Structure of two rat genes coding for closely related rolipram-sensitive cAMP phosphodiesterases. Multiple mRNA variants originate from alternative splicing and multiple start sites.

The products of two phosphodiesterase (PDE) genes (ratPDE3/IVd and ratPDE4/IVb) are present in the rat Sertoli cell in culture, and their expression is under the control of the gonadotropin follicle-stimulating hormone (Swinnen, J.V., Tsikalas, K.E., and Conti, M. (1991) J. Biol. Chem. 266, 18370-18377). To understand the basis of the sequence heterogeneity found in the 5'-region of the different cDNAs thus far characterized, the structure of the coding region of these two cAMP PDE genes was investigated. Analysis of five ratPDE3/IVd and ratPDE4/IVb genomic clones showed that the coding region of these genes expressed in the Sertoli cell is divided into 11 exons distributed over 35-45 kilobases of genomic DNA. The intron/exon boundaries agreed, with some exceptions, with the established consensus sequences and were located in the same position in the coding region of the two genes. Also present were similarities to the exon composition of the Drosophila melanogaster "dunce" gene, the ancestor of these mammalian cAMP PDEs. Multiple AUG codons and short open reading frames were present at the 5'-untranslated end of the ratPDE4/IVb mRNA, but not in the ratPDE3 mRNA. By using polymerase chain reaction amplification or Northern analysis, it was determined that at least two forms of ratPDE3/IVd mRNA are present in rat Sertoli and FRTL-5 thyroid cells, but not in the brain. These mRNA variants are generated by inclusion or removal of an intron sequence that produces a frameshift affecting the position of the initiation AUG codon. Both mRNA species were efficiently translated into cAMP PDE proteins with different molecular masses in a transient transfection assay in COS cells. Polymerase chain reaction amplification demonstrated that heterogeneity of ratPDE4/IVb mRNAs was present in the same location as in the ratPDE3/IVd mRNA. Two ratPDE4/IVb mRNAs with different 5'-ends were expressed in Sertoli and FRTL-5 cells and in the brain. This heterogeneity is caused by the presence of an intron promoter that controls the transcription of this mRNA in Sertoli and FRTL-5 cells, but not in the brain. Upstream exons and additional promoters are probably present and necessary to generate the brain-specific mRNAs. These findings demonstrate that the cAMP-specific PDE genes have complex structure and that cAMP PDE proteins with different amino termini are derived from these genes.

Structure of Two Rat Genes Coding for Closely Related Rolipram-sensitive cAMP Phosphodiesterases MULTIPLE mRNA VARIANTS ORIGINATE FROM ALTERNATIVE SPLICING AND MULTIPLE START SITES* (Received for publication, June 3, 1993, and in revised form, August 17, 1993) Lucia Monaco, Elena Vicini, and Marco ContiS From the Institute of Histology and General Embryology, Uniuersity of Rome, 00161 Rome, Italy The products of two phosphodiesterase (PDE) genes (ratPDE3lIVd and ratPDE4/IVb) are present in the rat Sertoli cell in culture, and their expression is under the control of the gonadotropin follicle-stimulating hormone (Swinnen, J. V., Tsikalas, K. E., and Conti, M. (1991) J. Biol. Chern. 266,18370-18377). To understand the basis of the sequence heterogeneity found in the 5'-region of the different cDNAs thus far characterized, the structure of the coding region of these two CAMP PDE genes was investigated. Analysis of five ratPDE3lIVd and ratPDE4/IVb genomic clones showed that the coding region of these genes expressed in the Sertoli cell is divided into ll exons distributed over 35-45 kilobases of genomic DNA. The introdexon boundaries agreed, with some exceptions, with the established consensus sequences and were located in the same position in the coding region of the two genes. Also present were similarities to the exon composition of the Drosophila melanogaster "dunce" gene, the ancestor of these mammalian CAMP PDEs. Multiple AUG codons and short open reading frames were present at the 5'-untranslated end of the ratPDE4/IVb mRNA, but not in the ratPDE3 mRNA. By using polymerase chain reaction amplification or Northern analysis, it was determined that at least two forms of ratPDE3lIVd mRNA are present in rat Sertoli and FRTL-5 thyroid cells, but not in the brain. These mFtNAvariants are generated by inclusion or removal of an intron sequence that produces a frameshift affecting the position of the initiation AUG codon. Both mRNA species were efficiently translated into CAMP PDE proteins with different molecular masses in a transient transfection assay in COS cells. Polymerase chain reaction amplification demonstrated that heterogeneity of r a t P D E m mRNAs was present in the same location as in the ratPDE3lIVd mRNA. Two ratPDE4/IVb mRNAs with different 5'-ends were expressed in Sertoli and FRTL-5 cells and in the brain. This heterogeneity is caused by the presence of an intron promoter that controls the transcription of this mRNA in Sertoli and FRTL-5 cells, but not in the brain. Upstream exons and additional promoters are probably present and necesdelle Ricerche and from the Faculty of Medicine, University of Rome, by * This work was supported by grants from the Consiglio Nazionale National Institutes of Health Grant HD20788, and by a grant from the "Istituto Pasteur, Fondazione Cenci Bolognetti." The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
The nucleotide seauencelsi reuorted in this uauer has been submitted to the GenBankTMiEMBL Data Bank with' accession number(s) B8 VOl278-VO1299.
Biology, Dept. of Gynecology and Obstetrics, Stanford Medical Center, $: To whom correspondence should be addressed: Div. of Reproductive 300 Pasteur Dr., Stanford, CA 94305. Tel.: 415-725-2452;Fax: 415-725-7102. sary to generate the brain-specific mRNAs. These findings demonstrate that the CAMP-specific PDE genes have complex structure and that CAMP PDE proteins with different amino termini are derived from these genes.
Recent studies on the molecular structure of cyclic-nucleotide phosphodiesterases (PDEs)' have uncovered an unsuspected complexity of the enzymes that degrade the second messengers CAMP and cGMP. At least 25 different PDE forms have been characterized and have been classified in five distinct families on the basis of their substrate specificity and regulation (1,2). 2 The family of PDEs that hydrolyze CAMP with high affinity (CAMP PDEs), type IV according to a nomenclature recently proposed (21, was originally identified on the basis of the affinity and selectivity for CAMP and specific inhibition by antidepressants like Rolipram and RO 20-1724 (3). Data from our and other laboratories have shown that the expression of these forms is regulated by hormones that act through the CAMP-dependent pathway (reviewed in Ref. 4). In spite of the considerable amount of data available, the definition of the exact biochemical properties of these CAMP PDEs has been elusive. A survey of the literature of the past 10 years has shown that molecular masses ranging from 29 to 80 kDa have been attributed to these forms (3,(5)(6)(7)(8). In addition, linear and nonlinear kinetics for apparently homogeneous PDE preparations have been observed. An explanation for these conflicting results has come with the cloning of Drosophila melanogaster "dunce" CAMP PDE (9) and, subsequently, of its mammalian homologues (10)(11)(12)(13). Analysis of cDNA clones derived from rat testis and brain has indicated that at least four groups of mRNAs encoding closely related proteins are present (10)(11)(12)(13). Thus, the presence of more than one kinetically related CAMP PDE form in a given tissue is a likely cause of contrasting physicochemical properties attributed to the cAMP PDEs.
The molecular cloning of the CAMP PDEs has uncovered, however, an additional level of complexity. Northern blot analysis of mRNA retrieved from different tissues demonstrates the presence of transcripts of different sizes for each of the four forms (10)(11)(12)(13)(14), with a maximum of five different transcripts for ratPDEl/IVc and five for ratPDE2DVa found in germ cells (15).
Furthermore, three classes of ratPDE2iIVa cDNAs with differ-' The abbreviations used are: PDEs, cyclic-nucleotide phosphodiesterases; kb, kilobase(s1; bp, base pair(s); PCR, polymerase chain reaction; Pipes, 1,4-piperazinediethanesulfonic acid. ent 5'-ends (RD1, RD2, and RD3) have been retrieved from brain libraries (lo), and two variant cDNAs for ratPDE3LVd with different 5'-ends have been detected in the testis (13). Comparison of the ratPDE4/TVb sequences retrieved from testis and brain libraries again demonstrates different nucleotide sequences in the 5"region (12,14). The presence of multiple mRNA species is compatible with two hypotheses. It is possible that different mRNAs produced by alternative splicing of exons in mRNA serve to generate CAMP PDE proteins with different amino or carboxyl termini. Regulatory domains affecting catalysis, like the domains binding calmodulin and cGMP, are present at the amino terminus of the calmodulin-regulated PDEs and the cGMP-stimulated PDEs, respectively (1G18). It is then conceivable that the generation of CAMP PDEs with different amino termini is a means to produce enzymes targeted for different regulatory signals. The alternative possibility is that differences at the 5'-end of the mRNAs are the result of the presence of different promoters and different transcription start sites. If this latter hypothesis were correct, different regulatory mechanisms would impinge on the gene and its expression and not on the protein product. That the genes encoding CAMP PDE have a complex organization and complex transcription regulation is documented by the characterization of the D. melunoguster gene (191, the probable ancestor of the mammalian CAMP PDEs (9,191. The dunce gene spans 148 kb of genomic DNA. Three different transcription start sites and alternative splicing generate six different mRNA species (19). In addition, two genes have been mapped within the first intron of this dunce gene (20). Similarly, a Dictyostelium discoideum PDE gene has three different start sites that are differentially utilized during development and differentiation of the slime mold (21).
To understand the functional significance of the cAMP PDE mRNA heterogeneity, we have characterized the structure of the two mammalian ratPDE3KVd and ratPDE4/IVb CAMP PDE genes expressed in the Sertoli cell and determined the origin of mRNA variants. Evidence consistent with the presence of multiple promoters and alternative splicing indicates a high degree of complexity of these genes in mammals and suggests that different promoters are employed in endocrine cells and in the brain to regulate CAMP PDE expression.
EXPERIMENTAL PROCEDURES Materials-The Charon 4A rat genomic library was purchased from Clontech; the pWE15 cosmid and A Dash I1 rat genomic library were from Stratagene. Nylon colony plaque hybridization filters, [y32PlATP, [CI-~~PI~CTP, and CY-~~S-~ATP were from DuPont NEN. Nylon membranes (Zeta-Probe) were from Bio-Rad. All restriction enzymes were purchased from Boehringer Mannheim, Life Technologies, Inc., or Promega Biotec. Plasmid and A preparation kits were from Qiagen. The first-strand cDNA synthesis kit and dextran sulfate were from Pharmacia LKB Biotechnology Inc. Oligo(dT)-cellulose T-3 was from Collaborative Research. Sequenase (version 2) was from United States Biochemical Corp. The GeneAmp PCR reagent kit and Taq DNA polymerase were from Perkin-Elmer Cetus Instruments.
All other chemicals used were analytical grade and purchased from Sigma or Bio-Rad. Oligonucleotides were synthesized using a Pharmacia Gene Assembler.
Isolation and Characterization of Genomic Clones-DNA isolations were carried out using standard methods (22). Three rat genomic libraries (cosmid pW15, A Dash 11, and A Charon 4A) were screened by plaque hybridization essentially as previously reported for the cDNA library screening (11). Briefly, Escherichia coli cell line LE392 was used as the host for plaque lifting. Replicate nylon colony plaque hybridation filters were prehybridized for 16 h a t 42 "C in 5 x SSC (1 x SSC = 0.67 M NaCl, 0.67 M sodium citrate), 10 x Denhardt's solution (1 x Denhardt's solution = 0.02% Ficoll, 0.02% polyvinylpyrrolidone, 0.02% bovine serum albumin), 50% formamide, 0.5% SDS, and 0.1 mg/ml salmon sperm DNA and then hybridized a t 42 "C for 16 h in the presence of 10% dextran sulfate with lo6 cpndml 32P-labeled probe (cDNA random primer-labeled or labeled PCR fragments). Filters were washed at room temperature in 2 x SSC and 0.5% SDS and at 55 "C in 0.5 x SSC and 0.5% SDS and exposed to Kodak XAR film overnight a t -70 "C with intensifying screens. Approximately 500,000 plaque-forming units were screened. Positive clones were purified by repeated screening, and the DNAs from resulting clones were purified on ion-exchange columns (Qiagen) according to the manufacturer's protocol. Restriction fragments of these clones were subcloned into the pBluescript KS(+) plasmid vector (Stratagene) for further characterization. Sequencing reactions were performed on both strands with the dideoxy chain termination method of Sanger et al. (23) using Sequenase DNA polymerase according to the protocol supplied by the manufacturer. Alignment of the restriction fragments obtained from each clone was performed by partial digestion and Southern blotting following established procedures (22).
Southern Blot Analysis-Genomic DNA was isolated from Wistar rat liver using the proteinase K digestion method (22). Genomic DNA (30 pg/lane) was restriction-digested with EcoRI, and DNA fragments were fractionated by electrophoresis on 0.8% agarose gel. The DNA was transblotted to nylon membrane (24). Hybridization conditions were the same as those reported above for the library screening.
RNA Preparation-Total RNA from brain tissue or from Sertoli cell cultures was purified by the guanidine thiocyanate extraction method of Chirgwin et al. (25) as previously described (11). Cell extracts were layered on a cushion of 5.5 M CsCl, and RNA was pelleted by centrifugation for 18 h at 100,000 x g in a Beckman SW 40 rotor at room temperature. Poly(A)+ RNAwas purified by affinity chromatography on an oligo(dT)-cellulose column (26).
Polymerase Chain Reaction Amplifications-PCRs to generate the probes used in library screenings were performed in a volume of 50 pl containing 500 ng of cDNA, 50 n m KCl, 10 mM Tris-HC1, pH 8.3, 1.5 rn 2.5 units of Taq polymerase. The reactions were performed for 1 min at 94 "C, 1 min at 50 "C, and 2 min at 72 "C for 25 cycles. At the end of the reaction, products were purified on a Sephadex G-50M column. To retrieve CAMP PDE cDNAs containing additional 5'-sequence, Agtll testis and Sertoli cell libraries (11,12) were used as template for PCR amplification. Primers used were ratPDE4lNb-specific oligonucleotides (oligonucleotide A and N) and primers corresponding to the A sequence flanking the cloning site (forward and reverse primers; Promega Biotec). Amplified fragments were subcloned in the pBluescript KS(+) plasmid and sequenced in both directions.
Reverse Dunscriptuse PCR-First-strand cDNAwas generated using the first-strand cDNA synthesis kit according to the supplied protocol (27). Briefly, 0.2 pg of poly(A)+ RNA was reverse-transcribed with Moloney murine leukemia virus reverse transcriptase using random hexadeoxynucleotides as primer. The completed first-stand cDNA reaction product was directly amplified by PCR following the addition of the specific primere and !lbq DNA polymerase. PCR was performed a8 de- TAEILE I Intron exon boundaries in ratPDE4 mRNA. Location of the exons is based on the numbering of the sequence reported in Fig. 2, where the A residue of the first ATG codon is base 1. Exon sequences are indicated by upper-case letters, and intron sequences by lower-case letters. Exon la is the putative exon present in the brain  Sambrook at al. (22). The oligonucleotides used (see Fig. 9) were endlabeled with T4 polynucleotide kinase and [-p3"?P1ATP. The radioactive probe was hybridized to 20 pg of Sertoli cell RNA overnight at 30 "C in 30 pl of buffer containing 40 mM Pipes, pH 6.8,0.4 M NaC1,l mM EDTA, and 80% formamide. The hybridization mixture was then precipitated, and the pellet was resuspended in 50 p1 of reverse transcription buffer (50 mM Tris-HC1, pH 7.5,60 m~ KCI, 10 mM MgCl,, 1 mM each dNTP, 1 mM dithiothreitol, 1 pl of RNasin, 50 pg/ml actinomycin D); 50 units of avian myeloblastosis virus reverse transcriptase was added, and the incubation was performed for 90 min at 42 "C. At the end of this incubation, template RNA was removed by digestion with DNase-free RNase, and the reaction product was extracted with phenoVchloroform and then ethanol-precipitated. The resulting pellet was resuspended in formamide loading buffer (80% formamide, 10 mM EDTA, 1 mg/ml xylene cyanol, 1 mg/ml bromphenol blue) and analyzed on a 6% denaturing urea-polyacrylamide gel.
Pansfection a n d Protein Detection by Western Ana1ysis"Transient transfection of COS or MA-10 cells was performed using the CaPOJ DNA method (28) following a procedure previously reported (29). The pCMV-ratPDE3.1 construct has been previously described (29). The pCMV-ratPDE3.2 construct was obtained by subcloning the ratPDE3.2 cDNA(13) into the EcoRI restriction site of the pCMV polylinker. Plamsids were purified using Quiagen columns according to the manufacturer's directions. Western analysis of extract from the transfected cells was performed with K116 antibody as previously described (14).

RESULTS
Organization of ratPDE4 lIVb and ratPDE3 lIVd Genes-A cDNA containing the major open reading frame of ratPDE4/IVb (11) was used to screen a genomic library prepared in cosmid pWE15. One positive clone (pWE15-ratPDE4.1), containing an insert of -25 kb, was identified and characterized by partial digestion and Southern blot analysis (Fig. 1). The fragments containing exon sequences were subcloned into the pBluescript KS(+) plasmid and sequenced using exon-specific primers. Comparison of the genomic fragment sequences with the corresponding cDNA sequence made it possible to map the positions of the exons in the gene (Fig. 1). To obtain the 3'-portion of this gene, a library constructed in A Dash I1 was screened using a DNA fragment prepared by amplifying a region of 155 bp at the 3'-end of cDNA (corresponding to bp + E 4 5 to +1699 of the ratPDE4lIVd cDNA oligonucleotides P and 0 in Fig. 2) as a probe. From this screening, four positive clones were retrieved, and one  was further characterized. EcoRI digestion of this clone revealed the presence of six genomic fragments, and each fragment was subcloned, sequenced, and mapped in the PDE4/IVb gene ( Fig. 1). Identical sequences were found in fragments of 4.3 kb of A Dash II-ratPDE4.2 and 5.2 kb of pWE15-ratPDE4.1, indicating that these clones overlap in this region. Comparison of the cDNA sequence with the genomic sequences allowed a tentative assignment of the introdexon boundaries (Table I). Most junctions followed the consensus proposed by Mount (30) and the GT-AG rules (31). The nucleotide sequences of the exons and introdexon junctions are reported in Fig. 2. This portion of the r a t P D E 4 m gene spans at least 35 kb, and it is arranged in 11 exons and 10 introns. The exons are located on five distinct EcoRI restriction fragments of 3.7 kb (including exons 1 and 2), 4.9 kb (exon 3), 5.2 kb (exons P 7 ) , 3.5 kb (exons [8][9][10], and 2.1 kb (exon 11). With the exception of the first and last exon, the lengths of the exons range from 93 to 183 bp, whereas the lengths of the introns, evaluated by restriction analysis of the genomic clones or PCR amplification, range from 0.5 to 10 kb.
To obtain all the coding fragments of the ratPDE3hVd gene, A Charon 4 and A Dash I1 libraries were screened with the following probes: a ratPDE3.1 cDNA containing the entire open reading frame (13), PCR fragment A B (obtained by amplifying the region corresponding to bp -3 to +260 on cDNA; see Fig. 4),

gctctttcctagccacttaatgcattttctt 232 t t c t t t a t a a a c a g T T C G A T G C~C C G G G A G C T G A C A C A C C T C T~G A T G A G~G A T C A G G~C~G T G T C~T A~T T T C~C A C -K E E S Y Q K L A M E T L E E L D W C L D Q L E T I Q T Y R S V S E M A S N K + A --A-F K R M L N R E L T H L S E M S R S G N Q V S E Y I S N T 318 GTTCTTAGgtaagatgttaacaggaaactcactgagcttttgc
... i n t r o n C . . . cactggcacacgtgtgctcaaaagcatgttcactgtt

A C C A T C A C C T C G C T G T G G G A T T~G C T C C T T~G A G~C A T T G C G A C A T C T T T~T C T T A C~~G~C G C C A G A C A C T~G~T G G T N H H L A V G F K L L Q E E H C D I F Q N L T K K Q R Q T L R K M V 1014 GATTGACATGqtgaggtgccagcctcatcccttacctacatttcatttt
... intron E . . . cttctcgactgacttgatgcgttttaattctg

t c t t t a c c a a g G T G T T A G C T G A T A T G T C C R R G C A C A T C T C C G G T G T V L A T D M S K H M S L L A D L K S M V E T K K V T S S G V
1113 TCCCTCCTGGA~CTATACGACCGGATACAGgtatgtgatgacgcaaagccaaagtaaacaaaaccaagcagta ... i n t m n I...ttatgtgta

attttatgggattttccaaaacttgatcatttcagGTTCTTCG~CATGGTA~TTGTGCAGACCTGAG~CCCTACC~GTCCTTGGAGTTGTATCG V L R N M V H C A D L S N P T K S L E L Y R 1212 G C A A T G G A C T G A T C G C A T C A T G G A G G A G T T T T T C C A A C A G 1312 TCTGTGGAAAAGTCCCAGgtatctacactgagattttttttcatgttgacgagctcgcc
... intron J... tagggcccatgcagtaacagat 1330 gtgttacttttcttcccagGTTGGTTTCATTGACTACATTGTCCATCCATTGTGG~CTGGGCAGACCTGGTTCAGCCTGATGCT~GACATTTTG

A G G T G A A C A T T C A T C C T T G G A G~T G C C R G C T G A G T G T 1811 TACTTGAGTTTGGAGC
letters. The location of the 3"splice site of intron A is marked by a downward arrow. This intron is spliced out in the brain mRNA, but h d i o n s The deduced amino acid sequence is shown below the nucleotide sequence using the single-letter amino acid code. Nucleotides are numbered from the first in-frame ATG codon (marked in boldface), and the numbers are reported on the left above the DNA sequence. The oligonucleotide primers used for PCR analyses are also indicated by arrows above the nucleotide sequence. Putative sites of transcription initiation are marked by white downward arrows. and PCR fragment ZIE (amplification product corresponding to bp 261435 on the ratPDE3.1 cDNA see Fig. 4). The genomic clones isolated were further characterized by the same strategy indicated for the ratPDE4/TVb gene. Clone Charon 4A-ratPDE3.1, retrieved by screening the library with the fulllength cDNA, covers most of the gene except the 5'-end. Clone , obtained by screening the library with PCR fragment I/E, overlapped with the Charon 4A-ratPDE3.1 clone.
No overlap was found with clone Dash 11-ratPDE3.2 (Fig. 3b), and one gap was present in the intron region between exons 2 and 3. The restriction map of the genomic clones isolated as well as the exons they include is shown in Fig. 3, whereas the nucleotide sequence of the ratPDE3/IVd introdexon boundaries is reported in Fig, 4. Introdexon boundaries of ratPDE3/ IVd contain the consensus 5'-and 3"splice sequences (Table 11) and were located in the same position as in ratPDE.Q/ZVb ( Genomic Southern Blot Analysis-Wistar rat genomic DNA digested with EcoRI was hybridized to ratPDE3lIVd and ratPDE4lIV4 cDNAs. The sizes of the genomic fragments hybridizing to ratPDE3lIVd and ratPDE4AVb cDNAs were comparable to those predicted from the restriction map of the corresponding genomic clones (Fig. 6). However, the doublet in the 2.8-kb region of ratPDE3lIVd and in the 3-and 5-kb regions of ratPDE4/IVb inferred from the restriction analysis of the genomic clones could not be completely resolved on the Southern genomic blot using the cDNAs as probe. Hybridization with exon-specific probes confirmed the presence of multiple fragments (data not shown).
Characterization of Different ratPDE3 lIVd and ratPDE4 I Nb mRNAs-Since none of the cDNAs encoding Sertoli cell ratPDE4/IVb (11) had an in-frame termination codon upstream from the first AUG codon, available cDNA libraries were used as template in a PCR to obtain cDNAs with more 5'-upstream sequence (see "Experimental Procedures"). Following this strategy, a clone with 100 base pairs of additional 5'-sequence of the ratPDE4AVb cDNA was isolated. This includes termination codons in all three reading frames, demonstrating that the AUG codon previously assigned as the first in-frame AUG codon of the ratPDE4AVb mRNA expressed in Sertoli cells was correctly identified (see Fig. 2).
As previously reported, ratPDE4/IVb cDNAs with different 5'-ends were retrieved from testis and brain libraries (11,12,14). Since the brain-specific sequence was not present in the r a t P D E m gene sequence obtained, reverse transcriptase PCR was performed with brain and Sertoli cell poly(A)+ RNAs. An antisense primer (oligonucleotide A) common to both tissue sequences was used. The sense primers were specific for the 5'-cDNA sequence of rat brain (oligonucleotide W) (12) and Sertoli cell sequence (oligonucleotide M). Amplification products were present only when the primers specific for each tissue were used, and the sizes of these fragments were identical to those predicted by the corresponding cDNAs (Fig. 7). Furthermore, oligonucleotide W, specific for the brain mRNA, did not hybridize either to the 3.7-kb EcoRI fragment containing the putative start site of ratPDE4AVb or to a 4-kb EcoRI fragment located upstream (data not shown). This suggests that the sequence present in the brain mRNA is, in the gene, more than 5 kb upstream from the Sertoli cell exon and that the start site of the Sertoli cell mRNA derives from intron sequences.
Sequencing of the 4.0-kb XbaI fragment of the ratPDE3lIVd gene containing exons 1 and 2 confirmed that the 87-bp sequence found in ratPDE3.1 cDNA is not a cloning artifact, but is rather due to unspliced intron sequence. The presence of the splicing variant mRNAs originating from this exon l/exon 2 boundary was further investigated by reverse transcriptase PCR. Poly(A)+ RNA was prepared from Sertoli and FRTL-5 thyroid cell cultures and from rat heart and brain tissues and used as template for reverse transcriptase PCR using the two oligonucleotides (B and A) flanking the putative intron as primers. Two products of 265 and 179 bp were amplified from the Sertoli cell and FRTL-5 mRNAs. These sizes correspond to those expected when the intron is spliced in or out (Fig. 8) and were identical to the products of amplification of the ratPDE3.1 and ratPDE3.2 cDNAs. In contrast, very little of the variant without the intron could be amplified from the heart mRNA. In spite of the fact that ratPDE3lIVd mRNA transcripts have been detected by Northern analysis of brain mRNA (131, the two primers could not amplify the proper fragments in any of the four different brain mRNA preparations used. That the ratPDE3lIVd mRNA is present in these brain mRNA preparations was demonstrated by the fact that primers in exons 3 and 4 of ratPDE3lIVd amplify fragments of the correct size (data not shown). This latter finding is compatible with the hypothesis that additional splice variants at the boundary of exons 1 and 2 of ratPDE3lIVd are expressed in the brain.
To determine whether the intron sequence is present only in high molecular mass unprocessed RNAs or in mature mRNA, a probe generated by PCR and corresponding exclusively to the intron sequence was labeled and used for Northern blot hybridization of Sertoli cell mRNA. A transcript of 6.7 kb hybridized to this probe (data not shown). This is the same size as the transcript that hybridizes to a cDNAcontaining the complete coding region (13), therefore confirming that this intron sequence ap-

Bintron A t A -E T L Q T R H S V S E M A S N K
. .. intron 1 ... a t c t t c t a c t c c t

t t c c t t c a c c c t a g W G T~C A R G T G G G G C C T C C A C G T T T T C~T f f i C G G A G C T G T C T G G~C G G C C T C T G A C T G T T A T C A T G C
ACACCATTTTTCAGgtaaccggcctctgactgt ... intron 1 ... ccggggaatgagttttaacaggaagacaaaaaatgttaagaggaaagat

D V L A K E L E D V N K W G L H V F R I A E L S G N R P L T V I M V A Y H N N I H A A D V V Q S T H V L L S T P A L E A V F T D L E I L A A I F A S A I H D V D H P G V S N Q F L I N T N S E L A L H 902 A C A A C G A C T C C T C C G T C T T A G A G A A T C A T C A T T T G G C T G T G G G C T T T~T T G C T C C A G~C T G T~T T T T C~T C T G A C~C A Y N D S S V L E N H H L A V G F K L L Q E E N C D I F Q N L T K K Q
1002 ARGRCAATCTTTARGGAAAATGGCCATTGACATTgtgagtaacaagaagagagatccgagttcttagagatgacatcaggttc ... intron ti ...

a g c t t t a g t c c t a a a a c t g t c t t t g c c c c g t c a g G T A C T A G C G A~C A T G T C A R A G C A C A T G A A T C T G C G
1102 AAGARGGTGACGAGCTCTGGCGTCCTCCTCCTTGAT~CTATTCTGA~GATCCAGgtaaatcagcatctcagtctgttcttgatgcacgtgttagtga 1159 ... intron I ... aaagaacagatgtggataaagaatgctccatccaccccagGTCCTC~TATGGTGCACTGTG~CCTGAGCAACCC~

R Q S L R K M A I D I V L A T D M S K H M N L L A D L K T M V E T K K V T S S G V L L L D N Y S D R I Q V L Q N M V H C A D L S N P 1202 CAAAGCCACTCCAGCTCTACCGCCAGTGGACGGACCGGAT~TG~TTCTTCCGTCAGGGGGACCGG~CGTGAGCGTGGCATGGAGAT~TCC T K P L Q L Y R Q W T D R I M E E F F R Q G D R E R E R G M E I S P
1302 CATGTGTGACARGCACAACGCCTCTGT~TCACAGgtaattcagcggcactgagaaaggacagaggtgcagaggtgcag ... intron J... pears in mature mRNA.

letters. The deduced
Mapping of mtPDE3lNd and mtPDE4lNb Danscription Start Sites in Sertoli Cells--In a preliminary attempt to determine the transcription initiation sites active in the Sertoli cell, primer extension and reverse transcriptase PCR experiments were performed using mRNA derived from Sertoli cells incubated for 24 h in the presence of 1 ~l l~ dibutyryl CAMP. This incubation was included to increase the relative abundance of the two mRNAs (13). Primer extension was performed with oligonucleotides K and C, which are complementary to the ratPDE4/TVb sequence (see Fig. 2). In the three independent experimenta performed, several extension products were observed with both primers. Extension with oligonucleotide K revealed major products of 615 and 550 bp (Fig. 9) and one or two faint bands of 750 and 840 bp, respectively. Two major (194and 249-bp) and two minor (480-and 340-bp) products were observed when oligonucleotide C was used. The predominant extension product of 615 bp using oligonucleotide K corresponds to the product of 249 bp extended with oligonucleotide C and terminates between bases -449 and -439. The CAP sequence TCAGTG'IT is present in this location and is 41 bp downstream from a sequence homologous to a TATA box (32).
An additional transcription start site upstream from this location is tentatively located between bases -673 and -663. Reverse transcriptase PCR using specific primers prepared according the genomic sequence confirmed the above data: PCR products were present amplifying RNA with sense oligonucleotide H or L and antisense oligonucleotide N, but not when amplification of the brain mRNA. Location of the exons is based on the numbering of the sequence reported in Fig. 4, where the A residue of the Exon sequences are indicated by upper-case letters, and intron sequences by lower-case letters. Exon l a is a putative exon inferred by the PCR first ATG codon is base 1. oligonucleotide F was used (Fig. 10). No products were evident amplifying RNA that had not been reverse-transcribed (data not shown), thus excluding potentially contaminating genomic DNA. The same set of experiments was conducted to map the start site of the PDE3 gene. Fig. 9 shows the results of one of the primer extension experiments. The two major products of extension corresponded to sizes of 320 and 460 base pairs. The two initiation start sites are located 463 and 603 base pairs upstream from the putative initiation AUG codon, respectively.
PCR analysis of the Sertoli cell mRNA confirmed that at least 400 base pairs were present in the ratPDE3m7d CAMP PDE message (data not shown).
Different mRNA Variants Are nanslated into Proteins of Different Mass-To test whether the different mRNA variants are translated into different protein products, ratPDE3.1, ratPDE3.2, and Sertoli cell ratPDE4 cDNAs were inserted into the pCMV eukaryotic expression vector and expressed in COS cells. The expression of Sertoli cell r a t P D E 4 M has been previously reported (14). Both ratPDE3KVd cDNA variants produced proteins with catalytic activity and immunoreactive bands recognized by the K116 antibody. The ratPDE3.1 cDNA directed synthesis of proteins of 72-74 kDa and a doublet of 67-68 kDa (Fig. 11). The ratPDE3.2 cDNAexpression led to the appearance of only the 67-68-kDa band doublet. The migration

FIG. 7. Evidence for diffemntid mplidng of retPDJWlVb in
Sertoli and FRTL.45 thyroid cells and in brain. An aliquot of 1 pg of poly(A)+ RNA extracted from Sertoli and FRTL-5 cell cultures and from brain tissue was reverse-transcribed into cDNA. The cDNA was then subjected to PCR using, as primers, the oligonucleotides depicted in the scheme. PCR products were analyzed by electrophoresis on a 3% agarose gel and ethidium bromide staining. Identical RNA samples were incubated in the absence of reverse transcriptase and then used for amplification with the same primers. The products of these control reactions were loaded on lane 3. HindIII-digested DNA fragments of A phage were used as size markers. The experiment was repeated three times with identical results. of the latter 67-68-kDa doublet is identical to that of the purified CAMP PDE from the Sertoli cell (13). DISCUSSION The data reported provide evidence that multiple CAMP PDE mRNAs are derived from the ratPDE3md and ratPDE4M genes and are expressed in a tissue-specific manner in the brain and in endocrine cells. These mRNA variants are derived from different transcription initiation sites and alternative exon usage andlor differential splicing of introns. Thus, intri- from Sertoli and FRTL-5 cells and from the brain and heart were reverse-transcribed into cDNA. The cDNA was then amplified using oligonucleotides A and B as primers. Plasmid containing ratPDE3.1 and ratPDE3.2 cDNAs were used as positive controls. PCR products were analyzed by electrophoresis on a 3% agarose gel with ethidium bromide staining. Amplification with RNA not reverse-transcribed did not produce any amplification product (data not shown). HindIII-digested DNA fragments of A phage were used a s size markers. The experiment was repeated four times with identical results. cate regulatory mechanisms serve to control the expression and function of the CAMP PDE genes in a tissue-specific manner.
The ratPDE3KVd and r a t P D E 4 M mRNAs expressed in the Sertoli cell derive from the assembly of 11 coding exons. Exons 5-10 of the two genes encode the catalytic region of the protein.
This assumption is based on deletion mutations performed on ratPDE3md cDNA (33) that demonstrated that removal of sequences in exons 1 4 and 11 produces a protein still catalytically active. Site-directed mutagenesis of ratPDE3md also shows that residues crucial for structure and function of the catalytic site are located in exons 7 and 8 (33). These are among the most conserved exons. The sequences of exons 7-9 and the location of the intron splicing are similar to those of exons 13 and 14 of the P-cGMP-specific retina PDE (341, suggesting that structural homologies are present between different families of PDE genes. In addition to the five exons corresponding to the catalytic domain, it was also noted that exon 2, adjacent to the alternative splicing site, contains a domain highly conserved in all rat and human CAMP PDE sequences available (10)(11)(12)(13)35,36). Exon 4 contains a highly charged sequence (KEKEXKKR) in ratPDE3md and ratPDE4M. This motif is absent in the Drosophila dunce PDE and ratPDE2ma, where an insertion of 8-20 amino acids has occurred.
It is likely that similar introdexon distribution is present in the other two CAMP PDE genes present in rodents (ratPDEl/ Tvc and ratPDE2ma). Davis et al. (10) have retrieved cDNAs with multiple 5'-ends from rat brain libraries. One of these   and extended with avian myeloblastosis virus reverse transcriptase. The primer-extended products were separated on a denaturing polyacrylamide gel as described under "Experimental Procedures." The major extension products for each specific primer are indicated by the arrows, followed by its estimated length. The sizes of the extended products were determined relative to the migration of A phage markers. Extension experiments were repeated three to four times with similar results.  (1 pg) was reverse-transcribed into cDNA and subjected to PCR. PCR products were analyzed by electrophoresis on a 3% agarose gel with ethidium bromide staining. Plasmid containing the 5'-end of the gene was used as a positive control. To control for genomic DNA contaminating the RNA preparations, the reverse transcriptase reaction was omitted in some experiments. The oligonucleotides used as primers and the sizes of the expected amplified fragments are also shown. HindIII-digested DNA fragments of A phage were used as size markers. The experiment was repeated three times with identical results. The transcription start sites identified by primer extension are located between oligonucleotides F and C. cDNAs, RD3, has a region of 99 base pairs removed. Since the boundaries of the sequence missing in RD3 are in a location identical to that of exon 5 of ratPDE3md and ratPDE4/lVb, this region probably corresponds to an exon also in the  FIG. 11. Western blot analysis of recombinant ratPDE3Wd PDE variants expressed in COS cells. Cells were transfected with pCMV-ratPDE3.1 and pCMV-ratPDE3.2 expression vectors. The ratPDE3.2 cDNA does not contain the first intron sequence, whereas ratPDE3.1 still contains the intron. After 48 h of culture, cells were harvested, and the PDE activity was immunoprecipitated using a polyclonal antibody. The characteristics of the antibody and the immunoprecipitation procedure have been previously reported (14,33). The proteins absorbed were eluted from the antibody with SDS and analyzed by SDS-polyacrylamide gel electrophoresis. After transfer to a polyvinylidene difluoride membrane, the blot was incubated with the same antibody and with radioactive protein A. Immunoprecipitated samples from cells processed for transfection with plasmid without the insert are also shown (Mock). ratPDE3.2, plasmid with ratPDE3 cDNA without the intron; ratPDE3.1, plasmid with ratPDE3 cDNA with the intron. ratPDE2ma gene. The significance of the removal of this exon is unknown, but it again points to an intricate splicing pattern of this CAMP PDE gene. Several introdexon boundaries are located in the same position in the D. melanogaster gene and in ratPDE3md and ratPDE4m (19). One major difference between the structures of the ratPDE3/IVd and r a t P D E 4 m genes (this report) and of the Drosophila dunce PDE gene is that additional noncoding and coding exons are present at the 5'-end of the latter gene. Exon 2 of ratPDE3md and ratPDEW MI is homologous to exon 6 of the Drosophila dunce gene. We should emphasize that the exon composition that we report is deduced from the characterization of the mRNAs expressed in the Sertoli cell. On the basis of our preliminary findings and the homology to the Drosophila dunce PDE gene, it is likely that the ratPDE3 and ratPDE4 mRNAs expressed in the brain contain additional 5'-exons (Fig. 12).
A distinctive feature of the ratPDE3md and ratPDE4IVb mRNAs derived from the Sertoli cell is the B'-untranslated sequence. Unlike what is observed for ratPDE4/IVb, the ratPDE3md 5'-untranslated region is very GC-rich (70%), suggesting a complex secondary structure of this mRNA. Furthermore, whereas the ratPDE3md 5'-untranslated sequence contains only one AUG codon, the ratPDE4lWb 5"untranslated sequence has nine AUG codons in the three reading frames. These AUG codons are all followed by stop codons determining the presence of short open reading frames. It has been proposed that the presence of these "mini-open reading frames" and of AUG-burdened 5"untranslated sequence may substantially reduce the translatability of the mRNA (37). According to the mRNA scanning hypothesis (37), r a t P D E 4 m mRNA should be translated at a slower rate than ratPDE3md.
It is striking that the beginning of the heterogeneity of the ratPDE3md mRNA was found at the exact same location as the tissue-specific r a t P D E 4 m mRNA alternative splicing the region absent in the brain mRNA. The stippled box in the Sertoli cell mRNA represents the coding region derived from the intron start site, type I mRNArepresents the coding sequence derived from the unspliced intron. The broken box marked IA? in the ratPDE3KVd gene represents by PCR amplification data on brain mRNA and by comparison with the the upstream exons utilized in the brain mRNA. Its presence is inferred Drosophila dunce gene.
boundary. This, together with the fact that the boundary of exons 1 and 2 is highly conserved in the two genes, is an indication that evolutionary pressure has maintained the intron sequence and the splicing variants. If splicing mRNAvariants are found at the same boundary in other species, it would be further proof that this splicing is conserved and physiologically relevant. PCR analysis of mRNAs from Sertoli and FRTL-5 cells shows that a subclass of ratPDE3liVd transcripts contains an intron sequence that is probably spliced out at a rate much slower than the other introns. That this intron splicing is delayed is indicated by the finding that oligonucleotide Ais able to amplify the mRNA containing this intron. This oligonucleotide was designed across the splice site between exons 2 and 3. Thus, if intron B had not been removed, amplification would not have occurred. That a substantial portion of the mature mRNA contains the intron sequence and that this intron is removed late are also indicated by the finding that a probe corresponding to the intron sequence hybridizes to the major Sertoli cell transcript of 6.7 kb. Since, in our experiments, nuclear RNAwas not separated from cytoplasmic mRNA, it is not know whether the intron sequence is present only in nuclear RNA or also in cytoplasmic RNA. Pending further studies to define this point, the splicing of this intron could be a mechanism by which mRNA translocation to the cytoplasm andor translation is delayed. This would be similar to what is shown for other mRNAs in which unspliced intron sequences serve to control the rate of translation (38,39).
It is also possible that the presence of the unspliced intron in the ratPDE3liVd mRNA serves to generate proteins with different amino termini since it is predicted that two different AUG codons are used in the two type I and I1 mRNAs (see Fig.  12). That both ratPDE3liVd mRNAs appear to be translated into functional proteins is documented by the transient transfection experiments. Expression of the cDNA lacking the intron (ratPDE3.2) produces a doublet of 67-68 kDa and an active PDE. The expression of a cDNA containing the intron sequence (ratPDE3.1) produces instead two polypeptides, one of 72-74 kDa and one of 67-68 kDa. Thus, the higher molecular mass polypeptide most likely originates from the mRNA with the intron sequence (type I mRNA). The cause of the appearance of the lower 67-68-kDa doublet after transfection of the cDNA containing the intron could be 2-fold. Some splicing of the intron might occur during the transient transfection, and both mRNAs would be translated into proteins of different molecular mass. Alternatively, both AUG codons of the type I mRNA might be used as the result of leaky scanning (37). Since the endogenous Sertoli cell ratPDE3Nd protein migrates as a doublet with a molecular mass of 67-68 kDa, we can conclude that this originates from an mRNA in which the intron has been removed (type I1 mRNA). In addition, unlike what is seen with transient transfection, stable transfection of the ratPDE3.1 cDNA produces only a protein of 67-68 kDa.3 The whole of these data would indicate that the type I1 mRNA is the form translated in vivo and that the presence of the intron has no bearing on the protein sequence in the Sertoli cell. However, amplification of the rat heart mRNA demonstrates that most of the RNA contains the intron sequence (type I mRNA), and a 72-74-kDa immunoreactive protein was detected by Western analysis of extracts from this t i~s u e .~ These latter findings are consistent with the hypothesis that, unlike what is observed in the Sertoli cell, ratPDE3liVd expressed in the heart is derived from the type I mRNA and that removal of intron A might be tissue-dependent.
Another interesting result is the fact that oligonucleotides corresponding to exons 1 and 2 of ratPDE3liVd can amplify a product from the testis, thyroid cells, and heart, but not from brain mRNA, whereas oligonucleotides in exons 3 and 4 and ratPDE4/IVb-specific oligonucleotides are able to amplify a band of the correct size from brain mRNA. Since previous Northern blot analyses have shown that ratPDE3liVd mRNAs are abundant in the brain (131, it has to be postulated that an additional splice variant in the region between exons 1 and 2 is present in the brain (see Fig. 12). This indicates that an even more intricate regulation of exon splicing occurs in the ratPDE3liVd gene in the brain and opens the possibility that additional promoters might be present, as we have suggested for ratPDE4/IVb. Experiments are underway to clarify this point.
Our data show that transcription of these genes in the Sertoli cell is directed by intron promoters and suggest that the expression of these CAMP PDEs in the brain is directed by additional upstream promoters. Once the structure of the genomic regions active in the brain is characterized, it will be possible to confirm whether these promoters are indeed brain-specific. In view of our finding that the expression of the CAMP PDEs is hormone-and m P -d e p e n d e n t i n endocrine cells (13, 141, it will be important t o define whether CAMP regulation of transcription of these PDE genes is dependent on one or more promoters, whether different promoter usage causes differences in CAMP sensitivity, and whether hormones can dictate which promoter is used. Certainly, the expression of these CAMP PDEs appears to be very finely regulated, implying that these enzymes play a crucial role in the function of the brain and of endocrine organs.