Production of antibodies to the human thyrotropin receptor and their use in characterising eukaryotically expressed functional receptor

ELSEVIER

Molecular and Cellular Endocrinology 102 (1994) 77-84

Production of antibodies to the human thyrotropin receptor and their use in characterising eukaryotically expressed functional receptor E. Harfst, M.S. Ross, S.S. Nussey, A.P. Johnstone

*

Department of Cellular and Molecular Sciences, St George’s Hospital Medical School, London, SW1 7 ORE, UK

(Received 22 November 1993; accepted 18 February 1994)

Abstract The structure of the human thyrotropin receptor expressed as a recombinant protein in eukaryotic cells was investigated by immunochemical and functional means using two types of polyclonal rabbit antisera: one raised against the large N-terminal extracellular region (residues l-415) expressed in E. coli and the other raised against a synthetic peptide (residues 313-330). Both types of antisera gave similar results, with the former being more effective. As expected from the lack of conformation of the immunogens, the antisera worked well in immunoblotting. Less predictably, the antisera also recognised the functional receptor in its native state (detected by flow cytofluorimetry and immunoprecipitation), and inhibited the binding of thyrotropin. Thus the region 313-330 is on the outside of the receptor molecule and falls within, or close to, the binding site of thyrotropin. None of the antisera stimulated CAMP production, showing that this is a very special property, largely restricted to certain human autoantibodies. The antisera were used to immunoprecipitate radioiodinated proteins from Chinese hamster ovary cell (CHO) lines expressing recombinant receptor. The most abundant and reproducible cell-surface molecule that correlated with the presence of full-length functional receptor was a glycopolypeptide of approximately 100 kDa, of which 15 kDa is attributable to carbohydrate, in good agreement with the size predicted for the polypeptide from the cDNA sequence. Three other molecular species were also variably detected at the cell surface: 55 kDa, 180 kDa and large molecular weight material at the top of the polyacrylamide gel. We attribute the 55 kDa molecules to breakdown of the intact receptor during handling in vitro; similarly we consider that the very large material is also an in vitro artefact, possibly due to aggregation; there is no obvious explanation for the 180 kDa band. From studies using a glucosidase inhibitor, it could be demonstrated that glycosylation is required for the expression of all of these molecules at the cell surface. Thus, our data favour the functional thyrotropin receptor being the single glycopolypeptide predicted from its cDNA sequence. We found no evidence that further processing of this polypeptide in recombinant cells is required to produce the functional receptor. Key words: Receptor thyrotropin,

(human); Recombinant

protein; Antibody; Binding competitive

1. Introduction The thyrocyte cell-surface receptor for the glycoprotein hormone thyrotropin (TSH) is an important molecule in controlling the growth and function of the normal thyroid, and in humans it is frequently a target of autoimmunity. The cDNA sequence of this receptor (Libert et al., 1989; Nagayama et al., 1989; Misrahi et al., 1990) predicts a polypeptide with an M, of approximately 84000, with six potential glyuxylation sites.

* Corresponding 784-2649.

author. Tel.: 081-672-9944 extn 55780; Fax: 081-

Elsevier Science Ireland Ltd. SSDI 0303-7207(94)00041-7

Hydrophobicity plots suggest that it spans the membrane seven times and it is closely related to two other G protein-linked receptors: the lutropin-choriogonadotropin and follicle-stimulating hormone receptors. These glycoprotein receptors each have large, structurally divergent, extracellularly located amino-terminal domains and this region of the TSH receptor (approximately 400 amino acid residues) has two unique insertions of 8 and 50 amino acids when compared with the other two receptors (residues 38-45 and 317-366 respectively; the numbering system of Nagayama et al. (1989) is used throughout this paper). It is generally assumed that sequences within the large N-terminal extracellular region of the TSH receptor interact with

18

E. Haflst et al. /Molecular and CellularEndocrinology102 (1994) 77-84

TSH and with receptor-binding autoantibodies (stimulator-y and non-stimulatory) present in the sera of patients with autoimmune thyroid diseases. However, we have recently demonstrated that this large extracellular region, produced in a eukaryotic expression system, is insufficient in itself to constitute the high affinity TSH-binding site, although it does contain epitopes for stimulating autoantibodies (Harfst et al., 1992b). In seeking to characterise more fully the structure of the TSH receptor (TSHR), the recombinant molecule has been produced in a number of expression systems. Full-length molecules have been produced in eukaryotic cells, both transiently and as stable lines (for example, Perret et al., 1990; Chazenbalk et al., 1990; Kosugi et al., 1991; Harfst et al., 1992a), and this material appears to be fully functional in binding TSH and pathological autoantibodies and in coupling to second messenger systems [although the lines of Chazenbalk et al. (1990) have only low adenylate cyclase responses]. In contrast, larger amounts of fragments of TSHR can be produced in prokaryotic systems, but there are significant problems of solubility and the products are without function, presumably reflecting the importance of correct folding and/or glycosylation (Takai et al., 1991; Loosfelt et al., 1992; Harfst et al., 1992b; Huang et al., 1992; Costagliola et al., 1994). For unknown reasons, the production of full-length TSHR in the eukaryotic baculovirus system has not been possible (Harfst et al., 1992a; Seetharamaiah et al., 1993), and although this system will produce truncated receptor (Huang et al., 1993; Seetharamaiah et al., 1993) there are again problems of solubility, little evidence for native structure or function exists, and in our hands the level of expression is no higher than that of the better stable lines or transient systems (unpublished data). Despite these efforts, the structure of the TSHR remains controversial, with different groups reporting evidence for single-subunit, two-subunit or multiple forms (Russo et al., 1991; Endo et al., 1992; Loosfelt et al., 1992; Ban et al., 1992). In the present report, we demonstrate that polyclonal antisera raised against prokaryotically expressed recombinant material and a peptide also recognise the native molecule (although the former are generally better) and that these sera can be used to probe the structure of the functional TSHR.

2. Materials and methods 2.1. Eukaryotic cell lines expressing functional human thyrotropin receptor (hTSHR)

The production and characterisation of a series of cell lines, derived from CHO-Kl, which express various levels of functional full-length hTSHR, coupled to

adenylate cyclase, using the glutamine synthetase amplifiable expression system has been described in detail (Harfst et al., 1992a; Harfst and Johnstone 1992). In the present study, lines FLD4 and FLE4.2 were used in further analyses. These contain, by Scatchard analyses, approximately 30000 and 75 000 receptors per cell, respectively. A further CHO cell line expressing the large extracellular region of hTSHR (residues l-415), produced using the same expression system and designated ExG2 (Harfst et al., 1992b), was also used. 2.2. Preparation of prokaryotically expressed extracellular region of hTSHR

Construction of a recombinant plasmid (using the plasmid pGEX-3X, Pharmacia, Milton Keynes, Bucks, UK) for the expression of the extracellular region of the hTSHR (residues 1-415) as a fusion protein with glutathione S-transferase and the characterisation of the protein product have been described in detail in Harfst et al. (1992b). The fusion protein was produced in milligram quantities for immunisation. Transformed E. coli from large scale cultures (750 ml) were suspended in l/100 culture volume of phosphate-buffered saline (PBS; 2.7 mM KCl, 1.5 mM KH,PO,, 137 mM NaCl, 9 mM Na,HPO,) containing 1% Triton X-100, 2 mM EDTA, 0.1% P-mercaptoethanol and 0.2 mM phenylmethylsulphonyl fluoride (PMSF) and sonicated at 100 W on ice for 3 x 10 s. Insoluble material was sedimented by centrifugation at 10000 x g for 20 min. Six rnls of 65 mM Tris-HCl containing 4% SDS, 20% glycerol, 10% P-mercaptoethanol and 0.1% bromophenol blue was added to the pellet and solubilisation achieved by sonication at 100 W for 2 x 10 s and heating the sample to 100°C for 5 min. The sample was subjected to polyacrylamide gel electrophoresis in sodium dodecyl sulphate (SDSPAGE) according to the method of Laemmli (1970); One third of the sample, containing an estimated 2 mg of fusion protein, was loaded onto a 10% polyacrylamide gel (90 ml separating gel and 60 ml stacking gel, 3 mm thickness) and separation of proteins carried out for 16 h at 50 V. The gel was placed in 50% methanol, 7% acetic acid containing 0.05% Coomassie Brilliant Blue for no more than 5 min and then rinsed in distilled water. The fusion protein (approximate M, 70 kDa) was cut from the gel as a strip of 0.5 cm thickness, cut further into 0.5 cm lengths and extruded through a syringe to break up the gel. The preparation was stored at -20°C. 2.3. Production of rabbit polyclonal antisera to the hTSHR

For each rabbit, the broken polyacrylamide gel containing 100-200 pg of bacterially expressed extracellu-

E. Ha@ et al./~Uolecularand CellularEndocrinolo~ IO2 (1994) 77-84

lar region of hTSHR was washed three times, by decantation, with 7% acetic acid in 40% methanol to remove the SDS and three times with PBS. It was then emulsified with Freund’s complete adjuvant and injected intramuscularly. The immunisation was repeated (but using incomplete adjuvant) three times at twoweekly intervals, and then twice more at six-weekly intervals. Sera containing useful levels of antibodies were obtained by bleeding at regular intervals after the fourth injection; the last two immunisations served to maintain this level of antibodies over a period of four months. The production of antisera against a synthetic peptide (corresponding to region 313-330) has been described previously (Harfst et al., 1992b). 2.4. Removal of carbohydrate from glycosylated

79

of 0.25 M sodium phosphate, pH 7.5. Two “Iodo-beads” (Pierce Chemical) and 0.5 mCi carrier-free Nalzl (Amersham) were added and the mixture incubated at room temperature for 5 min. The reaction was terminated by removing the cell suspension from the beads and the cells were washed three times, by centrifugation at 250 X g for 3 min, in phosphate-buffered saline (PBS), PBS containing 0.1% bovine serum albumin (BSA), and PBS again, 10 ml each. The final cell pellet was resuspended in 100 ~1 PBS and an equal volume of ice-cold 1% NP40 in PBS containing 2 mM PMSF, 5 pg/ml aprotinin, 10 pg/ml leupeptin was added and the mixture incubated on ice for 2 min. The nuclei and any insoluble material were removed by centrifugation at 10 000 X g for 2 min, and the supernatant was either stored at -20°C or taken directly for immunoprecipitation.

recombinant protein

2.7. Biosynthetic labelling of cells ExG2 or control CHO-Kl cells (5 X 106) were removed from monolayer culture with 5 mM EDTA, 5 mM EGTA in PBS, washed once in PBS (200 x g, 5 min) and solubilised in 1% Nonidet P40 (NP40) in PBS containing 2 mM phenylmethylsulphonylfluoride (PMSF), 5 pg/ml aprotinin and 10 pg/ml leupeptin (final volume 100 ~1) and centrifuged at 13 000 X g for 5 min to remove insoluble material. Half of each sample was incubated with 100 mU of endoglycosidase F (Calbiochem) in 0.1 M sodium phosphate buffer, pH 8.3, containing 1% NP40 and 0.1% sodium dodecyl sulphate (SDS) for 16 h at 37°C. The other half was incubated under identical conditions but in the absence of the enzyme. Samples were separated by SDSPAGE and Western blot analysis was carried out as described below. 2.5. Inhibition of glycosylation in eukaryotic cell lines Clone FLE4.2 cells (5 x lo6 per 9 cm tissue culture dish) were plated into growth medium containing 5% FCS and 1 mg/ml 1-deoxynojirimycin hydrochloride (DNJ) (Calbiochem, La Jolla, CA) and incubated at 37°C for 24 h. Cells were removed from monolayer with 5 mM EDTA, 5 mM EGTA in PBS and cellsurface proteins radiolabelled as described below. In each experiment, for comparison, FLE4.2 cells were also treated in an identical manner but without the inclusion of DNJ in the culture medium. 2.6. Radioiodination of cell-surface proteins The CHO lines were removed from monolayer culture using 5 mM EDTA, 5 mM EGTA in PBS and washed twice in PBS (200 X g, 5 min). Cells from one confluent 9 cm plate (approximately 3-5 X lo6 cells) were resuspended in 100 ~1 of PBS and added to 10 ~1

Proteins in ExG2 cells were labelled by incubation with [35Slmethionine and solubilised in NP40 as described in Harfst et al. (1992b). 2.8. Immunoprecipitation analyses Immunoprecipitation was carried out at room temperature in PBS containing 0.5% NP40, 0.1% BSA, 2 mM PMSF, 5 pg/ml aprotinin and 10 pg/ml leupeptin. Lysates of NP40-solubilised radiolabelled cells were precleared by incubating for 30 min with 10 ~1 non-immune rabbit serum and 100 ~1 of Protein ASepharose CL4B (a 1: 4 suspension in the above buffer). Following centrifugation (1 min at 2000 X g), precleared supematants (equivalent to 2-5 X lo6 cells per treatment) were incubated with 5 ~1 of non-immune rabbit serum or anti-receptor rabbit antiserum for 30 min before the addition of 50 ~1 of a 1: 4 suspension of protein A-Sepharose and a further incubation of 1 h with mixing. Precipitates were washed and analysed by SDS-PAGE and autoradiography as described in Harfst et al. (1992b). 2.9. Western blotting Cells were removed from monolayer culture using 5 mM EGTA, 5 mM EDTA in PBS for 5 min and washed with PBS by centrifugation at 200 X g for 10 min. The pellet was solubilised by sonicating at 1OOW for 5 set in 65 mM Tris-HCl, pH 6.8, containing 1% SDS, 5% P-mercaptoethanol, 2% glycerol and 0.001% bromophenol blue and heating at 100°C for 5 min. Western blotting was carried out essentially as described in Harfst et al, 1992b. Briefly SDS-PAGE separated samples (equivalent to 2 x 10’ cells per track) were electrophoretically transferred onto nitrocellulose

80

E. Ha@ et al. / Mokmdar and Cellular Endocrinology 102 (1994) 77-84

washed twice in the same buffer by centrifugation at 250 x g for 5 min, and then incubated with fluoresceinconjugated swine anti-rabbit Ig (Dako, High Wycombe, UK, l/40 dilution in the same buffer) for 1 h at room temperature. The cells were washed three more times, fixed in 2% paraformaldehyde in PBS and their fluorescence analysed on a FACScan flow cytofluorimeter (Becton Dickinson, Erembodegem, Belgium).

paper for 2 h at 0.5 A and blocked overnight in 5% milk powder in Tris-buffered saline (TBS; 200 mM NaCl, 50 mM Tris-HCI, pH 7.4). Blocked membranes were incubated for 2 h in rabbit serum diluted l/30 in TBS and then washed. Bound antibody was detected using ‘ZI-labelled, affinity-purified F(ab’), fragment of a horse anti-rabbit IgG. 2.10. Assay for stimulationof cellular CAMP

3. Results

FLD4 cells in 24-well plates (105/well) were incubated with the indicated concentrations of bTSH or serum and extracellular CAMP levels assayed as descibed in Page et al. (1990).

The antisera produced following immunisation with recombinant hTSHR protein made in E. coli reacted well with denatured receptor material, as demonstrated by immunoblotting of the extracellular region expressed in eukaryotic cells, both transiently and in stable lines (Fig. 1). The appropriate negative controls (original CHO-Kl cells and non-immune rabbit sera) demonstrated the specificity of the reactions. The glycosylated extracellular region was visualised as an approximately 60 kDa band (tracks A2, B3 and Cl); after treatment with endoglycosidase F, the size of the band decreased to approximately 45 kDa (track B4). This agrees with the size predicted from the known cDNA sequence (394 amino acids assuming that a signal peptide of 21 residues is removed; predicted molecular weight 44 847). Furthermore, it directly demonstrates the presence of carbohydrate accounting for approximately 15 kDa in the extracellular region, in agreement with the calculations of Costagliola et al. (1994) and, for the rat receptor, Ban et al. (1992). It is notable that the extracellular region expressed in baculovirus-infected insect cells had only 6 kDa attributable to carbo-

2.11. Radioligand binding assays The binding of highly purified bovine TSH (a kind gift of Dr J.G. Pierce, UCLA), radiolabelled with the Bolton-Hunter reagent, to FLD4 or FLE4.2 cells was carried out as described previously (Harfst et al., 1992a; Ha&t and Johnstone 1992). Experimental antisera and control non-immune sera were tested for their inhibition of this binding by adding them to the cells at the same time as the radiolabelled TSH, at various final concentrations up to a maximum of 6.7% (v/v). 2.12. Flow cytojluorimetry The CHO cell lines were removed from plates using 5 mM EDTA, 5 mM EGTA in PBS and l-5 X lo5 cells were incubated in l/50 dilution of antisera or non-immune sera in PBS containing 0.1% BSA, 0.1% sodium azide for 1 h at room temperature. The cells were

C

B

A 1

2

3

4

1

2

3

4

1

2

Fig. 1. Western blot analyses of the extracellular region of hTSHR expressed in eukaryotic cells, using rabbit antiserum to bacterially expressed recombinant protein (serum 14). Panel A. The stable ExG2 cell line (tracks 2 and 4) or the parental CHO-Kl cells (tracks 1 and 3) were reacted with non-immune rabbit serum (tracks 3 and 4) or with antiserum 14 (tracks 1 and 2). Pane1 B. CHO-Kl (track 11, CHO-Kl after treatment with endoglycosidase F (track 21, ExG2 (track 3) and ExG2 after treatment with endoglycosidase F (track 4) were reacted with antiserum 14. Pane1 C. Human 293 cells transiently expressing the extracellular region (track 1) and 293 cells transfected with a negative control plasmid (track 2) were reacted with antiserum 14. The positions of protein standards of known M, (X 10d3) are shown to the left.

E. Ha& et al. /h4olecular ana’ Celluh Endouimbgy

1

2

3

4

102 (1994) 77-84

81

5

-43

-29

Fig. 2. Immunoprecipitation of the extracellular region of hTSHR expressed in eukaryotic cells. ExG2 cells (tracks l-3) or control CHO-Kl cells (tracks 4 and 51 were biosynthetically labelled with [“Sjmethionine, solubilised and immunoprecipitated with non-ithmune rabbit serum (tracks 1 and 41, or rabbit antiserum to peptide 313-330 from the hTSHR (track 21, or rabbit antiserum to bacterially expressed recombinant protein (tracks 3 and 5). The positions of protein standards of known M, (X 10e3) are shown to the right. Fig. 3. Cytofluorimetric analysis of full-length hTSHR on the surface of recombinant CHO lines. Each panel presents the fluorescence histogram obtained for the indicated cell line after incubation with either non-immune rabbit serum (fainter dashed line) or rabbit antiserum to bacterially expressed hTSHR (bolder solid line).

hydrate (Seetharamaiah et al., 19931, suggesting abnormal glycosylation in this system. The antisera also immunoprecipitated the biosynthetically labelled recombinant extracellular region expressed in eukaryotic cells (cell line ExG2) as a 60 kDa band. Occasionally, for unknown reasons, this was observed as a doublet (Fig. 2) although it was more usually a single band (for example, Fig. 1 and Harfst et al., 1992b). Again, negative controls demonstrated the specificity of the reactions (Fig. 2). Thus, the antisera

123456

29-

recognise native receptor molecules. This conclusion was supported by flow cytofluorimetric analysis of the reaction of the antisera with recombinant full-length hTSHR expressed on the surface of eukaryotic cells. The antisera showed clear reaction, above that of non-

78

29-

11 12

9 10

29-

29-

Fig. 4. Immunoprecipitation of full-length hTSHR expressed at the surface of recombinant CHO lines. In four separate experiments, cell surface proteins were radio-iodinated, solubilised and immunoprecipitated with non-immune rabbit serum (NRS) or rabbit antiserum to bacterially expressed recombinant protein (serum 14). Track 1, CHO-Kl cells and NRS; track 2, CHO-Kl cells and antiserum 14; track 3, FLD4 cells and NRS; track 4, FLD4 cells and antiserum 14; track 5, FIE4.2 cells and NRS; track 6, FLE4.2 cells and antiserum 14; track 7. CHO-Kl cells and antiserum 14; track 8, FLEA.2 cells and antiserum 14: track 9, FLE4.2 cells and NRS; track 10, FLE4.2 cells and antiserum 14; track 11, FLE4.2 cells and antiserum 14; track 12, FLE4.2 cells after incubation with the glucosidase inhibitor DNJ, and antiserum 14. Arrowheads on the right indicate the positions of the 100 kDa and 55 kDa hTSHR bands. The positions of protein standards of known IU, (X 10T3) are shown to the left.

82

E. Ha& et al./Molecular and Cellular Endocrinology 102 (1994) 77-84

immune sera, with recombinant clones FLD4 and FLE4.2, but not with the original CHO-Kl cells (Fig. 3). Furthermore, the binding to FLE4.2 was higher than that to FLD4, in agreement with the difference in receptor numbers of these two lines characterised previously by Scatchard analysis of radioligand binding (Harfst and Johnstone, 1992). The antiserum raised against the bacterially expressed extracellular region could also be used to immunoprecipitate the cell-surface full-length hTSHR after radioiodination of FLD4 or FLE4.2 cells; examples from four independent experiments are shown in Fig. 4. The negative controls of non-immune sera and CHO-Kl cells showed variable amounts of a few nonspecific bands (the most prominant, at about 68 kDa, probably represents albumin absorbed from serum in the culture media). Tracks 6, 8, 10 and 11 show the application of the anti-hTSHR serum to FLE4.2 cells repeated in the four separate experiments. The antihTSHR serum precipitated additional bands from FLD4 and FLE4.2: the major reproducible band was approximately 100 kDa; other bands were also detected, in varying amounts, at approximately 55 kDa, 180 kDa and at the top of the gel. The intensity of these bands was greater for FLE4.2 than for FLD4 (cf. tracks 6 and 4), again reflecting the known differences in their receptor numbers (see above). The intensity of all of these receptor-associated bands decreased significantly if the cells were incubated in the glucosidase inhibitor DNJ for 24 h before the surface labelling (cf. tracks 11 and 12), demonstrating that their expression

0

s

2

P I

500

f

0

Serum

concentration

IX)

Fig. 5. Inhibition of binding of ‘%TSH to FLE4.2 cells by antihTSHR antisera. Cells in 24-well plates were incubated with a constant amount of radioiodinated TSH together with the indicated concentration of either non-immune rabbit serum (0 - - - - - - q), rabbit antiserum to peptide 313-330 from the hTSHR (0 -0) or rabbit antiserum to bacterially expressed hTSHR (A -A). The amount of radioactivity bound to the cells was then determined and the data represent the mean f SEM of triplicate wells, after subtraction of non-specific binding determined in the presence of 150 nM unlabelled TSH.

Serum

concentration

IX)

Fig. 6. Production of CAMP by FLD4 cells in the presence of experimental sera. Cells in 24-well plates were incubated with the indicated concentration of: rabbit 50 antiserum to peptide 313-330 n 1, rabbit 14 antiserum to bacterially from the hTSHR (m 01, or pre-immune serum from rabbit 50 expressed hTSHR (o(0 -----0) or rabbit 14 (O------O). The amount of CAMP produced was then determined and the data represent the meanf SEM of triplicate wells.

at the cell surface normally requires glycosylation. The non-specific bands (notably 68 kDa) were not affected by treatment with DNJ, thus providing an internal control for consistency of loading. The effect of the antisera on binding of radiolabelled TSH to full-length hTSHR on the surface of FLE4.2 cells was also investigated (Fig. 5). Both antisera raised against the synthetic peptide and against the bacterially expressed extracellular region inhibited the binding of TSH; the latter was substantially more effective. At the highest concentration used, non-immune sera also showed some inhibition of TSH binding; the use of IgG preparations instead of whole sera did not remove this non-specific inhibition. The production of CAMP by FLD4 cells was measured following incubation with the antisera to determine whether adenylate cyclase was stimulated on their binding to hTSHR. Even non-immune rabbit sera stimulated CAMP production, particularly at high concentrations; dialysis of the sera into the assay buffer before incubation with the cells had no effect on their stimulatory ability, demonstrating that this property resides in large molecules. Fig. 6 shows the data from one experiment; each serum was tested in at least three independent experiments, but in none of these was the effect of any of the antisera significantly higher than that of the non-immune controls.

4. Discussion The availability of large amounts of prokaryotically expressed material derived from the hTSHR, even though it is not of native structure and lacks function,

E. Harfst et al. /Molecular

and Cellular Endocrinology

has allowed the production of a specific antiserum, which has proved useful in characterising the finefiord full-length receptor expressed in eukaryotic cells. The antisera used in this study were raised against denatured protein or synthetic peptide and, hence, would only be expected to recognise linear epitopes. However, at least some of the antibodies in each of the sera show substantial reaction with native molecules, as demonstrated by flow cytofluorimetry, inhibition of TSH-binding, and immunoprecipitation using eukaryotically expressed functional receptor (Figs. 2-5). Hence, some of these linear epitopes, including the region 313-330 corresponding to our synthetic peptide, must be exposed on the surface of the native receptor. Furthermore, the inhibition of TSH-binding by antipeptide sera (Fig. 5) suggests that at least some of this region falls within, or is close to, the TSH binding site. Similarly, the inhibition of TSH-binding by the antiserum raised against bacterially expressed hTSHR suggests that at least some of these epitopes are within the TSH binding site; however, the nature of these epitopes is not known. The antiserum raised against bacterially expressed recombinant material was more effective than those against the synthetic peptide in all analyses: immunoprecipitation of both extracellular region (Fig. 2) and full-length receptor (Fig. 4); flow cytofluorimetry (Fig. 3 compared with Harfst and Johnstone 1992); immunoblotting and inhibition of TSH-binding (Fig. 5). This may be attributable to higher avidity or titre of the former, or to the presence of antibodies against many more epitopes. None of the antisera stimulated cAh4P production significantly more than did non-immune rabbit sera (Fig. 6). We strongly support the comments of Costagliola et al. (1994) on the absolute necessity for careful controls when claiming bio-activity for experimental antibodies. The stimulatory effect of non-immune rabbit sera was higher than that of normal human sera: at l/10 dilution, rabbit sera caused the production of 1.5 to 2-fold as much CAMP as normal human sera and was equivalent to the effect of 25-50 pU/ml TSH. By comparison, in the same experiment as that used to derive the data presented in Fig. 6, sera from some Graves’ patients (at l/10 dilution) caused the production of 250-300 pmoles/well CAMP. These human sera clearly have much lower titres of anti-hTSHR antibodies (probably with lower affinity) than do the experimental rabbit antisera, as demonstrated by their much weaker or undetectable activity in various immunochemical or functional assays, such as immunoprecipitation and flow cytofluorimetry (Harfst et al., 1992b and unpublished data). Hence, the ability of some anti-TSHR antibodies to stimulate CAMP production appears to be a very special property, largely restricted to certain human autoantibodies.

102 (1994)

77-84

83

Another group (Costagliola et al., 1994) have recently also found that polyclonal antisera raised against prokatyotically expressed extracellular region reacts with the native receptor as detected by immunofluorescence on CHO lines expressing full-length hTSHR. These antisera, like ours, did not stimulate cAh4P production. Our antiserum clearly inhibits the binding of TSH to CHO cells expressing functional full-length receptor more than does non-immune serum (Fig. 5), whilst those of Costagliola et al. (1994) do not. The ability of the antisera to react with native receptor allowed an investigation of the polypeptide composition of functional full-length hTSHR by immunoprecipitation. This was only possible following labelling of the recombinant protein to high specific activity by radioiodination; unlike the ExG2 cells, immunoprecipitation of biosynthetically labelled material from FLD4 or FLE4.2 cells was below the limits of detection (data not shown), and this presumably reflects the considerably higher level of expression of truncated compared with full-length recombinant material (Harfst et al., 1992b). Immunoprecipitation of recombinant full-length hTSHR demonstrated the existence of four receptor-associated bands (Fig. 4). All of these are present on the cell surface and all have been subjected to glycosylation during their production; this pattern was not affected by whether or not the material was boiled before SDS-PAGE, unlike the immunoblots of Ban et al. (1992). The 100 kDa band was the most reproducible and would agree with the molecular weight of 84502 predicted for the polypeptide from the cDNA sequence (assuming that the N-terminal 21 residues are removed as a signal peptide) plus carbohydrate accounting for 15 kDa (see Fig. 1); hence it is most likely that this is the primary product. The 55 kDa band was broader than the others and varied considerably in intensity. We think that this is most likely a breakdown product of the higher band(s) which occurs during the handling in vitro; this would be in agreement with the conclusions of others (Russo et al., 1991 from cross-linking studies; Ban et al., 1992 from immunoblotting). Whilst we cannot exclude the possibility that the 55 kDa form is due to physiologically relevant processing of the receptor (as proposed by Loosfelt et al., 19921, we think this unlikely. It is possible that the structure of the functional TSHR in recombinant lines differs from that in thyrocytes; it could be proposed, for example, that receptor processing may be defective in CHO cells compared with native thyroid cells. However, the 55 kDa band certainly does not appear to have functional relevance, since the receptor functions of TSH-binding and TSH-stimulation of adenylate cyclase have remained remarkably constant during the maintenance of FLD4 and FLE4.2 cells in continuous culture for over two years, whereas the amount of the

84

E. Hag% et al. / Molec&r and Cellular Earn

55 kDa form detected by immunoprecipitation varies between each experiment. It is notable that early studies of human thyroid membranes demonstrated that mechanical disruption caused the release of a soluble receptor fragment with a sedimentation coefficient of 4S, which corresponds to approximately 50 kDa (Dirmikis and Munro, 1973; Adlkofer et al., 1980). Endo et al. (1992) inmmnoblotted rat FRTGS cells with anti-peptide antisera and reported a single band of appro~~tely 100 kDa, although Ban et al. (1992) describe only a 230 kDa band from these same cells using similar antisera and technique. Unlike these authors, we have great difficulty in obtaining convincing bands by immunoblotting our CHO lines, despite their high receptor numbers and our effective antisera (FRTL-5 cells have only low levels of TSHR). In this we agree with Costagliola et al. (1994) who reported that immunoblotting of their CHO lines was only possible after purifying the crude membrane preparation on sucrose gradients and using affinity-purified antibody; by doing this they obtained bands of 95 kDa and 55 kDa, which agrees with our ~unoprecipitation data presented here. We can provide no evidence to support the idea of a very large (230 kDa) non-glycosylated precursor proposed by Ban et al. (19921, although this may be present inside the cell and thus not be accessible to radioiodination. The highest molecular weight band that we observe is at the top of the gel; its intensity is variable (Fig. 4) and this is most likely attributable to aggregation during in vitro handling. Other than this, the bands that we observe of 180, 100 and 55 kDa agree with those reported by Ban et al. (1992) using immunoblotting of recombinant receptor transiently expressed in Cos cells, although this same report mentions that only a 230 kDa band is detectable in stable recombinant CHO lines, contrary to the immunoblotting data of Costagliola et al. (1994). We have no explanation for the 180 kDa band; its intensity is also variable; a dimer of the receptor chain is a possibility, although this would have to remain intact under the reducing and denaturing conditions of SDS-PAGE. From these accumulated data we conclude that the functional TSHR is a single ~y~~l~eptide of approximately 100 kDa, of which 15 kDa is attributable to carbohydrate, which is easily broken down to a 55 kDa species during handling.

IO2 @?W4)77-84

Acknowledgements We thank Dr V. Ang for assistance with the production of antisera and Mr R. Jackson for CAMP assays. We are grateful to Dr G. Vassar% (IRIBHN, Universite Libre de Bruxelles) for allowing us to refer to his group’s work before publication. This work was supported by the Wellcome Trust. References Ban, T., Kosugi, S. and Kohn, L.D. (1992) Endocrinology 131, 815-829. Chazenbalk, G.D., Nagayama, Y., Russo, D., Wadsworth, H.L. and Rapoport, B. (1990) J. Biol. Chem. 265,20970-20975. Costagliola, S., Alcalde, L., Ruf, J., Vassar& G. and Ludgate, M. (1994) J, Mol. Endocrinol., in press. Dim&is, S. and Munro, D.S. (1973) J. Endocrinol. 58, 577-590. Endo, T., Ikeda, M., Ohmori, M., Anzai, E., Haraguchi, K. and Onaya, T. (1992) Biochem. Biophys. Res. Commun. 187,887-893. Harfst, E., Johnstone, A.P., Gout, I., Taylor, A.H., Waterfield, M.D. and Nussey, S.S. (1992a) Mol. Cell. Endocrinol. 83, 117-123. Harfst, E., Johnstone, A.P. and Nussey, S.S. (1992b) J. Mol. Endocrinol. 9, 227-236. Harfst, E. and Johnstone, A.P. (1992) Anal. Biochem. 207,80-84. Huang, G.C., Collison, K.S., McGregor, A.M. and Banga, J.P. (1992) J. Mol. Endocrinol. 8, 137-144. Huang, G.C., Page, M.J., Nicholson, L.B., Colt&m, KS., McGregor, A.M. and Banga, J.P. (1993) J. Mol. Endocrinol. 10, 127-142. Kosugi, S., Ban, T., Akamizu, T. and Kohn, L.D. (1991) J. Biol. Chem. 266,19413-19418. Laemmli, U.K. (1970) Nature 227,680-685. Libert, F., Lefort, A., Gerard, C., Parmentier, M., Perret, J., Ludgate, M., Dumont, J.E. and Vassar& G. (1989) Biochem. Biophys. Res. Commun. 165,1250-1255. Loosfelt, H., Pichon, C., Jolivet, A., Misrahi, M., Caillou, B., Jamous, M., Vannier, B. and Milgrom, E. (1992) Proc. Natl. Acad. Sci. USA 89, 3765-3769. Misrahi, M., Loosfelt, H., Atger, M., Sar, S., Guiochon-Mantel, A. and Milgrom, E. (1990) Biochem. Biophys. Res. Commun. 166, 394-403. Nagayama, Y., Kaufman, K.D., Seto, P. and Rapoport, B. (1989) B&hem. Biophys. Res. Commun. 165, 1184-1190. Page, S.R., Taylor, AH., Driscoll, W., Baines, M., Thorpe, R., Johnstone, A.P., Nussey, S.S. and Whitley, G.S.J. (1990) J. Endocrinol. 126, 333-340. Perret, J., Ludgate, M., Libert, F., Gerard, C., Dumont, J.E., Vassart, G. and Parmentier, M. (1990) Biochem. Biophys. Res. Commutt. 171,1044-1050. Russo, D., Chazenbalk, G.D., Nagayama, Y., Wadsworth, H.L., Seto, P. and Rapoport, B. (1991) Mol. Endocrinol. 5, 1607-1612. Seetharamaiah, G.S., Desai, R.K., Dallas, IS., Tahara, K., Kohn, L.D. and Prabhakar, B.S. (1993) Autoimmunity 14, 315-320. Takai, O., Desai, R.K., Seetharamaiah, G.S., Jones, C.A., Allaway, G.P., Akamizu, T., Kohn, L.D. and Prabhakar, B.S. (1991) Biochem. Biophys. Res. Commun. 179,319-326.