Alternative splicing in the N-terminal extracellular domain of the pituitary adenylate cyclase-activating polypeptide (PACAP) receptor modulates receptor selectivity and relative potencies of PACAP-27 and PACAP-38 in phospholipase C activation.

Pituitary adenylate cyclase-activating polypeptide (PACAP)-27 and PACAP-38 are neuropeptides of the vasoactive intestinal peptide/secretin/glucagon family. We previously described alternative splicing of the region encoding the third intracellular loop of the PACAP receptor generating six isoforms with differential signal transduction properties (Spengler, D., Waeber, C., Pantaloni, C., Holsboer, F., Bockaert, J., Seeburg, P. H., and Journot, L. (1993) Nature 365, 170-175). In addition, we demonstrated that the potencies of the two forms of PACAP are similar for adenylate cyclase stimulation, whereas PACAP-38 is more potent than PACAP-27 in phospholipase C activation. In the present work, we document the existence of a new splice variant of the PACAP receptor that was characterized by a 21-amino-acid deletion in the N-terminal extracellular domain. We demonstrate that this domain modulates receptor selectivity with respect to PACAP-27 and -38 binding and controls the relative potencies of the two agonists in phospholipase C stimulation.

We and others (15)(16)(17)(18)(19) recently cloned the rat PACAP 1 -R cDNA. In addition, we demonstrated that PACAP 1 -R hnRNA is alternatively spliced and gives rise to six variants (18,20). Two cassettes named "hip" and "hop" are possibly inserted at the end of the third intracellular loop of the receptor. The resulting variants were named PACAP 1 -R s (the shortest form, without cassette), PACAP 1 -R hip, PACAP 1 -R hop1, PACAP 1 -R hop2, PACAP 1 -R hip-hop1, and PACAP 1 -R hip-hop2. They display differential signal transduction properties upon expression into LLC PK1 cells and Xenopus oocytes. The short as well as the hop variants potently activate both AC and PLC. On the other hand, the hip variant does not stimulate PLC, whereas the hip-hop variants display an intermediate signal transduction pattern with an altered ability to activate PLC, which is best evidenced upon expression into Xenopus oocytes. Another remarkable property of PACAP 1 -R is the difference in the potencies of PACAP-38 and -27 in AC and PLC activation. Both agonists display similar EC 50 values in AC stimulation, whereas PACAP-38 is significantly more potent than PACAP-27 in PLC activation (3,18,(21)(22)(23). The combination of alternative splicing and activation by different ligands possibly induces an elaborated pattern of PLC stimulation by PACAP 1 -R.
Since the cloning of the first hormone G protein-coupled receptor almost a decade ago (24), numerous studies have addressed the importance of the different receptor domains in G protein activation and ligand binding (reviewed in Ref. 25). PACAP 1 -R belongs to the VIP-secretin-glucagon-calcitonin receptor family. These receptors share good homologies with each other but not with other members of the G protein-coupled receptor superfamily. At present, few structure-function studies have been performed on members of this family. Available data involve the extracellular domains of this family of receptors in ligand binding. For instance, two alternatively spliced forms of the rat calcitonin receptor were characterized. The C1b isoform displays a 37-amino acid insertion in the first extracellular loop as compared with the C1a isoform. The presence of this insertion induces a 40-fold reduction in affinity for the iodinated calcitonin, indicating that this domain is crucial for high affinity calcitonin binding (26). In addition, experiments with chimeric receptors suggested a role for the Nterminal extracellular domain of the calcitonin receptor in calcitonin binding (27). Similarly, it was demonstrated that the extracellular N-terminal domain of PACAP/VIP 1 -R is critical for VIP binding (28,29). Moreover, mutation of some residues conserved among the PACAP-VIP-secretin-glucagon receptor family (cysteine 63, aspartate 68, cysteine 72, tryptophan 73, cysteine 86, glycine 109) completely abolished VIP binding (30,31). This domain (residues 63-109) is therefore likely to interact with VIP. The authors have not assessed whether PACAP binding is also altered by the same mutations, and it is therefore not known whether the same domain also interacts with PACAP. Regarding the PACAP 1 -R, Cao and co-workers (32) demonstrated that the N-terminal extracellular and the first transmembrane domains are sufficient to confer high affinity PACAP binding. Interestingly, the PACAP 1 -R displays a 21-aa sequence in the N-terminal extracellular domain, which has no homologous domain in any other member of the VIP-secretinglucagon receptor family. The present study established the existence of a new splice variant missing the above mentioned 21 aa and described its functional properties regarding binding and activation of AC and PLC.

EXPERIMENTAL PROCEDURES
Screening of Mouse Genomic and Human cDNA Libraries-A genomic library (a generous gift from Dr. Philippe Soriano, Seattle) was constructed by partial digestion of genomic DNA from 129/Sv mouse strain with MboI and subcloning of the digested fragments into DashII (Stratagene). We screened 1.9 ϫ 10 6 clones with a random primed probe derived from the rat PACAP 1 -R cDNA (1) and isolated 18 positive clones. Clone D3313 containing exons 5, 6, and 7 was digested with NcoI, the resulting fragments were subcloned into pGEM5zf(ϩ) (Promega), and the exon-intron boundaries were sequenced using T7 DNA polymerase (Pharmacia). To determine the sequence of the junction between exon 4 and the 3Ј intron, we performed three independent PCRs on mouse genomic DNA using primers derived from the sequence of exon 4, i.e. 5Ј-TCACATGTTGGAAGCCTGCTC-3Ј, and from the 5Јterminal sequence of D3313, i.e. 5Ј-TGGATAGAAAGTCAAGGATGG-3Ј. The resulting 1.5-kilobase product was directly sequenced with a nested primer, 5Ј-GGTGAGATGGTCCTTGTGAGC-3Ј, using the fmol Sequence System (Promega). A human fetal (5-month-old abortion) brain cDNA library was constructed in Zap II (a generous gift from Dr. Peter J. Flor, Ciba-Geigy, Basel, Switzerland). We screened 1.2 ϫ 10 6 clones with a random primed probe derived from the rat PACAP 1 -R cDNA and isolated seven positives clones, which were excised with helper phage and sequenced as above. Full-length cDNAs were reconstructed using RT-PCR and restriction enzymes and subcloned into pRK5, a cytomegalovirus-based expression vector (33). All clones were sequenced on both strands as above.
Transfection of LLC-PK1 Cells-LLC PK1 (a generous gift from Dr. Falk Fahrenholz, Max Planck Institut fü r Biophysik, Frankfurt, Germany) were grown in Dulbecco's modified Eagle's medium with 10% fetal bovine serum and penicillin/streptomycin (Life Technologies, Inc.). Cells were trypsinized, centrifuged, resuspended in EP1x buffer (50 mM K 2 HPO 4 , 20 mM CH 3 CO 2 K, 20 mM KOH, 26.7 mM MgSO 4 , pH 7.4, with CH 3 C0 2 H) and mixed with plasmid DNA. After 15 min at room temperature, 300 l of cell suspension were transferred to a 0.4-cm electroporation cuvette (Bio-Rad) and pulsed using a Gene pulser apparatus (Bio-Rad) set at 960 microfarads/260 V. Cells were immediately transferred to a tube, diluted with Dulbecco's modified Eagle's medium, 10% fetal bovine serum, and plated at the desired density.
Binding of [ 125 I]PACAP-27-PACAP-27 was iodinated using the iodogene method. Briefly, 4 nmol of PACAP-27 were mixed with 1 mCi [ 125 I]NaI in phosphate buffer, pH 7.1, in tubes precoated with 15 g of iodogene. After 10 min at room temperature, the reaction was stopped by the addition of 400 l of 18% acetonitrile in 0.07% trifluoroacetic acid. Iodinated PACAP-27 was subsequently purified by two successive runs of high pressure liquid chromatography on a 50734 sphere 100RP18 Ncapt column (Merck) using a gradient of acetonitrile (18 -48% in 0.07% trifluoroacetic acid). The purified iodinated peptide was diluted in 50 mM Tris-HCl, pH 7.4, 5 mM MgCl 2 , 2.5 mg/ml bovine serum albumin and stored at Ϫ70°C. All binding experiments were performed on intact cells at 4°C. [ 125 I]PACAP-27 was diluted at the desired concentrations in HBS buffer (20 mM Hepes, pH 7.4, 150 mM NaCl, 4.2 mM KCl, 0.9 mM CaCl 2 , 0.5 mM MgCl 2 , 0.1% glucose, 0.1 mg/ml bovine serum albumin), and transfected cells were incubated for 4 h at 4°C with the different dilutions. At the end of the incubation period, cells were washed once with phosphate-buffered saline, 0.1 mg/ml bovine serum albumin. 500 l of 0.5 M NaOH was added to solubilize the cells, and radioactivity was measured using a ␥-counter (Kontron). Data were analyzed with the EBDA-Ligand software.
Stimulation of Luciferase Activity-Cells were cotransfected with a plasmid carrying a cAMP-regulated reporter luciferase cDNA (p⌬MC16-Luc) (18). Cells were washed and incubated overnight in Dulbecco's modified Eagle's medium without fetal bovine serum. PACAP-27 and -38 were added for 4 h before cells were lysed, and luciferase activity was determined by luminometry. Data were fitted to a four-parameter logistic equation to obtain response parameter (EC 50 , E max , n H ) with the Kaleidagraph software. Student's t test was per- formed with Statview software.
Total Inositol Phosphate Measurement-Transfected cells were washed twice with phosphate-buffered saline and incubated overnight with 1 Ci of myo[ 3 H]inositol in Dulbecco's modified Eagle's medium without fetal bovine serum. Medium was aspirated, and cells were incubated in HBS, 10 mM LiCl for 10 min at 37°C. 10-fold concentrated PACAP-27 or -38 dilutions were added, and the incubation was continued for a further 30 min. Medium was aspirated, and the reaction was stopped with 1 ml of 5% perchloric acid. Levels of inositol phosphate were determined as described previously (35), and data were fitted to a four-parameter logistic equation to obtain response parameter (EC 50 , E max , n H ) with the Kaleidagraph software. Student's t test was performed with Statview software.

Mouse PACAP 1 -R Gene Isolation and Characterization-Us-
ing the rat PACAP 1 -R cDNA as a probe, we isolated several genomic clones encoding the mouse PACAP 1 -R gene. It is composed of at least 18 exons and spans more than 50 kilobases.
Analysis of the exon-intron structure indicated that two separate exons encode a 21-aa sequence that is specific to PACAP 1 -R as compared with other members of the VIP-secretin-glucagon receptor family of G protein-coupled receptors (Fig. 1). In addition, splicing out of these exons would keep the reading frame and generate a shorter form of PACAP 1 -R missing the 21-aa in the N-terminal extracellular domain.
Existence and Distribution of the New PACAP 1 -R Splice Variant-To test whether such a splice variant may exist, we designed oligonucleotides flanking the 21-aa domain and performed RT-PCR on total RNA from various mouse tissues. In all tissues expressing PACAP 1 -R, we evidenced a major band corresponding to the size expected for the known PACAP 1 -R sequence (Fig. 2). In addition, the hypothalamus/thalamus and pituitary and adrenal glands displayed an additional band, the size of which was compatible with the deletion of 63 bp (Fig. 2). To establish the identity of this band, we isolated and subcloned both PCR fragments into pBlueScript and sequenced both strands using T3, T7, as well as internal primers. The sequences of these bands were identical except for the absence in the shortest fragment of 63 bp encoding the above mentioned

21-aa domain (data not shown). This demonstrates the existence of a previously uncharacterized splice variant of the Nterminal extracellular domain of the PACAP 1 -R.
Isolation of Human PACAP 1 -R cDNA-To isolate a fulllength cDNA encoding the new splice variant and to characterize its functional properties, we screened a fetal brain cDNA library. We chose a human library for use of the isolated clones in future pharmacological studies. We isolated seven independent clones encoding the human PACAP 1 -R. Most of these were partial clones, and several contained unspliced introns. Clone C19 contained the entire coding region and was the human homologue of the rat PACAP 1 -R s (18) (94% homology). This sequence was identical to the sequence published by Ogi and co-workers (36) except in the 5Ј region. These authors found a long repeat in the 5Ј-coding and non-coding regions that has no homologous sequence in the rat PACAP 1 -R. The sequence we found did not display this repeated sequence and was strictly homologous to the PACAP 1 -R sequences already published. The clone described by Ogi and co-workers (36) was therefore likely to result from rearrangement in the 5Ј region of their cDNA. The other clones we isolated were partial and corresponded to human homologues of the rat hip, hop1, and hop2 splice variants. This demonstrated that alternative splicing of the region encoding the third intracellular loop of PACAP 1 -R also occurred in human, at least in fetal brain. In addition, clone C24 started at codon encoding N60 and was otherwise identical to clone C19 in the coding region but for the deletion of the sequence encoding V89 to S109 in the putative extracellular N-terminal domain. Two additional clones displayed the same deletion. These results indicated that alternative splicing of the region encoding V89-S109 also occurred in human fetal brain. Using PCR and restriction enzymes, we reconstructed the human PACAP 1 -R hip and hop1 splice variants as well as the new variant. Since this latter variant was shorter than the variant referred as "s" (short), we named it "vs" for "very short" (Fig. 1). All reconstructed clones were entirely sequenced on both strands (data not shown) and subsequently subcloned into pRK5, a cytomegalovirus-based expression vector.
Functional Characterization-We transiently transfected LLC PK1 cells by electroporation and performed binding experiments and second messenger measurements on intact cells to characterize the functional properties of PACAP 1 -R vs as compared with those of PACAP 1 -R s. We used a dose of each plasmid, which resulted in expression of approximately 5000 receptors per cell. Scatchard plots on intact transfected cells were determined for each experiment to verify that both variants were expressed at similar densities at the cell surface.
Since the 21-aa domain is strongly acidic and since the C-terminal extension of PACAP-38 is rich in basic residues, we hypothesized a direct interaction between these domains. To test this model, we measured the displacement of [ 125 I]PACAP-27 by PACAP-(6 -38) at each splice variant on intact cells. Binding of PACAP-(6 -38) was not decreased by the absence of the 21-aa domain (Fig. 4).
Regarding signal transduction properties of the vs variant, potencies in stimulation of cAMP-stimulated luciferase activity by PACAP-27 and -38 were not significantly different for PACAP 1 -R vs and differed only 2-fold for PACAP 1 -R s (Table I and Fig. 5). On the other hand, total inositol phosphate measurements indicated that PACAP-27 and -38 were closely potent at PACAP 1 -R vs whereas, as previously shown, PACAP-38 was one order of magnitude more potent than PACAP-27 at PACAP 1 -R s (Table I and Fig. 6). DISCUSSION The partial structure of the mouse PACAP 1 -R gene presented in this study is in agreement with recently reported data (37). Interestingly, the exon-intron structure of the PACAP 1 -R gene is conserved as compared with that of other members of the VIP-secretin-glucagon receptor family except for two regions: (i) the third intracellular loop for which two cassettes named hip and hop are possibly inserted (18) and (ii) a 21-aa domain in the N-terminal extracellular region, which is also encoded by two exons. We previously demonstrated alternative splicing of the exons encoding the hip and hop cassettes. The present study demonstrates that the exons encoding the 21-aa domain are also alternatively spliced in the mouse PACAP 1 -R. Cloning of human PACAP 1 -R cDNAs confirmed the existence of the hip and hop splice variants (data not shown) as well as variants missing the 21-aa domain. Concurrent alternative splicing at both sites is possible since we also isolated clones with the hip or hop cassettes and missing the 21-aa domain (data not shown). In our previous study (18), RT-PCR and in situ hybridization experiments demonstrated that alternative splicing of the exons encoding the hip and hop cassettes occurs in all tissues expressing PACAP 1 -R, with the s and hop variants being the most abundant. On the contrary, alternative splicing of the N-terminal domain was restricted to pituitary and adrenal glands and hypothalamus/thalamus, which suggests specific function of the vs variant.
In agreement with previous studies (32), our results confirm that the N-terminal extracellular region of PACAP 1 -R is involved in ligand binding since alternative splicing of the 21-aa domain modulates selectivity for agonists binding. Note, the 21-aa domain is located in a region that has been shown to contain key residues for VIP binding at PACAP/VIP 1 -R (residues 63-109) (30,31). The mechanism by which the 21-aa domain influenced binding remains obscure. This domain behaved as an inhibitor of PACAP-27 binding. Since the only difference between PACAP-27 and -38 is the C-terminal extension of PACAP-38, and since both variants bound PACAP-38 with the same affinity, one can postulate that the inhibitory influence of the 21-aa domain was neutralized by the C-terminal extension of PACAP-38. Previous studies (38 -40) suggested that PACAP-38 interacts with its receptor through two distinct domains: (i) the N-terminal domain (amino acids 1-5), which is shared with PACAP-27 and is the major determinant of PACAP binding and AC activation, and (ii) the C-terminal extension specific to PACAP-38 interacts with a distinct site on the receptor as deduced from the high affinity binding of PACAP-(6 -38) as compared with PACAP- (6 -27). Note, the 21-aa domain displays 7 acidic residues, and the C-terminal extension of PACAP-38 displays 6 basic residues, suggesting a possible direct interaction. If such an interaction would occur, one can predict that deletion of the 21-aa domain should strongly impair PACAP-(6 -38) binding. In contrast, we showed that binding of PACAP-(6 -38) was not affected by deletion of the 21-aa domain, indicating that a direct interaction is not involved. Therefore, the C-terminal extension of PACAP-38 more likely interacts with the extracellular domains of the receptor at another site, the conformation and accessibility of which is not modified by the presence of the 21-aa domain.
The main functional consequence of the 21-aa deletion is an increased potency in PLC stimulation by PACAP-27, which is almost as potent as PACAP-38 at the vs variant. The physiological significance of PLC stimulation by PACAP-27, however, remains a matter of debate. In adult rats under resting physiological conditions, PACAP-38 is more abundant than PACAP-27 in all tissues expressing PACAP, indicating that the effects of PACAP are mainly due to PACAP-38 (41). However, in the anterior pituitary gland PACAP-38 is only two to four times more abundant than PACAP-27 (41), indicating a possible significant contribution of PACAP-27 to PACAP action. In the hypothalamus, PACAP-38 is 15 times more abundant than PACAP-27, but since PACAP is expressed at very high levels (575 ng of PACAP/g of wet tissue), PACAP-27 concentration is elevated and may significantly influence target cells (41). Interestingly, we demonstrated alternative splicing of the exons encoding the 21-aa domain in the pituitary gland and the hypothalamus. In those tissues where PACAP-27 is possibly involved in PACAP action, regulation of alternative splicing of exons encoding the N-terminal domain of PACAP 1 -R is therefore a possible way of tuning PLC stimulation by PACAP. The vs variant would significantly contribute to stimulation of PLC by PACAP-27, whereas insertion of the 21-aa domain in the N-terminal extracellular region of PACAP 1 -R would lead to impairment of the PLC stimulation by the same dose of PACAP-27.
The PACAP receptor therefore provided a unique example among G protein-coupled receptors of modulation of signal transduction properties by alternative splicing of extracellular and intracellular domains. Further experiments at the single cell level, however, will be necessary to correlate expression of the different PACAP 1 -R splice variants with functional parameters such as intracellular calcium concentration.