Characterization of a novel VPAC₁ selective agonist and identification of the receptor domains implicated in the carboxyl-terminal peptide recognition

Abstract

1. Vasoactive Intestinal Polypeptide (VIP) interacts with a high affinity to two subclasses of G protein coupled receptors named VPAC₁ and VPAC₂, and has a 3–10 fold preference for VPAC₁ over VPAC₂ receptors. Selective ligands for each receptor subclass were recently described. [R¹⁶]-PACAP (1–23) and [L²²]-VIP are two selective VPAC₁ agonists.

2. Chimaeric human VPAC₂-VPAC₁ recombinant receptors expressed in CHO cells were used to identify the receptor domains implicated in these two selective ligands recognition.

3. The VPAC₂ preference for [R¹⁶]-PACAP (1–27) over [R¹⁶]-PACAP (1–23) did not require the receptor's NH₂-terminus domain but involved the whole transmembrane domain.

4. In contrast, the selectivity of [L²²]-VIP depended only on the presence of the NH₂ terminus and EC₂ domains of the VPAC₁ receptor.

5. The present data support the idea that in the GPCR-B family of receptors the different selective ligands require different domains for their selectivity, and that the peptides carboxyl terminal sequence (amino acids 24–27) folds back on the transmembrane receptor domain, close to the peptides, aminoterminus.

Introduction

Vasoactive intestinal polypeptide (VIP) interacts with two receptor subtypes, previously named VIP₁ and VIP₂ and currently known as VPAC₁ and VPAC₂ (Harmar et al., 1998), that have been cloned in rat (Lutz et al., 1993; Ishihara et al., 1992), human (Couvineau et al., 1994; Svoboda et al., 1994) and mouse (Inagaki et al., 1994). These receptors belong to the class B of the G-protein coupled receptors that includes the calcitonin-, the corticotrophin releasing factor-, an insect diuretic hormone-, the gastric inhibitory peptide-, the glucagon-, the glucagon-like peptide I-, the growth hormone releasing factor-, the parathyroid hormone-, the pituitary adenylate cyclase activating polypeptide (PACAP)-, the secretin-, the VPAC₁- and VPAC₂- and also orphan receptors (Donnelly, 1997).

All these receptors have significant sequence homology (especially in the transmembrane domains) and possess a large extracellular NH₂-terminal domain with six conserved cysteine residues. They are all positively coupled to adenylate cyclase but some of them may also be coupled to other effectors (Van Rampelbergh et al., 1997; Spengler et al., 1993).

In situ hybridization reveals that both VIP receptor subtypes are transcribed – and thus probably expressed – in different tissues or cells (Usdin et al., 1994). This is particularly obvious in brain where their localization in many areas appear mutually exclusive (Vertongen et al., 1997). The development of agonists and antagonists selective for each receptor subclass is therefore of great pharmacological importance.

The neuropeptides VIP and PACAP recognize both receptor subtypes with a high affinity (Couvineau et al., 1996; Svoboda et al., 1994). The peptide with a NH₂-terminal histidine and a COOH-terminal isoleucine amide (PHI) or its longer form with a COOH-terminal valine amide (PHV) – that are co-synthesized and released with VIP – have a higher affinity for the human VPAC₂ than for the human VPAC₁ receptor (Couvineau et al., 1996; Gourlet et al., 1998), which is not the case for the rat receptors (Couvineau et al., 1996).

Several synthetic ligands, selective for each receptor subtype have been described recently (Gourlet et al., 1997a,1997b; 1998; Xia et al., 1997). Amongst them, [R¹⁶]-PACAP (1–23) (as example of truncated peptide) and [L²²]-VIP (as example of single change in the peptide sequence) are selective VPAC₁ receptor agonists (this study and Gourlet et al., 1998).

Little is known to date about the receptor determinants that are responsible for the different selectivity profiles observed for these related receptors.

In the present study, we constructed and stably expressed in Chinese hamster ovary (CHO) cells chimaeric receptors made by the replacement of parts of the human VPAC₁ receptor sequence by the corresponding sequence of the human VPAC₂ receptor to identify the receptor domains involved in this selectivity. The shortened original PACAP (1–27) derivative, [R¹⁶]-PACAP (1–23) and a mono-substituted VIP analogue [L²²]-VIP were used in this study as VPAC₁ selective ligands and were compared to [R¹⁶]-PACAP (1–27) and VIP respectively.

Methods

Wild-type receptors

The human VPAC₂ receptor cDNA was previously cloned and characterized in our laboratory (Svoboda et al., 1994). The human VPAC₁ receptor cDNA was cloned by PCR according to the sequence previously reported by Couvineau et al. (1994) using specific primers. Chimaeric receptor constructs The receptors studied in this work represented eight chimaeric constructs of wild type human VPAC₂ and VPAC₁ receptors, four of which (chimaeras C1, C4, C7 and C8) have been described previously (Juarranz et al., 1999a,1999b). Chimaeras were designed to replace portions of one wild type receptor cDNA with the corresponding portions of the other receptor. The following constructs were prepared (where the amino acids are numbered according to the human VPAC₁ and VPAC₂ receptor sequences): VPAC1, human wild-type VPAC₁ receptor (1–457); VPAC2, human wild-type VPAC₂ receptor (1–438); C1 (N-VPAC₂-VPAC₁), VPAC₂ receptor (1–127) and VPAC₁ receptor (144–457); C2 (N→IC₁ VPAC₂-VPAC₁), VPAC₂ receptor (1–158) and VPAC₁ receptor (175–457); C3 (N→TMD-2 VPAC₂-VPAC₁), VPAC₂ receptor (1–178) and VPAC₁ receptor (195–457); C4 (N→EC₁ VPAC₂-VPAC₁), VPAC₂ receptor (1–203) and VPAC₁ receptor (217–457); C5 (N→TMD-3 VPAC₂-VPAC₁), VPAC₂ receptor (1–227) and VPAC₁ receptor (241–457); C6 (N→TMD-4 VPAC₂-VPAC₁), VPAC₂ receptor (1–261) and VPAC₁ receptor (275–457); C7 (N→EC₂ VPAC₂-VPAC₁), VPAC₂ receptor (1–279) and VPAC₁ receptor (294–457); C8 (N→EC₃ VPAC₂-VPAC₁), VPAC₂ receptor (1–360) and VPAC₁ receptor (374–457).

The strategy was based on the use of PCR overlap extension. Briefly : (i) using human VPAC₂ and VPAC₁ receptor cDNA as templates and appropriate chimaeric primers, we first generated cDNA fragments overlapping at their 5′ or 3′ extremity; (ii) after purification of these fragments, using High Pure PCR Product Purification Kit (Boehringer, Mannheim), they were used in a round of PCR overlap extension. The use of a phosphorylated forward primer surrounding the ATG initiation codon produced a 5′ hemi-phosphorylated cDNA fragment. This particularity combined with the presence of a 3′-A overhang resulting from the terminal transferase activity of Taq polymerase, allowed the unidirectional cloning of the chimaeric receptors in pCR3.1-Uni (Invitrogen) that was suitable for both prokaryotic and eukaryotic expressions. All PCR reactions were performed using Expand Long Template system (Boehringer, Mannheim) in the Geneamp 2400 thermocycler (Perkin-Elmer). Successful construction of chimaeras was confirmed by nucleotide sequence determination. Receptor expression-stable transfection Recombinant plasmids were transfected into CHO cells by electroporation using a gene pulser (Electroporator II, Invitrogen, San Diego, CA, U.S.A.). Briefly, 4.10⁶ cells were preincubated on ice for 10 min with 20 μg of plasmid DNA in 0.25 ml of F12 nutrient mixture without serum (Gibco Life Technologies, Gent, Belgium). Electroporation was performed at 330 V and 1000 μF. After electroporation, cells were kept on ice for 10 min and then transferred into Petri dishes containing 10 ml of complete culture medium (Dulbecco-F12 supplemented with 10% foetal bovine serum, 2 mM L-glutamine, 100 μg ml⁻¹ penicillin and 100 μg ml⁻¹ streptomycin). After 48 h, cells were selected by addition of geneticin (400 μg ml⁻¹) for 2 weeks. Resistant cells were cloned and the final selection was made on the basis of their ability to express VIP stimulated adenylate cyclase activity. The CHO cells were routinely grown in Dulbecco-F12 medium enriched with 10% foetal bovine serum and geneticin was maintained in the stock cultures only. Peptide synthesis All the non-cyclic peptides were synthesized as carboxyl-terminal amides by solid phase methodology using the Fmoc (9-fluorenylmethoxy carbonyl) strategy on an Applied Biosystems Apparatus 431A (Foster City, CA, U.S.A.). The cleavage and the purification of the peptides have already been described (Robberecht et al., 1992). The purity of the material was at least 95% as judged by capillary electrophoresis and analytical reverse-phase chromatography and the peptide conformity was assessed by electrospray mass spectrometry and, when possible, by sequencing.

Membrane preparation and receptor identification

Stably transfected CHO cells were harvested with a rubber policeman and pelleted by low speed centrifugation; the supernatant was discarded and the cells lysed in 1 mM NaHCO₃ solution and immediately frozen in liquid nitrogen. After thawing, the lysate was first centrifuged at 4°C for 10 min at 400×g and the supernatant was further centrifuged at 20,000×g for 10 min. The pellet, resuspended in 1 mM NaHCO₃ was used immediately as a crude membrane fraction. Binding was performed as described (Ciccarelli et al., 1994; Vertongen et al., 1997). In all cases, non-specific binding was defined as the residual binding in the presence of 1 μM VIP. Binding was performed at 37°C in a 20 mM (pH 7.4) Tris-maleate buffer with 2 mM MgCl₂, 0.1 mg ml⁻¹ bacitracin and 1% bovine serum albumin. Bound radioactivity was separated from free by filtration through glass-fibre GF/C filters pre-soaked for 24 h in 0.01% polyethyleneimine and rinsed three times with a 20 mM (pH 7.4) sodium phosphate buffer containing 1% bovine serum albumin.

Tracer choice

We confirmed in the present work that ¹²⁵I-RO 25-1553 had a 3 fold higher affinity than ¹²⁵I-VIP for the VPAC₂ receptor (Vertongen et al., 1997), as well as for chimaeras C7 and C8. Consequently, the results obtained with the former tracer were technically more satisfactory (higher total over non-specific binding ratio). We therefore used [¹²⁵I]-VIP for characterization of the VPAC₁, C1 and C4 receptors, and [¹²⁵I]-RO 25-1553 (Ac¹-[Glu⁸, Lys¹², Nle¹⁷, Ala¹⁹, Asp²⁵, Leu⁶, Lys^27,28, Gly^29,30, Thr³¹]-NH₂(cyclo 21–25)) for characterization of the VPAC₂, C7 and C8 receptors. At the tracer concentrations used in these experiments (0.2–0.3 fold their K_d values), the unlabelled peptides IC₅₀ values are very close to their inhibition constants, K_i, and should therefore be independent of the nature of the tracer used. We verified that the unlabelled peptides IC₅₀ values were indeed independent of the tracer (¹²⁵I-RO 25-1553 or ¹²⁵I-VIP) chosen (not shown).

Specific tracer binding is proportional to the B_max over K_D ratio: it should theoretically be possible to compensate for low tracer affinities by increasing the membrane (and receptor) concentration. Unfortunately, non-specific ¹²⁵I-RO 25-1553 or ¹²⁵I-VIP binding also increased with increasing protein concentrations. We were therefore unable to obtain usable competition curves on CHO cell membranes expressing chimaeras C2, C3, C5 and C6, probably because their affinities for ¹²⁵I-VIP and ¹²⁵I-RO 25-1553 were too low.

Measurement of adenylate cyclase activity

Adenylate cyclase activity was determined according to the procedure of Salomon et al. (1974) on membranes from CHO cells stably transfected with the recombinant receptors. Membrane proteins (3–10 μg) were incubated in a total volume of 60 μl containing (mM): [α³²P]-ATP 0.5, GTP 10, MgCl₂ 5, EGTA 0.5, cyclic AMP 1, theophylline 1, phospho(enol)pyruvate 10, pyruvate kinase 30 μg ml⁻¹ and Tris-HCl 30 mM at a final pH of 7.8. The reaction was initiated by adding membranes and was terminated after a 15 min incubation at 37°C by adding 0.5 ml of a 0.5% sodium dodecyl-sulphate solution containing ATP 0.5 mM, cyclic AMP 0.5 mM and 20,000 c.p.m. [8-³H]-cyclic AMP. Cyclic AMP was separated from ATP by two successive chromatographies on Dowex 50 W×8 and neutral alumina.

Receptor density: absence of spare receptors?

The adenylate cyclase EC₅₀ value may underestimate the full agonists K_act if the receptor is expressed at a very high density (spare receptors). The maximal responses to VIP and PACAP (1–27) were comparable for all the constructs studied in this work, suggesting that, like wild type receptors, the chimaeric receptors could be expressed at high levels by the CHO cells. It is difficult to estimate correctly the density of VPAC receptors, because our tracers (labelled agonists) discriminate at least two receptor states in transfected CHO cells (Busto et al., 1999). We therefore used functional criteria to ensure that the results were not affected by the presence of spare receptors. Whenever binding studies were possible (VPAC₁, VPAC₂, C1, C4, C7 and C8 receptors), we verified that the IC₅₀ and EC₅₀ values were in reasonable agreement (EC₅₀⩾IC₅₀). RO 25-1553 behaves as a good partial agonist on clones expressing low human VPAC₁ receptor densities, and fully activates the adenylate cyclase at high VPAC₁ receptor expression levels (Gourlet et al., 1997b). The observation that it behaved as a partial agonist on chimaeras C2 to C6 therefore supported the hypothesis that the clones selected for this study did not express spare receptors.

Data analysis

All competition- and dose-effect curves were analysed by a non-linear regression program (Graph Pad Prism). They were compatible with recognition of a single binding site. IC₅₀ and EC₅₀ values were the ligand concentrations required for half maximal inhibition of tracer binding and half maximal adenylate cyclase activation, respectively. In order to calculate the peptide-receptor interaction free energy, we used the van't Hoff relationship: ΔG⁰=−RT ln(EC₅₀ or IC₅₀), where R is the gas constant and T, the absolute temperature. ΔΔG⁰ represents the difference between the control and modified peptide binding or activation free energies. Differences between mean IC₅₀, EC₅₀ values were tested by Student's t-test. P<0.05 was accepted as being significant. The IC₅₀ and EC₅₀ standard deviations were always below 0.1 log units (typically, 0.05–0.08 log units). We therefore assumed that the standard deviations of the ΔΔG⁰ values (corresponding to an IC₅₀ or EC₅₀ ratios s.d. <0.2 log unit) were below 1.17 kJ mol⁻¹ K⁻¹.

Results

VIP, PACAP and analogues were tested on recombinant human VPAC₁, VPAC₂, and chimaeric receptors for their ability to bind to the receptors and to activate adenylate cyclase in membranes from transfected CHO cells. The ligands and a schematic representation of the chimaeric receptors are shown in Table 1 and in the Figures, respectively.

Table 1.

Comparative amino acid sequences of the peptide ligands

VIP, PACAP (1–27) and [R¹⁶]-PACAP (1–27) are considered as non-selective reference ligands. As previously published, the VPAC₁ receptor had 5–10 fold lower IC₅₀ and EC₅₀ values for the three agonists than the VPAC₂ receptor. Surprisingly, four chimaeric receptors (C2, C3, C5 and C6) showed much higher EC₅₀ values for adenylate cyclase activation than the two wild-type receptors, and ¹²⁵I-VIP or ¹²⁵I-RO 25–1553 binding could not be valuably studied on these receptors (see Methods). If we consider only the chimaeric receptors on which binding studies were possible, the VPAC₁ receptor sequence 294–373 (TMD-5 to EC₃) seemed responsible for the higher affinity of the VPAC₁ receptor for these three agonists.

In order to identify the receptor region(s) that interact with the C-terminal peptide sequences, we compared a mono-substituted VIP analogue ([L²²]-VIP) with VIP, and the carboxyl-terminally truncated ligand ([R¹⁶]-PACAP (1–23)) with [R¹⁶]-PACAP (1–27).

[L²²]-VIP had a 10 fold lower affinity than VIP for VPAC₁ receptors, and a 120 fold lower affinity than VIP for VPAC₂ receptors. The effect of the [Y]→[L²²]-VIP substitution was 3– 4 fold greater on chimaeras C1 and C4 than on VPAC₁ receptor, and 10 to 20 fold greater on chimaeras C7 and C8 (Figure 1 and Table 2). Representing the differences between the VIP-receptor interaction free energy and the [L²²]-VIP-receptor interaction free energy (ΔΔG⁰) in the wild type and the different chimaeric receptors, we observed two ‘steps' in the binding affinity loss: one between VPAC₁ and C1 and the other between C6 and C7 (Figure 2, top). This suggested that the VIP tyrosine 22 recognized the VPAC₂ receptor N-terminal and EC₂ region. Comparison of the VIP and [L²²]-VIP EC₅₀ values (adenylate cyclase activation) further supported this hypothesis.

Figure 1.

Effect of VIP and [L²²]-VIP on tracer binding to membranes from CHO cells expressing the VPAC₁, VPAC₂ and VPAC₂-VPAC₁ chimaeric receptors C1, C4, C7 and C8. The values were the means of at least three experiments made in duplicate. The data are expressed as per cent of specific tracer binding in the absence of added ligand.

Table 2.

Interaction of the VIP and [L22]-VIP with the VPAC1, VPAC2 and VPAC2-VPAC1 chimaeric receptors (C1 to C8)

Figure 2.

Top panel: The differences between the free energy of binding or activation (±s.d.) of the different constructs by VIP and [L²²]-VIP were calculated from at least three experiments. The following binding ΔΔG⁰ differences were (≠) or were not (=) significantly different: (1): VPAC₁, ≠Cl=C4≠C7=C8=VPAC₂. Bottom panel: The differences between the free energy of binding or activation (±s.d.) of the different constructs by [R¹⁶]-PACAP (1–27) and [R¹⁶]-PACAP (1–23) were calculated from at least three experiments. The following binding ΔΔG⁰ differences were (≠) or were not (=) significantly different: VPAC₁=C1≠C4≠;C7=C8=VPAC₂.

The other selective ligand tested, [R¹⁶]-PACAP (1–23) had a similar (3 fold lower) affinity as the related non selective peptide [R¹⁶]-PACAP (1–27) for VPAC₁ and C1 receptors, but had a 25 fold lower affinity for C4, and was 100–150 fold less potent on chimaeras C7, C8 and VPAC₂ receptors (Table 3 and Figure 3 for binding data). It is possible to estimate the contribution of aminoacids 24–27 to [R¹⁶]-PACAP (1–27) recognition by calculating the difference between the binding free energy of [R¹⁶]-PACAP (1–27) and [R¹⁶]-PACAP (1–23) (ΔΔG°). As shown in Figure 2 (bottom), the receptor amino terminus did not contribute to the discrimination between [R¹⁶]-PACAP (1–23) and [R¹⁶]-PACAP (1–27) (VPAC₁ → C1). The chimaeric receptor's ability to discriminate the two peptides increased progressively, from C1 to C8; the last chimaera behaved like VPAC₂ receptors in this respect. Functional data (adenylate cyclase activation) on the eight chimaeric receptors did not allow us to pinpoint more accurately the receptor regions(s) involved in the C-terminal PACAP recognition.

Table 3.

Interaction of PACAP (1–27), [R16]-PACAP (1–27) and [R16]-PACAP (1–23) with the VPAC1, VPAC2 and VPAC2-VPAC1 chimaeric receptors (C1 to C10)

Figure 3.

Effect of [R¹⁶]-PACAP (1–27) and [R¹⁶]-PACAP (1–23) on tracer binding to membranes from CHO cells expressing the VPAC₁, VPAC₂ and VPAC₂-VPAC₁ chimaeric receptors C1, C4, C7 and C8. The values were the means of at least three experiments made in duplicate. The data are expressed as per cent of specific tracer binding in the absence of added ligand.

Discussion

It has been established, before the cloning of the receptors, that VIP acts through at least two functionally distinct receptors: the ‘classical' VIP receptor, exemplified by the receptor of the intestinal epithelial cells, now known as the VPAC₁ receptor and the helodermin-preferring VIP receptor, exemplified by the receptor of the lymphoblastic cell line SUP T₁ and now named VPAC₂ receptor.

The importance of the receptor NH₂-terminus for selective ligand recognition has already been shown for other related receptors: chimaeric constructions between the rat secretin and the rat VPAC₁ receptors, between the rat VPAC₁ and rat PACAP (PAC₁) receptors, and between human VPAC₁ and VPAC₂ receptors (Van Rampelbergh et al., 1996; Gourlet et al., 1996c,1996d; Vilardaga et al., 1996; Cao et al., 1995; Hashimoto et al., 1997; Holtmann et al., 1995; 1996a,1996b; Juarranz et al., 1999a,1999b) indicated that in all cases, the NH₂-terminal domain is a major contributor to the ligand selectivity.

The ‘Ala-Scan' of VIP performed by O'Donnell et al. (1994) focused our attention on the importance of amino acid [Y²²]. By substituting the tyrosine residue present in VIP by a leucine or an alanine residue (Gourlet et al., 1998), single mutated VIP analogues selective for the VPAC₁ receptors were obtained. The search for the shortest VIP or PACAP sequence that binds to the VIP and PACAP receptors led to the discovery that the carboxyl-terminally shortened peptides also have a preference for the VPAC₁ receptor (Gourlet et al., 1996b), and that PACAP (1–23) retained a relatively high affinity for the VPAC₁ receptor but not for the VPAC₂ receptor. Furthermore, the introduction in VIP and PACAP analogues of an arginine residue in position 16 increased their affinities for both VIP receptors (Gourlet et al., 1996a). [R¹⁶]-PACAP (1–23) was synthesized and behaved as a selective, high affinity VPAC₁ receptor ligand (this study).

The aim of the present work was to identify the receptor region(s) recognized by the C-terminal ligand sequence. To achieve this goal, we measured the ability of chimaeric receptors to discriminate two selective ligands, [L²²]-VIP and [R¹⁶]-PACAP (1–23) from the corresponding non-selective ligands VIP and [R¹⁶]-PACAP (1–27), respectively. Chimaeras C1, C4, C7 and C8 had a good affinity for VIP, PACAP (1–27) and [R¹⁶]-PACAP (1–27) in binding and functional studies. The chimaeric receptors C2, C3, C5 and C6, in contrast, had surprisingly low affinities for VIP and all the ligands tested: specific binding of the radioiodinated tracers was poor to undetectable, and the peptides potencies (EC₅₀⁻¹) for adenylate cyclase activation were very low (Tables 2 and 3).

The non selective peptides VIP, PACAP (1–27) and [R¹⁶]-PACAP (1–27) had a slightly weaker affinity and potency for VPAC₂ receptor as compared to VPAC₁ receptor. The binding determinant(s) involved appeared to belong to the receptor's transmembrane domain, between TMD-5 and EC₃, as chimaeras C1, C4 and C7 had binding and functional properties equivalent to VPAC₁, and chimaera C8 to the VPAC₂ receptor. We are unfortunately not able to be more precise: the intermediate chimaeras, C5 and C6, had a very weak affinity for the three ligands, suggesting that one of their common anchoring points was unavailable in these constructions. This might be due to incompatibilities between the VPAC₂ extracellular loops and VPAC₁ transmembrane domain: the extracellular loops EC₁ and EC₂, that are probably involved in two or three disulphide bridges (Vilardaga et al., 1997), have different lengths and sequences in the two receptors.

[L²²]-VIP, with a single (tyrosine to leucine) mutation in position 22, had an 11 fold lower affinity than VIP on VPAC₁ receptors and 120 fold lower affinity than VIP on VPAC₂ receptors. As shown in Table 2, the introduction of the N-terminal VPAC₂ receptor domain was sufficient to decrease the [L²²]-VIP affinity, as compared to VIP. This observation is in line with the results of Dong et al. (1999) indicating that the [L²²] of secretin is close to the receptor amino-terminal tail in the secretin-receptor complex. A further decrease of [L²²]-VIP affinity and potency was observed between chimaeras C4 and C7, the latter chimaeric receptor having an affinity ratio comparable to VPAC₂ receptors: Leu²² probably interacted, in addition, with a region found between EC₁ and TMD-5, that is, with TMD-3, TMD-4 or EC₂. We made the assumption that, as suggested above, chimaeras C5 and C6 had lost a ‘common contact' with all the peptides (contributing approximately 6.3 kJ mol⁻¹ to the peptide-receptor interaction). We calculated the [L²²]-VIP over VIP EC₅₀ ratios to estimate the additional contribution of Tyr²² to VIP recognition by the different chimaeras (Figure 2, top). Our results supported the hypothesis that Tyr²² was in contact with EC₂ (C6→C7), rather than with TMD-3 or TMD-4.

It is interesting to compare the present results with our previous work on the VIP₁ agonist (Table 1) (Juarranz et al., 1999a). The VPAC₁>VPAC₂ selectivity of the agonist peptide was due for the most part to preferential recognition of the VPAC₁ N-terminal and EC₁ domains, with a (smaller) contribution of a more distal receptor domain (probably EC₂: unpublished data). The Tyr→Leu replacement in position 22 explained the importance of EC₂, and part of the contribution of the N-terminal domain for the VIP₁-agonist selectivity. Further studies are needed to identify the additional positions that contribute to the VPAC₁>VPAC₂ selectivity of the VIP₁-agonist.

In contrast with [L²²]-VIP, [R¹⁶]-PACAP (1–23) had the same high affinities for the VPAC₁ receptor and chimaera C1. It therefore seems likely that the 3-dimensional structure of [R¹⁶]-PACAP (1–23) was conserved, allowing the correct positioning of Tyr²² with respect to the N-terminal receptor domain. The region involved in the [R¹⁶]-PACAP (1–23) selectivity (Table 3 and Figure 2, bottom) was the TMD-1 to EC₃ region. This result supported the hypothesis that the receptor core (TMD helices+EC loops) may participate to the recognition of the C-terminal sequence (amino acids 22–27 or 28) of the ligand.

The implication of both the NH₂-terminal domain and of TMD-1 to EC₂ in the selectivity of the VPAC₁-preferring peptides studied in this work suggests that the ligands might be ‘sandwiched' between the amino-terminal domain and the upper extracellular part of the core of the receptor. Alternatively, it is also possible that the extracellular loops play an important role in the folding of the N-terminal domain and in its positioning with respect to the peptide ligand.

The present results, taken together with published experimental data, allowed us to propose a general model of the peptide-receptor interaction, as follows: (1) N-terminal peptide binding. VIP, PACAP and secretin have identical sequences in positions 1–3, 6 and 7. We previously suggested that the aspartic acid in position 3 of secretin recognizes highly conserved amino acids in the secretin receptor TMD-2 (Vilardaga et al., 1996), corresponding to Lysine 173 and arginine 188 in the VPAC₁ receptor and to lysine 179 and arginine 172 in the VPAC₂ receptor. The hypothesis that the VIP amino terminal region (like secretin's) interacts with the transmembrane receptor domain is further supported by the observation that acylation of Histidine¹, that increases VIP's affinity for VPAC₂ receptor but not VPAC₁ receptors, did not increase the peptides' affinity for the VPAC₂-VPAC₁ chimaeric receptor (C1) but did increase its affinity for the ‘mirror image' chimaera (Juarranz et al., 1999a); (2) Aminoacids 8–15. We previously demonstrated, using a series of hybrid PACAP and secretin peptides and rat VPAC₁ and secretin chimaeric receptors that the positions 8 to 15 of secretin and PACAP interacted preferentially with the secretin and VPAC₁ receptors N-terminal domains, respectively (Gourlet et al., 1996b,1996c); (3) Aminoacids 22–27. Our present results suggest that the VIP carboxyl terminal region folds back towards the transmembrane region: Tyrosine 22 recognition was affected by the N-terminal domain and EC₂, and the C-terminal PACAP sequence (Ala²⁴, Ala²⁵, Val²⁶, Leu²⁷), necessary for high affinity recognition of the VPAC₂ transmembrane receptor domain (Gourlet et al., 1996b,1996d); (4) C-terminal extended peptides. The VPAC₂-selective peptide, RO 25-1553, differs from [acetyl-His¹]VIP in positions 9, 12, 19, 21 and 25–31. In contrast with [R¹⁶]-PACAP (1–23) recognition: the transmembrane receptor domain did not contribute to the selective RO 25-1553-VPAC₂ receptor interaction (Juarranz et al., 1999b). This suggests that Ala¹², Lys¹⁹ and the C-terminal extension of RO 25-1553, that markedly contribute to its selectivity interact with the extracellular N-terminal domain (in preparation).

Taken together, our results suggest that the different peptides that discriminate VPAC₁ from VPAC₂ receptors do so for different reasons, and can be used to visualise the relative receptor-ligand positioning. They are compatible with the hypothesis that the N-terminal peptide sequence forms a turn that is buried in the transmembrane domain; that amino acids 8–22 interact with the extracellular domain but fold back towards the extracellular loops, in order to allow amino acids 24–27 to interact with the transmembrane domain.

Abbreviations

CHO

Chinese hamster ovary

EC

extracellular loop

EC₅₀

concentration of agonist required for half maximal response

GPCR

G-protein coupled receptor

IC₅₀

concentration of ligand required for 50% inhibition of tracer binding

PACAP

pituitary adenylate cyclase activating polypeptide

TMD

transmembrane domain

VIP

vasoactive intestinal polypeptide