Thirty nine clinically confirmed Graves’ sera (two males, thirty seven females, mean age 40 years, range 9–67 years, two new and thirty seven treated, mean duration of treatment 1.7 years, range 1 month–8 years) were obtained on a consecutive basis at the thyroid clinic after screening of diabetes by means of urine and fasting blood glucose, and anyone who showed abnormal values on the tests was excluded. Three reference ICSA positive diabetic sera (one female, 4 year of age, duration of disease less than 1 month, two males, 12 and 26 year of age, 1 and 6 years’ duration each) were obtained from the diabetic clinic and also nine healthy normal controls (all males, age range 25–34 years) who had neither diabetic and/or thyroid disease parents nor siblings were included in this study.

ICSA were checked by the indirect immunofluorescense (IF) test using rat insulinoma (RINr) cells. Briefly, the RINr cells were grown on RPMI-1640 medium (Gibco, Chagrin falls, OH) supplemented with 10% fetal calf serum (FCS) in a humidified atmosphere of 5% CO₂, 37°C and the cells were harvested with 0.05% trypsin (Difco, Detroit, MI)-EDTA in phosphate buffered saline (PBS; also containing glucose 4.0nM, tris 2%, pH 7.8). Crude immunoglobulin (Ig) fractions were precipitated from the sera after heat inactivation with polyethyleneglycol (PEG), initially mixed with 30% PEG (MW. 4000) until the final concentration of the mixture was 15% in PEG. After centrifugation at 1500 g for 60 min at 4°C, the supernatant was removed by aspiration. The assay buffer was added and recentrifuged at 1500 g for 5 min at 4°C to remove any undissolved materials. When an Ig-amount adjusted experiment is needed, IgG was prepared through protein A-sepharose CL-4B (pharmacia, uppsala, Sweden). Harvested cells (5 ×10⁵ cells for each sample) were incubated with Ig’s (50ul each) for 30 min at 4°C, then resuspended in 50ul, fluorescein isothiocyanate (FITC) labelled anti-human IgG (Behring institute, Marburg, FGR), diluted 1:12 times in PBS and incubated for 30 min at 4°C. Before and after each step, the cells were washed three times with ice-cold PBS-2% FCS. On completion of the staining procedures, % binding was obtained on each duplicate sample through FACS IV (Becton-Dickinson, Mountain view, CA).

FRTL₅ cells were grown in Coon’s modified Ham’s F-12K medium (Flow lab, Irvine, Scotland) supplemented with 5% FCS and a six hormone mixture consisting of TSH (10 mlU/ml), insulin (10 ng/ml) hydroxycostisone (10⁻⁸ M), transferrin (5 μg/ml, somatostation (10 ng/ml) and glycyl-L-histidine-lysine acetate (10 ng/ml) (20). The cells were grown in a humidified atmosphere of 5% CO₂ at 37°C and were transferred biweekly using a collagenase (Sigma, ST. Louis, MO)-trypsin-chicken serum mixture in calcium-magnesium-free Hank’s balanced salt solution (HBSS).

For B-cells, EB virus cloned human B-lymphocytes were used. These B-cells, derived from the peripheral blood of patient with Graves’ disease were originaly producing antimicrosomal antibody, but after a few months of subculturing, these were no longer producing any Ig and Ig was not detected by sensitive microplate radioimmunoassay (detection range 5.0-0.0 lug/ml) even on 1,000 times concentration of culture media over uncultured medium (RPMI-1640 medium supplemented with 10% FCS). Harvested cells were stained as the same method of ICSA above and % binding was obtained on each duplicate sample.

Harvested FRTL₅ and RINr cells were reacted with Ig’s of equal volume of packed cells for 30 min at 4°C. Supernatants were obtained after centrifugation for 30 sec at 1,000g. Each sample of Ig’s unabsorbed and absorbed were then used to stain the appropriate cells as the same method above.

Pooled T-lymphocytes from the peripheral blood of 5 normal controls (males, aged 25–34 years) were obtained through a Ficoll-Conray gradient and sheep RBC rossette formation. Collected T-cells were then incubated with Na⁵¹CrO₄ (NEN, Boston, MA), 100uCi per 1 × 10⁶ target cells, for 1 hr at 37°C in 5% CO₂ and washed three times with PBS-2% FCS. The target cells were distributed to a U-bottom 96 microwell (Linbro, McLean, VA; 5 × 10⁴ cells/ml, 100μl/well), and incubated with sera, 100μl, diluted 1:10 times in RPMI-1640 medium for 60 min at 4°C. After centrifugation for 3 min at 500 g and aspiration of the supernatant, baby rabbit complement (Cedarlane lab, Hornby, Canada; 1:2 dilution in meida, 100μl/well) was added and reincubated for 3 hrs at 15°C with measurement of radioactivity on the supernatant thereafter.

a) TSH binding inhibiting antibody (TBIAb). Determination of TBIAb was carried out using solubilized porcine thyroid membranes.^21,22) In brief, 50μl of solubilized TSH receptor and 50μl of receptor purified ^l25I-TSH were incubated for 60 min at 37°C, and after adding 350μl of assay buffer, the mixture was precipitated with 500μl of PEG and centrifuged at 1500 g for 45 min at 4°C.

The supernatant was aspirated and the radioactivity of the pellet was measured, b) Assay for antimicrosomal and antithyroglobulin antibodies. Both antibodies were checked by the hemagglutination method (Fuji Zoki, Japan) and regarded as positive when the titer was above 1 to 100. As the hemagglutination method gives only roughly graded titers, the antibodies were also determined with the microplate assay. For microplate assay, U-shaped, 96 well microplates (Falcon, Oxnard, CA) were coated with a solubilized human thyroid microsome fraction by Triton X-100 and purified thyroglobulin (Tg), 10 ug/ml each, and incubated at room temperature (RT) for 60 min. Unbound antigens were removed by aspiration and the wells were filled with PBS-1 % bovine serum albumin (BSA) for 60 min. Then, 40μl of test samples were added and incubated for another 60 min at RT. After washing the wells 5 times with PBS-1 % BSA solution, the radioactivity of each well was measured to detect the respective antibodies. Determination of the Ig amount in B-lymphocyte cultured media was also performed the same as the above procedure except the use of anti-IgG antibody instead of microsome or Tg fraction.

The Bonferroni method was used for comparison of % binding on three different cell types and the chi-square test with Yate’s correction was employed for the data in Table 1 and Fig. 5.

When the sera which showed % binding three standard deviations above the normal controls (26.5 %, Fig. 3), were regarded as positive, 9 cases of 39, twenty three point one percent were ICSA positive in Graves’ disease. Also to determine the effect of the amount of Ig, pooled high and low ICSA activity sera were titrated equal in amounts of IgG through the protein A-sepharose CL-4B column, but still showed high and low ICSA activity respectively (Fig. 1).

The individual data of positiveness measured as % binding on each of the three different cells were correlated (p>0.01, each). When the same data were rearranged to % binding on RINr and B-cells against FRTL₅ cells (Fig. 2), the % binding of individual data shows FRTL₅ % binding as greatest and then B and RINr cells in decreasing order (y= −10.6 + 0.87×, y = 4.9 + 0.67×; p>0.01, both).

Statistically significant correlations were not found when RINr % binding was compared to the titers of each organ specific antibody including TBIAb measured by solubilized procine thyroid membranes (Fig. 3), antimicrosomal and antithyroglobulin antibodies by microplate assay and the lymphocytotoxicity test by complement mediated cytotoxicity (⁵¹ Cr release) (Fig. 4). Even though there were no correlations between individual data, when the results of % binding on RINr cells and TBIAb were combined and correlated to the data of thyroid related autoantibodies (table 1), group I, namely ICSA (+), TBIAb (+) group showed a high percentage of positiveness in antithyroglobulin and antimicorosomal antibodies, 88.9 and 100% compared to group II where 22.2 and 55.6% were positive and group III, where none was positive. Also all the patients with ICSA (+) showed TBIAb values above normal which was not true in reverse (Fig. 3).

Four Graves’ and three diabetic sera which were confirmed to be ICSA positive were employed for the absorption test. Percent RINr binding and % FRTL₅ binding were obtained before and after FRTL₅ and RINr cell absorption. Even though the procedures of the absorption test are prone to be rough to be accurate, absorptions of Ig’s to FRTL₅ cells in diabetic sera were less than 10% which were significantly different from those in Graves’ sera (mean 68%, range 47.6–91.3%, p>0.01). Also, % FRTL₅ binding of diabetic sera were significantly low and within the normal range except for one case with less than one month’ duration of disease who showed % binding above normal controls (^* in Fig. 5).