- Email: [email protected]
thyroid peroxidase
15
Molecular and Cellular Endocrinology, 53 (1987) 15-23 Elsevier Scientific Publishers Ireland, Ltd.
MCI? 01704
Structure-activity Y. Nakajima,
analysis of microsomal
antigen/
thyroid peroxidase
R.D. Howells, C. Pegg I, E. Davies Jones and B. Rees Smith Endocrine Immunology and I Department
Unit, Uniuersity of Wales College of Medicine,
ofSurge?, University ofNottingham.,Nottingham
CardifJ U.K.
(Received 16 February 1987; accepted 7 April 1987)
Key words: Thyroid; Microsomal antigen; Thyroid peroxidase; Thyroid autoi~unity
Summary
The interaction between thyroid microsomal autoantibodies and thyroid microsomal antigen/ thyroid peroxidase (TPO) has been studied using both intact antigen preparations and their water-soluble trypsin fragments. In an analysis of sera from 30 patients with Graves’ or Hashimoto’s diseases, microsomal antibodies showed similar reactivity towards trypsin fragments (with TPO activity) and intact detergent (sodium deoxycholate, DOC)-solubilized human microsomal antigen preparations (r = 0.96). This raised the possibility that both the peroxidase-active site and the major autoantigenic site(s) of microsomal antigen were present on the same trypsin fragments. Studies with porcine TPO showed that only a few sera contained microsomal ~tib~ies which cross-reacted strongly with the porcine preparations. Further analysis was carried out by i~unoprecipitation of 1251-labelled microsomal antigen followed by SDS-PAGE and autoradiography. These studies suggest that intact human microsomal antigen (a single-chain protein with M, = 110 000) contains an intrachain loop of amino acids formed by a disulphide bridge. Trypsin treatment cleaves the antigen close to its transmembrane section and releases a water-soluble fragment (M, = 100 000), containing the intact disulphide-linked loop of amino acids. Further trypsin action causes cleavage of the peptide bonds within the loop in some preparations. Consequently, three major water-soluble trypsin fragments (M, = 100 000, 73 000 and 68 000) are formed all of which contain an intact disulphide bridge and have microsomal antibody binding activities. The integrity of the disulphide bridge in intact antigen/TPO preparations and their trypsin fragments is essential for autoantibody binding activity.
Introduction Sera from patients with autoimmune thyroid disease frequently contain thyroid microsomal antibodies (Goudie et al., 1963), and there is evidence that these autoantibodies are involved in thyroid destruction (Bottazzo et al., 1986). Recent Address for correspondence: Dr. B. Rees Smith, Endocrine l~unolo~ Unit, 7th Floor Medicine, University of Wales College of Medicine, Cardiff CF4 4XN, U.K. 0303-7207/87/$03.50
studies have indicated that thyroid microsomal antigen is an integral membrane protein, consisting of a single 110 kDa subunit (Banga et al., 1985; Hamada et al., 1985; Kajita et al., 198.5; Furmaniak et al., 1986). Furthermore, the antigen appears to be identical to thyroid peroxidase (TPO) (Czarnocka et al., 1985, 1986; Portmann et al., 1985; Kotani et al., 1986). Trypsinization of detergent-solub~ed thyroid microsomal preparations causes the release of a water-soluble fragment of thyroid peroxidase with
0 1987 Elsevier Scientific Publishers Ireland, Ltd.
enzyme activity (Ohtaki et al., 1982), and the objectives of the present study were to investigate the interaction of microsomal antibodies with the trypsin fragments of human and porcine microsomal antigen/TPO. Our studies indicate that TPO activity and the major microsomal antibody binding site(s) are located on trypsin fragments which contain a disulphide bridge. Methods Microsomal antigen preparations Human thyroid tissue was obtained from patients undergoing partial thyroidectomy for Graves’ disease and porcine thyroid tissue from a local abattoir. A microsomal fraction was prepared by differential centrifugation of thyroid homogenates (Schardt et al., 1982) and solubilized by treatment with 1% sodium deoxycholate (DOC) in 50 mM NaCl; 10 mM Tris-HCl pH 8.3 and centrifuged at 100 000 X g for 1 h at 4 o C (Kajita et al., 1985). In some experiments the DOCsolubilized preparations were partially purified by chromatography on Sephacryl S-300 (Furmaniak et al., 1986). Purification of thyroid peroxidase activity Thyroid peroxidase (TPO) activity was measured by the guaiacol method (Hosoya and Morrison, 1967) and a trypsin fragment of the enzyme purified from porcine thyroid tissue by the method of Ohtaki et al. (1982). Briefly, 30 ml of DOCsolubilized thyroid microsomes (2 units of TPO per mg of protein) were treated with trypsin (0.01%) for 40 min at room temperature and after addition of trypsin inhibitor (0.02%), dialysed overnight against 10 mM sodium phosphate buffer pH 7.4, containing 0.1 mM KI. The dialysed material was then run on a 100 ml column of DEAE-cellulose and eluted with a linear salt gradient (O-O.3 M KC1 in 10 mM phosphate; 0.1 mM KI pH 7.4; 500 ml). Fractions containing TPO activity (20 units per mg of protein: 40% recovery) were then concentrated by ultrafiltration and run (20 ml h-l) on a 2.6 X 90 cm column of Sephacryl S-300 in 10 mM phosphate; 0.1 mM KI buffer pH 6.4. This preparation contained 150 units per mg; 35% recovery during gel filtration. Further purification was then achieved by chro-
matography on a 30 ml hydroxyapatite column using a linear gradient from 10 to 200 mM potassium phosphate pH 6.8 containing 0.1 mM KI. All procedures were carried out at O-4 o C. The specific activity of the final product was 200-300 units of TPO activity per mg of protein (protein concentration determined by the method of Bradford, 1976) with an overall recovery of 20%. This represented a purification factor of 100-200 times over the initial DOC-solubilized porcine thyroid microsomes. Analysis of the purified trypsin fragment by SDS-PAGE under non-reducing conditions showed the presence of three major bands, M, = 60 000, 70000 and 90 000 stained by Coomassie blue (Fig. 5). Insufficient human thyroid tissue was available to carry out extensive purification of human TPO activity. However, DOC-solubilized human thyroid microsomes partially purified by S-300 gel filtration in 1% DOC were treated with trypsin and re-run on Sephacryl S-300 in the absence of DOC. Microsomal antigen as measured by microsomal antibody binding and TPO activity co-eluted on Sephacryl S-300 prior to trypsinization. After trypsinization, the two activities co-eluted again, but in a greater volume than that of the nontrypsinized material. Measurement of autoantibodies to thyroid microsomal antigen Plastic tubes were coated (overnight at 4 o C in 0.1 M sodium bicarbonate buffer, pH 9.2) with thyroid microsomal antigen preparations under investigation or thyroglobulin. After washing, sera to be tested for microsomal antibodies (from normal donors or Graves’ or Hashimoto patients) diluted in 150 mM NaCl; 10 mM Tris-HCI pH 7.5 containing 5 mg ml-’ bovine serum albumin and 15 pg ml-l thyroglobulin (sufficient thyroglobulin to inhibit completely the binding of any thyroglobulin antibodies in the test sera to thyroglobulin contaminating the microsome preparations) were added (250 ~1 per tube in triplicate). After 2 h at 37” C, the tubes were aspirated, washed and 0.5 ml (50000 cpm) of 1251-labelled protein A (labelled by the method of Fraker and Speck, 1978, to a specific activity of 10 PCi per pg) added. After a further incubation of 2 h at 37 o C and washing, the amount of 1251-labelled protein
17 Labelling and immunoprecipitation of microsomal antigen preparations Purified or partially purified antigen preparations were labelled to a specific activity of 10 PCi per pg of protein using the Iodogen method of Fraker and Speck (1978). Aliquots (20 PCi) were pre-adsorbed (4’C, 16 h) with 100 ~1 of Staphylococcus aureus cells (Pansorbin-Calbiothem) and then incubated (37’C, 2 h) with diluted (20-fold in diluent containing 200 pg ml-’ human or porcine thyroglobulin) serum containing high levels of thyroid microsomal antibody. Pansorbin (100 ~1) was then added and after incubation for 2 h at room temperature and centrifugation (15 000 X g, 5 min), the cell pellets were washed. Electrophoresis sample buffer (4% SDS in 0.1 M Tris-HCl pH 6.8 containing 20% glycerol) with or without 10 mM dithiothreitol (DTT) was then added, and after heating (100 o C, 3 min) the samples were analysed by SDS-PAGE (5-15% gradient gels) followed by autoradiography (Kajita et al., 1985).
A bound to the antibody-antigen complex was determined. In the presence of human serum without detectable microsomal antibody labelled protein A binding was about 2% (see legends to Figs. l-3). This was due in the main part to the interaction of labelled protein A with IgG which contaminated microsome preparations. Human serum without detectable microsomal antibody was always included in each assay to assess the level of this ‘non-specific’ binding (legend to Figs. l-3). Removal of IgG contamination from the solubilized microsomes reduced the non-specific binding to about 0.2%. Preparation of thyroglobulin Thyroglobulin was purified from the supernatants obtained after sedimentation of human or porcine thyroid microsomes by precipitation between 1.52 M and 1.76 M ammonium sulphate (Derrien et al., 1948) followed by gel filtration on Sephacryl S-300.
.
. .
0
10 Binding
20 to intact
30 HUMAN
microsomal
.
0
. .
.
40 antigen
50
60
(%)
Fig. 1. Comparison of thyroid microsdmal antibody binding to intact DOC-solubilized human microsomal antigen and intact DOC-solubilized porcine TPO. The plastic tubes used in the assays were coated with 6 and 16 ag ml-’ of protein in the case of human and porcine preparations respectively (peroxidase activity 0.7 and 2 units per mg of protein for human and porcine preparations respectively). See text for full experimental details. The results of 30 sera from patients with Graves’ or Hashimoto’s disease are shown. The two parameters showed a significant correlation (r = 0.75; P < 0.001). Labelled protein A binding with eight individual normal sera gave a mean value of 2.3% (range 1.8-2.7s) in the case of human microsomal antigen, and 2.3% (range 2.1-2.5%)
in the case of porcine TPO.
18
Effect of reduction on thyroid microsomal antigen/ TPO DOC-solubilized human thyroid microsomal antigen or purified porcine TPO preparations were incubated with DTT (final concentration 10 mM) for 15 min at room temperature. Iodoacetamide
. . . . .
.
.
(400 mM in 2 M Tris-HCl pH 8.3) was then added to a final concentration of 40 mM and incubated for 10 min at room temperature. The mixture was dialysed overnight at 4” C against 1 1 of 1% DOC containing 0.1 mM KI (in the case of intact human microsomal antigen) or 1 1 of 10 mM phosphate buffer pH 6.8 containing 0.1 mM KI (in the case of purified porcine TPO), with one change of dialysis buffer. In control experiments DTT was inactivated prior to addition to microsomal antigen/TPO preparations by pre-incubation (15 min room temperature), with a 4-fold molar excess of iodoacetamide. After dialysis, both control and reduced samples were assayed for microsomal antibody binding and TPO activity. Results Microsomal antibody binding to intact DOCsolubilized S-300-purified human and porcine microsomes showed a significant correlation (r = 0.75; n = 30; P < 0.001) (Fig. 1). Over 20 of the 30 sera studied showed strong binding to human antigen, but only four bound well to porcine anti-
.
; .. O-
. . .
OL 0 Binding
.
l -
L trypsin
fragment (%1251-protein
of
human A
microsomal
.
40
30
20
10
to
.
antigen
bound)
Fig. 2. Comparison of thyroid microsomal antibody binding to intact DOC-solubilized human microsomal antigen and the trypsin fragment of human microsomal antigen. Intact human microsomal antigen was partially purified by S-300 gel filtration in 1% DOC. The trypsin fragment was prepared by digesting this preparation with trypsin and then rerunning on S-300 in the absence of DOC. The plastic tubes used in the assays were coated with 6 pg ml-’ of protein in the case of intact microsomal antigen (peroxidase activity 6 units per mg of protein) and 10 pg ml-i of protein in the case of the trypsin fragment (peroxidase activity 4 units per mg of protein). See text for full experimental details. Results of 30 sera from patients with Graves’ or Hashimoto’s disease are shown. The two parameters showed a highly significant correlation (r = 0.96; P i 0.001). Labelled protein A binding with eight individual normal sera gave a mean value of 2.3% (range 1.8-2.7%) in the case of intact antigen and 5.4 (range 4.5-6.0%) in the case of the trypsin fragment,
Fig. 3. Comparison of thyroid microsomal antibody binding to intact DOC-solubilized porcine TPO and the purified trypsin fragment of porcine TPO. Intact porcine TPO was partially purified by S-300 gel filtration. The trypsin fragment was prepared as described in detail in the text. Plastic tubes used in the assays were coated with 16 pg ml-’ of protein (peroxidase activity 6 units per mg) in the case of intact porcine TPO and 10 pg ml-’ of protein (peroxidase activity 200 units per mg) in the case of purified trypsin fragment. Results with 30 sera from patients with Graves’ or Hashimoto’s disease are shown. The two parameters showed highly significant correlation (r = 0.93; P < 0.001). Labelled protein A binding with eight individual normal sera gave a mean value of 2.3% (range 2.1-2.5%) in the case of intact TPO and 2.2% (range 1.8-2.596) in the case of the trypsin fragment.
19
gen. Microsomal antibody binding to intact DOC-solubilized ~crosomal antigen correlated well with binding to the antigen’s trypsin fragment in the case of human (r = 0.96; n = 30; P < 0.001) (Fig. 2) and porcine (r = 0.93; n = 30; P < 0.001) (Fig. 3) material. Considerably greater antibody binding was obtained with purified trypsin fragments of porcine TPO compared with non-purified intact porcine TPO (Fig. 3), and 12 out of the 30 sera showed relatively strong binding to the porcine trypsin fragment. SDS-PAGE analysis of intact DOC-solubilized human microsomal antigen labelled with *z51 indicated that a major band with Mr = 110000 was
~unopre~ipitated by ~crosom~ antibody (data not shown, but see Kajita et al., 1985). Similar analysis with trypsin fragments of human microsomal antigen indicated that three major bands were specifically immunoprecipitated by microsomal antibody and ran with 54,s of 68 000,73 000 and 100000 under non-reducing conditions (Fig. 4A). Under reducing conditions, major specifically immunoprecipitated bands with M, = 33 000, 36 000 and 90 000 were evident (Fig. 4B). Further experiments were carried out with lZ51labelled porcine thyroid peroxidase preparations. In the case of intact porcine TPO, partially purified by gel filtration’ on Sephacryl S-300, immunoprecipitation of the ‘251-labelled material
5
1
2
3
4
5
Fig. 4. Immunoprecipitation of the lab&d trypsin fragment of human microsomal antigen. (A) SDS-PAGE under non-reducing conditions; immunoprecipitation with: lane 1= Graves’ serum (patient DO) with high levels of microsomal antibody; lane 2 = Graves’ serum (patient SC) with high levels of microsomal antibody; lane 3 = Hasbimoto serum (patient GO) with high levels of microsomal antibody; lane 4 = normal pool serum; lane 5 = untreated material (non-i~unoprecipitated); lane 6 = labelled IgG marker. (B) SDS-PAGE under reducing conditions; i~unoprecipitation with: lane 1= labelled IgG marker; lane 2 = Graves’ serum (patient DO); lane 3 = Graves’ serum (patient SC); lane 4 = Hashimoto serum (patient GO); lane 5 = normal pool serum; lane 6 = untreated material (non-immunoprecipitatcd). The data shown in 4A and 48 are typical of two and three separate experiments respectively.
20
B Fig. 5. Analysis of the purified trypsin fragment of porcine TPO by SDS-PAGE under non-reducing conditions. (A) Unlabelled material, gel stained with Coomassie blue. (B) 12’1-Labelled material immunoprecipitated with: lane 1= normal pool serum; lane 2 = Hashimoto serum (patient KF) with high levels of microsomal antibody; lane 3 = untreated material (non-immunoprecipitated). The data shown are typical of two separate experiments.
with microsomal antibody failed to result in specific enrichment of any particular protein bands (data not shown). However, when the purified trypsin fragments of porcine TPO were labelled and immunoprecipitated, all the major components of the trypsin fragments (M, = 60 000,70 000 and 90000 under non-reducing conditions) were immunoprecipitated by microsomal antibody (Fig. 5). Under reducing conditions, a different pattern of both labelled and non-labelled bands was obtained (Fig. 6) with major labelled components at M, = 26 000, 34000, 55 000 and 82 000 (Fig. 6B). Coomassie blue stained major bands at &I, = 26 000 and 55 000 (Fig. 6A). Treatment of intact DOC-solubilized human microsomal antigen with DTT caused a marked loss of TPO activity (Fig. 7). Microsomal antibody
binding activity was also markedly diminished in experiments with over 20 sera with microsomal antibody activity (Fig. 7). Similar losses of TPO activity and microsomal antibody binding activity (four out of four sera studied) were observed on reduction of purified porcine TPO with labelled protein A binding falling from 17%, 13%, 12% and 10% to about 3% (normal serum value) in all four cases. Discussion
Our studies demonstrate that human thyroid microsomal antibodies do not usually cross-react well with porcine thyroid microsomal antigen preparations (Fig. 1). The trypsin fragment of human thyroid microsomal antigen reacted with
Fig. 6. Analysis of the purified trypsin fragment of porcine TPO by SDS-PAGE under reduciag conditions. (A) Unlabelled material, gel stained with Coomassie blue. (B) ‘251-LabeIled material immunoprecipitated with: lane 1 = IabeUed IgG marker; lane 2 = normal pool serum: lane 3 = Hashimoto serum (patient RF) with high levels of microsomal antibody; lane 4 = untreated material (non-immunoprecipitated). The data shown are typical of two separate experiments.
microsomal antibodies to a similar extent to the intact DOC-solubilized antigen (Fig. 2) and contained similar levels of peroxidase activity (Fig. 2, legend). This raised the possibility that both the auto~tigenic site or sites and the active site of the peroxidase enzyme were on the same water-soluble trypsin fragment(s). Immunoprecipitation of labelled human microsomal antigen preparations and analysis by SDS-PAGE under non-reducing conditions showed that the human trypsin fragment contained three bands with M, = 68 000, 73 000 and 100000 compared with a broad band at 110000 for the intact antigen (data not shown but see Kajita et al., 1985). Under reducing conditions, the M, of intact human microsomal antigen and the 100 kDa fragment were only slightly changed whereas the 68 kDa and 73 kDa trypsin fragment components showed marked changes in
M, to 33 000 and 36 000 (Fig. 4). This indicated that two of the major components of the human trypsin fragments consisted of peptide chains linked by disulphide bridges. Consequently, we can propose a structure for human microsomal antigen and its tryptic fragments as shown in Fig. 8. Intact human microsomal antigen is a single 110 kDa integral membrane protein with at least one loop of amino acids linked by an intrachain disulphide bridge. Trypsin treatment results in cleavage at a point or points close to the transmembrane section and this gives rise to a water-soluble fragment with M, = 100 000 under non-reducing conditions. A similar-sized fragment has been described recently by Ohtaki et al. (1986). Some of the water-soluble 100 kDa fragments have at least one additions trypsin cleavage point inside the disulp~de-inked loop of amino acids
22 intact
human
microsp,m,;Ka)ntigen 50
Trypsin
S
-ki
red;ction
Fig. 8. Proposed
untreated microsomal antigen
Microsomal antigen plus inactive DTT
Microsomal antigen plus active
DTT
Fig. 7. Effect of reduction on microsomal antibody binding and TPO activities in intact DOC-solubilized human microsomal antigen preparations. The 30 sera shown are from the same Graves’ and Hashimoto patients as in Figs. 1-3. See text for full experimental details. DTT treatment has similar effects on TPO and microsomal antibody binding activities (four sera studied) in purified porcine TPO preparations (see text).
and possibly additional cleavage points outside the loop. This gives rise to the fragments with A4, = 68 000 and 73 000 under non-reducing conditions (Fig. 8). On reduction of the disulphide bridge which forms the loop, the M, of the 100 kDa fragment falls slightly to 90000 presumably as a result of conformational changes only. However, reduction of the 68 kDa and 73 kDa fragments results in a fall of 50% in their M,s as the loop of amino acids formed by the disulphide bridge has been cleaved by trypsin (Fig. 8). The purified trypsin fragments of porcine microsomal antigen reacted more strongly with microsomal antibodies than intact porcine antigen preparations (Fig. 3). The reason for this is not clear, but the difference in reactivity between the two preparations was also observed in immunoprecipitation studies. It may be related to
I f
reduction
structure of human microsomal effects of trypsin and reduction.
+
reduction-
antigen
and
the relatively low concentration of TPO in the intact preparations. Analysis of the purified porcine trypsin fragments by SDS-PAGE under non-reducing conditions followed by Coomassie blue staining indicated the presence of. three major bands at M, = 60000, 70000 and 90000 (Fig. 5). When the preparations were labelled with rz51, all three bands were immunoprecipitated with microsomal antibody. This suggested that the purified porcine trypsin preparations contained three major fragments, all of which reacted with microsomal antibody. A different pattern of bands was obtained when SDS-PAGE was carried out under reducing conditions (Fig. 6) (major labelled bands at M, = 26 000, 34000, 55 000 and 82000). This indicated that some of t-he components of the porcine trypsin fragments were linked by a disulphide bridge or bridges, in a similar way to the trypsin fragments of human microsomal antigen (Fig. 8). Reduction of human or porcine microsomal antigen/TPO preparations resulted in loss of most of their TPO and microsomal antibody binding activities. The small residual amount of both activities (Fig. 7) may well have been due to a small amount of non-reduced material remaining. Our data suggest therefore that the major microsomal antibody binding site(s) and possibly TPO activity
23
depend on the integrity of the disulphide bridge shown in Fig. 8. Acknowledgements This work was supported by RSR Ltd. and the Medical Research Council. Dr. Y. Nakajima was in receipt of a Smith and Nephew Foundation Fellowship. We are most grateful to Dr. J. Furmaniak and Dr. F. Hashim for helpful advice and discussions and to Mrs. K. Earlam for excellent secretarial assistance. References Banga, J.P., Pryce, G., Hammond, L. and Roitt, I.M. (1985) Mol. Immunol. 22, 629-642. Bottazzo, G.F. and Doniach, D. (1986) Annu. Rev. Med. 37, 353-359. Bradford, M.M. (1976) Anal. Biochem. 72, 248-254. Czamocka, B., Ruf, J., Ferrand, M., Carayon, P. and Lissitzky, S. (1985) FEBS Lett. 190. 147-152.
Czamocka, B., Ruf, F., Ferrand, M., Lissitzky, S. and Carayon, P. (1986) J. Endocrinol. Invest. 9, 135-138. Goudie, R.B. and McCallum, H.M. (1963), Lancet ii, 1035-1038. Derrien, Y., Michel, R. and Roche, J. (1948) Biochim. Biophys. Acta 2, 454-470. Fraker, P.J. and Speck, J.C. (1978) Biochem. Biophys. Res. Commun. 80, 849-857. Furmaniak, J., Davies Jones, E., Buckland, P.R., Howells, R.D. and Rees Smith, B. (1986) Mol. Cell. Endocrinol. 48, 31-38. Hamada, N., Grimm, C., Mori, H. and DeGroot, L.J. (1985) J. Clin. Endocrinol. Metab. 61, 120-128. Hosoya, T. and Morrison, M. (1967) J. Biol. Chem. 242, 2828-2836. Kajita, Y., Morgan, D., Parkes, A.B. and Rees Smith, B. (1985) FEBS Lett. 187, 334-338. Kotani, T., Umeki, K., Matsunaga, S., Kato, E. and Ohtaki, S. (1986) J. Clin. Endocrinol. Metab. 62, 928-933. Ohtaki, S., Nakagawa, H., Nakamura, M. and Yamazaki, I. (1982) J. Biol. Chem. 257, 761-766. Ohtaki, S., Kotani, T. and Nakamura, Y. (1986) J. Clin. Endocrinol. Metab. 63, 570-576. Portmann, L., Hamada, N., Heinreich, G. and DeGroot, L.J. (1985) J. Clin. Endocrinol. Metab. 61, 1001-1003. Schardt, C.W., McLachlan, SM., Matheson, J. and Rees Smith, B. (1982) J. Immunol. Methods, 55, 155-168.
Molecular and Cellular Endocrinology, 53 (1987) 15-23 Elsevier Scientific Publishers Ireland, Ltd.
MCI? 01704
Structure-activity Y. Nakajima,
analysis of microsomal
antigen/
thyroid peroxidase
R.D. Howells, C. Pegg I, E. Davies Jones and B. Rees Smith Endocrine Immunology and I Department
Unit, Uniuersity of Wales College of Medicine,
ofSurge?, University ofNottingham.,Nottingham
CardifJ U.K.
(Received 16 February 1987; accepted 7 April 1987)
Key words: Thyroid; Microsomal antigen; Thyroid peroxidase; Thyroid autoi~unity
Summary
The interaction between thyroid microsomal autoantibodies and thyroid microsomal antigen/ thyroid peroxidase (TPO) has been studied using both intact antigen preparations and their water-soluble trypsin fragments. In an analysis of sera from 30 patients with Graves’ or Hashimoto’s diseases, microsomal antibodies showed similar reactivity towards trypsin fragments (with TPO activity) and intact detergent (sodium deoxycholate, DOC)-solubilized human microsomal antigen preparations (r = 0.96). This raised the possibility that both the peroxidase-active site and the major autoantigenic site(s) of microsomal antigen were present on the same trypsin fragments. Studies with porcine TPO showed that only a few sera contained microsomal ~tib~ies which cross-reacted strongly with the porcine preparations. Further analysis was carried out by i~unoprecipitation of 1251-labelled microsomal antigen followed by SDS-PAGE and autoradiography. These studies suggest that intact human microsomal antigen (a single-chain protein with M, = 110 000) contains an intrachain loop of amino acids formed by a disulphide bridge. Trypsin treatment cleaves the antigen close to its transmembrane section and releases a water-soluble fragment (M, = 100 000), containing the intact disulphide-linked loop of amino acids. Further trypsin action causes cleavage of the peptide bonds within the loop in some preparations. Consequently, three major water-soluble trypsin fragments (M, = 100 000, 73 000 and 68 000) are formed all of which contain an intact disulphide bridge and have microsomal antibody binding activities. The integrity of the disulphide bridge in intact antigen/TPO preparations and their trypsin fragments is essential for autoantibody binding activity.
Introduction Sera from patients with autoimmune thyroid disease frequently contain thyroid microsomal antibodies (Goudie et al., 1963), and there is evidence that these autoantibodies are involved in thyroid destruction (Bottazzo et al., 1986). Recent Address for correspondence: Dr. B. Rees Smith, Endocrine l~unolo~ Unit, 7th Floor Medicine, University of Wales College of Medicine, Cardiff CF4 4XN, U.K. 0303-7207/87/$03.50
studies have indicated that thyroid microsomal antigen is an integral membrane protein, consisting of a single 110 kDa subunit (Banga et al., 1985; Hamada et al., 1985; Kajita et al., 198.5; Furmaniak et al., 1986). Furthermore, the antigen appears to be identical to thyroid peroxidase (TPO) (Czarnocka et al., 1985, 1986; Portmann et al., 1985; Kotani et al., 1986). Trypsinization of detergent-solub~ed thyroid microsomal preparations causes the release of a water-soluble fragment of thyroid peroxidase with
0 1987 Elsevier Scientific Publishers Ireland, Ltd.
enzyme activity (Ohtaki et al., 1982), and the objectives of the present study were to investigate the interaction of microsomal antibodies with the trypsin fragments of human and porcine microsomal antigen/TPO. Our studies indicate that TPO activity and the major microsomal antibody binding site(s) are located on trypsin fragments which contain a disulphide bridge. Methods Microsomal antigen preparations Human thyroid tissue was obtained from patients undergoing partial thyroidectomy for Graves’ disease and porcine thyroid tissue from a local abattoir. A microsomal fraction was prepared by differential centrifugation of thyroid homogenates (Schardt et al., 1982) and solubilized by treatment with 1% sodium deoxycholate (DOC) in 50 mM NaCl; 10 mM Tris-HCl pH 8.3 and centrifuged at 100 000 X g for 1 h at 4 o C (Kajita et al., 1985). In some experiments the DOCsolubilized preparations were partially purified by chromatography on Sephacryl S-300 (Furmaniak et al., 1986). Purification of thyroid peroxidase activity Thyroid peroxidase (TPO) activity was measured by the guaiacol method (Hosoya and Morrison, 1967) and a trypsin fragment of the enzyme purified from porcine thyroid tissue by the method of Ohtaki et al. (1982). Briefly, 30 ml of DOCsolubilized thyroid microsomes (2 units of TPO per mg of protein) were treated with trypsin (0.01%) for 40 min at room temperature and after addition of trypsin inhibitor (0.02%), dialysed overnight against 10 mM sodium phosphate buffer pH 7.4, containing 0.1 mM KI. The dialysed material was then run on a 100 ml column of DEAE-cellulose and eluted with a linear salt gradient (O-O.3 M KC1 in 10 mM phosphate; 0.1 mM KI pH 7.4; 500 ml). Fractions containing TPO activity (20 units per mg of protein: 40% recovery) were then concentrated by ultrafiltration and run (20 ml h-l) on a 2.6 X 90 cm column of Sephacryl S-300 in 10 mM phosphate; 0.1 mM KI buffer pH 6.4. This preparation contained 150 units per mg; 35% recovery during gel filtration. Further purification was then achieved by chro-
matography on a 30 ml hydroxyapatite column using a linear gradient from 10 to 200 mM potassium phosphate pH 6.8 containing 0.1 mM KI. All procedures were carried out at O-4 o C. The specific activity of the final product was 200-300 units of TPO activity per mg of protein (protein concentration determined by the method of Bradford, 1976) with an overall recovery of 20%. This represented a purification factor of 100-200 times over the initial DOC-solubilized porcine thyroid microsomes. Analysis of the purified trypsin fragment by SDS-PAGE under non-reducing conditions showed the presence of three major bands, M, = 60 000, 70000 and 90 000 stained by Coomassie blue (Fig. 5). Insufficient human thyroid tissue was available to carry out extensive purification of human TPO activity. However, DOC-solubilized human thyroid microsomes partially purified by S-300 gel filtration in 1% DOC were treated with trypsin and re-run on Sephacryl S-300 in the absence of DOC. Microsomal antigen as measured by microsomal antibody binding and TPO activity co-eluted on Sephacryl S-300 prior to trypsinization. After trypsinization, the two activities co-eluted again, but in a greater volume than that of the nontrypsinized material. Measurement of autoantibodies to thyroid microsomal antigen Plastic tubes were coated (overnight at 4 o C in 0.1 M sodium bicarbonate buffer, pH 9.2) with thyroid microsomal antigen preparations under investigation or thyroglobulin. After washing, sera to be tested for microsomal antibodies (from normal donors or Graves’ or Hashimoto patients) diluted in 150 mM NaCl; 10 mM Tris-HCI pH 7.5 containing 5 mg ml-’ bovine serum albumin and 15 pg ml-l thyroglobulin (sufficient thyroglobulin to inhibit completely the binding of any thyroglobulin antibodies in the test sera to thyroglobulin contaminating the microsome preparations) were added (250 ~1 per tube in triplicate). After 2 h at 37” C, the tubes were aspirated, washed and 0.5 ml (50000 cpm) of 1251-labelled protein A (labelled by the method of Fraker and Speck, 1978, to a specific activity of 10 PCi per pg) added. After a further incubation of 2 h at 37 o C and washing, the amount of 1251-labelled protein
17 Labelling and immunoprecipitation of microsomal antigen preparations Purified or partially purified antigen preparations were labelled to a specific activity of 10 PCi per pg of protein using the Iodogen method of Fraker and Speck (1978). Aliquots (20 PCi) were pre-adsorbed (4’C, 16 h) with 100 ~1 of Staphylococcus aureus cells (Pansorbin-Calbiothem) and then incubated (37’C, 2 h) with diluted (20-fold in diluent containing 200 pg ml-’ human or porcine thyroglobulin) serum containing high levels of thyroid microsomal antibody. Pansorbin (100 ~1) was then added and after incubation for 2 h at room temperature and centrifugation (15 000 X g, 5 min), the cell pellets were washed. Electrophoresis sample buffer (4% SDS in 0.1 M Tris-HCl pH 6.8 containing 20% glycerol) with or without 10 mM dithiothreitol (DTT) was then added, and after heating (100 o C, 3 min) the samples were analysed by SDS-PAGE (5-15% gradient gels) followed by autoradiography (Kajita et al., 1985).
A bound to the antibody-antigen complex was determined. In the presence of human serum without detectable microsomal antibody labelled protein A binding was about 2% (see legends to Figs. l-3). This was due in the main part to the interaction of labelled protein A with IgG which contaminated microsome preparations. Human serum without detectable microsomal antibody was always included in each assay to assess the level of this ‘non-specific’ binding (legend to Figs. l-3). Removal of IgG contamination from the solubilized microsomes reduced the non-specific binding to about 0.2%. Preparation of thyroglobulin Thyroglobulin was purified from the supernatants obtained after sedimentation of human or porcine thyroid microsomes by precipitation between 1.52 M and 1.76 M ammonium sulphate (Derrien et al., 1948) followed by gel filtration on Sephacryl S-300.
.
. .
0
10 Binding
20 to intact
30 HUMAN
microsomal
.
0
. .
.
40 antigen
50
60
(%)
Fig. 1. Comparison of thyroid microsdmal antibody binding to intact DOC-solubilized human microsomal antigen and intact DOC-solubilized porcine TPO. The plastic tubes used in the assays were coated with 6 and 16 ag ml-’ of protein in the case of human and porcine preparations respectively (peroxidase activity 0.7 and 2 units per mg of protein for human and porcine preparations respectively). See text for full experimental details. The results of 30 sera from patients with Graves’ or Hashimoto’s disease are shown. The two parameters showed a significant correlation (r = 0.75; P < 0.001). Labelled protein A binding with eight individual normal sera gave a mean value of 2.3% (range 1.8-2.7s) in the case of human microsomal antigen, and 2.3% (range 2.1-2.5%)
in the case of porcine TPO.
18
Effect of reduction on thyroid microsomal antigen/ TPO DOC-solubilized human thyroid microsomal antigen or purified porcine TPO preparations were incubated with DTT (final concentration 10 mM) for 15 min at room temperature. Iodoacetamide
. . . . .
.
.
(400 mM in 2 M Tris-HCl pH 8.3) was then added to a final concentration of 40 mM and incubated for 10 min at room temperature. The mixture was dialysed overnight at 4” C against 1 1 of 1% DOC containing 0.1 mM KI (in the case of intact human microsomal antigen) or 1 1 of 10 mM phosphate buffer pH 6.8 containing 0.1 mM KI (in the case of purified porcine TPO), with one change of dialysis buffer. In control experiments DTT was inactivated prior to addition to microsomal antigen/TPO preparations by pre-incubation (15 min room temperature), with a 4-fold molar excess of iodoacetamide. After dialysis, both control and reduced samples were assayed for microsomal antibody binding and TPO activity. Results Microsomal antibody binding to intact DOCsolubilized S-300-purified human and porcine microsomes showed a significant correlation (r = 0.75; n = 30; P < 0.001) (Fig. 1). Over 20 of the 30 sera studied showed strong binding to human antigen, but only four bound well to porcine anti-
.
; .. O-
. . .
OL 0 Binding
.
l -
L trypsin
fragment (%1251-protein
of
human A
microsomal
.
40
30
20
10
to
.
antigen
bound)
Fig. 2. Comparison of thyroid microsomal antibody binding to intact DOC-solubilized human microsomal antigen and the trypsin fragment of human microsomal antigen. Intact human microsomal antigen was partially purified by S-300 gel filtration in 1% DOC. The trypsin fragment was prepared by digesting this preparation with trypsin and then rerunning on S-300 in the absence of DOC. The plastic tubes used in the assays were coated with 6 pg ml-’ of protein in the case of intact microsomal antigen (peroxidase activity 6 units per mg of protein) and 10 pg ml-i of protein in the case of the trypsin fragment (peroxidase activity 4 units per mg of protein). See text for full experimental details. Results of 30 sera from patients with Graves’ or Hashimoto’s disease are shown. The two parameters showed a highly significant correlation (r = 0.96; P i 0.001). Labelled protein A binding with eight individual normal sera gave a mean value of 2.3% (range 1.8-2.7%) in the case of intact antigen and 5.4 (range 4.5-6.0%) in the case of the trypsin fragment,
Fig. 3. Comparison of thyroid microsomal antibody binding to intact DOC-solubilized porcine TPO and the purified trypsin fragment of porcine TPO. Intact porcine TPO was partially purified by S-300 gel filtration. The trypsin fragment was prepared as described in detail in the text. Plastic tubes used in the assays were coated with 16 pg ml-’ of protein (peroxidase activity 6 units per mg) in the case of intact porcine TPO and 10 pg ml-’ of protein (peroxidase activity 200 units per mg) in the case of purified trypsin fragment. Results with 30 sera from patients with Graves’ or Hashimoto’s disease are shown. The two parameters showed highly significant correlation (r = 0.93; P < 0.001). Labelled protein A binding with eight individual normal sera gave a mean value of 2.3% (range 2.1-2.5%) in the case of intact TPO and 2.2% (range 1.8-2.596) in the case of the trypsin fragment.
19
gen. Microsomal antibody binding to intact DOC-solubilized ~crosomal antigen correlated well with binding to the antigen’s trypsin fragment in the case of human (r = 0.96; n = 30; P < 0.001) (Fig. 2) and porcine (r = 0.93; n = 30; P < 0.001) (Fig. 3) material. Considerably greater antibody binding was obtained with purified trypsin fragments of porcine TPO compared with non-purified intact porcine TPO (Fig. 3), and 12 out of the 30 sera showed relatively strong binding to the porcine trypsin fragment. SDS-PAGE analysis of intact DOC-solubilized human microsomal antigen labelled with *z51 indicated that a major band with Mr = 110000 was
~unopre~ipitated by ~crosom~ antibody (data not shown, but see Kajita et al., 1985). Similar analysis with trypsin fragments of human microsomal antigen indicated that three major bands were specifically immunoprecipitated by microsomal antibody and ran with 54,s of 68 000,73 000 and 100000 under non-reducing conditions (Fig. 4A). Under reducing conditions, major specifically immunoprecipitated bands with M, = 33 000, 36 000 and 90 000 were evident (Fig. 4B). Further experiments were carried out with lZ51labelled porcine thyroid peroxidase preparations. In the case of intact porcine TPO, partially purified by gel filtration’ on Sephacryl S-300, immunoprecipitation of the ‘251-labelled material
5
1
2
3
4
5
Fig. 4. Immunoprecipitation of the lab&d trypsin fragment of human microsomal antigen. (A) SDS-PAGE under non-reducing conditions; immunoprecipitation with: lane 1= Graves’ serum (patient DO) with high levels of microsomal antibody; lane 2 = Graves’ serum (patient SC) with high levels of microsomal antibody; lane 3 = Hasbimoto serum (patient GO) with high levels of microsomal antibody; lane 4 = normal pool serum; lane 5 = untreated material (non-i~unoprecipitated); lane 6 = labelled IgG marker. (B) SDS-PAGE under reducing conditions; i~unoprecipitation with: lane 1= labelled IgG marker; lane 2 = Graves’ serum (patient DO); lane 3 = Graves’ serum (patient SC); lane 4 = Hashimoto serum (patient GO); lane 5 = normal pool serum; lane 6 = untreated material (non-immunoprecipitatcd). The data shown in 4A and 48 are typical of two and three separate experiments respectively.
20
B Fig. 5. Analysis of the purified trypsin fragment of porcine TPO by SDS-PAGE under non-reducing conditions. (A) Unlabelled material, gel stained with Coomassie blue. (B) 12’1-Labelled material immunoprecipitated with: lane 1= normal pool serum; lane 2 = Hashimoto serum (patient KF) with high levels of microsomal antibody; lane 3 = untreated material (non-immunoprecipitated). The data shown are typical of two separate experiments.
with microsomal antibody failed to result in specific enrichment of any particular protein bands (data not shown). However, when the purified trypsin fragments of porcine TPO were labelled and immunoprecipitated, all the major components of the trypsin fragments (M, = 60 000,70 000 and 90000 under non-reducing conditions) were immunoprecipitated by microsomal antibody (Fig. 5). Under reducing conditions, a different pattern of both labelled and non-labelled bands was obtained (Fig. 6) with major labelled components at M, = 26 000, 34000, 55 000 and 82 000 (Fig. 6B). Coomassie blue stained major bands at &I, = 26 000 and 55 000 (Fig. 6A). Treatment of intact DOC-solubilized human microsomal antigen with DTT caused a marked loss of TPO activity (Fig. 7). Microsomal antibody
binding activity was also markedly diminished in experiments with over 20 sera with microsomal antibody activity (Fig. 7). Similar losses of TPO activity and microsomal antibody binding activity (four out of four sera studied) were observed on reduction of purified porcine TPO with labelled protein A binding falling from 17%, 13%, 12% and 10% to about 3% (normal serum value) in all four cases. Discussion
Our studies demonstrate that human thyroid microsomal antibodies do not usually cross-react well with porcine thyroid microsomal antigen preparations (Fig. 1). The trypsin fragment of human thyroid microsomal antigen reacted with
Fig. 6. Analysis of the purified trypsin fragment of porcine TPO by SDS-PAGE under reduciag conditions. (A) Unlabelled material, gel stained with Coomassie blue. (B) ‘251-LabeIled material immunoprecipitated with: lane 1 = IabeUed IgG marker; lane 2 = normal pool serum: lane 3 = Hashimoto serum (patient RF) with high levels of microsomal antibody; lane 4 = untreated material (non-immunoprecipitated). The data shown are typical of two separate experiments.
microsomal antibodies to a similar extent to the intact DOC-solubilized antigen (Fig. 2) and contained similar levels of peroxidase activity (Fig. 2, legend). This raised the possibility that both the auto~tigenic site or sites and the active site of the peroxidase enzyme were on the same water-soluble trypsin fragment(s). Immunoprecipitation of labelled human microsomal antigen preparations and analysis by SDS-PAGE under non-reducing conditions showed that the human trypsin fragment contained three bands with M, = 68 000, 73 000 and 100000 compared with a broad band at 110000 for the intact antigen (data not shown but see Kajita et al., 1985). Under reducing conditions, the M, of intact human microsomal antigen and the 100 kDa fragment were only slightly changed whereas the 68 kDa and 73 kDa trypsin fragment components showed marked changes in
M, to 33 000 and 36 000 (Fig. 4). This indicated that two of the major components of the human trypsin fragments consisted of peptide chains linked by disulphide bridges. Consequently, we can propose a structure for human microsomal antigen and its tryptic fragments as shown in Fig. 8. Intact human microsomal antigen is a single 110 kDa integral membrane protein with at least one loop of amino acids linked by an intrachain disulphide bridge. Trypsin treatment results in cleavage at a point or points close to the transmembrane section and this gives rise to a water-soluble fragment with M, = 100 000 under non-reducing conditions. A similar-sized fragment has been described recently by Ohtaki et al. (1986). Some of the water-soluble 100 kDa fragments have at least one additions trypsin cleavage point inside the disulp~de-inked loop of amino acids
22 intact
human
microsp,m,;Ka)ntigen 50
Trypsin
S
-ki
red;ction
Fig. 8. Proposed
untreated microsomal antigen
Microsomal antigen plus inactive DTT
Microsomal antigen plus active
DTT
Fig. 7. Effect of reduction on microsomal antibody binding and TPO activities in intact DOC-solubilized human microsomal antigen preparations. The 30 sera shown are from the same Graves’ and Hashimoto patients as in Figs. 1-3. See text for full experimental details. DTT treatment has similar effects on TPO and microsomal antibody binding activities (four sera studied) in purified porcine TPO preparations (see text).
and possibly additional cleavage points outside the loop. This gives rise to the fragments with A4, = 68 000 and 73 000 under non-reducing conditions (Fig. 8). On reduction of the disulphide bridge which forms the loop, the M, of the 100 kDa fragment falls slightly to 90000 presumably as a result of conformational changes only. However, reduction of the 68 kDa and 73 kDa fragments results in a fall of 50% in their M,s as the loop of amino acids formed by the disulphide bridge has been cleaved by trypsin (Fig. 8). The purified trypsin fragments of porcine microsomal antigen reacted more strongly with microsomal antibodies than intact porcine antigen preparations (Fig. 3). The reason for this is not clear, but the difference in reactivity between the two preparations was also observed in immunoprecipitation studies. It may be related to
I f
reduction
structure of human microsomal effects of trypsin and reduction.
+
reduction-
antigen
and
the relatively low concentration of TPO in the intact preparations. Analysis of the purified porcine trypsin fragments by SDS-PAGE under non-reducing conditions followed by Coomassie blue staining indicated the presence of. three major bands at M, = 60000, 70000 and 90000 (Fig. 5). When the preparations were labelled with rz51, all three bands were immunoprecipitated with microsomal antibody. This suggested that the purified porcine trypsin preparations contained three major fragments, all of which reacted with microsomal antibody. A different pattern of bands was obtained when SDS-PAGE was carried out under reducing conditions (Fig. 6) (major labelled bands at M, = 26 000, 34000, 55 000 and 82000). This indicated that some of t-he components of the porcine trypsin fragments were linked by a disulphide bridge or bridges, in a similar way to the trypsin fragments of human microsomal antigen (Fig. 8). Reduction of human or porcine microsomal antigen/TPO preparations resulted in loss of most of their TPO and microsomal antibody binding activities. The small residual amount of both activities (Fig. 7) may well have been due to a small amount of non-reduced material remaining. Our data suggest therefore that the major microsomal antibody binding site(s) and possibly TPO activity
23
depend on the integrity of the disulphide bridge shown in Fig. 8. Acknowledgements This work was supported by RSR Ltd. and the Medical Research Council. Dr. Y. Nakajima was in receipt of a Smith and Nephew Foundation Fellowship. We are most grateful to Dr. J. Furmaniak and Dr. F. Hashim for helpful advice and discussions and to Mrs. K. Earlam for excellent secretarial assistance. References Banga, J.P., Pryce, G., Hammond, L. and Roitt, I.M. (1985) Mol. Immunol. 22, 629-642. Bottazzo, G.F. and Doniach, D. (1986) Annu. Rev. Med. 37, 353-359. Bradford, M.M. (1976) Anal. Biochem. 72, 248-254. Czamocka, B., Ruf, J., Ferrand, M., Carayon, P. and Lissitzky, S. (1985) FEBS Lett. 190. 147-152.
Czamocka, B., Ruf, F., Ferrand, M., Lissitzky, S. and Carayon, P. (1986) J. Endocrinol. Invest. 9, 135-138. Goudie, R.B. and McCallum, H.M. (1963), Lancet ii, 1035-1038. Derrien, Y., Michel, R. and Roche, J. (1948) Biochim. Biophys. Acta 2, 454-470. Fraker, P.J. and Speck, J.C. (1978) Biochem. Biophys. Res. Commun. 80, 849-857. Furmaniak, J., Davies Jones, E., Buckland, P.R., Howells, R.D. and Rees Smith, B. (1986) Mol. Cell. Endocrinol. 48, 31-38. Hamada, N., Grimm, C., Mori, H. and DeGroot, L.J. (1985) J. Clin. Endocrinol. Metab. 61, 120-128. Hosoya, T. and Morrison, M. (1967) J. Biol. Chem. 242, 2828-2836. Kajita, Y., Morgan, D., Parkes, A.B. and Rees Smith, B. (1985) FEBS Lett. 187, 334-338. Kotani, T., Umeki, K., Matsunaga, S., Kato, E. and Ohtaki, S. (1986) J. Clin. Endocrinol. Metab. 62, 928-933. Ohtaki, S., Nakagawa, H., Nakamura, M. and Yamazaki, I. (1982) J. Biol. Chem. 257, 761-766. Ohtaki, S., Kotani, T. and Nakamura, Y. (1986) J. Clin. Endocrinol. Metab. 63, 570-576. Portmann, L., Hamada, N., Heinreich, G. and DeGroot, L.J. (1985) J. Clin. Endocrinol. Metab. 61, 1001-1003. Schardt, C.W., McLachlan, SM., Matheson, J. and Rees Smith, B. (1982) J. Immunol. Methods, 55, 155-168.