Chimeric Melatonin mt1 and Melatonin-related Receptors

Melatonin receptors bind and become activated by melatonin. The melatonin-related receptor, despite sharing considerable amino acid sequence identity with melatonin receptors, does not bind melatonin and is currently an orphan G protein-coupled receptor. To investigate the structure and function of both receptors, we engineered a series of 14 chimeric receptor constructs, allowing us to determine the relative contribution of each transmembrane domain to ligand binding and receptor function. Results identified that when sequences encoding transmembrane domains 1, 2, 3, 5, or 7 of the melatonin mt1 receptor were replaced by the corresponding domains of the melatonin-related receptor, the resultant chimeric receptors all displayed specific 2-[125I]iodomelatonin binding. Replacement of sequences incorporating transmembrane domains 4 or 6, however, resulted in chimeric receptors that displayed no detectable 2-[125I]iodomelatonin binding. The subsequent testing of a “reverse” chimeric receptor in which sequences encoding transmembrane domains 4 and 6 of the melatonin-related receptor were replaced by the corresponding melatonin mt1receptor sequences identified specific 2-[125I]iodomelatonin binding and melatonin-mediated modulation of cyclic AMP levels. To further investigate these findings, site-directed mutagenesis was performed on residues within transmembrane domain 6 of the melatonin mt1 receptor. This identified Gly258 (Gly6.55) as a critical residue required for high affinity ligand binding and receptor function.

Melatonin is the native ligand for high affinity G proteincoupled melatonin receptors (1,2). There are now almost 20 reported melatonin receptor sequences, which have been classified into three molecular subtypes, mt 1 , MT 2 , and Mel1c (3). Despite classification as molecular subtypes, they all have essentially identical binding affinities for melatonin (K i ϭ ϳ0.5 nM), and each mediates inhibition of stimulated cyclic AMP levels, mainly but not exclusively via the activation of G i proteins (4). During the cloning of the melatonin receptors a ho-mologue, the melatonin-related receptor, was identified, but despite containing 57% amino acid sequence identity with the TM 1 domains of the melatonin mt 1 receptor, it did not bind 2-[ 125 I]iodomelatonin or melatonin (5,6). The melatonin-related receptor remains an orphan GPCR with no ligand or signal transduction pathway yet identified.
The molecular structure and function of the melatonin mt 1 receptor has been investigated by site-directed mutagenesis, identifying the conserved His residue within TM5 as interacting with the 5-methoxy group of melatonin (7,8). Using the nomenclature of Ballesteros and Weinstein (9), this residue is located at position 5.46 (His 5.46 ), which corresponds to a position identified in the binding site of many other rhodopsin-like GPCRs, including Ser 5.46 (amino acid 207) in the hamster ␤ 2 adrenergic receptor (10). Ligand binding to the ␤ 2 adrenergic receptor has also been shown to involve an ionic interaction with Asp 3.32 (11,12), whereas the corresponding residue in the human melatonin mt 1 receptor, Met 3.32 , was found not to directly participate in melatonin receptor activation. (8). Therefore, the binding site of the melatonin receptor appears similar but not identical to those of other rhodopsin-like GPCRs. A recent molecular model proposed for the structure of the melatonin receptor-binding site hypothesized that two TM7 residues, Ser 7.38 and Ala 7.42 , may also be involved in melatonin binding interactions (13); however, site-directed mutagenesis of these residues recently suggested that this was not the case (8). Therefore, apart from His 5.46 , the residues forming the melatonin-binding site remain to be identified.
Many reports have been published that utilize chimeric receptor constructs to determine the contribution of different domains of GPCRs toward either ligand binding, binding selectivity, or receptor activation (14,15). This methodology is therefore an effective way to rapidly characterize GPCRs with poorly defined ligand-binding sites. The results from such chimeric receptor experiments provide a means to focus subsequent studies, such as site-directed mutagenesis, on the individual amino acid residues that provide ligand binding and signal transduction.
We have utilized the technique of recursive PCR (16 -18) to engineer 13 unique restriction endonuclease sites into the human melatonin mt 1 receptor DNA sequence, without altering the encoded amino acid sequence. This synthetic human melatonin mt 1 receptor sequence allowed the precise engineering of a series of 14 chimeric receptors with the human melatoninrelated receptor. These chimeric receptors enabled us to inves-tigate the relative contribution of each TM domain to ligand binding and receptor activation. The results from these studies were used to design site-directed mutagenesis experiments on the human melatonin mt 1 receptor. The data generated by site-directed mutagenesis identified a residue in TM6 as being critical for both ligand binding and receptor function.

EXPERIMENTAL PROCEDURES
Materials-Tissue culture media, supplements, and sera were purchased from Life Technologies Inc. 2-[ 125 I]Iodomelatonin (2200 Ci/ mmol) was obtained from NEN Life Science Products. Oligonucleotides for recursive PCR were synthesized and purified by Bioline, other oligonucleotides were form Sigma-Genosys. Melatonin and general reagents were obtained from Sigma. Molecular biology reagents and enzymes were purchased from Promega.
Engineering the FLAG Epitope onto the Human Melatonin mt 1 Receptor-The DNA sequence encoding the human melatonin mt 1 receptor (19) was engineered by PCR to introduce a FLAG epitope sequence (5Ј-GACTACAAGGACGACGATGACAAG-3Ј) directly after the ATG start codon. This was confirmed by DNA sequencing, and the engineered receptor sequence was ligated as a HindIII/KpnI restriction fragment into pcDNA3 (Invitrogen). This plasmid was named pcDNA3-mt 1 .
Construction of a Synthetic DNA Encoding the Human Melatonin mt 1 Receptor-A series of 16 overlapping oligonucleotides (60 -80 bases each with overlaps of 18 -20 bases) were designed to construct the DNA sequence encoding to human melatonin mt 1 receptor from the Bpu1102 I restriction endonuclease site (base 203) to one base following the stop codon (base 1054). Two sets of 8 oligonucleotides were assembled in two separate recursive PCRs, essentially by the method of Prytulla et al. (20). Recursive assembly PCR was performed using 5 pmol of each of 8 oligonucleotides in a 100-l volume with 2.5 units of native Pfu DNA polymerase (Promega). 10 l of the recursive assembly PCR reactions were then used as the templates in two PCR reactions to amplify the assembled DNA fragments (452 and 429 base pairs). Products were purified, ligated into pCR-Script (Stratagene), and confirmed by DNA sequencing. The products were recovered as Bpu1102 I/BamHI and BamHI/XbaI restriction fragments, respectively, and ligated into pcDNA-mt 1 . Correct ligation was confirmed by DNA sequencing, and the plasmid was named pcDNA3-mt 1 -(syn). The nucleotide modifications present in the synthetic human melatonin mt 1 receptor DNA sequence, relative to the published native sequence (2), are listed in Fig. 1.
Cloning and Engineering of the Human Melatonin-related Receptor-The two exons encoding the human melatonin-related receptor (5) were amplified by PCR from a genomic DNA library (CLONTECH). The two fragments were joined and engineered by PCR to introduce a FLAG epitope sequence directly after the ATG start codon. This product was ligated into pcDNA3 and confirmed by DNA sequencing, and the plasmid was named pcDNA3-MRR. PCR was subsequently used to truncate the melatonin-related receptor coding sequence, engineering a new TGA stop codon at bases 979 -982 (numbering excluding the FLAG epitope sequence), prior to the sequence encoding the large carboxylterminal domain. This PCR also introduced two restriction endonuclease sequences, a silent EcoRV restriction endonuclease site (bases 973-978) was placed immediately before the new stop codon, and a XbaI site was introduced 8 bases following the stop codon. This PCR product was cloned back into pcDNA3 and confirmed by DNA sequencing, and the plasmid was named pcDNA3-MRR-(trunc). The DNA encoding the carboxyl-terminal domain of the melatonin-related receptor was engineered by PCR to introduce a silent EcoRV restriction endonuclease site (bases 973-978) and a silent XbaI site (bases 1838 -1843) spanning the stop codon (bases 1840 -1842). This product was cloned back into pcDNA3 and confirmed by DNA sequencing, and the plasmid was named pcDNA3-MRR-(CT).
Generation of Chimeric Receptor Constructs-Segments of the human melatonin-related receptor were amplified by PCR from pcDNA3-MRR-(trunc) using oligonucleotide primers containing selected restriction endonuclease recognition sequences. Products were subcloned into pCR-Script, confirmed by DNA sequencing, and recovered as restriction products using the endonuclease recognition sites introduced by the PCR primers. These products were ligated into similarly restricted pcDNA3-mt 1 -(syn). When required the carboxyl-terminal domain of the melatonin-related receptor was obtained as an EcoRV/XbaI fragment from pcDNA3-MRR-(CT) and ligated onto appropriate chimeras. Confirmation of the successful construction of each chimeric receptor was performed by DNA sequencing. It should be noted that for some chimeras replacement of TM domains also included some sequence encoding the adjacent IL or EL domains, as illustrated in Fig. 2. In the following text we will therefore refer to chimeric receptors having replaced STM1 (segment including TM1) to STM7, to highlight that these sequences may encode loop(s) as well as the TM domains. TM will also now only be used where it specifically refers to the transmembrane region.
Generation of Site-directed Mutants-Two site-directed mutants of the human melatonin mt 1 receptor, Ala6.49Cys and Gly6.55Thr, were constructed using a PCR-based protocol with pcDNA-mt 1 -(syn) as the template. The amplification products were restricted, subcloned back into pcDNA3, and confirmed by DNA sequencing.
Transfection of COS-7 Cells-COS-7 cells were grown as monolayers in Dulbecco's modified Eagle's medium supplemented with 10% new born calf serum and 1% antibiotic/antimycotic solution, in 5% CO 2 at 37°C. Confluent plates were reseeded at 15,000 cells/cm 2 and used 24 h later for transfection. Cells were transfected using the DEAE-dextran method of Cullen (21). The expression of all receptor constructs was confirmed by the immunological detection of the engineered FLAG epitope as described previously (6). For ligand binding experiments, transfected cells were cultured for 72 h, washed twice with phosphate-buffered saline (pH 7.4), and harvested. Cells were pelleted in 1.5-ml microcentrifuge tubes (13,000 ϫ g, 1 min, 4°C) and stored frozen at Ϫ80°C.

2-[ 125 I]Iodomelatonin
Equilibrium Binding-2-[ 125 I]Iodomelatonin equilibrium binding experiments on transfected COS-7 cells were performed as described previously (22). Protein determinations were performed by the method of Bradford (23). All experiments were performed with duplicate or triplicate determinations, and the data were averaged. K d and IC 50 values were determined by Grafit software (Sigma), and K i values were calculated by the method of Cheng and Prusoff (24). All experiments were repeated on three or more occasions, and the mean values Ϯ S.E. were calculated.

Construction of G s /G i Chimeric G Protein ␣-Subunit-
The DNA encoding the ␣-subunit of G s (25), with the exception of the final 15 bases, was amplified by reverse transcription-PCR using messenger RNA isolated from HEK293 cells as the template. Based on the findings of Komatsuzaki et al. (26), a synthetic DNA linker was ligated onto the amplified product to essentially replace the DNA encoding the final 5 amino acid residues of G s with that encoding the final 5 amino acids of G i (27). The product was subcloned into pcDNA3 and confirmed by DNA sequencing.
Determination of Cyclic AMP Levels-COS-7 cells were co-transfected with receptor constructs and the chimeric G s /G i construct as described above. Following transfection cells were cultured for 24 h, trypsinized, and reseeded into 24-well tissue culture plates (1 ϫ 10 5 cells/well). Following a further 24-h incubation, cells were briefly washed with Dulbecco's modified Eagle's medium and incubated with appropriate concentrations of melatonin (in Dulbecco's modified Eagle's medium, 10 mM 3-isobutyl-1-methylxanthine) for 2 h at 37°C. Reactions were stopped by the addition of trichloroacetic acid (final concentration, 5%). Cyclic AMP levels were determined as previously reported (28). All experiments were performed with triplicate determinations, and the data were averaged. EC 50 values were determined by Grafit software (Sigma). Experiments were repeated two or three times, and where appropriate the mean values Ϯ S.E. were calculated.

2-[ 125 I]Iodomelatonin Binding Analysis of Chimeric Receptors Expressed in COS-7 Cells-Receptor constructs were transfected into COS-7 cells, and in all cases expression was
confirmed by immunological detection of the engineered FLAG epitopes (data not shown). The synthetic human melatonin mt 1 receptor displayed a 2-[ 125 I]iodomelatonin K d affinity value that was identical the native human melatonin mt 1 receptor (Ref. 19, Fig. 3). This demonstrated that modification of the nucleotide sequence of the synthetic mt 1 receptor sequence had no detectable effect on receptor expression. Both the human melatonin-related receptor and the truncated human melatonin-related receptor produced no detectable 2-[ 125 I]iodomelatonin binding ( Fig. 3 and data not shown). This confirmed the previous reports for both the human and ovine melatoninrelated receptors (5,6). These data also demonstrated that the lack of observed 2-[ 125 I]iodomelatonin binding to melatoninrelated receptors was not due to the presence of the extended carboxyl-terminal domain. Chimera [mt 1 (STM1 MRR)] displayed a 2-[ 125 I]iodomelatonin K d binding affinity close to that determined for the melatonin mt 1 -synthetic receptor but was expressed at a much reduced level (B max values, 14.2 and 506.9 fmol/mg protein, respectively). This implied that none of the residues that were changed in STM1 of this chimera relative to the melatonin mt 1 receptor sequence appeared to perturb the ligand-binding site but that one or more of these residues affected the level of receptor which bound the ligand at the cell membrane. Chimera [mt 1 (NTϩSTM1 MRR)] identified that when the aminoterminal extracellular domain of the melatonin-related receptor was also present in combination with STM1 then the determined K d remained close to that of the melatonin mt 1synthetic receptor but that the expression was further reduced to 1.14 fmol/mg protein. This additional reduction in B max may be due to the removal of the two Asn-linked glycosylation consensus sequences (Asn-Xaa-(Ser/Thr)) that are present in the amino-terminal domain of the melatonin mt 1 receptor but absent from the melatonin-related receptor sequence. The mutation or removal of such glycosylation sequences has been shown to reduce the B max values for some GPCRs (29), although the reduced expression seen in our studies could be due to other differences present in the amino-terminal sequences of the two receptors. Chimeras [mt 1 (STM2 MRR)], [mt 1 (STM3 MRR)], and [mt 1 (STM5 MRR)] all displayed K d and B max values close to that of the melatonin mt 1 receptor (within 3-fold), demonstrating that the residues that were changed in STM2, STM3, and STM5 of these chimeras relative to the melatonin mt 1 receptor sequence did not appear to be critical to either ligand binding or the detectable receptor level in the melatonin mt 1 receptor. Chimera [mt 1 (STM7ϩCT MRR)] displayed a 4.5-fold reduction in ligand binding affinity relative to the melatonin mt 1 -synthetic receptor and was expressed at a reduced level (B max ϭ 44.2 fmol/mg protein). This suggested that some of the residues that were changed in STM7 were required for high affinity ligand binding in the melatonin mt 1 receptor and that some residues affected the level of receptor which bound the ligand. Chimera [mt 1 (STM7 MRR)] in which the carboxyl-terminal domain of the melatonin-related receptor was truncated gave K d and B max values essentially identical to chimera [mt 1 (STM7ϩ CT MRR)], indicating that the reduction in B max relative to the melatonin mt 1 receptor did not appear to be caused by the presence of the carboxyl-terminal domain but was caused by the altered residues within STM7. Competitive displacement of 2-[ 125 I]iodomelatonin binding to chimera [mt 1 (STM7ϩCT MRR)] by a series of melatonin receptor ligands gave a very good correlation with binding data obtained for the synthetic melatonin mt 1 receptor (r 2 ϭ 0.99; slope ϭ 0.99). (Fig.  4). This demonstrated that the changes induced in STM7 caused a consistent (4 -5-fold) reduction in ligand binding affinity compared with the melatonin mt 1 receptor. Chimeras [mt 1 (STM4 MRR)] and [mt 1 (STM6 MRR)] displayed no detectable 2-[ 125 I]iodomelatonin binding. This implied that some residues that were changed in STM4 and STM6 of these chimeras relative to the melatonin mt 1 receptor sequence were critical for high affinity 2-[ 125 I]iodomelatonin binding to the melatonin mt 1 receptor.
To expand on these initial findings additional multiple domain chimeras were constructed and tested. Chimeras  (STM7ϩCT MRR)], respectively, but with a 7-9-fold increase in B max values. Therefore, in chimeras where either STM1 or STM7 was independently replaced, large reductions in B max values were observed, whereas in chimeras where STM1 and STM7 were replaced together, B max values were returned to near wild-type melatonin mt 1 receptor levels. This indicated that there may be an association involving STM1 and STM7 in both the melatonin mt 1 receptor and the melatonin-related receptor that increased the level of receptor that bound the ligand.
The final chimeras to be tested were those to further investigate the roles of STM4 and STM6 in melatonin receptor ligand binding. Chimeras [MRR (NTϩSTM4 mt 1 )] and [MRR (NTϩSTM6 mt 1 )] displayed no detectable 2-[ 125 I]iodomelatonin binding. Chimera [MRR (NTϩSTM4ϩSTM6 mt 1 )], however, displayed specific 2-[ 125 I]iodomelatonin binding, but this was not saturating at the highest concentration tested of 1500 pM (data not shown). Indeed, Scatchard analysis (30) indicated that the K d was substantially above 1500 pM. To obtain an affinity value for this chimera competitive displacement of specific 2-[ 125 I]iodomelatonin binding (1500 pM tracer) was performed using 2-iodomelatonin (1 ϫ 10 Ϫ13 to 1 ϫ 10 Ϫ5 M). Under the conditions of this study, where the tracer concentration was substantially below the K d of the tracer, the Cheng-Prusoff correction (24) states that the IC 50 obtained for the displacement curve is approximately equal to the K i . Using this principal the mean K i value determined for chimera [MRR (NTϩSTM4ϩSTM6 mt 1 )] was 28,800 Ϯ 10,800 pM, (n ϭ 3). The identification that chimera [MRR (NTϩSTM4ϩSTM6 mt 1 )] specifically bound 2-[ 125 I]iodomelatonin demonstrated the importance of residues in both STM4 and STM6 of the melatonin mt 1 receptor for the formation of the melatonin receptor ligandbinding site.

Melatonin-mediated Effects upon Cyclic AMP Levels in COS-7 Cells Transfected with Chimeric Receptors and G s /G i
Chimeric G Protein ␣-Subunit-COS-7 cells were co-transfected with the chimeric melatonin mt 1 /melatonin-related receptors and a chimeric G s /G i G protein ␣-subunit construct. In previous reported studies the use of chimeric G s /G i ␣-subunit has been shown to cause GPCRs that normally inhibit cyclic AMP levels, via activation of G i , to produce a ligand-mediated stimulation of cyclic AMP by promiscuous coupling to the G s /G i (26). This system was employed for functional studies in COS-7 cells because this cell line has been previously reported as being unsuitable for studies on GPCRs that signal via a G i -mediated reduction of cyclic AMP (31). Melatonin (1 ϫ 10 Ϫ5 M) was shown to elevate cyclic AMP levels ϳ2-fold in COS-7 cells co-transfected with the melatonin mt 1 -synthetic receptor but to have no effect upon cells co-transfected with the melatonin-related receptor or in mock transfected cells (Fig. 5). Melatonin elevated cyclic AMP levels in cells transfected with chimeric receptors involving the replacement of STM1, 2, 3, 5, or 7. Melatonin had no effect on cyclic AMP levels in COS-7 cells transfected with the individual STM4 and STM6 chimeras. Melatonin did however elevate cyclic AMP levels ϳ3-fold in COS-7 cells transfected with chimera [MRR (NTϩSTM4ϩSTM6 mt 1 )]. This again demonstrated that residues in STM4 and STM6 may play a very important role for both ligand binding and signal transduction in the human melatonin mt 1 receptor. 1 Receptor Expressed in COS-7 Cells-To investigate the roles of residues within TM6 of the human melatonin mt 1 receptor, two site-directed mutants were constructed, Ala6.49Cys and Gly6.55Thr. Both of these mutants exchanged the native melatonin mt 1 receptor amino acid residue for that in the same position of the human melatonin-related receptor. Saturation binding analysis of Ala6.49Cys produced similar K d and B max values as those obtained for the melatonin mt 1 -synthetic receptor (Fig. 6). Therefore Ala 6.49 probably does not play a critical role in ligand binding to the melatonin mt 1 receptor. Saturation binding analysis of mutant Gly6.55Thr also identified specific 2-[ 125 I]iodomelatonin binding, but this was not saturating at the highest concentration tested of 1500 pM. Using the same rationale as described above for chimera [MRR (NTϩSTM4ϩSTM6 mt 1 )] the affinity for mutant Gly6.55Thr was determined by competitive displacement of 2-[ 125 I]iodomelatonin binding (1500 pM) by 2-iodomelatonin (1 ϫ 10 Ϫ13 to 1 ϫ 10 Ϫ5 M) (Fig. 7). This produced a mean K i affinity value of 126,000 Ϯ 44,000 pM (n ϭ 3). These data demonstrated that Gly 6.55 was critical for providing high affinity ligand binding to the melatonin mt 1 receptor.

Dose Response Effect of Melatonin upon Cyclic AMP Levels in COS-7 Cells Transfected with Site-directed Mutant or Chimeric Melatonin Receptors and G s /G i Chimeric G Protein ␣-Sub-
unit-The dose-response effects of melatonin upon cyclic AMP levels in COS-7 cells co-transfected with the G s /G i chimeric G protein ␣-subunit and either the melatonin mt 1 -synthetic receptor, selected chimeric receptors, or the 2 site-directed mutant receptors were performed (Fig. 8). The results demonstrated that melatonin (1 ϫ 10 Ϫ12 to 1 ϫ 10 Ϫ4 M) caused a dose-dependent stimulation of cyclic AMP level for all receptor constructs. The melatonin mt 1 -synthetic receptor had a determined EC 50 affinity value of 186 pM, close to other reported values for melatonin-mediated activation of mt 1 receptors (2, 4). Chimera [mt 1 (STM7ϩCT MRR)] produced an EC 50 affinity value 7.3-fold lower than that of the mt 1 -synthetic receptor. This was consistent with the observed 4.5-fold reduction in ligand binding affinity previously described for this chimera (Figs. 3 and 4). Chimera [MRR (NTϩSTM4ϩSTM6 mt 1 )] produced an EC 50 affinity value ϳ830-fold lower than that of the mt 1 -synthetic receptor. This demonstrated that dual replacement of STM4 and STM6 of the melatonin-related receptor with the corresponding sequences of the melatonin mt 1 receptor could produce melatonin-mediated receptor activation at high dose. Site-directed mutant Ala6.49Cys produced a similar response to that of the mt 1 -synthetic receptor (within 2.2-fold), whereas mutant Gly6.55Thr displayed a profound loss of melatonin-mediated receptor activation with an EC 50 affinity value ϳ166,000-fold lower than that determined for the melatonin mt 1 -synthetic receptor. DISCUSSION The aim of this study was to expand the understanding of the molecular processes that underlie how melatonin receptors bind and are activated by melatonin. To achieve this, we employed a strategy involving chimeric receptors constructed from the human melatonin mt 1 receptor and the human melatoninrelated receptor. The rationale for these studies was based on the fact that melatonin receptors bind melatonin with high affinity (4), whereas the melatonin-related receptor, which shares considerable sequence identity with melatonin receptors, does not bind melatonin (5, 6). Our hypothesis for why the melatonin-related receptor does not bind melatonin was that this was due to the absence of a limited number of specific amino acid residues present in the melatonin receptor ligandbinding site and not due to a gross difference in the structures of the two receptors. If this hypothesis was accurate it would be possible to identify in which domains of the of melatonin receptor such residues occurred by the construction of chimeric receptors in which small regions of the melatonin mt 1 receptor were replaced with the corresponding region of the melatoninrelated receptor. A significant reduction in ligand binding affinity, compared with that of the melatonin mt 1 receptor, would show that one or more of the altered amino acid residues was important for the structure or function of the melatonin mt 1 receptor. Chimeric receptors in which STM2, 3, or 5 of the melatonin mt 1 receptor were replaced by the equivalent melatonin-related receptor sequences had little effect on either K d or B max when compared with the synthetic human melatonin mt 1 receptor. This demonstrated that the altered residues within these domains were not critical for high affinity ligand binding in melatonin receptors. It should be noted, however, that these studies were not designed to yield any information on the contribution of the conserved residues toward ligand binding. Indeed His 5.46 is conserved in both melatonin receptors and in the melatonin-related receptor and has been previously shown to be involved in ligand binding within the melatonin mt 1 receptor (7,8). Our findings do, however, suggest that any further site-directed mutagenesis studies that are designed to investigate possible ligand binding residues in STM2, 3, or 5 of melatonin receptors should concentrate on the residues that are conserved with the melatonin-related receptor. Chimeric replacement of STM1 also displayed little difference in K d value compared with the melatonin mt 1 receptor; however the B max was reduced by 36-fold. Therefore, one or more of the altered residues within this domain reduced the amount of receptor that was capable of binding the ligand. Immunological detection of the FLAG epitope on this chimera, however, indicated that the expression was indistinguishable from that of the melatonin mt 1 receptor; therefore, the loss of observed binding was probably due to a change in the structure of the expressed receptor and not due to impaired expression. Chimeric replacement of STM7 reduced both K d and B max values relative to the melatonin mt 1 receptor. This suggests that STM7 may be involved in the ligand-binding site of melatonin receptors. To further investigate the role of STM7 in the melatonin receptor-binding site, competitive displacement studies were performed using six bioisosteres of melatonin, including drugs that had a modified N-acetyl group (S20642, 5-methoxy tryptophol) or a modified 5-methoxy group (Jam09, N-acetyl tryptamine). The relative reductions in binding affinities for all of these drugs were identical, indicating that the changes produced in the STM7 chimera affected the general structure of the melatonin receptor-binding pocket and did not appear to affect specific interactions that might occur with, for example, the N-acetyl group. The chimeras involving the replacement of either STM4 or STM6 resulted in no detectable 2-[ 125 I]iodomelatonin binding. This indicated that one or more nonconserved residues in STM4 and STM6 may be critical in providing ligand binding in the melatonin mt 1 receptor.
To expand upon the above data, additional multiple domain chimeras were constructed. All of these chimeras were engineered to have the amino-terminal domain of the melatonin mt 1 receptor because of our previous observation that this produced higher expression levels than when the amino-terminal domain of the melatonin-related receptor was present. Two multiple chimeras were designed to investigate whether the reduction in B max observed for both STM1 and STM7 chimeras were related. These data identified that when both STM1 and STM7 were simultaneously replaced by melatonin-related receptor sequences, the B max levels were elevated to near melatonin mt 1 receptor levels. Therefore, high expression levels were seen when STM1 and STM7 were present as a pair from either the melatonin mt 1 receptor or the melatonin-related receptor but not when they were combined. It has previously been shown that TM1 and TM7 are adjacent in rhodopsin-like GPCRs (32,33). Therefore, the most logical explanation of the above phenomenon is that both melatonin receptors and melatonin-related receptors have some form of close association between TM1 and TM7 and that when this is perturbed the structure of either receptor is compromised. Potentially TM1 and TM7 could form specific interactions between their respective amino acid residues. There is evidence to suggest that different TM domains may interact in GPCRs (34), as illustrated by the reported interaction of TM2 and TM7 in the gonadotrophin hormone releasing hormone receptor (35). Because the melatonin and melatonin-related receptors are, however, not typical GPCRs in possessing Ala 7.50 in TM7, instead of the otherwise conserved Pro 7.50 , it is possible that this conveys unique aspects to the structure of these receptors, especially involving TM7. It is interesting to note that amino acid positions 1.56 (Ser 1.56 mt 1 , Ala 1.56 melatonin-related receptor) and 7.58 (Gln 7.58 mt 1 , Glu 7.58 melatonin-related receptor) both show nonconservation between the melatonin receptors and the melatonin-related receptor but are potentially closely positioned within the structure of these receptors. It is possible that residues such as these could therefore form unique interactions in these receptors. Additional site-directed mutagenesis studies may resolve the molecular mechanisms underlying the observed association of STM1 and STM7 in melatonin and melatonin-related receptors. The final three chimeric receptors that were constructed were to further investigate the role of STM4 and STM6 to ligand binding in the melatonin mt 1 receptor. We found that when STM4 and STM6 were replaced together in chimera [MRR (NTϩSTM4ϩSTM6 mt 1 )], this receptor construct could bind 2-[ 125 I]iodomelatonin, albeit with a relatively low affinity, K i 28,800 Ϯ 10,800 pM. Replacement of STM4 or STM6 alone resulted in no detectable 2-[ 125 I]iodomelatonin binding. The ligand binding results were complemented by measurement of the melatonin-mediated stimulation of cyclic AMP levels in COS-7 cells co-transfected with a G s /G i chimeric G protein ␣-subunit. All of the chimeras that bound 2-[ 125 I]iodomelatonin stimulated cyclic AMP levels, whereas the receptors that displayed no detectable ligand binding did not. The one exception to this was chimera [mt 1 (NTϩSTM1 MRR)], which bound 2-[ 125 I]iodomelatonin at a very low level (1.14 fmol/mg protein) but did not display changes in cyclic AMP level. This would appear to be solely due to the low expression level, because HEK293 cells that express mt 1 receptors at ϳ1 fmol/mg protein also failed to display melatoninmediated effects upon cyclic AMP levels (19).
The observation that chimera [MRR (NTϩSTM4ϩSTM6 mt 1  certain residues that were altered within these exchanged domains (STM4 and STM6) must be critical for providing ligand binding to the melatonin mt 1 receptor. The altered residues in STM4 mainly occurred within IL2 and EL2 and not TM4. These loop domains are not thought to be directly involved in ligand binding in rhodopsin-like GPCRs; therefore, the most probable explanation is that residues in these domains are required for the formation or maintenance of the melatonin receptor structure. This may possibly occur by interactions with other IL and EL domains. The altered residues present in STM6 included two residues within TM6 as well as some residues in IL3 and EL3. It is possible that residues in IL2 or EL2 may interact with residues in IL3 or EL3, respectively, thereby explaining why STM4 and STM6 were both required to produce detectable 2-[ 125 I]iodomelatonin binding in the chimeric receptors. Such putative interactions could be explored using more refined chimeras or by site-directed mutagenesis. To study altered residues within TM6, two site-directed mutants were constructed in the melatonin mt 1 receptor, Ala6.49Cys and Gly6.55Thr. These exchanged the melatonin mt 1 receptor residues for those present in the melatonin-related receptor. Mutant Ala6.49Cys displayed both a K d binding affinity and a functional EC 50 value close to those determined for the melatonin mt 1 receptor. This suggested that Ala 6.49 was not critical for ligand binding in the melatonin receptor. Mutant Gly6.55Thr, however, displayed a ϳ970-fold reduction in K d binding affinity and a ϳ166,000-fold reduction in functional EC 50 value relative to the determined melatonin mt 1 receptor values. Therefore, replacement of Gly 6.55 with a Thr residue severely compromised both high affinity ligand binding and ligand-mediated signal transduction in the melatonin mt 1 receptor. The positions of Gly 6.55 , Ala 6.49 , and His 5.46 are shown on a schematic projection of the human melatonin mt 1 receptor (Fig. 9). As can be seen, Gly 6.55 is predicted to point toward the central receptor core, and this suggests that mutant Gly6.55Thr affected ligand binding and receptor function by directly perturbing the structure of the melatonin receptor ligand-binding pocket. The most probable explanation for this would appear to be that the Thr residue was not tolerated in the melatonin receptor-binding pocket because of its hydroxyl containing side-chain structure.
We have presented data that Gly 6.55 is an important conserved residue required for ligand binding and receptor activation in melatonin receptors. Additional site-directed mutagenesis could further expand our knowledge of the molecular structure of TM6 and its role in the melatonin receptor ligandbinding site.