anti-PKD1 IgGs avidly bind nonphosphorylated PKD1 (upper panel, lanes 1 and 2) but have modest affinity for phospho-PKD1 (upper panel, lanes 3 and 4) (see "Experimental Procedures" and Ref. 30). Lanes 5-8 were probed with anti-DKF-2 IgGs. All lanes in the 2nd panel were incubated with IgGs that bind Ser(P) 744 in the PKD1 A-loop and Ser(P) 925 in the DKF-2 A-loop ("Experimental Procedures"). Loading controls are in the 3rd and bottom panels. D, cells expressing DKF-2 or DKF-1 were treated with 300 nM PMA or vehicle (basal) for 10 min prior to extraction with buffer containing 1% Triton X-100. Duplicate samples of cells were incubated with GF109203X for 1 h prior to addition of PMA. DKF-2, DKF-1, and endogenous PKD were immunoprecipitated, and in vitro kinase assays were performed as described under "Experimental Procedures." E, cells described in D were treated with 300 nM PMA for 10 min before lysis; DKF-2, DKF-1, and endogenous PKD were immunoprecipitated and assayed as described in D. GF109203X (3.5 M) was added to in vitro phosphorylation assays as indicated. F, Western immunoblot is shown. Each lane received 30 g of protein from lysates of cells expressing endogenous PKD1 or both PKD1 and recombinant DKF-2. Cells were treated with PMA (10 min before lysis) and GF109203X (1 h before PMA treatment) as indicated. Activated (phosphorylated) PKD1 and DKF-2 were detected as described in C above. Experiments were repeated three times, and similar results were obtained. Typical results are shown. OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 tein was solubilized from intact cells or membranes by extraction with buffer containing 1% Triton X-100. This suggests DKF-2 is concentrated at the cell surface by binding with lipids and/or intrinsic membrane proteins.

Regulation and Functions of DKF-2, a Novel PKD
Incubation of transfected cells with PMA induced large, stable increases (typically 5-14-fold) in DKF-2 catalytic activity ( Fig. 2D). Endogenous PKD1 was activated in parallel with DKF-2 in their shared intracellular environment ( Fig. 2, C-F). Another C. elegans PKD (DKF-1) was activated by PMA in a separate line of stably transfected cells (Fig. 2, A, bottom panel, and D). GF109203X, a potent inhibitor of DAG-activated PKCs, suppressed PMA-induced activation of both DKF-2 and PKD1, an established PKC effector (Fig. 2D). Similar results were obtained with a second PKC inhibitor, RO-318220 (data not shown). Results that verify the efficacy and selectivity of GF109203X are presented in supplemental Fig. S3 and associated text below. GF109203X had only a minor effect on in vitro catalytic activities of immunopurified DKF-2 and PKD1 (Fig.  2E). Thus, GF109203X does not directly inhibit the kinases. GF109203X did not alter phorbol ester stimulation of DKF-1 activity in intact cells or in vitro (Fig. 2, D and E). This selectivity control is consistent with the characterization of DKF-1 as a PMA/DAG-activated, PKC-independent PKD isoform (27). In contrast, DKF-2 activation is controlled by an upstream PKC.
DKF-2 proteins isolated from untreated cells exhibited an expected M r of 120,000 (Fig. 2, A, upper panel, lanes 1 and 2,  and C, upper panel, lanes 5 and 6). After exposure of cells to PMA, the band of DKF-2 polypeptides broadened, and the apparent molecular weight values increased (123,000 -130,000) in denaturing acrylamide gels (Fig. 2, A, S1B). A-loops were phosphorylated when DKF-2 and PKD1 proteins were recruited to membranes in PMA-treated cells (Fig. 2, C, 2nd panel, lanes 4 and 8, and F, lanes 2 and 5). Inhibition of PKCs with GF109203X blocked PMA-stimulated phosphorylation of DKF-2 and PKD1 (Fig. 2F, lanes 3 and 6). Evidently, accumulation of PMA in membranes enables C1 domain-dependent recruitment of both DKF-2 (see below) and PKC to the cell periphery. PMA, in concert with phosphatidylserine, activates PKC by expelling the pseudosubstrate peptide from the catalytic site (2). PKC can then phosphorylate the DKF-2 A-loop, triggering a conformational change that markedly increases catalytic activity of the D kinase.
DKF-2, PKDs, and DKF-1 Have Similar Substrate Specificities-Activated, detergent-solubilized DKF-2 was immunoprecipitated and assayed for kinase activity, using several possible peptide substrates. Syntide-2 (PLARTLSVAGLPGKK) was a good substrate for DKF-2 (Fig. 3A). In contrast, modified PKC⑀ and PKC pseudosubstrate peptides (in which Ser replaced Ala) were poorly phosphorylated. Myelin basic protein, a small protein used as a surrogate substrate for many Ser/Thr protein kinases, was minimally phosphorylated by DKF-2 (data not shown). Steady state kinetic analysis revealed that DKF-2 has an apparent K m of 23 M for Syntide-2 and catalyzes phosphorylation of the peptide with a k cat of 62 min Ϫ1 (Fig. 3B). DKF-2 exhibits a similar K m for ⑀-peptide (29 M), but k cat was markedly diminished to 6.8 min Ϫ1 . The k cat /K m ratio for DKF-2-mediated Syntide-2 phosphorylation was (a) 12-fold higher than the value obtained for ⑀-peptide phosphorylation (Fig. 3B) and (b) exceeded k cat /K m ratios obtained for -peptide and myelin basic protein by Ͼ30-fold (data not shown). Thus, Syntide-2 is a preferred DKF-2 substrate. Comparison of results presented in Fig. 3, A and B, with data in the literature (13,43,44) indicates that DKF-2, PKDs, and DKF-1 share similar substrate specificities and exhibit optimal activities with Syntide-2. . DKF-2 substrate specificity. A-C, PMA-activated DKF-2 was immunoprecipitated from cell extracts and assayed for kinase activity as described under "Experimental Procedures." A, assays were performed with the indicated substrates. Amino acid sequences for Syntide-2 and D-tide are given under "Results"; sequences of and ⑀ peptides are YRRGSRRWKKIY and RFARKGSLRQKNV, respectively. B, velocities of phosphorylation were measured as a function of Syntide-2 or ⑀-peptide concentration (using saturating ATP). Data were fitted by nonlinear regression and analyzed using PRISM software. C, cells were co-transfected with DKF-2 and either constitutively activated C. elegans PKC-1 or PKC-2 transgenes as described under "Results." The gradients (triangles) indicate results obtained using 0.2, 0.4, and 0.6 g of recombinant PKC expression plasmids. Expression of PKC-1 and PKC-2 was monitored by Western immunoblot analysis (bottom panel) as described previously (45,46). All assays performed for Fig. 3 were repeated three times and yielded similar results.
Syntide-2 contains the motif LXRXXS, which corresponds to a consensus phosphorylation site embedded in authentic PKD substrates (11,(13)(14)(15)(16). This motif was incorporated into a peptide named D-tide (AALVRQMSVAFFF) that was synthesized according to guidelines established by Cantley and co-workers (13). D-tide was robustly phosphorylated by DKF-2 (Fig. 3A). A scrambled peptide with the same amino acid composition was not a substrate. Thus, the LXRXXS motif is a target site for DKF-2.
We also addressed the question of whether DKF-2 itself is a preferred substrate (in situ) for a specific class of upstream PKCs. Because transfected cells contain multiple endogenous PKCs, a strategy was developed to bypass incubations with PMA or hormone. C. elegans expresses three DAGactivated PKCs named PKC-1, PKC-2, and TPA-1. Ca 2ϩ , DAG-activated PKC-2 (␣/␤-like), and PKC-1 (⑀/-like), which is activated by DAG alone, were studied (45,46). (For unknown reasons DAG-stimulated TPA-1 was not expressed in our experiments.) Constitutively activated kinases were generated by deleting pseudo-substrate sites as described by Parker and co-workers (47). Mutated cDNAs were transferred to the pcDNA3.1ϩ vector and co-transfected into cells expressing wild type DKF-2. PKC phosphotransferase activities were verified by in vitro kinase assays (data not shown). Transfection with increasing amounts of recombinant, activated PKC-1 transgene elicited 2-4-fold increases in DKF-2 catalytic activity in either AV-12 cells (Fig. 3C) or HEK293 cells (not shown). In contrast, activated PKC-2 had no effect on DKF-2 activity. Thus DKF-2 is differentially activated in intact cells by a member of the nPKC class of PKC isoforms.
The C1b Domain Mediates PMA-induced Membrane Targeting and Activation of DKF-2-Conserved C1 modules mediate DAG binding in ϳ100 proteins (48,49). However, C1 function must be tested experimentally because homology-based predictions can be misleading (e.g. C1 modules in c-Raf and PKC do not bind PMA or DAG). Mutation of a conserved Pro at position 11 in the C1 module provides an incisive test (41,50). Pro 11 confers high affinity PMA/DAG binding activity on authentic PLC effectors. Mutation of Pro 11 to Gly does not cause misfolding but decreases PMA binding affinity of C1 domains by 30 -300-fold (41,50). Corresponding Pro residues in DKF-2 (Pro 332 , Pro 484 ) were replaced with Gly to determine contributions of C1a and C1b to membrane recruitment and activation of DKF-2.
Cells expressing wild type (WT) or mutant DKF-2 proteins were treated with vehicle or PMA for 10 min and then fractionated into cytosol (supernatant) and total membranes (pellet) (see "Experimental Procedures"). All cytosol samples had low basal (background) kinase activity that was not affected by PMA. Measurements of membrane-associated, DKF-2 kinase activity ("Experimental Procedures") yielded the desired doseresponse curves for PMA-induced enzyme activation. Activity of WT DKF-2 increased ϳ3-fold when cells were treated with 30 nM PMA (Fig. 4A). PMA concentrations of 100 and 300 nM produced 7.5-and 10.3-fold increases, respectively, in kinase activity. A maximal, 14-fold enhancement of DKF-2 catalytic activity was observed at 3 M PMA. Higher concentrations of PMA did not yield further increments in kinase activity (data not shown). Half-maximal activation (K a ) of DKF-2 was achieved with 85 nM PMA.
Translocation of DKF-2 protein was independently assessed by comparing levels of DKF-2 protein in cytosol and membranes via Western immunoblot analysis (Fig. 4B). PMA was more potent in promoting DKF-2 translocation to membranes than in activating catalytic activity (Fig. 4B). A small increment in membrane-associated WT DKF-2 was evident at 10 nM PMA. Treatment with 30 nM PMA targeted ϳ65% of DKF-2 protein to membranes. Essentially all DKF-2 protein migrated to membranes when cells were exposed to 100 nM PMA. Complete DKF-2 translocation was maintained at higher PMA concentrations (0.3-3 M).
Replacement of Pro 332 with Gly (in C1a) generated a DKF-2 variant that was activated in parallel with WT enzyme at all PMA concentrations (Fig. 4A). Patterns of PMA-induced membrane translocation were similar for DKF-2 Gly 332 and WT kinase (Fig. 4B). In contrast, mutation of "Pro 11 " in C1b (DKF-2 Gly 484 ) severely attenuated kinase activation by PMA (Fig. 4A). Ratios of WT DKF-2 to DKF-2 Gly 484 catalytic activities were 9 and 4.5 at 100 and 300 nM PMA, respectively. Treatment of cells with 3 M PMA (a concentration well above the range used to induce physiological effects) produced a level of DKF-2 Gly 484 kinase activity that corresponds to 38% of the maximal value  OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 JOURNAL OF BIOLOGICAL CHEMISTRY 31279 measured for WT DKF-2 and DKF-2 Gly 332 . WT DKF2 and DKF-2 Gly 332 expressed this level of kinase activity when cells were incubated with only 60 nM PMA. Substitution of Gly for Pro 484 in the C1b domain potently inhibited translocation of DKF-2 to membranes in cells incubated with Յ100 nM PMA (Fig. 4B). In contrast, 100 nM PMA promoted complete transfer of WT DKF-2 and DKF-2 Gly 332 proteins to membranes. A small amount of DKF-2 Gly 484 migrated to membranes in response to 300 nM PMA. Increases in translocation of the C1bdefective kinase at 1 and 3 M PMA (Fig. 4B) correlated with elevations in kinase activity described above. At 1-3 M PMA, an equilibrium distribution is reached, in which similar amounts of DKF-2 Gly 484 are evident in cytosol and membranes. A DKF-2 variant (DKF-2 Gly 332 , Gly 484 ) that contains mutations in both C1 domains was neither translocated nor activated when cells were exposed to PMA (Fig. 4, A and B). Thus, C1b alone accounts for PMA-mediated targeting of DKF-2 to membranes at phorbol ester concentrations used to investigate DAG signaling. This ensures co-enrichment with PKC, thereby promoting efficient PKC-mediated activation of DKF-2. C1a exhibits weak affinity for PMA, which may make a minor contribution toward stabilizing association of DKF-2 with membranes.
The conclusion that the bombesin/DAG pathway activates DKF-2 via an upstream PKC is based on firmly established properties of GF109203X. GF109203X strongly inhibits all DAG-activated PKCs (4 novel PKCs and 4 cPKCs) but has no direct effect on DKF-2 (43,44,55). To strengthen our conclusion, we tested an additional PKC inhibitor and examined the possibility that GF109203X inadvertently inhibited other protein kinases. RO-318220, another compound that inhibits DAG-stimulated PKCs (56), suppressed bombesin-mediated DKF-2 activation to the same extent as GF109203X (supplemental Fig. S3). (RO-318220 was used throughout to verify results obtained with GF109203X (data not shown).) At GF109203X concentrations used in studies on DKF-2, the only other protein kinases that may be affected are p90 RSK (possible weak to moderate inhibition) and GSK-3 (possible weak inhibition) (57). A selective inhibitor of GSK-3 (SB 216763) had no effect on bombesin-mediated activation of DKF-2 (supplemental Fig. S3). Cells were transfected with a transgene encoding constitutively activated p90 RSK. Neither activation of endogenous WT kinase by hormones nor constitutively activated p90 RSK elicited an increase in DKF-2 activity (supplemental Fig. S3). Inhibition of ERK (a p90 RSK activator) via U0126 did not suppress bombesin-induced DKF-2 activation. Thus, neither p90 RSK, GSK-3, nor ERK is involved in a pathway that controls DKF-2 activity. The data strengthen conclu- sions and preclude concerns about "off target" effects of GF109203X on other protein kinases.
Because activated PLC generates phosphatidylinositol 3,4,5trisphosphate, Ca 2ϩ may play a role in DKF-2 activation. However, elevation of free Ca 2ϩ by incubation of cells with ionomycin failed to stimulate DKF-2 (supplemental Fig. S3). Conversely, chelation of intracellular Ca 2ϩ by 1,2-bis(2aminophenoxy)ethane-N,N,NЈ,NЈ-tetraacetic acid did not inhibit a bombesin-triggered increase in DKF-2 catalytic activity. Thus, results presented in Fig. 5, B and D, and supplemental Fig. S3 suggest that bombesin-stimulated activation of DKF-2 is predominantly controlled by three proximal upstream signaling molecules as follows: PLC, DAG, and PKC.
The PH Domain Is Not Essential for DKF-2 Translocation or Activation-PH domains target certain signaling proteins to membranes by binding phosphatidylinositol 3,4,5trisphosphate (or its metabolites) (39,42). DKF-2 contains a conserved PH module ( Fig. 1 and supplemental Table SII). Lys 643 corresponds to a conserved Arg or Lys that stabilizes binding of PH domains with phosphoinositides or proteins via electrostatic interactions. Substitution of Lys 643 with Ala or PH domain deletion caused 2-2.5-fold increases in basal DKF-2 catalytic activity (Fig. 6A). However, maximal activities of the DKF-2 mutants remained under PMA control. Incubation of transfected cells with PMA elicited 3-3.5fold increments in kinase activities of DKF-2 Ala 643 and DKF-2 ⌬PH (Fig. 6A). Moreover, WT and mutant kinases expressed similar amounts of catalytic activity after PMA stimulation. Neither mutation nor deletion of the PH domain altered PMAinduced targeting of DKF-2 proteins to membranes (Fig. 6B). Thus, the PH domain is not involved in PMA-induced translocation or activation of DKF-2.
Two A-loop Serines Differentially Regulate DKF-2 Activity-The predicted DKF-2 A-loop includes two serines that are candidate PKC substrates (Fig. 1A). Ser 925 and Ser 929 were replaced with Ala to investigate their roles in regulating DKF-2. Both mutants translocated to the periphery of PMA-treated cells (not shown). Effects of A-loop mutations were quantified by measuring phosphotransferase activities, using equal amounts of WT or mutant DKF-2 (Fig. 7, A and B). PMA treatment increased kinase activities of DKF-2 Ala 925 and DKF-2 Ala 929 ϳ4.4-fold (relative to basal values). Activation of DKF-2 Ala 929 and the WT kinase was accompanied by phosphorylation of Ser 925 (Fig. 7C and Fig. 2, C and F). Anti-Ser(P) 929 IgGs are not available. However, strong suppression of PMA-induced activation of DKF-2 Ala 925 by GF109203X (Fig. 7A) is consistent with PKC-mediated phosphorylation of Ser 929 . Replacement of Ser 925 or Ser 929 with Ala did not alter the sensitivity (fold activation) of DKF-2 to PMA or the mechanism of activation (translocation and PKC-mediated A-loop phosphorylation). Because equal amounts of WT and mutant DKF-2 proteins were assayed (Fig. 7B), the data presented in Fig. 7A correspond to specific activities. PMA-stimulated catalytic activity declined sharply in the DKF-2 Ala 925 mutant (73% decrease versus WT kinase), but only modestly (23%) in DKF-2 Ala 929 (Fig. 7A). Thus, Ser 925 is a dominant regulator of catalysis.
A-loop serines were also replaced with Asp. The negatively charged, carboxylate side chain of Asp mimics Ser(P). Basal (unstimulated) activities of DKF-2 Asp 925 and DKF-2 Asp 929 and WT kinase were similar (Fig. 7A). When cells were incubated with both PMA and GF109203X (to enable Asp kinase translocation in the absence of A-loop phosphorylation), DKF-2 Asp 925 and DKF-2 Asp 929 catalytic activities were ele- WT and mutant kinases were immunoprecipitated from cell extracts and assayed for phosphotransferase activity ("Experimental Procedures"). B, cytosol and membrane proteins were prepared (see "Experimental Procedures") from cells stably expressing WT or mutated DKF-2, after 10 min of incubation with PMA. Distribution of WT and mutant kinases between cytosol and membranes was determined by Western blot analysis using anti-DKF-2 IgGs as described under "Experimental Procedures." Note that the apparent molecular weight value for DKF-2 ⌬PH is 105,000. Experiments were repeated twice and similar results were obtained in each instance. Representative data are shown. OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 vated relative to WT DKF-2 activity (Fig. 7A). Unmasking of increased kinase activity by association of Asp 925 and Asp 929 mutants with membranes suggests optimal DKF-2 activation requires A-loop phosphorylation and additional PKC-independent post-translational modifications, protein-protein, or protein-lipid interactions.
PMA did not increase catalytic activity of the double mutant, DKF-2 Ala 925 ,Ala 929 (Fig. 7A). Thus, the inhibitory effect of the DKF-2 A-loop is apparently irreversible if phosphorylation of both regulatory serines is suppressed. Conversely, relaxed structural constraints in the A-loop of doubly mutated DKF-2 Asp 925 ,Asp 929 enable a PMA/PKC-independent, 4-fold increase in basal kinase activity (relative to WT DKF-2, Fig. 7A). DKF-2 Asp 925 ,Asp 929 activity increased further in cells treated with either PMA plus GF103209X (30% increase) or PMA alone (45% increase). This confirms that binding with membranes enhances DKF-2 activity by PKC-dependent and -independent processes.
routes DKF-2 for degradation. Phosphorylation of Ser 925 also promotes enzyme lability; substitution of Ala for Ser 925 yields a DKF-2 mutant that accumulates (ϳ2.5-fold relative to WT DKF-2) in untreated cells (Fig. 8B, lanes 1 and 3) and is refractory to PMA-mediated down-regulation (Fig. 8B, lanes 3, 4, 9, and 10). Replacement of Ser 925 or Ser 929 with Asp did not alter DKF-2 stability (Fig. 8B, lanes 5 and 6), indicating that Asp does not mimic Ser(P) in the degradation pathway. Typical ratios of relative amounts of DKF-2 Ala 929 , DKF-2 Ala 925 , and WT DKF-2, determined from multiple independent isolates of stable transfectants, are 6.3/2.8/1. Contributions of A-loop serines to DKF-2 instability are inverted relative to their roles in kinase activation. If activation is prolonged, Ser 929 will promote efficient targeting of a high proportion of DKF-2 to proteasomes; Ser 925 also routes DKF-2 for proteolytic destruction but with lesser efficiency.
Tandem, Cis-acting GATA Sequences Activate dkf-2 Gene Transcription in Intestinal Cells in Vivo-DKF-2 provides a prototype for investigating PKD gene expression and functions in vivo. First, cells engaged in dkf-2 gene transcription were identified. A 1.4-kb fragment of promoter/enhancer DNA, which flanks the 5Ј end of the dkf-2 gene, was inserted upstream from a green fluorescent protein (gfp) reporter gene in a C. elegans expression plasmid. An octapeptide nuclear localization signal was appended to the N terminus of GFP. C. elegans strains that transmit the gfp transgene in the germ line were generated, and promoter activity was monitored by fluores-cence microscopy. Accumulation of GFP in nuclei sharpens fluorescence signals and facilitates identification of individual cells in vivo.
Patterns of GFP expression were similar in multiple, independently created lines of transgenic animals. GFP accumulated in late embryos, all larval stages, and adult animals. dkf-2 promoter activity was robust but restricted to 18 cells comprising the intestine (Fig. 9, B-E) and two additional cells positioned near the posterior bulb of the pharynx (arrows, Fig. 9, B-G). Apparently, the remaining 98% of somatic cells of the nematode generate few or no DKF-2 transcripts. The same developmental and cell-specific expression patterns were observed in C. elegans that contained a transgene encoding GFP-tagged DKF-2 under control of dkf-2 promoter/enhancer DNA (not shown). These observations suggest the speculation that DKF-2 may place one or more gut functions under control of stimuli that switch on PLCs and PKCs.
Developmental and intestinespecific patterns of dkf-2 gene transcription parallel expression patterns described for ges-1, a C. elegans gene encoding a gut-specific esterase (58). A pair of "GATA" sites (TGATA(A/ G)), separated by 12 nucleotides, constitute a gut-specific enhancer that controls GES-1 mRNA and protein expression in intestine. An intestine-specific transcription factor ELT-2 binds TGATA(A/G) sequences, engages the RNA polymerase II complex, and activates transcription of ges-1 and many other genes that control specialized intestinal cell functions (59 -61). The dkf-2 promoter/enhancer contains a pair of conserved TGATAA sites that precede the initiator ATG codon by 231 and 201 bp (Fig. 9H). GATA sites are numbered in accord with the dkf-2 gene sequence in cosmid T25E12. Nucleotides 23505-23508 (ATAA) and 23532-23538 (CTGATAA) (Fig.  9H) were deleted from dkf-2 promoter/enhancer DNA by mutagenesis. Elimination of GATA sequences did not alter GFP expression in pharyngeal cells but extinguished dkf-2 promoter activity in intestinal cells (compare Fig. 9, F and G, with  B-E). Thus, dkf-2 gene expression is stringently governed by proximal GATA enhancer elements (and presumably ELT-2) in intestinal cells; these regulatory elements do not control DKF-2 expression in pharynx.

DKF-2 Deficiency Ameliorates Heat Stress and Extends Life
Span-Animals carrying a disrupted dkf-2 gene were obtained from the C. elegans Knock-out Consortium (Vancouver, British Columbia, Canada). Back-crossing into WT C. elegans eliminated extraneous mutations and yielded cloned lines of viable animals. Fragments of the mutated dkf-2 gene were amplified  OCTOBER 26, 2007 • VOLUME 282 • NUMBER 43 by PCR, using genomic DNA from individual animals as templates (27). Agarose gel electrophoresis documented loss of a 1.5-kb segment of dkf-2 DNA (Fig. 10A). Sequencing of the PCR product revealed that a 1546-bp deletion eliminated exons 8 -10, introns 8 and 9, and portions of introns 7 and 10 (see supplemental Table SI). Splicing of exon 7 to exon 11 in DKF-2 mRNA alters the reading frame and introduces a premature stop codon. The disrupted dkf-2 gene potentially encodes a 57-kDa protein composed of amino acids 1-398 from DKF-2 plus 102 artifactual amino acids (from an incorrect reading frame). If the mutant DKF-2 polypeptide were synthesized, it would lack C1b and the entire kinase domain. The 120-kDa DKF-2 polypeptide was readily detected when Western blots of WT C. elegans proteins were probed with anti-DKF-2 IgGs (Fig. 10B). Mutant animals lacked full-length DKF-2 (Fig. 10B) and the smaller, predicted chimera (not shown). Thus, the deletion mutant is homozygous null for DKF-2. The deletion allele of the dkf-2 gene is named dkf-2(pr3) in accord with C. elegans nomenclature guidelines.

Regulation and Functions of DKF-2, a Novel PKD
Because DKF-2 is expressed predominantly in gut, aspects of intestinal physiology were characterized in dkf-2(pr3) null animals. DKF-2-deficient nematodes consumed normal amounts of E. coli; rates of food ingestion and waste excretion were normal (data not shown). Synthesis and export of yolk proteins from intestine to oocytes was not impaired. Differentiation of intestinal cells and gut development were normal. Timing of the life cycle between embryonic and late larval (L4) stages of development was not affected. However, adult life span of DKF-2-depleted C. elegans increased substantially (Fig. 10C). For example, on day 12 only 55% of WT animals were viable, whereas 90% of dkf-2(pr3) null nematodes were alive. The 50% survival time (S 50 ) was 13 days for WT animals but increased to 17 days for DKF-2-depleted C. elegans (Fig. 10C). After correcting for the invariant length of embryogenesis and larval development (3 days), the data show DKF-2 depletion causes a 40% increase in adult longevity. The maximum adult life span increased from 16 to 19 days in animals lacking a functional dkf-2 gene.
The principal signaling pathway that regulates C. elegans aging is controlled by insulin/IGF-1-like peptides and a receptor tyrosine kinase named DAF-2 (62)(63)(64)(65). DAF-2 activation elicits AKT-1 mediated phosphorylation of DAF-16, a FOXO family transcription factor. Phosphorylated DAF-16 is retained in cytoplasm. Inactivation of the insulin-AKT-1 pathway promotes dephosphorylation and nuclear translocation of DAF- FIGURE 10. dkf-2 gene disruption increases life span and promotes nuclear accumulation of DAF-16, a stress-activated transcription factor. A, a segment of the dkf-2 gene was amplified from WT and mutant genomic DNA by PCR, using specific nested primers, as described previously (27). DNA products were sizefractionated in a 1% agarose gel and stained with ethidium bromide. Lane "WT" contains the expected 3.3-kb DNA; A 1.8-kb DNA fragment in a lane labeled "dkf-2(pr3) null" reveals a 1.5-kb deletion; lane M received DNA size markers. B, Samples of total protein (30 g) from WT and dkf-2(pr3) null animals were analyzed on a Western blot. The blot was probed with anti-DKF-2 IgGs and developed as described under "Experimental Procedures." C shows a survival curve for aging WT and DKF-2 depleted C. elegans. The assay was performed as described under "Experimental Procedures." D, a DAF-16-GFP transgene was introduced into WT and dkf-2(pr3) null animals. The location of DAF-16 was determined by immunofluorescence microscopy after L2 larvae 16. Nuclear DAF-16 activates transcription of genes encoding proteins that protect C. elegans against various stresses, thereby extending life span. Genetic crosses were used to construct C. elegans strains that express DAF-16-GFP (33) in intestinal (and other) cells in WT and dkf-2(pr3) null backgrounds. A prototypical stress, heat shock at 33°C, was applied for 2 h, and accumulation of DAF-16-GFP in cell nuclei was monitored by fluorescence microscopy. Nuclear GFP was not evident in unstressed WT or mutant embryos, larvae, or adult animals (data not shown). Heat stress induced accumulation of DAF-16 in many nuclei of both WT and DKF-2-deficient embryos and L1-L4 larvae (e.g. L1, L2 larvae in Fig. 10D, panels a and c).
In WT adults, heat stress does not promote translocation of DAF-16 into gut cell nuclei (Fig. 10D, panel b). In contrast, thermal stress caused robust accumulation of DAF-16-GFP in intestinal nuclei of adult, DKF-2-deficient animals (Fig. 10D,  panel d). Thus, heat-induced (protective) migration of DAF-16 to the nucleus is normally attenuated in the adult phase of the life cycle. DKF-2 is evidently involved in suppressing DAF-16 translocation/activation in stressed post-larval animals. Depletion of the gut kinase enabled entry of DAF-16 into intestinal cell nuclei and extension of adult life span. A link between DKF-2 and DAF-16 was confirmed and strengthened by measuring life spans of dkf-2(pr3);daf-16(mgDf50) double null mutants (Fig. 10E). Loss of DAF-16 function abrogates the effect of DKF-2 deficiency. Thus, DKF-2 is apparently an upstream, negative regulator of DAF-16 in intestine.

DISCUSSION
Characterization of DKF-2 yielded new insights into intrinsic biochemical properties of regulatory domains and mechanisms underlying PKD translocation, activation, and degradation. The DKF-2 C1b domain avidly binds PMA and targets the kinase to the cell periphery. C1b-dependent translocation enables phosphorylation and activation of DKF-2 by co-localized, PMA-stimulated PKC. The C1a module binds PMA weakly and is not essential for PMA-induced translocation to membranes. A critical regulatory role for C1a was discovered when DKF-2 activation was regulated by a classical intracellular signaling pathway, in which a bombesin-BB 2 R complex promotes synthesis of DAG. Both C1a and C1b bind a DAG analog (1-oleoyl-2-acetyl-sn-glycerol) with similar affinities. Mutation of a critical Pro residue in either C1 domain reduced bombesininduced DKF-2 activation by ϳ50%. A DKF-2 mutant defective in both C1 domains was not activated by saturating bombesin. Thus, C1a and C1b contribute equally and additively to translocation/activation of DKF-2 when PLC is activated in intact cells.
Mammalian PKDs also bind PMA via C1b (66). Binding between DAG and individual C1 modules in PKD isoforms has not been systematically studied. Investigations on C1-DAG interactions might reveal that different PKD isoforms are controlled by distinct mechanisms and/or suggest strategies for selective activation or inhibition of individual PKDs.
In mammalian PKD1, C1 domains inhibit catalytic activity by intramolecular occlusion of the active site; C1b-mediated association of PKD1 with membranes unmasks kinase activity by reversing pseudosubstrate-like inhibition (5, 53, 66, 67). PH domain disruption generates constitutively active, PMA/DAGindependent PKD variants (18,68). Consequently, C1 and PH domains are considered direct, negative regulators of PKD catalytic activity. The negative suppression concept and supporting experimental data indicate that inactivation of C1 or PH domains generates constitutively active PKD variants (5,18,67,68). However, inactivating mutations in DKF-2 C1 modules substantially reduced PMA, and DAG-stimulated catalytic activity and had no effect on basal kinase activity. Deletion or mutation of the PH domain did not alter PMA/DAG controlled translocation or activation of DKF-2. Thus, C1 and PH domains are not suppressors of DKF-2 activity. Compelling evidence (Figs. [2][3][4][5] indicates C1 domains positively regulate kinase activity by routing DKF-2 to a membrane that recruits activated PKCs. DKF-2 and PKD1-3 are rapidly and efficiently activated when co-localized PKC phosphorylates two conserved A-loop serines (5,9,10). Phosphorylation of either Ser 744 or Ser 748 promotes 50% maximal activation of PKD1. Results provided in Figs. 7 and 8 demonstrate that PKC-mediated phosphorylation of both A-loop serines is required for optimal DKF-2 catalytic activity. However, regulatory roles of Ser 925 and Ser 929 in the DKF-2 A-loop are not identical. Phosphorylation of Ser 925 by PKC switches on 70% of total DKF-2 catalytic activity; the remaining kinase activity is expressed when PKC phosphorylates Ser 929 . Thus, A-loop serines control DKF-2 activity in an unequal but additive fashion. Evidently, nonphosphorylated Ser 925 plays a central role in limiting access of substrates to the catalytic site.
A-loop serines also control enzyme stability. Results in Fig.  8A and supplemental Fig. S5 suggest prolonged phosphorylation of the regulatory serines targets DKF-2 for proteasome-mediated proteolytic degradation. This may be a homeostatic mechanism for circumventing toxicity caused by tonic activation of PLC/DAG-controlled effectors. In this instance, roles of Ser(P) 925 and Ser(P) 929 are inverted. Substitution of either Ser 925 or Ser 929 with Ala (a) resulted in substantially increased intracellular level of mutant kinase (relative to modestly expressed WT DKF-2) and (b) generation of large amounts of kinase resistant to PMA-induced degradation. Levels of DKF-2 Ala 929 were 2-2.5-fold higher than amounts of DKF-2 Ala 925 in unstimulated and PMAtreated cells. Thus, Ser(P) 929 evidently routes a high proportion (perhaps ϳ70%) of DKF-2 to proteasomes. Ser(P) 925 also targets DKF-2 for degradation but at a lower rate. It is not known if A-loop phosphoserines control levels of mammalian PKD isoforms.
Differential regulation of kinase activity and stability by Ser(P) 925 and Ser(P) 929 provides opportunities for controlling the amplitude and duration of DKF-2 activation by protein phosphatases. For instance, di-phosphorylated DKF-2 is expected to be optimally active for a short time. DKF-2 (Ser(P) 925 dephospho-Ser 929 ) will be highly active for a prolonged period, whereas DKF-2 (dephospho-Ser 925 Ser(P) 929 ) will briefly exhibit a modest level of activity. Development of antibodies that report levels of di-, mono-, and nonphosphorylated A-loop serines will be required to determine whether differential dephosphorylation modulates DKF-2 activity in situ.
A partial discrepancy between amounts of membrane-associated DKF-2 protein and lower than expected levels of kinase activity was observed in dose-response curves (see Fig. 4 and "Results"). This may reflect two discrete steps in DKF-2 activation. DKF-2 is efficiently routed to membranes because it binds PMA with high affinity (Fig. 4B). DKF-2 phosphorylation and activation depend on co-recruitment of PKC. Results in Fig. 4 can be explained if an upstream PKC binds membranes (PMA) with modest affinity. Then substantially higher levels of PMA might be required to fully activate a pool of DKF-2 that is preassembled at the cell periphery.
Two C. elegans genes, dkf-2 and dkf-1, encode PKDs. Many properties distinguish DKF-1 from the more closely related DKF-2 and PKDs (17,27). PMA-induced translocation of DKF-1 is mediated by C1a; PKC-independent phosphorylation of a single Thr (Thr 588 ) in the A-loop triggers expression of DKF-1 catalytic activity; and PMA/DAG-regulated kinase activity of DKF-1 is extinguished if the PH domain is mutated. DKF-1 is expressed in neurons and is involved in neuromuscular control of locomotion; DKF-2 accumulates predominantly in intestinal cells and modulates longevity by attenuating stress responses. Conserved regulatory domains of DKF-1, DKF-2, and PKDs can subserve divergent mechanisms and functions (17,27). Recent reports indicate properties of C1 domains may differ among PKD isoforms (68,69). In contrast to PKD1, C1 domains of PKD2 do not inhibit catalytic activity. PMA-induced translocation of PKD3 is mediated by C1a, whereas C1b modules selectively target PKD1 and -2 to membranes. Plasticity, induced by interactions between conserved regulatory domains and surrounding divergent segments of DKF/PKD polypeptides, may adapt individual PKDs for specific physiological roles. The demonstration that DKF-2 and DKF-1 link DAG signals to the regulation of markedly different physiological functions (in distinct differentiated cells) documents PKD isoform specialization in vivo.
GATA enhancer sequences control dkf-2 gene expression in intestine in vivo. ELT-2, a powerful, positive regulator of gutspecific gene expression, uniquely binds GATA DNA sequences in adult intestine (58,59,61). Our studies show that GATA enhancer sequences (and thus ELT-2) play a central role in linking a PLC/DAG-controlled signaling cascade to the regulation of stress responses and longevity.
Knowledge of transcriptional control of PKD genes is sparse. However, patterns of PKD mRNA expression were elucidated in embryonic murine tissues (70). Transcripts encoding individual PKD isoforms in lung and heart were differentially distributed in distinct cell types or tissue regions. Thus, association of GATA transcription factors 4 -6 (ELT-2 homologs) with heart and lung development (71,72) suggests the speculation that specialized functions of individual PKDs might be (partly) controlled transcriptionally by cell-specific GATA factors and GATA DNA sequences. The plausibility of extrapolating from C. elegans to mammals is supported by a report that shows the following: (a) ELT-2 mediates an anti-pathogen, innate immune response in C. elegans, and (b) GATA 6, an ELT-2 homolog, performs the same function (in response to the same pathogen) in human lung epithelial cells (73).
Studies on DKF-2-depleted animals linked a prototypic PKD to a specific in vivo physiological function. Disruption of the dkf-2 gene extends C. elegans adult life span by 40%. A potential mechanism for the increment in longevity was discovered. DKF-2 inhibits stress-induced translocation/activation of DAF-16, a FOXO transcription factor, in adult intestine. In thermally stressed, DKF-2-deficient adult animals DAF-16 rapidly entered gut cell nuclei. Nuclear DAF-16 stimulates expression of genes that protect C. elegans against oxidative and thermal stresses, misfolded proteins, toxins, etc. (62,64,65,74). Accumulation of DAF-16 in many nuclei elicits large, well documented increases in life span. However, intestine-specific nuclear translocation of DAF-16 also promotes increased longevity. Germ cell deficiency generates a signal (carried by a nuclear hormone receptor and regulatory protein KRI-1) that selectively recruits DAF-16 to gut nuclei in adults and life span increases 50% (75). Intestinal cells are prime targets for life span regulation because gut governs energy metabolism, innate immunity, stress responses, nutrient supply to oocytes and serves as adipose and endocrine tissue in C. elegans (62,65). Thus, it is likely that execution of a DAF-16regulated program of gene expression contributes to the prolonged life of dkf-2(pr3) null animals. In WT adults, DKF-2 apparently inhibits DAF-16 accumulation in nuclei. This favors survival of embryonic animals (developing in the parental uterus or externalized eggs) and larvae, which have full potential for reproduction when a transient thermal stress recedes. Vital metabolic resources are not expended on adults, which may have little or no remaining reproductive capacity. Selective protection of younger animals reduces competition for food because unprotected adults will decline more rapidly in a post-stress period.
Mammalian PKDs are activated and promote survival (via NFB activation and Hsp27 phosphorylation) when cells are exposed to oxidative or thermal stresses (11,76). These positive modes of PKD-mediated regulation should not be regarded as conflicting with observations on negative regulation of adult longevity by DKF-2. DKF-2 exerts regulatory control within complex in vivo signaling networks and must be interpreted accordingly. For example, limitation of longevity in adults may positively foster survival of larval animals. Results presented here address longevity and thermal stress and do not preclude the possibility that DKF-2 positively controls C. elegans survival or homeostasis in response to other stimuli. Recent studies demonstrate that DKF-2 dramatically induces pro-survival responses in a DAF-16-independent pathway. 3 Conversely, a complex containing PKC␦ and PKD attenuates signaling downstream from activated B cell receptors (77). Thus, PKDs also serve as signaling inhibitors in mammalian systems. DKF-2 is not an integral component of the insulin/IGF-1-DAF-16 signaling pathway. Rather, modulation of DAF-16 activity by DKF-2 reflects a novel intersection between discrete pathways controlled by different plasma membrane intercalated lipids, DAG, and phosphatidylinositol 3,4,5-trisphosphate. It is not known if DKF-2 phosphorylates DAF-16 or exerts regulation in an indirect manner. The absence of a DKF-2 consensus substrate motif in the DAF-16 amino acid sequence suggests that indirect mechanisms merit consideration. Irrespective of the identity of the primary phosphorylation target, a PKD (DKF-2) creates a novel branch in a PLC-PKC-controlled network that couples DAG signals to the regulation of stress responses and life span.