Multiplicity of the β Form of the cAMP-dependent Protein Kinase Inhibitor Protein Generated by Post-translational Modification and Alternate Translational Initiation*

Two distinct species of the thermostable inhibitor of the cAMP-dependent protein kinase, PKIα and PKIβ, exist that are the products of separate genes. The PKIβ form, as first isolated from rat testis, is a 70-amino acid protein, but the genomic sequence suggested that an alternate form might exist, arising as a consequence of alternate translational initiation. This species, now termed PKIβ-78, has been synthesized by bacterial expression, demonstrated to be equipotent with PKIβ-70, and also now demonstrated to occur in vivo. By Western blot analyses, six additional species of PKIβ are also evident in tissues. Two of these represent the phospho forms of PKIβ-78 and PKIβ-70. The other four represent phospho and dephospho forms of two higher molecular mass PKIβ species. These latter forms are currently termed PKIβ-X and PKIβ-Y, awaiting the full elucidation of their molecular identity. In adult rat testis and cerebellum, PKIβ-70, PKIβ-X, and PKIβ-Y constitute 39, 23, and 32% and 15, 29, and 54% of the total tissue levels, respectively. In adult rat testis, 35–42% of each of these three species is present as a monophospho form, whereas no phosphorylation of them is evident in cerebellum. PKIβ-78 is present at much lower levels in both rat testis and cerebellum (∼6 and 2% of the total, respectively) and almost entirely as a monophospho species. PKIβ-78, like PKIβ-70, is a high affinity and specific inhibitor of the cAMP-dependent protein kinase. PKIβ-Y and PKIβ-X, in contrast, also significantly inhibit the cGMP-dependent protein kinase.

The cAMP-dependent protein kinase (PKA) 1 plays a central role in the signal transduction of many hormones. The thermostable protein kinase inhibitor (PKI) is a highly specific competitive inhibitor of this enzyme that binds with high affinity to the protein substrate-binding site of the kinase (1). Its role as a regulator of PKA remains to be fully identified and appears to include not only the reduction of protein kinase activity, but also the trafficking of the protein kinase catalytic subunit between subcellular sites (2)(3)(4). Two genetically distinct forms of PKI (␣ and ␤) have been shown to be present in many tissues (5)(6)(7)(8)(9). Both forms act as specific pseudosubstrate inhibitors of PKA, have similar K i values in the subnanomolar range, and are likely both involved in intracellular trafficking (4). PKI␣ and PKI␤ share ϳ40% identity at the amino acid level (5), share very closely the same recognition determinants for the catalytic subunit of PKA (5,10) and for nuclear export (9), but are notably distinct in the C-terminal half of the molecule. PKI␣ and PKI␤ each have a distinct tissue distribution (6).
The PKI␤ form was first identified in rat testis (5) in a follow-up of observations made earlier by Means and co-workers (11,12). The PKI␤ form purified was one of what appeared to be multiple forms of the protein present in testis, as was apparent from the DEAE chromatographic profile of the partially purified protein. The form purified was shown by protein sequencing and amino acid analysis to be a 70-amino acid protein with a blocked amino terminus (5). Identification of the gene for this protein indicated, however, that in addition to the appropriate in-frame ATG codon, a second upstream in-frame ATG codon was present. If translated, this alternate in-frame ATG codon would give rise to an alternate translational product with a 7-or 8-residue amino-terminal extension with the additional sequence of MRTDSSEM-, with the specific size dependent upon whether or not the N-terminal methionine was processed. Translational initiation in eukaryotes is affected by the context of the nucleotides surrounding the AUG codon of the mRNA (13). The first and second in-frame AUG initiation codons in the PKI␤ gene have the nucleotide sequences Ϫ3 GU-GAUGA ϩ4 and Ϫ3 GAGAUGA ϩ4 . With a G at position Ϫ3 and an A at position ϩ4, both of these in-frame AUG codons have initiation sequences that would readily allow translation to be initiated at either site (13). Neither, however, has the optimum Kozak sequence ( Ϫ3 (A/G)CCAUGG ϩ4 ), and so read-through of the first upstream AUG codon, to allow initiation of translation at the second, appeared to be a significant possibility (see Fig.  1). The sequence of the amino-terminal amino acids of the isolated 70-amino acid PKI␤ is consistent with its being directly synthesized and then processed to remove the terminal methionine and to add a blocking group (14 -16). In contrast, if synthesized from the first in-frame AUG codon, the aminoterminal sequence context is such that processing of its methionine would be much less likely to occur. This paper addresses the existence of this alternate translational product and also the other forms of PKI␤ apparent from the DEAE profile. In total, it is now demonstrated that there exist at least eight isoforms of PKI␤, and in the accompanying paper (17), a ninth form has also been elucidated. The full nomenclature for these different species is developed subsequently in this paper. The initial 70-amino acid form isolated from rat testis, and originally termed PKI␤, is now more precisely termed PKI␤-70. The form that would be obtained by processing at the first upstream ATG site is termed PKI␤-78.

EXPERIMENTAL PROCEDURES
Bacterial Expression of PKI␤-70 and PKI␤-78 -Each of the coding sequences for PKI␤-70 and PKI␤-78 were subcloned into the Escherichia coli expression vector pET5a by engineering NdeI and BglII sites at the 5Ј-and 3Ј-ends of their coding sequences, respectively, using the polymerase chain reaction. The vectors were then transfected into the BL21(DE3) strain of E. coli, and proteins were expressed as described by Studier et al. (18). Each of these proteins was purified to homogeneity essentially as described by Van Patten et al. (5). In brief, for the purification of expressed PKI␤-78, bacteria from 3 liters of suspension (A 260 ϭ 0.8) were collected by centrifugation; lysed with a French press; and suspended in 25 ml of 10 mM Tris-Cl, pH 7.5, 10 mM NaCl, and 1 mM EDTA. The supernatant obtained following centrifugation was heated at 92°C for 10 min and then centrifuged. The supernatant from this heat treatment was adjusted to pH 5.0 and applied to a DEAE ionexchange column using the conditions for chromatography and elution as described by Van Patten et al. (5). Pooled fractions with PKI activity were subsequently purified to homogeneity by HPLC employing a Nova-Pak reverse-phase C 18 column and elution with an acetonitrile gradient in trifluoroacetic acid as described by Van Patten et al. (5). Purification of expressed PKI␤-70 was accomplished exactly as described for expressed PKI␤-78, except that the DEAE column was at pH 6.0, and the final HPLC was replaced by gel filtration using a Superdex 75 HPLC column and elution with 0.3 M ammonium acetate, pH 7.0. The two protein products were characterized by amino acid analysis and by determination of the sequence of at least the first 10 aminoterminal residues. The bacterial expression system for a third putative form of PKI␤, termed ␤2 by Scarpetta and Uhler (8) and containing the sequence of PKI␤-78 plus an additional 14-amino acid amino-terminal extension, was also created, and the expressed protein was purified by similar procedures; however, no evidence for its existence in rat tissue was observed in these studies.
Preparation of Tissue Extracts for Electrophoresis-For Western blot analyses of PKI isoforms, the tissues from adult Harlan Sprague Dawley rats were freeze-clamped with Wollenberger clamps precooled in liquid nitrogen immediately after dissection. The frozen tissue was powdered and then rapidly homogenized with a glass-Teflon homogenizer in 1 mM EDTA, pH 7.0 (1 ml/g of tissue), containing a mixture of protease inhibitors (0.5 mM (2-aminoethyl)benzenesulfonyl fluoride, 1 M leupeptin, 2 mM benzamidine, 0.1 mM L-1-tosylamido-2-phenylethyl chloromethyl ketone, and 20 milliunits/ml aprotinin). The extracts were heat-treated to 92°C for 10 min and spun at high speed in a microcentrifuge at 4°C for 2 min, and the supernatants were used for gel electrophoresis. Alkaline phosphatase treatment, where noted, was accomplished by incubating the tissue extract with 5 units/l alkaline phosphatase (calf intestinal alkaline phosphatase, Pharmacia Biotech Inc.) in 50 mM Tris-Cl, pH 8.0, and 0.125 mM EDTA plus the same mixture of protease inhibitors at 1:4 dilution for 2 h at 30°C. After treatment, the samples were directly applied to the isoelectric focusing tube gels.
One-and Two-dimensional Electrophoresis and Procedures for Western Blot Analyses of PKI Isoforms-Tissue samples for PKI isoform analyses were resolved by one-or two-dimensional gel electrophoresis, as noted. One-dimensional SDS-PAGE was in accord with Laemmli (19) using 12-18% gradient gels in 0.1% SDS and 0.375 mM Tris-Cl, pH 8.8. The first dimension for two-dimensional electrophoresis utilized 1.5 mm x 7.5 cm tube gels (Hoeffer Scientific Instruments) and 2% pH 4 -6 ampholytes (BDH), with anode and cathode buffers of 0.085% phosphoric acid and 20 mM sodium hydroxide, respectively. Following the first dimension, gels were equilibrated for 15 min in 0.125 M Tris-Cl, pH 6.8, 10% glycerol, 4.9 mM dithiothreitol, and 2% SDS and then applied to a 12-18% SDS-polyacrylamide gel. For Western blot analyses, following electrophoresis, the gels were equilibrated in 10 mM CAPS, pH 11.0, for 30 min, and the proteins then electrophoretically transferred to polyvinylidene difluoride membranes (Gelman Instrument Co.) for 1.5 h at 25 V. The membranes were incubated for 1 h in blocking buffer consisting of 0.5% casein and 0.2% Tween 20 in Tris-buffered saline (150 mM NaCl and 10 mM Tris-Cl, pH 7.4). The polyvinylidene difluoride membrane was incubated overnight with the designated primary antipeptide antibody (1:5000 dilution in blocking buffer) and then washed 1 ϫ 1 min and 3 ϫ 10 min with Tris-buffered saline containing 0.1% casein and 0.2% Tween 20. The membranes were then incubated for 2 h with biotinylated anti-guinea pig donkey IgG antibody (1:2500 dilution in blocking buffer); washed as before; incubated for 30 min with alkaline phosphatase-conjugated streptavidin; washed sequentially with Tris-buffered saline containing 0.1% casein and 0.2% Tween 20 (1 ϫ 1 min and 2 ϫ 10 min) and with 0.1 M Tris-Cl, pH 9.0, 100 mM NaCl, and 5 mM MgCl 2 (5 min); and then developed with the alkaline phosphatase nitro blue tetrazolium/5-bromo-4-chloro-3-indolyl phosphate substrate kit (Pierce 34041). For the SDS gels, molecular weight standards were as denoted in Fig. 2. The relative isoform composition was determined by scanning the Western blot two-dimensional gels and integrating the signals using an IS-1000 digital imaging system (Innotech Scientific Corp.).
Assay of PKI-Routine assays of PKI inhibitory activity were as described previously (22) using histone II-A as protein substrate and the PKA C␣ subunit obtained by bacterial expression (23).

Characterization of Bacterially Expressed PKI␤-78 and
PKI␤-70 -Both PKI␤-78 and PKI␤-70 were readily obtained by the creation of suitable bacterial expression vectors and were each purified to homogeneity. Their identity was established by amino acid analyses and amino-terminal sequencing. Their purity was established by SDS gel electrophoresis. We have previously demonstrated that, like the original PKI␣, PKI␤-70, as directly isolated from rat testis, is a specific inhibitor of PKA (5). The expressed PKI␤-78 and also the bacterially expressed PKI␤-70 were likewise specific high affinity inhibitors of PKA and did not inhibit even the highly homologous cGMP-dependent protein kinase (Fig. 1). K i values for all of these forms, as determined by Henderson analysis (24), were indistinguishable (expressed PKI␤-78, 0.14 nM; expressed PKI␤-70, 0.1 nM; and rat testis PKI␤-70, 0.14 nM). Each were competitive inhibitors versus protein substrate.
Identification of PKI␤-78 and Other Multiple Forms of PKI␤ in Tissues-The availability of a purified preparation of PKI␤-78 as a standard allowed us to address the issue of whether this possible alternate translational initiation form of PKI␤ occurred physiologically. Other forms of testis PKI also appeared to be a likelihood as evidenced by the earlier characterization of PKI by Mean and co-workers and antibody identification of multiple size forms (11,12,27). To make an evaluation of the physiological presence of PKI␤-78 and to expand the context of investigation to an assessment of the potential existence of other PKI␤ forms, specific anti-peptide antibodies were prepared against the peptides PKI␣-(5-22) (TTYADFIAS-GRTGRRNAI), PKI␤-(5-22) (SVISSFASSARAGRRNAL), and PKI␤-(60 -70) (GQPKKPLDEDK). The nomenclature for these peptides, as used previously (5,10), is based upon the sequence of the full PKI proteins. The first two of these peptides contain the pseudosubstrate regions of PKI␣ and PKI␤, respectively, and whereas some degree of overlapping antigenic specificity might have been expected between them, as shown below, it was not observed. The third peptide is from a region of PKI␤ in which there is absolutely no homology to PKI␣.
Western blot analyses with these anti-peptide antibodies revealed a complex pattern of PKI isoforms. In both testis and cerebellum, the prominent if not exclusive form of PKI␣ detected is the 75-amino acid species of PKI that is the form of inhibitor protein originally and extensively characterized from skeletal muscle and other tissues (1). This form is readily detected by anti-PKI␣-(5-22) antisera, but not by anti-PKI␤-(5-22), anti-PKI␤-(60 -70), or control sera (Fig. 2). This result demonstrates the specificity of the anti-PKI␣ peptide antisera. In cerebellum (Fig. 2), as has also been seen in some other tissues (data not shown), a small amount of a slightly lower molecular mass PKI␣ species is also evident. This latter form may be related to the PKI-IЈ species that has been previously described (28,29). Western blot analyses of a variety of rat tissues showed the expected profile of distribution of PKI␣, with a high predominance in brain and various muscle types (data not shown). These data were as anticipated from previous studies based upon mRNA distribution (6).
The pattern with PKI␤ is considerably more complex than is observed for PKI␣. Western blot analyses and one-dimensional SDS-PAGE showed the presence of at least three major PKI␤ size forms of ϳ7, 14, and 15.5 kDa in both testis and cerebellum (Fig. 2). These three forms were detected with both the anti-PKI␤-(5-22) and anti-PKI␤-(60 -70) antisera, indicating that all three contained, as a minimum, the homologous amino acid sequences matching both the amino-and carboxyl-terminal sequences of PKI␤-70. These forms were not detected with either control or anti-PKI␣-(5-22) antisera ( Fig. 2), indicating the specificity of the anti-PKI␤ antibodies, and this set of data demonstrates the absence of cross-reactivity of either the PKI␣ forms with the PKI␤ antibodies or vice versa. The sizes of these three forms identified here are closely similar to those of the testis PKI species previously identified by Means and co-workers (11,27). A fourth low abundance form of PKI␤, of slightly higher apparent molecular mass than the 7-kDa form, was also evident on the one-dimensional gels from testis and cerebellum (Fig. 2).
Western blot analyses of two-dimensional gels (first dimen-sion, isoelectric focusing; and second dimension, SDS-PAGE) showed that the pattern of PKI␤ isoforms is even more complex. In testis extracts, at least eight forms of PKI␤ are evident, with each of the three predominant forms and the fourth less abundant species, as separated based upon size ( Fig. 2), being composed of a pair of charge isoforms (Fig. 3a). In cerebellum, only the more cationic species of each pair was evident, except for the low abundance form that was present in both tissues as predominantly the more anionic species (Fig. 3b). There was a good correspondence between the gel migration of the cationic form of the three most abundant species in testis extracts with their counterparts in cerebellum, suggesting that these forms were likely identical in both two tissues. The diversity of forms in both testis and cerebellum was evident with both the anti-PKI␤-(5-22) and anti-PKI␤-(60 -70) antisera (Fig. 3, c and d) with a matching intensity of the spots. This is indicative that all of the species identified in testis and cerebellum contain key components (if not all) of the sequence that has been established for PKI␤-70. The nomenclature for these different forms, as used in this report, is presented in Fig. 4a. The basis for this nomenclature is as follows. The spot specified as PKI␤-70 (Fig. 4a) is identified as such based upon its comigration with exogenously added purified PKI␤-70 as isolated directly from rat testis (Fig. 3, e and f). This experiment compares the migration of the form in testis extracts with PKI␤-70 purified to homogeneity from rat testis (5). Bacterially expressed PKI␤-70 migrates slightly cationic to this form due to the presence of one additional charge contributed by the free (unblocked) amino terminus (data not shown).
The spot specified as PKI␤-78 (Fig. 4a) is identified as such based upon its comigration with exogenously added, bacterially  (29 -35) as substrate) were performed in the presence of the indicated concentrations of PKI isoforms diluted in 0.5 mg/ml bovine serum albumin in MES, pH 6.8. q and E, bacterially expressed PKI␤-78; f and Ⅺ, bacterially expressed PKI␤-70; å, PKI␤-70 purified from rat testis. All protein preparations were homogeneous. PKA was assayed as defined under "Experimental Procedures." Bovine lung PKG-1␣ was assayed as described previously (26). expressed, purified PKI␤-78 (Fig. 3, g and h). Given this comigration, it would appear most likely that during the in vivo synthesis of PKI␤-78, there is no processing of its amino terminus by the cleavage of the amino-terminal methionine and the addition of an amino-terminal blocking group. The absence of any processing of the amino terminus of PKI␤-78 is as might be predicted from its amino-terminal sequence (MRTD . . . ), which is unfavorable for such amino-terminal processing (15,16). The location of PKI␤-78 on the two-dimensional gel, in comparison with that of PKI␤-70, is consistent with the difference between them. PKI␤-78 is 8 amino acids greater in size and differs in charge by ϩ1. These data demonstrate that alternate translational initiation does occur under physiological conditions and that the PKI␤ transcript can give rise to both PKI␤-78 and PKI␤-70. The level of PKI␤-78 (dephospho) seen in tissues has varied from tissue preparation to preparation and has often been quite low or not detectable. The relative degree of alternate translational initiation is discussed subsequently in this paper following discussion of its phosphorylated form.
Two spots on the two-dimensional Western blot are designated as PKI␤-X and PKI␤-Y (Fig. 4a). These two forms of PKI␤ are detected by both the anti-PKI␤-(5-22) and anti-PKI␤-(60 -70) antisera and are therefore homologous to PKI␤-70 in at least these two regions. As evidenced below, these forms also exhibit high affinity PKA inhibitory activity. Based upon their degree of migration on SDS-PAGE, these two forms are apparently of higher molecular mass than either PKI␤-70 or PKI␤-78, although in this molecular mass range, such determinations based upon SDS-PAGE migration are not particularly reliable. Scarpetta and Uhler (8) have suggested the possible occurrence of a higher molecular mass form of PKI␤ derived by alternate RNA splicing that is identical to PKI␤-78, but with a 14-amino acid amino-terminal extension. We have synthesized this form by bacterial expression. On one-dimensional gels, this form migrates quite distinctively from either PKI␤-X or PKI␤-Y at a position intermediate between PKI␤-78 and PKI␤-X. PKI␤-X and/or PKI␤-Y is therefore not this species, except for the possibility that it might be derivatized with a significantly sized covalent modification.
In testis extracts, PKI␤-70, PKI␤-78, PKI␤-X, and PKI␤-Y each have a more anionic companion species. Treatment of the testis extract with alkaline phosphatase prompted a marked reduction in the anionic companion species of PKI␤-70, PKI␤-X, and PKI␤-Y, accompanied by an increased intensity of the more cationic forms (Fig. 5, upper panels). A similar result was also obtained using protein phosphatase 2A (data not shown). These data provide evidence that these anionic forms represent phosphorylated species, i.e. phospho-PKI␤-70, phospho-PKI␤-X, and phospho-PKI␤-Y, respectively. The anionic companion form of PKI␤-78 was notably less sensitive to alkaline phosphatase treatment (Fig. 5, upper right panel). To evaluate its possible identity, bacterially expressed purified PKI␤-78 was incubated with cyclin-dependent kinase (Cdk1; gift of Dr. Richard Vulliet, Department of Molecular Biosciences, University of California, Davis, CA), given the presence of a Ser-Pro sequence within the PKI␤ protein. 32 P-Labeled PKI␤-78, as produced by such a reaction, comigrates with the spot termed phospho-PKI␤-78 on the two-dimensional Western blot (Fig. 5, lower panels), thus providing evidence that PKI␤-78 likely also exists in phospho and dephospho forms. The phosphorylation of either PKI␤-70 or PKI␤-78 by Cdk1 was only achieved to a modest degree (ϳ2% stoichiometry) and with a very high K m value (Ͼ100 M). It is thus quite unlikely that under physiological conditions, Cdk1 catalyzes this PKI␤ phosphorylation, but other Ser-Pro-directed kinases, of which there are now many, might be the responsible in vivo catalyst. The 32 P-labeled PKI␤-78 species generated with Cdk1 and as used in the experiment presented in Fig. 5 is, however, likely to be a reliable marker to indicate the migration of a species of PKI␤-78 that contains 1 mol of phosphate (even though the phosphate incorporated may not necessarily be in the same site as in the native testis protein). A protein kinase capable of phosphorylating PKI␤-78 has been identified in male germ cells. 2  PKI␤-70 was also obtained by incubation with Cdk1, which likewise produced the characteristic change in isoelectric focusing migration (data not shown).
Quantification of PKI␤ Isoform Composition in Adult Rat Testis and Cerebellum-The relative distribution of the PKI␤ isoform in adult rat testis and cerebellum was determined from scans of the two-dimensional Western blots of adult rat testis and cerebellum extracts. A range of protein amounts (0.7, 1.2, 1.7, 2.3, and 3.5 g) was applied to the two-dimensional gels to validate that the signals for each of the PKI␤ isoforms were in the linear range of response. This allowed for an accurate determination of the relative percentage composition of the different forms. The results of quantification by this procedure are presented in Fig. 4 (b and c). As is evidenced, in testis, PKI␤-70, PKI␤-X, and PKI␤-Y constitute ϳ39, 23, and 32% of the total, respectively, each with a very similar proportion (35-42%) of monophospho versus dephospho forms. Rat cerebellum contains only the dephospho forms of PKI␤-70, PKI␤-X, and PKI␤-Y, with markedly less PKI␤-70 than in testis and with PKI␤-Y in the highest abundance. PKI␤-78 is present only in low amounts in either testis or cerebellum (ϳ2-6% of the total) and almost entirely as the monophospho species. The difference in the relative percentage of phosphorylated to nonphosphorylated PKI␤-78 in comparison with what is seen for the other PKI␤ species may in some manner be related to the differences in sensitivity that were observed with alkaline phosphatase treatment; the phospho-PKI␤-78 form was observed to be markedly more resistant to hydrolysis (see Fig. 5). Such a difference might arise, as an example, if the site of phosphorylation in PKI␤-78 was different than in the other phospho-PKI␤ forms.
Protein Kinase Inhibitory Activity of the Higher Molecular Mass Forms of PKI␤-The higher molecular mass PKI␤-X and PKI␤-Y species have been identified based upon their crossreactivity with both anti-PKI␤-(5-22) and anti-PKI␤-(60 -70) antisera. Each exhibits an antigenicity with the two antibodies very similar to that obtained with both PKI␤-70 and PKI␤-78. The PKI␤-(5-22) peptide, against which the first of these two antibodies was prepared, contains the pseudosubstrate PKA inhibitory site, and the peptide itself is nearly as equipotent a PKA inhibitor as the full-length protein (5). Given the observed antigenicity, it might be anticipated that PKI␤-X and PKI␤-Y would exhibit PKA inhibitory activity. This was documented by the further purification of PKI␤-X and PKI␤-Y by a combina- FIG. 4. Nomenclature of the PKI␤ isoforms and determined percentage composition of isoforms in adult rat testis and cerebellum. a, two-dimensional Western blot of rat testis PKI␤ isoforms depicting the nomenclature used; b and c, quantification of the percentage levels of PKI␤ isoforms in adult rat testis and cerebellum, respectively. As described under "Experimental Procedures," the level of staining for each isoform was quantified by densitometry and determined for a sequence of protein amounts applied to the two-dimensional gels so as to ensure that the staining was in the linear range. Total bar graph height depicts the amount of each isoform type composed of the phospho (shaded) and dephospho (black) forms. tion of DEAE chromatography, gel filtration, and affinity chromatography.
Rat cerebellum contains high levels of both PKI␣ and PKI␤. These are readily separated by DEAE chromatography (Fig.  6a), with first the elution of multiple forms of PKI␤ (1-3millisiemens conductivity), followed by the subsequent elution of PKI␣ (5-6-millisiemens conductivity). The multiplicity of PKI␤ forms was examined by Western blot analyses using both anti-PKI␤-(5-22) and anti-PKI␣-(5-22) antibodies (Fig. 6, b  and c). No antigenic cross-reactivity between the PKI␣ and PKI␤ forms eluted from the DEAE column was detected (data not shown). Three overlapping peaks in the 1-3-millisiemens conductivity range of PKI␤ were observed for the elution of rat cerebellum protein from the DEAE column. The first (fractions 12-14) contains predominantly PKI␤-70, the second (fractions 16 -18) contains PKI␤-X plus the lower levels of PKI␤-78, and the third (fractions 18 -21) contains PKI␤-Y. Fractions 21-23 also show the presence of a low abundance species of slightly higher apparent molecular mass than PKI␤-Y. This is likely to be "PKI␤-Z," a form that is a more prominent constituent of epididymal sperm and is discussed further in the accompanying paper (17). PKI␣ is eluted from the DEAE column at markedly higher conductivities (fractions 50 -67) and well separated from the PKI␤ forms (Fig. 6, a and c). Multiple charge forms of PKI␣ were evident in DEAE separation, as has been described previously for skeletal muscle and brain PKI␣ (29,30). The major forms of cerebellum PKI␣ comigrated with purified skeletal muscle PKI␣ (data not shown). A lower molecular mass species that is evident in later eluting fractions on the Western blots is likely identical to the skeletal muscle PKI␣-IЈ form described previously (28,29).
The DEAE elution of PKI␤ forms from rat testis (Fig. 6d) showed a similar pattern to that from rat cerebellum, made more complicated, however, by the presence of the phospho forms of PKI␤-70, PKI␤-X, and PKI␤-Y. These latter species are extensively present in testis, but absent in cerebellum (see Fig. 4). Three prominent peaks of inhibitor activity are evident from the DEAE elution of the rat testis extract. The first (fractions 14 -16) contains almost exclusively dephospho-PKI␤-70 and was the fraction that had been previously used in the first elucidation of the PKI␤ form of inhibitor protein (5). The second peak (fractions 19 -21) contains PKI␤-78 plus dephospho-PKI␤-X. Fractions 23-27 contain phospho-PKI␤-70, phospho-PKI␤-X, and dephospho-PKI␤-Y. Phospho-PKI␤-Y is then eluted in the subsequent fractions, 30 -34. Identification of these different forms was confirmed by two-dimensional analyses (data not shown). Adult rat testis has only very low activity levels of PKI␣ in comparison with the amounts of PKI␤ (ϳ2%). Measurements in the region of higher conductivity were not undertaken for the experiment presented in d, but have been fully described previously (5). mS, millisiemens.
Each of the two DEAE profiles of elution of PKI␤ from cerebellum and testis (Fig. 6, a and d) shows a strong coincidence between PKA inhibitory activity and the elution of the higher molecular mass PKI␤ forms. Further confirmation that both PKI␤-X and PKI␤-Y are inhibitors of PKA was obtained by subsequent gel filtration. Pools I (fractions [12][13][14], II (fractions 16 -17), and III (fractions 19 -22), obtained by the DEAE chromatography of cerebellum extracts (Fig. 6a), were each lyophilized and then applied to a Sephacryl S-100 gel filtration column. This purification fully separated the higher and lower molecular mass species of PKI␤ (Fig. 7, a-d) and demonstrated that both PKI␤-X (pool II) and PKI␤-Y (pool III) were as effective inhibitors of PKA as the lower molecular mass forms. Equivalent results were also obtained by gel filtration of the testis PKI␤ fractions obtained by DEAE fractionation, again illustrating the PKA inhibitory activity of PKI␤-X and PKI␤-Y (data not shown).
The affinity-purified preparations of PKI␤-X and PKI␤-Y were tested for protein kinase inhibitory activity. As is illustrated in Fig. 8, both are efficacious inhibitors of PKA. Adequate amounts of protein have not yet been available for a full analysis of inhibitory potency. Estimates of protein based upon Coomassie Blue staining on SDS gels (Fig. 8, inset a) and antigenicity suggest, however, that PKI␤-X and PKI␤-Y have PKA inhibitory potencies very similar to those of PKI␤-70 and PKI␤-78. The concentration of PKI␤-X used in the experiment depicted in Fig. 8 was ϳ18% of that of PKI␤-Y, accounting for the difference in the profile. PKI␤-Y and PKI␤-X were also tested as inhibitors of the cGMP-dependent protein kinase. In total contrast to what has been observed with both PKI␤-70 and PKI␤-78 (see Fig. 1) and also with PKI␣ (26), none of which inhibit PKG even at very high concentrations, PKI␤-Y and PKI␤-X were quite effective inhibitors of the cGMP-dependent protein kinase (Fig. 8). Based upon the IC 50 values, the K i values for PKG inhibition by these two forms are in the 1-5 nM range. In an earlier report by Szmigielski et al. (32), two forms  Fig. 6a), concentrated by lyophilization, and then applied to a Sephacryl S-100 gel filtration column (2.5 ϫ 90 cm) equilibrated in 10 mM MES and 200 mM NaCl buffer, pH 6.8. Gel filtration fractions were assayed for PKA inhibitory activity (a), or 2 l of the indicated fractions were assessed by one-dimensional SDS-PAGE and Western blotting with anti-PKI␤-(5-22) antibodies (b-d). Vertical lines in a denote, in sequence, the elution points of the standard molecular weight markers carbonic anhydrase (29,000), lactalbumin (14,200), and cytochrome c (12,400). of protein kinase inhibitor were isolated from rat cerebellum, termed by those authors types I and II. Type I was specific for PKA, whereas type II inhibited both PKA and PKG. Based upon these inhibitor potencies and their reported gel filtration elution profile (32), it is now apparent that type I is PKI␣ and type II is a combination of PKI␤-Y and PKI␤-X. This is fully consistent with the DEAE profile of PKI distribution in cerebellum that is illustrated in Fig. 5a. The low level of PKI␤-70 observed in this profile would likely have not been detected in the separation applied by Szmigielski et al. (32). Thus, the PKG inhibition now observed for PKI␤-Y and PKI␤-X (Fig. 8) is consistent with the past literature. The inhibition of PKG by PKI␤-Y is of note given the identified pseudosubstrate sequence that this inhibitor possesses. Short PKI␣ peptides, unlike their parent protein, do inhibit PKG to a modest degree (26,33), and PKI␣-based substrate peptides are quite good substrates for this enzyme (34). Clearly, it is either the other component parts of this PKI␤-Y that must aid in enhancing its interaction with PKG or some component part of PKI␤-70 and PKI␣ prohibits PKG interaction. Either is of interest given that these two protein kinases are closely homologous. Despite being less efficacious with PKA than with PKG, PKI␤-Y and PKI␤-X may be effective inhibitors of PKG physiologically since the cellular concentrations may well exceed the association constant for the interaction. DISCUSSION The results presented here open up a new level of complexity in the already intricate nature of the cAMP signal transduction pathway. As has been well established, multiple forms of the protein constituents of PKA exist (35,36). The three primary species of the catalytic subunit (C␣, C␤, and C␥) are the products of separate genes, and each has a distinctive cellular and tissue distribution pattern. As yet, however, there is little evidence that these three species in vivo catalyze the phosphorylation of distinct proteins, except possibly some yet to be fully explored indications that the C␥ isoform might exhibit some differences in substrate preference (37). It is possible that the major reason for the existence of multiple PKA catalytic forms is in order for their transcriptional expression to be differentially controlled and, in doing so, to create the different cellular milieu of PKA activity in each cell. Much, however, remains to be determined about the physiological events of PKA-dependent phosphorylation. With the presence of multiple protein substrates in a single cell, there is a need for not only each substrate to be phosphorylated in response to a cAMP signal, but also for there to be some preference in the order of their phosphorylation and by this means to orchestrate the integration of the cellular response (38). Possibly the existence of multiple catalytic species contributes to such events. There are also multiple forms of the PKA regulatory subunit. The two primary types, R I and R II , differ in some aspects of their cAMP sensitivity and in the existence in R II of cellular targeting sequences (39,40). Why there is a need for there to be ␣ and ␤ species of each, however, remains to be identified (41). Possibly they too exist to allow for differential transcriptional expression in different cell types and/or to allow for the complexity of response to the cAMP signal that occurs with each cell. Multiple species of AKAP proteins have also been identified that are involved in distributing the cAMP signal throughout the cell and by doing so would allow for a differential response dependent upon subcellular localization of the kinase substrates (39,42).
We have now demonstrated that PKI also exhibits a multiplicity of species. Previously, we established that there are two gene products, PKI␣ and PKI␤ (6), and these current studies now show that PKI␤ exists in eight or more different forms as related by at least covalent modification and alternate translational initiation. The first evidence of multiple species of a testis-derived PKI in fact preceded these current studies. PKI from testis was initially described by Means and co-workers (12,43), and as judged from its DEAE-cellulose elution characteristics, the form isolated by them was likely one (or more) of the isoforms of PKI␤. Subsequently, these same investigators generated antibodies against this purified testis PKI and, by their use, identified four forms of the protein in testis extracts. The two most prominent species had molecular masses of 9.3 and 15.6 kDa (27), closely similar to what we now observe for PKI␤. Our current data are fully complementary to the initial reports of Means and co-workers.
One mode for the formation of different PKI␤ isoforms appears to be by alternate translational initiation, giving rise to at least PKI␤-78 and PKI␤-70. According to the Kozak scanning model of translation, a ribosome binds to the 5Ј-end of the mRNA and proceeds downstream until an AUG codon is encountered (13). Most often, translation is initiated at the first AUG codon, although the efficiency of translation is dependent upon the sequence context of this AUG codon. Where there are two or more in-frame AUG codons within a short distance of each other (Ͻ20 codons), alternate translational initiation at the second AUG codon has also been observed. The synthesis of multiple forms of brain creatine kinase (44) and the yeast FIG. 8. Inhibitory potencies of PKI␤-X and PKI␤-Y. PKI␤-X and PKI␤-Y were purified from rat cerebellum by DEAE chromatography (see Fig. 7), gel filtration, and then PKA catalytic subunit affinity chromatography using the conditions described previously (31), except for the use of the recombinant PKA catalytic subunit. These purified preparations of PKI␤-Y (closed symbols) and PKI␤-X (open symbols) were then tested as inhibitors of PKA (E and q; assayed as described under "Experimental Procedures" using 6.4 nM purified catalytic subunit with Kemptide as substrate), PKG (å and Ç; assayed as described (26) using 3.2 nM purified bovine lung PKG-1␣ with [Ala 34 ]H 2 B-(29 -35) as substrate), and phosphorylase kinase (f and [itri]; assayed as described (46) using 0.42 nM purified skeletal muscle phosphorylase kinase with phosphorylase as substrate). The protein concentrations of the final PKI␤-X and PKI␤-Y solutions obtained were too low for accurate analysis, but by Coomassie Blue staining and antigenicity, they were estimated to be ϳ0.06 and 0.32 g/l, respectively. The x axis denotes the microliter equivalents of these solutions added to an 80-l assay. Inset a, Coomassie Blue-stained gel of the PKI␤-Y preparation; inset b, Western blot analysis of the PKI␤-X and PKI␤-Y preparations. protein MOD5 (45) are two examples where translation is initiated at both the first and second AUG codons within a short stretch of mRNA. It appears most likely that PKI␤-70 and PKI␤-78 arise by such an alternate initiation of translation. Both of the AUG codons in PKI␤, which would give rise to these two forms of the protein, have initiation sequences that would readily allow translation to be initiated at either site (13). Neither, however, has the optimum Kozak sequence, and so read-through of the first upstream AUG codon to allow initiation at the second is thus likely. Although it remains a possibility that PKI␤-70 could arise instead as a consequence of the proteolytic cleavage of PKI␤-78, this seems unlikely. The amino-terminal sequence of PKI␤-70 is consistent with its being directly synthesized and then processed to remove the terminal methionine and to add a blocking group (14 -16). In both testis and cerebellum, PKI␤-70 (as both phospho and dephospho forms) is much more prevalent than PKI␤-78 (Fig.  4, b and c). This may be a consequence of the second AUG codon being the preferred initiation site or because PKI␤-78 has a shorter half-life. The translation efficiency of a given AUG codon is affected by many parameters, including the sequence context of the AUG codon, secondary structure of the mRNA, and length of the 5Ј-untranslated region (13). PKI␤-70 and PKI␤-78 have equivalent PKA inhibitory activity. As one possibility, the production of these two forms of PKI␤ may be important to direct them to different intercellular locations, as occurs with the multiple forms of MOD5 protein that are produced by alternate translational initiation (45). The additional 8 amino-terminal amino acids of PKI␤-78 do not, however, conform to any established signal sequences.
The second mode for the formation of alternate PKI␤ isoforms is by their post-translational phosphorylation, which gives rise to the phospho forms of the four species. The degree of change in isoelectric point between the phospho and dephospho forms of each suggests that each is likely to be monophosphorylated. We have yet to determine which enzyme(s) catalyzes the phosphorylation of the PKI␤ isoforms. We have shown previously that PKI␣ could be phosphorylated by the epidermal growth factor tyrosine kinase receptor (47); however, since neither PKI␤-70 nor PKI␤-78 contains tyrosine residues, an equivalent phosphorylation cannot explain the formation of phospho forms observed with at least these two lower molecular mass species of PKI␤. The differential sensitivity of phospho-PKI␤-70 and phospho-PKI␤-78 to protein phosphatase treatment suggests that possibly more than one kinase may be involved, and the fact that the phospho forms are more predominant in testis than in cerebellum may indicate that the protein kinase responsible exhibits a distinctive tissue distribution pattern. One example of such a kinase is the male germ cell-specific Ser-Pro-directed kinase (Mak) that appears to be present only in post-meiotic germ cells in testis (48,49). As of yet, there is no evidence that phosphorylation changes the inhibitory potential of the PKI␤ forms and, as expanded upon below, is probably unlikely as a physiological mechanism of control.
The specific functions of the eight forms of PKI␤ remain to be elucidated. While possible, it would appear unlikely that these represent PKI forms with different levels of PKA inhibitory potency. The concentration of PKA and PKI in cells exceeds the K i value of inhibition by several orders of magnitude, and thus, even a very large change in the K i value between forms may not result in a greater or lesser degree of PKA activity under physiological conditions. It appears more probable that the different forms of PKI␤ would play some role in modulating the full cAMP response, other than the simple direct block of PKA activity, and be involved in such activities as modulating the intracellular site of the PKA catalytic subunit (2)(3)(4)9) or allowing for an altered pattern of interaction with its regulatory subunit.