Evidence for a second isoform of the catalytic subunit of cAMP-dependent protein kinase.

We have used a previously characterized mouse cDNA clone for the catalytic (C) subunit of cAMP-dependent protein kinase (Uhler, M. D., Carmichael, D. F., Lee, D. C., Chrivia, J. C., Krebs, E. G., and McKnight, G. S. (1986) Proc. Natl. Acad. Sci. U. S. A. 83, 1300-1304), which we designate C alpha, to isolate cDNA clones coding for a second isoform of the C subunit, C beta. C alpha cDNA clones hybridize to a 2.4-kilobase mRNA on Northern blots whereas C beta cDNA clones detect a 4.3-kilobase mRNA. Nucleotide sequence comparison between C alpha and C beta cDNA clones shows that the C beta cDNA codes for a protein which shows 91% identity with C alpha. Determination of mRNA levels for C beta in various tissues shows that it is most highly expressed in brain although it is detectable in all tissues examined. The presence of two genes coding for the C subunit of cAMP-dependent protein kinase may explain past reports of heterogeneity in C subunit protein preparations.

Many of the biological effects of cAMP are thought to involve activation of CAMP-dependent protein kinase via binding of cAMP to the regulatory (R)' subunit and subsequent release of the catalytic (C) subunit from the holoenzyme (1). Two different types of CAMP-dependent protein kinase have been distinguished based on their order of elution from DEAE-cellulose and have been termed Types I and I1 (2). The R subunit of the Type I enzyme (RI) has a molecular weight of 49 kDa on SDS-polyacrylamide gels, and the amino acid sequence of bovine skeletal muscle RI has been reported (3). At least two forms of the Type I1 subunit (RII) have been identified a 54-kDa RII found in most tissues (4) and a 51-kDa form found in brain, granulosa cells, and adrenal tissue (5, 6). In contrast, C subunit preparations isolated from the different types of cyclic AMP-dependent protein kinase and from different tissues have identical molecular weights of 40 kDa by SDS-polyacrylamide electrophoresis (7)(8)(9)(10). In addi-* This work was supported by United States Public Health Service Grant GM 32875. The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked ''aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide sequence(s) reported in this paper has been submitted to the GenBankTM/EMBL Data Bank with accession number(s) 502625.
$ tion these different C subunit preparations also have similar biochemical and enzymatic properties (7)(8)(9)(10). Nevertheless, as many as three forms of this protein have been separated by isoelectric techniques (9). We have previously described mouse cDNA clones for the C a subunit which is 98% identical with the published bovine heart C subunit protein sequence (11). We have used the Ca cDNA under low-stringency hybridization conditions to isolate cDNA clones coding for proteins related to the C subunit.
Here we report the isolation of cDNA clones coding for a protein (CP) which shows 91% amino acid identity with Ca.

MATERIALS AND METHODS
Construction and Screening of cDNA Libraries-Phage hgtl0 and Xgtll cDNA libraries from BALB/c mouse heart, brain, and S49 lymphoma cells were constructed as described (12). Phage plaques were transferred to nitrocellulose filters, denatured, and hybridized under low (25 "C) or high (42 "C) stringency in HYB buffer (50% formamide, 0.75 M NaCl, 0.075 M sodium citrate, 50 mM Na,HPO,, pH 7.5, 5 mM EDTA, 0.1% polyvinylpyrrolidone, 0.1% bovine serum albumin, 0.1% Ficoll, 0.1% SDS, 100 @g/ml of salmon sperm DNA). The nitrocellulose filters were washed at 20 "C (low stringency) or 65 "C (high stringency) in 0.9 M NaCl, 0.09 M sodium citrate, 50 mM Na,HPO,, pH 7.5, 0.1% SDS. A 700-base pair cDNA insert (01) was isolated from a S49 lymphoma cDNA library by low stringency hybridization using the 75-base pair SauSA/EcoRI Ca fragment employed previously to isolate the C a cDNA clone MC8 (11). The 91 cDNA insert was then used to isolate MCpl and MCP2 cDNA clones from a mouse heart library and to isolate MCP3 from a mouse brain cDNA library.
Subcloning and Sequencing of cDNA Inserts-The EcoRI cDNA inserts of MCP1, MCP2, and MCP3 were subcloned into the EcoRI site of pUC13 for restriction mapping. Restriction fragments of the MCPl, MCP2, and MCP3 inserts were then subcloned into M13 phage vectors and both strands were sequenced using the dideoxy termination method as described (11). No differences in overlapping sequences were found between the three CP cDNA clones.
Northern Blot Analysis-Samples were denatured in 20 mM MOPS, pH 7.0, 1 mM EDTA, 5 mM sodium acetate, 2.2 M formaldehyde, and 50% formamide at 65 "C for 15 min. The samples were loaded on a 1% agarose gel and run in the same buffer without formamide. The gel was then blotted to nitrocellulose, baked at 80 "C for 2 h, and hybridized for 16-20 h.
Quantitutwn of CCX and Cp mRNA Levels-Quantitation of CCX and CB mRNA levels was performed as described (11) except that the hybridizations were performed at 80 "C rather than 70 "C to increase the specificity of the hybridization. The SP6 transcript used to measure Cp mRNA level corresponded to the 5' EcoRI/ScaI fragment of MCP1.

RESULTS
In order to isolate cDNA clones coding for proteins related to the Ca protein, a 75-base pair Sau3AIEcoRI fragment of the Cn gene (11) was used to screen an S49 lymphoma cDNA library under low stringency hybridization conditions. A cDNA clone that hybridized to this probe but did not have a restriction map consistent with that of the Ca cDNA clones was then used under conditions of high stringency to isolate cDNA clones containing the entire coding region of CP. The composite restriction map for clones MCPl, MCP2, and MCP3 is shown in Fig. 1 and clearly distinguishes these cDNAs from the previously characterized C a cDNAs (11).
Sequence analysis of these three clones showed that they contained a 1053-nucleotide open reading frame identical in size to that found for the Car cDNA (Fig. 2). Within the coding regions, however, there is an 80% nucleotide homology between the mouse Ca and Cp cDNAs. The CP nucleotide sequence predicts a protein of 351 amino acids that has 91% identity to mouse Ca and half of the amino acid differences are conservative substitutions. The homology between Ca and Cp is weakest in the amino-terminal regions of the two proteins with only 73% identity over the first 70 amino acids. The two amino acid residues that have been implicated in ATP binding for the porcine C subunit, Lys7' and CYS'~, are conserved in both mouse Ca and Cp (13, 14). Comparison of the two protein sequences shows that Cp has a charge of +4 at pH 7.0, while C a would have a charge of +3. In spite of this charge difference, it is not clear that these two proteins could be resolved by isoelectric focusing techniques. The 192 nucleotides 5' of the initiation codon were 71% GC, similar to the 84% GC observed for the Ca cDNA (11).
In order to demonstrate the presence of both Ca and Cp mRNA in mouse cell lines and normal mouse tissues, the Ca  pituitary tumor cells or S49 lymphoma cells and 32P-labeled DNA markers were electrophoresed under denaturing conditions on a 1% agarose gel and blotted to nitrocellulose. One half was hybridized to a nick-translated Ca probe (the EcoRIIScaI fragment of MC411) and the other half was hybridized to a Cp probe (MCDl). Both halves were washed and exposed at -70 'C for 24 h. B, two Northern blots were prepared as described above using total RNA (6 pg) from mouse brain ( B ) , heart ( H ) , lung (L), testes ( T ) , or kidney ( K ) . One blot was hybridized with a Ca probe (the 400-base pair PuuII/PuuII fragment of MC411), washed, and exposed at -70 "C for 24 h. The other was hybridized with a Cp probe (the EcoRIIScaI fragment of MCpl), washed, and exposed at -70 "C for 48 h. The arrows indicate a 2.4-kb RNA species hybridizing to the Ca probe and a 4.3-kb RNA hybridizing to the Cp probe.

TABLE I Levels of Ca and Cp mRNAs in mouse tissues
Levels of Ca and Cp mRNA in mouse tissues were determined using an SP6 RNA transcript complementary to either Ca and Cp mRNA in a solution hybridization assay as described (11) except that the hybridizations were carried out at 80°C. The SP6 probe for Ca was the same as that used previously (11). The SP6 probe for Cp was generated using the EcoRI/ScaI fragment of MCp1. The absence of cross-hybridization between Ca and Cp SP6 SP6 probes was demonstrated using total nucleic acid samples from cells transfected with a Ca expression vector that contained levels of C a mRNA 10-fold higher than untransfected cells. Cp mRNA levels in these cells were identical to untransfected cells when measured using the SP6 solution hybridization assay. and Cp cDNAs were used for Northern blot analysis of C a and Cp mRNA. Fig. 3 shows that, whereas the C a mRNA is 2.4 kb in length (ll), the Cp mRNA is 4.3 kb in all tissues examined. In addition, the Northern blots show that the relative ratio of C a to Cp mRNA varies dramatically between different tissues. For example, there appear to be approxi-

+4.3kb
mately equal amounts of C a and Cp mRNA in anterior pituitary AtT-20 cells, but in S49 lymphoma cells the Ca mRNA predominates. In normal mouse tissues, brain appears to contain the largest amount of Cp mRNA although detectable levels appear in all tissues. Northern blot analysis under conditions of low stringency with either Ca or Cp probes shows only two bands at 2.4 and 4.3 kb (data not shown); however, this does not exclude the possible existence of mRNAs coding for other isoforms of C since they might be expressed in tissues we have not examined or have less nucleic acid homology to C a and C(3 than C a and C(3 have to each other. More quantitative estimates of C a and C(3 mRNA levels were obtained by using SP6 RNA probes in a solution hybridization assay under conditions where cross-hybridization of C a and Cp does not occur ( Table I). Since the DNA content of the tissue nucleic acid samples varies from one tissue to another, the results are given both in mRNA molecules/cell and in pg of mRNA/pg of total RNA for all the tissues examined. These results generally agree with those seen in Northern blot analysis and show that the level of Cp mRNA varies from 4 molecules/cell in intestine, lung, pancreas, and spleen to 20 molecules/cell in brain. The SP6 solution hybridization results confirm that the ratio of C a to Cp mRNA varies widely from tissue to tissue with testis having 20-fold more C a than Cp mRNA and spleen having approximately the same amounts of the two mRNAs. No consistent correlation between the Ca/Cp mRNA ratios (Table I) and the relative ratio of Type 1:Type I1 kinase is seen for mouse brain (1:9), heart (1:1), liver (l:l), and lung (61) (15). However, it should be noted that relative levels of C a and Cp protein in a particular tissue may differ substantially from that shown for the mRNAs depending on the translational efficiency and the rates of protein degradation.

DISCUSSION
Although many authors have assumed that C subunits from different tissues and holoenzyme types are identical (1,2,16), heterogeneity in C subunit preparations has been reported previously. Analysis of bovine heart Type I1 C subunit preparations using isoelectric focusing has shown three distinct isoelectric forms with PI values of 7.01, 7.48, and 7.78 (9). Since the C subunit is known to be both phosphorylated a t Thrlg7 and Ser338 and myristylated at its amino terminus (1,2), multiple isoelectric forms are not unexpected. However, Peters et al. (9) examined all three isoelectric forms and demonstrated similar phosphate content and amino-terminal blocking groups. Although the different isoelectric forms also had slightly different amino acid compositions, we found no clear correlation with the compositions of Ca and Cp presented here. A second report has compared the tryptic peptides generated from R and C subunits purified from either the Type I or Type I1 kinase from porcine skeletal muscle (10). Whereas the R subunits from either type of kinase had very different tryptic peptides, the C subunit preparations had very similar peptide maps with only minor differences. Finally, three forms of bovine heart C subunit complexed with the cyclic AMP-dependent protein kinase inhibitor protein were identified upon nondenaturing electrophoresis (17). The formation of one of the complexes required ATP while another showed no such dependence on ATP.
The multiple forms of the inhibitor protein-C subunit complexes were suggested to represent multiple isoelectric forms of the C subunit interacting with the inhibitor protein. Unfortunately it is difficult to strictly correlate any of these earlier demonstrations of C subunit heterogeneity with the Ca and C@ sequences shown here due to species differences and uncertainty regarding the contribution of post-translational phosphorylation and myristylation events to the observed heterogeneity.
The existence of two distinct C subunits of CAMP-dependent protein kinase raises several interesting questions about their functional significance. The high degree of amino acid sequence homology, despite considerable difference in nucleic acid sequence, suggests that after an early gene duplication event, Ca and Cp subunits acquired distinct and essential roles as part of the CAMP-dependent protein kinase system. This conclusion is supported by the demonstration that bovine tissues also express a Cp subunit' that shows more homology to mouse Cp (97% amino acid identity) than mouse Cp shows to mouse Ca.
It is possible that the amino acid sequence differences between Ca and Cp could lead to differences in substrate specificity between the two types of C subunit. The observation that substrate sequences on the amino-terminal side of M. 0. Showers, and R. A. Maurer, manuscript in preparation. phosphorylated serine residues generally fall into one of two classes, Arg-Arg-X-Ser or Lys-Arg-X-X-Ser, could be consistent with the existence of multiple forms of the C subunit (18). Since the regulatory subunits inhibit catalytic activity by binding to the same binding site as substrate proteins (2), one might also expect differences between Ca and Cp in their interactions with the various regulatory subunits. We suggest that Ca and Cp subunits do not represent Type I and Type I1 specific C subunits but that Ca interacts with both RI and RII while Cp may preferentially interact with a distinct R subunit isoform that has yet to be characterized. The crucial questions regarding the interaction of Cp with different types of R subunits and the possible differences in substrate specificities between Ca and Cp can be addressed now that cDNAs for Ca and CB have been isolated.