PKCu Is a Novel, Atypical Member of the Protein Kinase C Family”

We have isolated the full-length cDNA of a novel hu- man serinehhreonine protein kinase gene. The deduced protein sequence shows strong homology to conserved domains of members of the protein kinase C (PKC) subfamily. Homologies reside in the duplex zinc-finger-like cysteine-rich motif and in the protein kinase domain. The lack of the Cz domain of the Ca2+-dependent PKCs and the presence of a unique NH,-terminal sequence with a potential signal peptide and a transmembrane domain suggest that PKCp is a novel member of the subgroup of atypical PKCs. An open reading frame coding for 912 amino acids directs an in vitro translation product with an apparent lUr of 116,000. in vitro phorbol ester binding studies and kinase assays with lysates of cells overexpressing PKCp showed phorbol ester-inde-pendent kinase activity, autophosphorylation, and, in normal rat kidney (NRK) cells, predominant phosphorylation of a 30-kDa protein at serine residues. Southern analysis revealed that PKCp is a single copy gene located on human chromosome 21. There is constitutive low level expression of the human PKCp gene in normal tissues with a single transcript of 3.8 kilobases and el-evated expression levels in selected tumor cell lines. These data suggest a role of PKCp in signal transduction pathways related to growth control.

Protein phosphorylation, a fundamental process for the regulation of cell growth and diverse cellular functions, is catalyzed by a multitude of protein kinases (1)(2)(3)(4). A predominant role of protein kinases is in receptor-mediated signal transduction, where extracellular signals are amplified and propagated by a cascade of protein phosphorylation a n d o r dephosphorylation events that ultimately control the transcriptional activity of genes (5). Within the intracellular activation cascade, the Ca2+l phospholipid-dependent serinelthreonine kinases known as protein kinase C (PKC)' play an important role (6)(7)(8)(9). They are typically activated by the second messenger diacylglycerol and participate in cellular responses to various agonists like hormones, neurotransmitters, and growth factors (6,10). Molecular cloning of various PKC isoforms has established that PKC * This work was supported by German Ministry for Research and Technology (BMFT) Grant 017U 8601/31 and by Deutsche Forschungsgemeinschaft Grant FT.24713-1. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ~a d u e~i s e~n t~ in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
is a multigene family (11). To date, 10 members have been identified, which can be grouped in two major classes according to their dependence on Ca2+ ions. The first group, the conventional PKCs (CPKCLU, f31, f32, and 71, require Ca2+ to be activated in the presence of phosphatidylserine (6), whereas the second group, the novel PKCs (nPKCs), are Ca2+-independent (12-14). The molecular basis for this functional heterogeneity is located in the conserved C2 region, which is absent in the nPKC subfamily (12-15). All PKC isozymes share a conserved catalytic kinase domain in the COOH-terminal region and an NHz-terminal regulatory site (C1) (6,7). Common features of the C1 domain of cPKCs and nPKCs are a conserved pseudosubstrate site and two adjacent amino-terminal cysteine cluster that are responsible for phorbol ester binding (16). More recently, two novel members of the PKC family have been identified that do not fit into the above classification and might represent members of an independent subgroup. They have been termed atypical PKCs (*KC{ and A), because they lack the Cz region and contain only one cysteine-rich motif in the C1 region (6,171. aPKCs are dependent on phosphatidylserine but are not affected by diacyl~lycerol, phorbol ester, or Ca2+ (6). According to their different conditions of activation, each PKC subtype is likely to possess distinct functions in intracellular signaling pathways (7).
In search of new protein kinases participating in growth control and differentiation, we employed DNA probes, specific for the conserved catalytic kinase domain of serinelthreonine protein kinases, to screen cDNAs derived from various human tissues and established cell lines. Here we describe the cloning of a novel serine-specific protein kinase gene, which contains a putative amino-terminal transmembrane region but also shows significant homologies to the catalytic and regulatory domains of the cytoplasmic kinases of the PKC multigene family (10-15, 17).

EXPERIMENTAL PROCEDURES
Cloning and Characterization of cDNA Clones-A 145-bp PCR fragment (see "PCR Analysis") was used to screen a cDNA library derived from the human natural killer cell line YT, and cloned into the expression vector pCDM8. Colony hybridization was performed by transferring the bacteria to a nylon membrane (Schleicher & Schuell) and baking them on a prewet (5 x SSC, 5% SDS) Whatman paper in a microwave oven until filters were dry. Hybridization analysis with the 32P-labeled PCR fragment was performed according to standard p m edures (18). A positive cDNA clone (pkytl) was identified that contained a 150-bp 5'-and a 200-bp 3"extension of the above PCR fragment followed by a 2.9-kb noncoding sequence. A 23-bp 32P-labeled oligonucleotide, derived from the 5'-region of this clone, was used to screen an oii-g~dT)-primed human placenta cDNA library cloned into h7AP (Stratagene). Hybridization and stringent washing was performed as described (19). One positive clone, pBpl4, was excised and characterized by restriction analysis. The complete nucleotide sequence of the pBpl4 cDNA was determined by the dideoxy chain termination method (20) using the modified T7 DNA polymerase (Pharmacia LKB Biotechnology Inc.) with oligonucleotide primers designed according to the sequence obtained. Both strands were sequenced. The 5"deletion mutant of the PKCp cDNA was constructed by digesting pBpI4 with ApaI, removing the 334-bp ApaI fragment by agarose gel electrophoresis, and religating the larger pBpl4 fragment after elution from the agarose gel. Sequence analyses and data base searches were done with the HUSAR program (21) at the German Cancer Research Center, Heidelberg, Germany. PCR Analysis-128-fold and 192-fold degenerate oligonucleotides derived from the conserved motifs HTDLKPEN and GTPAYLAF'E of S e d Thr protein kinases were used to amplify an 145-bp fragment from a cDNA library prepared from the lymphoid cell line YT. PCR was performed a t 94 "C for 80 s, 55 "C for 150 s, and 72 "C for 180 s for 35 cycles. After subcloning and nucleotide sequence analyses, this fragment was used for screening to obtain a complete cDNA.
To determine the steady state level of PKCp mRNA, a quantitative PCR analysis from transcribed cDNAs of several cell lines was performed with 15-mer oligonucleotides spanning the region between nucleotide 2525 and nucleotide 3032 of the PKCp cDNA, resulting in a 507-bp fragment. PCR conditions were 94 "C for 90 s, 42 "C for 120 s, and 72 "C for 120 s for 38 cycles. 2 pl of the first strand cDNA reaction mixture were used as a template in a 100-pl reaction volume. PCR amplification of 5"specific sequences was performed under identical conditions using 20-mer oligonucleotides resulting in an amplification of a PCR fragment spanning nucleotides 13S524 of the cDNA. A314-bp fragment from the human GAPDH gene (22) was used as a control of PCR reactions and amplified in parallel to verify equal amounts of cDNA in each PCR reaction. PCR conditions were the same as for the PKCp cDNA.
I n Vitro Danscription a n d Danslation-Purified double-stranded DNA from the pBpl4 cDNA in pBluescript was linearized at the 3'-end with BarnHI for using the T7 promoter or at the 5'-end with XhoI for using the T3 promoter. Transcription and translation were performed according to the instructions from the manufacturers (Stratagene and Amersham Corp.). [35SlMethionine-labeled proteins were subjected to reducing SDS-PAGE (lo%), blotted onto a nitrocellulose membrane, and autoradiographed overnight.
Northern (RNA) Blot Analysis and cDNA Synthesis-Poly(A)+ RNAs from various cells were isolated using a Micro-RNA purification kit (Pharmacia). Northern blot analysis was performed using a [a32PldCTP-labeled PstI PKCp cDNA fragment spanning the 3"coding and noncoding region of the cDNA, according to standard procedures (18). RNA from several primary human tissues was analyzed with a commercially available RNA blot (Clontech). Autoradiographs were scanned using an Ultroscan XL scanner (Pharmacia LKB Biotechnology Inc.). cDNA synthesis for PCR analysis was performed using a first strand synthesis kit (Pharmacia) with 10% of the poly(A)+ preparation according to the manufacturer's protocol.
Southern Blot Analysis-Genomic DNAs (20 pg) from different tissues were digested with EcoRI, PstI, or XbaI and separated by agarose gel electrophoresis before blotting on a nylon membrane (Schleicher & Schuell). Hybridization was performed with the same labeled PstI fragment used for Northern analysis according to standard procedures. Final washing was performed in 0.1 x SSC a t 62 "C and 55 "C for the yeast DNA. Exposure times varied between overnight for the human DNA and 3 days for the yeast DNA.
Immunoprecipitation and Western Blot Analysis-A glutathione fusion protein was constructed by ligating a 770-bp SmaIIHindIII blunt fragment coding for amino acids 34-290 of PKCp in pGEX3. BALB/c mice were immunized with 80 pg of fusion protein using ABM multiplier (Linaris) and boosted twice to generate an immune serum. 2 pl of serum were used for immunoprecipitation (23) of PKCp from lysates of transfected cells. Immunoprecipitates were subjected to SDS-PAGE under reducing conditions, and Western blot analysis was performed with immune and preimmune serum a t a 1500 dilution and an alkaline phosphatase detection system (Dianova).
Cell Lines and Somatic Cell Hybrids-The cell lines A549, K562, HL60, Jurkat, HeLa, COS1, and L929 were derived from ATCC, Rockville, MD. The Kym I cell line is a human rhabdomyosarcoma cell line kindly provided by M. Sekiguchi (24). YT is a lymphoblastoid NK-like cell line kindly provided by M. Kronke, Munich, Germany. The human glioblastoma cell line 308 is a kind gift of B. Seliger, Mainz, Germany. Normal rat kidney (NRK) cells were obtained from T. Tamura, Giessen, Germany. The origin and characterization of the human mouse somatic cell hybrids, carrying selected human chromosomes (NI35, chromosomes 6, 7, 14, and 21; NI37, chromosomes 7, 14, and 21; NI48.26, chromosomes 7 and 14) has been described previously (25,26).
Generation of Stable NRK, H e k , a n d COS Dansfectants-The coding region of the PKCp cDNA was excised from the pBpl4 plasmid as a 3.2-kb EcoRYNsiI fragment. Overhanging 5'-ends were filled with the menow enzyme, and the DNA fragment was cloned into the mamma-lian expression vector pMAMneo (Invitrogen) or pBMGneo (27). 2 x lo6 cells were transfected with 5 pg of this plasmid using TransfectamTM reagent (Serva) according to the manufacturer's instructions. Cells were kept in RPMI 1640 medium supplemented with 5% fetal calf serum and 600 pg/ml G418 to select for transformants. After 3 weeks, clonal colonies were picked and expanded individually.
pMAMneotransfected NRK cells and pBMGneo-transfected HeLa and COS cells were induced by 2 p~ dexamethasone (48 h) and 5 p~ CdClz (12 h), respectively. In parallel, expression of PKCp was analyzed by Northern blotting and RT-PCR.
In Vitro Kinase Reaction and Phosphoamino Acid Analysis-Cells were harvested and washed once with phosphate-buffered saline, and aliquots of 5 x lo5 cells were resuspended in 50 pl of phosphorylation buffer (HEPES, pH 7.4, 120 mM NaCl, 5 mM KC1, 1 mM MgCl,, 5 mM glucose, 1 mM MnCl,, 2 mM sodium orthovanadate). 50 pglml phosphatidylserine was added to each reaction mixture. 50 nglml PMA was added when indicated. Cells were lysed by shearing several times using a 1-ml Luer syringe and a G23 gauge. Phosphorylation of cellular proteins was carried out by adding 2 pl of 100 p~ ATP, 1.5 pl of [y3*P]ATP (20 Ci/mmol) in a volume of 50 pl and incubated 15 min at 37 "C. Reactions were terminated by adding an equal volume of sample buffer, resolved by SDS-PAGE, and transferred to a nitrocellulose membrane. 32P-Labeled proteins were visualized by autoradiography.
To analyze phosphoamino acids, pertinent regions o f the SDS-PAGE were excised and eluted as described (28) and the 32P-labeled protein was hydrolyzed in 6 M HCI for 90 min at 110 "C. Supernatants were lyophilized, mixed with nonradioactive phosphoamino acid standards, and analyzed by one-dimensional electrophoresis on cellulose thin-layer chromatography plates (Merck, Silicagel) with pH 3.5 running buffer (pyridine:acetic acid:H20, 5:50:945).
Phorbol Ester BindingAssay-This was done essentially as described

RESULTS
Cloning and Sequencing of PKCp-cDNA derived from the human cell line YT was initially employed to identify new protein kinase genes. Using two degenerate primers, deduced from the second conserved domain within the catalytic site of Serf Thr protein kinases, a 145-bp fragment was amplified by PCR. Using a oligonucleotide derived from this sequence as a hybridization probe, a pCDM8-based cDNA library of YT cells was screened. One clone designated pkytl contained the initially amplified PCR sequence on a 3.4-kb insert. From the 5"region of pkytl, the oligonucleotide 5'-GCCTGCCCTTTTCACTT-GACA-3' was used as a hybridization probe to screen a AZAPbased cDNA library derived from human placenta tissue to obtain a full-length cDNA clone. Gel hybridization and restriction analysis of the plasmid DNA revealed that it contained a 3.8-kb insert. This cDNA, designated pBpl4, coded for a 236-bp 5'-untranslated region, an open reading frame of 912 amino acids followed by a TGA stop codon and a 787-bp untranslated 3"region. Sequence comparison of clone pBpl4 to the previously identified clone pkytl revealed identities between nucleotide 2194 and nucleotide 2669, representing 158 amino acids of the protein kinase domain from the cDNA.
The coding region of the pkytl cDNA sequence, identical with pBpl4, ends with a GAAG where GAA codes for glutamine, which is identical with Glu-811 in the PKCp cDNA. Downstream of the second G, which is part of the Ala-812 codon in the PKCp gene (see Fig. 11, the sequence of the pkytl gene continues with GTAA and a subsequent sequence that diverges completely from the pBpl4 cDNA. Within the nucleotide se-

L C D F G F A R I I G E K S F R R S V V G T P A Y L A P E V L R N K G Y N R 7 6 2 GGTGAAACTTTGTGATTTTGGTTTTGCCCGGATCATTGGAGAGAAGTCTTTCCGGAGGTCAGTGGTGGGTACCCCCGCTTACCTGGCTCCTGAGGTCCTAAGGAACAAGGGCTACAATCG
* * * * * * * *

N P U K E I S H E A I D L I N N L L P V K M R K R Y S V D K T L S H P U L P D Y~2
* * * * quence GAAG-GTAA, the AG-GT nucleotides represent the consensus region of a splice donor site (30). These data, together with the finding that no open reading frame could be identified within the 3"region of the pkytl cDNA, suggest that it is a partially spliced cDNA indicating an exodintron junction behind the G of codon Ala-812 (Fig. l ) .

A T T C T T G T C A~A A A A A A A A
Analysis of PKCp Structure-The open reading frame of the pBpl4 clone encompasses 2738 nucleotides from nucleotides 236 to 2974 of the cDNA. Upstream of the first in frame ATG, a purine (G) is located at position -3 and no T nucleotides are within 9 nucleotides upstream of the ATG. This is in accordance with the Kozak predictions (31) for an initiation codon. Based on this, the deduced protein comprises 912 amino acids with a calculated molecular mass of 102,000 daltons. The NH,-terminal region of PKCp up to codon 146 does not demonstrate significant nucleotide or amino acid homology to previously cloned kinase genes. Hydropathy analysis revealed a n NH2terminal hydrophobic sequence and a proteolytic cleavage site according to established rules (32) at amino acid Ala-25 ( p = 1.0) or Pro-33 ( p = 0.98). Thus, the mature protein would be comprised of 887 amino acids. A second hydrophobic region of 21 amino acids spans from Pro-35 to Ile-55. This sequence is followed by a transfer stop signal (arginine) at position 59 and might serve as a putative NH,-terminal transmembrane anchor (Fig. 1).
Located downstream from codon 146, PKCp contains two characteristic cysteine-rich repeats that are conserved among all members of the PKC family except PKC& in which only a single domain was identified (17) (Fig. 2 A ) . The cysteine-rich motif His-Xlz-Cys-Xz-Cys-X,o_l4-Cys-X~-Cys-X~-His-Xz-Cys-X,-Cys, which complexes with the heavy metal ions zinc and cadmium (33) and is responsible for phorbol ester binding (12- 14,16,17), is fully conserved in both amino-terminal repeats of PKCp (Fig. 2, A and B ) . Protein data base searches revealed strong overall homology t o the two cysteine domains of various PKC subtypes ranging from 44% (PKCB) to 66% (PKCa) for the CYS I domain and 58% (PKCa, nPKCB) to 68% (nPKCE, PKCy), for the CYS I1 domain. A further analysis of the cysteine cluster of PKCp and of PKC< revealed in each case a divergence in two motifs of the first His-Cys subdomain that are conserved in all other phorbol ester-binding PKCs (Fig. 2B). The motif H-KR-F is replaced by H-A-LIH-T-F (PKCp cluster Ilcluster 2) and H-L-F (PKCS). The motif Q-P-T of cysteine cluster 1 is replaced by A-P-A (PKCp) and by R-R-A (PKCb). The motif SN-P-T of cysteine cluster 2 is replaced by R-P-T (PKCp). Moreover, the two cysteine clusters of PKCp are separated by 79 amino acids, which is in contrast to a 15-and a 22-amino acid spacer in cPKCs and nPKCs, respectively (11)(12)(13)(14)(15) (Fig. 2 A ) . The tandemly repeated cysteine domains are followed by a region of approximately 270 amino acids that shows no significant homology to any known kinase genes. In particular, we found no homology to the second conserved domain (Cz) common to all conventional PKCs, the activation of which is Ca2+-dependent (11).
Comparison of the putative kinase domain of PKCp to those of other kinases (Fig. 2C) revealed the expected identity of those amino acids that are invariant among known kinase family members. Thus, the ATP binding consensus sequence, Gly-X-Gly-Xz-Gly-Xl,-Lys where X represents any amino acid, is conserved in the kinase domain of PKCp. Moreover, the invariant aspartate essential for kinase activity is located at the predicted position within the conserved motif HRD- (3).
In PKCp the X X X motif is KPE, which is found most often in serinefthreonine kinases, and not AAR or W, which are typical of tyrosine kinases (3).
The PKCp cDNA Directs the ~~n t h e s z s of a 116-kDa Protein-To confirm the predicted open reading frame, PKCp transcripts were synthesized in vitro and translated in a cell-free system in the presence of 35S-labeled methionine. The T7 sense transcript produced a polypeptide with an estimated M , of 115,000 as revealed from SDS-PAGE under reducing conditions (Fig. 3, lane 1 Fig. 3, lane 2 ) . These data confirm translation initiation of PKCp cDNA at the first ATG (position 236) in the open reading frame. In addition to the 115-kDa protein, several lower molecular weight polypeptides were found. These products might have been obtained by translation of incomplete transcripts due to premature termination of in vitro transcription. Alternatively, initiation of transcription from internal methionine start sites (34) or posttranslational processing of a larger PKCp species may have occurred. No polypeptides of comparable size were synthesized from the antisense T3 transcript or in the absence of exogenous RNA (Fig. 3, lanes 3 and 41, indicating that the synthesis of the identified protein is directed by the PKCp cDNA. Analysis of PKCp Gene Expression-Northern blot analysis of human poly(A)* RNA from various cell lines and human tissue using the 0.87-kb PstI PKCp cDNA fragment as a probe identifies a single transcript in two cell lines, Kym I and A549, as well as in several human tissues (Fig. 4, A and B ) . A transcript size of 3.8 kb indicates that the cloned cDNA is fulllength (Fig. 4A ). Moreover, constitutive expression of PKCp in all primary tissues examined indicates a broad dist~bution, although the mRNA level differed considerably between the various tissues. The highest steady state levels were detected in kidney, heart, and lung (Fig. 4A). Nevertheless, when compared to other genes, mRNA levels of PKCp in normal tissues appeared very low. Typically, a signal could only be detected after a longer (4-day) exposure of blots (Fig. 4, A and B), whereas GAPDH probes of similar specific radioactivity revealed signals after a 1-h exposure (data not shown). In contrast, Northern analysis of several established human tumor cell lines showed strong expression of PKCp in a lung carcinoma (A5491 and a rhabdomyosarcoma (Kym I) cell line. For comparison with the respective normal tissue (Fig. a), a 4-day exposure of Northern blots from the two cell lines is also shown in Fig. 48. Densitometric analysis of shorter exposures indicates 32fold and 4foid higher mRNA levels in the cell lines A549 and Kym I, respectively.
Because PKCp-specific mRNA was readily detectable by Northern blot analysis in only two out of seven cell lines, RT-PCR was employed to increase sensitivity of detection. PKCpspecific primer for amplification of a 508-bp DNA fragment were used. As a positive control of equal cDNA amounts, a GAPDH-specific fragment was amplified in parallel (Fig. 4C). As expected from Northern blot analysis of Kym I cells, high . In all these cells, PKCp-specific mRNA was not discerned by Northern analysis. No transcripts were detected by PCR in the myelomonocytic cell line HL60 (lane 5). In addition, 5"region specific primer for amplification of a 386-bp fragment coding for the putative leader and transmembrane sequence of PKCp were employed to compare transcript structure of Kym I cells with the cloned PKCp cDNA. A fragment of the expected size could be amplified from both cDNAs (Fig. 4 0 , lanes 1 and 21, which gives rise to identical cleavage products upon PuuII digestion (Fig. 4 0 , lanes 3 and 4 ) .
The PKCp Gene Is Located on Chromosome 21 and Is Highly Conserved between Species-Southern analysis of human DNA with the 0.87-kb PstI fragment of pBpl4 revealed one specific band, indicating that PKCp is likely to be a single copy gene (Fig. 5A). Using the whole cDNA fragment as a hybridization probe, five to seven hybridization signals were obtained, depending on the restriction enzymes used for digestion of genomic DNA (data not shown). This suggests a genomic structure of PKCp consisting of five to seven exons.
A chromosomal assignment of the PKCp gene was achieved by Southern blot analysis of human-mouse somatic cell hybrids differing in the content of human chromosome 21. The hybrids N135 and NI37, both containing human chromosome 21 (251, show the human PKCp-specific hybridization pattern, in addition to cross-hybridization with the homologous mouse gene, when probed with human PKCp-specific cDNA (Fig.  5A). In contrast, with the hybrid N148.26, which differs from N135 and N137 in the presence of chromosomes 6, 21, and 21, respectively, no human PKCp-specific signals were obtained (Fig. 5 A ) . Accordingly, an assignment of PKCp to chromosome 21 can be made on the assumption that no translocations of small human genomic fragments to other chromosomes had occurred that would have gone unnoticed in initial caryotype analysis (25).
Based on the observation of the strong cross-hybridization with mouse DNA, 20 pg of genomic DNA from yeast was analyzed by Southern blotting with the same 0.87-kb PstI fragment used as a probe for human DNAs. As for mouse DNA, one specific signal was obtained in yeast, although under lower stringency hybridization conditions (Fig. 5B 1 that PKCp or a closely related gene remained highly conserved during evolution. Protein Kinase Activity of PKCp and Phorbol Ester Binding-In order to verify kinase activity of the cloned cDNA gene product and to identify potential cellular substrates, the coding region of PKCp was cloned in the expression vectors . The human and mouse hlots reprrsrnt overnight exposures: the yrast hlot was exposed for 3 days. pMAMneo for expression in NRK cells and in pBMGneo for expression in HeLa and COS cells. G41R-resistant transfectants were assayed for PKCp transcription by Northern analysis (Fig. 6 R ) and/or by RT-PCR amplification. Constitutive in vitro kinase activity from total cellular extracts of transfectants was analyzed by SDS-PAGE and the resulting phosphoprotein pattern compared to vector-transfected cells. The data obtained with both COS and HeLa cells show a different phosphoprotein pattern with a novel band at 115 kDa in the PKCp transfectants ( Fig. 6 A , lanes I and 3 versus lanes 2 and 4 ) .
suggesting autophosphorylation activity of PKCp. The presence of PKCp protein in overexpressing COS transfectants was verified by immunoprecipitation analysis, revealing a single band at approximately 115 kDa (Fig. 6C). This protein band was absent in vector control transfectants. The differential intensity of the 115-kDa phosphoprotein band in COS ( Fig. 6 A , lane I ) and HeLa (Fig. 6 A , lane 3 ) transfectants is in accordance with a differential expression of PKCp at the mRNA level (Fig. 6R, lanes I and 3 ) . In addition to autophosphorylation, increased phosphorylation was observed in several other endogenous proteins in all PKCp transfectants. In NRK transfectants, autophosphorylation of PKCp is less evident because of constitutive phosphorylation of a n endogenous protein migrating at the same apparent molecular weight, resulting in only a 2-fold increase in the labeling intensity as measured by densitometric scanning. However, a prominent 30-kDa phosphoprotein was detected in PKCp-transfected cells (Fig. 7 A ) . A phosphoamino acid analysis of the excised 30-kDa protein of NRK PKCp transfectants revealed exclusive phosphorylation at serine residues (Fig. 7 B ) . As shown in lysates from HeLa or COS transfectants is shown in Fig. R. On  average ( n = 4), a n 1.2-fold increase WRS noted in these experiments for both cell types (HeLa, 1.13 T 0.30; COS, 1.27 2 0.23). Similar data were obtained with 1"HlPDRu binding to intact cells ( n = 2, data not shown). Together with the in urfro kinase assays, these data indicate that under the experimental conditions employed here PKCp does not bind effcirntly and is not activated by phorbol esters. and PKCp cDNA-transfected NRK cells (lane 1 ) were phosphorylated, subjected to SDS-PAGE, transferred to a nitrocellulose membrane, and exposed overnight. B, phosphoamino acid analysis. The phosphorylated SO-kDa protein shown in A was excised, eluted, acid-hydrolyzed, and The position of the amino acid standards is indicated. C, phorbol ester-separated in one-dimensional thin-layer electrophoresis as described.    (10). The more recently identified novel PKCs PKCG, E, q, and 0 (7, 12-15) lack Ca2+ binding sites, and their activation is Ca2+-independent. PKCi and h differ from these two major PKC subgroups by lack of phorbol ester binding/activation capacity. They represent members of a newly defined subgroup of PKCs, the atypical PKCs (6). Based on the homologies in kinase and regulatory domains, the kinase described here, designated PKCp, is a novel member of this subgroup of atypical PKCs. Similar to the nPKCs and aPKCs, PKCp lacks the Cz region of Ca2+-dependent PKCs (a, p, y ) (Fig. 2 A ) , suggesting that the kinase activity of PKCp is also Ca2+-independent. Of note is the observation that two zinc finger-like domains are highly homologous (4448%) to those of other members of the PKC family with a complete identity in the positions of the histidine and cysteine residues. However, whole cells and total cell extract from stable transfectants overexpressing the PKCp cDNA showed, on average, only a 1.2-fold increase in phorbol ester binding (Fig. 8). In contrast, it has been shown previously that transient expression of phorbol ester responsive PKCs typically gives rise to a 3-10-fold increase in phorbol ester binding (29). Irrespective of a potential function of PKCp as a phorbol ester receptor, analyses of PKCp kinase activity clearly suggest phorbol ester independence. Aconstitutive autophosphorylation of PKCp was noted for all transfectants (Figs. 6 and 7); autophosphorylation was not increased in the presence of PMA.2 Moreover, in lysates of NRK transfectants, phosphorylation of the major substrate, a 30-kDa protein, was also not enhanced by PMA stimulation (Fig. 7C). We therefore conclude that PKCp is a phorbol ester-independent kinase, although within the phorbol ester binding domain, the overall degree of homology between PKCp and other PKCs is not different from that within the group of phorbol ester-responsive PKCs. However, it was noted that PKCi and p differ a t two positions in the first histidinekysteine subdomain from all other PKC subtypes (Fig. 2B ). It remains to be determined whether or not these non-conservative substitutions indeed negatively affect the phorbol ester binding capacity of the PKC isoenzymes p and 5. Inefficient phorbol ester binding of PKCp might also be related to the unusual spacing of the tandem domains, which are 15-22 amino acids apart in phorbol ester binding PKCs, whereas in PKCp, the two domains are separated by 79 amino acids. Therefore, in addition to differences in the primary structure, it is conceivable that an appropriate spacing of both domains is required for efficient, high affinity phorbol ester binding. This reasoning is supported by the previous demonstrations that a single cysteine domain retains PMAbinding, however, its binding affinity is 10-20-fold lower as compared to the native enzyme (16). Moreover, high affinity binding could not be reconstituted by coexpression of two complementary mutants, suggesting interactive cooperation of both domains in cis (16). Our data obtained here suggest that two cysteine domains not only have to be in cis, but also in close proximity to allow interaction and to create a high affinity phorbol ester binding site. Binding studies with the purified PKCp will allow a more precise definition of the phorbol ester binding capacity and its relevance for its regulation of kinase activity.
Aside from apparent lack of PMA responsiveness, additional arguments in favor of a novel function of PKCp are obtained from an analysis of the deduced NH2-terminal structure of the gene product. A putative signal peptide and transmembrane domain suggest a potentially exclusive location of the mature PKCp protein at intracellular or cell surface membranes. This * F. J. Johannes and P. Oberhagemann, unpublished data. would be in stark contrast to the phorbol ester-dependent PKCs, which, upon activation, only transiently translocate to the cell surface or nuclear membrane (7). Accordingly, the potentially permanent membrane location, lack of the conventional regulatory domains of PKCs, and instead, two additional unique regions 5' of the cysteine cluster and 5' of the kinase domain, suggest a differential regulation of PKCp function and a distinct biological role of this PKC subtype. As no typical targeting motifs for intracellular organelles have been detected, it is conceivable that PKCp is directly associated with and activated by other cell membrane proteins and serves for example as a signal transducer for non-kinase receptors, such as cytokine receptors (35,36). In this context, it is interesting that the highly conserved PKCp gene appears to be located on human chromosome 21. Chromosome 21 has been shown earlier to carry several genes controlling cellular responsiveness to interferon a and y (26, 37, 38). At present, it is unknown whether PKCp has a functional relationship to these elusive interferon response controlling gene(s) or to any other cytokine receptor system.
A first indication of the biological role of PKCp comes from expression studies. When compared to the respective normal primary tissues, which show differential but always rather low expression of PKCp (Fig. 4A 1, a few tumor cell lines were identified that exert a significant overexpression of PKCp. In particular, a 32-fold increase in steady state mRNA levels was noted in the lung carcinoma cell line A549 (Fig. 4B). Overexpression of PKCp affects the phosphorylation pattern of endogenous proteins in transfected cells (Figs. 6 and 7). This may eventually lead to phenotypically apparent changes in cellular functions, morphology, or proliferative capacity. With the availability of stable transfectants and antibodies, it will be now possible to identify specific substrates and to analyze the functional role of PKCp in cell growth and differentiation.
A c k n o w~e d~n~s -~e thank Hans-Peter Geithe (Max Planck Institute for Biophysical Chemistry, Gottingen, Germany) for synthesizing the oligonucleotides used for sequencing and Gisela Link for expert technical assistance.
Note Added in Proof-We have now obtained data indicating that enhanced kinase activity of PKCp can be revealed upon phosphatidylserine/phorbol ester treatment in vitro when specific immunoprecipitates of PKCp rather than whole cell extracts are analyzed. This finding suggests that phorbol ester responsiveness of PKCp can be masked by other factors present in whole cell lysates.