Characterization of a Novel Protein Kinase D

Protein kinase D (PKD) isoforms are protein kinase C (PKC) effectors in diacylglycerol (DAG)-regulated signaling pathways. Key physiological processes are placed under DAG control by the distinctive substrate specificity and intracellular distribution of PKDs. Comprehension of the roles of PKDs in homeostasis and signal transduction requires further knowledge of regulatory interplay among PKD and PKC isoforms, analysis of PKC-independent PKD activation, and characterization of functions controlled by PKDs in vivo. Caenorhabditis elegans and mammals share conserved signaling mechanisms, molecules, and pathways Thus, characterization of the C. elegans PKDs could yield insights into regulation and functions that apply to all eukaryotic PKDs. C. elegans DKF-1 (D kinase family-1) contains tandem DAG binding (C1) modules, a PH (pleckstrin homology) domain, and a Ser/Thr protein kinase segment, which are homologous with domains in classical PKDs. DKF-1 and PKDs have similar substrate specificities. Phorbol 12-myristate 13-acetate (PMA) switches on DKF-1 catalytic activity in situ by promoting phosphorylation of a single amino acid Thr588 in the activation loop. DKF-1 phosphorylation and activation are unaffected when PKC activity is eliminated by inhibitors. Both phosphorylation and kinase activity of DKF-1 are extinguished by substituting Ala for Thr588 or Gln for Lys455 (“kinase dead”) or incubating with protein phosphatase 2C. Thus, DKF-1 is a PMA-activated, PKC-independent D kinase. In vivo, dkf-1 gene promoter activity is evident in neurons. Both dkf-1 gene disruption (null phenotype) and RNA interference-mediated depletion of DKF-1 protein cause lower body paralysis. Targeted DKF-1 expression corrected this locomotory defect in dkf-1 null animals. Supraphysiological expression of DKF-1 limited C. elegans growth to ∼60% of normal length.

A group of eight protein kinase C (PKC) 3 isoforms mediates actions of hormones and growth factors (GFs) that stimulate phospholipases (PLCs) ␤, ␥, and ⑀ (1-5). PKCs phosphorylate specific Ser/Thr hydroxyl groups in substrate/effector proteins, thereby modulating ion transport, secretion, cell growth, differentiation, gene transcription, and other physiological processes. Aberrant activation of PKCs is linked to tumor promotion, myocardial hypertrophy and infarction, and pathophysiological complications of type II diabetes (6 -11). Upon PLC activation, diacylglycerol (DAG) binding modules (C1 domains) ligate membraneintercalated DAG. Consequently, PKCs translocate from cytoplasm to a membrane surface enriched in phosphatidylserine (PS), a PKC activator. Association of PKCs with DAG and PS elicits expulsion of the pseudosubstrate domain (12) from the catalytic cleft and enables expression of intrinsic kinase activity (3,5,13,14). Ligand binding with Ͼ100 types of cell surface receptors triggers production of DAG in numerous mammalian cells. Knowledge of structures, activation mechanisms, and intracellular trafficking of PKCs is extensive. In contrast, information about in situ regulation, downstream effectors, interacting modulatory proteins, and the precise physiological roles of individual PKC isoforms is more limited.
The discovery of protein kinase D (PKD) isoforms (PKD, PKD2, and PKD3) revealed a new dimension in DAG-mediated signal transduction (15)(16)(17)(18). PKDs are composed of ϳ900 amino acids and contain tandem C1 domains and a Ser/Thr protein kinase module near their N-and C termini, respectively. PMA or DAG promotes redistribution of PKDs from cytoplasm to membranes and concomitant activation of PKD catalytic activity (19 -21). PKDs are distinguished from PKCs by (a) the presence of a pleckstrin homology (PH) domain, (b) the absence of a pseudosubstrate site, and (c) divergent substrate specificity (18 -23). Application of PKC-selective inhibitors and expression of constitutively active PKCs in transfected cells has disclosed that PKCs govern PKD activation (24 -26). Thus, PKDs are candidate physiological effectors for DAG-stimulated PKCs.
In mammals, PKD activation is associated with cell replication, tumorigenesis, Golgi vesicle fission and trafficking, adhesion, responses to stress, NFB activation, modulation of JNK activity, and other processes (5, 19 -21). Current concepts of regulation and physiological consequences of PKD activation are based predominantly on expression of wild type (WT) and mutant PKDs expressed in transiently transfected cells. Important insights into the properties of PKDs were acquired via transient transfection analysis, but inherent limitations of the methodology merit consideration. Typically, supraphysiological levels of WT/mutant PKDs accumulate in transfected cells. Thus, the potential for (a) mislocalization of activated or inhibitory ("dominant negative," "kinase dead") PKDs, (b) promiscuous phosphorylation of minor or nonrelevant substrates, and/or (c) association of PKD with atypical binding partners is increased. High level PKD expression may ablate counter-regulatory processes such as effector dephosphorylation or hinder operation of negative feedback loops. If these factors apply, data interpretation could be compromised and predictions of PKD physiological roles might be qualitatively or quantitatively imprecise.
Individual domains may specify subcellular location, level of catalytic activity, nuclear entry and exit, membrane association, or other attributes of PKDs (27)(28)(29)(30)(31)(32)(33). However, reports of complex or contradic-tory structure/function relationships have also confounded straightforward assignment of particular roles to specific domains in several instances. For example, both PKD that is diphosphorylated at Ser residues in the activation loop and PKD lacking phospho-Ser in the same loop are proposed as candidate mediators of NFB activation (25, 34 -36). Thus, a lack of information derived from (a) direct assessment or depletion of endogenous PKDs or (b) expression of near physiological levels of WT and mutant D kinases precludes definitive answers to the question of whether observed DAG/PKC/PKD signaling pathways are major or auxiliary (cell-specific or ubiquitous, etc.) contributors to overall cell/organism physiology. Full comprehension of effects of PKDs in homeostasis and hormone/GF actions will require more advanced knowledge of the complex regulatory interplay among PKD and PKC isoforms, rigorous analysis of modes of PKC-independent PKD regulation/ activation, and discovery of critical physiological changes in vivo that are specifically linked to alterations in PKD concentration and/or activity. At present, many central questions about PKDs remain unresolved including the following. 1) Are all PKD isoforms PKC effectors? 2) Can PKDs independently receive, amplify, and disseminate signals carried by DAG? 3) Are individual PKDs indispensable for specific physiological processes in intact animals? If so, which functions? 4) Will targeted overexpression in vivo reveal novel physiological roles for PKDs?
Answers to the posed questions may be acquired by coupling investigations on mammalian systems with complementary studies on PKDs in a model organism. Caenorhabditis elegans is an attractive choice because its cell biology, development, and physiology are characterized in exceptional detail (37)(38)(39)(40)(41)(42)(43). Moreover, C. elegans employs signaling molecules, mechanisms, and pathways that are conserved among eukaryotes (44 -49). Techniques for gene disruption, RNA interference (RNAi), and targeted mRNA/protein expression in specific cells are optimized for in vivo analysis in C. elegans (50 -54). Typically, information and insights derived from experimentation on C. elegans signal transduction networks complement and synergize with data and concepts obtained from investigations on parallel signaling pathways in mammals (39, 41, 44 -49). We now report the discovery, detailed characterization, and consequences of in vivo depletion of a novel D kinase family isoform (DKF-1). DKF-1 4 is a new prototype within the PKD family. In an accompanying article (89), we report on a series of unique regulatory properties in the C1a, C1b, PH, and kinase domains of DKF-1.

EXPERIMENTAL PROCEDURES
Isolation of cDNAs Encoding DKF-1-A BLAST search of the GenBank TM identified C. elegans expressed sequence tags (ESTs) that are homologous with cDNAs encoding human PKDs. Alignments of ESTs and cDNAs enabled design of a cDNA probe optimized for library screening. A 45-mer oligonucleotide (5Ј-CTGGATATGTGGTCTGT-TGGTGTCATTATT-TATGTCACGTTATCA-3Ј) was synthesized, end-labeled with 32 P, and used to screen a C. elegans cDNA library in gt10 bacteriophage as described by Hu and Rubin (55). Positive cDNA inserts were excised from recombinant phages and cloned in the plasmid pGEM7Z(ϩ) (Promega). One partial cDNA insert was repeatedly retrieved. Consequently, this (1.6 kbp) cDNA was excised by digestion with EagI and EcoRV, purified, and radiolabeled by random priming. A C. elegans cDNA library in bacteriophage ZAPII (Stratagene) was hybridized with the 32 P-labeled cDNA fragment. pBluescript SK phagemids that contained hybridizing cDNA inserts were excised from phage in Escherichia coli (in situ), amplified, and purified. Cloned cDNAs (2.3 kbp), which contain the complete coding sequence and 3Ј-untranslated region of DKF-1 mRNA, were isolated.
DNA Sequencing and Analysis-Genomic and PCR-amplified DNA and DKF-1 cDNAs were sequenced as described previously (56). Sequence comparisons and data base searches were performed with programs provided by NCBI servers at National Institutes of Health, Bethesda, MD. Protein domains were identified and characterized by using the SMART Web site (European Molecular Biology Laboratories, Heidelberg, Germany) (57). ClustalW programs were accessed at European Bioinformatics Institute, Hinxton, UK.
Purification of DKF-1 Fusion Protein Expressed in Escherichia coli-A DKF-1 cDNA fragment encoding amino acids 2-154 ( Fig. 1A) was synthesized via PCR as reported previously (58,59) using the high fidelity DNA polymerase PfuTurbo (Stratagene). Unique EcoRI (5Ј) and XhoI (3Ј) restriction enzyme sites were incorporated into the PCR primers. Product DNA was digested with EcoRI and XhoI and cloned into the bacterial expression plasmid pGEX-3X (GE Healthcare). Vector DNA encoding Schistosoma japonicum glutathione S-transferase (GST) precedes the cDNA insert. Fusion gene transcription/translation is controlled by an isopropyl-1-thio-␤-D-galactopyranoside (IPTG)-inducible tac promoter. Subsequent steps in fusion protein production are described in detail by Land et al. (60). Approximately 1.5 mg of highly purified DKF-1 fusion protein was isolated from a 1-liter culture of E. coli.
Production of Antiserum Directed against DKF-1-DKF-1 fusion protein was injected into rabbits (an 0.4-mg initial injection; 0.2 mg for each of four booster injections) at Covance Laboratories (Vienna, VA) at 3-week intervals. Antiserum was collected at 3-week intervals.
Growth and Synchronization of C. elegans-Bristol N2 WT C. elegans was grown, synchronized, harvested, and pulverized into powder in a mortar cooled with liquid N 2 . L1 larvae were harvested 6 h after hatching, L2 larvae at 20 h, L3 larvae at 29 h, L4 larvae at 40 h, and young adult C. elegans at 52 h; egg-laying adult animals were collected at 78 h. A complete description of the procedures is given in Hu and Rubin (55).
Preparation of Particulate and Cytosolic Fractions from C. elegans-Frozen, powdered nematodes were suspended in buffer and disrupted as reported previously (60). Homogenates were centrifuged at 100,000 ϫ g to separate and isolate supernatant solution (cytosol) and an insoluble pellet fraction. Cytosol and resuspended pellet were stored in liquid nitrogen. Details are provided in Ref. 60.
In Vitro Synthesis of DKF-1-A coupled transcription-reticulocyte lysate translation system (TNT TM , Promega Corp.) was programmed with 1 g of recombinant pBluescript that contains full-length DKF-1 cDNA. Incubation, denaturing electrophoresis, and autoradiography were performed as published previously (55,62).
Cell Culture-A cell line derived from a hamster subcutaneous tumor (AV-12) and human HEK293 cells were obtained from the American Type Culture Collection. Cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum in an atmosphere of 7% CO 2 and 93% air.
Electrophoresis of Proteins and Western Immunoblot Assays-Cytosol and particulate proteins were isolated from mammalian cells as stated previously (63). Triton X-100 was added or omitted from cell lysis buffer as indicated in the legend for Fig. 4 and under "Results." Proteins were size-fractionated by electrophoresis in a denaturing polyacrylamide (8% or otherwise indicated) gel as reported previously (64). BenchMarker TM prestained proteins (9 -182 kDa, Invitrogen) or Precision Plus Protein TM (10 -250 kDa, Bio-Rad) polypeptides were used as molecular size standards. Western blots of size-fractionated proteins were prepared, blocked, incubated with anti-DKF-1 IgGs (1:1000), and washed as indicated in previous studies (63,64). Unless noted otherwise, each lane in Western blots received 30 g of protein. Antigen⅐antibody complexes were visualized and quantified by using an enhanced chemiluminescence procedure and image analysis software (ImageQuant, GE Healthcare).
Purification of DKF-1 Fusion Protein Expressed in Sf9 Cells-Full-length DKF-1 cDNA was cloned downstream from a GST gene in a baculovirus transfer vector (pAcGHLT, Pharmingen). Recombinant baculovirus encoding GST⅐DKF-1 was generated and amplified as described in Islas-Trejo et al. (65). Procedures for infection, growth, and lysis of Sf9 cells are elaborated in detail elsewhere (60,65). Fusion protein was purified by affinity chromatography on GSH-Sepharose 4B (60). Fractions containing high levels of DKF-1 were pooled and stored at Ϫ80°C.
Expression of DKF-1 in Mammalian Cells-Full-length DKF-1 cDNA, excised from recombinant pBluescript plasmids (see above) by digestion with XbaI and Asp718, was cloned in the mammalian expression vector pCDNA3.1(ϩ) (Invitrogen, Carlsbad CA). Hamster AV-12 or human HEK293 cells were transfected with the recombinant DKF-1 transgene using Lipofectamine TM (Invitrogen). Stable cell lines that express modest amounts of DKF-1 protein were obtained by selection with 1 mg/ml G418 for 14 days. Individual drug-resistant clones were expanded and verified by independently determining DKF-1 protein levels and measuring PMA-stimulated kinase activity in cell extracts (see below).
Protein Determination-Protein concentrations were determined with the Bio-Rad DC protein assay reagents. Bovine albumin was employed as a standard.
Immunoprecipitation and in Vitro DKF-1 Kinase Assays-AV-12 cells were transfected by calcium phosphate precipitation (58,63) or uptake of Lipofectamine TM -recombinant DNA complexes. After 24 h, cells were harvested and homogenized in lysis buffer as reported previously (63). All operations (except kinase assays) were performed at 4°C. Cell lysates were centrifuged at 40,000 ϫ g for 30 min, and samples of supernatant solution (0.1-0.3 mg of protein) received 2 l (ϳ1 g) of affinity-purified anti-DKF-1 IgG. Incubation for 3 h yielded an optimal level of antigen⅐IgG complex formation. Subsequently, 25 l of protein A (or G)-Sepharose 4B beads (Zymed Laboratories Inc.) was added, and the incubation was continued for 60 min. Next, bead-bound immune complexes were washed three times with lysis buffer and then twice with kinase buffer containing 25 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , and 1 mM dithiothreitol.
Catalytic activity of DKF-1 was quantified by measuring incorporation of 32 P radioactivity from [␥-32 P]ATP into Syntide-2 peptide substrate (Calbiochem). Reaction buffer (30 l, containing 25 mM Tris-HCl, pH 7.4, 30 M peptide substrate, 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. Assays were terminated by the addition of 5 l of 0.2 M EDTA, pH 8.0, containing 60 mM NaF. Reaction mixtures were applied to P81 filter papers (Whatman) and subjected to extensive washing in 75 mM H 3 PO 4 ( 32 P-Syntide-2 binds P81 filters under acidic conditions, whereas 32 P i and [␥-32 P]ATP are washed away). After filters were washed and air-dried, 32 P radioactivity incorporated into Syntide-2 was measured in a scintillation counter. Phosphotransferase activity of purified GST⅐DKF-1 fusion protein from Sf9 cells was assayed as described above, except that 3 l of enzyme (ϳ50 ng protein) was used instead of an immunoprecipitate. Various combinations of PS, DAG, and Ca 2ϩ were added to kinase buffer to test their effects on catalytic activity. PKC⑀ peptide RFARKGSLRQKNV, PKC peptide YRRGSRRWK-KIY, Syntide-2 peptide PLARTLSVAGLPGKK, an optimum PKD peptide designed by Cantley et al. (23), AALVRQMSVAFFF, and myelin basic protein were tested as substrates for DKF-1. Potential phosphorylation sites are shown (above) in bold underlined italics.
To test the effects of PKC inhibitors on DKF-1 catalytic activity in situ, AV-12 cells were incubated with various concentrations of bisindolylmaleimide I (GF109203X) or bisindolylmaleimide IX (RO31-8220) (Axxora, LLC) for 1 h. Cells were then treated with specified concentrations of PMA for 10 min. In vitro kinase assays were performed on immunoprecipitated DKF-1 as described above. PKC inhibitors were tested in vitro by adding various amounts of GF109203X or RO31-8220 directly to the kinase reaction mixture.
Single Worm PCR Amplification of a Disrupted dkf-1 Gene-Deletions of 1-3 kbp in target genes were generated by mutagenesis with Psoralen and ultraviolet light by LiaoTeng Wang in the laboratory of Dr. Judith Kimble (Department of Genetics, University of Wisconsin). Individual adult worms from potential dkf-1 knock-out lines were extracted by incubation with 5 l of lysis buffer (50 mM KCl, 10 mM Tris-HCl, PH 8.2, 2.5 mM MgCl 2 , 0.45% Nonidet P-40, 0.45% Tween 20, 0.01% gelatin, and 5% proteinase K) at 65°C for 1 h. Large fragments of the dkf-1 gene were amplified by nested PCR using genomic DNA in diluted C. elegans extracts as template. PCR was first executed with an "external left" primer (5Ј-TCCGAAGAAGATGATCCAGG-3Ј) and an "external right" primer (5Ј-CAATTGTGCAATCACGCTCC-3Ј) pair. A second round of PCR used internal left (5Ј-AAACACAGTTCGAAGCGAGC-3Ј) and internal right (5Ј-AACTGCCACCCAATTTCAGG-3Ј) primers. Sizes of amplified DNA segments were estimated by electrophoresis in a 1% agarose gel. Exact deletion boundaries were determined by sequencing the amplified genomic DNA.
Preparation of Transgenic C. elegans-A DNA fragment (2.5 kbp) that precedes the 5Ј end of the dkf-1 structural gene was ligated to cDNA encoding the complete DKF-1 open reading frame in pBluescript SK by standard procedures. Subsequently, the fusion DNA (promoter ϩ open reading frame) was cloned into pPD 95.77, a C. elegans expression vector (52). A DKF-1 cDNA translation stop codon was eliminated by mutagenesis to permit in-frame ligation with DNA encoding green fluorescent protein (GFP). The GFP coding region is followed by translation termination and poly(A) addition signals. DKF-1 promoter/enhancer DNA ensures normal temporal developmental and cell-specific regulation and expression of DKF-1⅐GFP protein. Fluorescence signals emanating from individual cells of living or fixed worms are recorded via microscopy systems equipped for epifluorescence.
C. elegans was transformed by microinjecting the DKF-1⅐GFP kinasereporter plasmid (dkf-1P::DKF-1⅐GFP) as described previously (66). Transgenic C. elegans were identified by fluorescence microscopy and transferred to new plates to establish cloned populations. The dkf-1P::DKF-1⅐GFP chimeric gene was chromosomally integrated as described in a protocol provided by Dr. Michael Koelle (Department of Molecular Biophysics and Biochemistry, Yale University). Expression of DKF-1⅐GFP in C. elegans allows visualization of authentic promoter activation in individual cells in vivo. Fluorescence signals were captured as reported previously (60,65).
Ablation of DKF-1 Protein and Function by RNAi-Full-length DKF-1 cDNA was cloned into PPD129.36 (generously provided by Dr. A. Fire, Departments of Pathology and Genetics, Stanford University School of Medicine), a vector that has two opposing promoters for T7 RNA polymerase on sense and antisense DNA strands (67)(68)(69). E. coli HT115 (DE3) was transformed with recombinant plasmid, grown to A 595 ϭ 0.4 and seeded on plates containing 0.8 mM IPTG. L4-stage worms were transferred to plates containing transformed bacteria. Double-stranded RNAi ingested by C. elegans is disseminated to cells throughout the nematode via double-stranded RNA transporters (67)(68)(69). After 30 -36 h at 20°C, individual adult worms were transferred to fresh plates, and phenotypes of adults and offspring were observed by light microscopy.  analysis, mutagenesis, comparisons with sequences of several hundred phosphotransferases, and insights derived from crystal structures of prototypic Ser/Thr protein kinases (71)(72)(73)(74). Sequences in DKF-1 are tentatively linked to specific functions by analogy. A GXGXXGX 16 K motif (residues 433-455, Fig. 1A) provides sites that bind with phosphates of the substrate MgATP. Lys 455 is essential for catalysis and binding of ATP. A DFG tripeptide (residues 573-575, Fig. 1A) contributes a carboxyl side chain (Asp 573 ) that can mediate binding of divalent metal and ␥ phosphate of MgATP and stabilize pentavalent phosphorous in the transition state for the kinase reaction. Glu 599 , which is part of a conserved PPE motif (APE in other protein kinases), as well as Asp 611 and Arg 673 , stabilizes the catalytic core region. An XXDLKXX(N/D) motif is a Ser/Thr protein kinase "signature" sequence that guides protein substrate into an orientation favorable for catalysis (71)(72)(73)(74). The DKF-1 catalytic loop sequence, HCDLKPEN (residues 549 -556), diverges from the corresponding loop (YRDLKLDN) present in all members of the related PKC family. However, the HCDLKPEN motif is conserved among PKD isoforms (15)(16)(17)(18).

Characterization of cDNA Encoding C. elegans DKF-1-
The N-terminal portion of DKF-1 contains two Cys (and His)-rich sequences (HX 12 CX 2 CX 13 CX 2 CX 4 HX 2 CX 7 C, residues 99 -148 and residues 187-236, Fig. 1A) that constitute conserved regulatory regions (C1 domains) in DAG-activated PKCs (Fig. 1B). C1 domains fold into "finger-like" structures. Conserved spacing of six Cys and two His residues enables binding of two atoms of zinc in each finger. Zinc is essential for proper folding and function of PKCs.
Sequence Conservation and Divergence between DKF-1 and Human PKD Isoforms-The amino acid sequences of DKF-1 and protein kinases in standard data bases were aligned. DKF-1 is most closely related to mammalian PKD, PKD2, and PKD3 isoforms (50% similarity; Ն38% identical amino acids with each isoform) ( Table 1). No other protein kinases share Ͼ26% overall identity with DKF-1. Sequence homology varies markedly among domains ( Table 1). The catalytic domains of human PKD isoforms and nematode DKF-1 are maximally conserved (ϳ60% identity). DKF-1 and PKDs also share high levels of homology in the C1a and C1b regulatory regions (44 -58% identity, Table 1). Four His and 12 Cys residues, which are crucial for organizing/stabilizing zinc finger structures and binding physiological activators (DAG, PS) to C1 domains of PKCs and PKDs, are retained in perfect register in DKF-1. The PH domain and extreme C-terminal region of DKF-1 have lower but significant levels of sequence identity with corresponding segments in human PKDs (ϳ30 -33% identity plus ϳ20% conserved amino acid substitutions, Table 1). Conservation of amino acid sequences and the relative positions of domains along the polypeptide chain ( Fig. 1B, Table 1) indicate that DKF-1 is a member of the PKD family. However, major differences in structure/function relationships and mode of activation (elaborated in detail below and in an accompanying article (89)) indicate that DKF-1 is a novel, previously uncharacterized PKD isotype.
Organization of the C. elegans dkf-1 Gene-During the course of our investigations on DKF-1, the C. elegans Genome Project deposited genomic DNA sequence data for cosmid WO9C5 in GenBank TM and Wormbase. We established the organization of exons and introns for the dkf-1 structural gene by identifying genomic DNA sequences in recombinant cosmids that overlap with segments of DKF-1 cDNA. Our results and subsequent analysis by the C. elegans Genome Consortium are in agreement ( Table 2). The dkf-1 structural gene contains 11 exons and spans ϳ8 kbp of DNA (Table 2). 5 H. Feng and C. S. Rubin, unpublished observations. FIGURE 2. Preparation of IgGs that specifically bind recombinant kinase and endogenous DKF-1 from all developmental stages. A, a GST-partial DKF-1 (residues 2-154) fusion protein was expressed in E. coli, extracted, and purified to near homogeneity as described under "Experimental Procedures." The fusion protein has an apparent molecular weight of 43,000 in SDS-PAGE. A 10% polyacrylamide gel, stained with Coomassie Blue, is shown. Lane 1, total protein from uninduced E. coli; lane 2, total protein from E. coli induced with 1 mM IPTG; lane 3, soluble protein from induced E. coli; lane 4, column flow-through proteins; lanes 5 and 6, buffer wash; lanes 7 and 8, purified DKF-1 fusion protein eluted from GSH-Sepharose 4B beads by two successive applications of buffer containing 10 mM GSH. B, a Western blot was prepared as described under "Experimental Procedures." Each lane received 30 g of total protein from a mixed population (larvae and adults) of C. elegans. Lanes 1 and 3 were incubated with affinity-purified IgGs (0.5 g/ml, 1:500 relative to serum) directed against DKF-1. Lane 2 was incubated with preimmune serum (1:500); excess purified antigen (3 g) was present when lane 3 was probed with affinity-purified antibodies. Chemiluminescence signals were recorded on x-ray film. C, a Western immunoblot is shown. Lanes 1-3 received cytosolic proteins (30 g) isolated from AV-12 cells stably transfected with a DKF-1 transgene. Lane 4 contained 30 g of cytosolic proteins obtained from control cells. Lanes 1, 3, and 4 were probed with affinity-purified anti-DKF-1 IgGs. Excess purified DKF-1 antigen (3 g) was included when lane 3 was incubated with the IgGs. Production and Specificity of Anti-DKF-1 Immunoglobulins-A fusion protein that consists of 153 amino acids from DKF-1 (residues 2-154, Fig. 1A) appended to the C terminus of GST was produced in E. coli and purified as described under Experimental Procedures." Antisera directed against the 43-kDa DKF-1 fusion protein ( Fig. 2A) were produced in rabbits. IgGs that bind DKF-1 were purified from serum by affinity chromatography using Sepharose 4B beads derivatized with partial DKF-1 protein. Affinity-purified anti-DKF-1 IgGs bound (a) an 81-kDa protein in homogenates of C. elegans (Fig. 2B, lane 1) and (b) a protein of similar size in cytosol isolated from hamster AV-12 cells that contained a DKF-1 transgene (Fig. 2C, lane 1). Target antigen was not evident in extracts of nontransfected AV-12 cells (Fig. 2C, lane 4). Moreover, the 81-kDa polypeptide was not detected when Western blots of proteins from C. elegans and transfected AV-12 cells were probed with preimmune IgGs or anti-DKF-1 IgGs that were preincubated with excess DKF-1 fusion protein (Fig. 2, B and C, lanes 2 and 3). The size and identity of the C. elegans kinase were confirmed by in vitro translation. DKF-1 mRNA encoded an [ 35 S]Met, [ 35 S]Cys-labeled polypeptide that exhibited a M r of 81,000 in a denaturing polyacrylamide gel (Fig. 2D, lane 3).
Regulation of DKF-1 Expression during Development-Cytosolic and total particulate proteins were isolated from C. elegans at each stage of development. Western immunoblot analysis revealed that DKF-1 is expressed robustly in embryos. Lower but substantial amounts of the kinase are present as animals progress through four larval stages and adulthood (Fig. 2E). DKF-1 is isolated in the cytosolic fraction of C. elegans homogenates (Fig. 2E). DKF-1 protein accumulates principally in the cytoplasm of (a) AV-12 cells carrying a constitutively expressed dkf-1 transgene and (b) Sf9 cells infected with recombinant baculovirus that contains a dkf-1 cDNA insert (data not shown).
Purification and Properties of Recombinant DKF-1-A preparation of DKF-1 that is free of other PKDs, PKC isoforms, and unrelated but potentially interfering protein kinases was required for unequivocal characterization of the novel C. elegans kinase. Consequently a chimeric gene, encoding full-length DKF-1 polypeptide fused (in-frame) to the C terminus of GST, was created by inserting dkf-1 cDNA into a baculovirus transfer/fusion vector, pAcGHLT. Recombinant virus was generated and used to infect Sf9 cells. Cytosol was isolated 4 days after infection, and DKF-1 was purified by affinity chromatography. Purified DKF-1 fusion protein exhibited modest basal activity in the absence of co-factors. Phosphorylation of Syntide-2 (PLARTLSVAGLPGKK), a preferred substrate for mammalian PKDs, was only minimally increased by DAG or PS alone. However, a combination of PS and DAG synergistically increased DKF-1 phosphotransferase activity 15-20-fold (Table  3). Evidently, simultaneous binding of both co-factors is crucial for establishing/maintaining maximal kinase activity. DKF-1 phosphotransferase activity was not influenced by calcium which stimulates conventional PKCs (Table 3). Native nonfused DKF-1 was isolated and purified from extracts of transiently or stably transfected cells by immunoprecipitation. In vitro kinase assays performed with native or recombinant DKF-1 yielded essentially the same results under all conditions. DKF-1 and PKD Have Similar Substrate Specificities-AV-12 cells that express a DKF-1 transgene were incubated with 300 nM PMA and then lysed in buffer containing 1% Triton X-100. DKF-1 was immunoprecipitated from the detergent-soluble fraction of the lysate and assayed for kinase activity using several peptide substrates. Maximal kinase activity was evident when Syntide-2 served as DKF-1 substrate (Fig. 3A). In contrast, modified PKC⑀ and PKC pseudosubstrate peptides, in which Ser substitutes for Ala, are poor phospho-acceptors for DKF-1. A model small protein substrate, myelin basic protein, which is phosphorylated by many PKC isoforms and other regulatory kinases, is not a good target for DKF-1 (Fig. 3A). Using combinatorial libraries, Cantley's group (23) demonstrated that peptides containing Leu at the Ϫ5 position, an aliphatic residue at Ϫ4, a basic residue at Ϫ3, and a hydrophobic amino acid at the ϩ1 position (P-Ser site ϭ 0) constitute optimal substrates for mammalian PKDs. Each critical position in the motif is occupied by an appropriate amino acid in Syntide-2 (PLARTLS-VAGLPGKK). A distinct peptide designed by Cantley's (23) rules for optimal substrates, AALVRQMSVAFFF, is also a good phospho-acceptor for DKF-1-mediated catalysis (Fig. 3A). Similarities among preferred substrates for DKF-1 and PKDs support the idea that the newly discovered DAG/PS-stimulated kinase is a new member of the PKD family.   (Fig. 3B). DKF-1 and PKD Are Activated by Different Mechanisms in Intact Cells-Activation of intracellular PKD by hormones, GFs, phorbol esters or second messengers is blocked by permeable PKC inhibitors (19 -21, 26). Because purified mammalian PKDs are not markedly affected by the inhibitory drugs in in vitro kinase assays (75), these observations indicate that upstream PKCs regulate PKD activity and function. As expected, incubation of cells with a PKC inhibitor (GF109203X or RO31-8220) before and during treatment with PMA potently suppressed endogenous PKD activation (ϳ70 -75% decline, Fig. 3C, right). In contrast, robust stimulation of DKF-1 caused by PMA was not altered by treating cells with a high concentration of inhibitor (Fig. 3C, left). PMA-mediated stimulation of PKCs, PKDs, and DKF-1 was abolished by 1 M staurosporine, a nonselective inhibitor of a broad spectrum of protein kinases. Thus, suppression of mammalian PKD by GF109203X (or RO31-8220, data not shown) and the striking insensitivity of DKF-1 to these inhibitors (Fig. 3C) are not due to differences in accessibility of the kinase in situ. Instead, the data reveal inherent, fundamental differences between properties and modes of physiological regulation of the two PKD family isoforms. In vitro, GF109203X has no effect on DKF-1 kinase activity and (as expected) is only a weak, direct inhibitor of PKD (Fig. 3D). Results presented in Fig. 3, C and D, and previous studies (19 -21, 26, 75) show that PKD activity is controlled by upstream PKCs. PMA stimulates DKF-1 kinase activity when PKCs in the same intracellular environment are inactive.
Bombesin Elicits PKC-independent Activation of DKF-1-The preceding results suggest that physiological activators of DAG synthesis will promote DKF-1 activation in situ. This proposition was tested in cells expressing the bombesin BB 2 receptor. Bombesin receptors are present in many tissues and regulate a broad spectrum of physiological processes, including smooth muscle contraction, pancreatic secretion, gastrin release, satiety, lung development, thyrotropin secretion, lymphocyte chemotaxis, and activation of mammalian PKDs (28,(75)(76)(77). Bombesin binding with BB 2 , a serpentine (seven transmembrane segments) receptor, causes efficient and selective activation of heterotrimeric G q /G 11 GTP-binding proteins. G␣ q -GTP (or G␣ 11 -GTP) binds and activates PLC␤, thereby stimulating DAG synthesis at the plasma membrane. Typically, accumulation of DAG is rapid but transient, because the second messenger is robustly metabolized.
Incubation of cells with 100 nM bombesin peptide increased DKF-1 kinase activity ϳ9-fold within 2 min (Fig. 4A). Kinase activity declined ϳ50% over the succeeding 18 min. A basal level of phosphotransferase activity was fully restored after 150 min (data not shown). The observations are consistent with (a) a rapid but transient elevation in DAG and (b) an ability of members of the PKD family to remain active for an extended time after DAG levels decline. Bombesin-induced translocation of DKF-1 to membranes is documented in a companion article (89).
Thr 588 Regulates DKF-1 Catalytic Activity-Activation loops in mammalian PKD isoforms contain the sequence SFRRSV (amino acids

In vitro activation of C. elegans DKF-1
Phosphotransferase activity of GST-DKF-1 was measured as described under ЉExperimental Procedures.Љ The specified concentrations of PS, DAG, and/or Ca 2ϩ were added as indicated. For immunodepletion, a sample of DKF-1 was immunoprecipitated with 0.5 g of affinity purified anti-DKF-1 IgG that was bound to protein A-Sepharose 4B. Kinase activity remaining in the supernatant solution was then quantified. All assays were repeated three times. Typical results are shown.
Mutants were designed to test the importance of Thr 588 in kinase activation; replacement of Thr 588 with Ala will eliminate interactions mediated by the polar hydroxyl group and extinguish phosphorylation at the putative target site, whereas substitution of Thr 588 with Ser might sustain normal function and regulation. Mutation of Thr 588 to Ala sharply suppressed (Ͼ85%) basal and phorbol ester-stimulated DKF-1 phosphotransferase activity in situ (Fig. 5B, compare lanes 2 and 3 with lanes  7 and 8), whereas interchange of Ser for Thr 588 produced a PMA-activated kinase that was indistinguishable from WT DKF-1 (Fig. 5B, compare lanes 2 and 3 with lanes 11 and 12). The results demonstrate the central importance of Thr 588 in elevating and controlling DKF-1 kinase activity. Both WT DKF-1 and the Ala 588 mutant were targeted to mem-branes upon incubation of cells with PMA for 10 min (Fig. 5C, and immunolocalization data shown in an accompanying article (89)). Additional mutations (encoding Asp 588 or Glu 588 ) were incorporated into DKF-1 transgenes to mimic negative charge (P-Thr 588 ) that might be associated with PMA-activated WT DKF-1. These mutant proteins, WT DKF-1, and the Ser 588 and Ala 588 DKF-1 variants were expressed at similar levels in AV-12 cells (Fig. 5A, upper panel). However, in nonstimulated cells, the specific activity of DKF-1(Glu 588 ) was elevated ϳ3-fold relative to WT DKF-1 (Fig. 5B, lanes 2 and 9). Incubation of transfected cells with PMA caused a large increase in specific kinase activity of WT DKF-1 (Fig. 5B, lanes 2 and 3). In contrast, phorbol ester treatment did not alter DKF-1(Glu 588 ) activity (Fig. 5B, lanes 9 and 10). Like DKF-1(Glu 588 ), DKF-1(Asp 588 ) exhibits high basal activity that is not further enhanced by PMA. 5 Replacement of Thr 588 with nonacidic amino acids (e.g. Ala) yielded DKF-1 mutants that exhibit minimal basal activity and are unresponsive to PMA (Fig. 5B, lanes 7 and 8).
Possible linkage between phosphorylation of DKF-1 and expression of maximal DKF-1 catalytic activity with an optimal exogenous substrate was explored via a dephosphorylation strategy. PMA-activated DKF-1 was immunoprecipitated from cell extracts and incubated with highly purified protein phosphatase 2C (PP2C). Treatment with the phosphatase diminished DKF-1 phosphotransferase activity in a timeand temperature-dependent manner (Fig. 6). Under optimal conditions ϳ90% of the kinase activity was abolished by PP2C.

. DKF-1 is activated via bombesin receptors and PLC.
A, transfected AV-12 cells expressing bombesin BB 2 receptors and DKF-1 were incubated with 100 nM bombesin for the indicated time intervals. Subsequently, cells were lysed in buffer containing 1% Triton X-100, DKF-1 was isolated and purified by precipitation with anti-DKF-1 IgGs bound to protein G-Sepharose 4B beads, and kinase assays were performed (see "Experimental Procedures"). The inset is a Western immunoblot showing that similar amounts of DKF-1 protein were immunoprecipitated and assayed at each time point during the course of bombesin treatment. The blot was probed sequentially with anti-DKF-1 IgGs and secondary antibodies coupled to peroxidase. Chemiluminescence signals were recorded on x-ray film. B, transfected AV12 cells described in A were incubated with 5 M GF109203X, 5 M U73122, or vehicle for 1 h. Subsequently, cells were treated with 100 nM bombesin or buffer (control) for 2 min. DKF-1 was isolated from cell extracts by immunoprecipitation, and in vitro kinase assays were carried out as indicated under "Experimental Procedures." The lower panel presents a Western immunoblot demonstrating that similar amounts of DKF-1 protein were immunoprecipitated and assayed for each set of experimental conditions. Experiments in A and B were performed three times. Similar results were obtained in each repetition. After denaturing electrophoresis, incorporation of 32 P radioactivity into DKF-1 was assessed by autoradiography (lower panel) as described previously (88). B, immune complexes were assayed for kinase activity using Syntide-2 as substrate (see "Experimental Procedures"). Results from a typical set of in vitro kinase assays are depicted. C, AV-12 cells stably transfected with a DKF-1(T588A) transgene were exposed to vehicle (Ϫ) or 1 M PMA (ϩ) for 10 min as indicated. Cytosolic (S) and membrane (P) proteins were prepared from cells, purified by denaturing electrophoresis, and transferred to a polyvinylidene difluoride membrane as described under "Experimental Procedures." The effect of PMA on distribution of DKF-1(T588A) between cytosol and membranes was determined by probing the Western blot with anti-DKF-1 IgGs. Typical data are presented in A-C. Similar results were obtained from two additional replications of the experiments.
Treatment of cells with PMA markedly stimulated phosphorylation of DKF-1 protein in subsequent in vitro kinase assays (Fig. 5A, lower  panel, lanes 2 and 3). Thr 588 is essential for the phospho-acceptor activity DKF-1. Replacement of Thr 588 with Ala, Glu, Asp, or other amino acids (except Ser) nearly extinguished DKF-1 phosphorylation (Fig. 5A,  lanes 7-12). Collectively, the data strongly suggest that phosphorylation of the DKF-1 activation loop at Thr 588 triggers an alteration in the configuration of the catalytic domain that enables maximal rates of phosphorylation of extrinsic substrates and DKF-1 itself. Total loss of DKF-1 phosphorylation in the kinase-dead (Lys 455 to Gln) variant (Fig. 5A,  lower panel, lanes 13 and 14; Fig. 5B, lanes 13 and 14) further suggests that integrity of the ATP binding site is essential for DKF-1 activation by DAG/PMA. Acidic amino acid substitutions at Thr 588 mimic DAGinduced increases in DKF-1 activity to the extent that constitutively active kinases are produced in cell lines. The exact molecular configuration induced in the DKF-1 catalytic site when phosphate was incorporated at the hydroxyl group of Thr 588 was not achieved in DKF-1(Glu 588 ) (Fig. 5, lanes 9 and 10) or DKF-1(Asp 588 ) (not shown). Instead, constitutive kinase activity in these mutants was limited to ϳ30 -35% of the values determined for WT DKF-1 after exposure to saturating amounts of PMA. Despite this minor caveat, we conclude that the presence, precise context, and phosphorylation of a Thr or Ser hydroxyl group in the activation loop is evidently a stringent requirement for regulation of DKF-1 by PMA or DAG second messenger.
Replacement of Gln 584 with Ser yielded a DKF-1 activation loop sequence (SFRKTV) that closely resembled the motif containing two regulatory trans-phosphorylation sites in PKDs (SFRRSV). However, DKF-1 (Ser 584 ) was activated with the same time course and reached the same maximal activity as the WT enzyme (data not shown). Moreover, WT and DKF-1(Ser 584 ) activities were not inhibited by GF109203X or other PKC inhibitors (Fig. 5B, lanes 4 -6). Thus, DKF-1 and PKDs are regulated by distinct upstream activators and different mechanisms.
Substitution of Thr 588 with Ser did not affect properties of DKF-1. Both wild type and variant (DKF-1(Ser 588 )) kinases were stimulated ϳ8-fold in cells incubated with PMA (Fig. 5B, lanes 2, 3, 11, and 12). Preincubation of cells with 3.5 M GF109203X did not alter the magnitude of activation of DKF-1 or DKF-1(Ser 588 ) by subsequent addition of PMA. Thus, the presence of Thr or Ser at amino acid 588 in the activation loop (QFRK(T/S)V) enables PKC-independent activation of DKF-1.
dkf-1 Gene Promoter Activity and Cognate DKF-1 Protein Kinase Are Selectively Expressed in a Subset of Cells in Vivo-Promoter/enhancer DNA that precedes the authentic dkf-1 structural gene was coupled to the 5Ј end of full-length DKF-1 cDNA. The resulting dkf-1P::DKF-1 minigene was inserted upstream from GFP reporter cDNA in a C. elegans expression plasmid. Subsequently, the dkf-1P::DKF-1⅐GFP fusion gene was integrated into C. elegans genomic DNA, and stable transgenic lines of C. elegans were propagated. Levels of DKF-1⅐GFP were monitored by Western immunoblotting (Fig. 7A), and expression of DKF-1 fusion protein in individual cells in vivo was assessed by fluorescence microscopy (Fig. 7, B-D). A highly reproducible pattern of DKF-1 expression was observed as animals matured from embryo to adult. Intense fluorescence signals corresponding to DKF-1⅐GFP revealed robust kinase accumulation in both (a) a region bounded by the anterior and posterior bulbs of the pharynx (Fig. 7, B, B2, and C) and (b) a tail area that contains lumbar, dorsorectal and pre-anal ganglia (Fig. 7D).
Specifically, DKF-1 is differentially enriched in a cluster of cells that are immediately adjacent to the posterior pharyngeal bulb (Fig. 7B). Strong signals also emanate from cells positioned along the lateral surface of this bulb in animals carrying the dkf-1P::DKF-1⅐GFP transgene (Fig. 7B2, arrows). At the anterior pharyngeal bulb, DKF-1 accumulates selectively in bodies and in very thin processes (dendrites and axons) of two neurons (Fig. 7C). High cell density and complex organization of neuronal ganglia in the head (and tail) region of C. elegans preclude determination of the exact identities of individual cells that exhibit elevated dkf-1 promoter activity and, therefore, enrichment in DKF-1⅐GFP polypeptide.
Despite the limitations listed above, a comparison of our current results with the C. elegans anatomy data base is quite informative. This approach focuses attention on a small group of "candidate DKF-1 positive cells" that will guide further studies. Nearly all cells expressing DKF-1 in Fig. 7, B, B2, C, and D, appear to be neurons. Two fluorescent cells with similar sizes and locations (at the anterior edge of the isthmusposterior bulb, Fig. 7, BN and B, arrows) may be M2 motor neurons. M2 neurons innervate muscles in the anterior bulb of the pharynx. The location of the more posterior fluorescent neuron in Fig. 7C (upper arrow) approximates the position of the cell body of an NSM neuron. NSM is a multifunctional neuron involved in sensing food and physical stretch, locomotory behavior, and secretion of growth factors and hormones. DKF-1 also accumulates in a cell resembling I1 (Fig. 7C, lower  arrow), an interneuron that links sensory and effector neurons. Other candidate DKF-1-enriched cells in the pharyngeal region include: the AWB, ADL, and ADF chemosensory neurons; and AVB and AIA interneurons, which process sensory inputs and determine (directly or indirectly) the course of motor neuron actions at neuromuscular junctions, thereby governing body movement.
In C. elegans tail, DKF-1⅐GFP expression is differentially elevated in neurons located within the dense neuropile of several tail ganglia (Fig.  7D). This arrangement suggests that activation of the DKF-1 might FIGURE 6. DKF-1 is inactivated by incubation with protein phosphatase 2C. AV-12 cells that contain a DKF-1 transgene were incubated in 0. 3 M PMA for 10 min. Subsequently, cells were lysed, and DKF-1 was precipitated from cell extracts with affinitypurified anti-DKF-1 IgGs. Immune complexes (on protein A-Sepharose 4B beads) were incubated in the absence or presence of 1 g of soluble PP2C under the indicated conditions. Activity of divalent metal-dependent PP2C was quenched by the addition of 10 mM EDTA. After beads were washed five times with kinase reaction buffer, DKF-1 kinase activity was measured using Syntide-2 as substrate (see "Experimental Procedures"). A, precipitated DKF-1 was incubated with bovine serum albumin (Control) or PP2C in the presence or absence of 10 mM EDTA for the indicated time intervals. B, precipitated DKF-1 was incubated with PP2C for 20 min at the indicated temperatures. After washing (see above), DKF-1 mediated Syntide-2 phosphorylation was assayed, and data were normalized relative to untreated control samples that were held on ice during and prior to kinase assays. Panels below each graph show (via Western immunoblots) that the amount of DKF-1 was not altered during the course of immunoprecipitation, PP2C treatment, and phosphotransferase assay.
generate signals involved in tail muscle contraction (movement). The pattern of fluorescence depicted in Fig. 7D reveals that cell bodies and/or processes of phasmid neurons (PHA, PHB, and PHC), interneurons (PVC, DVA, DVB, PVQ, PVT) and motor neurons (VD13, DD6, VA12) are candidate sites for accumulation of DKF-1 protein. The phasmid neurons have openings in the tip of the tail that receive chemosensory information from the environment. Sensory signals are routed to the pre-anal ganglion via phasmid processes that pass through lumbar commissures and synapse with various interneurons, including DVA, PVQ, AVA, and PVC, as well as motor neurons (VA12, DA9 etc.). Because some of the interneurons run along the nerve cords and terminate in the "brain" of the worm (nerve ring around the pharynx), it is possible that DKF-1 phosphotransferase activity plays an important dual role in routing local sensory data to both distal and proximal sites of integration (e.g. pharyngeal nerve ring and pre-anal ganglion, respectively). In concert with other inputs, DKF-1 activation would promote an appropriate, fully coordinated pattern of motor neuron activation/ inactivation and muscle contraction/relaxation in response to positive or negative (e.g. attractants, toxins) environmental stimuli. This idea is supported by experimental data derived from "knock-out" animals carrying two disrupted copies of the dkf-1 gene (see Fig. 8 and under "Results" below).
Identities of cells that accumulate DKF-1 can be unequivocally established in future studies by determining (via confocal microscopy) patterns of co-expression of DKF-1⅐GFP with cell-specific marker proteins tagged with yellow, red, cyan fluorescent proteins, etc., in living, transgenic C. elegans. In addition, dkf-1 promoter activity/DKF-1 fusion protein accumulation patterns can be monitored in vivo using WT animals and a systematically prepared series of C. elegans variants that lack one potentially relevant cell (as a consequence of molecular genetics-or laser-mediated ablation technology) (78).
Depletion or Elimination of DKF-1 Protein in Vivo Reveals a Novel Physiological Role for a Member of the PKD Family-DKF-1 mRNA and protein levels were diminished in vivo by using RNAi methodology.
Full-length DKF-1 cDNA was cloned into an RNAi vector (pPD139.06). The cDNA insert is located between opposing, IPTG-inducible, T7 promoters on sense and antisense DNA strands. E. coli was transformed with recombinant plasmid and double-stranded RNA (dsRNA) production was induced by adding IPTG. C. elegans cells take up dsRNA from ingested, transformed E. coli. Ubiquitous dsRNA transporters disseminate "silencing" reagent (RNAi) throughout the tissues of C. elegans (79,80). Cells in the nematodes contain enzymes that cleave high molecular weight dsRNAs into small fragments (ϳ25 bp) that efficiently and selectively mediate degradation of a target transcript (DKF-1 mRNA in this instance) (81,82).
L4 stage worms were fed with bacteria producing RNAi directed against DKF-1, and phenotypes of these worms and their progeny were observed with a dissecting microscope. DKF-1 protein content decreased dramatically (ϳ90%) in vivo when cognate RNAi was ingested (Fig. 8A, upper panel). In contrast, the concentration of a control protein (PKC-1) was not altered (Fig. 8A, lower panel). A distinctive phenotype was discovered in DKF-1-deficient C. elegans. Normal C. elegans traverse the agar medium on culture plates with a sinusoidal motion. This pattern is compromised by DKF-1 RNAi; worms exhibited a "zigzag" motility pattern with markedly altered amplitude (Compare Fig. 8B, panel 1 with panels 2 and 3). Two days after RNAi feeding, the mobility of adult worms was impaired. Severe or complete loss of muscle contraction near the anus caused the partially paralyzed tail region of the animals to be dragged along by the upper body. Thus, DKF-1 deficiency apparently causes a neuromuscular defect (uncoordinated (unc) phenotype) in the tail region.
A C. elegans gene disruption project in the laboratory of Judith Kimble (Department of Genetics, University of Wisconsin) facilitated the isolation of animals with a disrupted dkf-1 gene (generously provided by Liaoteng Wang, University of Wisconsin). Back-crossing with wild type C. elegans eliminated extraneous mutations and yielded viable worms with a dkf-1(Ϫ/Ϫ) genotype. Fragments of the disrupted dkf-1 gene were generated by PCR amplification using genomic DNA tem- plates from individual worms in concert with a set of nested primers (Fig. 8C). Sequencing of PCR-amplified DNA revealed that a 1366-bp segment was deleted from the dkf-1 gene. This deletion corresponds to nucleotides 12736 -14281 in the sequence of cosmid W09C5. Comparison of cosmid (genomic) and cDNA sequences indicates that exon 3 and portions of two introns are deleted from the dkf-1 gene. The expected splicing of exon 2 to exon 4 in the disrupted gene will cause a frameshift in DKF-1 mRNA that places a translation stop codon after amino acid 83. The resulting transcript will encode a chimeric protein composed of amino acids 1-78 of DKF-1 plus five C-terminal amino acids from an incorrect reading frame. The calculated M r for the chimera is 9,600. Because the truncated DKF-1 polypeptide terminates before the first Cys-rich domain (C1a) and also lacks a kinase domain, the animals have a null phenotype for all established functional domains in DKF-1. dkf-1(Ϫ/Ϫ) C. elegans and WT worms that contain DKF-1 RNAi have the same "uncoordinated" phenotype ( Fig. 8B, panels 3 and 2, respectively). Null mutants move normally from the time of hatching to the egg-laying stage and have no obvious defects in embryogenesis. Adult dkf-1(Ϫ/Ϫ) C. elegans display uncoordinated tail movement at day 5 (2 days after the final larval stage, L4) of the life cycle. The phenotype persists until death. A link between DKF-1 protein and null phenotype (unc) was rigorously confirmed by demonstrating that introduction of a dkf-1P::DKF-1⅐GFP fusion gene into dkf-1(Ϫ/Ϫ) animals corrected the movement disorder. Cell-specific expression of the transgene restored normal sinusoidal motion (data not shown). Western immunoblots demonstrated that the predicted 107-kDa DKF-1⅐GFP fusion protein accumulates in rescued transgenic nematodes. Endogenous 81-kDa DKF-1 is evident in extracts derived from wild type animals but is absent in samples prepared from C. elegans carrying two copies of the disrupted dkf-1 gene (Fig. 8D).
DKF-1 Overexpression Restricts Post-larval Growth and Creates a "Small" Body Size Phenotype-An approach that often illuminates regulatory functions in vivo involves phenotype assessment in C. elegans overexpressing a gene of interest. During rescue of dkf-1 null animals with a chimeric transgene (see above), a novel phenotype emerged.
Although uncoordinated movement was compensated by synthesis of DKF-1⅐GFP, rescued C. elegans adults were distinguished from WT animals by size. The rescued nematodes have small cells and are substantially reduced in overall body size (Fig. 9A, b). In particular, abundant DKF-1 attenuates growth that occurs upon the L4 to adult transition. This idea was validated by creating stable transgenic C. elegans that overexpress DKF-1⅐GFP in appropriate cells in the context of a WT background (Fig. 9A, d). C. elegans body length was measured at L4 and adult stages. WT animals, dkf-1 null nematodes rescued with a dkf-1-gfp transgene, and WT nematodes overexpressing DKF-1 fusion protein were studied in parallel (Fig. 9B). At the last larval stage (L4), animals containing excess DKF-1 are ϳ20% shorter than WT worms. During the next 2 days, wild type C. elegans increased Ͼ50% in length. In contrast, worms with elevated DKF-1 content elongated only ϳ10% (Fig.  9B). Consequently, mature animals with constitutively elevated DKF-1 content are only ϳ60% as long as their wild type counterparts. Because the small worms are also thin relative to wild type C. elegans, their total internal volumes are likely to be diminished by Ͼ50%.

DISCUSSION
The C. elegans dkf-1 gene encodes a novel PMA/DAG-regulated Ser/ Thr protein kinase. Like cPKCs and nPKCs, the 81-kDa DKF-1 polypeptide possesses tandem N-terminal C1 domains that control phosphotransferase activity (see accompanying article (89)) in a C-terminal kinase module. Several divergent features distinguish DKF-1 from PKCs. DKF-1 lacks (a) a Ca 2ϩ -binding C2 domain inherent in cPKCs, (b) an N-terminal (C2-like) extension region that supports Ca 2ϩ -independent lipid binding in nPKCs, and (c) a pseudosubstrate inhibitory segment that is present in all PKC isoforms. DKF-1 contains a PH domain (a structural module not found in PKCs) and phosphorylates different effector proteins, because its optimal target amino acid sequence (LVRQMSVAF, critical amino acids Leu(Ϫ5), Arg(Ϫ3), and Met(Ϫ1) precede the C-terminal target Ser(0)) is not related to established PKC phosphorylation sites. Peptides and proteins that are effi-  (60)), was not altered. Note that the phosphorylation state of PKC1 (indicated by bands Ͼ90 kDa) varies, but total PKC1 protein is unchanged. B, worm phenotypes were revealed by light microscopy. Panel 1 shows tracks of WT C. elegans on an agar plate. Panels 2 and 3 depict highly atypical tracks produced by (homozygous null) dkf-1(Ϫ/Ϫ) animals and worms fed with DKF-1 RNAi, respectively. C, PCR analysis was performed on genomic DNA templates from individual WT and putative DKF-1 null C. elegans (see "Experimental Procedures"). Nested primers that hybridize only with the dkf-1 gene were used. Amplified DNA products were analyzed in a 1% agarose gel. Ethidium bromide-stained DNA is shown. ciently phosphorylated by PKCs are very poor substrates for DKF-1. Thus, DKF-1 is excluded from the PKC family.
Systematic structure/function analysis (reported in this and an accompanying article (89)) identified DKF-1 as a close, authentic relative of mammalian PKDs. DKF-1 and PKDs have the same domain composition (two C1 modules, PH domain and kinase domain). These discrete structural segments are incorporated in the same order along the polypeptide sequences of DKF-1 and all three PKD isoforms. Each DKF-1 domain shares a high level of amino acid sequence homology with corresponding regions in PKDs. DKF-1 and PKDs have similar substrate preferences in in vitro kinase assays. Moreover, except for members of their own family, the catalytic domains of DKF-1 and PKD are most homologous with catalytic regions of calmodulin-dependent kinases, death (apoptosis)-associated protein kinase, myosin light chain kinase, and ChK2, a protein kinase that coordinates DNA repair with retardation of the cell cycle. These relationships place DKF-1 and PKDs in the CAMK group of the kinome; In contrast, PKCs are prototypes for a distinct segment of the kinome, named the AGC group (83).
Hormones and GFs trigger activation of PKDs by stimulating PLCcatalyzed accumulation of DAG in membranes. Subsequent phosphorylation of PKDs at two sites in the activation loop (Ser 744 and Ser 748 in PKD) elicits conversion of an inactive enzyme to a potent phosphotransferase (29,30). nPKCs (e.g. PKC⑀, PKC) phosphorylate Ser in the activation loop (19 -21). Because PKDs and PKCs have (a) different substrate specificities and possibly (b) distinct intracellular destinations (after activation), this mode of signal transduction might enable diversification, specialization, and increased dissemination of DAG signals within cells.
DKF-1 is not activated by PKCs. Thr 588 in the DKF-1 activation loop is critical for an 8 -20-fold enhancement in kinase activity. However, treatment of cells with PKC inhibitors (at concentrations that fully suppress PKD activation by hormones and DAG and/or PMA) has no effect on PMA-stimulated activation of DKF-1. We conclude that DKF-1 is an example (perhaps the first) of a PKC-independent, DAG-activated protein kinase. Thus, DKF-1 may exert control over a subset of target effectors that receive no input from PKC-mediated signaling pathways. DKF-1-regulated signal transduction could also antagonize or synergize with signaling pathways governed by activated PKCs when cells containing both kinases initiate DAG production. Further investigation is needed to determine whether Thr 588 is a target for auto-and/or transphosphorylation. In either case, the DKF-1 enzyme and viable DKF-1 null animals are key components of a potentially important model system that may simplify future analysis of PKC-independent, PKD-mediated cell regulation. This minimally explored topic is especially relevant in view of reports that document bone morphogenetic protein-induced, PKD-mediated activation of stress kinases (84), H 2 O 2 -promoted, PKDregulated stimulation of the ASK1/JNK pathway (36), PKD2-controlled activation of NFB activity downstream from Bcr-Abl (34) and PKD3mediated basal glucose transport in myotubes (85). These processes proceed in the absence of PKC activation and without phosphorylation of Ser residues in the PKD activation loop.
Mammalian PKDs influence a diverse group of cell functions and pathologies, including extracellular signal-regulated kinase (ERK) and JNK signaling, induction of NFB by stress, transport vesicle budding from the trans-Golgi network, vesicular trafficking, integrin recruitment at focal adhesions, chromatin organization, apoptosis, cardiac myocyte hypertrophy, and cancer cell invasiveness (19 -21, 86, 87). Caveats and cautions apply when PKD physiological relevance is predicted principally from analysis of transfected, cultured cells. Inputs from PKD modulators (e.g. protein phosphatases, binding partners, DAG kinase, etc.), opposing or synergistic signaling pathways, redundancy introduced by other PKD isoforms, etc. can be compromised when supraphysiological levels of WT or mutant PKDs accumulate in transfected cells. Consequently, physiological effects elicited by PKD transgenes may reflect major or minor, unique or redundant, properly targeted or mislocalized signal transduction pathways. Moreover, the FIGURE 9. DKF-1 overexpression causes a dramatic reduction in C. elegans body size. Synchronized L4 worms from the indicated lines of C. elegans were grown for 2 days, thereby becoming mature adults of maximum size. Worm images were taken by phase contrast microscopy at ϫ40 magnification. A, images of a WT nematode (a); dkf-1(Ϫ/Ϫ) C. elegans that expresses a DKF-1⅐GFP transgene at a supraphysiological level (b); dkf-1(Ϫ/Ϫ) worm that express a high level of a distinct DKF isoform (DKF-2) (H. Feng and C. S. Rubin, unpublished observations) by transcribing a DKF-2-GFP transgene (c); WT C. elegans that expresses a high level of DKF-1⅐GFP (d); and a WT worm overexpressing a DKF-2-GFP transgene (e). Relative body length was measured at L4 and adult stages by ImageJ software; results are presented in B.
candidate PKD-controlled physiological processes listed above are also regulated by established (apparently) PKD-independent mechanisms as well. Thus, the observations summarized above provide invaluable guidance for further analysis of PKD physiological relevance, but definitive conclusions must be deferred until D kinases are linked to specific physiological functions in intact tissues and organisms. At present there are no published reports of experimental manipulation of PKD gene structure/expression in vertebrates in vivo.
We discovered that DKF-1 null C. elegans has a movement disorder (uncoordinated or unc) characterized by paralysis in the tail region. The phenotype is consistent with temporal and cellular patterns of DKF-1 expression and is reversed by restoration of normal DKF-1 content in animals carrying an integrated dkf-1 transgene. Thus, an indispensable connection between a PKD family member and a tissue function is established in vivo. The precise molecular basis for the unc phenotype is not yet known. A speculative possibility is that DKF-1-mediated phosphorylation/regulation controls contractile protein or ion channel functions in posterior body muscle (or transcription factors that regulate expression of these proteins), thereby accounting for paralysis. Other scenarios consistent with paralysis include (a) presynaptic dysfunction (e.g. lack of neurotransmitter release) in neurons or (b) post-synaptic aberrations in neurotransmitter receptors or downstream signaling proteins in muscle as consequences of the lack of DKF-1-catalyzed phosphorylation at neuromuscular junctions.
By exploiting authentic enhancer/promoter DNA and transgenesis, WT DKF-1 protein was modestly overexpressed in the appropriate cells and with normal kinetics during the nematode life cycle. A second stable phenotype was created; constitutively elevated DKF-1 blocked post-embryonic growth, and adult animals were only 50 -60% as large as WT controls. The molecular mechanisms underlying this remarkable size reduction remain to be determined. However, the results suggest a speculative connection between DKF-1 and a well characterized signaling pathway in C. elegans. Properly timed operation of the TGF␤/SMAD pathway is essential for the development of C. elegans of normal size (46,47). DKF-1 may inhibit or switch off TGF␤/SMAD signaling when growth is either temporarily halted during development or completed in adults. In principle, increased DKF-1 expression in the transgenic animals could result in premature and permanent phosphorylation/inactivation of one or more essential SMAD pathway signaling proteins, thereby terminating C. elegans growth at a late larval stage.
Generation of stable unc and small worm phenotypes enables genetic screens to identify in vivo effectors for a PKD family member. Simultaneously, a focused, functionally relevant list of proteins (e.g. proteins involved in secretion of the TGF␤ ligand analog, two receptors, a group of SMAD transcription, targeting and inhibitory proteins, etc.) can be evaluated as candidate substrates for DKF-1-catalyzed phosphorylation. Contemporaneous use of mutagenesis, screening, and a candidate target gene/protein approach for DKF-1 (and the potential for rescue of the DKF-1 null phenotype by expression of mammalian PKDs) should increase the rate of discovery and expansion of our knowledge (including determinations of biochemical, molecular, targeting, and regulatory properties) of authentic effectors for the PKD family.