Report

Control of Autoimmune Diabetes in NOD Mice by GAD Expression or Suppression in β Cells

See allHide authors and affiliations

Science  14 May 1999:
Vol. 284, Issue 5417, pp. 1183-1187
DOI: 10.1126/science.284.5417.1183

Abstract

Glutamic acid decarboxylase (GAD) is a pancreatic β cell autoantigen in humans and nonobese diabetic (NOD) mice. β Cell–specific suppression of GAD expression in two lines of antisense GAD transgenic NOD mice prevented autoimmune diabetes, whereas persistent GAD expression in the β cells in the other four lines of antisense GAD transgenic NOD mice resulted in diabetes, similar to that seen in transgene-negative NOD mice. Complete suppression of β cell GAD expression blocked the generation of diabetogenic T cells and protected islet grafts from autoimmune injury. Thus, β cell–specific GAD expression is required for the development of autoimmune diabetes in NOD mice, and modulation of GAD might, therefore, have therapeutic value in type 1 diabetes.

Type 1 diabetes, or insulin-dependent diabetes mellitus, is the consequence of progressive T cell–mediated autoimmune destruction of pancreatic β cells (1, 2). However, the initial events that trigger the destruction of β cells are incompletely understood. Several β cell autoantigens have been implicated in the triggering of β cell–specific autoimmunity (1, 3). GAD is the strongest candidate in both humans and the NOD mouse, which is considered the best animal model of the human disease (3–5). In NOD mice, GAD, as compared with other β cell autoantigens examined, provokes the earliest T cell proliferative response (4, 5). However, no unequivocal evidence exists to indicate that the β cell expression of GAD is required for the initiation of diabetes in NOD mice. To address this issue, we examined the effect of selectively suppressing GAD expression in the β cells of diabetes-prone NOD mice.

The suppression of β cell GAD expression was achieved by producing transgenic NOD mice with an antisense GAD transgene (6) for both isoforms of rat GAD cDNA (rGAD65 and rGAD67) under the control of the rat insulin promoter (RIP) (7) (Fig. 1A). Six lines of antisense GAD65.67 transgenic NOD mice were established, defined by relative amount of transgene expression (Fig. 1, B and C). The first three lines of transgenic mice with high, medium, and low levels of expression of antisense GAD65.67 in the β cells were designated H-AS-GAD-NOD, M-AS-GAD-NOD, and L-AS-GAD-NOD, respectively (Fig. 1, C and D), and the second three lines of antisense GAD65.67 transgenic mice were designated Hk-AS-GAD-NOD, Mk-AS-GAD-NOD, and Lk-AS-GAD-NOD, respectively (Fig. 1, B and C). Protein immunoblot analysis (8) revealed the complete suppression of β cell GAD expression in islets of H-AS-GAD-NOD mice, whereas moderate and low suppression was found in M- and L-AS-GAD-NOD transgenic mice, respectively (Fig. 1E). In contrast, GAD expression was detected equally in the brain tissue of transgene-negative NOD mice and the three lines of AS-GAD-NOD mice (Fig. 1E). The β cell–specific suppression of GAD expression was confirmed in H-AS-GAD-NOD mice, whereas different amounts of GAD expression were seen in transgene-negative, L-AS-GAD-NOD, and M-AS-GAD-NOD mice by immunohistochemical staining with antibodies to GAD and insulin (Fig. 1F) (9). The three lines of AS-GAD-NOD mice were indistinguishable from the transgene-negative littermates in pancreatic insulin content [H, 446 ± 46; M, 461 ± 51; L, 458 ± 47; control, 451 ± 42 (SD) (micrograms of insulin per gram of pancreas)] and plasma insulin concentrations [H, 4.1 ± 0.21; M, 4.3 ± 0.19; L, 4.6 ± 0.22; control, 4.5 ± 0.18 (SD) (nanograms of insulin per milliliter of plasma)]. Similar results regarding the suppression of GAD expression in β cells and insulin content in pancreas and plasma were obtained in the second three lines of antisense GAD65.67 transgenic NOD mice.

Figure 1

Structure of RIP-antisense GAD65.67 transgene, pedigree of antisense GAD transgenic mouse lines, the expression of antisense GAD mRNA, and the suppression of GAD expression in β cells of AS-GAD transgenic NOD mice. (A) Diagram of RIP-antisense GAD65.67 transgene structure: RIP, SV40 small t intron, and a polyadenylation site (I/A). (B) Expression of antisense GAD transcript by reverse transcriptase PCR. Two micrograms of total RNA was converted to cDNA with sense rat GAD67 primer (5′-ATGACGTCTCCTACGATACA-3′), and the cDNA was amplified with sense and antisense GAD67 primers (5′-CCCCTTGAGGCTGGTAACCA-3′). As an internal standard, hypoxanthine-guanine phosphoribosyl-transferase mRNA was amplified with the following primers: sense, 5′-GTAATGATCAGTCAACGGGGGAC-3′; and antisense, 5′-CCAGCAAGCTTGCAACCTTAACCA-3′. Lane M, 100-bp ladder. Each lane corresponds to the ear tag number shown in (C). Numbers at right are the size of amplified product in base pairs. (C) Pedigree of AS-GAD transgenic mice. Eleven positive founder mice were obtained; six mice were selected on the basis of the expression of the antisense transcript and backcrossed with NOD mice. Ear tag number: TN, transgene-negative littermates; C5, C57BL/6 mice; and NO, nontransgenic NOD mice. Transgene expression: H, high; M, intermediate; L, low; and −, no expression of transgene. (D) Northern (RNA) blot analysis (17) of antisense GAD transcripts in the first three different lines of transgenic NOD mice (H-, M-, and L-AS-GAD-NOD) and transgene-negative littermates (−). (E) Protein immunoblot analysis (8) of the suppression of GAD expression in pancreatic islets and brain tissue in the first three different lines of transgenic NOD mice (H-, M-, and L-AS-GAD-NOD) and transgene-negative littermates (−). (F) Immunohistochemical staining (9) of pancreatic islets from 10-week-old H-, M-, and L-AS-GAD-NOD mice and transgene-negative [Tg (−)] littermates. Serial islet sections were stained with either hematoxylin and eosin (HE), antibody to GAD, or antibody to insulin. Magnification, ×200.

To determine whether GAD expression in the β cells was required for the development of autoimmune diabetes in NOD mice, we monitored disease development in the three lines of AS-GAD-NOD mice and in transgene-negative littermates. None (0 of 15) of the H-AS-GAD-NOD mice developed diabetes by 40 weeks of age. In contrast, 67% (12 of 18) of the M-AS-GAD-NOD mice, 75% (12 of 16) of the L-AS-GAD-NOD mice, and 81% (17 of 21) of the transgene-negative littermates developed diabetes by the same age (Fig. 2A). We examined the islet histology of the above groups at 20 weeks of age. Over 80% of the examined H-AS-GAD-NOD islets were intact (Fig. 2, B and E), and less than 20% of the islets showed periinsulitis (Fig. 2B). In contrast, most of the M-AS-GAD-NOD and L-AS-GAD-NOD islets examined showed moderate to severe insulitis, as did transgene-negative littermates (Fig. 2, B and F). In the second three lines of AS-GAD-NOD mice, diabetes appeared in 2.8% (1 of 36) of the Hk-AS-GAD-NOD mice, 83.3% (15 of 18) of the Mk-AS-GAD-NOD mice, and 80.8% (21 of 26) of the Lk-AS-GAD-NOD mice by 40 weeks of age, whereas 85.7% (18 of 21) of the transgene-negative littermates developed diabetes (Fig. 2C). One animal in the Hk-AS-GAD-NOD group did develop diabetes; whether this small difference between the H-AS-GAD-NOD and Hk-AS-GAD-NOD groups is due to leakiness in the Hk-AS-GAD-NOD group or differences in the susceptibility genes is uncertain. We also examined the islet histology of the Hk-, Mk-, and Lk-AS-GAD-NOD mice at 19 to 20 weeks of age (Fig. 2D) and found no significant difference in the extent of insulitis from that in the H-, M-, and L-AS-GAD-NOD groups (Fig. 2B).

Figure 2

The effect of β cell–specific suppression of GAD expression on the development of diabetes and insulitis. (A) The incidence of diabetes in the first three different lines of AS-GAD-NOD mice, at the seventh generation with a NOD background. Cumulative incidence of diabetes was determined by positive glycosuria and confirmed by hyperglycemia (nonfasting blood glucose > 16.7 mM) on 2 consecutive days up to 40 weeks of age. (B) Histological examination of insulitis in H-, M- and L-AS-GAD-NOD mice and transgene-negative littermates. Histological examination of pancreatic islets at 20 weeks of age; shown are results from five randomly selected nondiabetic mice at 20 weeks of age (at least 20 islets per mouse examined). Grade: 0, normal islets; 1, mononuclear infiltration, largely in the periphery, in less than 25% of the islet; 2, 25 to 50% of islet showing mononuclear infiltration; 3, over 50% of islet showing mononuclear infiltration; and 4, small, retracted islet with few mononuclear cells. (C) Incidence of diabetes in the second three different lines of AS-GAD-NOD mice at the seventh generation with a NOD background. (D) Histological examination of insulitis in Hk-, Mk-, and Lk-AS-GAD-NOD mice and transgene-negative littermates. (E to H) Photomicrographs of representative pancreatic islet (E and F) and salivary gland (G and H) sections from H-AS-GAD-NOD mice and transgene-negative NOD littermates. Paraffin sections of pancreas or salivary gland were stained by HE. (E) H-AS-GAD-NOD pancreatic section (intact islets). (F) Transgene-negative NOD littermate pancreatic section (severe lymphocytic infiltration). (G) Salivary gland sections of H-AS-GAD-NOD mice (severe lymphocytic infiltration). (H) Salivary gland sections of transgene-negative NOD littermate (severe lymphocytic infiltration). Magnification, ×400.

Our findings indicate that β cell GAD expression is a requirement for the development of diabetes in NOD mice. The complete prevention of diabetes in H-AS-GAD-NOD mice and the near complete prevention of diabetes in Hk-AS-GAD-NOD mice are not likely to be due to a nonspecific effect of the antisense transgene incorporated into the chromosomal DNA, because low or moderate suppression of GAD expression in M-, Mk-, L-, and Lk-AS-GAD-NOD mice carrying the same antisense transgene did not result in the prevention of diabetes (Fig. 2, A and C). To examine this issue further, we developed another control for the antisense GAD transgenic NOD mice, namely, antisense transgenic NOD mice carrying the antisense endogenous murine leukemia proviral env region DNA under the control of the RIP. Endogenous retroviral env protein, a putative β cell autoantigen, is expressed in the β cells of NOD mice (10). These antisense transgenic NOD mice, unlike their GAD-suppressed counterparts, developed diabetes (79%, 15 of 19), as did the transgene-negative littermates (82%, 9 of 11) (11), even though the antisense transgene was highly expressed and effectively blocked the endogenous synthesis of viral protein. These results support the view that the prevention of diabetes in antisense GAD transgenic NOD mice is not due to the nonspecific effect of an antisense transgene incorporated into chromosomal DNA.

To determine whether β cell–specific suppression of GAD expression specifically affects β cell–specific autoimmunity, we examined the salivary gland, which also shows lymphocytic infiltration in diabetes-prone NOD mice. In contrast to the β cell, lymphocytic infiltration in the salivary gland of H-AS-GAD-NOD mice was not prevented (Fig. 2G), and sialitis was similar to that of transgene-negative littermates (Fig. 2H), indicating that autoimmunity was not affected in other tissues.

We next examined whether the suppression of GAD expression in the β cells inhibits disease development by blocking the generation of β cell–specific diabetogenic T cells. Splenocytes from 20-week-old nondiabetic female H-AS-GAD-NOD mice and age-matched nondiabetic transgene-negative littermates were transfused into 6- to 8-week-old NOD–severe combined deficiency disease (NOD.scid) mice (12). None of the NOD.scid recipients (0 of 8) of splenocytes from H-AS-GAD-NOD mice developed diabetes by 10 weeks after the transfer of splenocytes, whereas 90% (9 of 10) of the NOD.scid recipients of splenocytes from transgene-negative NOD mice developed diabetes within 9 weeks after transfer (Fig. 3A), as did NOD.scid recipients of splenocytes from acutely diabetic NOD mice. Similar results were obtained when we used splenocytes from Hk-AS-GAD-NOD mice. Thus, the generation of T cells capable of adoptively transferring diabetes is blocked in the absence of GAD expression in the β cells. In addition, we determined which T cell subsets (CD4+ and CD8+) are affected in H-AS-GAD-NOD mice. We found that the generation of both diabetogenic CD4+ and CD8+ T cells was blocked in the absence of GAD expression in the β cells (11).

Figure 3

The effect of β cell–specific suppression of GAD expression on the development of β cell–cytotoxic T cells and T cell immune responses to islet autoantigens. (A) Incidence of diabetes in 6- to 8-week-old female NOD.scidmice that received splenocytes (1 × 107 cells per mouse) isolated from 20-week-old H-AS-GAD-NOD mice (n = 8), age-matched transgene-negative littermates (n = 10), or newly diabetic NOD mice (n = 9). (Bto D) Splenic T cell proliferative response to islet antigens. Splenocytes isolated from 8-week-old (B), 12-week-old (C), and 15-week-old (D) female H-AS-GAD-NOD mice, female transgene-negative littermates, or female NOD mice were reacted with GAD peptide, recombinant human GAD65 protein, HSP60, porcine insulin, or ovalbumin, and the cells were incubated with 1 μCi of [3H]thymidine. Proliferation was determined by [3H]thymidine uptake. Data are expressed as stimulation indices (SI) ± SD of the mean from five individual mice, tested in triplicate. Cutoff value of SI was 2.0. *, P < 0.01; **, P < 0.05 as compared with transgene-negative littermates.

Intravenous or intrathymus immunization of NOD mice with GAD suppresses T cell responses to GAD, heat shock protein (HSP) 60, carboxypeptidase H, and peripherin (4, 5). To determine whether other β cell autoantigen-specific T cells developed in the absence of GAD in the β cells, we examined the proliferative response of splenocytes from 8- (Fig. 3B), 12- (Fig. 3C), and 15- (Fig. 3D) week-old H-AS-GAD-NOD mice, transgene-negative littermates, and control NOD mice to GAD and other β cell autoantigens (HSP60 and insulin) (13). In contrast to the transgene-negative control group, no proliferative response to GAD was detected in H-AS-GAD-NOD mice at any age tested. T cells only from the latter transgenic mice at 15 weeks of age showed a small but insignificant proliferative response to HSP60 or insulin (Fig. 3, C and D). Similar results were obtained when we used splenocytes from Hk-AS-GAD-NOD mice. Thus, β cell–specific suppression of GAD gene expression diminishes the T cell immune response to other β cell autoantigens as well as GAD.

The susceptibility of GAD-suppressed β cells to attack by diabetogenic T cells derived from acutely diabetic NOD mice was evaluated by transplanting GAD-suppressed islets from H-AS-GAD-NOD mice or GAD-expressing islets from young, transgene-negative male NOD mice into the renal subcapsular region of acutely diabetic NOD mice (14). All recipients (6 of 6) of GAD-expressing islets showed a recurrence of diabetes (Fig. 4A); most of the grafted islets showed massive infiltration by mononuclear cells within 1 week and were destroyed within 2 weeks (Fig. 4, B and C, bottom). In contrast, none of the recipients (0 of 7) of GAD-suppressed islets showed a recurrence of diabetes up to 40 days after transplantation, at the termination of the experiment (Fig. 4A). Furthermore, over 80% of the grafted GAD-suppressed islets remained intact, and about 20% showed periinsulitis (Fig. 4, B and C, top). Most of these islets were positively stained by antibody to insulin. When we transplanted GAD-suppressed islets from Hk-AS-GAD-NOD mice into the renal subcapsular region of acutely diabetic NOD mice, similar results were observed. In contrast, the transplantation ofenv-suppressed islets from antisense envtransgenic NOD mice resulted in their destruction within 2 weeks (11). Thus, the env-suppressed islets were not resistant to the cytotoxic effect of diabetogenic T cells, suggesting that the resistance of GAD transgenic NOD islets is a specific rather than a nonspecific effect. In keeping with these results, when splenocytes from acutely diabetic NOD mice were transfused into 6-week-old, irradiated, male H-AS-GAD-NOD mice and age- and sex-matched transgene-negative littermates, none of the H-AS-GAD-NOD mice (0 of 9) developed diabetes, whereas 71% (5 of 7) of the transgene-negative control recipients developed diabetes within 4 weeks after transfer (Fig. 4D), again demonstrating that GAD expression is required for autoimmune destruction of β cells.

Figure 4

Protection of GAD-suppressed β cells from autoimmune attack by diabetogenic T cells. (A) Prevention of the recurrence of diabetes by the transplantation of GAD-suppressed islets into the subrenal capsule of acutely diabetic NOD mice. Acutely diabetic NOD mice received islets (400 islets per mouse) from 4-week-old male H-AS-GAD-NOD (n = 7) or age- and sex-matched transgene-negative littermates (n = 6). Blood glucose was measured every other day after islet transplantation. (B) Insulitis grade in islet grafts from H-AS-GAD-NOD mice and transgene-negative littermates. The insulitis grades are described in Fig. 2B. (C) Photomicrographs of representative islet grafts from H-AS-GAD-NOD mice (top) (intact islets in the kidney capsule) and transgene-negative littermates (bottom) (massive infiltration of islets by mononuclear cells in the kidney capsule). Magnification, ×400. (D) Adoptive transfer of diabetes to H-AS-GAD-NOD mice or transgene-negative littermates by acutely diabetic splenocytes. Irradiated, 6-week-old male H-AS-GAD-NOD mice (n = 9) or transgene-negative littermates (n = 7) received splenocytes (1 × 107cells per mouse) from acutely diabetic NOD mice.

Previous studies involving GAD immunization (4,5, 15) and GAD-reactive T cells (16) support a role for GAD in the induction of autoimmune diabetes in NOD mice. Our data show that β cell–specific suppression of GAD expression is sufficient to nearly completely prevent autoimmune diabetes in NOD mice. This occurs in association with the suppression of GAD-reactive T cells. Thus, GAD expression is essential for the induction of diabetogenic T cells, and diabetogenic T cells cannot provoke diabetes in NOD mice in the absence of GAD from β cells.

  • * To whom correspondence should be addressed. E-mail: yoon{at}ucalgary.ca

REFERENCES AND NOTES

View Abstract

Navigate This Article