Characterization of PKIγ, a Novel Isoform of the Protein Kinase Inhibitor of cAMP-dependent Protein Kinase*

Attempts to understand the physiological roles of the protein kinase inhibitor (PKI) proteins have been hampered by a lack of knowledge concerning the molecular heterogeneity of the PKI family. The PKIγ cDNA sequence determined here predicted an open reading frame of 75 amino acids, showing 35% identity to PKIα and 30% identity to PKIβ1. Residues important for the high affinity of PKIα and PKIβ1 as well as nuclear export of the catalytic (C) subunit of cAMP-dependent protein kinase were found to be conserved in PKIγ. Northern blot analysis showed that a 1.3-kilobase PKIγ message is widely expressed, with highest levels in heart, skeletal muscle, and testis. RNase protection analysis revealed that in most tissues examined PKIγ is expressed at levels equal to or higher than the other known PKI isoforms and that in several mouse-derived cell lines, PKIγ is the predominant PKI message. Partial purification of PKI activities from mouse heart by DEAE ion exchange chromatography resolved two major inhibitory peaks, and isoform-specific polyclonal antibodies raised against recombinant PKIα and PKIγ identified these inhibitory activities to be PKIα and PKIγ. A comparison of inhibitory potencies of PKIα and PKIγ expressed inEscherichia coli revealed that PKIγ was a potent competitive inhibitor of Cα phosphotransferase activity in vitro (K i = 0.44 nm) but is 6-fold less potent than PKIα (K i = 0.073 nm). Like PKIα, PKIγ was capable of blocking the nuclear accumulation of Flag-tagged C subunit in transiently transfected mammalian cells. Finally, the murine PKIγ gene was found to overlap the murine adenosine deaminase gene on mouse chromosome 2. These results demonstrate that PKIγ is a novel, functional PKI isoform that accounts for the previously observed discrepancy between PKI activity and PKI mRNA levels in several mammalian tissues.

Attempts to understand the physiological roles of the protein kinase inhibitor (PKI) proteins have been hampered by a lack of knowledge concerning the molecular heterogeneity of the PKI family. The PKI␥ cDNA sequence determined here predicted an open reading frame of 75 amino acids, showing 35% identity to PKI␣ and 30% identity to PKI␤1. Residues important for the high affinity of PKI␣ and PKI␤1 as well as nuclear export of the catalytic (C) subunit of cAMP-dependent protein kinase were found to be conserved in PKI␥. Northern blot analysis showed that a 1.3-kilobase PKI␥ message is widely expressed, with highest levels in heart, skeletal muscle, and testis. RNase protection analysis revealed that in most tissues examined PKI␥ is expressed at levels equal to or higher than the other known PKI isoforms and that in several mouse-derived cell lines, PKI␥ is the predominant PKI message. Partial purification of PKI activities from mouse heart by DEAE ion exchange chromatography resolved two major inhibitory peaks, and isoform-specific polyclonal antibodies raised against recombinant PKI␣ and PKI␥ identified these inhibitory activities to be PKI␣ and PKI␥. A comparison of inhibitory potencies of PKI␣ and PKI␥ expressed in Escherichia coli revealed that PKI␥ was a potent competitive inhibitor of C␣ phosphotransferase activity in vitro (K i ‫؍‬ 0.44 nM) but is 6-fold less potent than PKI␣ (K i ‫؍‬ 0.073 nM). Like PKI␣, PKI␥ was capable of blocking the nuclear accumulation of Flag-tagged C subunit in transiently transfected mammalian cells. Finally, the murine PKI␥ gene was found to overlap the murine adenosine deaminase gene on mouse chromosome 2. These results demonstrate that PKI␥ is a novel, functional PKI isoform that accounts for the previously observed discrepancy between PKI activity and PKI mRNA levels in several mammalian tissues.
The cAMP-dependent protein kinases (PKAs) 1 comprise a subfamily of serine/threonine kinases that are activated by increases in intracellular concentrations of cAMP. Members of this family play a central role in the coordination of cellular responses to both hormones and neurotransmitters. Upon activation of receptors coupled to adenylate cyclase, intracellular concentrations of cAMP rise, and cAMP binds to each of two regulatory (R) subunits of the inactive tetrameric holoenzyme complex, releasing C subunit. Once released, C subunit phosphorylates both cytoplasmic and nuclear substrates that can alter the rate of cell division, cellular morphology, membrane ion permeability, metabolic enzyme activity, or levels of gene transcription (1)(2)(3).
Extensive biochemical characterization and molecular cloning studies have identified three C subunit (C␣, C␤, and C␥) and four R subunit isoforms (RI␣, RI␤, RII␣, and RII␤) (4,5). Although the amino acid sequences of C␣ and C␤ are highly similar (6), they differ significantly in their tissue distributions and interactions with R subunits. C␣ is widely expressed in mammalian tissues, and C␤ is expressed in cells of the nervous, endocrine, and reproductive systems (6,7). Importantly, PKA holoenzymes formed with RII␣ and C␤ have a 5-fold lower K a value for cAMP than RII␣ and C␣ containing complexes (8). Like the C subunits, the R subunit isoforms also show heterogeneity due to differences in their tissue distributions, subcellular localizations, and interactions with C subunit. The RII isoforms are localized to different parts of the cytosol via their interactions with cAMP-dependent protein kinase-anchoring proteins (9,10), and RI isoforms are generally cytoplasmic. Furthermore, holoenzymes containing RI␣ are less sensitive to increases in cAMP than those containing RI␤ (11,12). Several of these isoform-specific differences have been verified in whole animal studies (13)(14)(15).
In addition to the R subunits that inhibit the activity of the C subunit in a cAMP-regulated manner, there is a second level of regulation of PKA activity by protein kinase inhibitor (PKI) proteins. The PKIs are specific and potent inhibitors of the C subunit; however, unlike R subunits, PKI inhibition of C subunit is not relieved by cAMP. Due to the low levels of PKI activity found previously in tissues relative to C subunit activity and the high binding affinity of the PKIs for C subunit, it has been proposed that the PKIs may regulate basal PKA activity (16,17).
To date, biochemical characterization and molecular cloning of cDNAs encoding PKIs have demonstrated that at least two distinct PKI genes are expressed in mammals, PKI␣ and PKI␤ (18 -21). The PKI␣ isoform is highly expressed in heart, skeletal muscle, cerebral cortex, and cerebellum (18,21), whereas the PKI␤ isoform is most highly expressed in testis (21). Similar to R subunits, both PKI isoforms are pseudosubstrate, competitive inhibitors (19,22,23) and inhibit the C subunit through interactions within the substrate binding site of the C subunit (24). Specific amino acids conserved between PKI␣ and PKI␤1 (Phe 10 , Arg 15 , and the pseudosubstrate sequence (Arg 18 -Arg 19 -Asn 20 -Ala 21 )) (25,26) have been demonstrated to be important for binding and inhibition of the C subunit. Although both isoforms have a high affinity for C subunit, the murine PKI␣ and PKI␤1 isoforms differ significantly in their inhibitory potency, and individual residues important for this difference have been identified (16).
In addition to inhibiting C subunit phosphotransferase activity, the PKIs also serve to localize C subunit in the cell. It has been demonstrated that C⅐PKI complexes are more rapidly exported out of the nucleus than C subunit alone and that this process is both temperature-and ATP-dependent (27). Specifically, a nuclear export signal (NES) has been identified on PKI␣ corresponding to a leucine-rich sequence conserved between PKI␣ and PKI␤ (28).
Previous attempts to understand the cellular roles of the PKIs have been complicated by the heterogeneity of PKI activity in mammalian tissues. To help resolve this problem, we sought to identify the nature of PKI activity in tissues that expressed low levels of PKI␣ and PKI␤. In this report we describe the identification and characterization of a cDNA sequence that encodes a novel PKI isoform which is abundant, widely expressed, and a potent inhibitor of the C subunit of PKA. Because of its similarity to the known murine PKI isoforms, we have named it PKI␥. Like other members of the PKI family, PKI␥ inhibits cAMP-dependent gene transcription and nuclear accumulation of the C subunit. The results of this study suggest that PKI␥ is found at physiologically significant levels in many tissues and functions in a manner similar to previously characterized PKI isoforms.

MATERIALS AND METHODS
Isolation and Sequencing of a cDNA Clone Encoding Murine PKI␥-A full-length cDNA sequence coding for murine PKI␥ (I.M.A.G.E. Consortium Clone 419982/GenBank accession W91205) (29) was identified in a search of the expressed sequence tag data base for protein sequences homologous to murine PKI␣ using the basic local alignment search tool algorithm (30). This I.M.A.G.E. Consortium (LLNL) cDNA clone was obtained from Research Genetics Inc. It was sequenced in both directions by manual sequencing using Sequenase DNA polymerase (U.S. Biochemical Corp.). Sequence analyses were performed using DNASTAR software. The murine PKI␥ sequence has been submitted to the GenBank data base.
Northern Blot Analysis-Plasmids AR-1 and MtPKI.pcr were linearized with BamHI and BglII, respectively, and used to generate antisense RNA probes for PKI␣ and PKI␤, respectively, as described (31). The template for the PKI␥ antisense RNA probe was constructed by inserting the 231-bp BamHI/BglII fragment of pGEM-T.mPKI␥ (described below) into the BamHI/BglII site of pSP73 (Promega) to create pSP73.mPKI␥. This construct was linearized with BglII and used to generate PKI␥ antisense RNA probes. Mouse multiple tissue Northern blots (CLONTECH, Palo Alto, CA) were hybridized at 60°C with antisense cRNA probes for 10 -14 h. Following hybridization, blots were washed for 2 h at 70°C in 0.5 ϫ SET containing 0.1% sodium pyrophosphate, dried, and autoradiographed as described previously (31).
RNase Protection Analysis-RNase protection analysis was performed essentially as described previously (31). The polymerase chain reaction (PCR) was performed with oligonucleotides 5Ј GGA GAT CTC CAC CAT GAC TGA TGT GGA AAC TAC G 3Ј and 5Ј GGG AGA TCT TTA CTT GTC ATC GTC GTC CTT GTA GTC CCC GCT TTC AGA CTT GGC TGC 3Ј (Biomedical Core Facilities, University of Michigan) with pMAL-PKI␣ (32) as a template to amplify a fragment consisting of the full ORF of murine PKI␣ flanked by BglII sites. This amplified fragment was digested with BglII, isolated, and ligated into the BamHI/ BglII sites of pSP73 to create pSP73.mPKI␣. The resulting plasmid was linearized with BglII or HindIII and used as a template to synthesize antisense PKI␣ ORF RNA probes or sense RNA standards, respectively. MtPKI.pcr was used as a template to synthesize antisense murine PKI␤ ORF RNA probes or sense RNA standards as described (31). pSP73.mPKI␥ was linearized with BglII or BamHI and used as a template to synthesize antisense PKI␥ ORF RNA probes or sense RNA standards, respectively. Total RNA was isolated from mouse tissues and cell lines by using an acid guanidinium isothiocyanate/phenol/ chloroform protocol (33). T7 RNA polymerase was used to generate [␣-32 P]UTP-radiolabeled antisense RNA probes. The radiolabeled probes were incubated with 10 g of total RNA or varying amounts of sense RNA (0, 0.3, 1, 3, 10, or 30 pg) for 16 h at 50°C. Yeast tRNA was added to sense RNA samples to bring the RNA total to 10 g, and the samples were then treated with RNase A (20 g/ml) and RNase T1 (200 units/ml) (Sigma). The protected fragments were isolated and electrophoresed through 6% polyacrylamide sequencing gels. PhosphorImager quantitation was performed in a PhosphorImager apparatus and analyzed with IMAGEQUANT software (Molecular Dynamics).
Construction of PKI Mammalian Expression Vectors, Transient Transfection of HEK293 Cells, and Determination of C Subunit Inhibitory Activity-A mammalian expression plasmid encoding a carboxylterminal hemagglutinin (HA)-tagged murine PKI␣ protein was constructed by PCR. A PCR fragment encoding PKI␣ with a carboxylterminal 12CA5 epitope (YPYDVPDYA) and a one amino acid glycine linker was generated using primers 5Ј GGA GAT CTC CAC CAT GAC TGA TGT GGA AAC TAC G 3Ј and 5Ј GGG AGA TCT TTA AGC GTA GTC TGG GAC GTC GTA TGG GTA CCC GCT TTC AGA CTT GGC TGC 3Ј with pMAL-PKI␣ as a template. The resulting PCR fragment was digested with BglII, isolated, and ligated into BglII digested pCM-V.Neo (34) to create pCMV.HA-PKI␣. Likewise, the pCMV.mPKI␥ mammalian expression vector was constructed by PCR using the oligonucleotides 5Ј GGG AGA TCT CCA CCA TGA TGG AAG TCG AGT CCC 3Ј and 5Ј GGG GGA TCC TCA GGA TGA GGT GTT CGC ATC 3Ј with clone 419982 as a template. The resulting PCR fragment containing the coding region of PKI␥ flanked by a BglII site and a BamHI site was ligated into pGEM-T (Promega) to create pGEM-T.mPKI␥. pGEM-T.mPKI␥ was digested with BamHI and BglII, and a 231-bp fragment coding for mPKI␥ was isolated and ligated into the BglII site of pCM-V.Neo. Both pCMV.HA-PKI␣ and pCMV.mPKI␥ were sequenced to confirm the coding region sequence. The human PKI␣ and murine PKI␤1 mammalian expression plasmids have been described previously (20,35). The human PKI␣ protein is 97% identical to the murine PKI␣ amino acid sequence (35). HEK293 cells at 50% confluency in 10-cm plates were transfected using a calcium phosphate co-precipitation method (36) with 25 g per plate of either pCMV.hPKI␣, pCM-V.mPKI␤1, pCMV.mPKI␥, or pCMV.Neo. Forty-eight hours after application of DNA precipitates, plates were washed twice with ice-cold phosphate-buffered saline (PBS). Following the addition of 200 l of homogenization buffer (10 mM sodium phosphate (pH 7.0), 1 mM EDTA, 1 mM dithiothreitol, 250 mM sucrose) containing 1 mM phenylmethylsulfonyl fluoride, 1 g/ml pepstatin A, and 1 g/ml leupeptin (Sigma), cells were scraped into separate tubes and sonicated twice for 10 s. C subunit inhibitory activity was determined essentially as described (35). Extracts were heated to 95°C for 5 min. Various concentrations of heat-denatured extract were added to recombinant C␣ (1 nM) (32) in a phosphotransferase assay mix for 10 min at 30°C. The assay was initiated by addition of Kemptide substrate (30 M), incubated for an additional 20 min, and then terminated.
Partial Purification of PKI Activities from Mouse Heart-PKI activities from mouse heart were partially purified by a modification of previously described procedures (19,37,38). Mouse hearts (2.5 g) (Pel-Freeze) were frozen in liquid nitrogen, pulverized, and added to 8 ml of homogenization buffer containing 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, and 1 g/ml pepstatin A. The ice-cold suspension was homogenized with 20 strokes in a Dounce homogenizer (Wheaton), sonicated for 10 s at 4°C, and placed in a boiling water bath for approximately 5 min until the suspension temperature rose to 95°C. The resulting precipitate was removed by centrifugation for 1 h at 50,000 rpm in a TV-1665 rotor (Sorvall). The supernatant was adjusted to pH 4.0 by addition of glacial acetic acid and incubated on ice for 1 h. The resulting precipitate was removed by centrifugation for 15 min at 13,000 rpm in an HB-6 rotor (Sorvall). The supernatant was adjusted to pH 7.0 by addition of 1 M Tris base and dialyzed in dialysis tubing (molecular weight cut-off, 3500) against three 2-liter changes of 5 mM KPO 4 , 1 mM EDTA, pH 7.0. The dialysate (25 ml) was adjusted to pH 5.0 and absorbed to a 2-ml Bio-Scale-DEAE2 column equilibrated with 5 mM KPO 4 , pH 5.0, using the BioLogic system (Bio-Rad). The column was washed with 15 ml of 5 mM KPO 4 , pH 5.0, and 5 ml of 5 mM NaOAc, pH 5.0. Proteins were eluted with a 40-ml linear gradient of 5-1000 mM NaOAc, pH 5.0, as 1-ml fractions were collected. Fractions were adjusted to pH 7.4 with 2 M Tris, pH 8.0, and assayed for PKI inhibitory activity as described above.
Generation of Polyclonal Antibodies to PKI␣ and PKI␥-The pET9d.His 6 PKI␣ and pET9d.His 6 PKI␥ prokaryotic expression vectors were generated using PCR. Specific sense and antisense oligonucleotides were used to generate PCR fragments encoding the full-length ORFs of murine PKI␣ and PKI␥ each with an amino-terminal hexahistidine tag. pMAL-PKI␣ and clone 419982 were used as templates in the PKI␣ and PKI␥ PCR reactions, respectively. Fragments were digested with NcoI and BamHI, isolated, and ligated into pET9d (Novagen) that had been NcoI-and BamHI-digested. Escherichia coli (BL21(DE3)/ pLysS strain) were transformed with pET9d.His 6 PKI␣ and pET9d.His 6 PKI␥. E. coli cultures (1 liter) were grown and induced with 1 mM isopropyl-1-thio-␤-D-galactopyranoside for 3 h at 37°C. Pellets were resuspended in 20 ml of buffer A (20 mM Tris, pH 8.0, 300 mM NaCl, 0.1% Triton X-100) containing 1 mM phenylmethylsulfonyl fluoride, 1 g/ml leupeptin, and 1 g/ml pepstatin A. Each suspension was sonicated three times for 1 min, and imidazole was added to a final concentration of 20 mM. The bacterial lysate was centrifuged for 1 h at 100,000 ϫ g, and the supernatant was loaded onto a 2-ml nickel affinity resin (Qiagen) column. The column was washed with 10 column volumes of buffer A containing 20 mM imidazole and eluted with a step gradient of imidazole in buffer A. The majority of each His-tagged PKI eluted in the 80 mM imidazole elution. Purified His 6 PKI␣ and His 6 PKI␥ were separately conjugated to keyhole limpet hemocyanin and used to immunize rabbits for antibody production (Research Genetics Inc.).
Western Blotting-Antisera raised against His-tagged PKI␣ and PKI␥ were affinity purified on MBP-PKI␣ (32) and MBP-PKI␥ (see below) nitrocellulose blots essentially as described (39,40). Affinity purified anti-PKI␣ antibody recognized PKI␣ (14 kDa) but did not react with PKI␤ or PKI␥ on Western blots of HEK293 cell extracts from cells transfected with PKI expression constructs. Likewise, affinity purified anti-PKI␥ recognized PKI␥ (16 kDa), but no signal was detected in extracts from PKI␣-or PKI␤-transfected cells. Fractions from the DEAE column were concentrated in microconcentrators (Microcon-3, Amicon), denatured in SDS-PAGE buffer at 95°C for 5 min, resolved on 15% SDS-PAGE gels, and transferred to 0.2-m nitrocellulose membranes (BA-83, Schleicher and Schuell). Membranes were blocked for 4 h in PBS supplemented with 5% non-fat dried milk, 2% polyvinylpyrrolidone (PVP-40), and 0.1% Triton X-100 and subsequently incubated with either a 1:10 dilution of affinity purified anti-PKI␣ or anti-PKI␥ in PBS supplemented with 0.5% bovine serum albumin and 0.1% Triton X-100 for 2 h. Filters were washed three times for 10 min with TBST (50 mM Tris, pH 7.5, 150 mM NaCl, and 0.05% Tween 20), and then they were incubated with a 1:10,000 dilution of goat anti-rabbit alkaline phosphatase (Life Technologies, Inc.) in TBST supplemented with 5% non-fat dried milk as the secondary antibody for 2 h. Following the final set of three 10-min washes with TBST, the blots were developed with the nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate substrate system (Life Technologies, Inc.). 50 and K i Values-To create an MBP-PKI␥ prokaryotic expression vector, the murine PKI␥ coding region was amplified by PCR using oligonucleotides 5Ј GGG GAA TTC ATG GAA GTC GAG TCC TCC 3Ј and 5Ј GGG GAA TTC TTA GGA TGA GGT GTT CGC ATC 3Ј designed to create EcoRI recognition sites. The resulting PCR fragment was cut with EcoRI, isolated, and ligated into the EcoRI site of pMAL.cRI (New England Biolabs). This vector was sequenced and transformed into E. coli (XLI-Blue, Stratagene). The MBP-PKI␥ fusion protein was expressed and purified over a amylose affinity column as described previously (32). Protein concentrations of the purified fusion proteins were determined by the Bradford method (Bio-Rad Protein Assay). MBP-PKI␣ and MBP-PKI␥ were greater than 95% pure as determined by scanning densitometry of Coomassie Blue-stained SDS-PAGE gels. IC 50 values and K i values for MBP-PKI␣ and MBP-PKI␥ were determined essentially as described (16). Fifty pM recombinant C␣ was used in all phosphotransferase activity assays. K i values for murine PKI␣ and PKI␥ were determined using the Henderson method for tightly bound inhibitors (41).

Construction of Mammalian Expression Vectors for Flag-tagged C Subunits-Mammalian expression plasmids encoding amino-terminal
Flag-tagged murine C␣ and C␣Y235S/F239S (42) proteins were constructed using the PCR method. An amino-terminal C␣ PCR fragment containing an amino-terminal Flag epitope (DYKDDDDK) was generated using the primers 5Ј GGG GGA TCC ACC ATG GAC TAC AAG GAC GAC GAT GAC AAG GGC AAC GCC GCG GCC GCC AAG AA 3Ј and 5Ј AAG TAC TCC GGA GTC CCA C 3Ј with pGEM-4.C␣ (43) as a template. The resulting PCR fragment was digested with BamHI and BglII, isolated, and ligated into BglII-digested pCMV.Neo to create pCMV.Flag-C␣1. pGem-4.C␣ was digested with BamHI and BglII. An approximately 450-bp fragment encoding the carboxyl-terminal of C␣ was isolated and ligated into BglII digested pCMV.Flag-C␣1 to generate pCMV.Flag-C␣2. A 240-bp BglII fragment coding the central onethird of C␣ was isolated from the same digest and ligated into BglII digested pCMV.Flag-C␣2 to generate the wild type Flag-tagged C␣ expression vector pCMV.Flag-C␣3. To create the C␣Y235S/F239S mutant expression vector, pET9d.C␣Y235S/F239S (42) was digested with BglII, and the 240-bp fragment containing the mutated sites was ligated into BglII-digested pCMV.Flag-C␣2 to generate pCMV.Flag-C␣Y235S/F239S. Both expression plasmids were restriction mapped and sequenced.
Transient Transfection of NIH 3T3 Cells and Luciferase Assays-NIH 3T3 cells were grown on 10-cm plates to 50% confluency and transfected using a standard calcium phosphate method (36) with 1 g of pCMV.Flag-C␣3 or pCMV.Flag-C␣Y235S/F239S, 1 g of human chorionic gonadotropin-luciferase (HCG.Luciferase), 5 g of pRSV.␤gal, and the indicated amounts of pCMV.mPKI␥. The total amount of plasmid DNA was brought to 25 g with the parental vector pCMV.Neo. Twenty-one hours after transfection, cells were washed twice with ice-cold PBS, scraped into homogenization buffer, sonicated, and assayed for luciferase and ␤-galactosidase activities as described (8).
Immunofluorescence-COS-1 cells or CV-1 cells were grown in Dulbecco's modified Eagle's medium containing 10% fetal calf serum in 8-well tissue culture chambers on poly-L-lysine-coated glass slides (Lab-Tek) to 30% confluency and transfected using a standard calcium phosphate method (36). Cells were transfected with 2 g of pCMV.Flag-C␣3, 8 g of pCMV.RII␣, and either no PKI expression vector, 8 g of pCMV.HA-PKI␣, or 4 g of pCMV.mPKI␥. Total plasmid concentration was maintained at 25 g by addition of the parental vector, pCMV.Neo. Following a 12-h incubation with DNA precipitates, cells were washed once with Dulbecco's modified Eagle's medium and grown for 24 h. Indicated cells were then stimulated with forskolin (25 M) and 3-isobutyl-1-methyl-xanthine (500 M) in Dulbecco's modified Eagle's medium for 40 min at 37°C. Following stimulation, cells were washed twice with ice-cold PBS and fixed with 4% formaldehyde in PBS for 10 min at room temperature followed by a 1:1 mixture of methanol and acetone for 5 min. After washing three times with PBS, cells were incubated with an anti-Flag epitope antibody (M2) (Eastman Kodak) at a 1:2000 dilution in PBS supplemented with 1% bovine serum albumin, 1% horse serum, and 0.1% saponin (Sigma). After four washes with PBS supplemented with 0.1% saponin, a 1:3000 dilution of Cy3-F(abЈ) 2 fragment goat anti-mouse IgG (Jackson) was incubated with the cells for 1 h in the dark in PBS supplemented with 1% bovine serum albumin, 1% horse serum, and 0.1% saponin. Prior to examination by fluorescence microscopy (Olympus), cells were washed four times for 2 min in PBS plus 0.1% saponin and twice for 2 min in PBS. In control experiments, no fluorescence was detected in non-transfected cells.

Identification of cDNA Clone Encoding Murine PKI␥-A
comparison of PKI activity of tissue extracts with PKI␣ and PKI␤ tissue mRNA levels supported the idea that novel PKI isoforms might exist (18,20,21). For example, tissues such as kidney and liver show low levels of PKI␣ and PKI␤ mRNA but significant amounts of PKI inhibitory activity (data not shown).
To determine if evidence for other PKI isoforms existed, the amino acid sequence for murine PKI␣ (44) was used to search for homologous sequences in the NCBI GenBank expressed sequence tag data base using the basic local alignment search tool program. This search identified a murine full-length cDNA clone (I.M.A.G.E. consortium clone 419982) derived from a mouse embryo (embryonic day 13.5-14.5) cDNA library which encoded a protein having statistically significant homology (p Ͻ 0.001) to the murine PKI␣ amino acid sequence. The cDNA clone was fully sequenced (Fig. 1), shown to contain 1064 base pairs (bp), and characterized further.
Assuming that the first of two methionine codons encodes the initiator methionine, the murine PKI␥ cDNA contains an ORF of 231 nucleotides and a 3Ј-untranslated sequence with a putative polyadenylation signal AATAAA-(1042-1047) and a poly(A) tail (Fig. 1). The PKI␥ protein predicted by the ORF is 75 amino acids in length with a calculated molecular mass of 7.8 kDa and pI of 3.9. PKI␥ shows relatively low amino acid homology to the other known murine PKI isoforms: PKI␣ (35% identity) and PKI␤1 (30% identity). However, most of the residues demonstrated to play a role in the high affinity of PKI␣ for the C subunit are conserved. Not only is the inhibitory pseudosubstrate sequence (Arg 18 -Arg 19 -Asn 20 -Ala 21 ) found, but also Arg 15 , Phe 10 and Tyr 7 are conserved (16,25,26). Likewise, a sequence highly similar to the consensus PKI nuclear export signal "L 37 XL 39 XL 41 XXL 44 XHy 46 " (where X is any amino acid and Hy is any hydrophobic amino acid) is present in PKI␥ (28) (Fig. 2). Based on these elements of sequence homology, the new protein was designated PKI␥. Most of the homology between PKI␥ and the other known PKI isoforms occurs in the amino-terminal two-thirds of the protein, whereas little homology is seen in the carboxyl-terminal portion of the protein (Fig. 2).
Expression of PKI Isoforms in Mouse Tissues-To compare directly the expression patterns of PKI isoforms in mouse tissues, poly(A) ϩ Northern blots were probed using antisense RNA probes specific for the known murine PKI isoforms (PKI␣, PKI␤, and PKI␥). The PKI␣ probe detected a single 4.3-kb transcript in heart, brain, and skeletal muscle with lower levels in lung and kidney (Fig. 3A). Prolonged exposure showed barely detectable levels of message in spleen, liver, and testis (data not shown). These data are consistent with the previously reported Northern blot analysis of mouse and rat tissues (18,21). Hybridization with a PKI␤ probe detected two messages, a strongly hybridizing species of approximately 700 bp in testis and a weakly hybridizing 1.8-kb RNA in brain, spleen, lung, and testis (Fig. 3B). Low levels of the 1.8-kb message could also be detected in heart, liver, skeletal muscle, and kidney with longer exposures (data not shown). Previously, it was reported that a 1.8-kb PKI␤ transcript was expressed in all mouse tissues with highest levels in skeletal muscle (20). The discrepancy between these two studies may be due to differences in the age of the mice examined. Both a 700-bp and 1.8-kb PKI␤ transcript have been identified in rat testis, and the 700-bp message has been shown to be developmentally regulated (21).
It has yet to be determined whether this 700-bp rat message is the product of a highly homologous gene, due to alternative polyadenylation or due to alternative splicing of the rat PKI␤ gene. The PKI␥ probe detected a single 1.3-kb transcript in all tissues examined. Unlike the PKI␣ and PKI␤ transcripts, this message was expressed at high levels in all tissues tested. Highest levels were seen in heart, skeletal muscle, and testis; however, significant levels were also seen in spleen, lung, liver, and kidney (Fig. 3C). The presence of PKI␥ mRNA in these organs is important since they have low levels of PKI␣ and PKI␤ message (18,20,21).
RNase Protection Analysis of PKI Isoform Expression in Mouse Tissues and Mouse-derived Cell Lines-PKI activity has been extensively characterized in several tissues where PKI␥ mRNA is detected in Northern blot analysis including skeletal muscle (21,22), heart (38,45), and testis (19,37). However, it was surprising that PKI␥ had not been identified previously. One explanation could be that PKI␥ transcripts are widely expressed but of low abundance relative to PKI␣ and PKI␤ transcripts. To quantitate PKI␥ mRNA more accurately, RNase protection assays using isoform-specific, antisense RNA probes of similar length were performed. Antisense coding region probes of the PKI isoforms were chosen for these experiments so as to generate similar size protected fragments and to detect all possible coding region splice variants. For PKI␣, a protected fragment of 228 bp was seen prominently in heart, brain, and skeletal muscle with weaker bands in lung, kidney, and thyroid. Even with prolonged exposures, no fragments were detected in spleen, liver, or testis (data not shown). After hybridization with PKI␤ probe, a protected fragment of 254 nucleotides was observed in all tissues examined. The highest level of expression was observed in the testis with significantly lower levels in all other tissues (data not shown). Using the same total RNAs and conditions as those used for the PKI␣ and PKI␤ RNase protections, hybridization with the PKI␥ probe resulted in a protected fragment of 233 bp apparent in all mouse tissues studied with the highest levels found in heart and testis (Fig. 4A). Significant levels of PKI␥ mRNA were also seen in uterus, prostate, small intestine, and stomach (data not shown). The tissue distributions of the PKI isoforms as determined by RNase protection analysis are in good agreement with the Northern blot analysis with the exception of PKI␥ expression in the testis. The PKI␥ Northern blot analysis shows an intermediate level of PKI␥ expression in the testis (Fig. 3C), and the RNase protection analysis suggests that the testis is the organ of highest PKI␥ mRNA expression (Fig. 4A). PhosphorImager quantitation of sense RNA-protected frag-FIG. 1. Nucleotide and predicted amino acid sequence of murine PKI␥. Amino acid sequence of murine PKI␥ inferred from the nucleotide sequence is represented below the DNA sequence with the one-letter amino acid codes. Initiator ATG indicated was selected as the translation start site due to optimal alignment with the other known PKI isoforms. Nucleotide numbers are indicated at the left of the sequence, and amino acid numbers are indicated at the right of the sequence. The nucleotide sequence homologous to the murine adenosine deaminase genomic sequence starts at base 377 and extends to the end of the sequence. The putative polyadenylation signal is underlined once.
FIG. 2. Amino acid sequence alignment of murine PKI isoforms. Predicted protein sequence of murine PKI␥ is compared with the protein sequences of murine PKI␣ and murine PKI␤1 in the above alignment using DNASTAR software. The numbering of the three sequences begins with the predicted or known initiator methionine and is placed on the right of the diagram. Amino acid residues identical between any two of the three sequences are boxed. Residues known to be important in high binding affinity of PKI␣ for C subunit are indicated with a plus sign on the top line (Tyr 7 , Phe 10 , Arg 15 , and pseudosubstrate sequence). Hydrophobic residues shown to be important in the nuclear export of C subunit by PKI␣ are marked by an asterisk on the bottom line.
ments and tissue-protected fragments indicates that PKI␥ mRNA is expressed at comparable levels to PKI␣ and PKI␤ in all mouse tissues tested. In several tissues such as heart, lung, liver, and kidney, PKI␥ is the predominant PKI isoform transcript (Fig. 4, B-D).
To confirm the widespread nature of PKI␥ transcripts, total RNA from several mouse-derived cell lines was isolated and assayed for PKI isoform transcript expression by RNase protection analysis. As for the mouse tissues, hybridization with the PKI␥ probe generated a 233-bp fragment in all cell lines tested (data not shown). C 2 C 12 myoblasts, N1E-115 neuroblastoma and L939 cells showed the highest levels of expression, and TM4 Sertoli cells showed the least (Fig. 4E). Only the N1E-115 neuroblastoma cell line displayed a significant level of PKI␣ transcript, whereas no bands were detected in any of the cell lines in the PKI␤ RNase protection analysis (data not shown). Of interest, no evidence of PKI alternative splice variants was observed in these RNase protection experiments.
Separation and Identification of PKI Activities from Mouse Heart-Another possible explanation for why PKI␥ was not detected during prior PKI purifications is that it possesses significantly different biochemical characteristics from the other two known isoforms. Analysis of PKI␣, PKI␤1, and PKI␥ amino acid sequences shows that the three proteins vary considerably in their predicted isoelectric points; PKI␣ has a pI of 4.4, PKI␤1 has a pI of 5.1, and PKI␥ has a pI of 3.9. Previous purifications of PKI␣ and PKI␤1 have involved elution from a DEAE-cellulose column with a linear gradient of NaOAc (0 -350 mM) at pH 5.0 (19,38). Under these conditions, PKI␤1 eluted at lower ionic strengths than PKI␣. Since PKI␥ has a lower pI than PKI␣, it would be expected to elute later in the gradient. No inhibitory peaks were eluted after PKI␣, but it is possible that PKI␥ was never eluted from the column. To test this hypothesis and to verify that the PKI␥ protein is present in native mouse tissues, PKI activities from mouse heart were partially purified by a procedure similar to a previously described three-step purification scheme involving heat denaturation, acid treatment, and DEAE chromatography (19,37,38).
This purification was modified to extend the elution from 350 mM NaOAc to 1 M NaOAc. Mouse heart was chosen for this experiment due to its high levels of both PKI␣ and PKI␥ message as determined by Northern blot and RNase protection analysis. In vitro kinase inhibition assays were conducted on the mouse heart fractions obtained from the DEAE column.
Two peaks of inhibitory activity were detected, one that eluted at a theoretical salt concentration of 250 -325 mM NaOAc (9.9 -13.0 millisiemens/cm), and a second that eluted at a theoretical salt concentration of 450 -575 mM NaOAc (16.0 -20.0 millisiemens/cm) (Fig. 5A). To determine if fractions of these peaks were likely to contain PKI␣ or PKI␥, eukaryotic expression vectors for human PKI␣ and murine PKI␥ were transiently transfected into HEK293 cells, and PKI activities were partially purified as for mouse heart. Assay of the frac-  tions from PKI␣-transfected HEK293 cell extracts revealed a single inhibitory peak at approximately 250 mM NaOAc, overlapping the first heart inhibitory peak. Assay of the fractions from the HEK.mPKI␥ extracts likewise revealed a single inhibitory peak at approximately 500 mM NaOAc correlating exactly with the elution profile of the second heart inhibitory peak. No inhibitory peaks were identified in control HEK cells transfected with the parental expression vector pCMV.Neo (data not shown).
To confirm the identity of the inhibitors responsible for inhibitory peak 1 and peak 2 from mouse heart, isoform-specific PKI␣ and PKI␥ antisera were prepared from rabbits using His-tagged PKI␣ and His-tagged PKI␥ as antigens. The specificity of these antisera was determined using HEK293 cells transfected with pCMV.hPKI␣ (HEK.hPKI␣ extracts), pCM-V.mPKI␤1 (HEK.mPKI␤1 extracts), pCMV.mPKI␥ (HEK-.mPKI␥ extracts), and pCMV.Neo (HEK.Neo extracts). These extracts were resolved by 15% SDS-PAGE and transferred to nitrocellulose. Anti-PKI␣ sera recognized a single band with an apparent molecular mass of 14 kDa in the HEK.hPKI␣ extracts. No bands were detected in the other extracts. Similar results were obtained with the anti-PKI␥ sera that detected a single band with an apparent molecular mass of 16 kDa in the HEK.mPKI␥ extract but no bands in the other lanes. Preimmune sera from the two rabbits did not detect any bands in similar experiments (data not shown).
When samples from the heart inhibitory peak fractions were analyzed by Western blotting, the affinity purified anti-mPKI␣ antibody detected a single 14-kDa band in peak 1 fractions (Fig. 5B). This band was the same apparent molecular mass as transfected PKI␣. Likewise, the affinity purified anti-PKI␥ antibody recognized a single 16-kDa band in peak 2 fractions that co-migrated with transfected PKI␥ (Fig. 5C). Importantly, the intensity of the 14-kDa band in peak 1 fractions and the intensity of the 16-kDa band in peak 2 fractions correlated directly with the C subunit inhibitory activity of the fractions showing that PKI␣ is the inhibitor responsible for the majority of inhibitory peak 1 and PKI␥ is the inhibitor responsible for the majority of inhibitory peak 2. These results suggest that both PKI␣ and PKI␥ are expressed at significant levels in mouse heart and that PKI␥ can be distinguished from other PKI isoforms both immunologically and by its DEAE elution characteristics.
Kinetic Analysis of Murine PKI␣ and PKI␥-To study the inhibitory potency of murine PKI␥, full-length PKI␥ was PCRamplified, cloned into pMAL.cRI, and expressed in E. coli as an MBP-PKI␥ fusion protein. Previously, it has been shown that the presence of an amino-terminal MBP fusion does not affect the inhibitory efficacy of PKI␣ or PKI␤1 (16). To assess relative inhibitory efficacy, PKI␥ was compared with murine PKI␣ (16) in in vitro kinase inhibition assays. C subunit phosphotransferase activity was measured in the presence of increasing concentrations of either PKI␣ or PKI␥. A representative experiment is shown in Fig. 6A, and (Table I). To measure more accurately the difference in inhibitory potencies between these two tight binding inhibitors, K i values were determined by Henderson analysis (41) (Fig. 6, B and C). The K i values for PKI␣ and PKI␥ were determined to be 0.073 and 0.44 nM, respectively (Table I). The increase in slope with increasing Kemptide substrate concentrations in Fig. 6B suggests that as for PKI␣ and PKI␤1, PKI␥ is a competitive inhibitor of C subunit (16,19,22,23). Hence, PKI␥ is a potent, competitive inhibitor of C subunit; however, it is 6-fold less potent than PKI␣. PKI␥ possesses all the amino acid residues previously shown to be important in the high affinity of PKI␣ for C subunit (Tyr 7 , Phe 10 , Arg 15 , Arg 18 , and Arg 19 ) (16,25,26); however, it is possible that PKI␥ contains negative binding determinants. Mutagenesis of Thr 8 and Ser 12 in PKI␤1 to alanines caused a 4-fold increase in the inhibitory potency of PKI␤1. It has been postulated that these residues may block formation of the amino-terminal inhibitory ␣-helix of PKI␤1 (16). Thus, it is possible that Ser 8 and Ser 12 of PKI␥ are negative determinants of C subunit binding.
Mammalian Expression and C Subunit Inhibitory Activity of PKI␥-Expression experiments were performed to demonstrate that the PKI␥ cDNA sequence was capable of producing a relatively heat-and acid-stable inhibitor of PKA. In addition, these expression experiments sought to characterize the ability of PKI␥ to inhibit cAMP-dependent gene transcription and C subunit nuclear accumulation. pCMV.mPKI␥, a PKI␥ mammalian expression vector, was constructed. Initially, HEK293 cells were transiently transfected with pCMV.mPKI␥, pCMV.h-PKI␣, pCMV.mPKI␤1, or pCMV.Neo. Cell extracts were heated and acid-treated, and increasing amounts were added to an in vitro kinase inhibition assay to determine if they could inhibit recombinant C subunit. HEK293 cells transfected with either of the three PKI isoform expression vectors contained at least 10-fold higher levels of C subunit inhibitory activity than those FIG. 5. Partial purification and identification of PKI isoforms from murine heart. A, heat-and acid-treated mouse heart extract (25 ml) was chromatographed on a DEAE column (see "Materials and Methods"). Fractions (1 ml) were collected and assayed for C subunit inhibitory activity. Inhibitory activity is defined as the percent inhibition of total control kinase activity and is denoted by a solid line (q). Conductivity is represented by a dashed line. The open bar spans the fractions of inhibitory peak 1 (fractions 10 -15), and the hatched bar spans the fractions of inhibitory peak 2 (fractions 18 -23). B and C, Western blot analysis of PKI activities eluted from DEAE column. Aliquots from fractions containing significant inhibitory activity (peak 1, fractions 10 -15; peak 2, fractions 18 -23) were concentrated, resolved by 15% SDS-PAGE, and immunoblotted with either affinity purified anti-PKI␣ antibody (B) or affinity purified anti-PKI␥ antibody (C) using colorimetric detection. The sizes of the protein standards are indicated on the left of the figure in kilodaltons (kDa). The identity and the apparent molecular mass of the identified bands is indicated on the right with an arrow.
transfected with the parental vector pCMV.Neo. The extracts from cells transfected with pCMV.mPKI␥ showed greater inhibitory activity than pCMV.mPKI␤1-transfected cells but less inhibitory activity than the pCMV.hPKI␣-transfected cells (data not shown). The presence of the PKIs in cell extracts was verified with PKI-specific antibodies (data not shown).
In Vivo Inhibition of cAMP-dependent Gene Transcription by PKI␥-To verify the ability of PKI␥ to inhibit C subunit in vivo, NIH 3T3 cells were transiently transfected with a constant amount of the C␣ expression vector, either pCMV.Flag-C␣3 or pCMV.Flag-C␣Y235S/F239S, and increasing amounts of pCMV.mPKI␥. Each plate also received a constant amount of a cAMP-responsive reporter plasmid (HCG.luciferase) (46) and pRSV.␤gal to control for transfection efficiency. In this experiment, luciferase activity was used as an in vivo measure of free cellular C subunit. Transient transfection of either C subunit alone produced a 35-fold increase in luciferase activity (data not shown). One to two g of pCMV.mPKI␥ was required to completely inhibit luciferase activity in cells transfected with pCMV.Flag-C␣3. NIH 3T3 cells expressing the Flag-tagged C␣Y235S/F239S mutant, a C␣ mutant with reduced affinity for PKI␣ and PKI␤1 (42), showed significantly less inhibition of luciferase activity at all pCMV.mPKI␥ concentrations (Fig. 7). It has been shown previously that three PKI␣ residues outside the pseudosubstrate sequence, Tyr 7 , Phe 10 , and Arg 15 , contribute significantly to the high affinity of PKI␣ for C subunit. Specifically, Phe 10 of PKI␣ interacts with a hydrophobic pocket on the surface of the C subunit consisting of residues Tyr 235 and Phe 239 (24,42). The reduced ability of PKI␥ to inhibit the C␣Y235S/F239S mutant is consistent with the conservation of Phe 10 between PKI␣ and PKI␥ (Fig. 2). Interestingly, even though C␣Y235S/F239S is 2000-fold less sensitive to inhibition by PKI in vitro (18), it is still significantly inhibited in vivo. This is likely due to the fact that in vivo concentrations of C subunit reach micromolar levels, whereas nanomolar concentrations of C subunit are used in in vitro experiments.
PKI␥ Blocks Nuclear Accumulation of C Subunit-Both PKI␣ and PKI␤ are capable of actively exporting C subunit from the nucleus (27). In the case of PKI␣, this export of C⅐PKI complexes has been shown to require a nuclear export signal (NES) consisting of a hydrophobic sequence that is conserved between PKI␣ and PKI␤1 (28). Amino acid alignment of PKI␥ with PKI␣ and PKI␤1 reveals the possible presence of a NES in PKI␥ (Fig. 2). The most significant difference observed was a glycine at amino acid residue 39 where both PKI␣ and PKI␤1 contain leucines. To test the functionality of the NES of PKI␥, a transient transfection assay was developed. An amino-terminal Flag-tagged C␣ expression vector was constructed to determine the localization of C subunit in transfected cells. To verify that the Flag-tagged C subunit was catalytically active and that the Flag epitope was accessible both when the C subunit was free and when it was bound to a PKI, HEK293 cells were transiently transfected with the Flag-tagged C subunit alone or with a combination of Flag-tagged C subunit and PKI␣. Western blot analysis of these HEK293 cell extracts with the M2 anti-Flag antibody detected a single band with an apparent molecular mass of 41 kDa, the size expected for full-length C subunit (data not shown). No band was seen in extracts of HEK293 cells transfected with the parental vector alone (data not shown). When extracts of HEK293 cells transiently transfected with or without the Flag-tagged C subunit vector alone were subjected to immunoprecipitation with the M2 anti-Flag antibody, only the immunoprecipitates isolated from extracts  Table I.  Fig. 6A. The K i values were obtained from Henderson analyses (Fig. 6, B and C). IC 50  made from cells transfected with pCMV.Flag-C␣3 showed significant C subunit activity (data not shown). Importantly, boiled M2 antibody immunoprecipitates from extracts of HEK293 cells co-transfected with Flag-tagged C subunit and PKI␣ showed significant C subunit inhibitory activity (data not shown). Moreover, immunoprecipitation of this inhibitory activity was dependent on the presence of Mg 2ϩ and ATP in the immunoprecipitation buffer, both critical to the tight binding of C subunit to PKI (47). Hence, a catalytically active Flag-tagged C subunit is produced in cells transiently transfected with the pCMV.Flag-C␣3 expression vector, and it is recognized by the M2 anti-Flag monoclonal antibody when free in solution and when bound to PKI␣.
To study the effect of elevations in cAMP on the cellular localization of C subunit in the presence and absence of PKI␥, CV-1 cells were transiently co-transfected with expression vectors for RII␣, amino-terminal Flag-tagged C␣ and either no PKI, PKI␣, or PKI␥. Enough RII␣ expression vector was used in each transfection to completely inhibit transfected C subunit phosphotransferase activity (data not shown). Likewise, in transfections including PKI␣ or PKI␥, a sufficient amount of PKI expression vector was included so as to fully inhibit transfected cAMP-dependent kinase activity (data not shown). Following transfections, CV-1 cells were stimulated for 40 min with forskolin and isobutylmethylxanthine and then analyzed by indirect immunofluorescence with the M2 anti-Flag monoclonal antibody. Consistent with previous results, the Flagtagged C subunit was localized to the cytoplasm in untreated cells expressing Flag-tagged C subunit and RII␣ (Fig. 8A) (48). When identically transfected cells were treated with forskolin, C subunit was found to accumulate in the nucleus (Fig. 8B) (48). Co-transfection of PKI␣ or PKI␥ expression vectors with the Flag-tagged C subunit and RII␣ had no discernible effect on the cytoplasmic localization of C subunit in non-stimulated cells (Fig. 8, C and E). However, both PKI␣ and PKI␥ prevented the nuclear accumulation of C subunit in stimulated cells (Fig.   8, D and F) (49). Similar results were obtained in identically treated COS-1 cells (data not shown). In both cells lines, no significant anti-Flag immunofluorescence was observed in cells transfected with the parental vector alone (data not shown).

DISCUSSION
This study describes the identification, expression, and characterization of a new member of the PKI family of inhibitor proteins. The PKI family now includes at least three members: PKI␣, PKI␤, and PKI␥. The amino acid sequence of PKI␥ is 35% identical to PKI␣ and 30% identical to PKI␤1 suggesting that these three PKI isoforms are encoded by distinct genes. All members of the family possess an amino-terminal inhibitory region that includes a pseudosubstrate sequence and a central region containing a leucine and hydrophobic amino acid-rich NES. Most of the amino acid differences between the PKI isoforms occur in the carboxyl-terminal one-third of the molecule. All three PKI isoforms are approximately the same amino acid length suggesting that this size is important for physiological function.
The high degree of amino acid similarity between PKI␥ and the other known PKI isoforms in the amino-terminal inhibitory region suggests that the PKI isoforms may also have common biochemical characteristics. For instance, we anticipate that PKI␥ will require Mg 2ϩ and ATP for high affinity interactions with C subunit (47), that it will inhibit both C␣ and C␤ to similar extents (8), and that it will be specific for PKA (50,51). To understand the role of the PKIs in the PKA signal transduction system, it will be important to understand the differences between the PKI isoforms. Even though PKI␣ and PKI␤1 share many of the same amino acids in their inhibitory region, they differ significantly in their inhibitory potencies. As determined by Henderson analysis, murine PKI␤1 possesses a 32fold higher K i for C␣ than murine PKI␣ (16). This difference has been demonstrated to be due in part to the absence of a tyrosine at position 7 of PKI␤1 (16). Because Tyr 7 of PKI␣ is conserved in PKI␥, we anticipated that PKI␥ would bind and inhibit C␣ with a subnanomolar K i . Data from this study demonstrate that indeed PKI␥ is a tight binding inhibitor of C subunit with a K i of 0.44 nM. Still, it is 6-fold less potent than PKI␣. This decreased potency is probably due to potential negative determinants such as Ser 8 and Ser 12 . It is unlikely that a 6-fold difference in inhibitory potency could make a FIG. 7. Inhibition of luciferase gene transcription by PKI␥ in vivo. NIH 3T3 cells were transiently co-transfected with 1 g of an expression vector for either C␣ (E) or C␣Y235S/F239S (q) and the indicated amounts of a PKI␥ expression vector. All plates received 1 g of a cAMP-responsive reporter construct (HCG.luciferase) and 5 g of pRSV.␤gal to normalize for transfection efficiency. 21 h after removal of precipitates cells were harvested and assayed for luciferase and ␤-galactosidase activities. Luciferase activity was corrected for transfection efficiency by dividing by ␤-galactosidase activity and expressed as the percentage of relative light units (RLU) in the absence of the inhibitor. The error bars depict the standard deviation from the mean. This experiment was repeated three times, and a representative experiment is shown above. significant difference in a stimulated cell where free C subunit levels are micromolar; however, at resting cellular levels of cAMP, where there is little active kinase in the cell, this difference in inhibitory potency between PKI isoforms could be significant (16,17).
Unlike the other PKI isoforms, which do not contain any cysteines, PKI␥ possesses a single cysteine residue at amino acid position 13. Assuming that PKI␥ has an overall structure similar to PKI␣ (24), Cys 13 of PKI␥ would be located at the boundary of the amino-terminal ␣-helix and ␤-turn regions, two regions implicated in the high binding affinity of PKI␣ for C subunit. Previously, the ␤-turn region of PKI␣ has been hypothesized to be important in the proper positioning of Arg 15 with Glu 203 of the C subunit (24,32). Since the ␤-turn region of PKI␣ is believed to be flexible and not in a fixed conformation until binding to C subunit (24), it was possible that modification of this cysteine residue could affect the ability of PKI␥ to bind and inhibit C subunit. However, attempts at selective modification of Cys 13 with selective sulfhydryl-modifying reagents such as N-ethylmaleimide, iodoacetic acid, and iodoacetamide failed to show specific decreases in PKI␥ inhibitory activity (data not shown).
The original goal of this study was to identify novel PKI isoforms from tissues showing significant levels of PKI activity but low levels of PKI␣ and PKI␤ mRNA. Due to the relatively tissue-specific localization of PKI␣ and PKI␤ mRNAs, previous models of PKI function had assumed that some tissues and cell types did not require PKI for cAMP-mediated signal transduction. Assuming equivalent translation rates, results from Northern blot analysis and RNase protection analysis indicate that PKI␥ is widely expressed and may be the predominant PKI isoform in several tissues, including kidney and liver. In addition, determination of PKI isoform message levels in cultured cells showed that PKI␥ was expressed in all cell lines studied, and it was the major PKI transcript in all cell lines tested. These cell lines could afford the opportunity to examine the role of PKI␥ in the regulation of cAMP signaling.
PKI activity was first reported in 1965 as a heat-stable, trypsin-labile component of rabbit skeletal muscle extracts (52). The original observation that the inhibitory activity was stable to heat and acid treatment was used to devise a purification scheme (53). Rabbit skeletal muscle extracts were heated and loaded onto a DEAE column. PKI activity was eluted from this column using 0.25 M sodium acetate. Subsequently, this procedure was used to isolate PKI activity from a wide variety of tissues from many species. Using the results of these purifications and tissue purifications of C subunit, the level of PKI relative to C subunit in different tissues was calculated (54). For example, it was previously estimated that PKI activity in rat heart was sufficient to inhibit approximately 20% of the total heart C subunit. Similar results were obtained in other tissues with most tissues estimated to have significantly less total inhibitor than total C subunit. Importantly, this purification scheme limited the estimate of inhibitor activity from various tissues to those proteins that had the same chromatographic properties on DEAE-cellulose as the original material isolated from skeletal muscle.
To verify the existence of PKI␥ protein in mouse tissues, PKI activities from mouse heart were partially purified and identified in this study. Two inhibitory peaks were resolved following heat denaturation, acid treatment, and DEAE chromatography of the heart extracts. PKI␣ and PKI␥ were identified as the proteins responsible for inhibitory peak 1 and peak 2, respectively. It is likely that the previous purification procedure used to estimate PKI activity in tissues did not detect PKI␥ due to its higher affinity for DEAE-cellulose than the other PKI isoforms.
Interestingly, in the purification of PKI activity from rat testis, the yield of inhibitory activity after DEAE-cellulose chromatography is significantly lower than after other steps in the purification. This loss of PKI activity could reflect a selective loss of the PKI␥ isoform (37). Since the PKI␥ mRNA is abundant and widely expressed, previous studies significantly underestimated the PKI activities in several tissues. Recent results further challenge the belief that all cells have less total PKI than total C subunit. In situ hybridization analysis of mouse brain suggests that there is considerable heterogeneity of both PKI␣ and PKI␤ mRNA among different mouse brain regions (31). If this heterogeneity is also true of the PKI protein, then several regions of brain may have sufficient PKI to inhibit all of the C subunit present (i.e. cerebellar Purkinje cells and CA2 hippocampal neurons).
Results from this paper demonstrate that mouse heart contains significant protein levels of at least two PKI isoforms, PKI␣ and PKI␥. These results do not rule out the possibility of other uncharacterized PKIs, which may not bind DEAE-cellulose, or may not elute over the 5-1000 mM NaOAc gradient used in this study. Furthermore, multiple peaks of PKI activity have been detected by DEAE-cellulose chromatography of testicular extracts (19). Even though the heat stability of the known PKI isoforms has greatly aided in their purification, there is no a priori reason why all PKI isoforms should be heat stable.
During the course of these experiments the full-length PKI␥ cDNA nucleotide sequence was used to search the NCBI Gen-Bank data base. Significant sequence homology (p Ͻ 0.001) was detected between the 3Ј end of the mouse PKI␥ gene and the 3Ј end of the mouse adenosine deaminase (Ada) gene (55). Direct comparison of the two cDNA sequences demonstrated that the two genes overlapped in a tail-to-tail orientation with their coding sequences on opposite strands (Fig. 1), an uncommon occurrence in mammalian genomes. Adenosine deaminase is an important enzyme in purine metabolism, and adenosine deaminase deficiency is a major cause of autosomal recessive severe combined immune deficiency disease (56). Although tightly linked, there is no obvious functional relationship between PKI␥ and adenosine deaminase. To our knowledge no patients with deletions in the 3Ј end of the Ada gene have been identified. The mouse Ada gene has been localized to mouse chromosome 2 approximately 94 centimorgans from the centromere (57). The mouse PKI␤ gene (Prkacn2) has been localized to mouse chromosome 10 (58), and the mouse PKI␣ gene has not been mapped. Interestingly, the PKI␥ mRNA transcript was first observed during the characterization of the mouse Ada gene as an abundant, widely expressed 1.3-kb transcript transcribed from the antisense DNA strand that was co-amplified with the Ada gene during gene amplification (55). The low levels of amino acid identity and the distinct chromosomal localizations of the known PKI isoforms suggest that the members of the PKI family are distantly related.
PKI␥ mRNA is widely expressed and abundant in mammalian tissues but has not been characterized previously due to its low sequence homology with the other known PKI isoforms and its high affinity for DEAE-cellulose. The identification of a widely expressed, abundant PKI isoform suggests that PKI activity may be more critical to general cell function than previously believed. PKI␥ is differentially expressed in adult tissues and shows distinct interactions with C subunit, suggesting it may serve distinct roles. The availability of purification methods and antibodies capable of differentiating between PKI␥ and other PKI isoforms should help clarify their specific roles. Future studies should determine what, if any, specific functions the non-conserved carboxyl-terminal region of these proteins play. Verification of these specific roles will require the development of model systems selectively deficient in the expression of the three PKI genes.