Identification of novel splice variants of the human catalytic subunit cβ of cAMP-dependent protein kinase

Authors


B. S. Skålhegg, Institute for Nutrition Research, University of Oslo, PO Box 1046, Blindern, N-0316 Oslo, Norway. Fax: + 47 22851531, Tel.: + 47 22851548.

Abstract

Four different isoforms of the catalytic subunit of cAMP-dependent protein kinase, termed , , Cγ and PrKX have been identified. Here we demonstrate that the human gene encodes six splice variants, designated Cβ1, Cβ2, Cβ3, Cβ4, Cβ4ab and Cβ4abc. The splice variants differ in their N-terminal ends due to differential splicing of four different forms of exon 1 designated exon 1-1, 1-2, 1-3, 1-4 and three exons designated a, b and c. All these exons are located upstream of exon 2 in the gene. The previously identified human variant has been termed Cβ1, and is similar to the isoform identified in the mouse, ox, pig and several other mammals. Human Cβ2, which is the homologue of bovine Cβ2, has no homologue in the mouse. Human Cβ3 and Cβ4 are homologous to the murine Cβ3 and Cβ2 splice variants, whereas human Cβ4ab and Cβ4abc represent novel isofoms previously not identified in any other species. At the mRNA level, the splice variants reveal tissue specific expression. Cβ1 was most abundantly expressed in the brain, with low-level expression in several other tissues. The Cβ3 and Cβ4 splice variants were uniquely expressed in human brain in contrast to Cβ2, which was most abundantly expressed in tissues of the immune system, with no detectable expression in brain.

We suggest that the various splice variants when complexed with regulatory subunits may give rise to novel holoenzymes of protein kinase A that may be important for mediating specific effects of cAMP.

Abbreviations
KO

knockout.

Cyclic 3′,5′-adenosine monophosphate (cAMP) is a key intracellular signaling molecule, the main function of which is to activate the cAMP-dependent protein kinases [1]. Protein kinase A is a heterotetrameric enzyme containing a regulatory (R) subunit dimer and two catalytic (C) subunits. The holoenzyme is activated when four molecules of cAMP bind to the R subunit dimer, two to each R subunit, releasing two free active C subunits [2]. In human enzyme, four different R subunits (RIα, RIβ RIIα, RIIβ), and four different C subunits (, , Cγ and PrKX) have been identified [3]. The and subunits are expressed in most tissues, while the Cγ subunit, which is transcribed from an intron-less gene and represents a retroposon derived from the subunit [4], is only expressed in human testis [5]. PrKX is an X chromosome-encoded protein kinase, and was recently identified as a protein kinase A C subunit because it is inhibited by both protein kinase inhibitor (PKI) and regulatory subunit Iα (RIα), and because the RIα/PrKX complex is activated by cAMP [6].

Splice variants of both and have been identified. The splice variants encoded by the gene have been termed Cα1 (previously named [7]), Cα2[8] and Cα-s[9]. Originally, Cα2 was isolated from interferon-treated cells as a C-terminally truncated and enzymatically inactive subunit, probably representing a retroposon. Recently a novel Cα2 variant was reported [10], this splice variant is homologous to the Cα-s, which has been previously been identified and characterized in ovine [9] and human sperm [11]. Furthermore, Cα-s, now designated Cα2 is encoded with a truncated and nonmyristylated N-terminal end when compared to Cα1[9–11]. The variable parts of Cα1 and Cα2 are located upstream of exon 2 in the murine gene and are encoded by alternative use of different first exons [10]. In bovine, two splice variants of have been identified, termed bovine Cβ1[12] and bovine Cβ2[13]. Bovine Cβ1 is ubiquitously expressed whereas Cβ2 is expressed at low levels in most tissues with the highest expression in the spleen, thymus, and kidney [13]. The bovine splice variants contain variable N-terminal ends in which the nonidentical sequences are probably encoded by different forms of exon 1, as is the case with splice variants identified in the mouse, where three splice variants designated Cβ1, Cβ2 and Cβ3 have been identified [14]. Whereas mouse Cβ1 is ubiquitously expressed, mouse Cβ2 and mouse Cβ3 have so far only been identified in the brain. The mouse and bovine Cβ1 are similar along the entire sequence, demonstrating that they represent orthologous protein sequences. In contrast, mouse and bovine Cβ2 are not similar in the N-terminal region, indicating that their N-termini are encoded by unrelated exons. Thus, mouse and bovine Cβ2 are not orthologous proteins. Previous to this study, only a single splice variant of human with high homology (more than 98%) to mouse and bovine Cβ1 has been identified, demonstrating this isoform is the human Cβ1 splice variant. In the present study, we demonstrate that the gene encodes at least six different gene products, designated Cβ1, Cβ2, Cβ3, Cβ4, Cβ4ab and Cβ4abc. As is the case with the murine and bovine splice variants, all the human splice variants vary in the N-terminal region preceding that encoded by exon 2. All splice variants identified in mouse and bovine were identified in human (Cβ1, Cβ2, Cβ3 and Cβ4) in addition to two novel splice variants (Cβ4ab and Cβ4abc), that have previously not been identified in any other species.

Materials and methods

General protocols

Complementary DNA probes were radiolabeled using the Megaprime random priming kit and [α-32P]dCTP (Amersham) as instructed by the manufacturers to a specific activity of at least 1 × 109 c.p.m. Synthetic oligonucleotides were radiolabeled using T4 polynucleotide kinase (Pharmacia) and [γ-32P]ATP as instructed by the manufacturer.

DNA was either sequenced manually using Thermo Sequenase radiolabeled terminator cycle sequencing kit (Amersham) or by Medigenomix (Martinsried, Germany). Sequences were analyzed using the Wisconsin University gcg program package (uwgcg) and the basic local alignment and search tool (blast) [15]. Note: all numbering of nucleotide sequences refer to A in ATG translation of initiation as nucleotide +1.

Identification of cDNAs

The 5′ end of human cDNA was amplified from human total fetus and brain Marathon RACE-ready cDNAs (Clontech) using the Advantage KlenTaq Polymerase Mix (Clontech) as described by the manufacturer. Amplification was performed using adapter primer 1 (Clontech) and four different primers (1–4), complementary to the human cDNA sequence.

Primer 1, 5′-CAACCCAAAGAGAAGTAAGAAAGT GGTCTA-3′ corresponds to nucleotide 1272–1301 in the human cDNA [5]. Primer 2, 5′-TTGGTTGGTCTGCAAAGAATGGGGGATAGC-3′ corresponds to nucleotides 704–733 in the human cDNA [5]. Primer 3, 5′-TTTTCTCATTCAAAGTATGCTCTATTTGC-3′ corresponds to nucleotides 252–280 in the cDNA [5]. Primer 4, 5′-TATTTGCTTCAGTTTAACAACCTTCT GCTT-3′ corresponds to nucleotides 229–258 in the cDNA [5].

Five cycles were performed with 45 s, 94 °C; 2 min, 72 °C; five cycles with 45 s, 94 °C; 2 min, 70 °C; and 25 cycles with 45 s, 94 °C; 2 min, 68 °C; and a final extension of 10 min at 72 °C. The resulting products were separated by gel electrophoresis, subcloned into pCR2.1TOPO (Invitrogen) as instructed by the manufacturer and sequenced.

Amplification of gene fragments

A genomic fragment was amplified using an oligonucleotide corresponding to exon 1-3 (5′-GTTTAGGTGCAATCATTCTGCTGTTTG-3′, nucleotides − 9 to 15 in the Cβ3 cDNA, Fig. 1A) and a primer complementary to sequences in exon 2 (5′- AAAAAGTCTTCTTTGGCTTTGGCTAGA-3′ nucleotides 57–83 in the cDNA [5]). Another genomic fragment was amplified using a primer corresponding to exon 1-2 (5′- TGGCAGCTTATAGAGAACCACCTT-3′ nucleotides − 9 to 15 in the Cβ2 cDNA, Fig. 1A) and a primer complementary to sequence found in exon 1-3 (5′- CAATCCCATGTTGAACCTGGCA-3′ nucleotides −13 to 9 in the Cβ3 cDNA, Fig. 1A). PCR reactions were performed using the Expand Long Template PCR kit as instructed by the manufacturer using buffer 2. PCR was performed using human genomic DNA (Boehringer-Mannheim) as template with 1 min at 92 °C followed by 30 cycles of 10 s, 94 °C; 30 s, 60 °C and 10 min (extended with 20 s per cycle from cycle 11 to cycle 30), 68 °C, and a final incubation of 7 min at 68 °C. Products were separated by agarose gel electrophoresis and analyzed by Southern blotting using radiolabeled cDNAs and synthetic oligonucleotides corresponding to the different exons.

Figure 1.

 Nucleotide and deduced amino-acid sequences of novel splice variants (A) and comparison of the N-terminal amino-acid sequences of novel splice variants (B). (A) 5′ RACE on RACE ready cDNA, using four different oligonucleotide primers complementary to the previously published human cDNA, in combination with an anchor primer sequence (for primer sequences, see Materials and methods), were used to obtain five novel nucleotide sequences for the subunit. The sequences are designated Cβ2, Cβ3, Cβ4, Cβ4ab and Cβ4abc. The ATG start codon (bold underlined) is indicated in each clone. Starting at the ATG, the amino-acid sequence has been deduced and is given in one letter code below the nucleotide sequence. Only the 5′ end of the cDNAs are shown, including the variable parts and some of the nucleotides of the part common to all splice variants, encoded by exon 2. (B) The amino-acid sequences of the N-terminal parts of Cβ1 and Cβ2, Cβ3, Cβ4, Cβ4ab and C4βabc are shown. The myristylation motif G–N previously identified in Cβ1 is boxed. A protein kinase A autophosphorylation motif that has previously been identified in Cβ1, is shown in bold italic type and Ser10 which is potentially phosphorylated, is labeled by an asterisk. Note that there is a protein kinase A autophosphorylation motif, encoded by exon a, present in Cβ4ab and Cβ4abc.

Screening of P1-derived artificial chromosome library and subcloning of exon-containing sequences

The human P1-derived artificial chromosome library, RPCI-6, was screened and the isolated bacterial clone was grown in liquid culture and plasmid DNA was isolated using ion-exchange columns as described by the manufacturer (Qiagen, Hilden, Germany). Exon-containing DNA restriction fragments were identified by Southern blotting using radiolabeled cDNAs and synthetic oligonucleotides. Exon-containing fragments were excised from the gel and subcloned into the pZERO2.1 vector (Invitrogen) as instructed by the manufacturer.

Generation of splice variant specific probes, Northern blotting and Southern blotting

DNA fragments corresponding to the splice variant-specific parts of the cDNAs were amplified by PCR. The following primers were used for the different splice variants: Cβ1, 5′-GCTCTCCACCTCGCTGCCTTTCTT-3′ (Cβ1 cDNA [5] nucleotides 22–45) and primer 5′-CCAGCCCCCC TTCCCTTCCCTGAC-3′ (Cβ1 cDNA [5] nucleotides −47 to −24); Cβ2, primer 5′-TGGCAGCTTATAGAGAACCACCTT-3′ (Cβ2 cDNA Fig. 1, nucleotides −9 to 15) and primer 5′-ATTGATCTGTCCATAAGGCAGTAT-3′ (Cβ2 cDNA [5], nucleotides 154–178); Cβ3, primer 5′-TCACAGCTAGCAGTAAGAGCTG-3′ (Cβ3 cDNA Fig. 1, nucleotides −81 to − 61) and primer 5′-C AATCCCATGTTGAACCTGGCA-3′ (Cβ3 cDNA Fig. 1, nucleotides − 13–9); Cβ4, primer 5′-TCTCCAGT GTGTGTGTTTACAC-3′ (Cβ4 cDNA Fig. 1, nucleotides −114 to −94) and primer 5′-ATGATGAAAACCAACCTTTCCA-3′ (Cβ4 cDNA Fig. 1, nucleotides 95–116).

The primers were used for amplification of the fragments from cloned RACE products using Taq DNA polymerase (PerkinElmer) as described by the manufacturer. For generation of a probe specifically recognizing exons a and b, the primers 5′-GATATTTCTGAAGAGGAGCAAGCAG ATGCATCTGATGATTTGCGTG-3′ and 5′-CACGCAAATCATCAGATGCATCTGCTTGCTCCTCTTCAGA
AATATC-3′ (Cβ4ab cDNA Fig. 1A, nucleotides 8–53) were annealed, phosphorylated and ligated. A 1.5-kb fragment of cDNA (nucleotide 47–1547 of the clone in [5]) was used for recognizing the parts of the mRNA common to all splice variants. Two similar Northern blots containing RNA from various human sources were purchased from Clontech. One blot was hybridized using a probe specific for Cβ2, while the other blot was probed in succession with probes specific for Cβ3, Cβ4, exons a and b, and the 1.5-kb cDNA. Both blots were hybridized using glyceraldehyde-3-phosphate dehydrogenase cDNA as control. As an almost identical pattern of hybridization was obtained using glyceraldehyde-3-phosphate dehydrogenase on both blots, only one glyceraldehyde-3-phosphate dehydrogenase blot is shown (Fig. 3). All probes were hybridized in ExpressHyb hybridization solution (Clontech) as described by the manufacturer. A Southern blot containing EcoRI-digested DNA from various species (Clontech) and Southern blots containing human and mouse DNA digested with various enzymes were hybridized using the probe specific for Cβ2. The filters were prehybridized in 5 × Denhardt's solution, 5 × NaCl/Cit, 50 mm sodium phosphate buffer, pH 6.8, 0.1% SDS, 250 µg·mL−1 single stranded salmon sperm DNA, and 50% (v/v) formamide at 42 °C for 3 h, and hybridized for 16 h in a similar solution containing the radiolabeled common or Cβ2 probe. The membranes were washed four times in 2 × NaCl/Cit, 0.1% SDS for 5 min at room temperature, followed by two washes using 0.5 × NaCl/Cit, 0.1% SDS at 50 °C for 30 min. Autoradiography was performed at −70 °C using Amersham Hyperfilm MP and intensifying screens.

Figure 3.

 Tissue distribution of different splice variants. Northern blots containing various human tissues were hybridized using probes specific for Cβ1, Cβ2, Cβ4, exon a + b and a probe common to all splice variants (see Materials and methods). For comparison, the same blots were hybridized using a glyceraldehyde-3-phosphate dehydrogenase cDNA (GAPDH). All mRNAs had the same apparent length (4.4 kb).

Results

Nucleotide and deduced amino-acid sequence of novel Cβ splice variants

The 5′ ends of human cDNAs were amplified from human brain and total fetus RACE-ready cDNA using four different oligonucleotide primers complementary to the previously published human cDNA, in combination with an anchor primer sequence (for primer sequences, see Materials and methods). The resulting PCR products were subcloned and sequenced revealing five novel human splice variants, which were designated Cβ2, Cβ3, Cβ4, Cβ4ab and Cβ4abc(Fig. 1A) in addition to the variant previously identified, which is now designated Cβ1 (nucleotide sequence not shown). All the novel clones lack the 46 first protein-encoding nucleotides in the human Cβ1 cDNA sequence, and were identical starting from the sequence encoded by exon 2, which contains the two last nucleotides encoding amino acid 16 in Cβ1. Each of the clones contained a translation initiation codon and one or more in-frame upstream stop codons. Comparison of the deduced amino-acid sequences imply that the novel splice variants are all encoded with different N-terminal ends (Fig. 1A,B). The Cβ2 splice variant contains a 63-amino-acid sequence substituting the first 16 amino acids in Cβ1, and is homologous to the previously identified bovine Cβ2[13]. Furthermore, the human Cβ3 splice variant contains four amino acids in the N-terminal substituting the first 16 amino acids in Cβ1, and is similar to the previously identified murine Cβ3[14]. The human Cβ4 contains three amino acids substituting the first 16 amino acids in Cβ1, and is similar to murine Cβ2[14]. Finally, the splice variants Cβ4ab and Cβ4abc contain 18 and 21 amino acids, respectively, that substitute the first 16 amino acids of Cβ1. These splice variants show no homology to the N-terminus of any other C subunits identified so far.

Comparing the amino-acid sequences revealed that the novel splice variants are not encoded with the N-terminal features found in the Cβ1 subunit (Fig. 1B). Cβ1 is encoded with an N-terminal Gly (G) which is shown to be important for myristylation. An N-terminal G is also encoded in Cβ3, suggesting that Cβ1 and Cβ3 are the only splice variants that undergo myristylation in vivo[16]. Furthermore, Cβ1 is encoded with an autophosphorylation site at Ser10 (KKGS10) [17], which is not identified in any of the novel splice variants. Despite this, it should be noted that exon a encodes what may be a potential autophosphorylation site at Ser7 (RKSS7) in Cβ4ab and Cβ4abc(Fig. 1B).

Identification of exons encoding novel splice variants of human Cβ

All the cDNAs were similar from nucleotide 47 and downstream in the Cβ1 cDNA, which corresponds to the start of exon 2 in the murine gene. The identification of novel protein-encoding sequences upstream of exon 2, indicated the presence of several different exons upstream of exon 2. Thus, human genomic DNA was amplified using a combination of primers corresponding to exon 2 (antisense orientation, see Materials and methods) and the 5′ ends of the different novel cDNAs (sense and antisense orientation) in different combinations. A 17-kb PCR product was the result of an amplification using a primer corresponding to the 5′ end of Cβ2 cDNA (sense orientation) and the 5′ end of Cβ3 (antisense orientation). Furthermore, a 14-kb PCR product was the result of an amplification using a primer corresponding to the 5′ end of Cβ3 cDNA (sense orientation) and a primer corresponding to exon 2 (antisense orientation). These clones enabled us to physically map six novel exons in the gene that were designated 1-2, 1-3, 1-4, a, b and c to be located 31, 14.1, 14, 6, 4.0 and 3.3 kb upstream of exon 2, respectively (Fig. 2A). Furthermore, a P1-derived artificial chromosome library was screened using the 5′ ends of Cβ1 and Cβ2 cDNAs as probes. One of the clones identified, RPCI-6-228E23, contained both exon 1-2 and an exon containing the entire splice variant-specific part of the Cβ1 cDNA, which we termed exon 1-1.This P1-derived artificial chromosome clone was selected for detailed restriction mapping using CpG cutters. The digested P1-derived artificial chromosome DNA was separated by pulsed-field gel electrophoresis, transferred to Southern blot membranes and hybridized with exon 1-1 and 1-2, as well as Sp6 and T7 oligonucleotide probes. These results revealed a distance of ≈ 60 kb between exon 1-1 and 1-2 (Fig. 2A). All nucleotide sequences found in the different cDNAs could be identified in a continuous stretch of human genomic DNA, thereby supporting the notion that these cDNAs are products of the same gene. Exon 1-1 was shown to be homologous to the previously identified exon 1 A of the murine gene. As shown in Fig. 2B, exon 1-2 contains the entire Cβ2 specific sequence, and exon 1-3 contains the sequence specific for Cβ3 that is homologous to the previously identified exon 1B in the mouse gene. Finally, exon 1-4 was shown to contain the sequence specific for the human Cβ4 splice variant, and to be homologous to the murine exon 1C, which encodes the N-terminal end in the murine Cβ2 splice variant. Based on the Cβ4ab and Cβ4abc cDNA sequences, the exons a, b and c (Fig. 2B), were demonstrated to be alternatively spliced between exon 1-4 and exon 2, with either exons 1-4, a, b and 2 or exons1-4, a, b, c and 2 (Fig. 2C, lower panel). These cDNA sequences represent novel splice variants not identified in any other species.

Figure 2.

 Structure of the human genomic region encoding the novel splice variants (A) and nucleotide sequence of genomic regions encoding novel splice variants of (B). (A) Primers were made based on exon 2 and the most 5′ end of the different cDNAs, and used to amplify human genomic DNA by PCR. Two overlapping PCR products of 14 and 17 kb, respectively, were identified and mapped by Southern blotting and hybridization to oligonucleotides corresponding to the different cDNAs. As derived from the 14- and 17-kb PCR products, exons 1-2 1-3, 1-4 and exons a, b and c are located 31, 14.1, 14, 6, 4.0 and 3.3 kb upstream of exon 2. Based on restriction mapping of the P1-derived artificial chromosome clone RPCI-6-228E23, exon 1-1 is located approximately 60 kb upstream of exon 1-2. Exon 1-1 is specific for the splice variant, which encodes Cβ1. The exons are indicated as vertical lines. The introns are drawn to scale as indicated. (B) Protein encoding sequences are in bold capital letters, intron and 5′ untranslated sequences are in lower case letters. Translation initiation codons are underlined. Only the 5′ end of exon 2 is included. (C): Schematic representation of how the various human exons 5′ to exon 2 may be spliced. The upper panel describes a potential model in which four variants of exon 1, designated exons 1-1, 1-2, 1-3 and 1-4, may alternatively splice with exon 1 to encode the splice variant specific sequence in Cβ1, Cβ2, Cβ3 and Cβ4. The lower panel describes a model in which the exons a, b and c may splice with exons 1-4 and 1-3 upstream of exon 2 to encode the splice variant-specific sequences in Cβ4ab, Cβ4abc and Cβ3ab.

Tissue distribution of Cβ splice variants

To examine the tissue distribution of splice variants, exon specific DNA probes and a DNA probe common to all splice variants were hybridized to two similar Northern blots containing RNA from various human tissues. For comparison, the blots were hybridized to a cDNA encoding glyceraldehyde-3-phosphate dehydrogenase. In Fig. 3 (panel Cβ1) we show that Cβ1 is predominantly expressed in brain and kidney with low-level expression in several other tissues. Cβ2 is expressed at high levels in thymus, spleen and kidney in addition to a weak signal in other tissues (Fig. 3, panel Cβ2). In contrast to Cβ2, the exon 1-4 and mRNAs containing exons a and b appeared to be present exclusively in brain (Fig. 3, panels Cβ4 and exon a + b). Finally, probing the Northern blot with a probe common to all the splice variants (see Materials and methods), we observed ubiquitous expression of with the strongest signal in brain and a somewhat weaker signal in spleen and thymus, when compared to the glyceraldehyde-3-phosphate dehydrogenase signal (Fig. 3, panel Cβ common). Hybridization using a DNA fragment corresponding to the Cβ3 specific cDNA resulted in an almost undetectable signal in the brain and no detectable signals in any other tissues (data not shown).

The human Cβ2 splice variant is not present in the mouse

Previously, three splice variants of , designated Cβ1, Cβ2 and Cβ3, have been identified in the mouse [14]. Based on the present work, it is apparent that mouse Cβ2 is not homologous to either bovine or the human Cβ2. Instead, mouse Cβ2 is homologous to what we have now designated human Cβ4. Thus, we investigated whether a splice variant similar to human Cβ2 was present in the mouse genome. A Zoo-blot containing genomic DNA isolated from human, monkey, rat, mouse, dog, cow, rabbit, chicken and yeast was hybridized using a DNA fragment corresponding to exon 1-2 of human . In Fig. 4A (lanes 1–9), we show that a DNA fragment was detected using Cβ2 specific probe in human, monkey, dog, cow, and rabbit. In contrast, the Cβ2 specific probe did not recognize any fragments in the rat and mouse suggesting that the Cβ2 specific exon is not present in the murine genome. To further substantiate this observation, we isolated total RNA from human, wild-type mice and mice that are ablated (knockout, KO) for exon 1A of the gene [18]. The RNA was isolated from immune tissues and brain because we observed high-level expression of Cβ2 in human thymus, spleen and peripheral blood leukocytes and high levels of the other splice variants in the brain (Fig. 3). The Northern blots were probed with a cDNA probe (expected to recognize all known splice variants) and a Cβ2 specific probe (see Materials and methods). As expected, the probe complementary to all known splice variants hybridized to a 4.4-kb mRNA band in total RNA isolated from brain of wild-type and exon 1 KO mice in addition to RNA isolated from wild-type spleen (Fig. 4, upper panel, common, lanes 1, 2 and 3). However, it did not recognize mRNA isolated from exon 1 KO mouse spleen (lane 4). Finally, the probe expected to recognized all known splice variants, did hybridize to the 4.4-kb message in human peripheral blood leukocytes (lane 5). When probing the same filter with the Cβ2 specific probe (Fig. 4B, lower panel) Cβ2 message was only detected in human peripheral blood leukocytes (lane 5), whereas all the mouse tissues were negative for Cβ2 mRNA (lanes 1–4).

Figure 4.

 Species distribution of Cβ2 (A) and its lack of expression Cβ2in the mouse (B). (A) A Southern blot containing EcoRI-digested genomic DNA from various species was hybridized using a DNA probe corresponding to exon 1-2 (Cβ2 specific). A single hybridizing band identifying genomic sequence homologous to human exon 1-2 was identified in mammals such as monkey, dog, rabbit and human, but not in mouse and rat. (B) A Northern blot containing total RNA (20 µg protein per lane) isolated from wild-type (+/+) mouse brain and spleen (lane 1 and 3), brain and spleen of mice ablated (–/–) for Cβ1 (lane 2 and 4) and human peripheral blood leukocytes (lane 5) was probed with a probe expected to recognize all known splice variants ( common, upper panel) and a probe specific for the Cβ2 splice variant (Cβ2, lower panel). mRNA recognized by the two probes is indicated as 4.4 kb.

Discussion

Here we demonstrate that the human gene encodes five novel splice variants, designated Cβ2, Cβ3, Cβ4, Cβ4ab and Cβ4abc, in addition to the previously identified splice variant Cβ1[12]. All the splice variants contained a unique N-terminus, and showed tissue specific expression. As we found no evidence of an additional exon upstream of exon 1-1 and all the cDNA characterized had unique 5′ ends, it is reasonable to assume that the exons 1-1, 1-2, 1-3 and 1-4 each contain a separate promoter, and that the resulting mRNA products are due to use of alternative promoters. Despite this, we cannot rule out the possibility that two or more of these splice variants share a common promoter used to alternatively splice the different exons. Furthermore, we found two variants, Cβ4ab and Cβ4abc, which were the results of alternative splicing of either exon a and b, or exon a, b and c, between exon 1-4 and exon 2. The presence of the corresponding mRNA was confirmed by hybridizing a Northern blot with a probe complementary to the sequences found in exons a and b. This probe and the probe specific for Cβ4 bound to an RNA with the same apparent length located in human brain. The location of the exons a, b and c may suggest that they generate splice variants of in addition to those demonstrated here. Indeed, a short cDNA from human infant brain has been sequenced and demonstrated to contain a combination of exons 1-3, a, b and 2 (GenBank accession no. AA351487; see Fig. 2C). We were unable to produce such a cDNA, which could be due to low-level expression of Cβ3 in adult brain.

In all species examined, the various splice variants contain the same conserved catalytic domain encoded by the same sets of exons in the gene, which may be indicative of similar enzymatic features. This is in contrast to the exons upstream of exon 2, which encode N-terminal sequences with low or no homology. In fact, the N-terminus of Cβ1 shows higher homology to the Cα1 N-terminus than to any of the N-termini found in the other splice variants. The Cβ1 and 1 N-terminals are 98% homologous, form an α helix[19], and contain two sites for post-translational modification, an N-terminal G which is myristylated [16,20] and a conserved autophosphorylation site [17]. It should, however, be noted that both human and mouse Cβ3, possess an N-terminal G. Despite this, the G in mouse Cβ3 is not myristylated in vivo[14,20], because the G is not followed by the required amino acid [21]. Thus, because the mouse and human Cβ3 are 100% identical at the N-terminus, we suggest that human Cβ3 when expressed in vivo is not myristylated.

The N-terminal consensus autophosphorylation motif, KKGS(7–10), is identified in both Cα1 and Cβ1[7,12], but not in any of the other splice variants. Instead, we identified a potential autophosphorylation site, RKSS(3–6), in Cβ4ab and Cβ4abc that is encoded by exon a. To what extent this site represents a true autophosphorylation site, or if it is phosphorylated by other kinases, remains to be investigated.

The functional role(s) of the N-terminal domains and the post-translational modifications are elusive. However, expression of the N-terminally truncated and nonmyristylated C subunits such as the mouse Cβ2 and Cβ3 revealed that they are enzymatically active in vivo[14], suggesting that the α helix, the myristyl group and autophosphorylation are not crucial for catalytic activity [16,19,22]. This may also imply that the human Cβ3, Cβ4, Cβ4ab and Cβ4ab are active when expressed in vivo.

The increasing number of reports of C subunits with variable N-terminal ends lacking the ability to be myristylated and autophosphorylated in vivo may suggest differential features associated with the N-terminal domain. The myristyl group fills a hydrophobic pocket in the large lobe and was first suggested to be important for C subunit structure stability [16,23]. However, recent reports have shown that the absence or altered location of myristyl in Cα2 and Cα1, respectively, may influence hydrophobic properties, and thus, may be implicated in subcellular localization of the C subunits [9,11,24,25]. Based on this, we suggest that the N-terminally truncated and presumably nonmyristylated human Cβ3 and Cβ4 splice variants may display altered hydrophobic properties in vivo compared to Cβ1.

The human Cβ2 splice variant was similar to the previously identified bovine Cβ2 splice variant, but showed a somewhat different tissue distribution. Whereas the human Cβ2 subunit was absent from the brain, and expressed at high levels in immune tissues as examined by Northern blotting, the bovine Cβ2 variant was shown using the same method to be expressed in the brain and several other tissues, including immune tissues [13]. The diverging results may be caused by the specificity of the probes used. Furthermore, despite Thullner and colleagues’ identification [26] of bovine Cβ2 using a Cβ2 polyclonal antisera in rodents, we were unable to identify a similar splice variant in mice and rats as demonstrated both by Northern- and Zoo-blot analysis. Our results indicate that rodents lack the Cβ2 specific sequence in the genome and thus, do not express mRNA and protein for this subunit. The suggestion that rodents may lack the Cβ2 homologue is strengthened as mice ablated for Cβ1 do not express any homologue in peripheral tissues, including spleen and thymus that is recognized by the mouse cDNA [14]. Despite this, we can not rule out that a homologue not recognized by the probes used in this paper and in previous studies [14], but that is recognized by the bovine antiserum may exist. Finally, Cβ2 is expressed with an unusually long N-terminus. Thus, Cβ2 appears to be the most atypical splice variant and may have specific and unique features that will require further studies for a full characterization.

We suggest that tissue-specific expression of various splice variants when complexed with R subunits suggests the presence of novel protein kinase A holoenzymes with specific functional features that may be important as mediators of cAMP effects.

Acknowledgements

We acknowledge Professor G. Stanley McKnight, University of Washington, Seattle, Washington, USA for providing us with tissues from knockout mice. We are grateful for the expert technical assistance of Sissel Eikvar. This work was supported by the Norwegian Cancer Society, The Norwegian Research Council, Novo Nordic Foundation and Anders Jahre's Foundation for the Promotion of Science.

Ancillary