Hydrogen Bonding Interaction of Thyrotropin-releasing Hormone (TRH) with Transmembrane Tyrosine 106 of the TRH Receptor*

Thyrotropin-releasing hormone ("I, pyroglutamic acid-histidine-proline-amide) binds to a seven-trans-membrane-spanning, G protein-coupled receptor. We tested the hypothesis that %'06 of the third transmembrane helix of the TRH receptor (TRH-R) binds pyroglu- tamyl of by mutating Tyr" to Phe and replacing the ring carbonyl of the TRH pyroglutamyl moiety with a methylene group ([ProlITRH). Compared to the affin- ity of wild-type TRH-R for T R H , the affinities of [PhelOB]TRH-R for TRH and of wild-type TRH-R for [Pro'ITRH were 100,000- and 110,000-fold lower, respec-tively. The affinity of [PhelOBITRH-R for [Pro'ITRH was only 16-fold lower than that for TRH, demonstrating a lack of additivity of the effects of these changes in the receptor and ligand. These data provide compelling evidence

-fold lower than that for TRH, demonstrating a lack of additivity of the effects of these changes in the receptor and ligand. These data provide compelling evidence that the hydroxyl group of '&r1O6 of the TRH-R binds the TRH pyroglutamyl carbonyl group. To our knowledge, this represents the highest affinity, non-covalent bond yet observed between single functional groups of a GPCR and ligand and is the first delineation of a direct binding interaction between a residue in the transmembrane core of a GPCR and a specific moiety of a peptide agonist.
TRH1 (1) is a tripeptide (pyroglutamic acid-histidine-prolineamide, cGlu-His-ProNH2) that regulates functions of the ante- The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked "aduertisement" in accordance with 18  rior pituitary gland and the nervous system. TRH binds to a cell surface receptor that transduces the signal intracellularly by activating the phospholipase C-inositol-1,4,5-trisphosphate-calcium-1,2-diacylglycerol pathway (2,3). The TRH-R (4-7) is a member of the seven-transmembrane-spanning, guanine nucleotide-binding protein-coupled receptor family (8). Much of what is known about binding and activation of receptors in this family has been derived from studies of rhodopsin (9,10) and of receptors for small, nonpeptidic ligands, such as catecholamines (11)(12)(13) and acetylcholine (14). It appears that retinal binds to rhodopsin and that catecholamines bind to their receptors within a core formed by the receptor transmembrane helices. The large glycoprotein hormones, thyroid-stimulating hormone and luteinizing hormone/chorionic gonadotropin, by contrast, bind primarily to the extended extracellular domains of their GPCRs (15,16).
Less is known regarding the nature of the binding sites of GPCRs for small peptides. It has been shown that both extracellular and intracellular domains of the neurokinin-1 receptor appear to be required for high aEnity binding of agonists (17,18) and that a His residue at the extracellular surface of transmembrane helix five is involved in the binding of a nonpeptide antagonist (19,20). Mutation of a Lys residue in the third transmembrane helix of the endothelin type B receptor caused several hundred-fold decreases in affinities for the native endothelins and an approximately 16,000-fold decrease in affinity for the endothelin type B-selective agonist sarafotoxin 6c (21). The binding of endothelin type-A selective antagonists and endothelin type-B selective agonists appear to be mediated by different subdomains of their respective receptors (22). Mutation of a Lys residue in the fifth transmembrane helix of the angiotensin type 1 receptor caused a n 8-fold decrease in binding of angiotensin (23). A Val residue in the sixth transmembrane helix of the cholecystokinin-B/gastrin receptor has been demonstrated to be involved in binding a nonpeptide antagonist (24).
It has been demonstrated in the GPCRs for neurotransmitters that an aspartate in the third transmembrane helix is important for binding and appears to serve as a counterion to the positively charged ligand. In contrast, we have previously demonstrated that the interaction between TRH and its receptor is non-ionic (25). We postulated that a residue in the third transmembrane helix of TRH-R might be involved in binding TRH through a nonionic interaction. The Tyr106 residue contains a potential hydrogen bonding group. It has been demonstrated in the muscarinic receptor that a homologous tyrosine (26) is involved in binding (27). We hypothesized that the hydroxyl group of tyrosine was binding the <Glu moiety of TRH and tested this hypothesis by mutating Tyr106 to Phe and studying interactions of these receptors with TRH analogs in which the pyroglutamyl moiety was substituted.
EXPERIMENTAL PROCEDURES Materials-TRH and chlordiazepoxide were purchased from Sigma, and [Pro'ITRH and [Abu'ITRH were from Bachem. my~-[~HIInositol was from Amersham Corp., and L3H1MeTRH was from DuPont NEN. The expression vector pCDM8 was from Invitrogen. Dulbecco's modified Eagle's medium was from Life Technologies, Inc. Nu-Serum was from Collaborative Research. Restriction endonucleases were from New England Biolabs.
Mutagenesis-The full-length, mouse TRH-R cDNA in pCDM8 (pCDM8mTRHR) (4) was used for mutation and transfection. Mutants were prepared by the polymerase chain reaction (28), and plasmid sequences were confirmed by the dideoxy chain termination method (29).
Cell Culture and Dansfection-COS-1 cells were maintained and transfected as described (4). Cells were seeded 1 3 days prior to transfection at 3.0 x 1 0 6 to 1.2 x lo6 celld100-mm dish, transfected using the Dm-dextran method, and maintained in medium with 10% Nu-Serum for 2-3 days, at which time cells were harvested and seeded into 12-well plates at 100,000 celldwell in medium with 5% Nu-Serum. Mock transfection was performed as described but without plasmid DNA.
Inositol Phosphate Formation--Two to 6 days after transfection, cells in monolayer i n 12-well plates were labeled with my~-[~HIinositol. Stimulation of inositol phosphate formation was measured 1-2 days later for 1 h at 37 "C by methods previously described (30).
Receptor Binding Studies-One to 2 days after re-seeding into 12well plates, competition binding experiments were carried out under equilibrium conditions using 1 rn t3H1MeTRH and unlabeled TRH analogs in buffer with cells in monolayer as previously described (25

RESULTS AND DISCUSSION
To test the hypothesis that T y r ' " was important for binding,  (32,33). The similar extent of maximal stimulation by TRH analogs for WT receptors and by TRH for WT and Phe" receptors is consistent with the idea that the efficacies are similar. To confirm for WT TRH-Rs that relative potencies reflect relative binding affinities, activation and binding studies (Fig. 1, upper panel; Table I) were performed with intact cells under identical conditions using a TRH analog substituted at the <Glu moiety with a prolyl moiety ([Pro'lTRH) (see Fig. 2 for structures). The EC50 for stimulation of IP formation by [Pro'lTRH in cells expressing WT TRH-Rs was 110,000-fold higher than that for TRH. The Ki of WT TRH-Rs for [ProlITRH was 43,000-fold higher than for TRH. Thus, although the absolute values of the EC50s were lower than their respective Kj values, a common finding with GPCRs (34), the relative potencies and relative affinities were similar. These data are consistent with the idea that the relative order of potencies reflects the relative order of affinities of binding under these conditions. Of note, we have found similar results for all mutant TRH-Rs in which we have been able to assay binding directly.2 We measured the potencies of stimulation of I P formation by TRH analogs in cells expressing [PhelOG]TRH-Rs (Fig. 1, lower panel; Table I). The ECS0 for TRH stimulation of IP formation in cells expressing [PhelOGITRH-Rs was 100,000-fold higher than in cells expressing WT TRH-Rs. These data indicate the importance of T y r 1 O 6 in binding TRH. Substitution of Pro for <Glu results in a TRH analog in which the carbonyl group of <Glu has been substituted by a methylene group (Fig. 2). The    due interacted with the receptor without " l j + 0 6 , that is, when changes in both the receptor and agonist were studied together, is minimal in comparison to the approximately 100,000-fold decrease caused by mutating Tyr106 to Phe or by deleting the carbonyl moiety of the <Glu residue. The reason for this 16-fold decrease is not clear. There may be small conformational changes in the analog and [PhelO'ITRH-R such that an interaction with another contact site is slightly weakened. It is unlikely that substitutions at <Glu lead to large conformational changes in TRH because <Glu is not thought to be involved in stabilizing the structure of TRH (35, 36). It is possible that a mutation may indirectly alter a distant binding site by causing a conformational change in the receptor as opposed to altering a site of interaction directly. The similar decreases in affinity resulting from mutation of Tyr106 or substitution of the <Glu carbonyl, as well as the lack of additivity of these changes, indicate a direct effect. Moreover, we found that the Ki of [PhelM]TRH-R for the competitive TRH antagonist chlordiazepoxide was 2.8 2 1.0 (data not shown), which is not different from that of the endogenous TRH-R in GH3 cells (37,38) or the transfected WT TRH-R in COS-1 cells (4). This finding, taken together with the fact that WT and [Phe"]TRH-Rs demonstrate similar efficacies for TRH, confirms that the conformation of the mutant receptor has not been markedly altered. The data also show that Tyr106 does not bind chlordiazepoxide.
A Tyr residue homologous to T y r 1 O 6 of the TRH-R is strongly conserved in the rat, pig, and human muscarinic receptors (39).
Mutation of this Tyr to Phe in muscarinic receptors led to 11-15-fold decreases in agonist affinity and little or no change in antagonist binding (40). Analog binding data suggested that this Tyr is involved in binding the acetylcholine ester moiety (27). We have demonstrated a n approximately 100,000-fold decrease in agonist binding without affecting antagonist binding upon mutation of Tyr106, indicating that this Tyr is subserving a stronger interaction in the TRH-R than in the muscarinic receptor. An affinity decrease of 5 orders of magnitude corresponds to a loss of free energy of interaction of approximately 7 kcal/mol, which is consistent with loss of a strong hydrogen bond in a hydrophobic environment.
The relatively high solubility of the tripeptide TRH allowed for determination of affinities and potencies in the micromolar and millimolar range. This approach may not be possible with receptors for larger peptides or proteins. The possibility that small amounts of contaminating TRH could become significant at high concentrations of TRH analogs and bind to TRH-R is very unlikely because these analogs were synthesized de novo by solid state technique (41).
Relative potencies of activation reflect relative binding affinities for systems of comparable efficacy (31). However, variations in potencies relative to affinities of up to 30-fold have been found for some ligand-GPCR systems (42,43). Although this may cause a problem in interpretation of results in some ligand-receptor systems, it is unlikely that it could affect our conclusions. This is so because the 100,000-fold decrease in potency observed with [Phelo6]TRH-R is much greater than would be expected for a change in efficacy that does not affect maximal stimulation. Thus, although we have not measured the affinity of [Phelo6]TRH-R directly, the decreased potencies observed are almost certainly caused by decreased affinity.
We were able to use complementary site-specific mutations in the receptor and substitutions of specific groups in the ligand to test our hypothesis. A lack of additivity of similar effects of receptor mutants and ligand analogs is strong evidence that the changes in the receptor and the ligand are affecting the same bond. An appropriate choice of mutations and ligand substitutions can provide detail at a n atomic level as described in this work. This approach has been used in studies of binding sites of GPCRs for neurotransmitters (44) and nonpeptide substance P antagonists (20) involving interactions of substantially lower affinity than that described here.
In summary, we have found compelling evidence, based on the results of experiments using site-specific mutagenesis and selective analogs of TRH, that the hydroxyl group of Tyr106 binds the carbonyl group of the <Glu moiety of TRH. To our knowledge, this represents the highest affinity, non-covalent bond yet observed between single functional groups of a GPCR and a ligand, and is the first delineation of a direct binding interaction between a residue in the transmembrane core of a GPCR and a specific moiety of a peptide agonist.