Receptor Subtype-specific Docking of Asp6.59 with C-terminal Arginine Residues in Y Receptor Ligands*

Y receptors (YRs) are G protein-coupled receptors whose Y1R, Y2R, and Y5R subtypes preferentially bind neuropeptide Y (NPY) and peptide YY, whereas mammalian Y4Rs show a higher affinity for pancreatic polypeptide (PP). Comparison of YR orthologs and paralogs revealed Asp6.59 to be fully conserved throughout all of the YRs reported so far. By replacing this conserved aspartic acid residue with alanine, asparagine, glutamate, and arginine, we now show that this residue plays a crucial role in binding and signal transduction of NPY/PP at all YRs. Sensitivity to distinct replacements is, however, receptor subtype-specific. Next, we performed a complementary mutagenesis approach to identify the contact site of the ligand. Surprisingly, this conserved residue interacts with two different ligand arginine residues by ionic interactions; although in Y2R and Y5R, Arg33 is the binding partner of Asp6.59, in Y1R and Y4R, Arg35 of human PP and NPY interacts with Asp6.59. Furthermore, Arg25 of PP and NPY is involved in ligand binding only at Y2R and Y5R. This suggests significant differences in the docking of YR ligands between Y1/4R and Y2/5R and provides new insights into the molecular binding mode of peptide agonists at GPCRs. Furthermore, the proposed model of a subtype-specific binding mode is in agreement with the evolution of YRs.

This makes them interesting target molecules for therapeutic interventions. Consequently, knowledge of the mechanism of peptide-receptor interaction is of great relevance, and efforts in designing subtype-selective agonists are central. The effects of NPY, PYY, and PP are transmitted by at least five rhodopsinlike GPCRs named Y 1 receptor (Y 1 R), Y 2 R, Y 4 R, Y 5 R, and y 6 R. The y 6 R shows a restricted genomic presence, indicated by the lowercase letter, being functionally expressed in rabbits and mice but is truncated in humans and absent in rat (9 -11). Y 1 R, Y 4 R, and y 6 R can be combined to a subfamily sharing ϳ50% overall amino acid identity with each other but only 27-31% with Y 2 R and Y 5 R (12,13). Despite the low sequence similarity between YR subtypes, all bind the same family of peptides.
All of the four most intensely investigated YR subtypes (Y 1 R, Y 2 R, Y 4 R, and Y 5 R) have their own signature profile of ligand recognition, indicating that the ligand-binding pocket displays subtype-specific differences. In general, Y 1 R, Y 2 R, and Y 5 R preferentially bind NPY and PYY, whereas mammalian Y 4 R show a higher affinity for PP (14 -18). Y 1 R and Y 4 R require the complete N terminus of the ligand (15,19,20). Y 5 R accepts peptides with the deletion of the first amino acid (21), and Y 2 R even binds significantly shorter peptides (NPY ) or centrally truncated analogs ([Ahx [5][6][7][8][9][10][11][12][13][14][15][16][17][18][19][20][21][22][23][24] ]NPY) with only minor effects on affinity (19,22). However, the C-terminal pentapeptide of all natural ligands was identified as an essential region for binding to all YRs (23,24), and these residues of the ligand C terminus seem to represent a core contact domain. Interestingly, studies of analogs containing conformational constraints, when bound at the various YR subtypes, support the concept of structural differences between subtypes in this domain (25)(26)(27)(28). However, the two conserved Arg residues at positions 33 and 35 are important contact sites for all YRs.
Although many efforts have been made to characterize the relevance of the individual amino acids of the ligand, little is known about residues within receptor involved in ligand recognition. For the hY 1 R only, all extracellularly located Glu and Asp residues have been mutated to Ala to identify possible interaction partners of the Arg residues of the C-terminal pentapeptide of NPY (29 -31). In this study, we have now identified one conserved residue Asp 6.59 (see "Experimental Procedures" for position numbering scheme) and proved its importance in all YRs by replacing it with Ala, Glu, Asn, and Arg. Interestingly, the fine mapping of Asp 6.59 showed differences between Y 1 R, Y 2 R, Y 5 R, and Y 4 R, suggesting a subtype-specific binding pocket. Then using a complementary mutagenesis approach to receptors and ligands, we identified the interaction partner of YR-Asp 6.59 in the peptide ligands. Surprisingly, a receptor subtype-specific interaction was observed, where Asp 6.59 of Y 1 R and Y 4 R interacts with Arg 35 of NPY/PP, and Asp 6.59 of Y 2 R and Y 5 R contacts Arg 33 of NPY, suggesting differences in the docking of YR ligands to Y 2/5 R and Y 1/4 R. Further receptor subtypespecific interactions have been identified for [Ala 25 ]NPY/hPP and position Asp/Glu 5.27 in the second extracellular loop (ECL2) of YRs. These data are of great importance not only for YRs. We suggest a binding mode that involves at least two different events and contact sides. Furthermore, we identified one position on top of transmembrane domain (TMD) 6 that might be a general activation residue in many peptide-binding GPCRs, probably representing the second step. Additionally, these results are relevant to the evolution of GPCRs, because the multi-ligand/multi-receptor system has evolved from an ancestor ligand/receptor pair. We speculate that the Arg 35 / Asp 6.59 interaction may have evolved in the primordial YR first, and the alternate binding of the Asp 6.59 arose later, after the gene duplication and evolution of Y 2 /Y 5 Rs.

EXPERIMENTAL PROCEDURES
Peptide Synthesis-The following peptides were synthesized by automated solid phase peptide synthesis using the Fmoctert-butyl (9-fluorenylmethoxycarbonyl-tert-butyl) strategy (23) 35 ]pNPY) were performed. CD analysis of Ala substituted peptides were performed (23) and showed no significant differences to wild type pNPY and hPP.
Position Numbering System in YRs-The residues are numbered relative to the most conserved residue contained in the helix, as explained previously (32). On the basis of this scheme, the most conserved residue in TMD 6 is proline, designated with the index number 6.50 and hence referred to throughout as Pro 6.50 . Asp, located nine residues C-terminal of Pro 6.50 , is referred to Asp 6.59 to compare conserved residues in different receptors.
Generation of Mutant YRs-The subcloning of the human Y 2 R and Y 5 R cDNA into the eukaryotic expression plasmid pEGFP-N1 (Clontech, Heidelberg, Germany) has been described previously (33). Mutations were introduced with the QuikChange TM site-directed mutagenesis method (Stratagene, CA). For receptor quantification in enzyme-linked immunosorbent assay (ELISA) studies (see below) an N-terminal nine-amino acid residue epitope (YPYDVPDYA) derived from the influenza virus hemagglutinin protein tag was inserted downstream of the start codon of all YRs. Enhanced green fluorescent protein was fused to the human Y 2 R and Y 5 R, and enhanced yellow fluorescent protein (Clontech, Heidelberg, Germany) was fused to the human Y 1 R and Y 4 R cDNA for fluorescence detection. As previously shown the function of YRs is not altered by C-terminal fusion to fluorescent proteins (33). The correctness of all of the constructs was confirmed by sequencing of the entire coding region.
Cell Culture, Transfection, and Functional Assays-Baby hamster kidney cells were cultured in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum. COS-7 cells (African green monkey kidney cells) were maintained in Dulbecco's modified Eagle's medium containing 10% fetal bovine serum, 100 units/ml penicillin, and 100 g/ml streptomycin. The cells were grown as monolayers at 37°C in a humidified atmosphere of 5% CO 2 (34,35) (for details see supplemental materials). Signal transduction was studied by co-transfection of COS-7 cells with YRs and chimeric G proteins (G␣ qi4 , G␣ ⌬6qi4myr kindly provided by E. Kostenis) and subsequent measuring of the signal of the phospholipase C␤-pathway (inositol phosphate (IP) accumulation) as described (36,37). It has been demonstrated that replacement of the four or five C-terminal amino acids of G␣ q with the corresponding G␣ i residues (referred to as G␣ qi4 ) confers the ability to stimulate the phospholipase C␤ pathway onto G i -coupled receptors (38,39). For control, transfection of chimeric G proteins without receptor and only transfection of the receptor and subsequent stimulation experiments were performed and did not show any increased IP accumulation up to 100 M of the ligand (for details and controls see supplemental Fig. S1).
Immunological and Immunofluorescence Studies-To ligand-independently estimate plasma membrane localization, a cell surface ELISA was performed with wild type and mutant receptors carrying an N-terminal hemagglutinin tag (40) (for details see supplemental materials). Immunofluorescence studies were carried out to determine the subcellular distribution of wild type and mutant receptors. COS-7 cells were transferred into 24-well plates that contained sterilized glass coverslips and transfected. Approximately 48 h later, the cells were fixed with formaldehyde. The cells were treated with an anti-hemagglutinin monoclonal antibody (10 g of 12CA5/ml) and then incubated with a tetramethylrhodamine isothiocyanate-linked antimouse IgG secondary antibody. Fluorescence images were obtained with a confocal laser-scanning microscope (Leica TCS SP2; Heidelberg, Germany).

Identification of Conserved Residues in the Third Extracellular
Loop-By comparing the amino acid sequences of ECL3 of human, rat, and mouse YRs ( Fig. 1) and further available sequences, the conservation of residue Asp 6.59 became obvious for YRs, whereas Asp/Glu is not found in any other peptidebinding GPCRs except for the receptors of the neuropeptide FF (NPFF) and pyroglutamylated Arg-Phe-amide peptide (QRFP) family. The general receptor numbering system (32) was used to compare residue positions between receptors. The first digit indicates the transmembrane domain, and the last two digits are selected such that the most conserved residue in each helix has the number 50, e.g. proline in TMD 6 is referred to Pro 6.50 (Fig. 1). It became obvious that in many peptide-binding GPCRs of class A, position 6.59 (Ϯ one position) is a conserved trifunctional amino acid that qualifies for ligand binding. This includes receptors for orexin (Lys 6.58 /Lys 6.59 ), bradykinin (Asp 6.58 ), cholestocytokinin, (Arg 6.58 /Ala 6.59 ), endothelin (Lys 6.58 /Lys 6.59 ), and angiotensin (Asp 6.58 /Val 6.59 ).
Asp 6.59 Is a Functional Key Residue in All YRs-To study whether this conserved aspartate plays an important role in ligand binding, Asp 6.59 was substituted with Ala in Y 1 R, Y 2 R, Y 4 R, and Y 5 R, and wild type and mutant receptors were transiently expressed and functionally tested. As shown in Table 1, mutation of Asp 6.59 to Ala resulted in a dramatic loss of affinity, and the K d values could no longer be determined because specific binding was only 2-4% of wild type binding. To further characterize the mutant receptors, IP accumulation assays with chimeric G proteins were performed. IP formation revealed a more than 600-fold reduction in pNPY potency for Y 2 R-D6.59A, a 12-fold reduction for Y 4 R-D6.59A, and a 99-fold reduction for Y 5 R-D6.59A compared with a 461-fold reduction for Y 1 R-D6.59A (Table  1). In contrast, mutagenesis of nearby Asp residues in Y 2 R and Y 5 R to yield Y 2 R-D 6.61 A and Y 5 R-D 6.62 A showed wild type properties in both binding and signal transduction FIGURE 1. Amino acid sequences of the C-terminal segment of TMD6 and ECL3 of representative mammalian Y 1 R, Y 2 R, Y 4 R, Y 5 R receptors and the human NPFF 1/2 and QRFP receptor. The conserved aspartate at position 6.59 and additionally investigated aspartates at positions 6.61 and 6.62 of the Y 2 R and Y 5 R are marked in bold type. The general receptor numbering system (32) was used to compare residue positions between receptors. The first digit indicates the transmembrane domain, and the last two digits are selected such that the most conserved residue in each helix has the number 50, e.g. proline in TMD6 is referred to Pro 6.50 .  ( Table 1). ELISA and confocal microscopy studies demonstrated that cell surface expression levels of these and the following Asp 6.59 mutants were not significantly different when compared with the wild type receptors (Table 1 and supplemental Fig. S2). This implies that Asp 6.59 is a functional key residue in all YRs.
Interactions of Asp 6.59 Are YR Subtype-specific-To test for steric limitations, Asp 6.59 of all YRs were mutated to Glu, whose side chain has one methylene group more than that of Asp but maintains the carboxyl group. As shown in Table 1, this mutation revealed wild type-like binding properties and activity in the case of Y 4 R and Y 5 R (Fig. 2). Unexpectedly, Y 1 R-D6.59E displayed no ligand binding, but functional analysis revealed E max and EC 50 values comparable with Y 1 R wild type. The loss of agonist binding with the Y 1 R-D6.59E mutant was confirmed in two different cell lines (baby hamster kidney and COS-7 cells), and transfection was studied with two different vectors to exclude experimental or cell line specificities. D6.59E mutation in Y 2 R was associated with a nearly unchanged binding affinity, but a 20-fold right-shifted curve (EC 50 ϭ 199 Ϯ 97 nM) in the functional assay compared with Y 2 R wild type (EC 50 ϭ 9.7 Ϯ 4.8 nM). The E max values for YR-D6.59E were all in the range of the corresponding YR wild types proving full agonistic activity of these mutants.
To test the importance of the negative charge at position 6.59 and to investigate the participation of YR-Asp 6.59 in polar interactions, this residue was substituted with Asn. The charge-neutralizing mutation YR-D6.59N had a significant effect on receptor function (Fig. 2 and Table 1) of all receptors. In agreement with binding data, functional characterization of D6.59N mutants revealed a 20-fold (Y 4 R) to 175-fold (Y 2 R) increase in EC 50 values compared with the corresponding wild type YRs. Furthermore, Y 2 R-D6.59N displayed a reduced efficacy, whereas the Asn mutants of Y 1 R, Y 4 R, and Y 5 R showed E max values in the range of the respective wild type receptor. Comparing EC 50 values of D6.59N and D6.59A, no differences were found for Y 4 R and Y 5 R. In contrast, Asn substitution in Y 1 R and Y 2 R resulted in lower EC 50 values than did D6.59A substitution in the same receptor subtypes.
To further confirm the postulated ionic ligand-receptor interaction of Asp 6.59 in all YRs, this residue was mutated to Arg. It is reasonable to assume that introduction of a positively charged residue at position Asp 6.59 should have a contrary effect (repulsion) on the positively charged counter residue in the peptide ligand. Indeed, D6.59R abolished high affinity binding and further decreased agonist potency for all YRs when compared with D6.59N and D6.59A (Fig. 2 35 ]pNPY resulted in a more than 100-fold increase in EC 50 values with Y 2 R expressing cells (Table 2). To dissect which of the two most important Arg residues (Arg 33 and Arg 35 ) in NPY participated in a postulated electrostatic interaction with Asp 6.59 , the two pNPY-Ala peptides were functionally tested on COS-7 cells transfected with Y 2 R-D6.59A. In such a complementary mutagenesis approach, one can assume that the decreased potency of a mutant receptor, in which an interacting residue was substituted, remains unchanged when tested with wild type and mutant peptide. In other words, in the mutant receptor, one partner of a peptide-receptor bond has already been disrupted, and the mutation of the counterpart in the peptide should have no further impact on this interaction. Thus, [Ala 33 ]pNPY and [Ala 35 ]pNPY were tested on Y 2 R-D6.59A. Interestingly, concentration-response curves to [Ala 35 ]pNPY were further right-shifted compared with those to pNPY for cells expressing the mutant receptor (Fig. 3). This indicates that Arg 35 in pNPY contacts residues in Y 2 R that are different from Asp 6.59 . In contrast, the potency of [Ala 33 ]pNPY remained unchanged at this mutant receptor, implying that Arg 33 of pNPY and Asp 6.59 in Y 2 R form an electrostatic bond.
Interaction of Asp 6.59 with Arg 33 and Arg 35 Depends on the YR Subtype-To test whether Arg 33 of NPY interacts with Asp 6.59 not only in Y 2 R, we extended the two-sided mutagenesis approach to all other human YRs. As seen with the Y 2 R, replacement of Lys 4 , Arg 19 , and His 26 in NPY and Arg 26 in PP by Ala changed neither affinity nor activity at all other YRs.
[Ala 25 ]NPY/PP had no effect on Y 1 R or Y 4 R, respectively, but the NPY analog led to an approximately 10-fold loss of activity in Y 5 R (  35 ]pNPY was further reduced at the mutant receptor ( Fig. 3 and Table 2). This indicates that Asp 6.59 in Y 5 R also forms a salt bridge with Arg 33 of pNPY. In contrast, no major changes in the potencies of [Ala 35 ]peptides were seen at Y 1 R-D6.59A and Y 4 R-D6.59A, whereas [Ala 33 ]peptides caused a further right shift in the concentration-response curves ( Fig. 3 and Table 2). These data indicate that in the case of Y 1 R and Y 4 R, Asp 6.59 interacts with Arg 35 of peptide ligands.
Further Evidence for Different Subtype-selective Ligand-binding Pockets in YR-To generate further evidence for a different ligand-binding pocket in Y 2/5 R versus Y 1/4 R, the analogs [Ala 25 ]NPY/PP were studied in more detail. Indeed, only at Y 2 R and Y 5 R was a 10-fold loss in potency found, whereas Y 1/4 Rs showed wild type properties to these agonists (Fig. 4A). Thus, we mutated all relevant acidic residues within all four receptors to Ala. We found that mutation at Asp/Glu 5.27 in Y 1 R and Y 4 R has no significant effect on agonist potency (Fig. 4B), whereas mutation of Glu 5.24 in Y 2 R and Glu 5.27 in Y 5 R significantly reduced agonist potency (Table 2). Only the latter receptors are sensitive to Arg 25 mutation in the peptide agonist. Therefore, it appeared that Arg 25 interacts with the acidic residues of the ECL2 in a subtype-specific manner. Next, we tested the peptide [Ala 25 ]NPY on the E5.24A and E5.27A mutants in Y 2 R and Y 5 R, respectively. We anticipated no further changes in potency and efficacy of the mutant peptide, if an ionic interaction between Arg 25 and Y 2 R-Glu 5.24 and Y 5 R-Glu 5.27 exists. However, [Ala 25 ]NPY showed a further loss of agonist potency at both mutants (Table 2), rejecting the hypothesis that Arg 25 of the peptide forms an ionic interaction with an acidic residue of ECL2 of Y 2 R and Y 5 R. Similar results were seen with the [Ala 33 ]NPY and [Ala 35 ] NPY peptide analogs at Y 2 R-Glu 5.24 Ala and Y 5 R-Glu 5.27 Ala. Therefore, it is more likely that either -cation interactions form the second contact, or more than one Glu/Asp residue is involved. Accordingly, the complementary mutagenesis approach revealed that none of the three Arg residues of the ligand acts as the interacting partner, so a Y 2/5 R-selective intramolecular salt bridge can be postulated. This further supports the concept of a different binding pocket and again is nicely in agreement with evolution in which Glu 5.27 of the Y 4 R is not conserved, and Val is present at this position in the rat Y 4 R.

DISCUSSION
Identification of receptor residues that are involved in ligand binding is a fundamental step in elucidating and understanding the principles of ligand-receptor interactions. The only known detailed structure of a GPCR bound to its ligand has been reported for rhodopsin (41), in which the characterized seventransmembrane receptor is in its inactive state. Significant structural rearrangements, however, are required to obtain active GPCRs. Much effort has been made for many small ligand GPCRs, but little is known so far for peptide GPCRs (43)(44)(45). In peptide GPCRs one of the most challenging problems is the identification of the binding mode of agonists (42).
To identify receptor residues that are important for agonist binding of the whole YR family, we applied the approach of screening for residues conserved within the receptor family by comparing all of the available sequences for conserved positions. With this approach, Asp 6.59 was found to be the only YR-specific acidic residue in the ECLs that is conserved throughout all subtypes and species for which receptor sequences have been reported so far. This is in agreement with earlier reports suggesting that Asp 6.59 of YR at the junction of TMD6 and ECL3 plays an important role in agonist (29,31) and antagonist (30) binding at Y 1 R. Interestingly, in other peptidebinding GPCRs, position 6.59, in combination with position 6.58, is also frequently a residue conserved within a family. Thus, receptors binding orexin contain a conserved Lys 6.58 / Lys 6.59 , cholestocytokinin receptors Arg 6.58 /Ala 6.59 , endothelin receptors Lys 6.58 /Lys 6.59 /Leu 6.59 , bradykinin receptors Asp 6.58 / Phe 6.59 /Thr 6.59 , angiotensin receptors Asp 6.58 /Val 6.59 , and many more receptors show conserved trifunctional residues at position 6.58 and/or 6.59 that qualify for ligand binding. A further indication of a very distinct interaction at this position can be concluded because nearby Asp residues in the Y 2 R and Y 5 R (Asp 6.61 and Asp 6.62 ) do not contribute to ligand binding. Posi- One scenario might be that a first interaction site differs between the Y 1 R/Y 4 R and the Y 2 R/Y 5 R subfamilies and subsequently also causes a difference in the relative orientation of the peptide toward Asp 6.59 . We identified Arg 25 as a candidate that shows high subtype-specific binding properties. Interestingly, NMR studies demonstrate that Arg 25 is located at the same face of the C-terminal ␣-helix as is Arg 33 (53), which qualifies this residue to serve as a first contact point in Y 2/5 R (Fig. 5). Then because of the inflexibility of Arg 33 within the ␣-helix, this important position can interact with Asp 6.59 (proximity model). This is further supported by the importance of the ␣-helix of the ligand for Y 2 R binding (26). A second reasonable model suggests that the first interaction site is identical in all YRs but also different from the interaction of Asp 6.59 . In this model, one or more subtype-specific residues would constrain the peptide and restrict Asp 6.59 to interaction with only one of the Arg residues (hindrance model). A hydrophobic face that has been discussed previously for Y 1 R might serve as a first and common binding site (54). However, hydrophobic interactions work only at much shorter range than do ionic interactions, making such scenario not very likely. In a third model, the ligand nonspecifically binds first to the membrane and then enters the receptor (55), probably between TMD5 and 6 (56). The overall size of the binding pocket will then determine how deep the ligand can insert into the receptor. In the case of the Y 2 R/Y 5 R subfamily, residues might block the entrance and just allow the formation of the activating contact of Asp 6.59 via Arg 33 , whereas in the case of the Y 1 R/Y 4 R receptor, different docking might occur.
One candidate for such a subtype-selective sensitivity position is Asp/Glu 5.27 in ECL2, which might form a salt bridge, preventing the ligand from getting any deeper into the binding pocket.
In any case, a receptor subtype-specific event must occur first, and then the ionic contact of Asp 6.59 can be formed in a second step. As a fourth possibility, Y 2 R/Y 5 R ligand binding can be postulated to affect G␣ q binding in a different and selective manner. Ongoing work now seeks to identify further subtypespecific interaction partners, which subsequently will help to further elucidate the binding mechanism.
Evolutionary and structural comparison studies suggest that Y 1 R and Y 4 R form a distinct subfamily of YRs, and Y 2 R and Y 5 R orthologs showed closer structural relation to each other than to members of the Y 1 R/Y 4 R subfamily (13). This is in good agreement with the identified differences in the binding mode of the ligand. Further comparison of GPCRs revealed that only receptors that bind the RF-amide peptides also contain Asp 6.59 . RF-amide peptides such as neuropeptide FF, AF, and QRFP share a common C-terminal Pro-Gln-Arg-Phe-NH 2 motif. It can be speculated that this Arg has the same function as Arg 35 of NPY and hPP and interacts with the conserved Asp 6.59 of the NPFF 1 , NPFF 2 , and QRFP receptor. According to this, the evolutionarily older binding mode would be the interaction between Arg 35 and Asp 6.59 , and later, a second mode of binding characterized by the Arg 33 /Asp 6.59 interaction evolved.