Novel Alternative Splice Variants of cGMP-binding cGMP-specific Phosphodiesterase*

After our recent findings that the amino-terminal portion of rat cGMP-binding, cGMP-specific phosphodiesterase (cGB-PDE) differs from those of bovine and human cGB-PDEs, we found two forms of canine cGB-PDE cDNAs (CFPDE5A1 and CFPDE5A2) in canine lung. Each contained a distinct amino-terminal sequence, CFPDE5A1, possessing an amino-terminal portion with sequence similar to those of bovine and human, and CFPDE5A2, having one similar to that of rat. Other portions coding for the cGMP binding domains and the catalytic domain were conserved. Both CFPDE5A1 and CFPDE5A2 transcripts were detected in the cerebellum, hippocampus, retina, lung, heart, spleen, and thoracic artery. CFPDE5A1 transcripts were particularly abundant in the pylorus, whereas CFPDE5A2 transcripts were quite low in this tissue. CFPDE5A1 and CFPDE5A2 expressed in COS-7 cells had cGMP K m values of 2.68 and 1.97 μm, respectively, and both were inhibited by a low concentration of a cGB-PDE inhibitor, Zaprinast. Both CFPDE5A1 and CFPDE5A2 bound cGMP to their allosteric cGMP binding domains, and this cGMP binding was stimulated by 3-isobutyl-1-methylxanthine. Thus, two types of alternative splice variants of canine cGB-PDE have been identified and shown to have similar biological properties in vitro.

Cyclic nucleotides including cGMP are well known as second messengers and regulate many functions in various tissues (1)(2)(3)(4). Many kinds of cyclic nucleotide phosphodiesterases (PDEs) 1 have been reported as regulators of intracellular cAMP and cGMP concentrations, and based on amino acid sequence analysis and biochemical properties seven PDE families have been recognized in mammalian tissues (5,6). One of the PDE families, cGMP-binding, cGMP-specific PDE (cGB-PDE), is highly specific for cGMP and is involved in modulation of intracellular cGMP. cGB-PDE activity was found in lung, vascular and tracheal smooth muscle cells, spleen, and platelets (7)(8)(9)(10)(11). Recently, we cloned rat and human cGB-PDE cDNAs from lung cDNA libraries and investigated their tissue distribution (12,13). Our studies showed that high levels of rat cGB-PDE transcripts were observed in cerebellum and intestine besides tissues containing vascular smooth muscle cells. Although the cGMP/cGB-PDE pathway would be expected to play important roles in various tissues, the physiological roles and regulations of cGB-PDE are not well understood.
cGB-PDE has been purified from rat and bovine lung, and its characterization has been reported in earlier studies (14,15). Photoaffinity labeling studies have demonstrated that cGB-PDE contains two cGMP binding domains and one catalytic domain (16). The phosphorylation site for cAMP-and cGMPdependent protein kinases has been identified in cGB-PDE (17). Bovine cGB-PDE cDNA isolated from a lung library was revealed to consist of two kinds of domains and the phosphorylation site described above (18). Rat and human cGB-PDEs have also a potential phosphorylation site (12,13). The phosphorylation of cGB-PDE may participate in regulation of enzymatic activity and interactions with cellular factors.
In each PDE family, gene products derived from the subfamily genes and alternative splice variants have been reported (6). PDE4, which is the largest of the PDE families, consists of four genes and a large number of alternative splice variants. In many cases different gene products and alternative splice variants in each PDE family show different expression patterns in tissues and different subcellular localization (6, 19 -27). PDEs encoded by alternatively spliced mRNAs were reported to differ in their regulation by kinases and associated proteins (27,28). Some of PDE families are composed of splice variants having distinct amino-terminal domains, which are associated with some cofactors such as kinases, and by which their tissue expression patterns and subcellular localization are specified (28 -30). In contrast to these PDE families, splice variants of cGB-PDE have not been reported yet.
In our previous studies, the amino-terminal portion of rat cGB-PDE was shown to be different from those of bovine and human cGB-PDEs, although two cGMP binding domains and one catalytic domain were highly conserved among bovine, human, and rat cGB-PDEs (12,13,18). A cGB-PDE possessing the rat type of amino-terminal sequence had not been isolated previously from bovine and human lung cDNA libraries (13,18). Furthermore, cGB-PDE containing the sequence of the bovine type of cGB-PDE was not isolated from a rat lung cDNA library (12). In general, amino-terminal sequences of proteins, especially in cases where they are included in signal sequences, are less conserved among various species (31). However, bovine cGB-PDE has no signal sequence (18). It is expected that the amino-terminal differences among bovine, rat, and human cGB-PDEs result from alternative splicing.
In this work, we report the initial cloning of the cDNA encoding canine cGB-PDEs and the existence of amino-terminal variants of canine cGB-PDE cDNAs, which are thought to be derived from alternative splicing of the cGB-PDE gene. We * The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" 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 GenBank TM /EBI Data Bank with accession number(s) AB008467 and AB008468.
To isolate the complete cDNA of canine cGB-PDE, library screening was performed using standard methods. A canine lung cDNA library was constructed from canine lung mRNA using a Time Saver cDNA synthesis kit, gt10, and a Gigapack Packaging Extract kit (Stratagene) according to the manufacturer's instructions. A 32 P random primer labeling kit (Takara Shuzo) was used to radiolabel the two DNA fragments mentioned above. Plaques of the canine cDNA phage library plated onto 20 plates at 1.5 ϫ 10 5 plaque-forming units/plate were lifted using Hybond-Nϩ membrane and then screened by hybridization with the 32 P-labeled probes in 6 ϫ SSC (1 ϫ SSC is 0.15 M NaCl, 15 mM sodium citrate, pH 7.0), 0.5% SDS, 5 ϫ Denhardt's solution (1 ϫ Denhardt's is 0.02% each of bovine serum albumin, polyvinylpyrrolidone, and Ficoll 400), and 100 g/ml salmon sperm DNA at 65°C for 16 h. The filters were washed twice in 2 ϫ SSC, 0.5% SDS at room temperature for 10 min followed by a 30-min wash in 0.1 ϫ SSC, 0.5% SDS at 65°C. The filters were exposed to x-ray film at Ϫ70°C for 1 day. Plaques that hybridized to the probe were purified by two rounds of replating and rescreening. Inserted DNAs were subcloned into the EcoRI site of pBluescript II SK(ϩ) and used for sequencing.
Northern Blot Analysis-Male canines at 56 weeks of age were anesthetized with sodium pentobarbital before various tissues were excised and stored in liquid nitrogen until use. Total RNAs (20 g/lane) isolated from frozen tissues by acid guanidium-phenol-chloroform extraction (32) were subjected to electrophoresis in 1% agarose and 0.66 M formaldehyde gels. Gels were stained with ethidium bromide and photographed. The RNA fractions were transferred onto Hybond-Nϩ nylon membrane and fixed by ultraviolet irradiation using a UV cross-linker (Stratalinker, Stratagene). A 32 P-labeled DNA probe was prepared using cDNA encoding the catalytic domain of canine cGB-PDE (amino acid residues 610 -756 of CFPDE5A1) obtained by RT-PCR as described above. Hybridization was performed in 50% formamide, 4 ϫ SSC, 0.5% SDS, 5 ϫ Denhardt's, 100 g/ml salmon sperm DNA, and the probe at 42°C for 16 h. All blots were washed finally in 0.2 ϫ SSC and 0.1% SDS at 60°C for 1 h. The membranes were exposed to x-ray film at Ϫ70°C for 10 days.
RT-PCR and Southern Blot Analysis-To detect CFPDE5A1 and CFPDE5A2 mRNAs specifically in various canine tissues, RT-PCR was performed on 1 g of total RNAs prepared for Northern blot analysis as described above. A reverse transcriptase reaction was carried out using random hexamers at 42°C for 60 min according to the manufacturer's instruction for the GeneAmp RNA PCR Core kit (Perkin-Elmer). After the reverse transcriptase reaction, a cDNA fragment encoding CFPDE5A1 (amino acid residues 25-249) was amplified using the 5Јprimer 5Ј-ATGATCACCGGGACTTCACCTTCTC-3Ј and the 3Ј-primer 5Ј-TAATTTGGTCAACTTCTGCATTGAA-3Ј. The CFPDE5A2 cDNA fragment (amino acid residues 1-216 in addition to a 5Ј-untranslated region of 21 bp) was produced using the 5Ј-primer 5Ј-CCCCAACGT-TCTGTGCTCGCTATGTTGCCC-3Ј and the 3Ј-primer 5Ј-TAATTTGGT-CAACTTCTGCATTGAA-3Ј. PCR was carried out through 30 and 40 cycles for CFPDE5A1 and CFPDE5A2 cDNA, respectively, under conditions that each PCR amplification did not reach saturation. The reaction cycle was at 94°C for 1 min, at 52°C for 1 min, and at 72°C for 3 min. The PCR products were subjected to 2% agarose gel electrophoresis, and the fractions were transferred onto Hybond-Nϩ nylon membrane and fixed by ultraviolet irradiation. To detect both PCR products by Southern blot analysis, a probe cDNA encoding a part of canine cGB-PDE (amino acid residues 42-150 of CFPDE5A1), which does not contain the primer sequences for RT-PCR mentioned above, was obtained by PCR using the 5Ј-primer 5Ј-AAAAACTCGAGAGAAATGGT-CAATGCCTGGTTCGC-3Ј and the 3Ј-primer 5Ј-CTCGAGCACTGGTC-CCCTTCATCA-3Ј, with pBcPDE5-1 as a template. Hybridization was performed in 6 ϫ SSC, 0.5% SDS, 5 ϫ Denhardt's, 100 g/ml salmon sperm DNA, and the 32 P-labeled probe at 65°C for 16 h. All blots were washed finally in 0.1 ϫ SSC and 0.1% SDS at 65°C for 1 h. The membranes were exposed to x-ray film at room temperature for 20 min (CFPDE5A1 cDNA) or 4 h (CFPDE5A2 cDNA). Thus, signals of CFPDE5A2 transcripts were amplified strongly compared with those of CFPDE5A1 (see Fig. 4).
Each analysis included canine ␤-actin mRNA as a control for the RNA quality and quantity among different tissue samples. ␤-Actin cDNA fragment was amplified by PCR using the 5Ј-primer 5Ј-TTA-AGCTTGTAACCAACTGGGACGATATGG-3Ј and the 3Ј-primer 5Ј-AG-AAGCTTGATCTTGATCTTCATGGTGCTAGG-3Ј.
Generation of Antiserum-A rabbit polyclonal antiserum was raised against the poly-L-lysine-based multiple antigen peptide complex containing a 20-amino acid-long synthetic peptide of the sequence (NSPGNQILSGLSIEEYKTTL; amino acid residues 684 -703 of CFPDE5A1), which was derived from a portion in common with two types of canine cGB-PDEs. The multiple antigen peptide (MAP) complex was synthesized by the peptide synthesizer PSSM-8 (Shimadzu, Japan) using MAP resin TAKO8-WTGS (Shimadzu, Japan), and then mixed with Freund's complete adjuvant (Difco) for the first immunization and with Freund's incomplete adjuvant when boosted. After immunizing two Japanese White rabbits (Kitayama Laboratories, Japan) four times, antiserum was collected, and anti-cGB-PDE antibody IgG fraction was purified from the antiserum by affinity chromatography on a protein G column (Amersham Pharmacia Biotech). The antibody was eluted with 0.1 M glycine-HCl buffer, pH 2.7. The eluate was neutralized immediately with 1 M Tris-HCl, pH 9.0, and stored at Ϫ20°C until use.
Construction of Expression Plasmids-The 2905-bp SacI (blunt-ended)-XbaI cDNA fragment of pBcPDE5-1 was subcloned into XhoI (bluntended)-XbaI sites of pSVL expression plasmid, resulting in pSVL-CB. The 4228-bp XhoI-XbaI cDNA fragment of pRcPDE5-10 was ligated into the XhoI-XbaI sites of pSVL, resulting in pSVL-CR. To generate the amino-terminal truncated form of the canine cGB-PDE, PCR was performed using the 5Ј-primer 5Ј-AAAAACTCGAGAGAAATGGTCAAT-GCCTGGTTCGC-3Ј, the 3Ј-primer 5Ј-TAATTTGGTCAACTTCTGCAT-TGAA-3Ј (corresponding to amino acid residues 10 -216 of CFPDE5A2), and pSVL-CR as a template. The amplified cDNA fragment was digested with XhoI, and then the 300-bp XhoI cDNA fragment was subcloned into the XhoI site of the pBluescript II (SKϩ), and used for sequencing. After the XhoI cDNA fragment encoding the amino-terminal segment (amino acid residues 1-109 of CFPDE5A2) was deleted from pSVL-R, the 300-bp XhoI fragment obtained from the PCR product described above was inserted into the XhoI site of the deleted pSVL-CR, generating pSVL-CT.
Expression of cGB-PDE in COS-7 Cells-COS-7 cells were grown in Dulbecco's modified Eagle's medium (Life Technologies, Inc.) supplemented with 10% heat-inactivated fetal calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin at 37°C in 10% CO 2 , and were serially passaged before reaching confluence. The pSVL-CB, pSVL-CR, and pSVL-CT expression plasmids were transfected into COS-7 cells by the LipofectAMINE Reagent (Life Technologies, Inc.) according to the manufacturer's instructions. 48 h after transfection, cells were washed with ice-cold phosphate-buffered saline and scraped in an ice-cold homogenization buffer (20 mM Tris-HCl, pH 7.4, 2 mM magnesium acetate, 0.3 mM CaCl 2 , 1 mM dithiothreitol, 40 M leupeptin, 1.3 mM benzamidine, 0.2 mM phenylmethylsulfonyl fluoride, and 1 mM NaN 3 ). Freezethaw lysates were centrifuged at 100,000 ϫ g for 60 min to obtain cytosolic and particulate fractions. There was barely detectable latent lactate dehydrogenase activity (Ͻ5% of total activity) present in the particulate fraction, indicating that disruption of the cells was complete. cGMP hydrolytic activity in the lysate of mock transfected COS-7 cells was barely detectable, compared with those of pSVL-CB-and pSVL-CR-transfected cells.
Immunoblotting-5-15 l of cytosolic or particulate fraction prepared from transfected COS-7 cells was subjected to 10% SDS-polyacrylamide gel electrophoresis and transferred onto polyvinylidene difluoride membranes (Millipore). After blocking with 4% Block Ace (Snow Brand Milk Products, Tokyo, Japan) overnight, the blots were incubated with anti-cGB-PDE polyclonal antibody IgG fraction, biotinconjugated anti-rabbit IgG antibody, and then avidin-biotin peroxidase. Visualization of bands was performed by the ECL Western blotting detection reagents (Amersham Pharmacia Biotech).
PDE and Protein Assay-The PDE assay was performed by the radiolabeled nucleotide method (33). The assay buffer contained 50 mM Tris-HCl, pH 8.0, 5 mM MgCl 2 , 4 mM 2-mercaptoethanol, 0.33 mg/ml fatty acid-free bovine serum albumin (Sigma), 1 M unlabeled cGMP or cAMP, and 22 nM [ 3 H]cGMP or [ 3 H]cAMP. The reaction was started by mixing 0.1-10 l of enzyme solution into 500 l of assay buffer, and tubes were incubated at 37°C for 30 min. After boiling for 1.5 min, the mixtures were added to 100 l of 1 mg/ml C. atrox snake venom and incubated at 37°C for 30 min. The reaction was stopped by the addition of 500 l of methanol, and the resultant solutions were applied to a Dowex 1X8 -400 column. Aqueous scintillation mixtures were added to each elute and the radioactivity measured. The protein concentration of the cytosolic fractions of transfected COS-7 cells was determined by a Dc protein assay kit (Bio-Rad) using bovine serum albumin as a standard.
Relative V max Determinations-Relative V max values were determined according to the methods of McPhee et al. (22). Relative concentrations of cGB-PDE proteins expressed in COS-7 cells were calculated by immunoblotting as described above. The membranes were incubated with ECL reagents at room temperature for 1 min and then exposed to x-ray film for 10 -30 s, under conditions that each exposure to x-ray film did not reach saturation. The resultant films were scanned by the ScanJet 4c (Hewlett Packard) and DeskScan II program and quantitated using the NIH Image program. The optical densities versus amount of pSVL-CR-encoded protein were plotted to measure the relative concentrations of cGB-PDE proteins. Relative V max values were calculated from Lineweaver-Burk plots (34) using proteins that provided relatively equal enzymatic activity.
cGMP Binding Assay-The cGMP binding assay was performed in a total volume of 20 l by a modified version of the method described previously (35,36). 10 l of the cytosolic extract of transfected COS-7 cells was mixed with 10 l of an assay buffer to a final concentration of 10 mM sodium phosphate, pH 6.8, 4 mM EDTA, 25 mM 2-mercaptoethanol, 10 M 8-bromo-cGMP, 10 M cAMP, and 66 nM [ 32 P]cGMP (4,000,000 cpm/assay) in the presence or absence of 0.2 mM IBMX, followed by incubation at 0°C for 2 h. The reaction was stopped by the addition of 1 ml of ice-cold wash buffer (10 mM sodium phosphate, pH 6.8, and 1 mM EDTA) and then applied to Millipore HAWP filters (pore size 0.45 m). The filters were washed three times with 3 ml of ice-cold wash buffer and then counted on a scintillation counter. In all experiments, nonspecific binding was measured by incubation in the presence of 50 M unlabeled cGMP.

RESULTS
cDNA Cloning of Canine cGB-PDE-To investigate whether the cGB-PDE family has different gene products and alternative splice variants, we constructed a canine lung cDNA phage library and isolated canine cGB-PDE cDNAs from the library. To obtain two probes for the isolation of the full-length canine cGB-PDE cDNA, we performed RT-PCR using several oligonucleotide primers derived from bovine and rat cGB-PDE cDNA sequences and canine lung mRNA as a template. Specifically amplified 800-bp and 440-bp DNA fragments encoding parts of canine cGB-PDE were used to screen a canine lung cDNA phage library. Because we expected that variant forms of cGB-PDE possessing unique amino-terminal sequences existed in the canine lung library, many more phage clones were screened than those used in standard methods. Inserted cDNAs from six distinct positive clones were subcloned into pBluescript II SK(ϩ). Sequence analysis of their subclones revealed that all inserts possessed an open reading frame, and that two types of cGB-PDEs were taken from one canine lung cDNA library. Five of six clones carried a unique amino-terminal sequence similar to those of bovine and human cGB-PDEs, and the other possessed one similar to that of rat cGB-PDE. The bovine type of canine cGB-PDE cDNA encoded by pBcPDE5-1 was found to be 2905 bp in length and had an open reading frame of 2595 bp. The rat type cGB-PDE carried by pRcPDE5-10 was 4228 bp in length and possessed an open reading frame of 2499 bp. The pBcPDE5-1 encoded a protein of 865 amino acids with a predicted molecular mass of 98,292 Da, which is designated as CFPDE5A1 using accepted nomenclature (6), and pRcPDE5-10 encoded a protein, CFPDE5A2, of 833 amino acids with a predicted molecular mass of 94,748 Da (Fig. 1A). The aminoterminal portion of CFPDE5A1 (Met 1 -Thr 40 ) was larger than that of CFPDE5A2 (Met 1 -Arg 8 ), and their sequences were extremely different. Although triplet glutamine was observed in the amino-terminal portions of bovine and human cGB-PDEs, there was no triplet glutamine cluster in the unique amino terminus of bovine type of canine cGB-PDE (CFPDE5A1). Except for the unique amino-terminal portions, the sequence of CFPDE5A1 (Arg 41 -Asn 865 ), including cGMP binding domains and the catalytic domain, was identical to that of CFPDE5A2 (Arg 9 -Asn 833 ) (Fig. 1, A and B). Both proteins contained a potential phosphorylation site (Ser 92 and Ser 60 in the CFPDE5A1 and CFPDE5A2 proteins, respectively). The predicted amino acid sequences of the cGB-PDEs were highly conserved among bovines, humans, rats, and canines (Fig. 1A).
Expression of Canine cGB-PDE mRNA in Various Tissues-Expression patterns of two types of canine cGB-PDE transcripts were examined in various canine tissues by Northern blot analysis. The blots of total RNAs (20 g each) isolated from 56-week-old male canines were hybridized with the 32 P-labeled cDNA encoding the canine cGB-PDE catalytic domain to detect two types of cGB-PDE mRNAs. A 32 P-labeled cDNA probe strongly hybridized to the transcripts from the spleen, colon, pylorus, and thoracic artery. Moderate expressions of cGB-PDE transcripts were observed in the cerebellum, hippocampus, retina, lung, atria, left ventricle, right ventricle, kidney, and adrenal gland. (Fig. 2). CFPDE5A1 and CFPDE5A2 transcripts were not detectable separately in Northern blot analysis using their specific DNA probes encoding unique amino-terminal portions and 5Ј-untranslated regions. Therefore, we performed RT-PCR to distinguish between CFPDE5A1 and CFPDE5A2 transcripts in canine tissues. To know the relative amounts of CFPDE5A1 and CFPDE5A2 transcripts, the efficiency of PCR amplification using specific primer sets for CFPDE5A1 and CFPDE5A2 was first examined. PCR analysis was performed using the same amounts of pSVL-CB (CFPDE5A1) and pSVL-CR (CFPDE5A2) DNAs as a template under the same conditions in RT-PCR analysis, revealing that amplification using the primer set for CFPDE5A2 was more efficient than that using the primer set for CFPDE5A1 (Fig. 3A). In RT-PCR analysis using canine lung total RNA, under conditions that each reaction did not reach saturation, the PCR product derived from CFPDE5A1 transcripts was detected sufficiently by ethidium bromide staining. By contrast, the PCR product from CFPDE5A2 transcripts was barely detectable by the staining (Fig. 3B). These findings indicated that the amount of CFPDE5A1 transcripts was much greater than that of CFPDE5A2 in canine lung. To detect a small amount of CFPDE5A2 transcripts, RT-PCR and Southern blot analysis were performed using total RNAs from various tissues, resulting in the presence of the two independent, alternatively spliced CFPDE5A transcripts in various ca- nine tissues (Fig. 4). CFPDE5A2 transcripts were observed in the cerebellum, hippocampus, retina, lung, heart, spleen, and thoracic artery, whereas CFPDE5A1 transcripts were highly produced in the pylorus in addition to tissues producing CFPDE5A2 transcripts. Moreover, CFPDE5A2 transcripts were detectable in the cerebrum, pancreas, kidney, adrenal gland, and pylorus at a low level, whereas CFPDE5A1 transcripts were observed in the liver and colon in addition to tissues expressing CFPDE5A2 transcripts. The expression pattern of CFPDE5A1 transcripts was similar to that of CFPDE5A2 in most tissues. However, because CFPDE5A2 transcripts were detected by a combination of RT-PCR and Southern blot analysis with superior sensitivity compared with CFPDE5A1 transcripts (see "Experimental Procedures"), the amounts of CFPDE5A1 transcripts in canine tissues were suggested to be much greater than those of CFPDE5A2.
Phosphodiesterase Activities for cGMP and cAMP in Cytosolic Extracts of Transfected COS-7 Cells-To investigate the enzymatic properties of CFPDE5A1 and CFPDE5A2, expression studies were performed on COS-7 cells. The CFPDE5A1 and CFPDE5A2 cDNA fragments from pBcPDE5-1 and pRcPDE5-10 were introduced to the COS-7 cell expression vector pSVL to generate the recombinant plasmids pSVL-CB and pSVL-CR, respectively. Cytosolic extracts were prepared from transfected COS-7 cells and assayed for cyclic nucleotide hydrolytic activities using either 1 M cGMP or 1 M cAMP. COS-7 cells expressing CFPDE5A1 and CFPDE5A2 had approximately 200 -400-fold higher levels of cGMP hydrolytic activities compared with those of mock transfected cells (Fig.  5A). Both PDE5As were more specific for cGMP hydrolysis than cAMP on this condition (Fig. 5A).
cGMP Binding to CFPDE5A1 and CFPDE5A2 Expressed in COS-7 Cells-We investigated the cGMP binding activities in cytosolic extracts from transfected COS-7 cells. In general, cGMP binding has been reported in not only cGB-PDE but also cyclic nucleotide-dependent protein kinase (37) and the cGMP channel in various tissues (38). In this experiment, 8-bromo-cGMP and cAMP were included in the assay buffer to inhibit the cGMP binding to cyclic nucleotide-dependent protein kinase. Cytosolic extracts containing CFPDE5A1 and CFPDE5A2 exhibited increased amounts of the cGMP binding activity, compared with that of mock transfected cells (Fig. 5B). The inhibitor of cGMP hydrolysis, IBMX, stimulated the binding to both PDE5As, which is a typical property of cGB-PDE as indicated previously (18,35). Two types of PDE5As exhibiting the same enzymatic activities showed similar cGMP binding activities.
Subcellular Distribution of the CFPDE5A1 and CFPDE5A2 Proteins-To investigate the biological role of two types of amino-terminal sequences, the plasmid pSVL-CT coding for an amino-terminal truncated form (designated as CFPDE5A-T) of CFPDE5A1 and CFPDE5A2 was produced from pSVL-CB by deleting nucleotides encoding the amino-terminal 42 amino acids of the CFPDE5A1. The codon of the first Met in the truncated form, corresponding to Met 43 in CFPDE5A1 or Met 11 in CFPDE5A2, would be expected to be a starting point of the translation. To determine the subcellular distribution of CFPDE5A1 and CFPDE5A2, homogenates of COS-7 cells expressing these proteins were separated into cytosolic and particulate fractions using ultracentrifugation. cGMP hydrolytic activities derived from CFPDE5A1 and CFPDE5A2 that were expressed in COS-7 cells were observed in both fractions. The distribution ratios of their activities in the cytosolic and particulate fractions were 74 -88% and 12-26% for CFPDE5A1, and 85-86% and 14 -15% for CFPDE5A2, respectively. CFPDE5A-T showed a similar distribution ratio of cGMP hydrolytic activity.  2) as a template and the same primers as RT-PCR analysis (see "Experimental Procedures"). PCR amplification was carried out through 30 and 40 cycles for CFPDE5A1 and CFPDE5A2 cDNA, respectively. Amplified products were subjected to 2% agarose gel electrophoresis, and the fractions were detected with ethidium bromide staining. Panel B, RT-PCR analysis for CFPDE5A1 (lane 1) and CFPDE5A2 (lane 2) was performed on 1 g total RNA isolated from canine lung, as described under "Experimental Procedures." PCR products were detected with ethidium bromide staining.
A polyclonal antibody against a synthetic peptide corresponding to a part of the CFPDE5A protein was produced and used for immunoblot analysis of CFPDE5A1 and CFPDE5A2 expressed in COS-7 cells. This analysis demonstrated that CFPDE5A1 and CFPDE5A2 in cytosolic fractions had molecular masses of approximately 110 and 94 kDa, respectively (Fig.  6). The value of 110 kDa was higher than that estimated by the predicted amino acid sequence of CFPDE5A1 (98 kDa). By contrast, the apparent molecular mass of immunodetected CFPDE5A2 was similar to that predicted from the amino acid sequence. CFPDE5A1 and CFPDE5A2 expressed in particulate fractions showed the same mobilities as were found in cytosolic fractions.
Kinetic Properties of the CFPDE5A1 and CFPDE5A2 Enzymes-K m values were derived from Lineweaver-Burk plots of cGMP substrate for CFPDE5A1, CFPDE5A2, and CFPDE5A-T, which appeared in the cytosolic fractions of COS-7 cells (Fig. 7). CFPDE5As in all of these fractions exhibited levels of K m similar to those of purified cGB-PDEs from other animals as reported previously (5). However, the K m value for CFPDE5A1 (2.68 Ϯ 0.29 M) was significantly higher than those of CFPDE5A2 (1.97 Ϯ 0.057 M) and the CFPDE5A-T (1.73 Ϯ 0.17 M). The V max value relative to that of CFPDE5A2 was 0.48 Ϯ 0.032 for CFPDE5A1 and 0.70 Ϯ 0.082 for CFPDE5A-T (Table I). Although these differences in K m and relative V max values were statistically significant for the three proteins, they are unlikely to be biologically meaningful.
Kinetic inhibition of the CFPDE5A1, CFPDE5A2, and CFPDE5A-T by Zaprinast, which is known as a specific inhibitor for cGB-PDE (9), was measured by using 1 M cGMP as a substrate. The concentrations of Zaprinast for 50% inhibition of cytosolic CFPDE5A1, CFPDE5A2, and CFPDE5A-T enzymatic activities were 0.53 Ϯ 0.036 M, 0.31 Ϯ 0.10 M, and 0.35 Ϯ 0.095 M, respectively (Table I). These values were very similar to those of bovine and rat cGB-PDEs as reported previously (5,12,18). Thus, there was no significant difference in the IC 50 values among the three proteins. DISCUSSION We have cloned and characterized rat and human cGB-PDEs previously (11,13). Most of the predicted amino acid sequence of rat cGB-PDE was shown to be homologous to those of bovine , under conditions that PCR amplification did not reach saturation. The PCR products were subjected to Southern blot analysis using the 32 P-labeled DNA probe, which detected both PCR products. The membranes were exposed to x-ray film at room temperature for 20 min (CFPDE5A1 cDNA) or 4 h (CFPDE5A2 cDNA). DNA fragments for ␤-actin (680-bp product) were detected with ethidium bromide staining (panel C).
and human cGB-PDEs (11,13,18), whereas the amino-terminal portion of rat cGB-PDE was extremely different from those of bovine and human cGB-PDEs. In this current study, we isolated two kinds of canine cGB-PDEs (CFPDE5A1 and CFPDE5A2) possessing different and unique amino-terminal portions from a canine lung cDNA library. From the data of exon-intron organization of the human cGB-PDE gene (13), a unique amino-terminal segment and the following common region of human cGB-PDE were separated by an intron. Therefore, two variants of mRNA coding for canine cGB-PDEs would be expected to be alternatively spliced. We report here for the first time that the cGB-PDE family has at least two alternative splice variants. As a result, rat cGB-PDE isolated previously should be designated RNPDE5A2 according to accepted nomenclature (6).
Northern blot analysis showed that canine cGB-PDEs transcripts were expressed in many tissues. Although rat cGB-PDE transcripts were not detectable in the hippocampus, heart, kidney, and adrenal gland in our recent study (12), canine cGB-PDE transcripts were observed in such tissues. The tissue distribution of canine cGB-PDEs transcripts was not identical to that of rat cGB-PDE. The result of semiquantative RT-PCR analysis and the fact that five of six clones isolated from a canine lung cDNA library were CFPDE5A1 suggested that CFPDE5A1 mRNA is a major form in canine tissues. Purification of canine cGB-PDE protein from canine lungs according to the methods of Francis et al. (14) indicated that a major component from canine lung migrated as an approximately 110-kDa species (data not shown), which is the same molecular mass as that found in cytosolic extracts expressing CFPDE5A1. RT-PCR analysis showed that the tissue-specific expression pattern of CFPDE5A2 transcripts was not coincident with that of rat cGB-PDE transcripts, although CFPDE5A2 was the same variant form as rat cGB-PDE (RNPDE5A2). The expression of PDE5A2 transcripts in some tissues may be distinctive among various animals. The expression pattern of cGB-PDE transcripts in canine tissues suggests that cGMP may play important roles not only as a vasodilator and a neurotransmit-ter, but also as a mediator of other physiological processes in various tissues. The existence of different splice variants in canine and rat tissues suggests that distinct promoters and alternatively splicing factors regulate expression of both cGB-PDE transcripts in various tissues and different animals.
In numerous instances alternative splice variants in other PDE families have been reported to show unique kinetic and physical characteristics, and in some instances the subcellular localization of a PDE has been shown to be determined by the amino-terminal sequences. For example, one form of PDE4 FIG. 6. Detection of CFPDE5A1 and CFPDE5A2 in cytosolic and particulate fractions of COS-7 cells expressing two splice variants by immunoblotting. An anti-peptide antibody against canine cGB-PDE was raised and purified by affinity column as described under "Experimental Procedures." After cytosolic and particulate extracts of COS-7 cells expressing CFPDE5A1 and CFPDE5A2 were prepared, protein fractions containing the same amounts of enzymatic activity were subjected to 10% SDS-PAGE. Gels were used for electric transfer of proteins to polyvinylidene difluoride membranes for immunoblotting. The same result has been obtained with three different protein preparations.  variants possessing a unique amino-terminal sequence, named RD1, was known as a membrane-bound protein (29,30), whereas the other forms of PDE4 were localized both in cytosolic and particulate fractions. Because the canine cGB-PDEs reported here have two different amino-terminal sequences, canine cGB-PDE isoforms expressed in COS-7 cells might be expected to be observed in different subcellular fractions of COS-7 cells. However, the CFPDE5A1 and CFPDE5A2 proteins were found in both cytosolic and particulate fractions of COS-7 cells, suggesting that the amino-terminal moiety did not specify the subcellular localization of the cGB-PDE proteins. Previously, cGB-PDE activity has been reported to be present in the cytosolic fractions in several kinds of cells and various tissues (5,8,9,14,15). Immunoblot analysis in our study demonstrated that the cGB-PDE proteins were located in both cytosolic and particulate fractions of canine and rat platelets, and its activities were also found in both fractions (data not shown), suggesting that the appearance of the cGB-PDE proteins in the particulate fraction of COS-7 cells was not attributed to artificial overexpression. Regarding the PDE3 and PDE4 families, their activities of membrane-bound forms were regulated by several factors (27,28). Although cGB-PDE has neither transmembrane domains nor membrane-associated signals, it is possible that the cGB-PDE activity in the particulate fraction is regulated functionally in association with membrane. Immunoblotting demonstrated that the CFPDE5A1 protein expressed in both fractions of COS-7 cells migrated as a 110-kDa species, which is the same molecular mass as that of purified cGB-PDE from canine lung as described above. This value was higher than that estimated by the predicted amino acid sequence of CFPDE5A1. It possesses a potential myristoylation site in its unique amino-terminal portion, but not CFPDE5A2, suggesting that the CFPDE5A1 protein is subjected to post-translational modification.
cGB-PDE is composed of two allosteric cGMP binding domains and one catalytic domain (17,18). When cGMP binding sites are occupied by a substrate, cAMP-and cGMP-dependent protein kinases can then phosphorylate a specific serine residue of cGB-PDE (5,17,39). Partially purified cGB-PDE from guinea pigs was activated by the catalytic subunit of cAMP-dependent protein kinase (40,41). However, it has been unclear whether phosphorylation of cGB-PDE is directly related to its activation. Photoreceptor PDEs (PDE6), which are homologous to cGB-PDE, are known as membrane-associated proteins and are inhibited by ␥ subunits of PDE6 on retinal membrane (5). The PDE6 ␥ subunit is also noted to prevent the activation of cGB-PDE by cAMP-dependent protein kinase (42). It was reported that a form of the PDE4 family, named RNPDE4A5, interacted with SH3 domain of Src tyrosyl protein kinase, and that its activity is regulated by such proteins (28). It is possible that two distinct amino-terminal sequences of canine cGB-PDEs could be involved in differential regulation of the activities. To know whether amino-terminal diversity is related to biological differences between two canine cGB-PDEs or not, we investigated the enzymatic characteristics of two types of cGB-PDEs expressed in the cytosolic fraction of COS-7 cells. Analysis of enzymatic properties of the two cGB-PDEs expressed in COS-7 cells demonstrated that the cytosolic form of CFPDE5A1 showed a 1.5-fold higher K m value and a somewhat lower relative V max value compared with those of CFPDE5A2. Kinetic properties, especially the K m value, of a truncated form at the amino terminus, were similar to those of CFPDE5A2, and the IC 50 value for Zaprinast of CFPDE5A1 is of the same order of magnitude as that of CFPDE5A2. Qualitatively, the ratio of cGMP binding to catalytic activity was comparable between CFPDE5A1 and CFPDE5A2 under the experimental conditions used, and in both enzymes, the cGMP binding was increased by IBMX. However, the relative affinity and stoichiometry of cGMP binding were not assessed. The kinetic differences in catalytic characteristics, although small in magnitude, are reproducible and statistically significant. However, we consider that there is no evidence in these experiments to support the position that unique amino-terminal portions of cGB-PDEs are associated with biologically meaningful differences in enzymatic properties of the two alternative splice variants.
Thus, we demonstrated for the first time two types of canine cGB-PDEs and their characterizations. Although we found differences of expression patterns in some tissues between two types of variants by Northern blot and RT-PCR analysis, there were no functional and physiological differences between the two cGB-PDEs in this study. However, the discovery of alternative splice variants in the cGB-PDE family could open exciting new avenues of research in defining the regulation and interactions of cGB-PDE.