To identify specific interactions between either the tetrazole or carboxylate pharmacophores of non-peptide antagonists and the rat AT1 receptor, 6 basic residues were examined by site-directed mutagenesis. Three of the mutants (H183Q, H256Q, and H272Q) appeared to be like wild type. Lys and Arg mutants displayed reduced binding of the non-peptide antagonist losartan. Examination of their properties employing group-specific angiotensin II analogues indicated that their effects on binding were indirect. Interestingly, the affinity of losartan was not altered by a K199Q mutation, but the same mutation reduced the affinity of angiotensin II, the antagonist [Sar1,Ile8]angiotensin II, and several carboxylate analogues of losartan. An Ala substitution reduced the affinity of peptide analogues to a larger extent as compared to the affinity of losartan. Thus, the crucial acidic pharmacophores of angiotensin and losartan appear to occupy the same space within the receptor pocket, but the protonated amino group of Lys is not essential for binding the tetrazole anion. The binding of the tetrazole moiety with the AT1 receptor involves multiple contacts with residues such as Lys and His that constitute the same subsite of the ligand binding pocket. However, this interaction does not involve a conventional salt bridge, but rather an unusual lysine-aromatic interaction.
INTRODUCTION
The octapeptide hormone Ang II plays a central role in the regulation of blood pressure(
1
, 2
, 3
). Two distinct angiotensin receptor subtypes, AT1 and AT2, have been identified(2
). Because the AT1 receptor is solely responsible for mediating response of cardiovascular system to Ang II it is a major target for drug design efforts for the treatment of hypertension, congestive heart failure, and cardiac hypertrophy (1
, 2
, 3
). The AT1 receptor is a G-protein-coupled receptor characterized by a putative seven-transmembrane helical structure motif (Fig. 1A; (4
). Besides Ang II-peptide analogues, a family of non-peptide antagonists binds to the AT1 receptor with high affinity (Fig. 1B). These non-peptides presumably function like the classical G-protein-coupled receptor antagonists that utilize residues located in the putative transmembrane helices for their binding. Because Ang II is substantially larger, its binding is likely to involve the extracellular loops of the AT1 receptor in addition. Lys has been suggested to be involved in binding of the C-terminal carboxylate group of Ang II-peptides but has not been implicated in binding the non-peptide ligands(5
). Furthermore, residues located in transmembrane helices 3-7 involved in binding non-peptide antagonists do not influence peptide binding(5
, 6
, 7
, 8
, 9
). Because pharmacological competition between non-peptides and peptides is clearly established, an overlapping but non-identical binding site model is currently favored(3
, 5
, 6
, 7
, 8
, 9
).The non-peptide AT1 receptor-antagonists available today are rational improvements of the imidazole carboxylic acid lead compounds, S8307 and S8308(
3
, 10
). They contain the core structure of S8307 with a biphenyl-acidic extension, which is thought to be a better mimic of the Ang II structure(3
). Modeling of the three-dimensional structural overlay of several non-peptides indicates that a common geometry of critical pharmacophores is present in all high affinity angiotensin receptor antagonists(11
). Peptide antagonists have higher affinity toward the AT1 receptor, suggesting that they bind in a unique conformation to the receptor(2
, 11
). Therefore it is possible that the same pharmacophore geometry exists in the receptor bound conformation of Ang II-peptide antagonists as well (Fig. 1B). Presumably the same residues on the receptor are utilized for docking the pharmacophore presented by peptide and non-peptide antagonists. For example, the α-carboxyl group of the Ang II-peptides and the carboxylic acid or carboxamide and sulfonamide, or tetrazole group on the biphenyl portion of various non-peptides, are equivalent in structure-activity relationship(3
, 11
, 12
, 13
). In this report we present evidence to suggest that the peptide and non-peptide antagonists indeed occupy the same AT1 receptor binding pocket, but may employ dissimilar interactions for stabilization of bound ligands. The relevance of these results to the pharmacophore overlay of peptide and non-peptide AT1 receptor ligands is discussed.MATERIALS AND METHODS
A rat AT1 receptor gene with 41 unique restriction sites was designed by strategies used previously(
14
), and synthesized and cloned into the shuttle expression vector pMT2(). The synthetic gene encodes 359 amino acids of the rat vascular AT1 receptor with an 8-residue epitope tag (ETSQVAPA) at the C terminus. The epitope is the binding site for the monoclonal antibody 1D4, which can be used to detect polypeptide expression in transfected cells as described previously(16
). Total membrane preparations from transfected COS cells were frozen at −85°C in 50 mM HEPES, 12.5 mM MgCl2, 1.5 mM EGTA, and 10% glycerol until assayed(16
). The K and Bmax of the receptor were estimated by I-[Sar1,Ile8]Ang II equilibrium binding and Scatchard plot analysis. Extensive analysis of affinity, expression levels and agonist-induced inositol phosphate formation suggests that the AT1 receptors expressed from cDNA and the epitope-tagged synthetic DNA exhibit identical properties. Mutations were constructed in the synthetic gene by the technique of restriction fragment replacement, and all of the mutants were confirmed by DNA sequence analysis(16
, 17
).AT1 receptor binding was determined using total membrane prepared from transfected COS cells as described earlier(
18
). [Sar1,Ile8]Ang II and [Sar1,Ile8]Ang II-amide were radioiodinated by the lactoperoxidase method, and the radiolabeled peptides were purified by the reverse-phase high performance liquid chromatography method(19
). The specific activity of both these peptides was 2200 Ci/mmol. [3H]Losartan (specific activity 42.3 Ci/mmol) was obtained from Amersham. All binding data were analyzed and IC values determined by nonlinear regression analysis. K values for radiolabeled [Sar1,Ile8]Ang II were estimated from competition binding data with 8-10 different concentrations of the corresponding unlabeled [Sar1,Ile8]Ang II using the equation: K = IC/(1 - L/K), where L is concentration of radioligand, and IC is the concentration of competing ligand required to reduce specific radioligand binding by 50%, and K is the dissociation constant for I-[Sar1,Ile8]Ang II(20
). K values (nanomolar) represent mean ± S.E., n = 3-10.RESULTS AND DISCUSSION
Multiple interactions of Ang II with the receptor contribute to its binding enthalpy. Among these, the ion pair interaction of the C-terminal carboxyl group of Ang II with the receptor is well defined. The contractile response to [acetyl-Asn1,Val5]Ang II, an agonist analogue of Ang II that contains a single carboxylate at the peptide C terminus, is pH-sensitive, and the modification of its carboxylate to hydrogen bonding or hydrophobic groups reduces biological activity(
1
, 21
). Structure-activity analysis and modeling studies of non-peptide antagonists predict that a similar ion pair interaction is essential for them to bind with high affinity to the AT1 receptor(3
, 11
, 13
). We decided to define this interaction by systematic mutagenesis of receptor combined with group-specific modification of the ligand, since an earlier study (5
) did not investigate all potential residues in the receptor that could be involved in this interaction. We restricted our search to basic residues (Arg, Lys, and His in its protonated state) located in the putative transmembrane helices and the extracellular loops of the AT1 receptor because the majority of small molecule ligands bind to G-protein-coupled receptors within this region(22
). Basic residues conserved in all losartan-selective AT1 receptors were selected for mutagenesis (Fig. 1A). Several different single residue substitutions at each position were tested.Parameters Defining the Salt Bridge Interaction with the Ligands
Binding of the radiolabeled antagonist I-[Sar1,Ile8]Ang II by the expressed AT1 receptor is potently inhibited by I-[Sar1,Ile8]Ang II > Ang II > [Sar1,Ile8]Ang II-amide > losartan > Ang II-amide, as shown in Table 1. In addition, the binding of [Sar1,Ile8]Ang II is pH-sensitive (Fig. 2). The influence of pH on [Sar1,Ile8]Ang II, [Sar1,Ile8]Ang II-amide, and losartan binding shown in Fig. 2 is reversible. In addition, Marshall et al.(
2
) have reviewed the body of evidence suggesting that the conformation of Ang II is not affected by pH changes in this range. Therefore, the profile represents pH-sensitive interaction between the ligand and receptor. However, in [Sar1,Ile8]Ang II-amide and Ang II-amide the loss of the C-terminal negative charge results in a 10-fold loss of binding affinity. This illustrates that [Sar1,Ile8]Ang II and Ang II binding by the AT1 receptor involves a salt bridge interaction, which confirms earlier bioassay results (Refs. 1
, 2
, and 21
). The profile of losartan binding to the AT1 receptor is somewhat similar to that of [Sar1,Ile8]Ang II-amide. It is comparatively distinct from that of [Sar1,Ile8]Ang II, with a sharp optimum near pH 7.0, and is especially sensitive to higher pH values (Fig. 2). Because the substitution of the tetrazole moiety with azoles or sulfonamides of higher pKa results in a loss of binding affinity toward the native AT1 receptor, it was predicted that the tetrazole moiety is involved in an ion pair interaction with the receptor(12
, 13
). This putative salt bridge between losartan and the AT1 receptor is unassigned as yet. Assuming that the same residue of the receptor is involved in the putative salt bridge interaction with all AT1 receptor-specific ligands, its substitution with a neutral hydrogen-bonding residue must result in a 10-fold loss of affinity for [Sar1,Ile8]Ang II and Ang II, without affecting the affinity for [Sar1,Ile8]Ang II-amide and Ang II-amide. Because losartan and related non-peptides likely make fewer contacts with the AT1 receptor than do peptide ligands, the predicted effect of such a mutation is likely to be greater on their binding. On the other hand, substitution of a residue involved in some other interaction with the ligands should lead to a decrease in the affinity of all types of ligands.Three Different Mutations Cause a Decrease of [Sar1,Ile8]Ang II Binding Affinity
Since substitution of Gln for Arg, His, or Lys results in the removal of the positive charge without substantial change in the side chain size, the effect of Gln substitution on ligand binding was examined initially (Table 1). The mutants H183Q, H256Q, and H272Q had no effect on the binding of either peptide or non-peptide ligands.
Substitution of 3 residues in the putative transmembrane domain (Lys, Arg, and Lys) leads to a decrease of [Sar1,Ile8]Ang II binding affinity. All of the Arg mutants expressed receptor proteins at the same level, which were glycosylated as for the wild-type receptor. The expressed mutant proteins did not specifically bind any of the peptide and non-peptide antagonists. Therefore, the role of Arg in ligand binding remains unclear. The K102Q mutant showed loss of binding affinity toward all ligands (Table 1). Interestingly, K102Q mutation did not alter pH profile of [Sar1,Ile8]Ang II binding (Fig. 2) and the ability to discriminate differences between Ang II and [Sar1,Ile8]Ang II, indicating that Lys does not interact with the carboxylate group of the Asp1 side chain of Ang II (Table 1). Thus, these results are consistent with the requirement of a positively charged, long side chain at this position for stabilization of the AT1 receptor conformation.
The substitution of Lys with Gln did not significantly alter the binding of losartan. Surprisingly, however, the affinity of this mutant (K199Q) for [Sar1,Ile8]Ang II and Ang II was about 10- and 48-fold lower, respectively, than that of the wild-type receptor (Table 1, Fig. 3). The effect of pH on [Sar1,Ile8]Ang II binding of the K199Q mutant is shown in Fig. 2. The binding profile shows a shift of maximal binding to pH 7.0. When the pH is raised to 8.0, specific binding dropped to about 40% of that at pH 7.0 for the mutant, while it did not change for the wild type. This indicates that the positive charge of Lys plays a predominant role in the pH dependence of [Sar1,Ile8]Ang II binding. Similar profile is observed when [Sar1,Ile8]Ang II-amide binds to wild-type AT1 receptor (Fig. 2). This clearly establishes the complementarity of interaction between Lys and C-terminal carboxylate of [Sar1,Ile8]Ang II. However, the pH-binding profiles of losartan with the mutant K199Q and the wild-type receptor were identical. [Sar1,Ile8]Ang II binding to wild-type receptor at pH 6-9 is consistent with a salt bridge involving a Lys, but the pH optimum of 6.5-7.0 for losartan binding suggests that its negatively charged tetrazole group (pKa = 6, see (
13
) interacts with a basic residue that deprotonates closer to neutral pH. These observations lead us to question if Lys is indeed the common counterion for both carboxylate- and tetrazole-containing ligands. We investigated this question by substitution of Ala (K199A), Glu (K199E), or Arg (K199R) for Lys (Fig. 3, Table 1). Electrostatic interaction and hydrogen bonding could account for interaction of [Sar1,Ile8]Ang II and [Sar1,Ile8]Ang II-amide with these mutants (21
). We conclude that the effects of modifying the Lys positive charge are consistent with its being the exclusive counterion for peptidyl ligands. This conclusion supports the observation of Yamano et al.(5
).The role of the positive charge of Lys in losartan binding is unclear because a Gln substitution does not affect its binding and an Arg substitution unexpectedly led to reduction of binding affinity. It is possible that the binding interaction of tetrazole might be different from a conventional salt bridge interaction. This viewpoint is supported by the observation that the affinity of a variety of carboxyl-containing non-peptide antagonists is reduced (5-fold for Exp7711 and 14-fold for L-159,810; see (
11
) in the K199Q mutant (Table 1, Fig. 3). Therefore, the carboxyl-containing non-peptides display all properties of a salt bridge interaction with Lys. Moreover, this indicates that peptide to non-peptide differences do not drive non-peptide antagonists to bind at a different basic residue on the receptor.Losartan Binding: Alternate Counterion or Multiplicity of Interaction?
Assuming that the tetrazole group of losartan occupies the same space as the carboxyl group of EXP7711 within the receptor binding pocket, then the enthalpic contribution of Gln or Lys side chains should be identical for losartan binding. Alternatively, because of its larger size, the tetrazole may make additional contacts with potential neighboring residues. In a molecular model of the AT1 receptor, the His side chain from helix 6 (Fig. 1A) appeared to point toward Lys, and, thus, could potentially act as an additional counterion residue or as a hydrogen bond donor. As shown in Table 1, single-residue replacement of His with Ala, Gln, and Glu indicated that its contribution to losartan binding affinity is small if any. However, double mutant K199A/H256A showed a loss of affinity that is larger than the additive single mutant affinities toward both the ligands (Table 2). Substitution of an Arg at position 256 caused 30-fold loss of losartan binding affinity without significantly affecting [Sar1,Ile8]Ang II binding. In the double mutant K199A/H256R, however, the binding affinities of losartan and [Sar1,Ile8]Ang II were significantly improved, in comparison to the double mutant K199A/H256A. Hence, for [Sar1,Ile8]Ang II binding, an Arg substituted for His can restore the loss of the Lys side chain. However, the restoration of losartan binding affinity is only partial, indicating that the configuration of the residue at position 199 is also important to losartan binding. We conclude that in both K199R and H256R mutants, the Arg substitution might disrupt the normal function of His and Lys, respectively, in losartan binding.
These results indicate that the positive charge of the 199 residue is the most important electrostatic interaction required for [Sar1,Ile8]Ang II binding affinity. By contrast, losartan binding requires a side chain bearing an amino group, but it is not necessary that this group be protonated. The straightforward explanation may be that both Lys and His participate in tetrazole binding. The role of His may be secondary in the wild-type receptor. Therefore, in the K199Q and K199A mutants, removal of the charge is presumably compensated by His through a change of its pKa(), which then allows it to function as an alternate counterion in the same microenvironment as Lys (see (
24
) for similar examples). In the double mutant K199A/H256A, this environment for tetrazole binding may be altered substantially, since the Ala side chain is smaller than that of either Lys or His (Table 2). This could result in the formation of a cavity in the receptor structure(25
), such that the tetrazole group probably cannot make van der Waals contacts with either of the substituted Ala residues. The cavity in the pocket is expected to affect [Sar1,Ile8]Ang II binding affinity as well (Table 2). The Arg substitution in the mutant K199A/H256R probably allowed the restoration of [Sar1,Ile8]Ang II binding by filling the cavity in addition to restoring the positive charge. The positive charge of Arg could satisfy losartan but the bulkier Arg side chain may cause steric problems, explaining partial improvement of affinity compared to that of K199A/H256A mutant.A conventional hydrogen bond interaction between tetrazole and receptor must also be considered because such an interaction is probably involved in the binding of [Sar1,Ile8]Ang II-amide and Ang II-amide(
21
). However, since conventional hydrogen bonds are weak interactions, loss of such an interaction in the mutant K199A, for example, could not account for the observed 10-fold decrease in binding affinity. The observation that a Gln can perfectly substitute for Lys must indicate that either the -amide group of Gln or the -amino group of Lys are involved in a direct interaction with the tetrazole of losartan. Receptor/ligand interactions in which sp2 nitrogen atoms form a stacked interaction with aromatic rings or form an amino/aromatic hydrogen bond are known(26
, 27
). Lys with its sp3 nitrogen cannot form stacked interaction but could interact with planar aromatic rings. This might explain why Lys and Gln side chains have the same effect on losartan binding affinity and on the apparent pH dependence of binding. Such interactions may be further stabilized by additional interactions, with His, for instance. A Lys-aromatic interaction between receptor and ligand has not been reported before. In support of this suggestion, Duncia et al.(12
) have reported that an acidic aromatic group substituted in the meta position of the terminal phenyl ring in biphenyl series has a binding affinity comparable to that of tetrazole analogues. Furthermore, disruption of the conjugation in the terminal ring of the biphenyl reduces the binding affinity(3
, 13
). Therefore, aromaticity is a crucial determinant. Thus, exploitation of the novel interaction in the tetrazole binding subsite of the AT1 receptor might be worthy of consideration in future drug design.The conclusion that the same space is occupied by tetrazole and the carboxylate groups seems justified. This provides an experimental basis for docking peptides and non-peptide ligands to a common site on the AT1 receptor in modeling the receptor-ligand complex. In structurally related peptide hormone receptors for the neurokinins, research so far has shown that critical binding determinants for peptide ligands do not participate in the binding of the non-peptide ligands(
22
). Similar conclusions on the interaction of losartan with AT1 receptor have appeared(5
, 6
, 7
, 8
, 9
). However, unlike the non-peptide antagonists of the neurokinin receptor, losartan was developed through model-based drug development and pharmacophore overlay(3
), a concept that is further supported by the experiments in this report.Acknowledgements
We are indebted to the insightful suggestions of the late Dr. F. Merlin Bumpus. We thank Drs. A. Chiu and J. Duncia of Du Pont-Merck Co.; Dr. W. Greenlee of Merck-Sharp-Dohme Co. for a generous gift of antagonists; Dr. Kunio Misono for assistance in synthesis and characterization of peptides; Dennis Wilk, Robert Gaivan, and Xiao-Pu Liu for technical assistance; and Christine Kassuba for editorial assistance. We also acknowledge helpful suggestions and critique of Drs. S. Acharya, U. Chandrasekharan, and Y.-H. Feng.
REFERENCES
- Geneset J. Koiw E. Kuchel O. Hypertension: Physiology and Treatment. McGraw-Hill, New York1977: 183-201
- Wesson L.G. Fanelli Jr., G.M. Recent Advances in Renal Physiology and Pharmacology. University Park Press, Baltimore1974: 215-256
- Pharmacol. Rev. 1993; 45: 205-251
- Nature. 1991; 351: 233-236
- Biochem. Biophys. Res. Commun. 1992; 187: 1426-1431
- Proc. Natl. Acad. Sci. U. S. A. 1993; 90: 5133-5137
- J. Biol. Chem. 1994; 269: 16533-16566
- Proc. Natl. Acad. Sci. U. S. A. 1994; 91: 7046-7050
- J. Biol. Chem. 1994; 269: 20815-20818
Furakawa Y. Kishimoto S. Nishikawa K. (July 20, 1982, and October 19, 1982) U. S. Patents 4,340,598 and 4,355,040
- J. Comput. Aided Mol. Design. 1994; 8: 491-512
- J. Med. Chem. 1990; 33: 1312-1329
- J. Med. Chem. 1991; 34: 2525-2547
- Proc. Natl. Acad. Sci. U. S. A. 1986; 83: 599-603
- Methods Enzymol. 1990; 185: 537-566
- J. Biol. Chem. 1994; 269: 6743-6752
- Proc. Natl. Acad. Sci. U. S. A. 1977; 74: 5463-5467
- J. Biol. Chem. 1987; 262: 7076-7080
- J. Pharmacol. Exp. Ther. 1986; 2339: 71-77
- Biochem. Pharmacol. 1973; 22: 3099-3108
- J. Med. Chem. 1986; 29: 1968-1971
- Annu. Rev. Biochem. 1994; 63: 101-132
- Protein Sci. 1994; 3: 211-226
- Nature. 1994; 367: 639-642
- Science. 1992; 255: 179-183
- Nature. 1994; 362: 350-353
- J. Mol. Biol. 1994; 239: 315-331
Article Info
Publication History
Published online: February 03, 1995
Received in revised form:
December 9,
1994
Received:
October 6,
1994
Identification
Copyright
© 1995 ASBMB. Currently published by Elsevier Inc; originally published by American Society for Biochemistry and Molecular Biology.
User License
Creative Commons Attribution (CC BY 4.0) | How you can reuse
Elsevier's open access license policy
Creative Commons Attribution (CC BY 4.0)
Permitted
- Read, print & download
- Redistribute or republish the final article
- Text & data mine
- Translate the article
- Reuse portions or extracts from the article in other works
- Sell or re-use for commercial purposes
Elsevier's open access license policy