KEPI, a PKC-dependent Protein Phosphatase 1 Inhibitor Regulated by Morphine*

cDNAs encoding KEPI, a novel protein kinase C (PKC)-potentiated inhibitory protein for type 1 Ser/Thr protein phosphatase (PP1), were identified. They were found among morphine-regulated brain mRNAs identified using subtracted differential display techniques. Full-length rat, mouse, and human cDNA and genomic sequences were elucidated with library screening and data base searching. Rat, mouse, and human KEPI cDNAs encode 164–165 amino acid proteins with calculated isoelectric points of 5.2. Each species' amino acid sequence contains consensus sequences for phosphorylation by PKC (KVT72VK), protein kinase A (RKLS154), and casein kinase II (S43SRE, S120EEE). Multiple KEPI N-terminal myristoylation consensus sites provide potential regions for membrane anchoring. Subcellular fractionation and Western analyses revealed that most KEPI immunoreactivity was associated with P2 and P3 membrane-enriched fractions and little in cytosolic fractions. 2.6-kb KEPI mRNAs were detected in brain, especially in the cerebral cortex and hippocampus, and in heart and skeletal muscle. Brain KEPI mRNA was up-regulated by both acute and chronic morphine treatments. The human KEPI gene contains four exons extending over more than 100 kb of genomic sequence on 6q24-q25, near the μ opiate receptor gene. These sequences displayed sufficient homology with the porcine PP1 inhibitor CPI-17 that we asked whether KEPI could share the ability of CPI-17 to modulate PP1 activity in a PKC-dependent fashion. Recombinant mouse KEPI is phosphorylated by PKC with aKm of 2.6 μm and at 1/2 of 20 min. Phospho-KEPI inhibits PP1α with an IC50 of 2.7 nm, a potency more than 600-fold greater than that displayed by unphosphorylated KEPI. Neither phospho- nor dephospho-KEPI inhibits protein phosphatase 2A. Up-regulation of KEPI expression by morphine, an agonist at PKC-regulating G-protein-coupled μ receptors, provides a novel signaling paradigm in which the half-lives of serine/threonine phosphorylation events can be influenced by activities at Gi/Go-coupled receptors that modulate KEPI expression, KEPI phosphorylation, and KEPI regulation of PP1 activity.

Morphine activation of G i /G o -coupled receptors alters second messengers, ion fluxes, and gene expression patterns in ways that could contribute to long term consequences of receptor occupancy such as tolerance, dependence, and addiction (8 -10). We have developed and used modifications of subtractive hybridization and differential display (SDD)-PCR approaches to seek morphine-regulated genes. These approaches thus provide a number of short cDNAs that correspond to morphine-regulated mRNAs.
To define interesting novel genes that might correspond to these apparently morphine-regulated cDNAs, we have screened mouse and rat brain cDNA libraries with probes derived from pools of short cDNAs that were morphine-regulated in initial SDD experiments. One of these cDNAs hybridized to a novel 2.6-kb mRNA species that was expressed specifically in brain, heart, and muscle and was up-regulated by acute and chronic morphine treatments in brain. The sequence of this cDNA displayed homologies with the previously elucidated CPI-17 and PHI/PNG genes (11,12), which are PKC-potentiated inhibitors of the major serine/threonine type 1 protein phosphatase, PP1. We now report elucidation of this novel gene, termed "KEPI" (kinase-enhanced PP1 inhibitor), describe its ability to serve as a PKC substrate, and document its phosphorylation-dependent ability to inhibit PP1 selectively. Definition of this protein and related sequences provides an expanded view of an emerging family of potential PKC-dependent PP1 inhibitors. Elucidating KEPI regulation by morphine also provides a novel mechanism for longer term effects of morphine on a variety of cellular signaling pathways, including those that might contribute to tolerance, dependence, and addiction. mice were implanted subcutaneously with pellets containing 75 or 25 mg of morphine, respectively, or matched placebo controls (National Institute on Drug Abuse, Division of Basic Research). Animals were sacrificed 4 days after implantation, and brain regions (rat) or whole brains (mice) were dissected. RNA was isolated from frozen brain tissues using RNAzol B (Tel-Test, Friendswood, TX), and SDD was carried out as described (8,14). Radiolabeled PCR fragments displaying substantial intensity differences between morphine and saline treatments were excised, eluted, and subcloned into the PCRII vector (Invitrogen). Northern blot (8) and RNA dot-blot analyses (15) were performed as described. KEPI hybridization was analyzed by ImageQuaNT TM (Storm 860, Amersham Biosciences) and normalized to ␤-actin expression.
Screening and Sequencing-Plasmids containing SDD fragments were purified with Qiagen-tips (Qiagen, Valencia, CA). cDNA inserts were excised with appropriate enzymes, purified from agarose gels, and labeled with [␣-32 P]dCTP using a multiprimer labeling kit (Amersham Biosciences). Approximately 1.2 ϫ 10 6 plaques from a neonatal mouse brain cDNA library, cloned in the EcoRI/XhoI sites of the Uni-ZAP XR vector (Stratagene, La Jolla, CA), and a similar number of plaques of a rat cerebral cortex cDNA library, cloned into EcoRI site of -ZAP II vector (Stratagene), were screened with pools of 5-20 radiolabeled rat SDD probes. Positive plaques were identified, plasmids excised, and clones purified as described (16). Longer cDNA clones were subcloned further as smaller EcoRI restriction fragments, and both strands were subjected to dideoxynucleotide dye termination sequencing using 373A automatic DNA sequencing and primer walking (PE Applied Biosystems, Foster City, CA). Sequencher (Version 3.0, Gene Code Co.)-assembled sequence data and GCG and NCBI tools provided comparisons. We focused on a 2.2-kb murine cDNA insert that we termed initially B16.
Production and Purification of a Fusion Protein Including the B16 Open Reading Frame (His-KEPI)-A translation initiation site adaptor with sequence GCGCGGGGcCATGgCGGTGGTG, which provided a new NcoI restriction site (underlined), was used to amplify a 2-kb segment of the B16 cDNA clone including its entire open reading frame. PCR products were digested with NcoI at the translation initiation site and HindIII at the 3Ј-UTR end of the sequence and cloned into NcoI/ HindIII sites of the bacterial expression vector pET30a (Novagen, Madison, WI) to form pET30aB16. The initiation methionine of pET30aB16 was positioned so that the B16 open reading frame followed the His tag peptide in-frame. BL21(DE3) Escherichia coli cells were transformed with pET30aB16, grown at 37°C in LB/kanamycin for 2-3 h, induced using 1 mM isopropyl-1-thio-␤-D-galactopyranoside, grown at 37°C for 3 h, lysed by sonication, boiled for 5 min, and centrifuged at 6,000 ϫ g for 10 min. Fusion protein was purified from the supernatant by metal chelation chromatography as described (HisTrap kit, Amersham Biosciences), quantitated using Bradford assays (Bio-Rad), and its purity was confirmed by SDS-PAGE (15% gel).
PP1 and PP2A Assays-A protein phosphatase assay system (Invitrogen) provided differential measurements of PP1 and PP2A activities in dephosphorylating the glycolytic enzyme phosphorylase (17). Phosphorylase A was labeled with [␥ 32 P]ATP according to the manufacturer's protocol, purified by PD-10 Sephadex column chromatography (Amersham Biosciences), and [ 32 P]phosphorylase A protein concentrations were measured using Bradford assays. Dephosphorylation reactions were carried out with 0.3 unit of rabbit muscle PP1␣ or 5 ng of bovine kidney PP2A 1 (Calbiochem) in 60 l containing 6 M [ 32 P]phosphorylase A and various concentrations of phospho-HisB16-KEPI or unphosphorylated HisB16-KEPI as inhibitors. Maximal phosphatase activities were assessed in reactions using no inhibitors. Nonspecific blank values were determined in reactions with no phosphatases. Samples were incubated at 30°C for 20 min, proteins were precipitated with 160 l of 25% trichloroacetic acid and centrifugation at 12,000 ϫ g for 5 min at 4°C, and released free 32 P in 200-l supernatant samples was quantitated by scintillation counting. Triplicate data were analyzed by nonlinear regression using GraphPad.
KEPI Antiserum, Subcellular Fractionations, and Western Analyses-A 15-amino acid peptide with the sequence HQQGKVTVKYDR-KEL, which corresponded to residues 66 -80 of mKEPI protein, was synthesized, conjugated to keyhole limpet hemocyanin, and used to immunize rabbits that produced KEPI-specific antiserum G631 after initial immunization and boosting immunizations (Genemed Synthesis, Inc., San Francisco). Specificity was documented by enzyme-linked immunosorbent assay titers of Ͼ1:10,000 against the immunizing peptide; initial results of immunohistochemical experiments documenting specific, apparently specific, patterns of brain immunoreactivity 2 ; results of Western analyses including effective competition for Western staining by preincubation with the 15-amino acid KEPI peptide; and the absence of immunoreactive KEPI (iKEPI) in tissues that express high levels of other members of the KEPI family.
To assess regional distribution of iKEPI, rat tissues were homogenized in 50 mM Tris-HCl, pH 7.5, 150 mM NaCl containing 1% Nonidet P-40, 0.25% sodium deoxycholate, 10 g/mg bovine serum albumin, 300 g/mg phenylmethysulfonyl fluoride, 20 g/mg aprotinin, 10 g/mg leupeptin, and 100 M Na 3 VO 4 . Supernatants were cleared by microcentrifugation at 10,000 ϫ g for 5 min. Samples were electrophoresed using 10% SDS-PAGE, elecrotransferred to nitrocellulose membranes, and protein blots incubated with KEPI peptide antiserum diluted 1:1,000-fold for detection of iKEPI using alkaline phosphatase-conjugated goat anti-rabbit IgG second antibody as described previously (15). In experiments to determine specificity, primary antisera were incubated with 1 g/ml KEPI peptide for 5 min prior to application to the blots.

Identification of Mouse, Rat, and Human B16-KEPI cDNA
Sequences and Initial Polymorphisms-cDNA clones that hybridized to pooled radiolabeled SDD probes were found in both rat and mouse brain cDNA libraries. Positive clones were further selected by hybridization with single short cDNAs. One of the longest inserts was the 2201-nucleotide clone B16. This sequence displayed a consensus sequence for translational initiation (19). No in-frame stop codon was found in the 5Ј-regions of clone B16. Alignment of mouse EST sequences AL605782 and AA671272 provided an additional 390 nucleotides and a 5Ј-UTR to form an extended B16 cDNA. This extended B16 cDNA contains 2546 nucleotides and an in-frame stop codon located at nucleotide position 335 in the 5Ј-UTR of this sequence. 164 amino acids are encoded between the translation initiation codon (ATG) at nucleotide position 518 and the translation termination codon (TGA) at nucleotide position 1010 ( Fig. 1). A poly(A) signal (AATAAA) is located at nucleotide positions 2507-2512, 13 bp upstream of the poly(A) tail (20). An additional poly(A) signal (AATAAC) is located at nucleotide position 1096 -1101, 17 bp 5Ј from the shorter poly(A) site at 1118. Mouse and rat EST sequences display both sites for poly(A) addition, potentially varying the lengths of their 3Ј-UTRs. Rat sequences were derived from a partial rat B16 cDNA whose 5Ј-sequence was augmented by information from rat EST sequences BF567111 and BF560736. The extended rat cDNA sequences encode an open reading frame of 164 amino acids which displays 96% identity with mouse B16 amino acid sequences. The calculated molecular masses of the mouse and rat translation products are 17.75 and 17.98 kDa, respectively ( Fig. 2A). Human EST sequences AW290962, AA627753, AA846873, BE171609, AA788675, AW182882, C02977, AA368804 and N32557, together with genomic exons overlapping the coding region, allow assembly of a human B16 cDNA that encodes 165 amino acids. The additional amino acid in the human sequence, Gln-60, is not found in either rodent sequence. The human B16 protein has a predicted unmodified molecular mass of 17.84 kDa and displays 91% amino acid identity with mouse sequence. Mouse and rat cDNA and EST sequences reveal polymorphisms at a tetranucleotide CCTT repeat at nucleotide positions corresponding to position 2240 in the murine 3Ј-UTR. Mouse EST sequences contain 1, 9, or 10 CCTT repeats at this site, whereas rat EST sequences contain 1, 2, or 3 CCTT repeats. However, human EST sequences available to date each display a single CCTT sequence at the corresponding position.
Amino Acid Sequence Analysis of KEPI-Analyses of the murine B16 amino acid sequence (Genetics Computer Group) reveal a hydrophilic molecule with no signal peptide leader sequence, no clear transmembrane domain, and no clear sites for N-linked glycosylation. The predicted protein is acidic, with an isoelectric point of 5.22 (Genetics Computer Group Isoelectric). The N-terminal residues 20 -24, RVFFQ, contain a basic amino acid followed by two hydrophobic residues, recently described as a consensus PP1 binding motif (21). The N-terminal 50 amino acids could form a secondary structure with 12 predicted turns interrupting a more regularly spaced ␤-sheet structure (Genetics Computer Group PlotStructure). 10 N-terminal region glycine residues, Gly at positions 6, 7, 12, 13, 14, 15, 28, 31, 32, and 35, provide potential N-myristoylation sites (Genetics Computer Group Motif). There is a predicted 45% chance that C-terminal region amino acids 90 -140 form a coiled-coil structure (Genetics Computer Group CoilScan). Potential phosphoacceptor sites for PKC are identified at the Thr-72, for PKA at Ser-154, and for casein kinase II at Ser-43 and Ser-120 (Fig. 1). Each of these potential phosphorylation and acylation motifs is conserved in human, mouse, and rat B16 sequences ( Fig. 2A).
Evolution Analysis of B16 as KEPI, a Member of the PKCpotentiated PP1 Inhibitor Protein Family-Murine B16 sequences from residues 70 to the C terminus are 67% identical to those of PNG/PHI (human PNG, X91195; mouse PNG, NM_008889) and 51% identical to porcine CPI-17 (AB008376). They are also 38% identical to human hypothetical protein FLJ29251 (AK000258) that we have named "GCPI", for gut and cerebral cortex PP1 inhibitor. 3 They display 29% identity with the Drosophila CG17124 gene product (AE003629) (12,(22)(23)(24). The 13 C-terminal amino acids near the Thr-72 potential site for PKC activity are absolutely conserved in each member of this gene family (Fig. 2B) The K/RXTXK/R sequences that constitute typical PKC phosphoacceptor sites are also conserved in each of the family members except the Drosophila CG17124 protein, which displays an alternative PKC recognition motif FLTAK at an analogous position (Fig. 2B).
The gene family that contains KEPI can be tentatively divided into two subfamilies based on protein isoelectric points. Human KEPI, human PNG/PHI, and Drosophila CG17124 gene products are acidic proteins with pI values of 5.22, 5.53, and 5.86, respectively. On the other hand, hCPI-17 and human FLJ29251/GCPI gene products are more basic with pI values of 10.7 and 7.9, respectively. Table I compares these family members. A plausible evolution analysis (Genetics Computer Group EvolutionTree) demonstrates that the PKC-potentiated PP1 inhibitor family is conserved from invertebrate insects (Drosophila) to vertebrate mammalian species separated by more than 500 million years of evolution (25). Because only a single member is currently identified in the Drosophila genome, it is possible that an ancestral gene was duplicated and modified in eukaryotic organisms in response to pressures for greater signaling pathway complexities.
KEPI mRNA Expression and Up-regulation-Northern analyses using a radiolabeled hybridization probe consisting of a 700-bp EcoRI fragment from the mouse B16-KEPI 3Ј-UTR just 3Ј to the shorter polyadenylation site detect a 2.6-kb mRNA expressed in several brain regions, spinal cord, heart, and muscle (Fig. 3A). A 270-bp radiolabeled rat hybridization probe containing sequences 5Ј to the shorter poly(A) signal site recognizes an additional, less intensely expressed 1.2-kb mRNA, consistent with the lower level expression of a shorter mRNA with alternative polyadenylation (data not shown). KEPI mRNA was detected in heart and muscle, but not in liver, kidney, testis, ovary, intestine, stomach, or spleen (some hybridization data not shown). KEPI mRNA is up-regulated in mouse brain after chronic morphine treatment (Fig. 3B). KEPI mRNA is regulated differentially in different rat brain regions after acute and chronic morphine treatments (Fig. 3C). After acute morphine treatments, KEPI is most prominently up-regulated in hippocampus (120 Ϯ 16%, n ϭ 5). After chronic morphine treatment, KEPI is up-regulated most significantly in thalamus (75.6 Ϯ 6.2%, n ϭ 6). In brains of homozygous receptor knockout mice, however, expression is no different from expression in wild type mice (Fig. 3B).
Association of Endogenous KEPI with Membranes-Western analyses of iKEPI revealed bands with apparent mobilities of 45 and 37 kDa on SDS-PAGE bands. Immunoreactivities in both bands were blocked equally by preadsorption of the serum with the immunizing peptide. Immunoreactivities of both apparent mobilities were detected in rat cerebral cortex, midbrain, thalamus, brain stem, cerebellum, and heart. Faint 37 kDa bands were detected in proteins extracted from muscle, lung, liver, kidney, spleen, and testis (Fig. 4). No significant cross-reactivities with the 20 or 23 kDa SDS-PAGE mobilities of CPI and PHI, respectively (11,26), were noted in these brain extracts.
KEPI immunoreactivity was associated with membrane fractions and little with cytosolic fractions. The higher molecular mass 45-kDa immunoreactive species was much more prominent in the P2 crude synaptosomal fraction, whereas lower molecular mass 37-kDa species was found in the P3 crude microsomal fractions (Fig. 4). However, after prolonged storage in SDS sample buffer, much of the iKEPI in extracts from P2 fraction migrated at the 37 kDa position of the lower band (data not shown). Molecular masses observed in Western analyses could be higher than those calculated from KEPI peptide sequence lengths because of deviant SDS-PAGE migration of acidic polypeptides (27) and/or post-translational modifications of KEPI. Differential migration displayed by iKEPI extracted from different subcellular fractions supports substantial roles for post-translational modifications.
Preparation and Purification of mKEPI Fusion Protein-A recombinant murine KEPI fusion protein consisting of KEPI fused to 43 pET30a vector amino acids including its His tag,  thrombin recognition sequences, and enterokinase recognition sequences yielded a 32 kDa band after SDS-PAGE of extracts of isopropyl-1-thio-␤-D-galactopyranoside-induced expressing BL21 E. coli (Fig. 5, left). This was again larger than the 23-kDa unmodified mass predicted from the sequence as often found for acidic proteins. The heat stability of KEPI allowed cell extract boiling and HisTrap metal chelation chromatography, facilitating purification of this fusion protein to homogeneity (Fig. 5, right). Yields of His-KEPI reached as high as 10 mg of protein/liter of bacterial culture.

PKC Phosphorylation of KEPI-Recombinant
His-KEPI protein can be phosphorylated by purified rat brain PKC preparations. Under these conditions, recombinant protein was phosphorylated with apparent K m ϭ 2.6 M, V max ϭ 37.8 nM/mg/ min, and t 1/2 of 20.7 min (Fig. 6A). At plateau time points, a near stoichiometric ratio of phosphorylation, 0.81 Ϯ 0.12 mol of phosphorus/mol of His-KEPI was achieved. When the mobility of 32 P-labeled His-KEPI was analyzed by SDS-PAGE and autoradiography (Fig. 6B), a dominant 32 kDa 32 P-labeled band corresponding to His-KEPI was observed.
Inhibition of PP1 Dephosphorylation Activity by KEPI-Phospho-KEPI, produced by KEPI preincubation with PKC, inhibited the ability of rabbit muscle PP1␣ to dephosphorylate 32 P-labeled phosphorylase. This activity was inhibited by PKCphosphorylated recombinant KEPI fusion protein with an IC 50 of 2.7 Ϯ 0.7 nM (Fig. 7A). Unphosphorylated KEPI could only inhibit rabbit muscle PP1␣ at much higher concentrations, displaying IC 50 values of 1.8 M. KEPI thus increased its in- FIG. 3. Northern analyses of KEPI tissue distribution and morphine regulation. Panel A, Northern analyses of total RNA extracted from mouse tissues and brain regions of mice treated with chronic morphine or placebo pellets. Olf, olfactory bulb; Ctx, cortex; Cer, cerebellum; Brs, brain stem; Mid, midbrain; Tha, thalamus; Hth, hypothalamus; Str, striatum; Hip, hippocampus; Spc, spinal cord; Hrt, heart; Mus, muscle; Liv, liver; Kid, kidney. Panel B, wild type (wt) or receptor knockout (ko) mice. mRNAs hybridizing to a radiolabeled 0.7-kb EcoRI fragment of mKEPI were detected by phosphorimaging. Panel C, quantitation of regulation of KEPI mRNA in rat brain regions after treatments with acute and chronic morphine.
KEPI Genomic Structure-The three exons encoding the human KEPI C-terminal polypeptide sequences are found among GenBank sequences AL096708.34 of BAC RP5-1179L24 and Celera hCG166079, localized to chromosome 6q24.3-25.3. The first exon encoding the KEPI N-terminal peptide sequence is found in GenBank sequences AL138890.9 and AL355497 of the overlapping BACs RP5-932N18 and RP11-472G23, as well as Celera genomic scaffold GAx54KREAVE16.1 which again localize to 6q25.2-26. The genomic contig AL096708 contains each of these BAC sequences, sequence-tagged sites including stSG9583, WI-22584, stSG27036, stSG1514, stSG6339, and stSG9891 and links to ESTs of the UniGene Hs.12599 locus. We assembled sequences of the three overlapping genomic BAC clones covering approximately 330 kb and identified four KEPI exons on more than 100 kb of chromosome 6 sequence (Fig. 8). The sequences of predicted exon/intron junctions are shown in Table II. Intron 1 is 71 kb, intron 2 is 2 kb, and intron 3 is 32 kb in length. Neighboring genes can also be tentatively identified. Exon sequences that display homologies to the major histocompatibility complex class I-related ligand UL16-binding protein gene, hypothetical genes LOC135250, LOC135251, LOC135252, LOC135254, LOC135256, and processed pseudogenes for RNA polymerase B transcription factor 3 (BTF3) and prohibin (PHB) were identified 5Ј to the KEPI gene. The hypothetical gene LOC135258 was identified 3Ј to KEPI gene in these BAC sequences. Interestingly, KEPI maps on chromosome 6 at 157.5 megabases, 4 megabases centromeric to the 161.5-megabase locus for the receptor OPRM1 (AL136444) locus and is transcribed in the same orientation, toward the 6q telomere (NCBI human genome map viewer build 27).

DISCUSSION
Drug-regulated genes include several whose products are involved in signals emanating from activated G i /G o -coupled receptors. These include the G protein ␤ 1 subunit (14), PP2B/ calcineurin (8), and the novel kinase-enhanced PP1 inhibitor KEPI described here. Elucidating these genes raises interest in roles that they or their family members might play in normal cellular functions and in drug-altered functions including those that contribute to addictions.
One approach to these pathways examines the balance between protein kinase and phosphatase activities which regulates dynamic protein phosphorylation process (28). PP1 and PP2A are among the most prominent Ser/Thr protein phosphatases in mammalian cells and can be distinguished from types 2B and 2C phosphatase by their preferential dephosphorylation of phosphorylase A (17). PP1 is thus a signal-transducing enzyme that can modulate the physiological activities of numerous proteins by regulating their states of phosphorylation or dephosphorylation. PP1 is enriched in neuronal postsynaptic densities (29) and acts in neurons to regulate phosphorylation of many neurotransmitter receptors, voltagegated ion channels, ion pumps, and transcription factors (30,31). PP1 can also regulate other cellular processes, including protein synthesis, metabolism, muscle contraction, and cell division (17). The dephosphorylating activities of PP1 are regulated by a group of heat-stable inhibitory proteins (32). These PP1 inhibitory proteins are themselves phosphorylated and regulated by several protein kinases. PKA phosphorylates DARPP-32 (dopamine-and cAMP-regulated phosphoprotein, M r 32,000), inhibitor-1, and NIPP-1 (nuclear inhibitor of PP1) (33). cGMP-dependent protein kinase (protein kinase G) phosphorylates G substrate (34). Glycogen synthase kinase-3 phosphorylates inhibitor-2 (35). Calcium/phospholipid-dependent protein kinase (PKC) phosphorylates CPI-17, PNG/PHI, and KEPI (11,36). PP1 inhibitory activities can be enhanced up to 1,000-fold upon phosphorylation of specific threonine residues in DARPP-32, inhibitor-1, G substrate, CPI-17, PHI/PNG and KEPI (11,17,26,33). In contrast, phosphorylation of NIPP-1 and inhibitor-2 reduces their potencies in inhibiting PP1 (35,37). The activities of PPI inhibitory proteins are returned to the basal level by dephosphorylation (38) by several different phosphatases. Phosphatase 2B/calcinurin dephosphorylates DARPP-32 at its PKA phosphorylation site and virtually eliminates its PP1 inhibitory activity (39). PP1 inhibitory proteins could thus play intermediate regulatory roles between kinases and phosphatases to modulate phosphorylation states of multiple target molecules in response to extracellular signals. Observations that KEPI displays potential phosphorylation sites for PKA and casein kinase II could also indicate that KEPI may be under even more complex regulation by phosphorylation or dephosphorylation at several of its sites KEPI appears well positioned to play a significant role in inhibiting one of the major cellular phosphatases, PP1, when PKC is activated in brain or heart. Such activities are also displayed by several other members of its gene family. CPI-17 is expressed heavily in smooth muscle, where its activation by PKC phosphorylation makes it a potent inhibitor of PP1-myosin light chain complex. When phosphorylated by PKC, PNG/ PHI inhibits myosin-and glycogen-associated PP1 holoenzyme activities in multiple tissues (11,26). We have demonstrated here that recombinant KEPI fusion protein can specifically inhibit rabbit muscle PP1␣ after PKC phosphorylation. In this in vitro test, KEPI inhibition of phosphatase was just as potent as values reported for porcine CPI or human PHI (11,12). In very recent initial work, we have also found that the KEPI family member FLJ29251/"GCPI" is also a potent PKC-dependent PP1 inhibitor found in tissues including gastrointestinal and cerebral cortex. 3 Different tissue and subcellular localizations of members of this family of PKC-dependent PP1 inhibitors provide the opportunity to exert exquisite tissue-specific control over temporal profiles of PP1 activity. Some of the detailed molecular mechanisms for KEPI inhibition of PP1 could share similarities with those of other family members. KEPI shares limited amino acid sequence identity with DARPP-32. By analogy to DARPP-32, a potential KEPI inhibitory domain surrounding a potential phospho-Thr-72 site might bind to the PP1 active site and inhibit its dephosphorylating enzymatic activity (40,41). The N-terminal residues 20 -24 (RVFFQ) of KEPI contain a consensus motif (basic amino acid followed by two hydrophobic residues) that could represent its PP1 binding domain (21). Conceivably, such a KEPI domain could allow it to compete for PP1 binding with its other partners, such as spinophilin, to alter the way in which PP1 could otherwise selectively dephosphorylate molecules such as ligand-and voltage-gated ion channels that are often located near spinophilin (42). In addition to this possible PP1 binding motif, the acidic cluster of KEPI residues 88 -125 shares 50% similarity with DARPP-32 residues 104 -141. Conceivably, the complex structure predicted for KEPI's N-terminal 50 amino acids could allow its centrally localized PKC recognition site to be more accessible for phosphorylation and dephosphorylation. If KEPI were indeed membrane-anchored by myristoylation at its N-terminal glycine (43), its flexible N-terminal region might even allow the C-terminal coiled-coil structure predicted for KEPI to interact with other proteins.
The physiological in vivo significance of KEPI's in vitro potencies could depend on several features, including the cellular levels of KEPI expression, repertoires of expressed PP1 isoforms (44), and expression of other members of this gene family. It will be interesting to compare its expression with those of the FLJ29251/GCPI protein and other members of the more acidic or basic subfamilies of kinase-enhanced PP1 inhibitors. PKC is a heterogeneous enzyme with a dozen isoforms implicated in many cellular responses (45). The exact repertoire of PKC isozymes expressed by the cells of interest could also influence in vivo activities of KEPI. PKC and KEPI subcellular localizations could also vary. Agonist activation of SH-SY5Y cell receptors can lead to membrane translocation and activation of the PKC calcium-and diacylglycerol-regulated ␣ form, its DAG-regulated ⑀ form, and its form (5,46,47). Subcellular localization could also be altered in vivo by differential N-terminal KEPI myristoylation. Acylation could change KEPI's membrane anchoring and thus change its proximity to membrane-bound PKC. Differential phosphorylation by casein kinase II at the SXXE motifs found at Ser-43 and Ser-120 or by PKA at its R/KXXS motifs, including Ser-154, could also contribute to in vivo variations in KEPI activities and enrich the current picture. Even information available to date indicates that the presence of KEPI could result in significantly lengthened half-lives of important protein phosphorylation events when PKC is activated.
Longer protein phosphorylation half-lives could be especially effective in amplifying and prolonging signals from agonist activation of G i /G o -coupled receptors when phosphorylation rates were also enhanced. Agonist-exposed cells that express G i /G o -coupled receptors and high levels of KEPI might thus accumulate many more phosphates on phosphoacceptor sites of important cellular proteins than agonist-exposed cells that contain the same receptors but lack KEPI or its family members. Such mechanisms could enhance "feed forward" mechanisms that could profoundly alter the dynamic properties of phosphorylation events.
These events could change still further after drug treatments. After opiate treatments, for example, cells with upregulated KEPI expression could display more prominent effects of subsequent PKC activation by agonists occupying G i / G o -coupled receptors expressed on their surfaces. Cells coexpressing receptors and KEPI could provide an especially interesting case. In such neurons, the consequences of morphine activation of PKC pathways are likely to be enhanced by KEPI up-regulation if morphine has been present for some time. This up-regulated KEPI could increase the level of phosphorylation of neural signaling molecules such as glutamate receptors, Ca 2ϩ channels, cAMP response element-binding protein and other transcription factors, GABA A receptors, Na ϩ channels, Na ϩ /K ϩ -ATPase (31), and other proteins. Phosphorylated signaling molecules are in turn candidates to contribute to behavioral manifestations such as sensitization, the greater locomotor stimulation that a repeated dose of morphine provides when compared with an initial dose. Most brain KEPI expression is not receptor-dependent, however, because homozygous receptor knockout mice express whole brain KEPI levels indistinguishable from those found in wild type animals. These data combine with data indicating that KEPI expression in different brain regions is influenced differently by acutely and chronically administered morphine. They suggest a more sophisticated role for the regulation of this protein than just morphine or receptor dependence alone. These data also fit with our current observations that iKEPI can be found in neuronal processes and neurons that are likely to contain several different neurotransmitters and are located multifocally in both regions which contain high densities of receptors and in those that do not. 2 KEPI provides an apparent brain counterpart to PKA-regulated inhibitors of PP1, including DARPP-32. Because KEPI is regulated by drugs, the phosphorylation of DARPP-32 is also regulated by psychostimulants (48) and its expression enhanced by antidepressants (49). The KEPI gene has 4 exons, whereas the human DARPP-32 gene appears to display 8 exons. 4 Increasing data that implicate DARPP-32 function in phosphoregulatory events that contribute to aspects of addictive behavior make DARPP-32's PKC-regulated counterparts, including KEPI, candidates to provide similar roles for PKC-regulating G protein-coupled receptors. Because these G i /G ocoupled receptors include many of the receptors that are activated directly or indirectly by most drugs of abuse, including opiates, cannabanoids, and psychostimulants, KEPI and its family members seem especially well positioned to play significant roles in these processes.