Journal of Neurochemistry, 2005, 92, 375–387 doi:10.1111/j.1471-4159.2004.02867.x

Mutagenesis analysis of the 5-HT2C receptor and a Caenorhabditis elegans 5-HT2 homologue: conserved residues of helix 4 and helix 7 contribute to -dependent activation of 5-HT2 receptors

Jinling Xie,1 Serghei Dernovici and Paula Ribeiro

Institute of Parasitology, McGill University, Ste Anne de Bellevue, Quebec, Canada

Abstract agonist binding affinity and significantly lower constitutive An alignment of serotonin [5-hydroxytryptamine (5-HT)] G pro- activity compared with wildtype. Mutagenesis of S7.45 in the tein-coupled receptors identified a at position 4.45 (helix C. elegans receptor increased serotonin binding affinity by up to

4) and a small polar residue (serine or cysteine) at 7.45 (helix 7) 25-fold and decreased Emax by up to 65%. The same mutations that occur exclusively in the 5-HT2 receptor family. Other of the cognate C7.45 in rat 5-HT2C produced a smaller fourfold serotonin receptors have a hydrophobic amino acid, typically a change in the affinity for serotonin and decreased agonist methionine, at 4.45 and an invariant asparagine at 7.45. The efficacy by up to 50%. Substitutions of S/C7.45 did not produce functional significance of these class-specific substitutions was a significant change in the basal activity of either receptor. All tested by site-directed mutagenesis of two distantly related mutants tested exhibited levels of receptor expression similar 5-HT2 receptors, Caenorhabditis elegans 5-HT2ce and rat to the corresponding wildtype based on measurements of 5-HT2C. Residues 4.45 and 7.45 were each mutated to a specific [3H]-mesulergine binding or flow cytometry analyses. methionine and asparagine, respectively, or an alanine and the Taken together, these results suggest that K4.45 and S/C7.45 resulting constructs were tested for activity. A K4.45M mutation play an important role in the conformational rearrangements decreased serotonin-dependent activity (Emax) of the rat leading to agonist-induced activation of 5-HT2 receptors. 5-HT2C receptor by 60% and that of the C. elegans homologue Keywords: Caenorhabditis elegans, G protein-coupled by 40%, as determined by a fluorometric plate-based calcium receptor, 5-HT2C, mutagenesis, serotonin. assay. The rat mutant also exhibited nearly sixfold higher J. Neurochem. (2004) 92, 375–387.

Serotonin [5-hydroxytryptamine (5-HT)] is a ubiquitous classes of 5-HT receptors belong to the large superfamily of neuroactive agent of both vertebrates and invertebrates. In seven transmembrane-spanning G protein-coupled receptors mammals, 5-HT regulates a variety of physiological phe- (GPCR). nomena in the CNS and periphery, including cognition, As a group, 5-HT2 receptors are characterized by having a sleep, pain perception, mood, feeding behavior, sexual relatively lower affinity for indolealkylamines, including behavior, temperature regulation and gastrointestinal func- tion (Weiger 1997). Among invertebrates, 5-HT acts as both a neurotransmitter and hormone and mediates feeding, Received May 19, 2004; revised manuscript received September 3, locomotion, circadian rhythm, defense behavior and meta- 2004; accepted September 10, 2004. bolic activity across various invertebrate phyla (Walker et al. Address correspondence and reprint requests to Paula Ribeiro, Insti- tute of Parasitology, McGill University, 21 111 Lakeshore Road, Ste 1996). This diversity of effects is mediated by multiple 5-HT Anne de Bellevue, Quebec, Canada H9X 3V9. receptors, a total of seven structurally distinct receptor E-mail: [email protected] classes (5-HT1–7), each of which is further divided into 1The present address of Jinling Xie is Vanderbilt University Medical several subtypes (Boess and Martin 1994). With the excep- Center, Department of Pharmacology, 452 Preston Research Building, tion of the mammalian 5-HT3 ionotropic receptor and a 23rd Avenue South at Pierce, Nashville, TN 37232-6600, USA. Abbreviations used: FACS, fluorescence-activated cell sorting; GPCR, recently identified nematode (roundworm) 5-HT-gated chlor- G protein-coupled receptor; 5-HT, 5-hydroxytryptamine; 5-HT2ce, Ca- ide channel (Ranganathan et al. 2000) all other known enorhabditis elegans 5-HT2 receptor; TM, transmembrane domain.

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 375 376 J. Xie et al. serotonin itself, and are preferentially linked to the Gq/phosp- pCIneo and includes the complete coding sequence of 5-HT2ce fused holipase C-b pathway of signal transduction. Three 5-HT2 at the C-terminal end to a FLAG epitope (Hamdan et al. 1999) For subtypes have been identified in mammals, 5-HT2A, 2B and studies of rat 5-HT2C, a cDNA encoding the full-length unedited , which differ on the basis of primary structure and receptor was obtained from the American Type Culture Collection pharmacological profiles (Roth et al. 1998). In addition, (ATCC, Manassas, VA, USA) and modified by PCR to introduce an N-terminal FLAG epitope. The resulting construct was subcloned 5-HT2 receptors have been cloned from invertebrates, including between the NheI/NotI sites of pCEP4 mammalian expression vector Drosophila (Colas et al. 1995), the snail Lymnaea (Gerhardt (Invitrogen, Burlington, Canada), confirmed by DNA sequencing and et al. 1996) and two nematodes (roundworms), Caenorhabditis used for site-directed mutagenesis. All point mutations were generated elegans (Hamdan et al. 1999) and the pig parasite Ascaris suum with the QuickChange mutagenesis kit (Stratagene, La Jolla, CA, (Huang et al. 1999, 2002). The Drosophila and Lymnaea USA), according to the recommendations of the manufacturer. The receptors show some binding characteristics of the mammalian mutations were verified by sequencing the full-length cDNAs. 5-HT2B prototype, whereas the nematode receptors have a distinctive pharmacological profile and may constitute a Cell culture and transfection separate subtype of 5-HT2 receptor (Hamdan et al.1999). COS7 were grown in Dulbecco’s modified Eagle’s medium The finding of 5-HT2 receptors in the lower invertebrates supplemented with 10% bovine fetal serum (Invitrogen) and supports the notion that the 5-HT2 class diverged early in 20 mM HEPES buffer at 37C in a humidified environment containing 5% CO . HEK293(EBNA1) cells were cultured in evolution, at least before the separation of nematodes. 2 Dulbecco’s modified Eagle’s medium containing L-glutamine and An impressive amount of research over the last few years has supplemented with 10% fetal bovine serum (Invitrogen), 1 mM been aimed at unveiling the structural organization of the sodium pyruvate, 250 lg/mL G418 and 20 mM HEPES buffer 5-HT2 receptor, particularly the 5-HT2A and 2C subtypes. In (Invitrogen). For transfection, cells were seeded in HEPES-buffered the absence of high-resolution crystallographic data, which are Dulbecco’s modified Eagle’s medium containing 10% dialysed fetal still lacking for all biogenic amine GPCRs, most of the bovine serum and cultured overnight to approximately 80% information available on 5-HT2 structure is derived from confluency. Unless indicated otherwise, cells were transfected in mutagenesis analyses and comparisons with bovine rhodopsin, 100-mm culture dishes (1.5–2.5 · 106 cells and 3 lg plasmid DNA/ the class I GPCR prototype (Palczewski et al. 2000; Teller dish) using FuGENE 6 (Roche Diagnostics, Laval, Canada), et al. 2001). This research has identified a number of primary according to the specifications of the manufacturer. Transfection agonist binding residues located mainly on transmembrane efficiency was monitored routinely by using a green fluorescence protein-encoding plasmid (pTracer) and was typically 40–50%. domains (TM) 3, 5 and 6, which are believed to constitute a core of the receptor’s binding pocket (Choudhary et al. 1995; Binding assays Almaula et al. 1996; Roth et al. 1997; Kristiansen et al. 2000; Binding assays of 5-HT2ce wildtype and mutants were performed Shapiro et al. 2000; Visier et al. 2002a; Ebersole et al. 2003). on crude membrane preparations of transiently transfected COS7 A number of other residues identified on TM2, 3, 6 and 7 and cells. Rat 5-HT2C wildtype and mutants were transiently expressed intracellular loops have been implicated in receptor activation in HEK293(EBNA1) cells. In both cases, transiently transfected and G protein coupling (Sealfon et al. 1995; Herrick-Davis cells were harvested 48 h post-transfection and lysed by brief et al. 1997, 1999; Roth et al. 1997; Prioleau et al. 2002; sonication in ice-cold TEM buffer (50 mM Tris-HCl, pH 7.4, Shapiro et al. 2002; Visiers et al. 2002a,b) Despite these 0.5 mM EDTA, 10 mM MgCl2). Binding assays were performed advances, however, a great deal remains to be learned about the with aliquots (5–10 lg protein/reaction) of a 28 000 g crude structural organization of GPCRs, in particular the membrane preparation in a total volume of 200 lL TEM buffer containing [3H]-mesulergine (75–86 Ci/mmol; Amersham, Baie subtle differences between the various receptor subtypes. In d’Urfe´, Canada) as the radiolabeled ligand. Saturation curves were this study, we have used site-directed mutagenesis to test a generated from a minimum of seven different labeled ligand number of TM4 and 7 residues which were found to occur concentrations in the presence and absence of 10 lM exclusively in 5-HT2 receptors, both vertebrate and inverteb- (Sigma, Oakville, Canada) for measurements of non-specific rate, and thus were postulated to have functional significance. A binding. Competition studies were performed by testing seven to mutagenesis analysis of two distantly related 5-HT2 receptors, eight concentrations of unlabeled competitor in the presence of a rat 5-HT2C and C. elegans 5-HT2ce, implicated a conserved constant amount of [3H]-mesulergine. All test ligands were prepared lysine of TM4 (K4.45) and a small polar residue of TM7 (S or in 0.1% ascorbic acid. Reaction mixtures were incubated at room C7.45) in the conformational activation of both receptors. temperature for 90 min. Reactions were then terminated by rapid filtration over 1.0 lm Filtermat (Molecular Devices, Sunnyvale, CA, USA) pre-soaked in 0.3% polyethylenimine and subsequently washed three times with ice-cold buffer (50 mM Tri-HCl, pH 7.4) Experimental procedures using a Skatron ClassicCell Harvester (Molecular Devices). All saturation and competition kinetic parameters were determined by Site-directed mutagenesis computer-assisted non-linear regression analysis using the Prism A C. elegans 5-HT2ce expression construct was used as a template for software package (GraphPad Software Inc., San Diego, CA, USA). site-directed mutagenesis. The construct was made previously in

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 Mutagenesis analysis of the serotonin 5-HT2 receptor 377

Ca2+ assays cytoplasmic helix 8 were identified and their boundaries were Receptor signaling activity was measured in intact cells using a adjusted manually to coincide with those of the rhodopsin template. fluorometric plate-based assay for detecting changes in intracellular Variable loop regions were built by searching the Sybyl protein calcium. A preliminary test determined that the rat 5-HT2C receptor databank with Composer using default parameters. The N-terminus produced robust calcium responses in both transfected HEK293(EB- (5-HT2C residues 1–53; C. elegans 5-HT2ce residues 1–54), third NA1) and COS7 cells upon agonist stimulation. However, the intracellular loop (5-HT2C residues 239–305; C. elegans 5-HT2ce HEK293(EBNA1) cells were easily removed from the plate during residues 238–364) and C-terminus (5-HT2C residues 388–460; the washing procedure, which increased experimental variability. C. elegans 5-HT2ce residues 445–683) were not built due to lack of COS7 cells produced considerably less well-to-well variation and structural information. In addition, a fragment of the predicted thus were selected for the study. COS7 cells were cultured overnight extracellular loop 2 of 5-HT2C (5-HT2C residues 208–211) could in HEPES-buffered Dulbecco’s modified Eagle’s medium contain- not be built and was omitted. The two model structures were refined ing 10% dialysed fetal bovine serum and transiently transfected in by energy minimization in the subroutine Powell using the Kollman- 100-mm dishes as described above (1.5–2.5 · 106 cells and 3 lg All Atom force field with a non-bonded cut-off of 8 A˚ and a plasmid/dish). Approximately 24 h post-transfection, cells were dielectric constant of 1. Minimizations were performed first with the harvested, seeded in black-walled, clear-base 96-well plates at a carbon backbone fixed, after which the entire model was minimized density of 80 000 cells/well and cultured for another 24 h before the until a convergence gradient value of 0.1 kcal/mol.A˚ was reached. To calcium assay. Cell density was optimized in preliminary experi- assess the effects of the mutations of interest, the appropriate residues ments within a range of 10 000–100 000 cells/well. In some were modified with the Biopolymer module and the models were experiments, cells were transfected directly into 96-well plates (1.5– reminimized to convergence. 2 · 104 cells and 50 ng of plasmid DNA/well), using FuGENE6, and assayed in the same plates 48 h post-transfection. The assay was Numbering of G protein-coupled receptor amino acid residues performed with the use of a FLIPR calcium assay kit, according to Amino acids of rat 5-HT2C and C. elegans 5-HT2ce are identified the recommendations of the manufacturer (Molecular Devices). according to the system of Ballesteros and Weinstein (1995). Each Briefly, cells were washed once in Hanks balanced saline supple- amino acid within a TM region is identified by the TM number mented with 2.5 mM probenecid (pH 7.4) and incubated with (1–7) followed by the position in the TM helix relative to an 100 lL of calcium dye reagent for 60 min at 37C. After incubation, invariant reference residue, which is arbitrarily assigned the number cells were placed immediately in a FlexStation plate fluorometer 50. Residues of interest to this study are identified as K4.45 equipped with a multichannel injector (Molecular Devices) and set (5-HT2C K175 and C. elegans 5-HT2ce K175) and S/C7.45 (5-HT2C 362 419 at kex ¼ 485 nm, kem ¼ 520 nm. Agonist or a vehicle was added in C and C. elegans 5-HT2ce S ). a volume of 50 lL/well at an injection speed of 80 lL/ s. Fluorescence measurements were taken at 1.52 s intervals before Other methods and after agonist addition for a total of 60 s per well. Each assay Expression levels of the various receptors in COS7 and HEK293(EB- plate included a mock-transfected control, wildtype receptor and a NA1) were monitored by in situ immunofluorescence, as described test mutant all transfected at the same time and seeded at the same previously (Hamdan et al. 2002), using a monoclonal antibody cell density. The raw data were analysed with the SoftmaxPro directed against the FLAG epitope (anti-FLAGM2; Sigma; 5 lg/mL) software package (Molecular Devices). The baseline was defined as and a secondary FITC-conjugated antibody (goat anti-mouse IgG; the average of the first five recordings on each curve. Functional Sigma; 1 : 200 dilution). The level of receptor expression on the cell responses were measured as peak fluorescence levels after subtrac- surface was measured by fluorescence-activated cell sorting (FACS) tion of the baseline. Basal (spontaneous) receptor activity was analysis on a FACScan flow cytometer (Becton-Dickinson, Oakville, determined by subtracting the baseline in the mock-transfected Canada) according to standard procedures. FACS analyses were control from that of the test sample also measured on the same assay performed on both COS7 and HEK293(EBNA1) cells transiently plate. All measurements were derived from at least three separate transfected with FLAG-tagged 5-HT2C-expressing constructs and experiments each performed in sets of three to four replicates. cell viability was assessed based on exclusion of propidium iodide. Secreted alkaline phosphatase reporter gene assays were performed in Three-dimensional modeling 293CRE-SEAP cells (Durocher et al. 2000) transiently transfected Theoretical three-dimensional models of the C. elegans 5-HT2ce with 5-HT2C wildtype or mutant expression constructs, as described receptor and rat 5-HT2C were generated with the homology previously (Hamdan et al. 2002). Protein was determined according modeling program Composer of the Biopolymer module of Sybyl to the method of Bradford (1976) using a Protein Assay Kit (Bio-Rad, 6.9 (Tripos Inc., St Louis, MO, USA). The models were produced by Mississauga, Canada). Statistical tests for significance were conduc- using the coordinates of bovine rhodopsin (1hzx) (Teller et al. 2001) ted using unpaired two-tailed Student’s t-tests. as a structural template. The primary sequence of each receptor was first aligned with the sequence of rhodopsin and the alignment was inspected to insure that seed residues matched perfectly, including Results the reference residues designated at position 50 of each helix (Ballesteros and Weinstein 1995) and conserved motifs, such as the Identification of 5-HT2-specific residues E/DRY and NPxxY motifs. The structural alignment was subse- A total of 38 vertebrate and invertebrate 5-HT receptor quently performed using default parameters. Structurally conserved regions corresponding to the seven transmembrane helices and sequences (Table 1), belonging to all six major classes of

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 378 J. Xie et al.

Table 1 5-Hydroxytryptamine (5-HT) rece- Receptor Accession no. Receptor Accession no. ptor sequences 5-HT1A_RAT J05276 5-HT5B_MOUSE X69867 5-HT2A_RAT P14842 5-HT1A_FUGRU O42385 5-HT2B_RAT P30994 5-HT2A_CIRGR P18599 5-HT2C_RAT P08909 5-HT_Balanus D83547 5-HT4_RAT U20907 5-HTdroA_Drosophila Z11489 5-HT6_RAT L03202 5-HTdroB_Drosophila Z11490 5-HT7_RAT L22558 5-HT2Dro_Drosophila X81835 5-HT1A_HUMAN P08908 5-HT_Spisula AAL23575 5-HT-1Da_HUMAN P28221 5-HTap1_Aplysia AF041039 5-HT-1Db_HUMAN M81590 Ap5-HTB1_Aplysia L43557 5-HT1E_HUMAN P28566 Ap5-HTB2_Aplysia L43558 5-HT2A_HUMAN P28223 5-HTLym_Lymnaea L06803 5-HT2B_HUMAN P41595 5-HT2Lym_Lymnaea U50080 5-HT2C_HUMAN P28335 5-HT1_Dugesia BAA22404 5-HT4_HUMAN Q13639 5-HT4_Dugesia BAA22403 5-HT5A_HUMAN P47898 5-HT2Asc_Ascaris AF005486 5-HT6_HUMAN P50406 5-HTce_C. elegans U15167 5-HT7_HUMAN P34969 5-HT2ce_C. elegans AF031414 5-HT5A_MOUSE Z18278 5-HT_Haemonchus AAO45883

List of vertebrate and invertebrate 5-HT receptor sequences analysed in this study. Details of the multisequence alignment are provided in the text.

serotonergic GPCRs, were aligned with the program Mac- functional binding sites was not significantly altered by the Vector (version 7.0) using the ClustalW method. The analysis mutagenesis. identified several amino acids that are unique to 5-HT2 Agonist and antagonist binding affinities were determined receptors (Fig. 1). Among these residues is a TM4 lysine at by competition against [3H]-mesulergine. A number of position 4.45 (K4.45). The cognate amino acid in other 5-HT2-selective ligands were tested. In addition, we tested receptor subtypes is hydrophobic and almost always a ligands that bind preferentially to other 5-HT receptors to methionine. In TM7, the 5-HT2 sequences are characterized determine whether the mutagenesis altered the pharmacolo- by having a small polar residue, a serine or cysteine, at gical profile of 5-HT2C. The analysis revealed that the position 7.45 of the helix. The 5-HT2C subtype has a binding of serotonin was influenced by mutations of K4.45 cysteine at this position whereas other 5-HT2 receptors, and C7.45 (Fig. 2). Compared with the wildtype, the including C. elegans 5-HT2ce, have a serine. Most seroton- K4.45M mutant showed a significant sixfold increase in ergic GPCRs and other amine receptors have a highly serotonin binding affinity (p £ 0.0005). The C7.45N substi- conserved asparagine at 7.45 (Fig. 1). K4.45, S7.45 and tution produced a fourfold increase in affinity (p £ 0.005) C7.45 were designated as 5-HT2-specific residues and were whereas an alanine mutation decreased binding affinity targeted for site-directed mutagenesis. fourfold (p £ 0.05). In addition to serotonin, the mutagenesis produced modest but significant changes in the binding Mutagenesis of the rat 5-HT2C receptor affinities of other serotonergic ligands (Table 2). The Four mutant forms of 5-HT2C were generated, each K4.45M mutant exhibited approximately fivefold higher carrying a single point mutation. The mutagenesis was affinity for the serotonin derivative, 5-methoxytryptamine, designed to change the native amino acid to that normally and threefold higher affinity for the 5-HT2 agonist, 1-(4- found at the same position in other serotonin receptors. iodo-2,5-dimethoxylphenyl)-2-aminopropane (DOI). This K4.45 was mutagenized to a methionine (K4.45M) and mutation had no effect on any of the 5-HT2 antagonists C7.45 to an asparagine (C7.45N). An alanine substitution tested, including mesulergine, , and of C7.45 (C7.45A) was also produced. Wildtype and mianserin, or selective ligands of other 5-HT receptor mutant receptors were tested for binding activity using the subtypes [methiothepin, 8-hydroxydipropylaminotetralin selective 5-HT2 ligand, [3H]-mesulergine. All constructs (8-OH-DPAT)]. In the case of the C7.45N mutant, the most exhibited saturable [3H]-mesulergine binding and similar pronounced effects were a significant fivefold decrease in the 3 kinetics. The Bmax values for [ H]-mesulergine varied by affinity for and a smaller two- to threefold decrease about twofold (Fig. 2), suggesting that the density of in the affinity for the antagonists mianserin and ketanserin.

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 Mutagenesis analysis of the serotonin 5-HT2 receptor 379

Fig. 1 Multiple sequence alignment of all six major classes of sero- regions shown are representative of an alignment of 38 sequences tonergic G protein-coupled receptors, including mammalian and and include the conserved E/DRY motif, transmembrane (TM) 4 and a invertebrate 5-HT2 sequences. The 5-HT2 class consists of three portion of TM7. Accession numbers for all the sequences used in the mammalian subtypes (5-HT2A, 2B and 2C) and four invertebrate analysis are provided in Table 1. The positions of 5-HT2-specific sequences, including the Caenorhabditis elegans 5-HT2ce receptor. residues tested in this study are indicated by arrows. The rat and C. elegans sequences used in this study are marked. The

Activity assays of the rat 5-HT2C receptor assayed on the same plate. This basal activity could be Wildtype and mutant 5-HT2C constructs were transiently inhibited by the inverse agonist, mianserin (not shown), expressed in COS7 and tested for calcium activity using a consistent with the notion that the receptor was spontane- fluorometric plate-based assay. Stimulation of the wildtype ously activated. Compared with the wildtype, the C7.45N receptor with serotonin produced a rapid, transient increase and C7.45A mutants had slightly elevated baselines but the in the level of intracellular calcium (Fig. 3). The agonist difference was not statistically significant. In contrast, the response peaked within 10 s of stimulation and returned to K4.45M mutation reduced the receptor’s basal activity to less near basal level after 60 s. Control cells transfected with than 10% of the wildtype level (Fig. 4). The average basal empty vector did not respond to serotonin. All three mutants activity of the K4.45M mutant was 300 ± 94 relative tested caused a significant decrease in the magnitude of the fluorescence units. agonist response. The most pronounced effect was observed In other experiments, we tested whether residues at with the K4.45M mutant. A kinetic analysis of the various positions 4.45 or 7.45 contributed to the selectivity of receptor species revealed that the K4.45M mutant decreased 5-HT2 receptors for Gq and the inositol triphosphate/Ca2+ the Emax for serotonin by nearly 60% compared with the signaling pathway. Cells transiently expressing 5-HT2C wildtype receptor (Fig. 4). Substitutions of C7.45 also wildtype or a mutant were assayed for cAMP-mediated decreased agonist efficacy significantly but to a lesser extent; signaling by using a reporter CRE-SEAP assay (Durocher asparagine and alanine mutations of C7.45 decreased Emax by et al. 2000) in both the presence and absence of forskolin. 42 and 54%, respectively. The results showed no evidence of cAMP signaling in the In addition to agonist-stimulated activity, the rat 5-HT2C wildtype or any of the three mutants (data not shown). receptor is known to have spontaneous (ligand-independent) activity, which is sensitive to inhibition by inverse Measurements of wildtype and mutant 5-HT2C (Herrick-Davis et al. 1999). In this study, we observed that expression the transiently expressed wildtype receptor exhibited an There is increasing evidence that some loss-of-function average basal activity of 4034 ± 638 relative fluorescence mutations of GPCRs are associated with receptor desensiti- units above the corresponding mock-transfected control zation and the loss of receptor molecules from the cell

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 380 J. Xie et al.

Fig. 3 Time course of serotonin-induced elevation of intracellular calcium in cells expressing 5-HT2C wildtype (WT) or the K4.45M mutant. A control transfected with vector only is also shown. Trans- Fig. 2 Saturation kinetics and inhibition of specific [3H]-mesulergine fected cells were pre-loaded with a calcium fluorescent indicator and binding to rat 5-HT2C. Kinetic parameters K and B were obtained d max then placed in a FlexStation (Molecular Devices) plate fluorometer set from saturation binding curves, using at least seven concentrations of at kex ¼ 485 nm, kem ¼ 520 nm. The baseline was recorded for 17 s radioligand. The K values for serotonin (5-HT) were obtained from i at which point serotonin (10 lM) was injected into each well at a speed competition assays against [3H]-mesulergine. Serotonin competition of 80 lL/ s. Fluorescence was monitored at 1.52-s intervals for up to curves for the wildtype (WT) receptor and three test mutants are 60 s. Data were normalized by subtracting the baseline in each test shown. Each curve is the average of duplicate determinations from a sample. The representative results shown are from a single experi- typical experiment repeated at least three times. The data were nor- ment that was repeated at least three times. malized relative to the level of maximum binding obtained in the absence of competitor. Kd, Bmax and Ki values were calculated by non- linear curve-fitting analyses with the Prism (GraphPad Software Inc.) surface (Kristiansen et al. 2000; Wilbanks et al. 2002). To software package. All kinetic parameters are the means and SEM of at test whether the mutations described here had similar effects least three independent determinations, each in duplicate. *Statisti- on the expression of the 5-HT2C receptor we performed cally different from WT at p £ 0.05. in situ immunofluorescence and flow cytometry analyses of the various populations of transfected cells. Each of the rat 5-HT2C constructs was engineered with an N-terminal (extracellular) FLAG epitope, which was targeted for the study. A first in situ immunofluorescence study of non- Table 2 Inhibition of [3H]-mesulergine binding to wildtype (WT) rat permeabilized transfected COS7 revealed essentially the same 5-HT2C and mutants pattern of surface expression in all test cells examined (not shown). This was subsequently confirmed by fluorescence- Ki (nM) activated cell sorting (FACS) analysis using live transfected Compound WT K4.45M C7.45N cells (Figs 5a and b). The results showed similar levels of cell surface fluorescence in all mutant and wildtype-expressing Methoxy-5-HT 51.3 ± 6.5 9.38 ± 0.31a 42.0 ± 4.0 cells but not in the mock-transfected controls or negative DOI 134 ± 13 40.2 ± 2.4a 94.6 ± 11.9 a controls lacking primary antibody. A slightly higher level of Mianserin 1.35 ± 0.25 1.78 ± 0.21 4.45 ± 0.28 Ketanserin 38.9 ± 8.9 39.3 ± 3.6 128 ± 14.5a expression was observed in cells transfected with the C7.45A Lisuride 12.7 ± 1.3 28.9 ± 2.4 59.4 ± 14.5a mutant, consistent with the elevated Bmax value of this mutant 3 Metergoline 0.29 ± 0.11 0.14 ± 0.08 0.23 ± 0.02 in the [ H]-mesulergine binding assays (see Fig. 2). However, Methiothepin 0.73 ± 0.10 1.25 ± 0.40 1.6 ± 0.38 this small difference in expression levels was not statistically 8-OH-DPAT > 10 000 > 10 000 > 10 000 significant at p £ 0.05. The assays were repeated with HEK293(EBNA1) transfected with the same 5-HT2C-expres- 3 Ki values were obtained from competition assays against [ H]-mesu- sing constructs and, again, no difference could be detected lergine using seven to eight concentrations of each drug. Data are the between the various mutants and wildtype (not shown). Thus, means and SEM from two to three independent experiments. Drugs that produced ‡ twofold differences between the WT and mutants the results suggest that the effects of the mutations on calcium were tested three separate times, each in duplicate. activity were due to changes in the activation of the receptor DOI, 1-(4-iodo-2,5-dimethoxylphenyl)-2-aminopropane. rather than changes in the level of protein expression or the aStatistically different from WT at p £ 0.05. loss of receptor molecules from the surface.

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 Mutagenesis analysis of the serotonin 5-HT2 receptor 381

(a)

(b)

Fig. 5 Flow cytometry analysis of COS7 cells expressing 5-HT2C wildtype (WT) or mutants. Cells were transfected with vector only (mock) or a construct designed to express 5-HT2C fused to an N-terminal (extracellular) FLAG epitope. (a) Cell surface expression Fig. 4 Serotonin (5-HT)-induced stimulation of intracellular calcium in was estimated by flow cytometry, using a monoclonal anti-FLAG cells expressing rat 5-HT2C wildtype (WT) or mutants. The data on antibody followed by a FITC-conjugated secondary antibody. Mock- each curve are the means and SEM of three to four replicates from a transfected cells and cells incubated with secondary antibody only typical experiment repeated at least three times. The mock-trans- (blk) were used as controls. Data were normalized relative to the WT fected control (vector) did not respond to serotonin. The results were level and are shown as the means and SEM of three independent standardized relative to the maximum response produced by the WT experiments. (b) Typical histogram plots produced by flow cytometry analysis of cells expressing WT 5-HT2C (solid line), mutant 5-HT2C assayed at the same time and on the same plate. EC50 and Emax values were obtained by computer-assisted non-linear curve-fitting (broken lines) or the mock-transfected control (shaded). Only two analyses of dose–response curves. Basal activity was determined by mutants (K4.45M and C7.45A) are shown. subtracting the baseline of the mock-transfected control assayed on the same plate from that of each test sample. Under the conditions tested, the basal activity of the WT receptor was 4034 ± 638 relative saturation curve for [3H]-mesulergine along with a summary fluorescence units. EC50, Emax and basal activity values are the means of binding kinetics. Non-specific binding was measured in and SEM of at least three independent experiments, each in sets of the presence of 10 lM mianserin and represented approxi- three to four replicates. *Statistically different from WT at p £ 0.05. mately 10% of total binding. As in the case of the rat receptor, mutagenesis of C. elegans 5-HT2ce did not alter Mutagenesis and characterization of the Caenorhabditis expression of functional binding sites in a significant manner. elegans 5-HT2ce receptor Bmax values of mutants and wildtype were similar within a The analysis of the rat 5-HT2C receptor suggested that range of 1.0–2.4 pmol/mg. positions 4.45 and 7.45 influenced serotonin binding and Competition assays against [3H]-mesulergine confirmed efficacy. To test whether this was true in other 5-HT2 the low affinity of the C. elegans receptor for serotonin receptors, we repeated the analysis with a C. elegans 5-HT2 (Fig. 6). The Ki values for serotonin and the related homologue, 5-HT2ce, which has exceptionally low affinity derivative 5-methoxytryptamine were 9.7 and 7.5 lM, for the natural ligand (Hamdan et al. 1999). The C. elegans respectively, nearly two orders of magnitude above the sequence was mutagenized at the same sites and three values obtained with the rat receptor. The K4.45M mutation mutants (K4.45M, S7.45N and S7.45A) were generated and did not change the Ki values significantly. In contrast, the assayed for [3H]-mesulergine binding. [3H]-mesulergine had S7.45N mutation markedly increased the affinity for both not been previously tested on this C. elegans receptor but ligands. Compared with the wildtype, the S7.45N mutant was shown here to bind saturably and specifically to both the exhibited 25-fold higher affinity for serotonin (Ki, 0.39 lM) wildtype and mutant species. Figure 6 shows a typical and nearly 11-fold higher affinity for 5-methoxytryptamine

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 382 J. Xie et al.

Fig. 7 Calcium assays of the Caenorhabditis elegans 5-HT2ce receptor. Cells expressing wildtype (WT) or mutant receptor were stimulated with serotonin and assayed for intracellular calcium using a fluorometric method, as described above (see Fig. 3). Typical sero- tonin dose–response curves for the WT, S7.45N mutant or a mock- transfected control (vector) are shown. Each curve is the average and Fig. 6 Binding studies of the Caenorhabditis elegans 5-HT2ce SEM of three to four replicates from a single experiment repeated receptor. (a) Typical [3H]-mesulergine saturation curve of the wildtype three times. The corresponding EC and relative E values are the (WT) receptor. Specific binding (squares) was calculated after sub- 50 max means and SEM of three independent experiments performed in sets traction of the non-specific component (triangles). Non-specific binding of three to four replicates. The basal activity was calculated as in Fig. 3 was determined in the presence of 10 lM mianserin. The data are the and is presented as the mean and SEM of three to five independent average of duplicate determinations from a typical experiment experiments. The basal activity of the WT C. elegans receptor was repeated three times. (b) Inhibition of specific [3H]-mesulergine binding 1280 ± 606 relative fluorescence units. *Statistically different from WT by serotonin (5-HT). Results were normalized relative to the level of at p £ 0.05. maximal binding measured in the absence of competitor. Data shown are typical competition curves for the 5-HT2ce WT and S7.45 mutant and were repeated at least three times. Ki values were estimated from competition curves for serotonin and the derivative, 5-methoxytrypta- was decreased by about 40% compared with the wildtype. mine. All kinetic parameters are shown as the means and SEM of at The C. elegans receptor expressed in COS7 cells exhibits a least three independent experiments, each performed in duplicate. basal activity of 1280 ± 606 relative fluorescence units *Values significantly different from WT at p £ 0.05. above that of the corresponding mock-transfected control. None of the substitutions tested had a significant effect on

(Ki, 0.69 lM). The alanine substitution (S7.45A) had no this basal activity. significant effect on the binding affinity of serotonin at p £ 0.05. Three-dimensional modeling of K4.45 and S/C7.45 Mutant and wildtype forms of the C. elegans 5-HT2ce Theoretical three-dimensional structures of the rat and receptor were transiently transfected in COS7 cells and C. elegans receptors were produced by homology mode- tested for calcium activity. As in the case of the rat receptor, ling, using the coordinates of the 2.8 A˚ crystal structure of all three mutations tested produced a decrease in the bovine rhodopsin as a template. The two receptors share magnitude of the serotonin-induced response compared with only about 66% overall homology at the level of primary the wildtype (Fig. 7). Substitutions of S7.45 had the sequence (Hamdan et al. 1999). However, the regions greatest impact on agonist activity. The S7.45A mutant surrounding 4.45 and 7.45 within the helical bundle are had 55% less agonist-induced activity, whereas the S7.45N similar in the two models. K4.45 is located within the mutant exhibited both a decrease in EC50 (from 1340 to transmembrane portion of TM4, nearly two turns of the 90 nM) and a decrease in Emax of about 65%. The K4.45M helix above the predicted cytoplasmic boundary (position mutant also exhibited diminished activity but to a lesser 4.38). The lysine resides in a predominantly hydrophobic extent; the efficacy of the agonist response in this mutant region formed by non-polar residues of TM4 (positions

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 Mutagenesis analysis of the serotonin 5-HT2 receptor 383

Fig. 8 Computer-generated models of (a) the rat 5-HT2C receptor and (b) the Ca- enorhabditis elegans 5-HT2ce receptor show proximity of the K4.45 side chain to aspartate 3.49 of the E/DRY motif and a neighboring residue at 3.45. (c) Predicted orientation of K4.45 relative to residues of the E/DRY motif in the 5-HT2C model. Arginine 3.50 is held by interactions with both aspartate 3.49 and a conserved TM6 glutamate (6.30). K4.45 interacts with the side chain carboxylate of D3.49 and the peptide carbonyl of A3.45. For clarity, only the cytoplasmic ends of helices 3, 4 and 6 are shown.

4.42–4.50) and neighboring helices, in particular TM3 and the cytoplasmic end of TM6 (Visiers et al. 2002a). These TM2. Several of these non-polar amino acids are conserved in interactions are also predicted in our two models (Fig. 8). the two receptors, for example F2.42 in TM2, I3.46 and L3.48 Position 7.45 occurs approximately midway in helix 7. in TM3. The predicted orientation of the lysine side chain is Our rat and C. elegans models suggest that 7.45 resides shown in Fig. 8. In both models the side chain of K4.45 is within a short stretch of amino acids that also includes a oriented towards the cytoplasmic end of TM3 and comes conserved glycine (G7.42) and is predicted to disrupt the within a relatively short distance (< 5 A˚ ) of aspartate 3.49, alpha helical nature of the TM7 segment (Fig. 9). The which is conserved in both receptors. The rat 5-HT2C model 5-HT2C model shows the side chain of its cysteine 7.45 predicts a direct contact between K4.45 and D3.49. The turned towards the interior of the helical bundle. In contrast, estimated distance between the e amino nitrogen of K4.45 and the side chain hydroxyl of serine 7.45 in the C. elegans the side chain carboxylate of D3.49 in 5-HT2C is 2.6 A˚ .In model is turned upwards and may form an intrahelical addition, the lysine side chain forms a direct contact with the H-bond with the peptide carbonyl of G7.42. In both models, peptide carbonyl of an adjacent residue at 3.45 in both 7.45 occurs within relative spatial proximity of the TM6 5-HT2C (N–O distance of 2.3 A˚ ) and C. elegans models aromatic cluster, in particular 6.48 and the (N–O distance of 2.7 A˚ ). Aspartate 3.49 belongs to the neighboring 6.44, which are conserved in all invariant E/DRY motif of rhodopsin-like GPCRs. A number amine GPCRs. The indole ring of W6.48 is shown nearly of models have proposed that the motif’s arginine, R3.50, perpendicular to the plane of the membrane, as predicted by forms a network of electrostatic interactions with the the rhodopsin structure and several recent models of amine neighboring D3.49 and a conserved glutamate (E6.30) at GPCRs (Visiers et al. 2002a). Immediately below 7.45 on

Fig. 9 Spatial orientation of C7.45 in the rat 5-HT2C model (right panel) and S7.45 in the Caenorhabditis elegans 5-HT2ce model (left panel) relative to residues of the aro- matic cluster motif. Only helices 6 and 7 are shown.

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 384 J. Xie et al. the same side of the TM7 helix is residue N7.49 of the closer to the cytoplasmic boundary, between positions 4.38 NPxxY motif. The side chain of N7.49 is turned towards and 4.41, and are not typically conserved (Javitch et al. conserved TM2 aspartate 2.50 in both models (not shown), 2000) whereas K4.45 is located within the predicted consistent with existing mutagenesis and modeling evidence transmembrane region and is type specific. Although the (Sealfon et al. 1995; Visiers et al. 2002a). possibility of a structural role cannot be ruled out, we postulated that K4.45 was more likely to be involved in a particular aspect of receptor activation. Discussion To interpret the results of the mutagenesis, we generated The growing numbers of invertebrate GPCR sequences three-dimensional models for both the rat 5-HT2C receptor available in the database provide a new wealth of information and the C. elegans homologue, using the coordinates of the for receptor structure–function studies. Multisequence com- 2.8 A˚ crystal structure of bovine rhodopsin as a template. parisons of mammalian receptors with lower vertebrate and There are a number of 5-HT2A and 5-HT2C models in the invertebrate orthologues allow for identification of residues literature (Chambers and Nichols 2002; Prioleau et al. 2002; that are conserved across phylogeny and thus are likely to have Shapiro et al. 2002; Visiers et al. 2002a,b) but the orienta- structural or functional significance. In this study, we have tion of K4.45 relative to the surrounding helices has not been taken advantage of recently cloned invertebrate 5-HT receptor discussed. Our models suggest at least two explanations for sequences to identify a few conserved residues that distinguish the results of the K4.45M mutagenesis. One possibility is the entire 5-HT2 group from other serotonergic GPCRs. that the loss of activity in the mutants was an indirect effect The lysine at position 4.45 is conserved only in the 5-HT2 caused by the introduction of a methionine in this predom- receptors. The majority of amine GPCRs, not only seroton- inantly non-polar environment. Replacing the native lysine ergic but also adrenergic, and muscarinic, have with a methionine in the 5-HT2C model moved the side a hydrophobic residue at this site, usually a Ile, Leu or Met. chain away from D3.49 and towards hydrophobic residues of A Met substitution of K4.45 significantly decreased agonist TM2, in particular A2.38 and F2.42 (not shown). A closer efficacy of the rat 5-HT2C receptor by 60% and that of the C. packing of these helices in the mutant may hinder the ability elegans homologue to a lesser extent, by about 40%. In of the receptor to become activated. A second explanation for addition, the mutation increased serotonin binding affinity of the results of the mutagenesis is that the native lysine is 5-HT2C and decreased this receptor’s basal activity without required for receptor activation, possibly due to its apparent affecting the level of receptor protein expression on the cell proximity to the E/DRY motif. The motif contributes to a surface. How a mutation at 4.45 could have such a network of electrostatic interactions that stabilize the inactive pronounced effect on activity is difficult to explain at state of rhodopsin-like GPCRs by holding the cytoplasmic present. Serotonergic 5-HT2 receptors have been the subject ends of TM3 and 6 close together (Visiers et al. 2002a,b). A of extensive research but the majority of these studies have large body of modeling and experimental evidence suggests focused on helices 3, 5, 6 and 7, which comprise the key that the activation of GPCRs is associated with a weakening functional domains of GPCRs. Considerably less is known of these interactions and the separation of the two helices about other helices, in particular helix 4. There is evidence (Ballesteros et al. 2001; Angelova et al. 2002; Shapiro et al. that the extracellular (upper) half of TM4, beginning at the 2002; Visiers et al. 2002a,b). Among the 5-HT2 receptors, in invariant tryptophan 4.50, plays a role in ligand binding and particular, mutations of D3.49 and the motif’s invariant may contribute to the receptor’s binding pocket (Roth et al. arginine 3.50 have been shown to cause significant inhibition 1997; Javitch et al. 2000). In contrast, the cytoplasmic half of agonist-stimulated activity (Shapiro et al. 2002; Visiers of TM4, where 4.45 resides, has not been widely investi- et al. 2002b). Ala and Glu substitutions of D3.49 in the 5- gated. Two recent studies of the D2 and muscar- HT2A receptor both decreased agonist efficacy (Visiers et al. inic M1 receptors reported that position 4.45 had no effect on 2002b), suggesting that the precise length and positioning of either ligand binding or receptor activation (Javitch et al. the acidic side chain at this site is critical for receptor 2000; Lu et al. 2001). However, these receptors both have a activation. Our models suggest that K4.45 may be suffi- methionine at 4.45. A lysine may function differently in the ciently close to this region to influence the orientation of 5-HT2 receptors. Recent GPCR models have proposed that D3.49 within the motif. This could facilitate the transition to basic residues located near the cytoplasmic interface of TM an active state by destabilizing the motif and/or stabilizing segments may interact with charged headgroups of mem- active forms of the receptor during these conformational brane phospholipids, thereby anchoring the receptor onto the rearrangements. Moreover, the direct contact between K4.45 membrane (Visiers et al. 2002a). If K4.45 also functions in and D3.49 predicted by the 5-HT2C model would be this manner, the decrease in activity associated with the expected to facilitate spontaneous activation of this receptor, mutagenesis may be related to changes in the interaction which could explain the loss of constitutive activity observed between the receptor and the membrane. However, the in the rat mutant. It should be emphasized that the rhodopsin anchoring Arg/Lys residues of the TM4 helix are clustered structural template has a glycine at 4.45 and there are no

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 Mutagenesis analysis of the serotonin 5-HT2 receptor 385 interactions between this site and any of the surrounding TM6 have been widely implicated in GPCR activity (Visiers helices (Palczewski et al. 2000; Teller et al. 2001). Thus, the et al. 2002a). Rhodopsin studies and several recent models predicted orientation of the lysine side chain in our two suggest that the indole ring of W6.48 shifts from a models must be viewed with caution. Nevertheless, together perpendicular to a parallel plane when the receptor is with the mutagenesis data, this study suggests that the activated by an agonist (Lin and Sakmar 1996; Visiers et al. cytoplasmic end of TM4 may be more important for the 2002a). This is thought to mediate a conformational switch activation of the 5-HT2 receptors than is currently believed. that serves to relay the signal from one end of the receptor to Whereas the M4.45 of other receptors appears to have no the other (Visiers et al. 2002a). A possible explanation for the direct functional role (Javitch et al. 2000; Lu et al. 2001), the results of the mutagenesis is that C7.45 is required to facilitate 5-HT2-specific lysine may facilitate conformational activa- this rearrangement of the neighboring tryptophan, either by tion and may also contribute to the natural propensity of stabilizing a favorable conformation in the surrounding region 5-HT2C towards spontaneous activity. This has important or by contributing directly to the conformational shift. Our implications for the study and modeling of 5-HT2 receptors models do not show any intramolecular contacts that might and deserves further investigation. suggest a direct involvement in this process. However, there In addition to K4.45, 5-HT2 receptors can be distinguished may be new contacts formed following agonist activation that by the lack of a highly conserved TM7 asparagine at position help the transition to the active state. 7.45. This asparagine is part of a NSxxNP(7.50)xxY motif, An inspection of the C. elegans model suggests that the which is common to a majority of amine GPCRs, including orientation of S7.45 in this receptor may be different from other 5-HT receptors, as well as muscarinic and gonadotropin that of the cognate cysteine in 5-HT2C. The side chain receptors (Konvicka et al. 1998; Lu et al. 2001; Angelova hydroxyl of S7.45 is turned upwards and is linked to the et al. 2002; Prioleau et al. 2002; Visiers et al. 2002a). peptide carbonyl of G7.42 by an intrahelical H-bond. As in Whereas the NP(7.50)xxY portion of this motif is also the case of 5-HT2C, an Ala substitution of 7.45 decreased conserved in the 5-HT2 sequences, the asparagine at 7.45 is agonist efficacy by about 50% and therefore the serine is also replaced with a smaller polar residue, typically a serine or, in required for full activation of this receptor. On the other the case of 5-HT2C, a cysteine. Residues of this motif, hand, the asparagine mutagenesis had a more pronounced notably N7.49 and Y7.53, form intramolecular contacts with effect on the C. elegans receptor than 5-HT2C, decreasing neighboring helices that contribute both to the stability of the agonist efficacy by as much as 65%. Replacing the native inactive state and the transition to an active conformation serine with an asparagine in the C. elegans model removed (Kristiansen et al. 2000; Sealfon et al. 1995; Rosendorff the additional intrahelical bond with G7.42 and changed the et al. 2000; Prioleau et al. 2002; Visiers et al. 2002a). orientation of H-bond-forming groups in the asparagine side Position 7.45 has not been tested in any of the 5-HT chain towards the interior of the helical bundle (not shown). receptors. However, in the glycoprotein and muscarinic It is possible that the interaction between S7.45 and G7.42 is GPCRs the cognate N7.45 is required for receptor activation needed to stabilize the active conformation of this receptor. and may contribute directly to the binding crevice (Angelova In addition, the bulkier side chain of N7.45 may hinder the et al. 2000, 2002; Lu et al. 2001). In this study, we have repositioning of the indole ring of W6.48, which could found that the smaller polar residue of the 5-HT2 sequences explain the decrease in efficacy seen in both the rat C7.45N also plays an important role in GPCR activity. Ala and Asn and C. elegans S7.45N mutants. substitutions of C7.45 in the rat 5-HT2C receptor both One unexpected effect of the S7.45N mutation in the decreased efficacy by about 50%, suggesting that this residue C. elegans receptor was that the loss of efficacy was is required for full activation of the receptor. Interestingly, the associated with a significant increase in the affinity for mutations had no apparent effect on basal activity, in contrast serotonin. S7.45 does not contribute to serotonin binding in to mutations of the neighboring NPxxY motif, which either the C. elegans receptor as an alanine substitution produced decreased or increased basal activity (Kristiansen et al. 2000; no significant change in binding affinity. In contrast, the Prioleau et al. 2002). Thus, C7.45 appears to be involved asparagine mutation increased binding affinity 25-fold and mainly in the process of agonist-induced activation and does also increased potency (i.e. decreased EC50) 15-fold, despite not contribute to the stability of the inactive state. C7.45 may the loss of agonist efficacy. A similar trend, although less facilitate the transition to an active conformation, in part, by pronounced, was observed in the rat 5-HT2C receptor; the contributing to the binding of serotonin as an Ala mutation asparagine substitution of C7.45 decreased Emax but decreased serotonin binding affinity about fourfold. Other increased serotonin binding affinity approximately fourfold residues of the mid-TM7 helix, in particular F7.38 and Y7.43, and potency about twofold. There is general consensus that have also been implicated in both agonist binding and the core of the binding pocket in the 5-HT receptors is activation (Roth et al. 1997). In addition, the effect of C7.45 formed mainly by residues near the extracellular ends of may stem from its spatial proximity to W6.48 and F6.44 of the TM6, 3 and 5 (Almaula et al. 1996; Roth et al. 1997; TM6 aromatic cluster. The conserved aromatic residues of Shapiro et al. 2000; Ebersole et al. 2003). However, there

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 386 J. Xie et al. are additional interactions with neighboring helices, TM7 in engineered mutations in transmembrane helices 6 and 7. J. Biol. particular, that can influence the positioning of the ligand Chem. 277, 32 202–32 213. within the pocket. Mutations of conserved TM7 aromatic Ballesteros J. A. and Weinstein H. (1995) Integrated methods for the construction of three-dimensional models and computational pro- residues, F7.38 and Y7.43, increased the Ki for agonists bing of structure-function relations in G protein coupled receptors. several fold (Roth et al. 1997), an indication that the Meth. Neurosci. 25, 366–428. extracellular (upper) half of the helix may contribute directly Ballesteros J. A., Jensen A. D., Liapakis G., Rasmussen S. G. F., Shi L., to the binding site. Our mutagenesis data suggest that 7.45 Gether U. and Javitch J. A. (2001) Activation of the b2-adrenergic may also play a role in binding and, moreover, the type of receptor involves disruption of an ionic lock between the cyto- plasmic ends of transmembrane segments 3 and 6. J. Biol. Chem. polar residue at 7.45 may influence the affinity for the natural 276, 29 171–29 177. ligand. The smaller serine is not required for binding and Boess F. G. and Martin I. L. (1994) Molecular biology of 5-HT recep- appears to be associated with a lower-affinity state. This may tors. Neuropharmacology 33, 275–317. be one of the reasons why some 5-HT2 receptors, notably Bradford M. (1976) A rapid and sensitive method for the quantitation of 5-HT2A and this C. elegans receptor, have typically lower microgram quantities of protein utilizing the principle of protein dye binding. Anal. Biochem. 72, 248–254. affinities for serotonin. On the other hand, a cysteine and, to a Chambers J. J. and Nichols D. E. (2002) A homology-based models of greater extent, an asparagine at 7.45 may reach further into the human 5HT2A receptor derived from an in silico activated G the receptor’s binding crevice and contribute interactions that protein-coupled receptor. J. Comp.-Aid. Mol. Des. 16, 511–520. increase both affinity and potency. We postulate that the Choudhary M. S., Sachs N., Uluer A., Glennon R. A., Westkaemper R. naturally occurring N7.45 of most 5-HT receptors is an B. and Roth B. L. (1995) Differential and ergopeptine binding to 5-hydroxytryptamine2A receptors: require an important contributor to the serotonin binding site and is thus aromatic residue at position 340 for high affinity binding. Mol. worthy of further investigation. Pharmacol. 47, 450–457. In summary, this study has determined that type-specific Colas J. F., Launay J. M., Kellermann O., Rosay P. and Maroteaux L. TM4 and 7 residues are required for activity of two (1995) Drosophila 5-HT2 serotonin receptor — coexpression with evolutionarily distant 5-HT2 receptors. That the mutagenesis fushi-tarazu during segmentation. Proc. Natl Acad. Sci. USA 92, 5441–5445. should produce similar effects in such different microenvi- Durocher Y., Perret S., Thibaudeau E., Gaumond M. H., Kamen A., ronments reinforces the functional importance of these Stocco R. and Abramovitz M. (2000) A reporter gene assay for residues for the entire 5-HT2 group. Our sequence align- high-throughput screening of G-protein-coupled receptors stably or ments have identified a number of other residues that are transiently expressed in HEK293 EBNA cells grown in suspension conserved across the 5-HT2 sequences and have been culture. Anal. Biochem. 284, 316–326. Ebersole B. J., Visiers I., Weinstein H. and Sealfon S. C. (2003) tentatively designated as type specific. Examples include Molecular basis of partial agonism: Orientation of indoleamine I4.60, G5.42, T5.61, V6.40 and F6.41. Also noteworthy is ligands in the binding pocket of the human serotonin 5-HT2A the absence of a highly conserved (Y4.33) of most receptor determines relative efficacy. Mol. Pharmacol. 63, 36– 5-HT GPCRs, which is replaced with a variable residue in 43. the 5-HT2 sequences. Some of these amino acids, for Gerhardt C. C., Leysen J. E., Planta R. J., Vreugdenhil E. and Van Heerikhuizen H. (1996) Functional characterization of a 5-HT2 example Y4.33 and T5.61, occur within predicted intracel- receptor cDNA cloned from Lymnaea stagnalis. Eur. J. Pharma- lular loop regions and thus may play a role in the interactions col. 311, 249–258. between the receptor and its G protein. These residues are Hamdan F. F., Ungrin M. D., Abramovitz M. and Ribeiro P. (1999) currently under investigation in our laboratory. Characterization of a novel serotonin receptor from Caenorhabditis elegans: Cloning and expression of two splice variants. J. Neuro- chem. 72, 1372–1383. Acknowledgement Hamdan F. F., Abramovitz M., Mousa A., Xie J. L., Durocher Y. and Ribeiro P. (2002) A novel Schistosoma mansoni G protein-coupled This work was supported by a grant from the Canadian Institutes of receptor is responsive to histamine. Mol. Biochem. Parasitol. 119, Health Research (CIHR) to PR. 75–86. Herrick-Davis K., Egan C. and Teitler M. (1997) Activating mutations of the serotonin 5HT2C receptor. J. Neurochem. 69, 1138–1144. References Herrick-Davis K., Grinde E. and Niswander C. M. (1999) Serotonin 5-HT2C receptor RNA editing alters receptor basal activity: Almaula N., Ebersole B. J., Zhang D., Weinstein H. and Sealfon S. C. Implications for serotonergic signal transduction. J. Neurochem. (1996) Mapping the binding pocket of the serotonin 5-hydroxy- 73, 1711–1717. 2A receptor: Ser3.36(159) provides a second interaction Huang X., Duran E., Diaz F., Xiao H., Messer W. S. Jr and Komuniecki site for the protonated amine of serotonin but not of R. (1999) Alternative-splicing of serotonin receptor isoforms in the diethylamide or . J. Biol. Chem. 271, 14 672–14 675. pharynx and muscle of the parasitic nematode, Ascaris suum. Mol. Angelova K., Narayan P., Simon J. P. and Puett D. (2000) Functional Biochem. Parasitol. 101, 95–106. role of transmembrane helix 7 in the activation of the heptahelical Huang X. Y., Xiao H., Rex E. B., Hobson R. J., Messer W. S., lutropin receptor. Mol. Endocrinol. 14, 459–471. Komuniecki P. R. and Komuniecki R. W. (2002) Functional Angelova K., Fanelli F. and Puett D. (2002) A model for constitutive characterization of alternatively spliced 5-HT2 receptor isoforms lutropin receptor activation based on molecular simulation and

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387 Mutagenesis analysis of the serotonin 5-HT2 receptor 387

from the pharynx and muscle of the parasitic nematode, Ascaris Roth B. L., Willins D. L., Kristiansen K. and Kroeze W. K. (1998) 5- suum. J. Neurochem. 83, 249–258. hydroxytryptamine(2)-family receptors (5-hydroxytryptamine(2A), Javitch J. A., Shi L., Simpson M. M., Chen J. Y., Chiappa V., Visiers I., 5-hydroxytryptamine(2B), 5-hydroxytryptamine(2C)) — where Weinstein H. and Ballesteros J. A. (2000) The fourth transmem- structure meets function. Pharmacol. Ther. 79, 231–257. brane segment of the dopamine D2 receptor: Accessibility in the Sealfon S. C., Chi L., Ebersole B. J., Rodic V., Zhang D., Ballesteros J. binding-site crevice and position in the transmembrane bundle. A. and Weinstein H. (1995) Related contribution of specific helix 2 Biochemistry 39, 12 190–12 199. and 7 residues to conformational activation of the serotonin 5- Konvicka K., Guarnieri F., Ballesteros J. A. and Weinstein H. (1998) A HT2A receptor. J. Biol. Chem. 270, 16 683–16 688. proposed structure for transmembrane segment 7 of G protein- Shapiro D. A., Kristiansen K., Kroeze W. K. and Roth B. L. (2000) coupled receptors incorporating an Asn-Pro/Asp-Pro motif. Bio- Differential modes of agonist binding to 5-Hydroxytryptamine(2A) phys. J. 75, 601–611. serotonin receptors revealed by mutation and molecular modeling Kristiansen K., Kroeze W. K., Willins D. L., Gelber E. I., Savage J. E., of conserved residues in transmembrane region 5. Mol. Pharmacol. Glennon R. A. and Roth B. L. (2000) A highly conserved aspartic 58, 877–886. acid (Asp-155) anchors the terminal amine moiety of Shapiro D. A., Kristiansen K., Weiner D. M., Kroeze W. K. and Roth B. and is involved in membrane targeting of the 5-HT2A serotonin L. (2002) Evidence for a model of agonist-induced activation of 5- receptor but does not participate in activation via a ‘salt-bridge hydroxytryptamine2A serotonin receptors that involves the dis- disruption’ mechanism. J. Pharm. Exp. Therap. 293, 735–746. ruption of a strong ionic interaction between helices 3 and 6. J. Lin S. W. and Sakmar T. P. (1996) Specific tryptophan UV-absorbance Biol. Chem. 277, 11 441–11 449. changes are probes of the transition of rhodopsin to its active state. Teller D. C., Okada T., Behnke C. A., Palczewski K. and Stenkamp R. E. Biochemistry 35, 11 149–11 159. (2001) Advances in determination of a high-resolution three- Lu Z. L., Saldanha J. W. and Hulme E. C. (2001) Transmembrane dimensional structure of rhodopsin, a model of G-protein-coupled domains 4 and 7 of the M-1 muscarinic acetylcholine receptor are receptors (GPCRs). Biochemistry 40, 7761–7772. critical for ligand binding and the receptor activation switch. Visiers I., Ballesteros J. and Weinstein H. (2002a) Three-dimensional J. Biol. Chem. 276, 34 098–34 104. representations of G protein-coupled receptor structures and Palczewski K., Kumasaka T., Hori T. et al. (2000) Crystal structure of mechanisms. Meth. Enzymol. 343, 329–371. rhodopsin: a G protein-coupled receptor. Science 289, 739–745. Visiers I., Ebersole B. J., Dracheva S., Ballesteros J., Sealfon S. C. and Prioleau C., Visiers I., Ebersole B. J., Weinstein H. and Sealfon S. C. Weinstein H. (2002b) Structural motifs as functional microdomains (2002) Conserved helix 7 tyrosine acts as a multistate conforma- in G-protein-coupled receptors: Energetic considerations in the tional switch in the 5HT2C receptor — Identification of a novel mechanism of activation of the serotonin 5-HT2A receptor by ‘locked-on’ phenotype and double revertant mutations. J. Biol. disruption of the ionic lock of the arginine cage. Int. J. Quantum Chem. 277, 36 577–36 584. Chem. 88, 65–75. Ranganathan R., Cannon S. C. and Horvitz H. R. (2000) MOD-1 is a Walker R. J., Brooks H. L. and Holden-Dye L. (1996) Evolution and serotonin-gated chloride channel that modulates locomotory overview of classical transmitter molecules and their receptors. behaviour in C. elegans. Nature 408, 470–475. Parasitology 113, S3–S33. Rosendorff A., Ebersole B. J. and Sealfon S. C. (2000) Conserved helix Weiger W. A. (1997) Serotonergic modulation of behaviour — a phy- 7 tyrosine functions as an activation relay in the serotonin 5HT2C logenetic overview. Biol. Rev. Camb. Philos. Soc. 72, 61–95. receptor. Mol. Brain Res. 84, 90–96. Wilbanks A. M., Laporte S. A., Bohn L. M., Barak L. S. and Caron M. Roth B. L., Shoham M., Choudhary M. S. and Khan N. (1997) Identi- G. (2002) Apparent loss-of-function mutant GPCRs revealed as fication of conserved aromatic residues essential for agonist bind- constitutively desensitized receptors. Biochemistry 41, 11 981–11 ing and efficacy at 5-HT2A receptors. Mol. Pharmacol. 52, 259– 989 . 266.

2004 International Society for Neurochemistry, J. Neurochem. (2005) 92, 375–387