IMMUNOLOGY:
GAD, a Single Autoantigen for Diabetes

Type I diabetes is an autoimmune disease that affects 0.3% of the world's population. It is caused by autoaggressive T cells that infiltrate the pancreas and eventually destroy the insulin-producing b-islet cells. This results in an increase in glucose levels, which are normally kept in check by insulin. Autoimmune diabetes usually affects young people, who are then dependent on an artificial source of insulin for life. The identity of the self proteins in the pancreatic islets that target the cells for autoimmune destruction has long been debated. On page 1183 of this issue, Yoon et al. report a real breakthrough in understanding the etiology of type I diabetes (1). They show that a single self protein expressed by b-islet cells, glutamic acid decarboxylase (GAD), controls the development of diabetes in the nonobese diabetic (NOD) mouse (a good animal model of human type I diabetes).

There are two forms of GAD, GAD65 and GAD67, and both forms are expressed in brain cells (where they are involved in production of the neurotransmitter GABA) and in b-islet cells, where their function is not clear (2). GAD65 has come under the scrutiny of diabetes researchers because some of the earliest autoantibodies found in prediabetic patients are GAD-specific, although other autoantibodies, such as those directed against insulin, are also present (3, 4). Furthermore, intrathymic, intravenous, or oral administration of GAD65 has, in some instances, significantly delayed the onset of disease in NOD mice (5-9). Although intriguing, these observations still do not implicate GAD in the initiation of the disease process.

The simple and bold strategy adopted by Yoon et al. was to determine whether the development of diabetes in NOD mice required the expression of GAD. They accomplished this by generating transgenic NOD mice that expressed a GAD antisense gene exclusively in b-islet cells such that expression of both GAD isoforms was prevented in these cells (but not in brain cells). This strategy would have failed if GAD had an essential function in b-islet cells, which apparently it does not. The investigators were thus able to observe a strict correlation between the presence of GAD protein in b-islet cells and the development of diabetes. In those animals that efficiently expressed the antisense transgene, there was no b-islet GAD expression, and the mice remained free of diabetes (see the figure). In contrast, mice that poorly expressed the transgene or that expressed a transgene with irrelevant information became diabetic. Furthermore, the autoimmune inflammation of salivary glands (which do not express GAD) that is normally observed in NOD mice was not diminished by efficient expression of the antisense transgene. So, the specific absence of GAD in b-islet cells offers the cells protection from autoimmune attack. The authors go on to show that because of the absence of GAD in b-islet cells, there was no anti-GAD T cell response. T cells from GAD-less NOD mice (in contrast to those from animals expressing GAD in b-islets) did not transfer the disease when injected into T cell-deficient NOD mice. Moreover, when transplanted into diabetic NOD mice, GAD-less islet cells but not normal islets were spared from immune attack. Finally, there were fewer T cells reactive to other b-islet-specific autoantigens such as insulin in GAD-less NOD mice but not in nontransgenic control animals. These results show that GAD is the essential autoantigen that initiates the disease by activating GAD-specific T cells. As the disease progresses, T cells reactive against additional b-islet-specific antigens become activated.

The demonstration that a single self protein initiates autoimmune diabetes could have important consequences for therapeutic strategies--provided, of course, these findings can be extended to the human disease. It is likely that GAD is also the initiating autoantigen in human type I diabetes because GAD-specific autoantibodies are among the first to appear in the prediabetic phase in human patients. Transplantation of human islets rendered GAD-less by introduction of an antisense transgene in vitro might benefit diabetic patients. Alternatively, if the relevant GAD autoantigenic peptides were expressed in all tissues of the body, including the thymus, it may be possible to induce GAD-specific tolerance. It is conceivable that such an approach might eliminate type I diabetes from the human population. This is not a new idea; several groups have generated NOD mice with ubiquitously expressed transgenes. Surprisingly, in light of the Yoon results, these animals still developed diabetes. In fact, it was found that the mice were not tolerant to GAD (10, 11). One study showed that despite the presence of the entire GAD protein in hematopoietic cells, autoantigenic GAD peptides were not presented on the surface of these cells (11). Thus, the GAD in pancreatic islet cells may be proteolytically cleaved to generate a completely different set of peptides not found in other cells. Given the results presented in the Yoon report, it may be worthwhile to express the relevant GAD epitopes before the immune system develops, thereby inducing effective tolerance by intrathymic deletion of immature T cells (12, 13). This would require inserting a transgene into the germ line, clearly taboo in humans. To treat human diabetes in this way one would have to effectively induce tolerance in mature T cell populations in young individuals with a genetic predisposition to the disease. This is not easy: Tolerance induction in mature T cells is often preceded by a brief effector phase before the T cells become anergic or are deleted, which would pose a risk of accelerating the disease (14, 15).

The finding by Yoon et al. that GAD is the initiating antigen of autoimmune diabetes is a major step toward formulating new therapies to treat this disease. The invaluable NOD mouse will enable the benefits of these new therapeutic strategies to be firmly established before their application in humans.

References

1. J.-W. Yoon et al., Science 284, 1183 (1999).
2. S. Baekkeskov et al., Nature 347, 151 (1990) [Medline].
3. S. Baekkeskov et al., ibid. 298, 167 (1982) [Medline].
4. L. C. Harrison, M. C. Honeyman, H. J. DeAizpurura, Lancet 341, 1365 (1993) [Medline].
5. D. L. Kaufman et al., Nature 366, 69 (1993) [Medline].
6. R. Tisch et al., ibid., p. 72 [Medline].
7. J. F. Elliott et al, Diabetes 43, 1494 (1994) [Medline].
8. J. Tian et al, Nature Med. 2, 1348 (1996) [Medline].
9. S. W. Ma et al, ibid. 3, 793 (1997) [Medline].
10. L. Geng, M. Solimena, R. A. Flavell, R. S. Sherwin, A. C. Hayday, Proc. Natl. Acad. Sci. U.S.A. 95, 10055 (1998) [Medline].
11. D. Jeske et al., in preparation.
12. J. W. Kappler, N. Roehm, P. Marrack, Cell 49, 273 (1987) [Medline].
13. P. Kisielow, H. Bluthmann, U. D. Staerz, M. Steinmetz, H. von Boehmer, Nature 333, 742 (1988) [Medline].
14. A. Lanoue, C. Bona, H. von Boehmer, A. Sarukhan, J. Exp. Med. 185, 405 (1997) [Medline].
15. B. Rocha and H. von Boehmer, Science 251, 1225 (1991) [Medline].