Introduction

Top Abstract Introduction Materials and Methods Results Discussion References

Prostacyclin (PGI₂), the major product of arachidonic acid in vascular endothelium (Habib et al., 1997), transduces its potent antiplatelet, vasodilator, and antiproliferative actions via a G protein-coupled receptor (GPCR), the prostacyclin receptor (IP). Human IP (hIP) is coupled to stimulation of both adenylyl cyclase and phospholipase C (PLC) (Boie et al., 1994; Namba et al., 1994; Smyth et al., 1996). Agonist-induced phosphorylation and desensitization of hIP is mediated primarily by protein kinase C-dependent receptor phosphorylation (Smyth et al., 1996, 1998), whereas activation-dependent receptor sequestration may be phosphorylation independent and proceed via at least two distinct pathways (Smyth et al., 2000)

Most GPCRs contain one or two potential N-linked glycosylation sites (Asp-X-Ser/Thr, where X represents any amino acid except a proline (Gavel and Heijne, 1990), located at the N terminus and/or extracellular loops (Marshall et al., 1972; Bause et al., 1983). The carbohydrate moieties of glycoproteins are generally are believed to be important for intracellular trafficking, stability, secretion, and/or protein folding, enzymatic activity, and additional structural functions (Frost et al., 1991; Rodriguez et al., 1995; Ray et al., 1998; Walsh et al., 1998; Zhou and Tai, 1999; Boer et al., 2000; Kataoka et al., 2000). All eicosanoid receptors that have been cloned so far (prostaglandin D₂, prostaglandin F₂, PGI₂, two thromboxane A₂ (TP) and four prostaglandin E2 (EP) (Coleman et al., 1994), have N-glycosylation sites. This modification seems to play important functional roles in the GPCR superfamily although the impact varies from receptor to receptor. Glycosylation of TP and EP3 are necessary both for correct sorting to the plasma membrane and for normal ligand binding (Huang and Tai, 1998; Walsh et al., 1998). In contrast, whereas plasma membrane localization of glycosylation-deficient -adrenergic receptors (Rands, 1990) or EP3 receptors (Boer et al., 2000) was reduced, ligand binding and signal transduction were preserved.

Two potential N-glycosylation sites (N⁷ and N⁷⁸) are located in the hIP's N terminus and first extracellular loop, respectively. We previously demonstrated that hIP is expressed as a glycoprotein in HEK 293 cells (Smyth et al., 1996); however, the functional importance of this receptor modification has not been examined. In this study, we sought to determine the functional importance of N-glycosylation of hIP. We used site-directed mutagenesis of the N⁷ and N⁷⁸ residues of the hIP to explore the significance of N-glycosylation for hIP function. Our results indicated that N-glycosylation of hIP, especially at N⁷⁸, is important for membrane expression, ligand binding, and signal transduction.

	Materials and Methods

Top Abstract Introduction Materials and Methods Results Discussion References

(R)-Phycoerythrin and Cy3-conjugated AffiniPure F(ab')2 fragment goat anti-mouse IgG (H+L) were from Jackson ImmunoResearch (West Grove, PA). Anti-human golgin-97 was from Molecular Probes (Eugene, OR). PEGFP-N3 was obtained from CLONTECH Laboratories, Inc (Palo Alto, CA). Poly-lysine and tunicamycin were purchased from Sigma (St. Louis, MO). All cell culture reagents, G418, Albumax were from Invitrogen (Carlsbad, CA). N-[1-(2,3-Dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate, FuGENE transfection reagent, complete protease inhibitor tablets, and 4-nitrophenyl phosphate were obtained from Roche Biochemicals (Indianapolis, IN). AG 1-X8 resin (formate form) was purchased from Bio-Rad (Hercules, CA). Dr. Kathleen Metters (Merck Frosst, Quebec, Canada) generously donated the hIP cDNA. PNGase F was from New England BioLabs Inc (Beverly, MA). Oligonucleotides were from Cybersyn (Philadelphia, PA).

Generation of Green Fluorescent Protein-hIP Fusion Protein and Site-Directed Mutagenesis. The green fluorescent protein (GFP) was fused to the C-terminal end of the hemagglutinin (HA) tagged hIP (HAhIP) to generate the construct HAhIP-GFP as described previously (Smyth et al., 1996; Smyth et al., 2000). Primers were designed to mutate N⁷ to Q⁷ (sense, 5' GCG GAT TCG TGC AGG CAG CTC ACC TAC GTG CGG) and N⁷⁸ to Q⁷⁸ (sense: 5' GTG GCC TAT GCG CGC CAG AGC TCC CTG CTG GGC), in HAhIP-GFP. Mutagenesis was performed using the MORPHTM site-specific plasmid DNA Mutagenesis kit (Eppendorf-5 Prime, Inc., Boulder, CO). The mutated receptors were termed N⁷-Q⁷,N⁷⁸-Q⁷⁸ and N⁷,N⁷⁸-Q⁷,Q⁷⁸. Mutagenesis was confirmed by DNA sequencing.

Cell Culture and Transfection. COS-7 and HEK 293 cells (American Type Culture Collection; Manassas, VA) were maintained in Dulbecco's modified Eagle's medium supplemented with 10% heat-inactivated fetal bovine serum, 50 U/ml penicillin, 50 µg/ml streptomycin, 25 mM HEPES, and 2 mM glutamine. COS-7 cells were transfected transiently with FuGENE transfection reagent in six-well cell culture plates. For stable transfections, HEK cells were seeded at 1.5 × 10⁶ cells/100-mm dish and, the next day, transfected with 10 µg/dish DNA by liposome-mediated transfer (N-[1-(2,3-dioleoyloxy)propyl]-N,N,N-trimethylammonium methylsulfate), as described previously (Smyth et al., 1996, 1998).

Western Blotting. Cells were lysed (radioimmuniprecipitation assay: 50 mM Tris/5 mM EDTA, pH 8.0, containing 150 mM NaCl, 1% Nonidet P40, 0.1% SDS, 0.5% deoxycholic acid, 1 tablet/50 ml complete protease inhibitor cocktail), drawn through a 23-gauge needle six times and centrifuged at 14,000 rpm. Proteins were resolved on 8% SDS polyacrylamide gels and transferred to nitrocellulose. Receptors were visualized by treating immunoblots, first blocked with 5% nonfat milk in Tris-buffed saline/Tween 20 (50 mM Tris/250 mM NaCl, pH 7.6 containing 1% Tween 20), for 2 h at room temperature, followed by incubation with anti-HA (1:1500 dilution) and then incubation with a horseradish peroxidase-conjugated anti mouse IgG (1:5000 dilution). Antigen-antibody complexes were visualized with the use of enhanced chemiluminescence.

Deglycosylation of IP. Removal of asparagine-linked mannose, hybrid and complex oligosaccharides with PNGaseF was carried out according to the manufacturer's instructions. Membrane proteins (30 µg) were denatured in 0.5% SDS, 1% -mercaptoethanol at 100°C for 10 min. Deglycosylation was carried out for 60 min at 37°C in 50 mM sodium phosphate buffer, pH 7.5, containing 1% Nonidet P-40, and reactions were stopped by the addition of SDS-polyacrylamide gel sample buffer.

Immunofluoresence Microscopy. Cells were seeded into lysine coated two chamber slides (Nalge Nunc International Corp, Naperville, IL). One to two days later, cells were fixed and permeabilized in 100% methanol at 80°C for 7 min. The Golgi apparatus was stained with anti-human golgin-97 (1 µg/ml, 60 min) followed by Cy3-conjugated anti-mouse (1:1000 dilution, 60 min). GFP and Cy3 fluorescence was examined by confocal microscopy within 1 week.

Radioligand Binding. Membrane proteins were prepared from confluent 100-mm dishes as follows. Cells were washed once with phosphate-buffered saline and scraped into 20 mM Tris, pH 7.4, containing 4 mM EDTA, and complete protease inhibitor. Cells were lysed and drawn although a 23-gauge needle 10 times, and membrane fractions were collected by centrifugation at 115,000g for 1 h at 4°C. The resulting pellet was resuspended in the same buffer and stored at 80°C for further use. Radioligand binding studies were carried out using membrane proteins (50 µg/reaction) in 10 mM HEPES, pH 7.4, contain 10 mM MnCl as described previously (Smyth et al., 1996). Nonspecific binding was measured in the presence of a 500-fold excess of unlabeled iloprost. Saturation binding data were analyzed using Prism 2.0 (Graphpad Software, San Diego, CA) to calculate K_d and B_max values for binding.

Flow Cytometry. Cells were detached from tissue culture plates with 1 mM EDTA in PBS and washed with PBS containing 1% BSA and 0.1% NaN₃. The detached cells were resuspended in PBS and 1% BSA containing anti-HA monoclonal antibody (or a control isotype IgG1) for 60 min, followed by (R)-phycoerythrin-conjugated AffiniPure F(ab')2 fragment goat anti-mouse IgG) for 30 min at 4°C. Cells were analyzed for fluorescence intensity by flow cytometry, after one further wash. Dead cells were detected by low forward- and right-angle scatter and excluded from the analysis.

Internalization of HahIP. Cells were seeded on 0.1 mg/ml lysine-coated, glass-bottomed plates (MatTek Corp., Ashland, MA), for confocal microscopy, or 24-well dishes for ELISA and, 24 to 48 h later, treated with iloprost (1 µM). Real time internalization of receptors was examined by confocal microscopy of GFP. Cell surface localization of HAhIP-GFP was quantified by ELISA after fixation of the cells (4% paraformaldehyde in PBS, 4°C, 10-15 min), as described previously (Smyth et al., 2000).

Measurement of cAMP. Cells, grown to confluence in 12-well plates coated with 0.1 mg/ml lysine, were treated with iloprost (10 min at 37°C). Reaction were terminated by aspiration and cAMP was extracted with ice-cold 65% ethanol for 30 min. Samples were dried under vacuum and reconstituted in assay buffer, and cAMP was quantified by radioimmunoassay, as described previously (Smyth et al., 1996, 1998). Dose-response data were analyzed using the CalcuSyn dose-effect analysis program (Biosoft, Milltown, NJ) to calculate EC₅₀ values.

Inositol Phosphate Production. Cells, grown to 70 to 80% confluence in 12-well plates coated with 0.1 mg/ml lysine, were labeled overnight with 2 µCi/ml [myo-³H]inositol in DMEM (without inositol) containing 0.5% Albumax. Thirty minutes before stimulation, cells were treated with 20 mM LiCl at 37°C. After stimulation for 10 min at 37°C, the reactions were terminated by aspiration. Total inositol phosphates were extracted, recovered, and quantified as described previously (Smyth et al., 1996, 1998).

Statistical Analysis. Data were compared by Student's t test or analysis of variance for multiple comparisons. P value of <0.05 was considered significant.

	Results

Top Abstract Introduction Materials and Methods Results Discussion References

Expression and Characterization of N-Glycosylation HAhIP-GFP and Mutant Receptors in COS-7 Cells and HEK Cells. We have previously shown that HAhIP is glycosylated in stably transfected HEK 293 cells (Smyth et al., 1996) and that the addition of the HA (Smyth et al., 1996) and the GFP tags (Smyth et al., 2000) does not effect receptor signaling or localization. In the current study, COS-7 cells were transiently transfected with HAhIP-GFP and the receptor was examined by Western blotting. Wild-type hIP resolved as three bands (Fig. 1A). Only the lowest band (68 kDa, which represents the deglycosylated receptor plus the 27-kDa GFP) remained after treatment of cells with tunicamycin (2 µg/ml), an inhibitor of glycosylation, indicating that, similar to the HAhIP in HEK 293 cells, HAhIP-GFP was glycosylated in COS-7 cells. Thus, hIP glycosylation is not confined to a particular cell line.

View larger version (53K): [in this window] [in a new window]   Fig. 1.   Expression of HAhIP-GFP and mutant receptors in COS-7 and HEK cells. Cells were lysed and 30 µg total protein was loaded onto 8% polyacrylamide gels, transferred to polyvinylidene difluoride membranes and immunoblotted with anti-HA antibody. Molecular masses are in reported in kiloDaltons. A, transient expression of HAhIP-GFP in COS-7 cells. Lane 1, wild-type HAhIP-GFP; lane 2: wild-type HAhIP-GFP + tunicamycin (2 µg/ml); lane 3: N7-Q7; lane 4: N78-Q78; lane 5: N7, N78-Q7,Q78. B, screening of the HEK 293 cell lines stably transfected with mutant HAhIP-GFPs. 1, wild-type HAhIP-GFP cell; 2, a-e, N7-Q7 clones; 3, f-i, N78-Q78 clones; 4, j-m, N7,N78-Q7,Q78 clones. C, analysis of PNGase F digested mutant HAhIP-GFPs. Lanes 1-4, HAhIP-GFP, N7-Q7, N78-Q78, N7, N78-Q7, Q78; lanes 5-8: HAhIP-GFP, N7-Q7, N78-Q78, N7, N78-Q7, Q78 digested with PNGase F.

Role of N-Linked Carbohydrates in IP Cell Surface Expression. Analysis of GFP localization by confocal microscopy demonstrated that, similar to the wild-type HAhIP-GFP, each of the three mutant receptors were localized to the plasma membrane (Fig. 2A). However, the relative intensity of membrane localization appeared to be reduced in each of the mutated cell lines, compared with the wild-type. In addition, dense cytoplasmic localization of receptor, which colocalized with a marker for the Golgi apparatus, was evident in each mutant cell lines (Fig. 2B) demonstrating retention of the glycosylation deficient mutants in the Golgi. Colocalization, which was minimal in wild-type cells (Fig. 2B, a) was evident when glycosylation at N⁷ was absent (Fig. 2B, b) and more extensive when N⁷⁸ was mutated (Fig. 2B, c and d).

View larger version (80K): [in this window] [in a new window]   Fig. 2.   Confocal microscopy HEK 293 cells expressing mutant HAhIP-GFPs. a, wild-type HAhIP-GFP cells; b, N7-Q7 cells; c, N78-Q78 cells; d, N7, N78-Q7, Q78 cells. A, arrows indicate dense cytoplasmic localization of the GFP-tagged receptors. B, cells were treated with anti-golgin 97 to stain the Golgi apparatus. Colocalization of GFP-tagged receptors (green) with the Golgi (red) can been seen (yellow) when the data was merged. Arrows highlight areas of colocalization

View larger version (12K): [in this window] [in a new window]   Fig. 3.   Flow cytometric analysis of cell-surface expression of mutant HAhIP-GFPs. Cell surface HA expression was examined in each of the four cell lines, by flow cytometry, using the anti-HA monoclonal antibody or a control isotype. Specific fluorescent intensity (FI) of each cell line was corrected for the control FI. The percentage of specific FI relative to the wild-type HAhIP-GFP (100%) was calculated in each cell line. Data are the mean ± SEM of three independent experiments. *P < 0.05; **P < 0.01 compared with wild-type HAhIP-GFP. #P < 0.05 compared with N7-Q7 cells.

Role of IP Glycosylation in Ligand Binding. Ligand binding to HAhIP-GFP mutant receptors was examined in saturation binding experiments using membranes prepared from each of the four cell lines. Specific [³H]iloprost binding was evident in all cases except the double mutant (Fig. 4, Table 1), although radioligand binding was reduced in both N⁷-Q⁷ and N⁷⁸-Q⁷⁸ compared with the wild-type receptor. The extent of receptor glycosylation seemed to produce differential effects on radioligand binding. Thus, although HAhIP-GFP and N⁷-Q⁷ had similar K_d values for [³H]iloprost binding, the latter showed a much lower B_max (Table 1). In contrast, N⁷⁸-Q⁷⁸ cells had a low B_max value, similar to that of N⁷-Q⁷ cells, but the ligand affinity was greatly reduced compared with both the wild-type HAhIP-GFP and N⁷-Q⁷ cells. Finally, binding was not detected in N⁷,N⁷⁸-Q⁷,Q⁷⁸ cells.

View larger version (13K): [in this window] [in a new window]   Fig. 4.   Saturation binding of 3H-iloprost to membranes prepared from HAhIP-GFP mutant cell lines. Membrane proteins (50 µg) were incubated at 30°C for 30 min with [3H]iloprost. Nonspecific binding was quantified in the presence of 5 µM unlabeled iloprost. Data are from a representative experiment that was repeated with similar results. a, wild-type HAhIP-GFP cells; b, N7-Q7 cells; c, N78-Q78 cells.

Analysis of Intracellular Signaling. Human IP couples to activation of both adenylyl cyclase and phospholipase C in HEK 293 cells and these features are unaffected by the addition of the HA and/or GFP tags (Smyth et al., 1996, 2000). Intracellular signaling of HAhIP-GFP was dramatically altered by mutation of the N-glycosylation sites. Generation of cAMP was reduced in the N⁷-Q⁷ cell line. Iloprost induced a concentration-dependent increase in cAMP generation in HAhIP-GFP cells (EC₅₀ = 0.51 ± 0.1 nM, n = 3), which reached a maximum level of 150.4 ± 15 pmol cAMP/10⁶ cells at 5 nM iloprost (Fig. 5). In contrast, cAMP production at 5 nM iloprost was reduced to 75 ± 20 pmol cAMP/10⁶ cells (n = 3) in N⁷-Q⁷ cells and the concentration-response curve was shifted to the right, demonstrating an increase in EC₅₀. A low level of cAMP was generated through N⁷⁸-Q⁷⁸ at very high concentrations of iloprost only while N⁷,N⁷⁸-Q⁷,Q⁷⁸ did not transduce this signal (Fig. 5). Thus, similar to membrane localization and radioligand binding, iloprost-induced activation of adenylyl cyclase was reduced relative to the extent of receptor glycosylation. In contrast, iloprost-induced generation of inositol phosphate was absent in each of the three mutant cell lines regardless of their glycosylation state (Fig. 6).

View larger version (15K): [in this window] [in a new window]   Fig. 5.   Intracellular cAMP production in HEK 293 cells expressing mutant HAhIP-GFP. Cells were stimulated with increasing concentration of iloprost for 10 min at 37°C. Intracellular cAMP was extracted and quantified as outlined under Materials and Methods. The results were calculated as the fold increase over basal levels (cAMP production in the absence of agonist stimulation). Data are from one experiment that was repeated three to five times with similar results.

View larger version (23K): [in this window] [in a new window]   Fig. 6.   Total inositol phosphate production in HEK 293 cells expressing mutant HAhIP-GFP. Cells were treated with increasing concentrations of iloprost for 10 min at 37°C, and inositol phosphates extracted and quantified as outlined under Materials and Methods. The results were calculated as the fold increase over basal levels (inositol phosphate production in the absence of agonist stimulation). The inset shows inositol phosphate generation resulting from stimulation with 1 to 10 µM iloprost. Data are mean ± SEM from three to four independent experiments, each performed in duplicate.

Role of N-Linked Carbohydrates in IP Internalization Recently, we reported that the hIP is internalized in response to agonist stimulation (Smyth et al., 2000). Iloprost induced sequestration of HAhIP-GFP was evident only in the wild-type and N⁷-Q⁷ cell line and not in N⁷⁸-Q⁷⁸ or N⁷,N⁷⁸-Q⁷,Q⁷⁸ (Fig. 7), suggesting a role for glycosylation in receptor trafficking. Similar results were obtained by confocal microscopy (data not shown).

View larger version (12K): [in this window] [in a new window]   Fig. 7.   Internalization of cell surface wild-type and deglycosylated hIP. Cells were treated with 1 µM iloprost for 60 min. Surface HA expression was measured by ELISA and calculated as the percentage of absorbance in the absence of iloprost. Data are expressed as mean ± SEM from five independent experiments, each performed in triplicate. *P < 0.01 with reference to control.

	Discussion

Top Abstract Introduction Materials and Methods Results Discussion References

Many GPCRs, including the hIP (Smyth et al., 1996), have been shown to undergo N-linked glycosylation. However, the role played by this post-translational modification in GPCR biology is not clear. The hIP contains two asparagine residues, N⁷ and N⁷⁸, that are potential N-linked glycosylation sites. We examined the role of this modification in hIP cellular localization and function.

We have previously demonstrated that hIP, stably expressed in HEK 293 cells, is a glycoprotein and can be converted to the deglycosylated form with PNGase F (Smyth et al., 1996). In the current study, glycosylated forms of HAhIP-GFP were evident in both transiently transfected COS-7 cells and stably transfected HEK 293 cells, demonstrating that glycosylation of hIP is not cell specific. We generated three mutant HAhIP-GFP receptors in which N⁷ or N⁷⁸ was replaced with Q, alone or in combination. These mutant receptors were transiently expressed in COS-7 cells, stably expressed in HEK 293 cells and examined by Western blotting. Both glycosylation sites were used, because mutation of either asparagine residue led to reduced receptor glycosylation. However, the loss of glycosylation was more extensive in N⁷⁸ mutants, demonstrating the greater importance of this site for hIP glycosylation. Receptor glycosylation followed the sequence HAhIP-GFP > N⁷-Q⁷ N⁷⁸-Q⁷⁸ > N⁷,N⁷⁸-Q⁷,Q⁷⁸. Interestingly, glycosylation at N⁷ was not apparent in the N⁷⁸-Q⁷⁸ mutant, expressed in COS-7 cells (Fig. 1A, lane 3), suggesting some interaction between the two sites.

N-linked glycosylation has been demonstrated to play a role in subcellular distribution of many membrane-associated proteins (Nagai et al., 1997; Ray et al., 1998; Zhou and Tai, 1999). The cellular trafficking of a glycoprotein often depends on its correct glycosylation; carbohydrate moieties are thought to act as tags for the correct sorting of proteins to their subcellular target (Scheiffele et al., 1995; Gut et al., 1998; Benting et al., 1999). Non/partially glycosylated or improperly folded proteins are, in contrast, retained in the endoplasmic reticulum (Hammond and Helenius, 1995). Studies have demonstrated that the loss of the carbohydrate moieties associated with membrane receptors led to reduced membrane expression (Ray et al., 1998; Walsh et al., 1998; Boer et al., 2000; Kataoka et al., 2000). In agreement with these reports, cell surface expression of HAhIP-GFP was reduced in all three mutant cell lines (Fig. 3). Indeed, direct examination of mutant receptors, by confocal microscopy, indicated that each was retained in the cytoplasm, which colocalized with a Golgi marker, although membrane localization was still evident. Impaired membrane localization, together with increased retention in the Golgi, followed a similar sequence to glycosylation, namely HAhIP-GFP > N⁷-Q⁷ > N⁷⁸-Q⁷⁸  N⁷,N⁷⁸-Q⁷,Q⁷⁸. Thus, it seems that glycosylation of hIP plays a role in its correct translocation through the Golgi apparatus to the plasma membrane. However, membrane localization of each mutant receptor was evident, despite the progressive loss of glycosylation, suggesting that pathway does not exclusively determine the subcellular localization of HAhIP.

A direct functional role for GPCR glycosylation has been demonstrated (Frost et al., 1991; Huang et al., 1995; Rodriguez et al., 1995; Pang et al., 1999; Elleman et al., 2000; Kataoka et al., 2000; Nagai et al., 2000; Nagayama et al., 2000; Zhou et al., 2000). The partial membrane localization of HAhIP-GFP mutant receptors, despite major differences in glycosylation, allowed us to examine directly the role played by glycosylation in hIP ligand binding and signal transduction. It is generally thought that the seventh transmembrane domain, highly conserved among the prostanoid receptor family (Ushikubi et al., 1995) forms a critical portion of the ligand binding pocket for GPCR (Baldwin et al., 1995). However, reports have demonstrated that N-linked glycosylation of TP and the EP3 isoform of the PGE₂ receptor are necessary for optimal ligand binding (Huang et al., 1995; Huang and Tai, 1998; Walsh et al., 1998), although this was not the case for EP3 (Boer et al., 2000). As with surface localization, binding of iloprost to HAhIP-GFP was reduced in all three glycosylation deficient mutant receptors, albeit to varying extents (Table 1). However, the loss of cell surface receptor localization (20-40%, Fig. 3) could not account for the dramatic decrease in B_max for each mutant receptor (>90%, Table 1). Thus, the deficiency in glycosylation seems to affect directly ligand binding. Whereas the B_max value was reduced in cells expressing the N⁷-Q⁷ mutant, the K_d value remained unchanged, indicating that this receptor was capable of binding iloprost with high affinity. In contrast, a reduced B_max, together with an increased K_d, in N⁷⁸-Q⁷⁸ expressing cells, and loss of specific binding in N⁷,N⁷⁸-Q⁷,Q⁷⁸ cells, suggests a direct relationship between the extent of glycosylation and the receptors ability to ligate its agonist. Although it may be argued that the mutant cell lines simply expressed less receptor protein, compared with the wild-type controls, this is not likely to adequately explain the changes in ligand binding. Each receptor was transfected under the same conditions using the same CMV promoter and cell lines were selected for further work on the basis of equivalent expression by Western blotting and confocal microscopy (Figs. 1B and 2). It is more likely that the progressive reduction in ligand binding was coincident with reduced glycosylation. This is particularly evident when N⁷-Q⁷ was compared with N⁷⁸-Q⁷⁸; both cell lines have the same B_max value for iloprost binding, demonstrating equivalent receptor expression levels, but the loss of glycosylation at N⁷⁸ dramatically reduced binding affinity. These data suggest that glycosylation is required for optimal binding of ligand to hIP and that N⁷⁸ may play a more significant role in determining the efficiency with which the receptor binds its ligand.

We have previously shown that hIP can couple to both activation of adenylyl cyclase, through Gs, and increased generation of inositol phosphates, probably via Gq-PLC activation (Smyth et al., 1996). Not surprisingly, signaling through the glycosylation-deficient HAhIP-GFPs was reduced compared with the fully glycosylated control (Figs. 5 and 6). Coupling to cAMP generation followed the same pattern as glycosylation; N⁷-Q⁷ was more efficient compared with either N⁷⁸ mutant, the double mutant being the weakest of the three. It may be that the shift in the cAMP dose-response curve observed in the N⁷-Q⁷ mutant cells was simply a result of the 10-fold reduction in B_max (Table 1) and not as a result of changes in glycosylation per se. Although we cannot discount this possibility, the marked difference in cAMP production in N⁷⁸-Q⁷⁸ cells compared with N⁷-Q⁷ cells, despite their equivalent B_max (Fig. 5, Table 1), demonstrates a direct role of glycosylation at N⁷⁸ in signal transduction and implies a similar explanation for the impaired coupling of N⁷-Q⁷ to Gs. In contrast to cAMP production, activation of PLC is absent in all three cell lines, regardless of glycosylation efficiency, even when iloprost was used at concentrations of up to 10 µM. Thus, normal glycosylation of hIP is required for optimum Gs-coupling and, similar to ligand binding, glycosylation at N⁷⁸ seems to be more significant for signal transduction through this pathway. The absolute dependence of the Gq-PLC pathway on receptor glycosylation indicates that glycosylation may direct hIP coupling to multiple G proteins, raising the intriguing possibility that the cell may control the formation of different receptor-G protein pairs via changes in receptor glycosylation.

Agonist-induced sequestration of HAhIP from the cell membrane into the intracellular space proceeds in a phosphorylation-independent manner that seems to involve both dynamin-dependent and -independent pathways and is mediated through the C-terminal region of the receptor (Smyth et al., 2000). We examined receptor sequestration in each of the three glycosylation deficient HAhIP-GFPs. Iloprost stimulated receptor internalization in both the HAhIP-GFP and N⁷-Q⁷ cells, but not in N⁷⁸-Q⁷⁸ or N⁷,N⁷⁸-Q⁷,Q⁷⁸ cells. It may be argued the N⁷⁸ mutants do not internalize simply because of impaired ligand binding. However, both N⁷-Q⁷ and N⁷⁸-Q⁷⁸ cells display similar B_max values for iloprost binding, although the K_d value is reduced in the latter case (Table 1). Given the high dose of iloprost (1 µM) used in the sequestration experiments, both receptors would be expected to substantially bind ligand. However, sequestration is evident only in the case of N⁷-Q⁷, suggesting that glycosylation plays a direct role in hIP trafficking from the plasma membrane to the cytoplasm. The mechanism by which the absence of carbohydrate moieties at the N terminus can affect sequestration, mediated through the C terminus, is unclear and currently under investigation. Certainly, it seems that sequestration occurs independently of coupling to PLC; N⁷-Q⁷ did not couple to increased generation of inositol phosphates, but was sequestered in a manner similar to the that of wild-type receptor. Note that despite the ability of N⁷⁸-Q⁷⁸ to mediate some activation of cAMP production, treatment of N⁷⁸-Q⁷⁸ cells with a high concentration of iloprost did not elicit a sequestration response (Figs. 5 and 7). Sequestration of hIP may, therefore, be at least partially independent of signal transduction and require glycosylation at N⁷⁸.

In summary, our findings demonstrate that both potential N-linked glycosylation sites are used in the hIP N⁷⁸ to a greater extent. N-linked glycosylation of hIP, particularly at N⁷⁸, seems to play roles in membrane localization, ligand binding, signal transduction, and receptor sequestration. The mechanism through which glycosylation of hIP affects its function is not known. Recently, a new family of proteins known as receptor activity modifying proteins was described. These proteins physically interact with GPCRs (Leuthauser et al., 2000) and alter their glycosylation state, ligand binding, and signal transduction properties (Fraser et al., 1999). Although it is not known whether receptor activity modifying proteins exist for hIP, the interaction of such an accessory protein may depend on glycosylation. How glycosylation at the N terminus can affect events throughout the receptor, including at distal regions such as the C terminus, is not clear but suggests a further level of complexity through which the cell can control and alter its responses to GPCR activation.