Abstract
The amino-terminal ectodomain of human thyrotropin receptor (TSHR) contains six potential N-linked glycosylation sites (N-Xaa-S/T). This study was designed to evaluate the functional role of TSHR carbohydrates in detail. Because our previous mutagenesis study by Asn to Gln substitutions suggested the critical role of the first and third glycosylation sites (amino acids 77 and 113) for expression of the functional TSHR, we first constructed TSHR mutants having these two glycosylation sites to elucidate whether these two sites are sufficient for TSHR function and expression; this mutant however proved to be nonfunctional. Also the expression levels and function of TSHR mutants with a Ser/Thr to Ala substitution at the first or third glycosylation site were found to be intact. These data indicate that our previous data appear to result from amino acid substitution itself, not from disruption of glycosylation. The next series of the mutants was therefore constructed to identify at least how many glycosylation sites are necessary. Neither TSH binding nor cAMP response was detected in TSHR mutants with three glycosylation sites. However, the mutants with four glycosylation sites were fully functional in terms of TSH binding and cAMP production, although the expression levels were 30 to 40% of that in wild-type TSHR. Finally, Western blot revealed that all six glycosylation sites are actually glycosylated. These data indicate that 1) TSHR ectodomain contains six N-linked carbohydrates, and 2) glycosylation of at least four sites appears necessary for expression of the functional TSHR.
The thyrotropin receptor (TSHR) is physiologically the primary regulator of differentiated function and proliferation of thyroid follicular epithelial cells and pathogenetically a target for the immune system in autoimmune thyroid diseases such as Graves' disease and Hashimoto thyroiditis. This receptor as well as the receptors for lutropin (LHR) and follitropin (FSHR) comprise a unique subfamily of a G protein-coupled receptor superfamily, characterized by a large amino-terminal ectodomain (350–400 amino acid residues), which is thought to be the high-affinity hormone-binding site (Segaloff and Ascoli, 1993; Rapoport et al., 1998).
The ectodomain of TSHR is heavily glycosylated with asparagineN-linked oligosaccharides that represent 30 to 40% of its molecular weight (Rapoport et al., 1996). The functional importance of oligosaccharides in TSHR function and expression has been shown by our previous studies with in vitro site-directed mutagenesis (Russo et al., 1991) and tunicamycin treatment (Nagayama et al., 1998). Thus, inhibition of N-linked glycosylation blocks cell surface expression of the functional TSHR. We have also demonstrated that the addition and processing of oligosaccharides in the endoplasmic reticulum (ER) and the Golgi apparatus are crucial for protein folding and intracellular trafficking, respectively (Nagayama et al., 1998).
There are six potential N-linked glycosylation sites (the consensus sequence is Asn-Xaa-Ser/Thr for glycosylation on an asparagine residue, where Xaa is any amino acid except Pro) (Kornfeld and Kornfeld, 1985) at amino acid positions Asn-77, -99, -113, -177, -198, and -302 of human TSHR ectodomain (Fig.1). Among them, five sites are highly conserved in human, dog, mouse, bovine, and rat TSHR (Nagayama et al., 1989; Parmentier et al., 1989; Akamizu et al., 1990; Stein et al., 1994; Silversides et al., 1997). Asn-113 is an exception and unique in human TSHR. In our previous study (Russo et al., 1991), an amino acid substitution of Asn to Gln at the first or third glycosylation sites (amino acids 77 and 113, respectively) of human TSHR disrupted TSH binding and TSH-stimulated cAMP synthesis. Although these data suggested that these two sites appeared important for cell surface expression of the functional TSH, the possibility cannot be excluded that the amino acid substitution itself rather than disruption of glycosylation might affect the receptor function in these studies. Furthermore, the precise number and location of N-linked carbohydrates have not yet been determined, except that the sixth site has recently been shown to be actually glycosylated in human TSHR (Tanaka et al., 1998).
The present study was therefore designed to study further the role ofN-linked carbohydrates in human TSHR; we constructed new TSHR mutants that had either a Ser/Thr to Ala substitution at amino acids 79 and 115 instead of an Asn to Gln substitution at amino acids 77 and 113, or a different number of glycosylation sites. We show herein that 1) our previous data obtained from an Asn to Gln substitution at the first and third glycosylation sites seem to result from an amino acid substitution itself, not from disruption of glycosylation; 2) all six glycosylation sites are actually glycosylated; and 3) glycosylation of at least four sites appears critical for cell surface expression of the functional receptor.
Materials and Methods
Cell Lines.
Chinese hamster ovary (CHO) cells were grown at 5% CO2 at 37°C in Ham's F-12 medium with 5% fetal calf serum, penicillin (100 U/ml), and streptomycin (100 μg/ml). CHO-Lec8 (ATCC CRL-1737) cells, which lack UDP-galactose translocases and are incapable of transporting UDP-galactose from the cytosol to the Golgi, thereby producing sialic acid/galactose-deficient (GlcNAc2Man3GlcNAc2compared with Sia2Gal2GlcNAc2Man3GlcNAc2in CHO cells) oligosaccharides (Stanley and Siminovitch, 1977), were maintained in minimal essential medium supplemented with 10% fetal calf serum and antibiotics as described above.
Construction and Expression of Mutant TSHRs.
TSHRS79A and TSHRT115A cDNAs (Fig. 1) were constructed by site-directed mutagenesis with overlapping extension method by polymerase chain reaction using wild-type (wt) TSHR cDNA (Nagayama et al., 1989) as a template, and TSHRS79A/N99Q/T115A cDNA was constructed using TSHRS79/T115A cDNA (see below) as a template. The mutations and adjacent regions were confirmed by automated DNA sequencing (Hitachi SQ-5500DNA sequencer; Hitachi Electronics Engineering Co., Tokyo, Japan). TSHRN99/177/198/302Q cDNA was constructed by combining the EcoRI/Aat II-fragment of TSHRN99Q cDNA (Russo et al., 1991) (Fig. 1) and the AaT II/XbaI-fragment of TSHR6xQ cDNA. TSHRN99/177/198Q cDNA was from the EcoRI/Bfr I-fragment of TSHRN99/177/198/302Q cDNA and Bfr I/XbaI-fragment of wt TSHR cDNA. TSHRN77/99/113Q cDNA was from the EcoRI/AaT II-fragment of wt TSHR cDNA and the AaT II/XbaI-fragment of TSHR6xQ cDNA. TSHRN99/302Q cDNA was from the EcoRI/Bfr I-fragment of TSHRN99Q cDNA and the Bfr I/XbaI-fragment of TSHRN302Q cDNA. TSHRN177/198Q cDNA was from the EcoRI/AaT II-fragment of wt TSHR cDNA and the AaT II/XbaI-fragment ofTSHRN99/177/198Q cDNA. TSHRS79/T115A cDNA was from the EcoRI/SnaB I-fragment of TSHRS79A cDNA and the SnaB I/XbaI-fragment of TSHRT115A cDNA. TSHRS79/T115A/N302Q cDNA was from theEcoRI/AaT II-fragment of TSHRS79/T115A cDNA and the AaT II/XbaI-fragment of TSHRN302Q cDNA. These mutant TSHR cDNAs as well as the mutant TSHR cDNAs with a single glycosylation site disrupted (Russo et al., 1991) (Fig. 1) were ligated into the eukaryotic expression vector pCAGGS (Nagayama et al., 1998) in which expression of the mutant TSHRs is controlled by the constitutive CAG promoter.
These expression plasmids as well as pCAG-TSHR expressing wt TSHR (Nagayama et al., 1998) were transfected together with pSV2neo into the cells with Lipofectin reagent (Life Technologies Inc., Grand Island, NY). The cells were selected with 500 μg/ml G418 (Geneticin; Wako, Osaka, Japan). Pooled clones of CHO cells were used for125I-TSH binding and cAMP measurements (see below). Surviving clones of CHO-Lec8 cells (∼12 clones for each transfection) were also isolated with cloning cylinders, and the clonal cell lines expressing the highest levels of the receptor, determined by125I-TSH binding, were selected.
125I-TSH Binding and cAMP Measurement.
125I-TSH binding to intact cells and intracellular cAMP measurement were performed with125I-bovine TSH (TRAb kit; Cosmic, Tokyo, Japan) and with a cAMP radioimmunoassay kit (Yamasa, Tokyo, Japan), respectively, as previously described (Nagayama et al., 1998). Unlabeled TSH used in TSH-binding study was of bovine origin (Sigma, St. Louis, MO).
Western Blot Analysis.
Extraction of the crude cell membrane and immunoblotting were performed as previously described (Nagayama et al., 1998) with mouse anti-human TSHR monoclonal antibody A11 (Nicholson et al., 1996).
Results
As mentioned above, we have previously shown that disruption of the first or third glycosylation sites (amino acids 77 and 113) by an Asn to Gln substitution impaired cell surface expression of the functional human TSHR (Russo et al., 1991). From these data we expected that two oligosaccharides at amino acids 77 and 113 might be sufficient for cell surface expression of the functional TSHR. To explore this possibility, we constructed the mutant TSHR harboring only these two glycosylation sites, TSHRN99/177/198/302Q. Somewhat unexpectedly, neither TSH binding nor cAMP response to TSH stimulation was observed in pooled clones of CHO cells stably transfected with TSHRN99/177/198/302Q (Figs. 1 and 2; Table 1).
These data suggest either that 1) N-linked oligosaccharides at amino acids 77 and 113 are not important for TSHR function, i.e., our previous data resulted from amino acid substitutions introduced per se but not from disruption of glycosylation, or 2) although interpretation of our previous data was correct, TSHR cannot tolerate multiple mutations at glycosylation sites. That is, these two sites are critical, but not sufficient, for cell surface expression of TSHR.
To distinguish these two possibilities, the TSHR mutants TSHRS79A and TSHRT115A were next constructed in which a Ser/Thr to Ala substitution was introduced at amino acids 79 or 115 instead of an Asn to Gln substitution at residues 77 or 113 to disrupt the first and third glycosylation sites. The expression levels and function of TSHRS79A and TSHRT115A stably expressed in CHO cells were indistinguishable from those of wt TSHR in CHO cells (Figs. 1 and 2; Table 1). These data strongly indicate that the former possibility is correct. Thus, an amino acid substitution at amino acid 77 or 113 disrupted the receptor structure.
It is so far unknown how many potential N-linked glycosylation sites actually have oligosaccharides in TSHR. Western blot analysis was therefore performed to address this issue. TSHR is well known to cleave into two subunits (Buckland et al., 1982; Loosfelt et al., 1992; Rapoport et al., 1998), and the anti-TSHR monoclonal antibody we used (A11) recognizes the A-subunit (Nicholson et al., 1996). We first found it difficult to detect a size difference between TSHR with six and five oligosaccharides expressed in CHO cells (data not shown). This is presumably because TSHR A-subunit has highly heterogenous carbohydrates and is detected as a “broad” band in Western blot as previously described (Fig.3, lane 2) (Nagayama et al., 1998). Therefore, TSHRs were expressed in CHO-Lec8 cells in which oligosaccharides are sialic acid/galactose deficient. In these cells TSHR A-subunit is homogenous and can be detected as a “sharp” band by Western blot (Nagayama et al., 1998). Western blot analysis with the clonal cell lines stably expressing the highest levels of each mutant as well as wt TSHR clearly showed that the mobility of immunoreactive bands of all the mutants with a single glycosylation site disrupted is slightly but reproducibly increased compared with that of wt TSHR (Fig.3, lane 3 versus lanes 4–9), indicating that all six sites are actually glycosylated in the context of the native receptor expressed in CHO-Lec8 cells.
From these data together with our previous study (Russo et al., 1991), it is now clear that TSHR ectodomain has six N-linked oligosaccharides, and that disruption of N-linked glycosylation at any single site has little effect on the receptor function and expression. These results suggest the quantitative rather than qualitative (location-specific) importance of oligosaccharides in TSHR.
To identify at least how many glycosylation sites are necessary for cell surface expression of the functional TSHR, the next series of mutant TSHRs with three or four N-linked glycosylation sites was constructed. TSH binding or TSH-induced cAMP production was not detected in TSHR mutants with three glycosylation sites, TSHRN99/177/198Q, TSHRN177/198/302Q, TSHRS79A/N99Q/T115A, and TSHRS79/T115A/N302Q (Figs. 1 and 4; Table1). However, in TSHR mutants with four glycosylation sites, TSHRN99/302Q, TSHRN177/198Q, and TSHRS79/115A, TSH binding and cAMP production were clearly observed; TSH-binding affinity and the EC50 for cAMP response were indistinguishable from those in wt TSHR, whereas the expression levels of these mutants were 30 to 40% of that in wt TSHR (Figs. 1 and 4; Table 1). These data indicate that glycosylation of at least four sites appears necessary for cell surface expression of the functional TSHR.
Discussion
In this article, we show several new findings with respect to the functional role of N-linked carbohydrates on human TSHR. First, our results obtained with Ser/Thr to Ala substitutions revealed that interpretation of our previous data showing impairment of cell surface expression of the functional TSHR by disruption of the first or third glycosylation site (Russo et al., 1991) are likely incorrect. An Asn to Gln substitution at these two sites might have altered the conformation of TSHR. Similar data have been described for FSHR (Davis et al., 1995). Therefore, our previous (Russo et al., 1991) and present results show that disruption of a single N-linked glycosylation site has little effect on TSHR expression and function.
We also demonstrated that all six potential N-linked glycosylation sites on TSHR ectodomain are actually glycosylated in CHO-Lec8 cells. However, it is uncertain that TSHR endogenously expressed in thyroid cells also contains six N-linked carbohydrates. It has been reported that the number of the actual glycosylation sites in LHR is different in the distinct cell types used (Zhang et al., 1995; Davis et al., 1997).
We next showed that although wt TSHR and TSHR mutants with a single glycosylation site disrupted show normal cell surface expression, elimination of two or more glycosylation sites causes progressive reduction in cell surface expression, i.e., an increasing number of mutated glycosylation sites is associated with decreasing surface expression of the receptor. These data together with our previous report (Russo et al., 1991) indicate that the number of sites glycosylated rather than specific sites of glycosylation seems to determine the efficiency of cell surface expression of TSHR. However, it should be noted that the possibility cannot be excluded that the combined effect of three or more simultaneous amino acid substitutions at the glycosylation sites may be structurally deleterious, although a single amino acid substitution is tolerable.
We have recently shown that TSHR is first synthesized as a ∼84-kDa polypeptide chain to which high mannose type carbohydrates attach in the ER that are processed to mature, complex type carbohydrates in the Golgi apparatus (Rapoport et al., 1996; Nagayama et al., 1998). The mature receptor with complex type carbohydrates (∼120 kDa) is then cleaved into two subunits (A and B) on cell surface (Misrahi et al., 1994; Rapoport et al., 1998). (A-subunit is ∼55 kDa in CHO cells and ∼43 kDa in CHO-Lec8 cells; Fig. 3.) Because one of the functional roles of carbohydrates on membrane proteins in the ER is to attach to lectin-like molecular chaperones such as calnexin and calreticulin, which facilitate correct protein folding (Helenius, 1994), we speculate that four or more carbohydrate chains on each TSHR polypeptide may be quantitatively necessary to bind enough amounts of molecular chaperones required for TSHR to fold correctly in the ER. In contrast, the mutant TSHRs with three or less carbohydrates may not fold correctly and may be trapped in the ER by the “quality control” system, a function of the ER that ensures selective transportation of the properly folded proteins from the ER to the Golgi.
Decreased number of carbohydrate moieties seems to affect cell surface expression of TSHR but not to impair TSHR function in terms of TSH-binding affinity and the EC50 for cAMP response in our present study. It has also recently been demonstrated that deglycosylation of native TSHR with PNGase F treatment does not affect autoantibody binding to TSHR (Atger et al., 1999), although the efficacy of PNGase F treatment has not been verified. Thus, it is plausible that the carbohydrates may not be part of ligand-binding site in TSHR and may not be necessary to maintain the three-dimensional structure of TSHR after the completion of correct folding.
In receptors structurally and functionally related to TSHR, the similar results have been demonstrated for FSHR. Thus, FSHR is glycosylated on two of three potential N-linked glycosylation sites and glycosylation of at least one site is necessary for cell surface expression of the functional FSH, whereas deglycosylation of native FSHR does not impair FSH binding (Davis et al., 1995). In contrast, the number of actual glycosylation sites and functional role of carbohydrates are controversial in LHR (Zhang et al., 1995; Davis et al., 1997). However, for both receptors binding of calnexin to receptor with high mannose type carbohydrates has been demonstrated (Rozell et al., 1998).
The functional role of N-linked carbohydrates varies among other members of the G protein-coupled receptor superfamily. Studies with in vitro site-directed mutagenesis or with tunicamycin treatment have revealed that impaired glycosylation does (Rands et al., 1990;Goke et al., 1994; Kaushal et al., 1994; Davidson et al., 1995; Garcia Rodriguez et al., 1995; Couvineau et al., 1996; Ray et al., 1998; Walsh et al., 1998; Ho et al., 1999; Jayadev et al., 1999; Pang et al., 1999) or does not (van Koppen and Nathanson, 1990; Fukushima et al., 1995; Unson et al., 1995; Bisello et al., 1996; Innamorati et al., 1996; Kimura et al., 1997) affect receptor expression and/or function. For most receptors in the former group, impaired glycosylation is associated with reduced cell surface expression of otherwise normal receptors (Rands et al., 1990; Goke et al., 1994; Kaushal et al., 1994;Davidson et al., 1995; Garcia Rodriguez et al., 1995; Ray et al., 1998;Walsh et al., 1998; Jayadev et al., 1999), as for TSHR. Exceptions are the receptors for vasoactive intestinal peptide, secretin and calcitonin, in all of which significance of specific sites forN-linked glycosylation is reported (Couvineau et al., 1996;Ho et al., 1999; Pang et al., 1999). Furthermore, deglycosylation causes a decrease in ligand binding, not cell surface expression, of secretin and calcitonin receptors (Ho et al., 1999; Pang et al., 1999). At present it seems difficult to predict the role ofN-linked glycosylation in receptor function from the primary amino acid sequences.
In summary, we demonstrate herein that all six potentialN-linked glycosylation sites on TSHR ectodomain are actually glycosylated, and that glycosylation of at least four sites appears necessary for cell surface expression of the functional TSHR.
Acknowledgments
We thank Dr. J. P. Banga (Kings College School of Medicine, London) for providing mouse anti-TSHR monoclonal antibody A11, and Prof. Basil Rapoport (Cedars-Sinai Medical Center, Los Angeles, CA) for the cDNAs for the mutant TSHRs and also for the critical review of the manuscript.
Footnotes
-
Send reprint requests to: Dr. Yuji Nagayama, M.D., Department of Pharmacology 1, Nagasaki University School of Medicine, 1-12-4 Sakamoto, Nagasaki 852-8523, Japan. E-mail:nagayama{at}net.nagasaki-u.ac.jp
- Abbreviations:
- TSHR
- thyrotropin receptor
- LHR
- lutropin receptor
- FSHR
- follitropin receptor
- ER
- endoplasmic reticulum
- CHO
- Chinese hamster ovary
- wt
- wild-type
- Received February 16, 2000.
- Accepted June 26, 2000.
- The American Society for Pharmacology and Experimental Therapeutics