Respective Roles of Calcitonin Receptor-like Receptor (CRLR) and Receptor Activity-modifying Proteins (RAMP) in Cell Surface Expression of CRLR/RAMP Heterodimeric Receptors*

Receptor activity modifying proteins RAMP1, RAMP2, and RAMP3 are responsible for defining affinity to ligands of the calcitonin receptor-like receptor (CRLR). It has also been proposed that receptor activity-modifying proteins (RAMP) are molecular chaperones required for CRLR transport to the cell surface. Here, we have studied the respective roles of CRLR and RAMP in transporting CRLR/RAMP heterodimers to the plasma membrane by using a highly specific binding assay that allows quantitative detection of cell surface-expressed CRLR or RAMP in the Xenopus oocytes expression system. We show that: (i) heterodimer assembly is not a prerequisite for efficient cell surface expression of CRLR, (ii)N-glycosylated RAMP2 and RAMP3 are expressed at the cell surface and their transport to the plasma membrane requiresN-glycans, (iii) RAMP1 is not N-glycosylated and is transported to the plasma membrane only upon formation of heterodimers with CRLR, and (iv) introduction ofN-glycosylation sites in the RAMP1 sequence (D58N/G60S, Y71N, and K103N/P105S) allows cell surface expression of these mutants at levels similar to that of wild-type RAMP1 co-expressed with CRLR. Our data argue against a chaperone function for RAMP and identify the role of N-glycosylation in targeting these molecules to the cell surface.

Functional properties of G protein-coupled receptors (GPCR) 1 can be altered upon formation of receptor homodimers or heterodimers (1). In 1998, McLatchie et al. (2) demonstrated that ligand binding of a GPCR, namely calcitonin receptor-like receptor (CRLR), depends on association with receptor activitymodifying proteins (RAMP). CRLR, originally identified as an orphan GPCR, was shown to form a high affinity receptor to calcitonin gene-related peptide (CGRP), when associated with RAMP1, or, to specifically bind adrenomedullin (AM), when associated with RAMP2 or RAMP3. RAMP are type I transmembrane proteins that share ϳ30% of amino acids identity and a common predicted topology with the short cytoplasmic C termini, one transmembrane domain, and the large extracellular N termini that are responsible for the acquisition of the RAMP-specific receptor phenotypes (3)(4)(5). The discovery of RAMP has demonstrated that ligand-GPCR interaction may require accessory proteins that are either directly involved in the formation of receptor-binding pockets or participate in the acquisition by their partner receptor of specific conformations required for ligand binding. More recently, a second GPCR, namely calcitonin receptor (CTR), was demonstrated to form heterodimeric complexes with RAMP. CTR/RAMP1 and CTR/ RAMP3 heterodimers revealed the pharmacological profiles of receptors specific to amylin (6,7). This finding, as well as the ubiquitous and abundant RAMP expression across tissues, suggested that RAMP might be involved in defining functional properties of different GPCRs. However, the direct evidence for RAMP interaction with GPCRs other than CRLR and CTR is still lacking.
The molecular mechanisms that guide the processes of assembly, intracellular trafficking, and cell surface expression of RAMP with their partner receptors remain poorly understood. Several studies performed in HEK293T cells suggested that CRLR-RAMP heterodimerization is required for CRLR plasma membrane targeting (2,3). Therefore, it was proposed that RAMP play the role of molecular chaperones for CRLR cell surface expression. However, Kuwasako et al. (8) and Hilairet et al. (9) have respectively reported that when expressed alone, a significant number of RAMP3 or CRLR reach the cell surface of HEK293T cells. Buhlmann et al. (10) have demonstrated that in the embryonic kidney TSA cells CRLR is efficiently expressed at the cell surface with or without RAMP. This apparent discrepancy was attributed to endogenous CRLR and RAMP expression detected in several of the commonly used mammalian expression systems, including HEK293T cells (2,8,9). However, the fact that cell surface expression of CTR, a second established RAMP partner, does not require RAMP co-transfection suggested that the proposed hypothesis requires further confirmation.
In the present study, we assessed the respective roles of CRLR and RAMP in cell surface expression of CRLR/RAMP receptors by using a binding assay allowing quantitative detection of CRLR or RAMP at the cell surface. This assay is based on binding of an iodinated antibody of known specific activity to an epitope placed in the extracellular parts of CRLR and RAMP. For these experiments, we took advantage of the Xenopus oocyte expression system in which CRLR/RAMP heterodimers are expressed as fully functional receptors for CGRP or AM (2). Importantly, in this expression system, a significant interaction with endogenously expressed CRLR and/or RAMP can be excluded (see below). Using this assay, we show that assembly in heterodimers is not a prerequisite for cell surface expression of CRLR, whereas RAMP are efficiently transported to the plasma membrane only when N-glycosylated. Transport of nonglycosylated RAMP can, however, be restored by coexpression with CRLR. This was also true for co-transport to the cell surface of RAMP with CTR but not with several other GPCR tested, thus providing evidence: (i) for the primary role of GPCRs in trafficking of GPCR/RAMP heterodimers to the cell surface, (ii) for the critical role of N-glycosylation for trafficking RAMP to the cell surface and, (iii) for a selective association of RAMP with CRLR and CTR.

EXPERIMENTAL PROCEDURES
cDNA Constructs and FLAG Epitope Insertion-Mouse cDNAs for RAMP1 (GenBank TM accession number NP_005846), RAMP2 (Gen-Bank TM accession number NP_005845), RAMP3 (GenBank TM accession number NP_005847), CRLR (GenBank TM accession number NP_061252), receptor to parathyroid hormone and parathyroid hormone-related peptide (PTH/PTHrP-R; IMAGE clone 4238969), and receptor to glucagon (GluR; IMAGE clone 4241407) and rat cDNAs for CTR, V2 type receptor to vasopressin (V2R), and V1a type receptor to vasopressin (V1aR) were used in this study. The corresponding proteins were tagged with the FLAG reporter octapeptide DYKDDDDK that is recognized by the anti-FLAG M2 mouse monoclonal antibody (Sigma). The sites for FLAG insertion were chosen in the extracellular parts of the proteins, as determined, according to the predicted membrane topology of these proteins (Proscan software; www.expasy.org/tools). For CRLR, RAMP1, RAMP2, and RAMP3, the FLAG epitope was inserted between the residues flanking the predicted signal peptide cleavage sites, so that in the processed proteins the FLAG epitope undertakes the most N-terminal position of the proteins (SignalIP software; www. cbs.dtu.dk/services/SignalP). The predicted positions for signal peptide cleavage are flanked by amino acids Ala 26 and Cys 27 for RAMP1, Ala 44 and Ser 45 for RAMP2, Gly 26 and Cys 27 for RAMP3, Gln 24 and Ala 25 for CTR, and Ala 22 and Glu 23 for CRLR. The cleavage of the signal peptides in FLAG-tagged RAMP1 and RAMP3 proteins was tested in in vitro protein translation experiments that were performed in the absence or presence of canine pancreatic microsomal membranes (Promega) with the following treatment of the RAMP3 protein with the endoglycosydase F. After separation on SDS-PAGE (13%), a decrease of 1-2 kDa in the molecular masses of RAMP1 and RAMP3 synthesized in the presence of canine pancreatic microsomal membranes was observed, thus indicating that these proteins are correctly processed in the presence of FLAG (data not shown).

Binding Analysis of CRLR and RAMP Cell Surface Expression-
Complementary RNAs for RAMP and GPCRs were synthesized in vitro using SP6 polymerase. Equal amounts of RAMP and GPCR cRNAs were injected into Xenopus oocytes (10 ng of total cRNA/oocyte). Injected oocytes were kept into modified Barth solution containing 88 mM NaCl, 1 mM KCl, 2.4 mM NaHCO 3 , 0.8 mM MgSO 4 , 0.3 mM Ca(NO 3 ) 2 , 0.4 mM CaCl 2 , 10 mM Hepes-NaOH (pH 7.2). The cell surface expression of CRLR, as well as wild-type and mutant FLAG-tagged RAMP, was quantitatively determined by specific binding of [ 125 I]M2IgG 1 iodinated anti-FLAG M2 antibody (11). The anti-FLAG M2 antibody was iodinated using the Iodo-Beads iodination reagent (Pierce) and carrier-free Na 125 I (Hartmann), according to the Pierce protocol. Iodinated antibody had a specific activity of 2-10 ϫ 10 17 cpm/mol. Binding of the iodinated antibody to oocytes expressing the FLAG-tagged CRLR or RAMP was determined 24 -48 h after the cRNA injection, as described (11). Specific binding was calculated as the difference of the binding between the oocytes injected with FLAG-tagged cRNAs and the noninjected oocytes.
Determination of EC 50 Values of CRLR/RAMP Heterodimers in the Xenopus Oocytes-Combinations of wild-type and/or FLAG-tagged RNAs were injected into Xenopus oocytes (10 ng of total cRNA/oocyte). Injected oocytes were kept into modified Barth solution for 20 h. The EC 50 values were determined by measurements of CGRP-and AMgenerated Cl Ϫ currents generated by cystic fibrosis transmembrane regulator (CFTR), a cAMP-activated chloride channel, according to Chraibi et al. (12). Current was measured under two-electrode voltage clamp at a potential oscillating between Ϫ40 mV and Ϫ80 mV. Chloride conductance was then calculated.
Immunoprecipitation-Injected with wild-type or mutant FLAGtagged GPCRs and/or RAMP, the Xenopus oocytes were incubated overnight in modified Barth solution containing 1.0 mCi/ml [ 35 S]methionine (PerkinElmer Life Sciences). The microsomes were prepared as described by Geering et al. (13), and the immunoprecipitations were performed with anti-FLAG M2 antibody under nondenaturing conditions, resolved by 8 -13% SDS-PAGE, and revealed by fluorography.

RESULTS
To assess the respective roles of CRLR and RAMP in cell surface expression of CRLR/RAMP heterodimers, we have developed a quantitative assay based on the binding of 125 Ilabeled M2 anti-FLAG monoclonal antibody directed against a FLAG reporter epitope introduced into the extracellular N termini of the mouse CRLR, RAMP1, RAMP2, and RAMP3 proteins (see "Experimental Procedures"). To ensure that the insertion of a FLAG epitope did not significantly change the function of CRLR/RAMP heterodimers, we measured the CGRP-and AM-stimulated currents generated by the CFTR, a cAMP-activated chloride channel. CFTR was co-injected in the Xenopus oocytes with either wild-type or FLAG-tagged CRLR and/or RAMP, and the effector concentrations for half-maximal response (EC 50 ) for CGRP and AM were determined. As shown in Fig. 1A, the CRLR-FLAG/RAMP1wt and CRLRwt/RAMP1-FLAG heterodimers have the EC 50 values of 28 Ϯ 9 pM (n ϭ 4) and 56 Ϯ 10 pM (n ϭ 5), respectively, not different from that observed for the CRLRwt/RAMP1wt heterodimers (67 Ϯ 17 pM, n ϭ 5). Similarly, the presence of FLAG, did not change the EC 50 values for AM (Fig. 1B) of the CRLRwt/RAMP2-FLAG (477 Ϯ 61 pM, n ϭ 4) and CRLRwt/RAMP3-FLAG (653 Ϯ 419 pM, n ϭ 3) heterodimers, as compared with the CRLRwt/ RAMP2wt (339 Ϯ 47 pM, n ϭ 3) and CRLRwt/RAMP3wt (244 Ϯ 151 pM, n ϭ 4) heterodimers, respectively. Importantly, the oocytes injected with CRLR alone have no detectable CGRP-or AM-induced Cl Ϫ currents (data not shown), thus ruling out a significant interaction with endogenously expressed RAMP. Also, the absolute CFTR current values were not significantly different between the combinations of either wild-type or FLAG-tagged CRLR and RAMP (data not shown), indicating that neither ligand affinities nor efficiency of membrane targeting are modified by FLAG insertion.
The respective roles of CRLR and RAMP in transport to the cell surface of CRLR/RAMP heterodimers were assessed in two sets of experiments in which we compared the cell surface expression of CRLR, RAMP1, RAMP2, and RAMP3 expressed either individually or in combinations. In the first group, the cell surface expression of FLAG-tagged CRLR expressed with or without wild-type RAMP was compared. As shown in Fig.  2A, CRLR is efficiently expressed at the cell surface without RAMP (1.72 Ϯ 0.23 fmol/oocyte, n ϭ 85), and co-expression with either RAMP1, RAMP2 or RAMP3 did not change the cell surface expression level of CRLR (1.56 Ϯ 0.22 fmol/oocyte (n ϭ 54) for CRLR/RAMP1, 1.97 Ϯ 0.34 fmol/oocyte (n ϭ 56) for CRLR/RAMP2, and 1.69 Ϯ 0.22 fmol/oocyte (n ϭ 89) for CRLR/ RAMP3, not significantly different from CRLR expressed alone). In the second group, the cell surface expression of FLAG-tagged RAMP1, RAMP2, and RAMP3 expressed with or without wild-type CRLR was compared. As shown in Fig. 2B, RAMP2 and RAMP3 are transported to the cell surface without CRLR (0.61 Ϯ 0.16 fmol/oocyte (n ϭ 55) and 0.51 Ϯ 0.10 fmol/oocyte (n ϭ 86), respectively), whereas RAMP1 cell surface expression was at a nearly undetectable level (0.14 Ϯ 0.19 fmol/oocyte (n ϭ 46)). Co-expression of RAMP with CRLR leads to efficient transport to the plasma membrane of RAMP1 (2.17 Ϯ 0.39 fmol/oocyte (n ϭ 37), p Ͻ 0.001 as compared with RAMP1 expressed alone) and increases to some extent but significantly RAMP2 and RAMP3 cell surface expression (1.39 Ϯ 0.40 fmol/oocyte (n ϭ 57), p Ͻ 0.05 and 0.75 Ϯ 0.07 fmol/oocyte (n ϭ 84), p Ͻ 0.05 as compared with RAMP2 and RAMP3 expressed alone, respectively). These experiments show that: (i) assembly in heterodimers is not a prerequisite for efficient cell surface expression of CRLR, (ii) RAMP1 is transported to the plasma membrane only upon formation of heterodimers with CRLR and, (iii) individually expressed RAMP2 and RAMP3 are transported to the plasma membrane, but their maximal cell surface expression requires association with CRLR.
Interestingly, the cell surface expression of individually expressed RAMP2 and RAMP3 correlated with the presence of N-glycosylation consensus sites in known RAMP2 and RAMP3 amino acid sequences from different species (mouse, human, and rat), whereas for RAMP1, the N-glycosylation consensus sites are not present, and the RAMP1 protein is intracellularly retained (5). Because the N-glycosylation has been shown to play a variety of roles, including facilitating protein trafficking to the plasma membrane (14), we hypothesized that N-glycosylation of RAMP2 or RAMP3 was the basis for the observed difference between RAMP1 and RAMP2 or RAMP3. To assess the role of N-glycosylation in RAMP2 and RAMP3 cell surface expression, we eliminated the four N-glycosylation consensus sites present in both RAMP2 and RAMP3 mouse proteins by site-directed mutagenesis (see "Experimental Procedures") and compared the cell surface expression of wild-type and mutant RAMP2 and RAMP3. As shown on Fig. 3A, the RAMP2-⌬4N N-glycosylation mutant is expressed at the cell surface at a nearly undetectable level (0.15 Ϯ 0.02 fmol/oocyte (n ϭ 95) for RAMP2-⌬4N as compared with 1.73 Ϯ 0.14 fmol/oocyte (n ϭ 91) for wild-type RAMP2, p Ͻ 0.001). Importantly, co-expression of RAMP2-⌬4N with CRLR completely restored the cell surface expression of this mutant (Fig. 3A). Similarly, RAMP3-⌬4N N-glycosylation mutant is not transported to the cell surface (0.09 Ϯ 0.02 fmol/oocyte (n ϭ 85) for RAMP3-⌬4N as compared with 1.93 Ϯ 0.16 fmol/oocyte (n ϭ 85) for wild-type RAMP3, p Ͻ 0.001), and co-expression of this mutant with CRLR partially restored its cell surface expression (Fig. 3B). The cell surface expression of single N-glycosylation mutants of RAMP2 or RAMP3 was significantly decreased by more than 50% (data not shown). Immunoprecipitation of [ 35 S]methionine-labeled proteins (Fig. 3C) performed on oocytes injected with either wild-type or mutant RAMP confirmed that RAMP2 and RAMP3 are N-glycosylated. Importantly, wild-type and mutant RAMP are expressed at a similar level, thus indicating that the loss of cell surface expression of RAMP2-⌬4N and RAMP3-⌬4N was not due to decreased protein stability or degradation of the mutants. Thus, these experiments show that: (i) cell surface expression of both wild-type RAMP2 and RAMP3 requires the N-glycosylation, (ii) similarly to RAMP1, RAMP2-⌬4N and RAMP3-⌬4N mutants require CRLR for their cell surface expression, and (iii) co-transport to the cell surface expression of RAMP1, RAMP2-⌬4N, and RAMP3-⌬4N with their partner receptors can be used as a stringent test for GPCR-RAMP heterodimerization.
To further assess the possibility that N-glycosylation of RAMP2 and RAMP3 may be the basis of the observed difference between RAMP1 and RAMP2/RAMP3 cell surface expression, we took advantage that three of four of the N-glycosylation sites of RAMP2 and RAMP3 are within the partially conserved stretches of amino acid that are also present in RAMP1 (Fig. 4A). We therefore introduced N-glycosylation sites in the sequence of RAMP1 (D58N/G60S, Y71N, or K103N/ P105S) and examined their cell surface expression. As shown in Fig. 4B, the three mutants were able to traffic to the cell surface unlike wild-type RAMP1. For two mutants (RAMP1-Y71N and RAMP1-K103N/P105S), the level of cell surface expression of RAMP was equal or superior to that of the RAMP1/ CRLR heterocomplexes. We next verified that the mutated RAMP1s were indeed N-glycosylated. As shown in Fig. 4C, wild-type RAMP1 was, as expected, not glycosylated, whereas RAMP1(Y71N) and RAMP1(K103N/P105S) were. Interestingly, the RAMP1-D58N/G60S mutant was only partially Nglycosylated, and its cell surface expression level reached only ϳ40% of that of Y71N and K103N/P105S mutants (Fig. 4B). This experiment confirms the critical importance of N-glycosylation for cell surface expression of the RAMP proteins.
Because the existence of other RAMP partners has been demonstrated (CTR) or proposed (5,(15)(16)(17), we were interested to compare the effects of different GPCRs on RAMP expression at the cell surface. The aim of these experiments was 2-fold: (i) to check the selectivity of RAMP association to different GPCRs and (ii) to compare the role of other RAMP partners in trafficking to the cell surface of GPCR/RAMP heterodimers. We have chosen three receptors that share a high degree of amino acids homology with CRLR, namely CTR (ϳ65% homologous), PTH/ PTHrP-R (ϳ45% homologous), and GluR (ϳ45% homologous) and two GPCRs that do not exhibit the significant degree of homology with CRLR, namely, vasopressin V1aR and V2R. As shown in Fig. 5 (A-C), in addition to CRLR, only CTR was able to co-transport to the cell surface the RAMP1, RAMP2-⌬4N, and RAMP3-⌬4N, despite the similar levels of protein expression observed for all receptors (data not shown). These data, as well as previous observations that individually expressed PTH/ PTHrP-R, GluR, V1aR, and V2R are efficiently targeted to the cell surface of the Xenopus oocytes (18 -21), suggested that RAMP are not associated with these receptors. As an additional test, we co-immunoprecipitated FLAG-tagged RAMP1 with coexpressed wild-type receptors, using the anti-FLAG M2 antibody. As shown in Fig. 5D, RAMP1 is physically associated only with CRLR and CTR but not with other receptors, thus confirming the data of cell surface binding analysis. These experiments show that among tested receptors RAMP selectively associates with CRLR and CTR and that, similar to CRLR, CTR is responsible for the efficient RAMP trafficking to the cell surface.

DISCUSSION
In 1998, McLatchie et al. (2) proposed that RAMP play the role of molecular chaperones required for CRLR cell surface expression and that association with RAMP results in differential N-glycosylation of CRLR. The fully glycosylated mature form of CRLR in CRLR/RAMP1 heterodimer and the coreglycosylated immature forms of CRLR in CRLR/RAMP2 and CRLR/RAMP3 heterodimers have been proposed as specific CGRP or AM receptors, respectively. However, further studies performed by Aldecoa et al. (22) and by Hilairet et al. (9) have demonstrated that differential N-glycosylation is not a prerequisite for ligand selectivity of CRLR/RAMP heterodimers (9,22), and Hilairet et al. (9) have proposed that CRLR/RAMP protein-protein interactions are instead responsible for expression of CGRP and AM receptor phenotypes (9). Similarly, several studies in which a significant amount of individually expressed CRLR or RAMP was detected at the cell surface have indicated that the chaperone function of RAMP should be reevaluated (8 -10).
A New Assay to Assess Cell Surface Expression of CRLR and RAMP-In this study, to assess the respective roles of CRLR and RAMP in cell surface expression of CRLR/RAMP heterodimers, we used a specific binding assay allowing quantitative detection of CRLR and RAMP at the cell surface. We and others have previously used a similar approach to study the process of assembly and trafficking to the plasma membrane of the ␣-␤-␥ subunits of the epithelial sodium channel (11,23) and of the ␣-␤ subunits of Na,K-ATPase (24). This binding assay, when performed in the Xenopus oocytes, has the following advantages: (i) CRLR/RAMP1, CRLR/RAMP2, and CRLR/ RAMP3 heterodimers are expressed in the Xenopus oocytes as the fully functional CGRP and AM receptors, respectively (2). (ii) The number of exogenously expressed plasma membrane proteins reaches as much as ϳ50% of the total number of proteins expressed at the oocyte surface, indicating that a significant interaction with endogenously expressed proteins can be excluded (25). For example, in our experiments, the CRLR expression level was ϳ2 fmol/oocyte, whereas Zampighi et al. (25) have estimated the total number of protein particles in the oocyte membrane as ϳ10 fmol/oocyte. This fact and the absence of CGRP-or AM-induced effects in CRLR-injected oo-cytes clearly demonstrate that endogenous RAMP, if present, were not able to significantly influence cell surface expression of exogenously expressed proteins. This favorably compares the Xenopus oocytes with most of the mammalian expression systems in which endogenous RAMP and/or CRLR are expressed at detectable levels (2). (iii) the Xenopus oocytes are character-  RAMP2-⌬4N (B), or RAMP3-⌬4N (C) cell surface expression was determined by the binding of 125 I-labeled anti-FLAG antibody. Shown are the means Ϯ S.E. of three to four experiments performed with 12 oocytes/experimental condition. **, statistical significance Ͻ 0.001. D, for co-immunoprecipitation the oocytes were metabolically [ 35 S]methionine-labeled for 12 h, and RAMP1-associated GPCRs were co-immunoprecipitated using anti-FLAG M2 antibody. Two major CRLR forms with apparent molecular masses of ϳ 57 and 62 kDa were co-immunoprecipitated with RAMP1, whereas a single CTR form with a molecular mass of ϳ58 kDa was detected.
ized by a stringent cellular quality control system that is responsible for the correct assembly, folding, intracellular trafficking, and cell surface expression of oligomeric membrane proteins (for a review see Ref. 26). In our experiments, the stringency of oocytes quality control was demonstrated by the fact that RAMP1 wt, RAMP2-⌬4N, and RAMP3-⌬4N N-glycosylation mutants were not able to reach the cell surface (Fig. 3). It is important to note that ⌬4N mutants were not degraded to any extent, suggesting that they folded properly. Our data rather suggest that the nonglycosylated mutants were still able to associate properly with their cognate partner, i.e. CRLR (presumably in the ER compartment), and that they are able to traffic normally to the membrane.
Critical Importance of RAMP N-Glycosylation for Cell Surface Expression-Our data suggest that (i) assembly in heterodimers is not a prerequisite for CRLR, RAMP2, and RAMP3 cell surface expression, whereas RAMP1 trafficking to the plasma membrane requires association with CRLR and (ii) CRLR is responsible for the maximal expression of CRLR/ RAMP heterodimers at the cell surface. Although statistically significant, this effect is relatively small in absolute terms (ϳ110% for RAMP2 and ϳ40% for RAMP3). Our findings partially differ from previous observations in HEK293T cells, in which RAMP have been proposed as molecular chaperones required for CRLR cell surface expression. We can offer the following explanations for such a difference. Previous estimations of CRLR cell surface expression in HEK293T cells were done by analysis of intracellular and submembrane distribution of GFP-tagged CRLR (8) or by an estimation of the fully glycosylated mature form of CRLR, considered to be cell surface-expressed receptors (9). These two methods do not allow precise cell surface quantification. Another approach was fluorescence-activated cell sorter analysis of Myc-tagged CRLR in HEK293T (2), but a recent report has shown that an important number of CRLR alone can reach the cell surface (9). This finding was attributed to the presence of a significant expression of endogenous RAMP in this cell line, a factor that can be excluded in the present study. Finally, Buhlmann et al. (10), using the Myc-tagged CRLR in the embryonic kidney TSA cells, have obtained a similar CRLR cell surface expression whether the receptor was expressed alone or in combination with RAMP1 or RAMP2.
We also demonstrate that cell surface expression of individually expressed RAMP2 and RAMP3 requires N-glycosylation of the proteins. In agreement with recent data by Gujer et al. (27) that efficient CRLR cell surface expression and ligand binding require receptor N-glycosylation, our results indicate that N-glycosylation may play an important role in CRLR/ RAMP heterodimer cell surface expression and receptor function. This is strongly supported by our ability to confer an efficient cell membrane trafficking to RAMP1 by just introducing a single N-glycosylation site in the sequence (Fig. 4). Only the RAMP1-(D58N/G60S) mutant was not fully competent to reach the surface in parallel with poor efficiency of that specific glycosylation site. We do not know the mechanism for this phenotype. It is, however, interesting to observe that the mutation is introduced just after a highly conserved cysteine residue. This is consistent with the finding that a localized folding event on the nascent chain, such as disulfide bond formation, which blocks access to the oligosaccharyl transferase, may be a determinant of glycosylation site usage (28). It will be interesting to examine the disulfide bond formation of RAMP and its consequence on glycosylation. 2 RAMP Heterodimerize with CRLR and RAMP but Have No Affinity for Several More Distantly Related GPCRs-We also tested several other GPCRs as potential partners for RAMP. The interest in these experiments came from the widely discussed possibility that pharmacological profiles of GPCRs, other than CRLR and CTR, may be dependent on RAMP association (5,(15)(16)(17). However, in the absence of information about the protein domains required for CRLR-and CTR-RAMP heterodimerization, the prediction for the possible partners from hundreds of GPCRs identified in the Human Genome Project (29) is a difficult task. CRLR and CTR, belong to Family B of GPCRs, which also includes receptors for secretin, pituitary adenylate cyclase polypeptide type I, vasoactive intestinal peptide, PTH/PTHrP, glucagon and glucagon-like peptides, and receptors for other hormones (for GPCR phylogenic tree see www.gpcr.org). Thus, together with CRLR and CTR, we tested for RAMP association the closely related PTH/PTHrP-R and GluR and Family B-unrelated V1aR and V2R. These experiments failed to find any significant interaction between RAMP and receptors other than CRLR and CTR, thus arguing against a general role of RAMP in the regulation of GPCRs function. Importantly, similarly to CRLR, CTR was able to co-transport to the cell surface expression the RAMP1 and RAMP2-⌬4N and RAMP3-⌬4N mutants, thus indicating that co-transport of a GPCR with RAMP to the cell surface can be used as a common criterion for the formation of GPCR/RAMP heterodimers.