Abstracts 103

HUMAN MONOCLONAL ANTIBODIES AGAINST EPITOPES ON HUMAN INSULIN OBTAINED BY EPSTEIN-BARR VIRUS (EBV) TRANSFORMATION OF LYMPHOCYTES FROM A NEWLY DIAGNOSED DIABETES TYPE I PATIENT. A. Refn and J. Zeuthen, Novo Research Institute, DK-2880 Bagsvaerd, Denmark. Peripheral blood lymphocytes (PBL) from a newly diagnosed diabetes type I patient were isolated, and cultures of lymphoblastoid cell lines (LCL) were obtained by transformation with B95-8 virus. The patient was a 15 year old girl who had only received fast-acting human insulin for 2 days before the blood sample was obtained. Supernatants from a total of 960 independent cul- tures of LCL plated in microtitre plates were screened in ELISA against RINmSF rat insulinoma cells (53 positive) and human insulin (57 positive). Two cell lines were selected for further study based on their strong reaction in ELISA with human insulin (Ins33 and Ins44). The epitopes on insulin defined by the two antibodies were analyzed in ELISAwith insulin of different species origins. Ins33 bound equally well to bovine and human insulins and somewhat weaker to porcine insulin. In contrast, Ins44 did not bind bovine insulin. Neither of the two antibodies bound human proinsulin. Based on its binding characteristics, Ins44 is suggested to bind to an epitope including the A- chain loop (A8-10). whereas the different epitope bound by Ins33 was not further defined. Attempts were made to estimate the binding affinity of the two antibodies in a competitive ELISA system, and no significant binding against human insulin could be observed (i.e. K < lo5 1 M-l).LCL are known to be unstable producers of immunoglobulin, and therefore stable secretion of the Ins33 and Ins44 antibodies was achieved by the EBV-hybridoma technique by fusion of the cell lines with the human fusion partner KR-12 (oua",HPRT-). Conclusion: The present study shows the feasibility of the production of human diabetes related autoimmune antibodies, in this case against insulin. Due to their extremely low affinity against native human insulin it is plausible that the authentic immunizing antigen is different from native human insulin.

ICA NEGATIVE FIRST DEGREE RELATIVES OF TYPE 1 DIABETICS: PREVALENCE AND RELATIVE CONSTANCY OF COMPETITI~VE ANTI-INSULIN AUTOANTIBODIES. A.G. Ziegler, I. Reske, R.A. Jackson, J.S. Soeldner and G.S. Eisenbarth. Joslin Diabetes Center, Boston, MA USA Nine-hundred-and-thirty-six ICA negative (<40 JDF units) non-diabetic first degree relatives of patients with Type I diabetes and 46 ICA negative subjects with new onset diabetes prior to insulin therapy have been studied for insulin autoantibodies (CM) using a competitive RIA. Forty-one percent (19/46) of ICA- new onset diabetics and 2.8% (26/936) of ICA- first degree relatives at first assay exceeded 39nU/ml, the upper limit of normal. Normal CIAA range is based upon 74 controls, including 14 control children less than age 5 yrs. and 7'between ages 5 and 10 yrs. Range of elevated CIAA was 42 to 1353nU/ml, mean 190+52nU/ml. Of new onset ICA- diabetic infants (5 yrs. of age 100% (lo/LO) were pos?*We-for CIAA (age 5-10 yrs. 3/6 = 50% positive; positive).

age lo-15 yrs. 6/13 = 46% positive; age >15 yrs. O/17 Of ICA negative non-diabetic relatives 11/363 = 3% offspring,

12/415 = 2.9% siblings, and 3/158 = 1.9% parents were found to have elevated CIAA levels. Follow-up sera for repeat CIAA determination (time between samples 0.4 to 5.5 yrs., mean 2.1 yrs.) has been obtained on 12 CIAA+/ICA- non-diabetic first degree relatives. positive on the second test.

Ten relatives were found to be again The correlation coefficient of the magnitude

of CIAA positivity between first and second sample was highly significant (R=O.98, p<O.OOl). In nine of the ICA-/CIAA+ relatives intravenous glucose tolerance tests (IVGTT) were performed to date. CLAA+ relatives had a significant decrease of the first phase insulin secretion (1+3 min.) in comparison to our normal control population (27+8 vs. 50 percentile, p<O.O5). In sunnnary, the remarkable constancy zf competitive insulin

104 Abstracts

autoantibodies in ICA- prediabetic subjects, their low prevalence in ICA- relatives compared to ICA- new onset diabetics, their identification of ICA- relatives with low IVGTI response , and their relation to age at onset of diabetes suggest that these antibodies will aid in the prediction of Type I DM and the rate of autoiumune beta cell destruction.

GENETIC DETERMINATION OF E3JHANCED PRODUCTTON OF INSULIN ANTV83DIES BY MTCE REJECTING INSULIN TRANSFECTED CELLS. A.G. Ziegler, D.J. Gross, M. Grinbergs and G.S. Eisenbarth. Joslin Diabetes Center. Boston. MA USA To study potential mechanisms involved in the production of antibodies to insulin, we transplanted different strains of mice with histoincompatible non-islet cells (MTZO-INS and 3T3-INS) synthesizing homologous insulin in contrast to immunization with non-transfected cells and insulin in Freund's adjuvant. The pituitary cell line MT20 and the fibroblast cell line NIH- 3T3 was transfected with the rat II insulin gene (coding sequence identical to that of the mouse insulin gene). No antibodies to insulin were found after subcutaneous injection of AtT20-control (without the integrated rat insulin gene) or different concentrations of rat insulin (50ng, 5000ng) in complete Freund's adjuvant. Agter the subcutaneous injection of living AtT2O'INS or 3T3-INS cells (10 cells/iniection containina "5Ona immunoreactive insulin) insulin antibodies (IA) rapidly appeared in 3 strains after AtT20-Ins injections (Balb/c, C3H, C57, but not SLJ mice) and in 2 strains after 3T3 injection (Balb/c, C3H). No antibodies were detected after injection of insuliln in Freunds's adjuvant.

INSULIN ANTIBODIES X+SEM i-&J/ml (NORMAL RANGE t39nU/ml) STRAIN PRE POST INJECTION

AtT20-INS 3T3-INS AtT20- Rat-Ins Rat-Ins CONTROL 5Ong+Adj SOOOng+Adj

Balb/cJ -5+5 128+11 56+13 15+8 4+9 4+8 C3H/HeJ 674 98715 60718 lT3 ST11 1376 SJL/J 6710 11Tll C57BL/6J

-773 3T7 i4 nTd. 2T5 4s4 -3T9 4TlO 21 7

Antigen presentation Within cellg enhances 'fhe response, since subcutaneous injections in adjuvant of similar and even 100x concentrations of rat insulin were not able to induce anti-insulin antibodies.

NOVEL MONOCLONAL ANTIBODIES AGAINST RAT ISLETS (RISL): INITIAL CHARACTERIZATIDN. J. Zielasek, A. Rabizadeh and G.S. Eisenbarth. Joslin Diabetes Center, Boston, Massachusetts, U.S.A. We have recently reported that cytoplasmic ICA react equivalently with both human and rat islets in acetone fixed frozen sections utilizing protein A for detection. In order to further characterize the ICA antigen of rat tissue we have produced 8 new mouse monoclonal antibodies (McAbs) against rat islets by immunizing 6-8 week old female Balb/c mice with islets from CD rats. This new series of McAbs was termed RISL (for rat islet) and showed the following characteristics: they react with islets in frozen pancreatic sections of the WF rat. Similar to I& staining with seven out of the eight McAbs was sensitive to pronase. nuo McAbs (RISL-62 and RISL-65) showed cross reactivity with CD rat kidney tissue. The other RISL antibodies show no cross reactivity with non islet pancreatic tissues. In regard to species specificity, none of the RISL McAbs reacted with NOD or human islets. On RIN tumor sections one McAb (RISL-60) showed the polar staining pattern that we previously observed with serum from all NOD mice between 1 week and 6 months of age and approximately 30% of ICA+ first degree relatives of Type I diabetics and new onset diabetics. We are currently studying the biochemical characteristics of the antigens detected by the RISL series of McAbs and evaluating their ability to block human ICA.