RNA Editing of the Human Serotonin 5-Hydroxytryptamine 2C Receptor Silences Constitutive Activity*

RNA transcripts encoding the serotonin 5-hydroxytryptamine 2C (5-HT2C) receptor (5-HT2CR) undergo adenosine-to-inosine RNA editing events at up to five specific sites. Compared with rat brain, human brain samples expressed higher levels of RNA transcripts encoding the amino acids valine-serine-valine (5-HT2C-VSV) and valine-glycine-valine (5-HT2C-VGV) at positions 156, 158, and 160, respectively. Agonist stimulation of the nonedited human receptor (5-HT2C-INI) and the edited 5-HT2C-VSVand 5-HT2C-VGV receptor variants stably expressed in NIH-3T3 fibroblasts demonstrated that serotonergic agonists were less potent at the edited receptors. Competition binding experiments revealed a guanine nucleotide-sensitive serotonin high affinity state only for the 5-HT2C-INI receptor; the loss of high affinity agonist binding to the edited receptor demonstrates that RNA editing generates unique 5-HT2CRs that couple less efficiently to G proteins. This reduced G protein coupling for the edited isoforms is primarily due to silencing of the constitutive activity of the nonedited 5-HT2CR. The distinctions in agonist potency and constitutive activity suggest that different edited 5-HT2CRs exhibit distinct responses to serotonergic ligands and further imply that RNA editing represents a novel mechanism for controlling physiological signaling at serotonergic synapses.

RNA editing is a post-transcriptional modification resulting in an alteration of the primary nucleotide sequence of RNA transcripts by mechanisms other than splicing. The enzymatic conversion of adenosine to inosine by RNA editing has been identified within an increasing number of RNA transcripts, indicating that this modification represents an important mechanism for the generation of molecular diversity. Several of these editing events have been shown to have significant consequences for cellular function. Transcripts encoding the Bsubunit (GluR-B) of the ␣-amino-3-hydroxy-5-methyl-4-isoxazolepropionic acid subtype of glutamate receptor undergo RNA editing events that modulate both the ion permeation and electrophysiological characteristics of this glutamate-gated ion channel (1)(2)(3). Mice that are deficient in their ability to edit GluR-B transcripts die at 3 weeks of age due to epileptic seizures, suggesting that editing of GluR-B RNA is important in the modulation of normal glutamatergic neurotransmission (4). These results suggest that the consequences of editing events within other, diverse RNA molecules might also have important ramifications for cellular function.
The monoamine 5-hydroxytryptamine (serotonin; 5-HT) 1 interacts with a large family of receptors to induce signal transduction events important in the modulation of neurotransmission (5). The 2C subtype of serotonin receptor (5-HT 2C R) is a member of the G protein-coupled receptor superfamily and stimulates phospholipase C, resulting in the production of inositol phosphates and diacylglycerol (6). We have recently shown that RNA transcripts encoding the rat, mouse, and human 5-HT 2C R undergo adenosine-to-inosine RNA editing events at five positions, termed A, B, C, D, and E (previously termed CЈ) (7,8), resulting in an alteration of amino acid coding potential within the putative second intracellular loop of the encoded protein. Editing at the A site, or at the A and B sites concurrently, converts an isoleucine to a valine at amino acid 156 of the human receptor, while editing at the B position alone generates a methionine codon at this site (Fig. 1A). Editing at C converts asparagine 158 to a serine; editing at E generates an aspartate at this site, and conversion at both C and E generates a glycine triplet. Editing at D results in the substitution of a valine for an isoleucine at position 160.
We have previously demonstrated a decrease in 5-HT potency when interacting with the rat 5-HT 2C R isoform 5-HT 2C-VSV , which is simultaneously edited at the A, B, C, and D positions encoding valine, serine, and valine at positions 157, 159, 161, respectively. This decrease in potency was reflected as a rightward shift in the dose-response curve for [ 3 H]inositol monophosphate accumulation (7). We proposed that the decreased potency resulted from a reduced G protein coupling efficiency induced by the introduction of novel amino acids into the second intracellular loop, a region known to be important for G protein coupling (9 -16). In the present study, we show that the patterns of 5-HT 2C R editing differ in human versus rat brain, resulting in the generation of novel receptor isoforms in human tissue. Radioligand binding and functional studies were designed to more fully test the hypothesis that the decreased agonist potency of edited isoforms reflects alterations in G protein coupling. The results show that differences in the ability to spontaneously isomerize to the active R* conformation play a role in the altered coupling properties of the fully edited receptor, suggesting that RNA editing of the 5-HT 2C R represents a novel mechanism for regulating neuronal excitability by stabilizing receptor signaling and enhancing the signal: noise ratio at serotonergic synapses.

EXPERIMENTAL PROCEDURES
Preparation of RNA, Reverse Transcription-PCR, and Editing Efficiency Analyses-Total RNA isolated from whole human brain was obtained from CLONTECH (Palo Alto, CA). Reverse transcription-PCR amplification of 5-HT 2C R messenger mRNA was performed as described previously (7) with the antisense oligonucleotide 5Ј-GCAGTAACAT-CAAAGCTTGTCGGCG-3Ј (coordinates 723-747 relative to the translation start site of the human 5-HT 2C R cDNA; Ref. 17) employed for cDNA synthesis. The sense oligonucleotide primer 5Ј-CCAGGGAAT-TCAAACTTTGGTTGCTTAAGACTGAAGC-3Ј (coordinates Ϫ30 to Ϫ5; Ref. 18) and the original cDNA synthesis primer or, in some cases, 5Ј-ATTAGAATTCTATTTGTGCCCCGTCTGG-3Ј (coordinates 372-389; Ref. 17) and the cDNA priming oligonucleotide (boldface sequence indicates introduced EcoRI restriction sites) were used for PCR amplification of human 5-HT 2C R sequences. Amplification proceeded for 75 s at 94°C, for 75 s at 50°C, and for 150 s at 72°C for 35 cycles. Products were purified on a 2% agarose gel, and primer extension analyses at the A and C/D sites were performed as described (7). For sequencing analysis, PCR fragments were digested with EcoRI and HindIII and unidirectionally subcloned into pBKSII Ϫ (Stratagene). Individual cDNA isolates were sequenced using the fmole sequencing system (Promega) as described previously (7).
Preparation of Human 5-HT 2C R Isoform DNA and Expression in Cell Culture-Human 5-HT 2C R variants were prepared by oligonucleotidedirected mutagenesis of human INI receptor DNA (a gift of Dr. Alan Saltzman) as described previously (10,19). These constructs were unidirectionally cloned into the eukaryotic expression vector pCMV2 (a gift of Dr. David Russell) at the EcoRI and XbaI sites. Transient experiments in NIH-3T3 cells were performed as described previously (7). For transient expression into the COS-7 cell line, cells were plated in 24-well plates at a density of 1 ϫ 10 5 cells/well 24 h prior to transfection. Two l of Lipofectamine reagent ® (Life Technologies, Inc.) and 0.5 g of plasmid were combined in 400 l of serum-free DMEM and added to each well for 5 h. Cells were washed with phosphate-buffered saline, and complete medium (DMEM, 10% fetal bovine serum, 100 units/ml of penicillin, and 100 g/ml of streptomycin) was added for 24 h. Stable cell lines were generated by co-electroporation of human isoforms in pCMV2 (60 g) and pRC/CMV (Invitrogen, Carlsbad, CA) (6 g) using electroporation conditions described previously (7). Single clones were selected using 1 mg/ml G418 (Geneticin ® ) in DMEM containing 10% dialyzed calf serum, 100 units/ml penicillin, 100 g/ml streptomycin (Life Technologies), and 1% calf serum (Hyclone ® ; Hyclone Laboratories, Logan UT). Cells were maintained in DMEM with 10% calf serum (Hyclone ® ), 100 units/ml penicillin, 100 g/ml streptomycin, 500 g/ml G418, and 1 M 2-bromolysergic acid diethylamide.
Phosphoinositide Hydrolysis Assay-Transiently transfected cells were prepared for inositol monophosphate analysis as described previously (NIH-3T3 cells, 7; COS-7 cells, 21). For stable cell lines, cells were washed three or four times with Hanks' buffer prior to plating in a 24-well plate (Falcon 3047, Becton-Dickinson Laboratories, Lincoln Park, NJ) in DMEM containing 10% dialyzed calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin. 24 h later, cells were labeled for 16 -20 h with 1 Ci of myo-[ 3 H]inositol/ml (20 -25 Ci/mmol, NEN Life Science Products) in serum-free, inositol-free DMEM containing 100 units/ml penicillin and 100 g/ml streptomycin. For isolation of inositol monophosphates from both transient and stable transfectants, agonists were added in the presence of 10 mM lithium chloride and 10 M pargyline. Incubations continued for 30 min unless otherwise noted. [ 3 H]Inositol monophosphates were isolated as described previously (22). Concentration response curves were analyzed using GraphPad Prism ® software (GraphPad Software Inc., San Diego, CA).
Radioligand Binding-Radioligand binding analyses were performed in transiently transfected cells as described previously (7,21). Stable cell lines were prepared by washing three or four times in Hanks' buffer and replacing the medium with DMEM containing 10% dialyzed calf serum, 100 units/ml penicillin, and 100 g/ml streptomycin for 24 h. Subsequently, cells were placed into serum-free DMEM containing 100 units/ml penicillin and 100 g/ml streptomycin for 16 -18 h prior to analyses. [ 3 H]Mesulergine binding was assayed in crude membrane preparations in 50 mM Tris-HCl, 10 mM MgCl 2 , pH 7.5, as described previously (23). Competition binding was performed with 0.5 nM [ 3 H]mesulergine in the presence of increasing concentrations of nonlabeled competitor at 37°C for 30 min. In some experiments, 100 M guanosine 5Ј-(␤,␥-imido)triphosphate (Gpp(NH)p) was added during the incubation. IC 50 values were determined by fitting data to a sigmoidal curve with variable slope using GraphPad Prism ® (GraphPad Software, Inc., San Diego, CA); one-site and two-site binding curves were compared using the F ratio. IC 50 values were converted to K i using the transformation of Cheng-Prusoff (24).

RESULTS
The 5-HT 2C R Editing Pattern in Human Brain Is Distinct from Rat Brain-To determine the editing pattern for 5-HT 2C R transcripts in human brain, whole brain total RNA was amplified by reverse transcription-PCR. The extent of editing at specific sites and the editing patterns within individual RNA species were determined by primer extension analysis and by subcloning and sequencing individual cDNA isolates, respectively. These studies revealed that the extent of editing in human brain at the A and D positions was similar to the levels previously identified within rat brain RNA, while editing at position B was lower in human brain (Ref. 7; Fig. 1B). By contrast, the editing efficiency at the C position was much higher in 5-HT 2C R transcripts isolated from human compared with rat brain (60 versus 35%). Editing at the E position (previously referred to as CЈ) was barely detectable in rat brain mRNAs from a variety of brain regions (Ͻ5%; Ref. 7), but this position was edited with much higher efficiency in human brain transcripts (Fig. 1B).
Due to increased levels of editing at the C and E positions, the pattern of editing observed within individual human brain 5-HT 2C R transcripts differed from that found in the rat (Fig.  1C). For example, the 5-HT 2C-VSV isoform, which comprised approximately 10% of the 5-HT 2C R transcripts isolated from whole rat brain, represented almost 40% of the messages derived from whole human brain (Fig. 1C). The enhanced editing at the E position in humans was also reflected by the appearance of variant RNAs encoding the novel 5-HT 2C-VGV , 5-HT 2C-VGI , and 5-HT 2C-IGI protein isoforms (Fig. 1C). RNA encoding the 5-HT 2C-VNV variant, which represented approximately 50% of the transcripts isolated from rat brain (7), only comprised 8% of the total 5-HT 2C R RNAs found in human brain (Fig. 1C). Together, these results indicated that editing of the 5-HT 2C R RNA differs considerably between rodent and human species and further suggested that the isoforms specific to human brain may have distinct functional roles.
Differential Signaling Profiles of the Human Edited Receptor Isoforms-To assess the functional consequences of editing within human 5-HT 2C R RNAs, we transiently expressed six of the major human isoforms in NIH-3T3 fibroblasts and examined the ability of these receptors to stimulate the phospholipase C signal transduction cascade by measuring the accumulation of [ 3 H]inositol monophosphates. Results from these studies revealed that the human 5-HT 2C-VSV receptor exhibited a 5-fold shift in potency for 5-HT compared with the 5-HT 2C-INI receptor (Fig. 2). The 5-HT 2C-VGV receptor exhibited a more substantial rightward shift in the dose-response curve for 5-HT, with an EC 50 value of 59 versus 2.3 nM for the 5-HT 2C-INI receptor variant. The EC 50 values for all other isoforms were not significantly different from the EC 50 observed for the 5-HT 2C-INI receptor.
To more carefully examine the properties of the human edited receptors, stable cell lines of the human 5-HT 2C-INI, 5-HT 2C-VSV , and 5-HT 2C-VGV receptor isoforms were generated in NIH-3T3 cells. The density of receptors in these cell lines was 258 Ϯ 87 for 5-HT 2C-INI , 173 Ϯ 28 for 5-HT 2C-VSV , and 1054 Ϯ 260 fmol/mg protein for the 5-HT 2C-VGV isoform. Similar to the phenotype observed in the transient transfections, 5-HT exhibited a lower potency when interacting with either the 5-HT 2C-VSV or 5-HT 2C-VGV isoforms as compared with the 5-HT 2C-INI receptor (Fig. 3A). The 5-HT 2C-VGV receptor-expressing cells, however, did not exhibit the substantial EC 50 shift seen in the transient experiments. It is possible that the higher receptor expression in the 5-HT 2C-VGV receptor line resulted in a significant receptor reserve for the 5-HT response, causing the EC 50 value to appear inappropriately low. To address this possibility, 5-HT 2C-VGV receptor cells were treated with the alkylating agent phenoxybenzamine (25). This treatment inactivated approximately 60% of the 5-HT 2C-VGV receptors, resulting in a statistically equivalent density between the three cell lines, thereby allowing a more direct comparison of EC 50 values. Subsequent stimulation of this partially inactivated population with 5-HT revealed that stably expressed 5-HT 2C-VGV receptors now exhibited a decreased maximum response (20% of untreated cells) and a 29-fold lower EC 50 for 5-HT when compared with the 5-HT 2C-INI isoform (Fig. 3A). This potency shift agrees closely with the 26-fold difference determined in transient transfection analyses (Fig. 2).
Two Agonist Affinity States Are Observable Only for the 5-HT 2C-INI Isoform-According to current theory of receptor/G protein interaction, the coupling of receptor with G protein results in an increase in the affinity of agonists for the receptor. Competition binding analyses were performed to evaluate differences in agonist affinity states for the 5-HT 2C-INI and 5-HT 2C-VGV receptors (Fig. 4). Competition binding regularly revealed a low and high affinity state upon interaction of 5-HT with the 5-HT 2C-INI receptor (Fig. 4A); competition binding curves could never be fit to a two-site model for 5-HT 2C-VGV receptors ( Fig. 4B; Table I). Data with the 5-HT 2C-VSV receptor was variable, sometimes revealing two sites, but more often best fit by a one-site model (data not shown). The 5-HT high affinity population of 5-HT 2C-INI receptors could be shifted to low affinity when binding analyses were performed in the presence of 100 M Gpp(NH)p, while the curves for the edited 5-HT 2C-VGV receptor remained unchanged. In the presence of Gpp(NH)p, significantly higher affinities were consistently observed for agonists when the 5-HT 2C-INI receptor was compared with the 5-HT 2C-VGV variant (Table I). This was not the case, however, when competition analyses were performed with mianserin, an inverse agonist at the 5-HT 2C R and a ligand that is predicted to preferentially interact with uncoupled receptors. and whole rat (white bars) brain is shown. Editing efficiencies at the A, C, and D sites were determined by a combination of primer extension (n ϭ Ն2 independent analyses) and sequencing analyses (n ϭ 50 independent cDNA clones for human samples); values represent the mean Ϯ S.E. Editing efficiencies at the B and E sites were determined by sequencing of 50 independent cDNA clones from human brain. C, the relative expression level, presented as a percentage of total isoform expression derived from a sequencing analysis of 50 (human) or 100 (rat) brain cDNA clones, is compared for 5-HT 2C R RNA derived from human (black bars) versus rat (white bars) brain. Rat data are from Burns et al. (7). These results again support the hypothesis that the 5-HT 2C-INI receptor couples more efficiently to G proteins, even in situations where no agonist is present to induce the coupled state.
Constitutive Activity Differences between Human 5-HT 2C R Isoforms-Rat 5-HT 2C Rs have been shown to exhibit constitutive activity (22,23), defined as the ability of a receptor to stimulate downstream signaling events in the absence of agonist occupation (reviewed in Ref. 26). The binding studies described previously suggested that the edited receptors might exhibit differential abilities to stimulate basal levels of inositol phosphate hydrolysis. Studies of constitutive receptor activity were performed in COS-7 cells, where transient expression gave reproducibly high levels of receptor expression, which is needed to accurately evaluate agonist-independent activity. An analysis of transient transfection experiments in which the nonedited and edited receptors were expressed at similar densities indicated that the 5-HT 2C-INI isoform elicited 5-fold greater levels of basal [ 3 H]inositol monophosphate generation compared with the 5-HT 2C-VGV receptor (Fig. 5). The addition of 1 M methysergide, a neutral antagonist, blocked 5-HT-stimulated inositol phosphate formation but had no effect on the basal activity of the 5-HT 2C-INI receptor (data not shown). The disparity in basal activity mirrors the differences in "precoupling" ability reflected in the competition binding experiments, again suggesting that the 5-HT 2C-INI receptor interacts more efficiently with the G protein linked to phospholipase C. Importantly, these analyses also revealed that similar maximal responses were obtained for both receptor isoforms (Fig. 5), indicating that the primary effect of editing may be to alter agonist-independent activity rather than affecting the intrinsic ability of receptor isoforms to promote G protein coupling. DISCUSSION RNA editing is a post-transcriptional process that contributes to molecular diversity within cells. 5-HT 2C R RNA transcripts undergo adenosine-to-inosine RNA editing events resulting in the generation of distinct amino acids within the second intracellular loop of the protein (7), a region known to be important for G protein coupling (9 -16). Editing within rat 5-HT 2C R RNA is mediated by the coordinated actions of a family of adenosine deaminases termed ADARs (adenosine deaminases that act on RNA; Refs. 7 and 27). These editing events are conserved among rodent and human species (8), suggesting that they serve an important role in receptor function.
Editing in the human brain creates a different complement of receptor isoforms due to an increase in the percentage of isoforms edited at position 158, most notably the 5-HT 2C-VSV and 5-HT 2C-VGV isoforms. Transient transfection of the principal human isoforms revealed that 5-HT exhibited the lowest potency at the novel 5-HT 2C-VGV receptor. The 5-HT 2C-VSV receptor was the only other isoform to show a significant difference in 5-HT potency relative to the nonedited isoform. In cell lines stably expressing the 5-HT 2C-INI , 5-HT 2C-VSV , or 5-HT 2C-VGV receptors, agonists exhibited a decreased potency when interacting with the edited receptor isoforms with the effect again being most dramatic upon interaction with the 5-HT 2C-VGV receptor. The potency difference between 5-HT 2C-VGV and 5-HT 2C-INI receptors was augmented when a majority of the 5-HT 2C-VGV receptors were inactivated with phenoxybenzamine, indicating that the observed potency phenotypes are not due to variations in receptor reserve between different receptor populations. Instead, these results are consistent with the interpretation that certain edited receptors exhibit a reduced G protein coupling capacity.
As a first step in defining the mechanism responsible for the reduced potency of agonists at edited 5-HT 2C Rs, competition binding profiles were examined to exploit the differential affinities of an agonist for G protein-coupled versus G proteinuncoupled receptor populations. The 5-HT competition binding curves for 5-HT 2C-INI receptors were shallow and best fit by a two-site competitive binding model. The high affinity state of 5-HT, representing approximately 50% of the binding, was eliminated by the addition of the GTP analog, Gpp(NH)p. In contrast, the 5-HT competition binding curves with 5-HT 2C-VGV receptors were best fit by a one-site competitive binding model, and the affinity approximated the low affinity site found with the 5-HT 2C-INI isoform. For the 5-HT 2C-VSV receptor, the curves sometimes fit two sites (data not shown), but most often the one-site competitive binding model gave the best fit. These results demonstrate that, compared with the 5-HT 2C-VSV and 5-HT 2C-VGV receptor populations, a relatively large proportion of 5-HT 2C-INI receptors exists in a G protein coupled state, providing the first direct evidence that RNA editing generates 5-HT 2C Rs that differ in the efficiency of coupling to G proteins. Thus, there appear to be graded states of precoupling ability for the edited 5-HT 2C Rs, with 5-HT 2C-INI receptor being most efficacious at G protein coupling, the 5-HT 2C-VSV variant being intermediate, and 5-HT 2C-VGV receptors existing predominantly in the uncoupled state.
The ternary complex model of receptor/G protein interaction has recently undergone revision to accommodate the finding that some receptors have the ability to induce effective G protein coupling in the absence of an agonist, termed constitutive activity (28). The model predicts that receptors have the capacity to spontaneously isomerize from an inactive conformation, termed R, to an active state, R*, with the R* version of the receptor having the ability to interact with G proteins. In experimental paradigms studying the properties of constitutively active mutant receptors, it has been observed that these receptors exhibit both a higher level of basal activity and a greater potency for agonists (28,29). Since the nonedited 5-HT 2C R exhibits constitutive activity (22,30), we considered the possibility that the agonist potency differences generated by RNA editing may reflect differential abilities to isomerize to the R* state, with the 5-HT 2C-INI receptor existing more readily in the R* form compared with the edited 5-HT 2C-VSV and 5-HT 2C-VGV isoforms. To evaluate alterations in constitutive activity, the basal activity of the nonedited 5-HT 2C-INI receptor was compared with the fully edited 5-HT 2C-VGV isoform, which exhibited the greatest shift in 5-HT potency. In COS-7 cells transiently expressing 5-HT 2C-INI receptors, basal phospho-

TABLE I Relative affinities of agonists for 5-HT 2C-INI and 5-HT 2C-VGV receptors
The   inositide hydrolysis was substantially increased in comparison with mock-transfected cells. In contrast, the 5-HT 2C-VGV receptor isoform supported only a modest degree of basal [ 3 H]inositol phosphate formation, about 20% of that produced by 5-HT 2C-INI receptors expressed at a comparable density. These results indicate that the fully edited 5-HT 2C-VGV receptor has a much lower intrinsic ability to support agonist-independent phosphoinositide hydrolysis than does the nonedited receptor isoform. Two alternative possibilities could explain the differences in 5-HT 2C-INI and 5-HT 2C-VGV receptor constitutive activity. One hypothesis is that large differences in the fraction of receptors residing in the R* conformation might underlie the distinction in constitutive activities. An alternative explanation is that both receptor isoforms exhibit the same ability to convert to R*, but the combination of amino acids within the 5-HT 2C-INI receptor promotes more efficient interaction with the G protein, due to either the favorable conformation of the second intracellular loop or actual contact of the nonedited amino acids with the G protein. In experiments designed to test the latter hypothesis, we found that 5-HT treatment elicited an increase in [ 3 H]inositol phosphate formation of the same magnitude in cells expressing equal densities of 5-HT 2C-INI or 5-HT 2C-VGV receptors, strongly suggesting that the two receptor isoforms have equivalent ability to couple to G proteins when bound by agonists. Thus, these results indicate that 5-HT 2C-INI receptors have a greater propensity to spontaneously isomerize into or maintain the active R* conformation than 5-HT 2C-VGV receptors. Thus, we propose that one consequence of RNA editing of the 5-HT 2C R is to silence its constitutive activity, thus increasing the signal:noise ratio at sites where editing efficiency is high.
The extended ternary complex model predicts that receptor isoforms with high levels of constitutive activity would exhibit higher affinity for agonists, even in the absence of G protein coupling (28,29). As a first step at evaluating agonist affinities in the absence of G proteins, we compared the binding affinities in the presence of Gpp(NH)p, assuming that this reflects the affinity of the G protein-uncoupled receptor. The affinity of 5-HT for the 5-HT 2C-VGV receptor was 5-fold lower than for the 5-HT 2C-INI isoform; significant differences were also found for the other agonists tested but not for the inverse agonist mianserin. These observations suggest that some structural perturbation allows the edited receptors to assume an inactive receptor conformation with lower affinity for agonists. Studies of agonist affinities using purified receptor protein are needed to definitively show that the agonist affinity differences do not depend on G protein interactions.
Recent structure-activity studies of other G protein-coupled receptors point to the importance of amino acid residues in positions analogous to the edited sites in the dynamics of receptor/G protein interactions. Amino acid residues that reside at positions analogous to Ile/Val 156 and Ile/Val 160 in other G protein-coupled receptors are important in terms of general G protein coupling ability and receptor regulation (10,12,16). For example, within the m1 and m3 muscarinic receptors and the ␤2 adrenergic receptors, the amino acid at Ile/Val 160 must be bulky and hydrophobic to induce a productive G protein interaction (10), consistent with our evidence that editing at position 160 contributes to a low efficiency of G protein coupling. Molecular modeling of the gonadotropin-releasing hormone receptor has shown that the Ile corresponding to the 5-HT 2C R Ile/Val 156 is critical for forming a "cage" around the arginine of the highly conserved DRY sequence in the second intracellular loop (16). It was proposed that the "arginine cage" stabilizes interaction with the adjacent aspartate residue, thereby enabling the receptor to remain in an active state.
Alteration of the Ile/Val 156 position of the 5-HT 2C receptor in the context of other edited amino acids might produce different conformations of this critical arginine, allowing some edited isoforms to couple more efficiently than others. The expression of more than 12 predominant isoforms in the human brain further suggests that unique degrees of coupling ability might be produced with each distinct edited receptor.
The consequences of RNA editing within 5-HT 2C R RNA are summarized in the model presented in Fig. 6, with A representing the 5-HT 2C-INI receptor and B representing the 5-HT 2C-VGV receptor isoform. In this model, the 5-HT 2C-INI receptor has a much greater capacity to isomerize to the active R* conformation (represented in boldface type and described by the constant J). The R* form of the receptor interacts efficiently with G proteins in the absence of agonist (H); upon agonist binding, the response is enhanced. For edited 5-HT 2C-VGV receptors, however, the J constant is much smaller, reflecting a decreased spontaneous ability of these receptors to convert to the coupling-competent state. It is only upon binding of agonist that these receptors achieve the ability to significantly convert to a form able to elicit efficient G protein interaction. Our current working hypothesis is that the edited receptor has a lower value for the J constant, while the M constant, an index of the intrinsic ability of an isoform to couple to G proteins, remains relatively constant.
In summary, the current results demonstrate that the 5-HT 2C-INI receptor isoform exhibits a greater level of constitutive activity than does the edited 5-HT 2C-VGV receptor isoform and therefore possesses a greater propensity to spontaneously isomerize to the R* state. The differential degrees of constitutive activity and the accompanying secondary alterations in agonist potency and G protein coupling could have important implications for the physiological effects of 5-HT. Brain regions that contain the nonedited receptor would be more sensitive to 5-HT that may be tonically released at distinct sites, possibly generating considerable noise in the system. In addition, the magnitude of signal produced by firing of presynaptic serotonergic terminals may be reduced because of a high basal tone at the constitutively active 5-HT 2C-INI receptor. Region-specific generation of edited isoforms, coupled with the possibility that individuals have different editing patterns, suggests that the repertoire of expressed edited 5-HT 2C Rs might determine the response to endogenous serotonin as well as control the signal:noise ratio at central serotonergic synapses.