Functional Characterization of Structural Alterations in the Sequence of the Vasodilatory Peptide Maxadilan Yields a Pituitary Adenylate Cyclase-activating Peptide Type 1 Receptor-specific Antagonist*

Maxadilan is a vasodilatory peptide derived from sand flies that is an agonist at the pituitary adenylate cyclase-activating peptide (PACAP) type 1 receptor. Surprisingly, maxadilan does not share significant sequence homology with PACAP. To examine the relationship between structure and activity of maxadilan, several amino acid substitutions and deletions were made in the peptide. These peptides were examined in vitro for binding to crude membranes derived from rabbit brain, a tissue that expresses PACAP type 1 receptors; and induction of cAMP was determined in PC12 cells, a line that expresses these receptors. The peptides were examined in vivo for their ability to induce erythema in rabbit skin. Substitution of the individual cysteines at positions 1 and 5 or deletion of this ring structure had little effect on activity. Substitution of either cysteine at position 14 or 51 eliminated activity. Deletion of the 19 amino acids between positions 24 and 42 resulted in a peptide with binding, but no functional activity. The capacity of this deletion mutant to interact with COS cells transfected with the PACAP type 1 receptor revealed that this peptide was a specific antagonist to the PACAP type 1 receptor.

Pituitary adenylate cyclase-activating peptide (PACAP) 1 was first isolated from ovine hypothalamus because of its potent activity in stimulating cAMP production in rat anterior pituitary cells (1). Like other neuropeptides, PACAP possesses various physiological functions, including neurotransmission, vasodilatory actions, and endocrinological effects. Cloning of the PACAP receptors reveals three distinct subtypes (2)(3)(4). Two of the three receptors also function as receptors for VIP. The PACAP type 1 receptor is not considered a receptor for VIP as its affinity for this ligand is 1000 times less than its affinity for PACAP. Assessment of the physiological significance of PACAP has been hampered by the lack of a specific potent antagonist.
Maxadilan is a vasodilator peptide isolated from salivary gland extracts of the New World sand fly, Lutzomyia longipalpis, a vector of the protozoan disease leishmaniasis (5). The peptide aids the fly in blood feeding. It is produced as a 63amino acid peptide that may undergo C-terminal cleavage and amidation to a 61-amino acid peptide. It contains four cysteine residues that participate in the formation of two disulfide bonds between positions 1 and 5 and positions 14 and 51 ( Fig.  1) (6). Initial studies had suggested that maxadilan shared properties with CGRP, but binding of maxadilan to CGRP receptors could not be demonstrated. It has been demonstrated recently that maxadilan binds specifically to PACAP type 1 receptors. This result was surprising as maxadilan and PACAP do not share significant sequence homology (7).
In this structure-activity study, we examined the role of the two disulfide bonds in maxadilan by amino acid substitution of the individual cysteine residues. In other experiments, a series of deletion mutants was constructed. The activity of these substitution and deletion mutants was assessed by inhibition of 125 I-maxadilan binding to rabbit brain, crude membranes derived from rabbit brain induction of cAMP formation in PC12 cells, and in vivo by the induction of erythema in rabbit skin. The deletion mutant max.d.4 was further characterized in COS cells transiently transfected with the PACAP receptor cDNAs and found to be a potent PACAP type 1-specific antagonist. The far-ultraviolet CD spectra of maxadilan and max.d. 4 in the absence and presence of TFE suggested that these peptides fold into stable structures even in the water environment.

MATERIALS AND METHODS
Peptides-Recombinant maxadilan produced in Escherichia coli contained the four additional amino acids glycine, serine, isoleucine, and leucine at the N terminus as a result of construction in the pGEX vector designed for cleavage with thrombin. Termed GSIL-maxadilan, it was purified to homogeneity using reverse-phase high pressure liquid chromatography and prepared as described previously (8). 125 I-Maxadilan with a specific activity of 2000 Ci/mmol was prepared using the chloramine-T method (9). 125 I-PACAP-27 was purchased from NEN Life Science Products, and 125 I-VIP was purchased from Pharmacia Amersham Biotech. PACAP-27, PACAP-38, PACAP-(6 -27) and PACAP-(6 -38) were obtained from Peninsula Laboratories, Inc. (Belmont, CA).
Construction of Mutant Maxadilans-The maxadilan gene was inserted initially into the pGEX-2T plasmid and termed pGEX-max. The deletion mutant max.d.4, missing amino acids 24 -42, was prepared as follows. Because construction of the desired deletion mutant by PCR would necessitate the synthesis and use of lengthy primers, ClaI and PstI sites, which would not alter the amino acid sequence, were introduced into the maxadilan coding sequence. Beginning with pGEX-max * This work was supported by an agreement between Shiseido Co. Ltd. and the Massachusetts General Hospital/Cutaneous Biology Research Center, a grant from the Cutaneous Biology Research Center, and National Institutes of Health Grant RO1 AR42005 (to E. A. L.). The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "advertisement" in accordance with 18 (5), PCR was used to introduce a ClaI site using CCGG-ATCCTGTGATGCAACATGCCAATTTCGAAAG GCCAT CGATG as a 5Ј-primer and GGGAATTCACTTGCCGGCTTTAAATTCC as a 3Јprimer. These primers had BamHI and EcoRI sites at their 5Ј-ends to facilitate subcloning and were used to amplify the maxadilan coding sequence from pGEX-max. The purified 200-base pair product was ligated with pGEX-max predigested with BamHI and EcoRI and termed preparation A. The PstI site was introduced into preparation A using CCATCGATGACTGCCAGAAGCAGGCGCATCATAGCAATGTTCTG-CAG and GGGAATTCACTTGCCGGCTTTAAAA ATTCC as 5Ј-and 3Јprimers, respectively, in PCR. These latter primers were designed with ClaI and EcoRI linkers to facilitate subcloning. The PCR product was digested using ClaI and EcoRI. The gel-purified fragment was ligated to preparation A predigested with ClaI and EcoRI and termed preparation B. Deletion of amino acids 24 -42 was performed by cutting preparation B at the unique BstXI site and introducing PstI site. The blunt-ended product was self-ligated. This construct was designated max.d. 4.
To construct max.d.1, preparation A was digested with NspI; the resultant product was treated with a DNA blunting kit (Takara Shuzo Co., Kyoto, Japan) to blunt the 3Ј-cohesive end; and agarose electrophoresis was used to purify a 250-base pair fragment. This fragment was ligated to BamHI linkers (Takara Shuzo Co.), digested with BamHI and EcoRI, and separated by agarose gel electrophoresis; and an ϳ180base pair fragment was purified from the gel. This fragment was ligated with preparation A predigested with BamHI and EcoRI to construct max.d. 1.
Point mutations in the cysteine residues were introduced using PCR with oligonucleotides containing single codon changes, and their names, max.p.x, are indicated in Fig. 2. All mutants were analyzed by restriction mapping and sequencing of the mutated region before use. The mutants max.d.2 and max.d.3 were made by Peptide Institute Inc. (Osaka, Japan).
COS Cell Transfection-Plasmid preparations encoding the three PACAP receptor cDNAs were prepared and purified twice over cesium chloride gradients. One day before transfection, COS cells were plated in 6-well plates (3 ϫ 10 5 cells/well for binding assays) or 12-well plates (1.5 ϫ 10 5 cells/well for assay of cAMP). On the day of transfection, the medium was aspirated, and the cells were rinsed gently twice with 1 ml of 37°C phosphate-buffered saline without Ca 2ϩ or Mg 2ϩ . DNA solution (475 l of phosphate-buffered saline without Ca 2ϩ or Mg 2ϩ , 1 g of DNA, and 25 l of 10 mg/ml DEAE-dextran solution) was added to the well. Cells were incubated 37°C for 30 min with occasional swirling. Three milliliters of medium (10% Nuserum, penicillin/streptomycin, and glutamine in Dulbecco's modified Eagle's medium) containing 80 mol of chloroquine was added, and the cells were incubated for 3 h at 37°C. DNA/DEAE-dextran/chloroquine solution was aspirated, and the cells were treated with 1 ml of 10% dimethyl sulfoxide made with complete medium for 2.5 min. Fresh medium was added, and the cells were assayed 3 days later.
Preparation of Tissue Membranes-Rat and rabbit brain tissues were obtained from Pel-Freez Biologicals (Rogers, AR). Membrane fractions from tissues were prepared as described previously (9). Briefly, tissue was placed in 10 volumes of ice-cold 50 mM Tris-HCl buffer (pH 7.6) containing 0.32 M sucrose, 5 mM EDTA, 1 g/ml leupeptin, 1 g/ml pepstatin A, 2 g/ml bacitracin, and 10 g/ml phenylmethylsulfonyl fluoride. Tissue was homogenized with a Polytron PT 3000 (Brinkmann Instruments) for 30 s at power level 8 at 4°C, and the homogenate was centrifuged for 10 min at 1000 ϫ g at 4°C. The supernatant was removed, and the pellet was resuspended in 15 ml of the homogenizing buffer, homogenized again using the Polytron at the same setting as the first homogenization, and centrifuged at 1000 ϫ g for 10 min at 4°C. The combined supernatant was centrifuged at 30,000 ϫ g for 20 min at 4°C. The pellet was washed two times by successive suspension in 50 mM Tris-HCl buffer containing 1 mM MgCl 2 , 0.3% bovine serum albumin, 1 g/ml leupeptin, 1 g/ml pepstatin A, 2 g/ml bacitracin, and 10 g/ml phenylmethylsulfonyl fluoride.
Binding Study-Crude membranes (250 -400 g) were incubated for 2 h at 4°C in a final volume of 0.5 ml consisting of 50 mM Tris-HCl buffer (pH 7.6) containing 0.3% bovine serum albumin, 1 mM MgCl 2 , 1 g/ml leupeptin, 1 g/ml pepstatin A, 2 g/ml bacitracin, 10 g/ml For the binding assay using transfected COS cell, cells were incubated with 50 pM labeled peptide in the presence of a 1 M concentration of the indicated competitor ligand. At the end of incubation, samples were assayed for protein-bound radioactivity by vacuum filtration through Whatman GF/C glass microfiber filters pretreated with 0.5% polyethyleneimine. Filters were washed three times with 3 ml of buffer consisting of 50 mM Tris-HCl buffer (pH 7.5), 1 mM MgCl 2 , and 0.3% bovine serum albumin at 4°C. The radioactivity trapped on the filters was measured using a ␥-counter. Nonspecific binding represented between 10 and 20% of total binding. Binding is shown normalized to control; all experiments were performed in triplicate on at least three occasions; and the data are displayed with S.D. values.
Measurement of Cyclic AMP-PC12 cells or transfected COS cells grown to confluence in 12-well plates were stimulated for 10 min with the indicated concentrations of ligand in Hanks' medium with 1 mM 3-isobutyl-1-methylxanthine and 0.1% bovine serum albumin. The incubation was terminated by aspiration of the medium and addition of 400 l of ice-cold 50 mM Tris and 4 mM EDTA (pH 7.5). The cells were harvested, transferred to 1.5-ml Eppendorf tubes, boiled for 5 min, and spun at top speed in an Eppendorf centrifuge for 5 min. Supernatants were collected, and 50-l aliquots were assayed using a kit for determination of cyclic AMP following the instructions of the manufacturer (Amersham Pharmacia Biotech). All experiments were performed in triplicate on at least three occasions, and the data are displayed with S.D. values.
Cell Culture-PC12 cells were kindly provided by Dr. Barry Kosofsky (Massachusetts General Hospital). The cells were cultured in Dulbecco's modified Eagle's medium containing high glucose supplemented with 5% fetal bovine serum, 10% horse serum, penicillin, and streptomycin. The cells were propagated in a humidified 37°C incubator with 5% CO 2 .
Erythema Formation Assay-Erythema formation activity on rabbit skin was determined as described previously (5). Briefly, the indicated quantities of maxadilan and associated mutants were injected intracutaneously in a volume of 50 l of phosphate-buffered saline into the shaved back of a rabbit. Erythema formed on the skin of the rabbits was observed 30 min and 2 h after the injection. The smallest quantity of peptide that produced erythema detectable by eye is indicated.
CD Spectrum Measurement-Far-UV circular dichroism spectra were recorded on a Jasco J-500 spectropolarimeter. A water bath was used to maintain the temperature at 25°C. Ellipticity is reported as the mean residue molar ellipticity ([q]) with units of degrees cm 2 /dmol and was calculated from the following equation: [q] ϭ ([q] obs ϫ MRW)/(10 ϫ l ϫ C), where [q] obs is the observed ellipticity in millidegrees, MRW is the mean residue molecular weight of the polypeptide (molecular weight divided by the number of amino acids), C is the concentration of the sample in mg/ml, and l is the optical path length of the cell in cm. All spectra were measured at 250 -190 nm using a 1-cm CD cell.
Chemicals-All other reagents were of analytical grade from Sigma.

Characterization of Cysteine Substitution
Mutants-Cysteine substitution mutants were assayed for erythema formation, inhibition of 125 I-maxadilan binding to rabbit brain crude membrane, and accumulation of cAMP in PC12 cells. Monosubstitution of cysteine 1 (max.p.1) or 5 (max.p.2) or disubstitution of cysteine 1 and 5 (max.p.3) resulted in conservation of activity as compared with wild-type maxadilan. Monosubstitution of cysteine 14 (max.p.4) or 51 (max.p.5) or disubstitution of 14 and 51 (max.p.6) or all cysteines (max.p.7) abolished activity, indicating that the second disulfide bond between cysteines 14 and 51 is needed to maintain the conformation necessary for maxadilan to interact with its receptor (Fig. 2).
Characterization of Deletion Mutants-Four deletion mutants (max.d.1, which lacks the N-terminal ring structure; max.d.2, which consists of only the N-terminal ring structure; max.d.3, which consists of the N-terminal ring structure and the second ␣-helix structure; and max.d.4, which lacks 19 amino acids in the region between two ␣-helices) were examined for their ability to induce erythema formation, to inhibit 125 I-maxadilan binding to rabbit brain crude membrane, and to induce cAMP in PC12 cells. max.d.2 and max.d.3 did not inhibit 125 I-maxadilan binding to rabbit brain crude membrane, indicating that these two mutants could not hold the proper structure to bind to PACAP receptors. max.d.1 acted similarly to the wild-type peptide. max.d.4, which lacks 19 amino acids between positions 24 and 42, bound to the receptor, but failed to stimulate cAMP production and erythema formation (Fig. 3).
Binding of max.d. 4 to PACAP Receptor Subtypes-Binding of 125 I-maxadilan to cells expressing PACAP type 1 receptors could be inhibited by max.d.4 and the PACAP antagonist PACAP-(6 -38) (Fig. 4A). The binding of 125 I-VIP to cells expressing VIP type 1 receptors was not affected by either max.d.4 or PACAP-(6 -38) (Fig. 4A). The binding of 125 I-VIP to FIG. 3. Structure, activity, and binding of deletion mutants of maxadilan. Erythema formation, inhibition of 125 I-maxadilan binding to rabbit brain, and cAMP production in PC12 cells by the individual peptides are indicated. WT, wild type. cells expressing VIP type 2 receptors was affected by PACAP-(6 -38), but not by max.d.4 (Fig. 4A). max.d.4 competed with the binding of 125 I-PACAP to cells expressing PACAP type 1 receptors, but not those expressing VIP type 1 or 2 receptors (Fig. 4B). The binding of max.d.4 to PACAP type 1 receptors was reversed by increasing concentrations of agonist (data not shown).
max.d. 4 Inhibits cAMP Accumulation in a Receptor Subtypespecific Fashion-max.d.4 inhibited PACAP-38-induced accumulation of cAMP only in COS cells expressing PACAP type 1 receptors, whereas PACAP-(6 -38) inhibited accumulation of this second messenger in COS cells transfected with both PACAP type 1 and VIP type 2 receptor cDNAs (Fig. 6A). The induction of cAMP in VIP type 1 receptor cDNA-transfected COS cells was not inhibited by either max.d.4 or PACAP-(6 -38). These results are consistent with the binding results of transfected COS cells. The concentration-dependent inhibition of PACAP-38-induced production of cAMP in COS cells expressing PACAP type 1 receptors once again indicated that max.d.4 was more potent than PACAP-(6 -38) and PACAP-(6 -27) (Fig. 6B).
CD Spectra-The double minima at 220 and 207 nm of CD spectra of maxadilan and max.d.4 were observed in both aqueous solution and 25% TFE. The ratios of molar ellipticity ([] 220 / [] 207 ) of maxadilan and max.d.4 in 25% TFE were 0.93 and 0.99, respectively (Fig. 7).

DISCUSSION
Maxadilan was initially compared with CGRP because both peptides are potent vasodilators producing long-lasting erythema when injected into rabbit or human skin (10). The N terminus of maxadilan had sequence similarity to CGRP, and structural prediction indicated overall similarity between the two peptides. Both peptides possess the first ring structure formed by disulfide bonding, followed by an ␣-helix. Another ␣-helix structure is predicted at the C terminus. In the case of CGRP, deletion of the disulfide ring structure resulted in an antagonist (11,12). This result suggested that the disulfide ring might be a transduction domain or might be important in the maintenance of the proper structure for signal transduction. However, the result of cysteine substitution mutants of maxadilan indicated that the first ring structure can be deleted without changing the binding and transduction capacity of the peptide (Fig. 2). An analogous result was obtained with the mutant max.d.1, in which the disulfide ring structure was deleted with maintenance of erythema induction, inhibition of 125 I-maxadilan binding to rabbit brain crude membranes, and induction of cAMP in PC12 cells (Fig. 3). These results are consistent with the observation that maxadilan and CGRP bind to different receptors.
Sequence alignment of the maxadilan and PACAP family of peptides is displayed in Fig. 8. All PACAP family peptides possess FT residues at positions 6 and 7, although the significance of these residues in terms of binding specificity is not obvious because these peptides bind to their own receptors with high affinity. Maxadilan also possesses FT residues, at positions 34 and 35. In Fig. 8, the sequences for maxadilan and PACAP peptides are arbitrarily arranged such that the shared FT residues are aligned. With this alignment, the C-terminal lysines and valine are also aligned. PACAP-27 exhibits 68% similarity to VIP. PACAP is at least 1000-fold more potent than VIP in terms of binding to PACAP type 1 receptors. To identify which part of the PACAP molecule is responsible for the recognition of PACAP type 1-binding sites, PACAP and VIP sequences have been compared. It has been reported that replacement of Asp 9 and Leu 13 in VIP by serine and tyrosine, respectively, increased, by two to five times, the affinity of VIP for 125 I-PACAP-27-binding sites (13). A simultaneous change at both positions increased by 10-fold the affinity of the peptide, but [Ser 9 ,Tyr 13 ]VIP still had 40 -200 times less affinity than PACAP (14). It has been known in multiple substituted analogues of PACAP by VIP demonstrated that Gly 4 , Ile 5 , and Ser 9 , in combination, play an important role in the recognition of the PACAP type 1 receptor (15). In the aligned sequences, maxadilan does not have Gly, Ile, Ser, or Tyr at the corresponding residues. N-terminal histidines are present in all peptides of the PACAP/VIP family except growth hormone release factor. The absence of histidine in position 1 decreased the affinity for both 125 I-VIP-binding sites and 125 I-PACAP-binding sites (13,15). Maxadilan does not possess histidine in the corresponding site. Despite these differences, maxadilan still binds to PACAP type 1 receptors with high affinity, suggesting that not only individual amino acids, but also the entire structure is needed for optimal receptor binding.
The deletion mutation analysis indicated that max.d.4 was a possible PACAP receptor antagonist (Fig. 3). To examine in more detail the interaction of max.d.4 with PACAP/VIP receptors, the three known receptors were expressed transiently in COS cells. max.d.4 competed with the binding of 125 I-maxadilan and 125 I-PACAP to cells expressing PACAP type 1 receptors, but not those expressing VIP type 1 or 2 receptors (Fig. 4,  A and B). These results indicate that max.d.4 binds specifically to the PACAP type 1 receptor.
Both PACAP and maxadilan have been shown previously to induce the accumulation of cAMP (17). The ability of max.d.4 to inhibit cAMP accumulation induced by 1 nM PACAP-38 was examined in COS cells transfected with the individual receptor clones. max.d.4 inhibited the accumulation of cAMP only in COS cells expressing PACAP type 1 receptors (Fig. 6A). The inhibition of cAMP accumulation by max.d.4 in COS cells expressing PACAP type 1 receptors was concentration-dependent (Fig. 6B). Approximately 300 nM max.d.4 was necessary to inhibit production of cAMP elicited by 1 nM PACAP-38 by 50%, whereas the obtained affinity (IC 50 ) of max.d.4 for the PACAP type 1 receptor was ϳ6 nM (Table I). This discrepancy in sensitivity may be due to the different affinity between PACAP-27 and PACAP-38 and the different affinity between the labeled and unlabeled ligands.
max.d.4 is a PACAP type 1-specific antagonist, indicating that the two ␣-helix structures are important in differentiating PACAP type 1 receptors from VIP type 1 and 2 receptors. Competitive binding data and inhibition of cAMP production induced by PACAP-38 indicated that max.d.4 is more potent than previously known PACAP antagonists (Table I). The deletion mutant max.d.3, which lacks the first ␣-helix structure and the following sequence, did not bind to PACAP type 1 receptors, indicating that proper folding of the two ␣-helical domains is needed for receptor recognition. This result is compatible with the finding that the second disulfide bond is important for ligand-receptor binding (Fig. 2). The importance of the ␣-helix structure has also been shown for PACAP. Our results, as well as those of others, reveal that receptor binding affinity was higher with PACAP-38 than with PACAP-27. It has also been reported that the helix-breaking Gly 20 -Gly 21 substitution in PACAP-27 reduced the affinity of the resulting peptide 500-fold, indicating that the two highly hydrophilic residues (Lys 20 and Lys 21 ), located in the central portion of a hydrophobic domain in PACAP-27, were important for PACAP receptor recognition (13,15).
It is known that with the addition of TFE to aqueous solution, PACAP-27 and/or PACAP-38 acquires characteristics of increased ␣-helix structure (16). In the case of maxadilan, the double minima at 220 and 207 nm of CD spectra were observed not only in 25% TFE, but also in aqueous solution (Fig. 7). The CD spectrum for max.d.4 was similar to that for maxadilan, suggesting that the deletion from Leu 24 to Leu 42 did not induce a dramatic change in secondary structure. It is known that the ratio of molar ellipticity ( In conclusion, deletion of amino acids 24 -42 of maxadilan created a PACAP type 1 receptor-specific antagonist, max.d. 4. This mutant antagonized both PACAP-and maxadilan-induced cAMP production in PACAP type 1 receptor-transfected cells and is more potent than the PACAP antagonists PACAP-(6 -27) and PACAP-(6 -38). As maxadilan does not bind to 125 I-VIP-binding sites, which are otherwise virtually indistinguishable for VIP and PACAP, further study of the structureactivity relationships between maxadilan, PACAP, and VIP may provide important information about ligand-receptor interactions. The use of max.d.4 as a PACAP type 1 receptorspecific antagonist along with maxadilan as a PACAP type 1 receptor-specific agonist should be helpful in defining further the function of this receptor and its importance in vascular, endocrine, and neurological phenomena.