The Ligand Binding Site of the Neurokinin 2 Receptor SITE-DIRECTED MUTAGENESIS AND IDENTIFICATION OF NEUROKININ A BINDING RESIDUES IN THE HUMAN NEUROKININ 2 RECEPTOR*

Thirteen residues in the human neurokinin 2 (NK2) receptor were identified as potential ligand-binding residues by molecular modeling and amino acid se- quence analysis. Site-directed mutagenesis was used to alter these residues in order to ascertain their impor- tance in binding neurokinin A (NKA), the physiological peptide ligand for the NK2 receptor, and the non-pep- tide NK2 receptor selective antagonist SR48968. Four sites appear to be critical for NKA binding (Gln'", HidBs, IleZoz, and Gl$7s). The mutant receptors Gln'" 4 His, nezoz + Val, Glg7s + Pro, and - Thr maintain their affinity for SR48968, despite being unable to bind the peptide ligand. His'g8 + Ala and His'Oe + Leu no longer bind NKA or SR48968. We have also identified a residue (Leuzgz) which appears to play a minor role in the binding of substance P (SP) and neurokinin B (NKB) to the NK2 receptor. The mutant receptor Leum2 + Ser binds NKFS and SP with approximately a &fold greater affinity in comparison with the wild type receptor while the affinity of NKA remains unaffected. The results suggest that intramembranous residues, as well as residues which lie close to the extracellular side of transmembrane helices 3,5, and 6, form part of the NK2 receptor binding site. Binding of SP and NKB to the

Thirteen residues in the human neurokinin 2 (NK2) receptor were identified as potential ligand-binding residues by molecular modeling and amino acid sequence analysis. Site-directed mutagenesis was used to alter these residues in order to ascertain their importance in binding neurokinin A (NKA), the physiological peptide ligand for the NK2 receptor, and the non-peptide NK2 receptor selective antagonist SR48968. Four sites appear to be critical for NKA binding (Gln'", HidBs, IleZoz, and Gl$7s). The mutant receptors Gln'" 4 His, nezoz + Val, Glg7s + Pro, and -Thr maintain their affinity for SR48968, despite being unable to bind the peptide ligand. His'g8 + Ala and His' Oe + Leu no longer bind NKA or SR48968. We have also identified a residue (Leuzgz) which appears to play a minor role in the binding of substance P (SP) and neurokinin B (NKB) to the NK2 receptor. The mutant receptor Leum2 + Ser binds NKFS and SP with approximately a &fold greater affinity in comparison with the wild type receptor while the affinity of NKA remains unaffected. The results suggest that intramembranous residues, as well as residues which lie close to the extracellular side of transmembrane helices 3,5, and 6, form part of the NK2 receptor binding site. Binding of SP and NKB to the NK2 receptor may also be influenced by residues near the extracellular side of helix 7. These results suggest that some regions of the binding site for NKAin the NK2 receptor are not used for binding SP in the N K 1 receptor. However, it also seems that the NKA binding site includes regions that are also used by other G-protein-coupled receptors such as rhodopsin and the &-adrenergic receptors.
The neurokinins, substance P (SP),' neurokinin A (NKA), and neurokinin (NKB) are neuropeptides of 10-11 amino acid residues in length which share the concensus C-terminal sequence Phe-X-Gly-Leu-Met-NH,. They are widely distributed in both the central and peripheral nervous systems. These peptides each act upon three distinct membrane receptors, neurokinin 1 (NKl), neurokinin 2 (NK21, and neurokinin 3 (NK3), respectively, which belong to the superfamily of G-proteincoupled receptors (GPCRs) which possess seven transmembrane a-helices. However, despite the high degree of sequence identity between the three receptors (approximately 50-60%) and between the three ligands, each receptor displays a clear selectivity between the three peptides (1). The rank order of potency of the three neurokinin agonists for the NK2 receptor is NKA > NKB > SP; for the NK1 receptor, SP > NKA > NKB; and for the NK3 receptor, NKB > NKA > SP.
An understanding of the molecular basis of peptide-receptor interactions can contribute to the design of novel ligands with desirable pharmacological properties. Site-directed mutagenesis, coupled with innovative ligand design, has already been used to successfully identify the principle ligand binding residues in the P-adrenergic receptor, a member of a GPCR subfamily that bind a class of much smaller ligands (2, 3). The ligand binding site in this receptor subclass is believed to lie in the extracellular third of the seven transmembrane helical bundle. However, it is not clear whether such a binding site will be common to those receptors, such as the NK2 receptor, which bind much larger ligands. It seems that these larger peptide ligands might bind primarily to extramembranous regions of the receptor (4, 5). They also, presumably, form many more sites of receptor-ligand interaction compared to small ligands such as adrenaline.
In this paper, we have employed a three-dimensional model of a human NK2 receptor (based on Ref. 61, alongside an alignment of 10 neurokinin receptor sequences, in order to identify potential sites for NKA binding and selectivity. In order to reduce the likelihood of disrupting the structure of the receptor protein, these sites have been substituted either to residues with similar properties or else to residues found in the NK1 or NK3 receptors. Our results suggest the involvement of the transmembrane helices in NKA binding to the NK2 receptor.
[ ['251110dohistidyllNKA ([1251]NKA; specific activity = 1800-2200 Ci/ mmol) was purchased from Amersham International and 'H-SR48968 (specific activity = 25 Ci/mmol) was purchased from Du Pont NEN. pEF-BOS was the kind gift of Dr. S. Nagata (7). cDNA for the human ileum NK2 receptor was kindly provided in the cloning vector pUC119 by Glaxo. General reagents were purchased from Sigma. Cell culture reagents were purchased from Life Technologies, Inc. and Boehringer Mannheim.
Site-directed Mutagenesis of NK2 Receptor cDNA-A 1260-base pair BamHUNdeI restricted fragment of the NK2 receptor cDNA was introduced into the polylinker region of pEF-BOS using BstXI adaptors. Point mutations were created by the method reported by Kunkel (8) using single stranded DNA of the NK2 receptor cDNA in pEF-BOS. All mutations were confirmed using dideoxy chain termination DNA sequencing (9). At least two individual clones were selected for binding assessment.
COS Cell Expression-Wild type and mutated receptors were expressed transiently in COS-1 cells by Lipofectaminem lipid encapsulation of NK2 receptor cDNA in pEF-BOS. Transfections were performed using 12 pg/ml Lipofectamine and 1 pg/ml DNAin Opti-MEM for 5 h at 37 "C, 10% CO, and terminated by the addition of Dulbecco's modification of Eagle's medium containing 20% fetal calf serum. The following

Ligand Binding
Site of the Neurokinin 2 Receptor day, the transfected cells were plated at a density of 5 x 104-105 cells per well into 24-well plates and maintained for a further 2 days in Dulbecco's modified Eagle's medium, 10% fetal calf serum a t 37 "C and 10% CO,. Ligand Binding Assays-Binding assays were performed on intact COS-I cell monolayers, 72 h post-transfection. At this stage, the medium was removed and the cells were washed twice with phosphatebuffered saline (pH 7.4). For competition studies the cells were incubated for 90 min a t room temperature in binding buffer (Dulbecco's modified Eagle's medium, 20 m~ Hepes, 0.1% (w/v) bovine serum albumin, 400 pg/ml bacitracin) containing 0. To terminate binding, the ligandhuffer mixture was removed and the cells were washed manually three times with ice-cold phosphate-buffered saline, solubilized with 1% SDS, 0.1 M NaOH, neutralized with 0.1 M HCl, and collected for counting. For / 3 counting scintillation fluid was added. Computerized data analysis was carried out using GraphPAD inplot software.
Sequence Alignment-The amino acid sequences of 10 neurokinin receptors were extracted from the OWL protein sequence data base (11) and aligned using the program MALIGN (12). Competition studies were performed with 0.1 n~ [12511NKA using unlabeled NKA, NKB, and SP to determine whether the mutant receptors display differences in neurokinin selectivity compared to the wild type receptor. Fig. 2 shows competition curves for each of the three unlabeled neurokinin peptides for both the wild type receptor and the Leuzg2 + Ser mutant. While NKA binding was unaffected by this mutation, both NKI3 and SP had increased affinity for the mutant receptor compared to the wild type protein. However, the substitution Leuzg2 + Ile had no significant effect on the relative affinities of the three neurokinins for the receptor. Table I shows the IC,, values for the wild type and each of the  19 mutant NK2 receptors. Values for the Glnl" + His, His'g8 + Ala, His'" + Leu, IleZo2 4 Val, Gl$73 --j Pro, and Gly273 + Thr mutants were unobtainable since they no longer bound detectable levels of [lz5I1NKA at 0.1 I " . Indeed, no specific binding of [lz5I1NKA was detectable upto 2 nM (data not shown). However, Scatchard analysis shows that, with the exception of His1'' + Ala and Hidg8 "-f Leu, these mutants maintained their ability to bind the NK2 receptor selective non-peptidic antagonist 3H-SR48968 as illustrated in Fig. 3 (Kd and B , , values are shown in Table 11). Kd values for 3H-SR48968 binding are comparable to those obtained for the wild type receptor.  (Tables I and 11). Fig. 4 shows the alignment of the neurokinin receptor sequences. The asterisks (*) show the positions that are completely conserved. The N and C termini show the greatest degree of variability along with the central region of the third cytoplasmic loop and the second extracellular loop. The crosses (+) show the positions substituted in this work. The region of the second extracellular loop on the N-terminal side of helix 5 is particularly attractive as a target for mutagenesis since the residues are highly variable and an optimal sequence alignment is only possible by including gaps. It is a prime target for probing the receptor in search of residues or epitopes that might confer the specificity that each particular neurokinin receptor has for its ligand. However, the substitution of Glylgo and Glylgl for residues found at the equivalent positions in the NK1 and NK3 receptors had little effect on the binding affinities of the peptide agonists even though these positions were substituted to residues with very different properties (i.e. Asn and Lys). It may be that the entire structure of this loop is Ligand Binding Site of the Neurokinin 2 Receptor important rather than the side chains of one or two individual residues. In the NK1 receptor, several regions of the N terminus and the second and third extracellular loops have been shown to affect high affinity binding of the three neurokinins (13). The remaining residues targeted for site-directed mutagenesis are on the predicted internal face of the extracellular third of helices 3-7 (Fig. 5). Met117 lies deep in transmembrane helix 3. The substitution of this residue for Leu has no effect on receptor binding characteristics. Likewise, the residues LeulGn and C Y S '~~ (helix 4) and Leulg4 (helix 5) do not appear to be involved in NKA binding.

Binding affinities of wild type and mutant receptors for SR48968
Gin"' at the top of helix 3, IleZo2 on helix 5, and Gl$73 at the top of helix 6, all seem to play important roles in NKA-receptor interaction but not in the binding of the selective non-peptidic antagonist SR48968. The maintenance of SR48968 binding shows that the mutants are fully expressed and strongly suggests that the receptor structure is not significantly disrupted. This result is interesting since two of these residues are at equivalent positions to putative ligand binding site residues in other GPCR classes. Gin"' is in the equivalent position to the putative Schiff base counterion (Glu113) in rhodopsin which is thought to be responsible for the absorption characteristics of this GPCR (14, 15). Within the neurokinin group of receptors, the equivalent residue to Gin"' in the NK1 receptor (Hidos) has been shown to be involved in high affinity agonist binding (13). IleZo2 on the other hand is equivalent to Se904 in the &-adrenergic receptor which has been implicated in binding the catechol moiety of adrenaline (2). This suggests that the different GPCRs use a similar region as a ligand binding site despite the diversity of their ligands.
IleZo2 was targeted due to the significance of Se9O4 in the &-adrenergic receptor. However, the equivalent residue in the rat and mouse NK2 receptors is a phenylalanine. Hence it is surprising that the conservative substitution Ilezo2 + Val has such a dramatic effect on NKA binding without affecting  showing the location of the putative NKA binding residues (dot surfaces). SR48968 binding. This may indicate a possible point of a direct peptide receptor interaction. However, this seems unlikely since the residue is not conserved in the rodent NK2 receptors. Another possibility is that the mutation indirectly affects a nearby ligand-receptor interaction. In our three-dimensional model, IleZo2 contacts the conserved residue His267 on helix 6 which has been implicated in NK1 receptor ligand binding (16).
The substitution of Gly273 to proline or threonine, the equivalent residues in NK1 and NK3, respectively, abolishes NKA but not SR48968 binding indicating it as a potentially critical residue in NKA binding. The substitution of the small glycine residue by proline or threonine in the NK2 receptor may sterically hinder NKA binding. Alternatively, the more conformationally flexible glycine residue may be necessary for maintaining the correct local structure in this region of NK2. The mutation of the adjacent residue Sel-274, however, does not seem to affect NKA binding, even when substituted for tyrosine, a much larger residue. In our three-dimensional model Glf73 faces helix 7 whereas Sel.274 faces toward helix 5.
Analysis of the sequence alignment (Fig. 4) reveals that a 27-residue region that includes helix 7 (residues 285-311; human NK2 receptor numbering) is completely conserved in the neurokinin subfamily of GPCRs apart from two positions (292 and 293) which lie near the extracellular side of the helix. The substitution of Leuzg2 + Ser slightly affects the affinity of the NK2 receptor for NKB and SP despite not altering NKA or SR48968 binding. The substitution Leuzg2 + Ile, however, has no significant affect on agonist selectivity. This may imply that this residue is part of the NKl3 and SP binding site in the NK2 receptor. In the NKl receptor, substitution of the equivalent residue (IleZg0) has no effect on agonist selectivity (18).
It seems, therefore, that the neuropeptide binding site in the NK2 receptor involves residues on the extracellular third of helices 3, 5, 6, and 7 close t o the bilayer interface. This is in common with other GPCR classes (19). Fig. 6 shows where these residues lie in our current three-dimensional model which will be used to identify further sites that may be involved in ligand binding.