The structure of a bovine lung cGMP-binding, cGMP-specific phosphodiesterase deduced from a cDNA clone.

Polymerase chain reaction (PCR) methodology and cDNA library screening were used to isolate a cDNA clone encoding a cGMP-binding, cGMP-specific phosphodiesterase (cGB-PDE) from bovine lung. Degenerate oligonucleotides based on cGB-PDE peptide sequences were used as primers for a PCR reaction with bovine lung cDNA as the template. An 824-base pair PCR product was recovered and used as a probe to screen a bovine lung cDNA library. A 4.5-kilobase pair cDNA clone encoding a full-length cGB-PDE was isolated. The open reading frame of this cDNA predicted an 875 amino acid (AA), 99,525-Da polypeptide. By Northern analysis, the cGB-PDE cDNA hybridized to a single lung 6.9-kilobase mRNA. The identity of the cGB-PDE cDNA was verified by comparison of the deduced AA sequence with several peptide sequences obtained from cGB-PDE. COS-7 cells transfected with cGB-PDE cDNA overexpressed cGMP-binding and cGMP-PDE activities characteristic of lung cGB-PDE. The sequence of cGB-PDE contained a segment (AA 578-812) that was homologous to the putative catalytic region conserved among all mammalian PDEs and a segment (AA 142-526) that was homologous to the putative cGMP binding region of the cGMP-stimulated PDE and the photoreceptor PDEs. As noted also for these PDEs, two internally homologous repeats were contained within the putative cGMP binding region of cGB-PDE. The amino-terminal 142 residues of cGB-PDE showed no significant homology to other PDEs and contained the serine (AA 92) which is phosphorylated by cGMP-dependent protein kinase.


228
Cyclic nucleotide phosphodiesterases (PDEs)' constitute a complex family of enzymes which catalyze the hydrolysis of 3':5'-cyclic nucleotides to the corresponding nucleoside 5'monophosphates. The multiple PDEs differ in their amino acid (AA) sequences, physicochemical properties, substrate specificities, sensitivities to inhibitors, immunological reactivities, and modes of regulation. Comparison of the AA sequences of PDEs has led to the proposal that most PDEs are chimeric multidomain proteins, possessing distinct catalytic and regulatory domains (1). A segment of sequence approximately 250 AA residues in length is conserved among all mammalian phosphodiesterases characterized to date and is located in the more carboxyl-terminal portion of the PDE molecules. This segment is hypothesized to contain the catalytic site (1, 2), and limited proteolysis studies aimed at mapping discrete functional domains of several PDEs have supported this interpretation (1,3,4). Domains of the PDE molecules which interact with allosteric/regulatory molecules are thought to be located within the more amino-terminal regions (1,4,5).
The cGMP-binding, cyclic nucleotide phosphodiesterases comprise a heterogeneous subfamily of PDEs, all of which exhibit cGMP-binding sites that are distinct from sites of cyclic nucleotide hydrolysis. Although the members of this subfamily share several similar biochemical properties, they differ in many respects including substrate specificities, subunit compositions, tissue distributions, mechanisms of regulation of enzymatic activity, and physiological roles. The cGMP-binding PDEs can be categorized into at least three subgroups: the widely distributed cGMP-stimulated cyclic nucleotide phosphodiesterases (cGS-PDEs), which most likely exist as homodimers of 100-105-kDa subunits and which hydrolyze both CAMP and cGMP (6); the cGMPspecific photoreceptor phosphodiesterases (rod outer segment phosphodiesterase (ROS-PDE), whose holoenzyme structure aPyP is composed of two large subunits (Y and p, both of which are catalytically active, and two smaller y subunits (7,8), and The abbreviations used are: PDEs, 3':5'-cyclic nucleotide phosphodiesterases; cGB-PDE, cGMP-binding, cGMP-specific phosphodiesterase; cGS-PDE, cGMP-stimulated phosphodiesterase; ROS-PDE, rod outer segment phosphodiesterase; CONE-PDE, cone phosphodiesterase; cAK, CAMP-dependent protein kinase; cGK, cGMPdependent protein kinase; PAGE, polyacrylamide gel electrophoresis; PCR, polymerase chain reaction; ssDNA, single-stranded DNA; MOPS, 3-[N-morpholino]propanesulfonic acid; bp, base pair; kb, kilobase; ORF, open reading frame; CaM-PDEs, calcium/calmodulinstimulated phosphodiesterases; cGI-PDE, cGMP-inhibited phosphodiesterase; MeOxMeMIX, 3-isobutyl-1-methyl-1-methoxymethylxanthine; IBMX, 3-isobutyl-1-methylxanthine; CAP, E. coli catabolite activator protein; cpm, counts/min; HPLC, high performance liquid chromatography; pfu, plaque-forming unit(s). 163 22864 cGMP-binding, cGMP-specific Phosphodiesterase cone phosphodiesterase (CONE-PDE) (9), which is composed of two identical a' subunits and three smaller subunits); and the cGMP-binding, cGMP-specific phosphodiesterase (cGB-PDE), which has been shown to be a homodimer comprised of two 93-kDa subunits (10). Comparison of the deduced amino acid sequences of cGS-PDE (11), the a and / 3 subunits of 13), and the a' subunit of , reveals that each sequence possesses, in addition to a homologous catalytic region located near the carboxyl terminus, a second conserved segment of approximately 340 residues located closer to the amino terminus (5,13). This additional conserved segment is not present in any non-cGMP-binding PDEs and has been proposed to constitute an allosteric, cGMP binding region (1,5). Photoaffinity labeling studies designed to map the catalytic and cGMP-binding sites of cGS-PDE support this hypothesis (3, 15). Full-length amino acid sequence has not been previously reported for cGB-PDE, but it has been predicted that its sequence contains a segment homologous to the putative cGMP binding regions of the cGS-PDE and photoreceptor PDEs (5).
cGB-PDE has been purified to homogeneity from rat (16) and bovine lung (10) and has been shown to be present in a variety of tissues and species including rat and human platelets (17), rat spleen (18), guinea pig lung (19), vascular smooth muscle (20), and sea urchin sperm (21). The enzyme exhibits specific cGMP-hydrolytic (K, = 5 p~) and cGMP-binding (half-maximal = 0.2 p~) activities (10) and is believed to be a chimeric multidomain protein. This proposal is supported by the finding that DEAE chromatography of a partial achymotryptic digest of cGB-PDE separates cGMP-binding fragments from a cGMP hydrolytic fragment (10). In addition to cGMP-binding and hydrolytic domains, cGB-PDE has been shown to contain a single site which can be readily phosphorylated by cGMP-dependent protein kinase (cGK) and, with a lower rate, by CAMP-dependent protein kinase (cAK) (22). The primary amino acid sequences of the phosphorylation site and of the amino-terminal end of the cGMPbinding fragments generated by chymotryptic digestion of cGB-PDE have been reported (10,22).
We report here the cloning of the cDNA and analysis of the deduced amino acid sequence of cGB-PDE. This deduced sequence, in combination with previous biochemical studies of the enzyme, provides new and important insights into the location and structure of individual domains of this enzyme, and into the structural relationship of cGB-PDE to other PDEs. This is an important step in ultimately understanding the mechanisms by which the enzymatic activity of cGB-PDE is regulated.

Methods
Purification of cGB-PDE and Catalytic Subunit of cAK-cGB-PDE was purified using a previougly described method (10) or a modification of that method. For all preparations, processing of bovine lung extract, DEAE-cellulose chromatography, and Blue Sepharose" CL-6B chromatography were performed as previously described. Zinc chelate affinity adsorbent chromatography was performed using either an agarose or Sepharose-based gel matrix. The resulting protein pool from the zinc chelation step was processed as previously described or was subjected to a modified purification procedure. If the modified procedure was performed, the protein pool was dialyzed against 20 mM sodium phosphate, pH = 6.8, with 2 mM EDTA and 25 mM P-mercaptoethanol (PEM), for 2 h, and loaded onto a 10-ml preparative DEAE-Sephacel column equilibrated in PEM buffer. The protein was eluted batchwise with 0.5 M NaCl in PEM, resulting in an approximately 10-15-fold concentration of protein. The concentrated protein sample was loaded onto an 800-ml (2.5 X 154 cm) Sephacryl S-400 gel filtration column equilibrated in 0.1 M NaCl in PEM, and eluted at a flow rate of 1.7 ml/min. The purity of the protein was assessed by Coomassie staining after sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE) (23). Approximately 0.5-3.0 mg of pure cGB-PDE were obtained per 10 kg of bovine lung.
Catalytic subunit of cAK was purified to homogeneity according to the method of Flockhart and Corbin (24).
Limited Amino Acid Sequence Analysis of cGB-PDE-The methods used to obtain the peptide sequences KISASEFDRPLR and TSPRFDNDEG(E/D)Q2 have been described elsewhere (10,22). The peptide sequences REXDANRINYMYAQYVKNTM3 and QSLAAXVVP3 were obtained during analysis of proteolytic fragments of phosphorylated cGB-PDE. Phosphorylation of cGB-PDE was performed prior to digestion in order to allow for 32P labeling of resultant phosphopeptides, but several non-labeled fragments were also isolated during these studies. For REXDANRINY-MYAQYVKNTM: approximately 100 pg of purified cGB-PDE was phosphorylated in a reaction mixture containing 9 mM MgC12, 9 p M [3ZP]ATP, 10 p~ cGMP, and 4.2 pg of purified catalytic subunit of cAK in a final volume of 900 pl. The reaction was incubated for 30 min at 30 "C and stopped by addition of 60 p1 of 200 mM EDTA. This solution was incubated for 30 min at 30 "C with 3.7 pl of a 1 mg/ml solution of a-chymotrypsin in 10 mM potassium phosphate, pH = 6.8, with 2 mM EDTA (KPE buffer). This limited proteolysis was stopped by addition of 50 pl of 10% SDS and 25 p1 of P-mercaptoethanol, and the volume of the sample was reduced to less than 400 p1 by boiling. The sample was loaded onto an 8% preparative SDSpolyacrylamide gel (23), followed by electrophoresis at 50 mAmps. Fragments were electroblotted onto Immobilon polyvinylidene difluoride (Millipore), identified by Coomassie Blue staining, and a non-32P-labeled 50-kDa band was excised from the membrane for automated gas-phase amino acid sequencing (25).
The peptide sequence QSLAAXVVP3 was obtained as follows: approximately 200 pg of purified cGB-PDE was added to 10 mM MgCI,, 10 p~ [32P]ATP, 100 p~ cGMP, and 1 pg/ml purified catalytic subunit of cAK (24) in a final volume of 1.4 ml. The reaction was incubated for 30 min at 30 "C and was terminated by the addition of 160 p1 of 0.2 M EDTA. Nine p1 of 1 mg/ml Staphylococcal aurew V8 protease in KPE were added, followed by a 15-min incubation at 30 "C. Proteolysis was stopped by addition of 88 p1 of 10% SDS and 45 pl of p-mercaptoethanol. The sample was boiled, and the digestion products were separated by electrophoresis on a preparative 10% SDS-polyacrylamide gel (23), run at 25 mAmps for 4.5 h. Proteins were electroblotted onto Immobilon and stained, and a n~n -~' Plabeled 28-kDa protein band was excised from the membrane for automated gas-phase amino acid sequencing (25).
Purified cGB-PDE was previously found to be resistant to aminoterminal amino acid sequencing (10). However, the peptide sequence AGPGSARPQQXWD3 was obtained by direct Edman degradations of approximately 0.5 pg of one preparation of purified cGB-PDE dissolved in 50% formic acid. (These results imply that partial proteolysis occurred 3 residues from the amino terminus.) Sequencing was performed on an Applied Biosystems Model 470A gas-phase sequencer equipped with an on-line model 120A phenylthiohydantoin-amino acid analyzer. This sequence was also obtained from cGB-PDE electroblotted onto a polyvinylidene difluoride membrane. Prior to electroblotting, 0.5 pg of purified cGB-PDE was phosphorylated under the conditions described in the previous paragraph. Phosphorylation was terminated by addition of approximately 0.01 volumes of 5% SDS in 7 M @-mercaptoethanol, followed by heating at 60 "C for 10 min. The sample volume was reduced by centrifugation under vacuum before electrophoresis.
The sequence GTREMVNAXFAERVHTP was obtained from a peptide purified after digestion of cGB-PDE with Achromobacter protease I. The digest of approximately 20 pg of cGB-PDE was performed in 200 pl of 2 M urea, 50 mM Tris-HC1, pH 8.5, using a The parentheses surrounding the E/D (position 11 in the peptide sequence) indicate that the automated gas-phase amino acid sequencing analysis yielded ambiguous results, and that either a D or E residue could have occupied this position in the amino acid sequence. A degenerate oligonucleotide was designed based on this sequence containing an E residue at position 11, and this oligonucleotide was used as a primer for production of the PCR clone (described under "Experimental Procedures"). However, subsequent sequencing of the cGB-8 cDNA revealed that the cGB-PDE contains a D residue at the corresponding position in the protein sequence.
X denotes an unidentified amino acid residue.

cGMP-binding, cGMP-specific Phosphodiesterase 22865
protease/substrate ratio of 1:50, for 12 h at 37 "C, followed by reduction and pyridylethylation as described previously (26). The resultant peptide mixture was purified by reversed-phase HPLC on a RP-300, C8 column (2.1 X 100 mm (Pierce Chemical Co.)) using a Hewlett-Packard model 1090 HPLC system. Production of an Initial cDNA Clone by the Polymerase Chain Reaction (PCR)-Degenerate oligonucleotides corresponding to the amino acid sequences NYMYAQYV and FDNDEG(E/D)Q2 (obtained from purified cGB-PDE as described above) were synthesized using an Applied Biosystems Model 380A DNA Synthesizer. After ethanol precipitation, oligonucleotides were combined at 400 nM each in a PCR containing 50 ng of first strand bovine lung cDNA, 200 p~ deoxynucleotide triphosphates, and 2 units of Taq polymerase. The initial denaturation step was carried out at 94 "C for 5 min, followed by 30 cycles of a 1-min denaturation step at 94 "C, a 2-min annealing step at 50 "C, and a 2-min extension step at 72 "C. PCR was performed using a Hybaid Thermal Reactor. The resulting PCR product (referred to as the "PCR clone") was purified by electrophoresis on a 1% low melting point agarose gel run in 40 mM Tris acetate, 2 mM EDTA, and isolated using the Gene Clean" DNA purification kit according to the manufacturer's protocol. The PCR product (20 ng) was ligated into the EcoRV site of the pBluescript KS (+) subcloning vector and sequenced.
Construction and Screening of cDNA Library-Library construction was performed using standard techniques (27). Specifically, polyadenylated RNA was prepared from bovine lung as described (11). First strand cDNA was synthesized using avian myeloblastosis virus reverse transcriptase with random hexanucleotide primers (27). Second strand cDNA was synthesized using Escherichia coli DNA polymerase I in the presence of E. coli DNA ligase and E. coli RNase H. Selection of cDNA >500 base pairs (bp) was performed by Sepharose' CL-4B chromatography (27). EcoRI adaptors were ligated to the cDNA using T4 DNA ligase. Following heat inactivation of the ligase, the cDNA was phosphorylated using T4 polynucleotide kinase. Unligated adaptors were removed by Sepharose" CL-4B chromatography (27). The cDNA was ligated into EcoRI-digested, dephosphorylated X-ZapIf arms and packaged with Gigapack" Gold extracts according to the manufacturer's protocol (Stratagene). The titer of the unamplified library was 9.9 X 10' plaque forming units (pfus)/ml with 18% nonrecombinants. The library was amplified by plating 50,000 pfus onto 20 150-mm plates, resulting in a final titer of 5.95 X lo6 pfu/ml, with 21% nonrecombinants.
Library screening was performed using standard methods (27). Specifically, the library was plated onto 24, 150-mm plates at 50,000 pfu/plate, and screened with a 32P-labeled cDNA clone obtained by PCR (see above). The probe was prepared using the method of Feinberg and Vogelstein (28), and the 32P-labeled DNA was purified using Elutip-D" columns according to the manufacturer's protocol. Plaque-lifts were performed using 15-cm nitrocellulose filters. Following denaturation and neutralization (27), DNA was fixed onto the filters by baking at 80 "C for 2 h. Hybridization was conducted at 42 "C overnight in a solution containing 50% formamide, 5 X SSC (1 X ssc is 0.15 M NaCl, 0.15 M sodium citrate, pH 7), 25 mM sodium phosphate, pH 7.0, 2 X Denhardt's solution, 10% dextran sulfate, 90 pg/ml yeast tRNA, and approximately lo6 cpm/ml 32P-labeled probe (5 X 10" cpm/pg). The filters were washed three times in 2 X SSC, 0.1% SDS at room temperature for 15 min/wash, followed by a single 20-min wash in 0.1 X SSC, 1% SDS at 45 "C. The filters were exposed to x-ray film at -70 "C for several days.
Plaques that hybridized with the labeled probe were purified by at least three rounds of replating and rescreening. Insert cDNAs were subcloned into the pBluescript SK(-) subcloning vector by the in vivo excision method (29) according to the manufacturer's protocol (Stratagene). Southern blots were performed as described (27) to verify that the rescued cDNA hybridized to the PCR probe. Two unique cDNAs were isolated, designated cGB-8 and cGB-10.
DNA Sequencing-Putative cGB-PDE cDNAs, subcloned into the pBluescript SK(-) vector, were sequenced (30) using Sequenase@ Version 2.0 or TaqTrack' kits. Universal T3/T7 primers or specific oligonucleotides designed to prime the cDNA insert were used as primers in the sequencing reaction. Template DNA was prepared by alkaline denaturation (31) of plasmid DNA, or by generation of singlestranded DNA (ssDNA) using T7 gene 6 exonuclease (32). For the latter procedure, 30 pg of plasmid DNA containing the cDNA insert were digested with either ApaI or SmaI. The digested DNA was purified by phenol-chloroform extraction and ethanol precipitation, followed by digestion with T7 gene 6 exonuclease (75 units/lO pg of DNA) in a solution containing 50 mM Tris-HC1, pH 8, 20 mM KC1, 5 mM MgCl,, and 10 mM @-mercaptoethanol in a final volume of 20 pl, The resulting ssDNA was purified by phenol-chloroform extraction and ethanol precipitation, and quantitated by measuring AZW, resulting in a typical yield of 10-13 pg of ssDNA. Sequencing of the entire cGB-8 cDNA clone was completed for both strands.
Northern Analysis-Polyadenylated RNA was purified from total RNA preparations (11, 27) using the Poly(A) Quick' mRNA purification kit according to the manufacturer's protocol. RNA samples (5 pg) were loaded onto a 1.2% agarose, 6.7% formaldehyde gel. Electrophoresis and RNA transfer were performed as previously described (11). Prehybridization of the RNA blot was conducted for 4 h at 45 "C in a solution containing 50% formamide, 5 X SSC, 25 mM sodium phosphate, pH 7, 2 X Denhardt's solution, 10% dextran sulfate, and 0.1 mg/ml yeast tRNA. A random hexanucleotide-primer-labeled probe (5 X 10' cpmlpg) was prepared as described above (28) using the 4.5-kilobase pair (kb) cGB-8 cDNA clone excised from pBluescript by digestion with AccI and SacII. The probe was heat denatured and injected into the blotting bag (6 X lo5 cpm/ml) following prehybridization. The Northern blot was hybridized overnight at 45 "C, followed by one 15-min wash with 2 X SSC, 0.1% SDS at room temperature, and three 20-min washes with 0.1 X SSC, 0.1% SDS at 45 "C. The blot was exposed to x-ray film for 24 h at -70 "C. The size of the RNA that hybridized with the cGB-PDE probe was estimated using a 0.24-9.5-kb RNA ladder that was stained with ethidium bromide and visualized with UV light.
Transient Transfection and Expression in COS-7 Cell.-A portion of the cGB-8 cDNA was cut out of the pBluescript SK(-) subcloning vector using XbaI. XbaI cut at a position in the pBluescript polylinker sequence located 30 bp upstream of the EcoRI cloning site, and at position 3359 within the cGB-8 insert. The resulting 3404-bp fragment, which contained the entire coding region of cGB-8, was then ligated into the XbaI cloning site of the expression vector pCDM8 (33). Competent E. coli MC1061/P3 cells were transformed with the resulting ligation products, and transformation-positive colonies were screened for the presence of the pCDM8 plasmid containing the cGB-8 insert in the proper orientaton using PCR and restriction enzyme analysis of miniprepped (27) plasmid DNA. The resulting expression plasmid construct is referred to as pCDM8-cGB-PDE. The DNA was purified from large scale plasmid preparations using Qiagen pack-500 columns according to the manufacturer's protocol. COS-7 cells were cultured in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 50 pg/ml penicillin, and 50 pg/ml streptomycin at 37 "C in a humidified 5% COz atmosphere. Approximately 24 h prior to transfection, confluent 100-mm dishes of cells were replated at one-fourth or one-fifth the original density. In a typical transfection experiment, cells were washed with buffer containing 137 mM NaCl, 2.7 mM KCl, 1.1 mM potassium phosphate, and 8.1 mM NaH2POI, pH 7.2 (PBS), followed by the addition of 4-5 ml of Dulbecco's modified Eagle's medium containing 10% NuSerum to each dish. Transfection with 10 pg of DNA mixed with 400 pg of DEAE-dextran in 60 pl of TBS (Tris-buffered saline: 25 mM Tris-HC1, pH 7.4, 137 mM NaCl, 5 mM KCl, 0.6 mM Na2HP0,, 0.7 mM CaC12, and 0.5 mM MgCl,), was performed by dropwise addition of the mixture to each plate (27). The cells were incubated at 37 "C, 5% CO, for 4 h, and then treated with 10% dimethyl sulfoxide in PBS. After 2 min, the dimethyl sulfoxide was removed, the cells were washed with PBS, and incubated in complete medium. After 48 h, cells were suspended in 0.5-1 ml of ice-cold homogenization buffer (40 mM Tris-HC1 (pH 7.51, 15 mM benzamidine, 15 mM ,5"mercaptoethanol, 0.7 pg/ml pepstatin A, 0.5 pg/ml leupeptin, and 5 mM EDTA) per dish of cells, and disrupted using a Dounce homogenizer. The resulting whole cell extracts were assayed for phosphodiesterase activity, cGMP binding activity, and total protein concentration. The number of repetitions of each assay is given in the legend to Fig. 3.
Phosphodiesterase, cGMP-binding, and Protein Assays-Phosphodiesterase activity was measured using a modification of the assay procedure described for the cGS-PDE (6). Incubation mixtures contained 40 mM MOPS, pH 7, 0.8 mM EGTA, 15 mM magnesium acetate, 2 mg/ml bovine serum albumin, 20 p~ cGMP or CAMP, [3H] cGMP or [3H]cAMP (100,000-200,000 cpm/assay), and 10-40 p1 of COS-7 cell extract, in a total volume of 250 pl. For the PDE inhibition studies, 25 pl of the indicated PDE inhibitor, diluted to the appropriate concentration, was added to the reaction mixture. For each PDE inhibitor tested, two types of control experiments were performed 1) an inhibition profile was obtained using purified bovine lung cGB-PDE diluted into homogenization buffer. This control was used to ensure that the phosphodiesterase assay conditions and the preparations of inhibitors used in these studies produced similar patterns of inhibition as had been previously reported for purified cGB-PDE (10); 2) a PDE inhibition profile was obtained using purified bovine lung cGB-PDE diluted into 10-40 pl of extract from mock-transfected COS-cells. This control was used to determine if any factors present in COS cell extracts affected the rate of cGMP hydrolysis or efficacy of the PDE inhibitors. Reaction mixtures were incubated for 10 min at 30 'C, and stopped by boiling. Next, 10 pl of 10 mg/ml Crotalus atror venom was added followed by a 10-min incubation at 30 "C. Nucleoside products were separated from unreacted nucleotides as described (6). In all studies, less than 15% of the total [3H]cyclic nucleotide was hydrolyzed during the reaction.
The cGMP-binding assay, modified from the previously described assay (lo), was conducted in a total volume of 80 pl; 60 pl of cell extract (see above) was combined with 20 pl of a binding mixture such that the final concentrations of the components of the mixture were 5 p M CAMP, 10 p~ 8-bromo-cGMP, and 1 or 2.5 p~ [3H]cGMP. The CAMP and 8-bromo-cGMP were added to block any [3H]cGMP binding attributable to cAK and cGK, respectively. Binding was measured in the absence and presence of 0.2 mM IBMX, as indicated. Excess unlabeled cGMP or CAMP, diluted in water to the appropriate concentration, was added when indicated. The reaction was initiated by the addition of the cell extract, followed by incubation for 60 min at 0 "C. Reaction mixtures were filtered as described (10). Assay blanks were determined by parallel incubations with homogenization buffer replacing cell extracts, or with a 100-fold excess of unlabeled cGMP. Similar results were obtained with both methods. Total protein concentration of the cell extracts was determined by the method of Bradford (34) using bovine serum albumin as the standard.
the DNA INSPECTOR program from Textco. SwissProt and Sequence Analyses-DNA sequence analysis was performed using GenEmbl databases (February, 1992) were searched using the FASTA (35) program supplied with the Genetics Computer Group (GCG) Software packages, and regions of homology between two sequences were located with the COMPARE (36) program. The DOT PLOT dot matrix program was used to plot the results of the COMPARE program. The DOT PLOT parameters used in these studies were window = 30 and stringency = 15. Painvise sequence alignments were conducted using the ALIGN (distributed by the National Biomedical Research Foundation (37) and BESTFIT (38) (included in the GCG package) programs. The ALIGN program calculates a maximum alignment score for a pair of real sequences and the distribution of maximum scores for random permutations of the two sequences. The alignment score is the number of standard deviations by which the maximum score for the real sequences exceeds the average maximum score for the random permutations. The probability of a score 2 5 standard deviations occurring by chance is 2.87 X lo-?. ALIGN program default parameters were used in these studies (mutation data matrix = +6, break penalty = 10, random runs = 100). Multiple sequence comparisons were performed using the Progressive Alignment Algorithm (39) implemented in the PILEUP program (GCG software).
The following strategy was utilized in specifically defining the boundaries of the putative catalytic and cGMP-binding regions of cGB-PDE. First, dot matrix analysis was used to compare the sequence of cGB-PDE to each of the other individual PDE sequences shown in Fig. 5. The results of this analysis provided a rough estimate of the location of the catalytic region, conserved among all PDEs, and the noncatalytic cGMP binding region, conserved only in cGB-PDE, cGS-PDE, and the photoreceptor-PDEs. Next, a multiple sequence alignment including the entire sequence of all nine PDEs described in Fig. 5 was performed (not shown) and used as a guideline in rationally assigning boundary residues. The amino-and carboxylterminal boundaries of the catalytic region were arbitrarily assigned to the first and last residues, respectively, which were identical in at least six out of nine PDEs. Similarly, the amino-terminal and carboxyl-terminal boundaries of the cGMP-binding region were assigned to the first and last residues, respectively, which were conserved in at least three of the five cGMP-binding PDEs shown in Fig. 6. The amino-terminal boundary of the cGMP-binding region (Leu"' in cGB-PDE) was preceded by a 14-residue gap introduced by the alignment program. Similarly, the carboxyl-terminal boundary (G1dZ6 in cGB-PDE) was followed by a 20-residue gap introduced by the alignment program.
Dot matrix analysis of the deduced cGB-PDE sequence was used to approximate the location of the internally homologous repeats within the cGMP-binding region. Next, painvise alignment of the regions spanning residues 160-350 and 351-526 of cGB-PDE was performed. The amino-terminal boundaries of the repeats were arbi-trarily assigned to residues GluZZ8 (repeat a ) and Glu4I0 (repeat b ) because these residues were preceded by a 4-residue gap introduced by the alignment program. Similarly, the carboxyl-terminal boundary residues, Glu311 (repeat a), and Glu5" (repeat b) were followed by a 2residue gap introduced by the alignment program, and a region of only 11% sequence identity.

RESULTS AND DISCUSSION
cGB-PDE cDNA Isolation and Characterization-Degenerate oligonucleotides were designed based on the sequences NYMYAQYV and FDNDEG(E/D)Q2 obtained from peptides purified from proteolytic digests of bovine lung cGB-PDE. The resulting oligonucleotides were then used as primers in a PCR reaction with bovine lung cDNA as the template, resulting in the production of an 824-bp DNA fragment (referred to as the PCR clone). DNA sequencing of the PCR clone revealed a single large open reading frame (ORF) encoding a polypeptide which contained the cGB-PDE peptide sequences and possessed a high degree of homology to the deduced AA sequences of cGS-PDE (ll), and the photoreceptor-PDEs (12)(13)(14). In order to obtain a cDNA encoding a full-length cGB-PDE, a bovine lung cDNA library in X-ZAP I1 was screened using the 32P-labeled PCR clone as a probe. Two distinct cDNA clones, designated cGB-8 and cGB-10, were isolated. cGB-10, which was 1019 bp in length, contained residues 264-1283 of cGB-8. The sequence of cGB-10 was identical to the corresponding sequence of cGB-8, except for a single base difference at position 123 in the cGB-10 sequence. This single base discrepancy could have been due to an error of cDNA synthesis. The PCR clone contained residues 489-1312 of cGB-8. During the cloning process, a third putative cGB-PDE cDNA clone was identified and plaque purified. The size and nucleotide sequence of this third clone cGMP-binding, cGMP-specific Phosphodiesterase 22867 appeared to be identical to cGB-8 by restriction map analysis (data not shown). The cGB-8 cDNA clone was 4474 bp in length and contained a large ORF of 2625 bp (Fig. 1). A translation initiation site (ATG) was identified at position 99-101 in the nucleotide sequence. It was preceded by an in-frame stop codon, and the surrounding bases were compatible with the Kozak consensus initiation site for eukaryotic mRNAs (40). The stop codon, TAG, was located at position 2724-2726, and was followed by 1748 bp of 3"untranslated sequence. The sequence of cGB-8 did not contain a transcription termination consensus sequence (41) or a poly(A) tail; therefore, the clone did not represent the entire 3' sequence of the corresponding mRNA. This was consistent with the fact that in Northern analysis of bovine lung polyadenylated RNA, a 32P-labeled cGB-PDE cDNA hybridized to a single 6.9-kb species (Fig. 2).
The ORF of the cGB-8 cDNA encoded an 875-AA polypeptide with a calculated molecular mass of 99,525 daltons. This was in good agreement with the molecular mass of a cGB-PDE monomeric subunit, 93 kDa, estimated by SDS-PAGE analysis of purified protein (10). Also, the AA sequences of six peptides obtained from proteolytic digests of purified bovine lung cGB-PDE (boned in Fig. 1) corresponded exactly to the AA sequence encoded by cGB-8, thus providing strong evidence that cGB-8 encodes cGB-PDE.
Expression of cGB-PDE in COS-7 Cells-The ability of cGB-PDE cDNA to encode a functionally intact phosphodiesterase was demonstrated by transfecting COS-7 cells (42) with the cGB-8 cDNA clone. Transfection of COS-7 cells with the pCDM8-cGB-PDE construct (see "Experimental Procedures" for explanation) resulted in the expression of approximately 15-fold higher levels of cGMP phosphodiesterase activity than in mock-transfected cells, or in cells transfected with pCDM8 vector alone (Fig. 3A). No increase in cAMP phosphodiesterase activity over mock or vector-only transfected cells was detected in extracts from cells transfected with pCDM8-cGB-PDE. (Fig. 3A). Thus, under the assay conditions described in this report (20 p~ substrate) the expressed cGB-PDE cDNA protein product hydrolyzed cGMP, whereas no hydrolysis of cAMP was detected. This degree of substrate specificity was also observed with the nonrecombinant cGB-PDE purified from bovine lung, which exhibited no detectable cAMP hydrolysis at concentrations Up to 50 p M (10).
The relative potencies of PDE inhibitors for inhibition of cGMP hydrolysis by the expressed cGB-PDE cDNA protein product are shown in Fig. 3B. Zaprinast, a relatively specific inhibitor of cGMP-specific phosphodiesterases (10, 43, 44) had an IC50 value of 2.5 PM. As expected for cGB-PDE (lo), dipyridamole, an effective inhibitor of selected CAMP-and cGMP-specific phosphodiesterases (43), was also a potent inhibitor of the expressed cGB-PDE (IC50 = 3.5 p~) .
The relative potencies of this battery of PDE inhibitors for inhibition of cGMP hydrolysis by the expressed cGB-PDE cDNA protein product (Fig. 3B) were identical to those relative potencies reported for cGB-PDE purified from bovine lung (10). These findings provided strong evidence that the cGB-8 cDNA encodes cGB-PDE.
Interestingly, the absolute IC50 values for all inhibitors tested were 2-7-fold higher for the recombinant cGB-PDE expressed in COS-7 cells as compared to cGB-PDE purified from bovine lung. This difference could not be attributed to the effects of any factors present in COS-7 cell extracts on cGMP hydrolytic activity, since purified bovine lung cGB-PDE exhibited kinetics of inhibition identical to those previously reported, when assayed as a pure enzyme, or when assayed in the presence of extract from mock-transfected COS-7 cells (described under "Experimental Procedures") (data not shown). This apparent difference in pharmacological sensitivity could have been due to a subtle difference in the structure of the expressed cGB-PDE cDNA protein product and the cGB-PDE purified from bovine lung, such as a difference in post-translational modXication at or near the catalytic site. Alternatively, this difference could have been due to an alteration of the catalytic activity of bovine lung cGB-PDE during the purification.
The cGMP binding activity of the expressed cGB-PDE cDNA protein product was characteristic of the cGB-PDE purified from bovine lung in that it was stimulated by IBMX.
In the presence of 0.2 mM IBMX and 1 p~ [3H]cGMP, extracts from COS-7 cells transfected with pCDM8-cGB-PDE exhibited &fold higher cGMP binding activity than did extracts from mock or vector-only transfected cells (Fig. 3C). This cGMP binding activity was stimulated approximately 1.8-fold by the addition of the IBMX (Fig. 3C). This effect of IBMX, a competitive inhibitor of cGMP hydrolysis, in stimulating cGMP binding 2-5-fold is a distinctive property of the cGMP-binding phosphodiesterases, having been observed for the cGB-PDE from rat (21), and bovine lung (lo), as well as for the photoreceptor-PDEs from Rana catesbiana rod (8) and bovine cone (9), and for the bovine heart cGS-PDE (15). IBMX-stimulated binding occurs under conditions in which no cGMP hydrolysis is detected, thereby eliminating the possibility of substrate protection as the mechanism for increased binding (10). Rather, it is thought that IBMX elicits this effect on cGMP binding by generating a conformational change in the PDE upon interacting with the catalytic site (49). The finding that recombinant cGB-PDE, as expressed in COS-7 cells, exhibited IBMX-stimulated binding suggested that its conformation is similar to that of the nonrecombinant enzyme. No IBMX stimulation of the low level of background cGMP binding was observed.
The binding activity of the protein product encoded by the cGB-PDE cDNA was highly specific for cGMP as compared to CAMP. Less than 10-fold higher concentrations of unlabeled cGMP were required to lower [3H]cGMP binding by 50%, whereas approximately 100-fold higher concentrations of cAMP were required for the same effect (Fig. 30). 100-fold excess concentrations of unlabeled cAMP were also required to produce 50% inhibition of the cGMP binding activity of the nonrecombinant cGB-PDE purified from bovine lung (10).

AA Sequence Analysis and Domain Structure Assignment-
A search of the SWISS-PROT and GenEmbl data banks revealed that no protein or nucleic acid sequences other than those of PDEs displayed significant homology to the deduced cGB-PDE sequence. Like all mammalian phosphodiesterases sequenced to date, cGB-PDE contained a conserved segment of sequence, approximately 250 residues in length, located in the carboxyl-terminal half of the protein. Several pieces of evidence supported the proposal that this region of conserved sequence comprises the catalytic site of PDEs (for review, see Ref. 1). This segment included residues 578-812 (Fig. 4)   The catalytic region of cGB-PDE more closely resembled the catalytic region of the photoreceptor-PDEs than it did the corresponding region of other PDEs (45-48% sequence identity). This finding was not surprising since previous studies demonstrated that the hydrolytic activities of cGB-PDE and the photoreceptor-PDEs share several characteristics, including a strong specificity for cGMP relative to CAMP, and sensitivity to the PDE inhibitors zaprinast and dipyridamole (10,50). Several segments of sequence contained residues that were invariant among the photoreceptor-PDEs and cCB-PDE, but that were not present in other PDEs (Fig.  5). These regions of high homology among cGB-PDE and the photoreceptor-PDEs may serve important roles in conferring the specificity for cGMP hydrolysis as compared to CAMP hydrolysis, or in conferring sensitivity to specific pharmacological agents.

A r q L y l Gly Thr A r q Glu net Val Aan Ala Trp Phe A l a Glu A r q Val His Thr Ile Pro Val Cys L y s Glu Gly I1e Lys Gly His The Glu
The structural similarity between cGB-PDE and the other cGMP-binding PDEs (cGS-PDE and the photoreceptor-PDEs) was not limited to the conserved catalytic region, but also included an additional region of sequence homology located in the amino-terminal half of the protein. Optimization of the alignment (see "Experimental Procedures") between cGB-PDE, cGS-PDE, and the photoreceptor-PDEs indicated that this conserved amino-terminal segment included residues 142-526 of cGB-PDE (Fig. 4). Earlier comparison of the deduced AA sequences of the cGS-PDE and the photoreceptor-PDEs led to the proposal that this segment of conserved sequence comprised a noncatalytic cGMP binding region (5). Alignment of the proposed cCMP binding region of cGB-PDE with the corresponding region of the photoreceptor-PDEs and cGS-PDE produced Dayhoff alignment scores of 14-18 standard deviations (see "Experimental Procedures" for explanation of alignment scores) with 26-28% sequence identity. A multiple sequence alignment of the proposed cGMP binding regions of several PDEs is shown in Fig. 6. Within this region containing approximately 400 residues, 38 positions were invariant among all cGMP-binding PDEs listed. Neither the catalytic nor the noncatalytic cGMP binding regions of cGB-PDE showed statistically significant homology to the cyclic nucleotide-binding domains of the E . coli catabolite gene activator protein (CAP) family of proteins (which includes the CAP protein, cAK, cGK, and the cyclic nucleotide-gated channels) (for review, see Ref. Sl). Additionally, neither region contained the guanine nucleotidebinding elements of G-proteins (14).
The proposal that the cGMP binding region is located within the conserved segment of cGB-PDE sequence near the amino terminus is supported by previous biochemical studies of the purified lung cGB-PDE. Partial proteolysis of purified cGB-PDE with chymotrypsin produced two fragments (35 and 45 kDa) which exhibited ['HIcGMP binding activity and specific photoaffinity labeling with ["'PIcGMP (10). The fragments were identical in their amino-terminal sequences, TSPRFDNDEG, suggesting that the 35-kDa fragment was derived from a second chymotryptic cleavage of the 4.5-kDa fragment. If the molecular weights determined by SDS-PAGE were accurate, both fragments would span most of the more amino-terminal conserved segment. These data provided direct evidence that this segment of conserved sequence serves as a cGMP binding region of cGB-PDE. It is of interest to note that earlier studies of purified cGR-PDE suggested that the dimerization domain is located within or adjacent to this cCMP binding region (10).
Like the photoreceptor-PDEs and cCS-PDE, the proposed cGMP-binding domain of cCB-PDE contained two internally homologous repeats. Optimization of the alignment of these two segments of cGB-PDE sequence (see "Experimental Procedures") suggested that the repeats span at least residues 228-311 (labeled repeat a in Fig. 4) and 410-500 (repeat b in Fig. 4). Alignment of cGB-PDE repeat a with cGR-PDE repeat b gave a Dayhoff alignment score of 16 with 43% sequence identity (Table I). Since boundary assignments were not rigorously defined (see "Experimental Procedures"), some degree of internal homology may extend beyond the core of conserved sequence defined by regions a and b. Alignment of the repeated segments of cCB-PDE with the corresponding regions of cGS-PDE and the photoreceptor-PDEs showed that both segments a and 6 of cGB-PDE were homologous to each segment a and b from all the cCMP-binding PDEs (Table  I). Fig. 7 shows a multiple sequence alignment of the regions corresponding to repeats a and b from several cGMP-binding PDEs.
The internally homologous repeats present in the cGMP binding region of cGB-PDE, as well as of the other cGMPbinding PDEs, may represent two similar, but distinct, cCMP-binding sites. A tandem repeat pattern is also present in the sequences of cAK and cGK and, in these instances, represent two distinct cyclic nucleotide-binding sites on both kinases (51). Studies of the association and dissociation patterns of cGMP binding to cGB-PDE are suggestive of the existence of more than one type of cCMP-binding site (10, 49). Additionally, Scatchard analysis of cGMP binding indicated the existence of two classes of non-catalytic sites for cGMP binding to cGS-PDE (IS), and ROS-PDE (8, 52). If the tandem repeats of sequence in the cGMP-binding PDEs represent two distinct cGMP-binding sites, amino acids essential for cGMP binding should he conserved in both regions  a and 6 of each enzyme. Seven invariant residues exist in all a and 6 regions from all cGMP-binding PDEs listed in Fig. 7. These residues are also present in the corresponding regions of the a and j3 subunits of the human (53,54) and mouse ROS-PDEs (55) (not shown in Fig. 7). Three of these 7 residues possess charged side groups which may be important for interacting with cGMP.
Three regions of cGB-PDE, beyond the boundaries of the proposed cGMP binding region and proposed catalytic region, showed no significant sequence similarity to other PDEs (Fig.  4). The site (Ser'') of phosphorylation of cGB-PDE by cGK (22) was located within the unique amino-terminal region (Fig. 4). The location of the phosphorylated serine near the amino terminus of the deduced AA sequence of cGB-PDE is in good agreement with previous studies of purified cGB-PDE. The peptide sequence of the site of phosphorylation of cGB-PDE, KISASEFDRPLR, (the phosphorylated serine is underlined) was obtained from a tryptic [32P]phosphopeptide derived from purified cGB-PDE, which had been phosphorylated by cGK and [-p3*P]ATP (22). Limited proteolysis of cGB-PDE, phosphorylated by either cGK or catalytic subunit of cAK and [y3'P]ATP, with chymotrypsin or S. Aureus protease produced 80-85-kDa cGB-PDE fragments which lacked radiolabel (56). The radioactivity remained associated with small fragments of cGB-PDE (10-15 kDa), providing biochemical evidence that the phosphorylated serine was located within 10-15 kDa of one extremity of the cGB-PDE monomer.
Since the amino-terminal segment containing the phosphorylation site of cGB-PDE showed no homology to other phosphodiesterase sequences, and the effect of phosphoryla-   cGS , cGB-PDE . Numbers to the right indicate the position of the last residue in that line. Alignment was performed using the PILEUP program. -8 (59-462), CONE-a' (60-461), tion by cGK on cGB-PDE function is unknown, the fact that several phosphodiesterases have been reported to be regulated by phosphorylation (57,58) could be coincidental. A recent report showed that incubation of partially purified cGMPbinding, cGMP-specific phosphodiesterase from guinea pig lung with the catalytic subunit of cAK resulted in a 10-fold increase in the V,,, for cGMP hydrolysis (59). It is of interest to note that binding of cGMP to the allosteric site on cGB-PDE from bovine lung is required for its phosphorylation (22). Perhaps regulation of cGB-PDE activity involves a complex interplay between the phosphorylation site and the cGMP binding region.

ROS
cDNA cloning and analysis of the deduced AA sequence of cGB-PDE has provided new insight into the structural relationship of cGB-PDE to other PDEs and into the location of functional domains within the cGB-PDE molecule. The availability of the AA sequence of cGB-PDE and the cDNA encoding cGB-PDE provides the basis for future studies aimed at more precisely characterizing the structure and function of the individual domains of cGB-PDE. These studies could lead