Role of asparagine-linked oligosaccharides in protein folding, membrane targeting, and thyrotropin and autoantibody binding of the human thyrotropin receptor.

The amino-terminal ectodomain of thyrotropin (TSH) receptor (TSHR) is heavily glycosylated with asparagine-linked (N-linked) oligosaccharides. The present studies were designed to evaluate how acquisition and processing of N-linked oligosaccharides play a role in the functional maturation of human TSHR. A glycosylation inhibitor tunicamycin, which inhibits the first step of N-linked glycosylation (acquisition of N-linked oligosaccharides), and a series of mutant Chinese hamster ovary (CHO)-Lec cells defective in the different steps of glycosylation processing were used. Inhibition of acquisition of N-linked oligosaccharides by tunicamycin treatment in CHO cells stably expressing TSHR produced nonglycosylated TSHR, which was totally nonfunctional. In contrast, all of the TSHRs synthesized in mutant CHO-Lec1, 2, and 8 cells (mannose-rich, sialic acid-deficient, and galactose-deficient oligosaccharides, respectively) bound TSH and produced cAMP in response to TSH with an affinity and an EC50 similar to those in TSHR expressed in parental CHO cells (CHO-TSHR; sialylated oligosaccharides). However, Lec1-TSHR and Lec2-TSHR were not efficiently expressed on the cell surface, whereas the expression levels of Lec8-TSHR and CHO-TSHR were essentially identical. All of the TSHRs expressed in CHO-Lec cells cleaved into two subunits. Finally, anti-TSHR autoantibodies from Graves' patients interacted with all of the TSHRs harboring different oligosaccharides to a similar extent. These data demonstrate that acquisition and processing of N-linked oligosaccharides of TSHR appear to be essential for correct folding in the endoplasmic reticulum and for cell surface targeting in the Golgi apparatus. We also show that complex type carbohydrates are not crucially involved in the interaction of TSHR with TSH and anti-TSHR autoantibodies.

In human TSHR, the receptor ectodomain has asparaginelinked (N-linked) oligosaccharides, which represent 30ϳ40% of its molecular weight (15). Earlier studies with in vitro sitedirected mutagenesis suggest that, among six potential Nlinked glycosylation sites, the first (amino acid 77) and third (amino acid 113) glycosylation sites are essential for the cell surface expression and function of the full-length TSHR (16). The possibility, however, cannot be excluded in this paper that the amino acid substitutions introduced per se affected receptor function (9). Subsequent studies on the functional role of Nlinked oligosaccharides of TSHR have been performed with the truncated form of TSHR ectodomain expressed in mammalian cells and insect cells, yielding controversial data (15,17,18). However, it is difficult to interpret these data, because conformational integrity of TSHR protein has not been verified in these studies. Conformationally intact TSHR protein showing high affinity TSH binding has so far been expressed only in mammalian cells as the full-length receptor or the ectodomain fused to a membrane-anchoring peptide (19 -21).
The process of N-linked glycosylation of proteins has been well documented (see Fig. 1) (22). Briefly, a dolichol pyrophosphate precursor (Glc 3 Man 9 GlcNAc 2 ) is at first transferred to Asn side chain of Asn-X-Ser/Thr consensus sequence for Nlinked oligosaccharides in a nascent polypeptide in the endoplasmic reticulum. Processing is initiated by the removal of the three terminal glucose residues and at least one mannose residue in the endoplasmic reticulum, followed by transportation to Golgi apparatus, where mannose residues are further trimmed, and N-acetylglucosamine, galactose, and sialic acid residues are sequentially added. The newly synthesized glycoproteins then exit the Golgi and are transported to their final destination.
In the present studies, to further clarify the functional roles of N-linked oligosaccharides in human TSHR, we evaluated the functional properties of nonglycosylated TSHR and TSHRs with different types of N-linked oligosaccharides. Nonglycosylated TSHR was produced by treatment of TSHR expressed in Chinese hamster ovary (CHO) cells with tunicamycin, and TSHRs with different oligosaccharides were expressed in a series of mutant CHO-Lec cell lines harboring mutations in the distinct steps in carbohydrate processing (see Fig. 1) (23).
Transfection and Cell Culture-The human full-length TSHR cDNA (24) was ligated into the expression vector pCAGGS (25) to yield pCAG-TSHR. The plasmid was transfected into the cells described above with Lipofectin reagent (Life Technologies, Inc.). The cells were selected with 500 g/ml G418 (Geneticin; WAKO, Osaka, Japan) and pooled. Surviving clones (12 clones for CHO cells and 24 for CHO-Lec cells) were also isolated with cloning cylinders, and the clonal cell lines expressing the highest levels of the receptor, determined by [ 125 I]TSH binding (see below), were selected.
Tunicamycin Treatment-The cells were treated with 5 g/ml tunicamycin (WAKO) for 3 days. The medium was replaced daily to purge the cells of wild-type TSHR. Tunicamycin inhibits transfer of dolichol pyrophosphate precursor to Asn in the consensus sequence for N-linked glycosylation (Asn-X-Ser/Thr).
Immunoblot analysis of TSHR-Extraction of the crude cell membrane and immunoblotting were performed as described previously (28) with mouse anti-TSHR monoclonal antibodies A10 and A11 (26).
[ 125 I]TSH Binding and cAMP Measurements-125 I-labeled bovine TSH was from the TRAb kit (Cosmic, Tokyo, Japan). [ 125 I]TSH binding to intact cells and intracellular cAMP measurement with a cAMP radioimmunoassay kit (Yamasa, Tokyo, Japan) were performed as described previously (11). Unlabeled TSH used in TSH binding studies was of bovine origin (Sigma). [ 125 I]TSH binding to the detergent-solubilized extracts was performed with polyethylene glycol (PEG) precipitation previously described (29). Briefly, the detergent-solubilized extract was prepared by solubilization of the crude membrane (11) with 1% Triton X-100 at 4°C for 1 h. One-hundred l of detergent-solubilized extract from 100 g of crude membrane was incubated with 100 l of [ 125 I]TSH (ϳ15,000 cpm) for 1 h at 37 C, and solubilized extract and [ 125 I]TSH complexes were precipitated with PEG.
Thryoid Stimulatory Autoantobody (TSAb) and TSH Binding Inhibitory Autoantibody (TBIAb) Assays-Two Graves' sera positive for TSAb activity, another two Graves' sera positive for TBIAb, and a mixture of three normal sera were used. Preparation of crude IgG fraction with PEG and measurements of TSAb activity were performed as described previously (30). For the TBIAb assay, detergent-solubilized extract (100 l) from 100 g of crude membrane was incubated with 100 l of serum and 100 l of [ 125 I]TSH (ϳ15,000 cpm) for 1 h at 37 C, and solubilized extract and [ 125 I]TSH complexes were precipitated with PEG as described above. TBIAb activity was calculated as follows: [1 Ϫ (radioactivity with test IgG/radioactivity with normal IgG)] ϫ 100 (%).

Immunoprecipitation and Immunoblot of TSHR in the Clonal Cell Lines
Stably Expressing TSHR-Using clonal cell lines stably expressing the highest levels of receptors, we determined the molecular masses and glycosylation patterns of TSHR expressed in CHO cells treated with and without tunicamycin and in a series of CHO-Lec cells. In immunoprecipitation experiments (2-h pulse followed by 2-h chase), TSHR protein expressed in parental CHO cells (CHO-TSHR) was visualized as an ϳ95-kDa precursor and an ϳ120-kDa mature protein ( Fig. 2A, lane 1) as previously reported (27). Both decreased in size to ϳ85 kDa on N-glycosidase F treatment ( Fig. 2A, lane 3), a size compatible with that of the core polypeptide chain of TSHR (24) and also identical to that of nonglycosylated TSHR ( Fig. 2A, compare with lane 4; see below). The ϳ95-kDa protein was sensitive to endo H, indicating that this has high mannose type oligosaccharides, whereas the ϳ120-kDa species possesses complex type oligosaccharides because of its resistance to endo H ( Fig. 2A, lane 2).
With respect to mutant cell lines, CHO-Lec1 cells produced a TSHR that migrated as an ϳ90-kDa band, slightly smaller than a TSHR with high mannose oligosaccharides ( Fig. 2A,  compare lanes 1 and 7) and was sensitive to endo H ( Fig. 2A,  lane 8). This finding is consistent with the fact that proteins synthesized in CHO-Lec1 cells have oligosaccharides of mannose-rich intermediates (Man 5 GlcNAc 2 ), and high mannose type oligosaccharides in the ϳ95-kDa TSHR in parental CHO cells are presumably Man 8 -9 GlcNAc 2 . CHO-Lec8 cells pro- duced a TSHR migrating as a doublet with molecular masses of ϳ90 and ϳ95 kDa ( Fig. 2A, lane 10); the ϳ90-kDa species is likely a TSHR with galactose-and sialic acid-deficient oligosaccharides (GlcNAc 2 Man 3 GlcNAc 2 ) because of its resistance to endo H ( Fig. 2A, lane 11). In CHO-Lec2 cells, TSHR was visualized as a doublet of ϳ105 and ϳ95 kDa ( Fig. 2A, lane 13). The ϳ105-kDa species is likely a TSHR with sialic acid-deficient oligosaccharides (Gal 2 GlcNAc 2 Man 3 GlcNAc 2 ).
Although it is well known that TSHR cleaves into two subunits (12)(13)(14), TSHR subunits were detectable only after longer chase and N-glycosidase F digestion in our immunoprecipitation (27). Therefore, the subunit structures of TSHRs with different oligosaccharides were analyzed on immunoblotting.
On immunoblotting, the single polypeptide of TSHR, for example, the ϳ95and ϳ120-kDa species in immunoprecipitation of CHO-TSHR, were not visualized. Instead, several nonspecific reactions were observed between 70 and 120 kDa (Fig.  2B). However, the diffuse "A subunit" with the mean molecular mass of ϳ55 kDa could be clearly detected in CHO-TSHR. The majority of TSHRs expressed on the cell surface probably cleave into two subunits at the steady state levels. A similar diffuse band, slightly smaller than that in CHO-TSHR, was also present in Lec2-TSHR (Fig. 2B, lane 5). The diffuse banding pattern of TSHR A subunit in CHO and CHO-Lec2 cells may be attributable to different degrees of glycosylation of the core polypeptides. These oligosaccharides may contain various outer chain sequences including different sizes of polylactosamine (22). In contrast, TSHR A subunit was observed as a doublet in Lec1-TSHR and particularly in Lec8-TSHR (Fig. 2B,  lanes 3 and 4), in which oligosaccharides must be homogeneous. This finding agrees with our recent hypothesis of "two cleavage sites in TSHR" (27). Thus, these data suggest that all of the TSHRs expressed in CHO-Lec cells can cleave into two subunits. Cleavage seems to occur at the cell surface and to be mediated by matrix metalloprotease(s) (31). No cleaved product was observed in tunicamycin-treated, nonglycosylated TSHR (Fig. 2B, lane 2), suggesting that nonglycosylated TSHR may not be expressed on the cell surface.
These data are in agreement with the expected consequences of the mutations represented by the cell lines used.
Effect of Tunicamycin on Expression and Function of TSHR Stably Expressed in CHO Cells-Pooled clones of CHO cells stably expressing TSHR were treated with 5 g/ml tunicamycin for 3 days and subjected to [ 125 I]TSH binding and TSHinduced cAMP production experiments. As shown in Fig. 2A, 5 g/ml tunicamycin completely inhibited transfer of dolichol precursor to Asn in the nascent TSHR. Although effect of 3-day treatment with tunicamycin should be evaluated with immunoblot, not immunoprecipitation, immunoblot showed no specific signal as mentioned above. However, this period of treatment has been previously demonstrated to be enough to purge the cells of wild-type recombinant LHR (9). As shown in Fig.  3A, [ 125 I]TSH binding determined with intact cells was of high affinity in CHO-TSHR. Untransfected CHO cells also showed nonspecific TSH binding of low affinity. This specific, high affinity TSH binding was largely abolished in CHO-TSHR treated with tunicamycin. Consistent with TSH binding data, little or no TSH-induced cAMP production was observed in tunicamycin-treated CHO-TSHR (Fig. 3C). In addition, as shown in Fig. 3B, the solubilized extract from tunicamycintreated CHO-TSHR also showed no TSH binding, although the solubilized extract from CHO-TSHR cultured in the absence of tunicamycin demonstrated significantly higher TSH binding than untransfected CHO cells. Thus, together with the immunoblot data, nonglycosylated TSHR seems to accumulate intracellularly in an unfolded form, thereby suggesting an essential role for acquisition of oligosaccharides in correct folding and membrane expression of TSHR.
[ Fig. 4A, [ 125 I]TSH binding experiments with pooled clones of cells revealed that Lec8-TSHR bound to TSH with an affinity and a B max similar to those of CHO-TSHR. In contrast, specific TSH binding was significantly decreased in Lec1-TSHR and Lec2-TSHR. However, not only the detergent-solubilized extracts from CHO and CHO-Lec8 cells but also those from CHO-Lec1 and CHO-Lec2 cells demonstrated significantly higher [ 125 I]TSH binding than that from untransfected CHO cells (Fig. 4B). Consistent with the TSH binding data in intact cells, CHO-TSHR and Lec8-TSHR produced intracellular cAMP in response to TSH stimulation to a similar extent, and Lec1-TSHR and Lec2-TSHR showed blunted TSH-induced intracellular cAMP synthesis (ϳ10 and ϳ30% of that in CHO-TSHR, respectively) (Fig. 4C).

I]TSH Binding and cAMP Production in CHO and CHO-Lec Cells Stably Expressing TSHR-As shown in
To further evaluate the function of TSHR expressed in CHO-Lec cells, particularly in Lec1 and Lec2, clonal cell lines expressing the highest levels of receptors were used. The clonal cells lines of Lec1-TSHR and Lec2-TSHR, as well as those of CHO-TSHR and Lec8-TSHR, demonstrated specific TSH binding, as shown in Fig. 5A. In these cells, the binding affinity for TSH was inversely correlated with the receptor number expressed. Thus, TSH binding affinity was highest in Lec1-and Lec2-TSHR and lowest in CHO-TSHR (Table I). In functional studies, however, the EC 50 for TSH-induced cAMP production 1, 4, 7, 10, and 13) or treated with endo H (lanes 2, 5, 8, 11, and 14) or N-glycosidase F (lanes 3, 6, 9, 12, and 15) and subjected to 7.5% SDS-polyacrylamide gel electrophoresis under reducing conditions and autoradiography. B, crude membrane preparations (50 g) of the cells were subjected to 7.5% SDS-polyacrylamide gel electrophoresis under reducing conditions. After transfer to a membrane, proteins were probed with A10 and A11. Signals were developed with streptoavidin-horseradish peroxidase and 3,3Ј-diaminobenzidine/H 2 O 2 .

FIG. 2. Immunoprecipitation and immunoblot of TSHR expressed in CHO and CHO-Lec cells. A, the cells were metabolically labeled with [ 35 S]methionine/cysteine (2-h pulse and 2-h chase) followed by immunoprecipitation of TSHR with anti-TSHR monoclonal antibodies A10 and A11 (see "Materials and Methods"). Precipitated samples were left untreated (lanes
was very similar in all of the cells (Fig. 5B and Table I), thereby suggesting that the apparently lower binding affinity for TSH in CHO-TSHR appears to be related to its higher expression levels as previously reported (32). Therefore, we interpret these data as that TSH binding is likely of high affinity in TSHR expressed in all of the cells.
These data indicate that Lec8-TSHR is correctly folded and fully expressed on the cell surface, and that Lec1-TSHR and Lec2-TSHR accomplish conformational maturation but are trapped intracellularly. However, some of these receptors can reach the cell surface and normally transduce a signal.
TSAb and TBIAb Activities in the Clonal Cell Lines Stably Expressing TSHR-The effect of different oligosaccharides on the ability of TSHR to interact with anti-TSHR autoantibodies was examined. In the TBIAb assay shown in Fig. 6A, two Graves' IgG, one with potent TBIAb activity (Graves 1) and the other with moderate activity (Graves 2) in CHO-TSHR, comparably inhibited [ 125 I]TSH binding in all of the CHO-Lec cells. TSAb activity in two Graves' sera (Fig. 6B, Graves 3 and Graves 4) was also similar in all of the cells. Therefore, receptor recognition by autoantibodies does not involve complex type oligosaccharides in TSHR. DISCUSSION In the present studies, using a glycosylation inhibitor, tunicamycin, and a series of mutant CHO-Lec cells defective in the different steps of glycosylation process, we evaluated how acquisition and modification of oligosaccharides affect the functional maturation of TSHR.
Our results indicate that nonglycosylated TSHR produced by inhibition of the first step of glycosylation with tunicamycin  treatment is not folded correctly. In contrast, high affinity TSH binding to Lec1-TSHR clearly indicates that TSHR with mannose-rich intermediates (Man 5 GlcNAc 2 ) is conformationally intact. Therefore, the acquisition of N-linked oligosaccharides appears to allow the nascent, highly unfolded TSHR protein to accomplish conformational maturation in the endoplasmic reticulum. Several roles of addition of N-linked oligosaccharides to nascent proteins have been demonstrated; for example, Nlinked oligosaccharides attach to lectin-like molecular chaperons such as calnexin and calreticulin, facilitating correct protein folding, and also play a role in the "quality control" system of the endoplasmic reticulum that ensures selective transportation of the properly folded proteins for the Golgi complex (33). Therefore, unfolded proteins such as nonglycosylated TSHR mentioned above may be trapped in the endoplasmic reticulum.
Impaired cell surface expression in Lec1-TSHR suggests that processing of N-linked oligosaccharides from high mannose type to complex type in the Golgi apparatus appears to participate in cell surface targeting of TSHR. However, efficient cell surface expression of TSHR is not necessarily dependent on completion of complex oligosaccharides, because TSHR with sialic acid-and galactose-deficient oligosaccharides (Lec8-TSHR) is efficiently expressed on the cell surface. Contrarily, TSHR with sialic acid-deficient oligosaccharides (Lec2-TSHR) is not efficiently expressed on the cell surface. We cannot explain these data at present, although oligosaccharides terminating in galactose residues may somehow interfere with cell surface targeting of TSHR in the Golgi apparatus. Alternatively, lack of sialylation in the trans-Golgi network may affect a role of the trans-Golgi network as an important sorting site for different proteins destined for the apical or basolateral surfaces of cell membranes (34). For any reason, these data are in agreement with the facts that N-linked oligosaccharides generally play critical roles in protein folding and intracellular trafficking in the endoplasmic reticulum and the Golgi apparatus, respectively (33,35). However, our data on TSHR may not be generalized to other membrane proteins, because (i) the membrane receptor for Newcastle disease virus is reported not to be properly expressed in mouse mammary carcinoma Had-1 cells deficient in UDP-galactose transporter like Lec8 cells (36), and (ii) the expression levels of rat LHR were decreased by 30 -40% of that in CHO cells in all of the CHO-Lec cells. 2 Our data also indicate that complex type oligosaccharides do not appear to constitute TSH or autoantibody binding sites or to participate in signal transduction, although glycosylation is reported to be related to function in some proteins (37)(38)(39). It has recently been reported that the truncated type of TSHR ectodomain, which is not secreted, possesses high mannose type carbohydrates and is incapable of binding to TSH or anti-TSHR autoantibodies in CHO cells (15). However, whether this lack of binding was related to polypeptide misfolding or to the type of oligosaccharides on the protein was unknown in this study. From our results, it is now clear that misfolding is the primary event. Thus, TSHR ectodomain by itself may not be able to accomplish correct folding, thereby being trapped in the endoplasmic reticulum and unable to acquire complex oligosaccharides. Recent studies also suggest that only the TSHR ectodomain fused to a membrane-anchored peptide seems to be correctly folded (19 -21).
A nonessential role of galactose and sialic acid in function of TSHR shown in Lec8 cells is reminiscent of our previous finding (4,40) of high affinity TSH binding to TSHR expressed in 293 human embryonal kidney cells. These cells express Nacetylgalactosaminyl-transferase and N-acetylgalactosaminyl-N-acetylglucosaminyl-mannosyl-sulfotransferase and produce N-linked oligosaccharides terminating in sulfate in Ͼ70% of the chains (41). Of interest is that some glycoproteins synthesized in the thyroid gland have been reported to be sulfated (42,43). TSHR expressed in the thyroid gland may also be, at least in part, sulfated. Furthermore, N-linked oligosaccharides in bovine pituitary TSH that we used in the present studies are exclusively sulfated, whereas human pituitary TSH terminates both in sulfation and sialylation (44). Biological activity of sulfated TSH has been reported to be more potent than sialylated TSH (45). The carbohydrates, especially the terminal sialic acids, of TSH, have also been demonstrated to attenuate its biological activity (45,46). In this regard, it will be interesting to compare the relative potency of TSH with different types of oligosaccharides to bind and stimulate TSHR with various oligosaccharides.
In FSH receptor, although removal of N-linked oligosaccharide from the wild-type receptor does not affect FSH binding (5,7), the deglycosylated receptor produced by in vitro mutagenesis or tunicamycin treatment shows no FSH binding activity (7). Thus, the N-linked oligosaccharides seem to be necessary for the proper folding rather than hormone binding in FSH receptor. However, the role of N-linked oligosaccharides in LHR is controversial; although high affinity human chorionic gonadotropin binding to LHR does not appear to involve the carbohydrates on LHR, it remains unclear whether the oligo- saccharides are necessary for the correct folding of the nascent LHR (6,8,9).
In conclusion, our data show that acquisition and processing of N-linked oligosaccharides in TSHR are essential for correct folding in the endoplasmic reticulum and for intracellular trafficking in the Golgi apparatus. However, binding of TSH and anti-TSHR autoantibodies to TSHR does not appear to involve the oligosaccharides of the receptor.