Tolerance induction of autoreactive T cells against pancreatic β cell-specific autoantigens such as glutamic acid decarboxylase 65 (GAD65) and insulin has been attempted as a method to prevent autoimmune diabetes. In this study, we investigate whether adenoassociated virus (AAV) gene delivery of multiple immunodominant epitopes expressing GAD500–585 could induce potent immune tolerance and persistently suppress autoimmune diabetes in NOD mice. A single muscle injection of 7-wk-old female NOD mice with rAAV/GAD500–585 (3 × 1011 IU/mouse) quantitatively reduced pancreatic insulitis and efficiently prevented the development of overt type I diabetes. This prevention was marked by the inactivation of GAD500–585-responsive T lymphocytes, the enhanced GAD500–585-specific Th2 response (characterized by increased IL-4, IL-10 production, and decreased IFN-γ production; especially elevated anti-GAD500–585 IgG1 titer; and relatively unchanged anti-GAD500–585 IgG2b titer), the increased secretion of TGF-β, and the production of protective regulatory cells. Our studies also revealed that peptides 509–528, 570–585, and 554–546 in the region of GAD500–585 played important roles in rAAV/GAD500–585 immunization-induced immune tolerance. These data indicate that using AAV, a vector with advantage for therapeutic gene delivery, to transfer autoantigen peptide GAD500–585, can induce immunological tolerance through active suppression of effector T cells and prevent type I diabetes in NOD mice.
The etiology of autoimmune diabetes (type I diabetes) is both complex and multifactorial (1, 2). Both CD4+ and CD8+ T cells comprise the effector arm, with underlying functional defects in APC (macrophages, dendritic cells, B lymphocytes) shown to be essential components in the selection and activation of the autoimmune repertoire (3, 4). Autoreactive T cells infiltrate the pancreas and are targeted against multiple autoantigens, including glutamic acid decarboxylase (GAD)3 and insulin, which are known to be major β cell autoantigens that play an important role in the development of type I diabetes (5, 6). The destruction of β cells apparently entails both necrotic and apoptotic events in response to the invasion of the autoreactive T cells (7, 8).
The NOD mouse is one of the best animal models for human type I diabetes; it spontaneously develops the disease with many features in common with human type I diabetes. Based on the etiology of this disease, many autoantigen-based strategies have been used to tolerize autoreactive T cells and prevent autoimmune type I diabetes in NOD mice. However, the selection of autoantigen and the need for repeated administration limited the use of these strategies. For GAD65, which has been implicated as a strong candidate both in human and in NOD mouse in triggering of β cell-specific autoimmunity, there are different reports. Some results showed that i.m. injection of a plasmid encoding GAD65 resulted in the prevention of diabetes (9, 10, 11). However, others showed that injection of a plasmid encoding GAD65 was not effective (12, 13, 14). Some work even suggested that certain GAD65 peptides can accelerate the disease (15). These different results might be due to different route of Ag administration, different amounts of autoantigen gene expression in the vaccinated mice, different peptide usage and different epitope presentation in certain local microenvironment, or different time point of intervention. Generally, intervention before the onset of insulitis can result in fine preventive effects. However, in clinical settings, it is not easy to find the potential patients before the onset of insulitis.
Of particular importance are the findings that recombinant adenoassociated virus (rAAV)-mediated gene delivery into skeletal muscle can achieve sustained and stable production of secreted proteins (16, 17). rAAV vectors become increasingly recognized as having some superiority to other viral and nonviral gene delivery systems with regard to safety, efficiency, lack of need for repeated viral administration, duration of action without known pathology, and only the occasional induction of modest immune responses (18, 19). The advantages of rAAV would make it valuable for clinical use in type I diabetes, if administration of rAAV-expressing β cell autoantigen proteins can tolerize the autoreactive T cells and prevent the development of diabetes.
In this study, to induce potent immune tolerance and to obtain persistent therapeutic effects for autoimmune diabetes, we first analyzed the epitope character of GAD65 and selected one peptide, GAD500–585, as an immunotherapy target based on the fact that GAD500–585 is composed of multiple immunodominant epitopes such as previously described I-Ag7-restricted GAD65 determinants 509–528 and 524–543, and Kd-restricted GAD65 determinants 546–554, and that young NOD mice pretreated with 509–528, 524–543, or 546–554 all showed reduced self reactivity and delayed onset of type I diabetes (20, 21, 22). Then we developed rAAV expressing GAD500–585. The role of rAAV/GAD500–585 injection in preventing or delaying the development of insulitis and diabetes in NOD mice and the possible mechanisms were finally studied.
Materials and Methods
Plasmid construction, viral packing and production, cellular transfection, and transduction
The rAAV (serotype 2) vector plasmids (obtained from National Molecular Virology Lab of China) are depicted diagrammatically. Murine cDNAs for GAD500–585 were cloned into pSNAV plasmid. rAAV2 production, titer determination, and infectivity were performed, as described (23). Transfection and transduction of BHK-21 cells (baby hamster kidney cell line) were performed, as described (17). A rAAV virus encoding enhanced GFP was also produced for viral control use (Fig. 1).
Peptides and Ag
GAD peptides 509–528, 524–538, 530–543, 539–553, 546–558, 570–585, and 546–554 were synthesized at Genemed Synthesis, and were purified using HPLC. Peptide purity was determined by capillary electrophoresis, and the amino acid composition was verified by mass spectrometry. An irrelevant Kd-restricted 9-mer peptide PA was kindly gifted by S. Wang (Institute of Basic Medical Science, Beijing, People’s Republic of China) and was used as negative control for CD8+ T cell determinant 546–554. OVA was purchased from Sigma-Aldrich.
Recombinant GAD500–585 protein was produced from Escherichia coli BL-21 containing the bacterial expression vector pET28a+ vector. This vector contains a 6× histidine tag. Recombinant protein was purified on Ni-NTA beads, as described in the product literature (Qiagen). The presence and purity of GAD500–585 were confirmed by SDS-PAGE and Western blot analysis using anti-GAD mAb (BD Pharmingen). An irrelevant his-tag Ag (UP), which is also prepared and purified from E. coli BL-21 in an identical manner to the rGAD500–585, was used as an edotoxin control.
Specific pathogen-free NOD MarTac mice and NOD-scid mice (Taconic Farms) were housed and bred under specific pathogen-free conditions. In our colony, at the Beijing Institute of Basic Medical Science, female NOD MarTac mice become clinically diabetic starting at ∼12 wk of age. The incidence of insulin-dependent diabetes mellitus in these mice is >85–90% by 27 wk of age. Blood glucose levels were determined biweekly or every other day (for NOD-scid recipient), with animals considered type I diabetes when levels exceeded 11.3 mmol/L.
rAAV vector administration
Seven-week-old female NOD mice were injected i.m. into the caudal muscle of the pelvic limb. These injections used 100 μl of saline, saline containing 3 × 1011 infection units of either rAAV or rAAV/GAD500–585 per mouse.
In vitro proliferation assay
Splenocytes were collected from NOD mice 3–5 wk later after vectors or saline injection. Single cell spenocytes (5 × 105) were cultured in 100 μl of RPMI 1640 medium supplemented with 1% horse serum and 10−5 mol/L 2-ME in a 96-well microplate in the presence of 20 μg/ml mouse rGAD500–585 protein or OVA control for 72 h. For peptide-induced proliferation assay, 8 × 105 splenocytes were cultured in 200 μl of RPMI 1640 medium supplemented with 1% horse serum and 10−5 mol/L 2-ME in the presence of 50 μg/ml different peptide, respectively, for 96 h. The cells were pulsed with [3H]thymidine (1 μCi/well) for the last 16–18 h. The proliferation was quantified by determining the [3H]thymidine incorporation using a liquid scintillation counter (PerkinElmer Wallac Trilux 1450).
Phenotypic analysis of isolated splenocytes was performed on a FACScan. All Abs were purchased from BD Pharmingen. In this study, we used FITC-conjugated anti-mouse CD4 mAb (H129.19) and PE-conjugated anti-mouse CD25 mAb (3C7). T cells were analyzed by two-color staining with corresponding combinations of mAbs. All flow cytometric analyses were performed using appropriate isotype controls.
Assay of Abs against GAD500–585
Sera were collected from NOD mice at the time when they were sacrificed. rGAD500–585 protein (5 μg/ml, in 0.1 mol/L NaHCO3, pH 8.5) was coated onto a 96-well microplate at 4°C for 20–22 h. Then the plate was blocked with 10% FCS for 2–3 h at 37°C. After washing, serially diluted sera were added onto the plate for 1 h at 37°C. Unbinding Ab were washed away. Then 2 μg/ml goat anti-mouse IgG, goat anti-mouse IgG1, or goat anti-mouse IgG2b mAb (all purchased from Sigma-Aldrich) were added and were incubated for 1 h at 37°C. After extensive washing, peroxidase conjugate rabbit anti-goat Ab (Sigma-Aldrich, 1/3000 diluted) were added and incubated for another hour at 37°C. Finally, the color was developed by incubation with o-phenylenediamine. The OD was read at 492 nm with an ELISA reader (Bio-Rad). Ab titer was determined as multiplying the value of negative control by 3.
ELISA of cytokine production
Lymphocyte cytokine secretion in response to GAD500–585 autoantigen was determined, as described (24). Briefly, splenocytes (5 × 105) were incubated in 96-well flat-bottom microtiter plates in the presence of 20 μg/ml rGAD500–585 protein. Supernatants were harvested after 48 h. The levels of IL-4, IFN-γ, and IL-10 were determined in triplicate in 0.1 ml of supernatant by sandwich ELISA. Briefly, purified rat anti-mouse IL-4 (11B11; 4 μg/ml), rat anti-mouse IL-10 (JES5-2A5; 4 μg/ml), or rat anti-mouse IFN-γ mAb (R4-6A2; 4 μg/ml) were coated overnight on 96-well microtiter plates at 4°C. Then the plates were blocked with 10% FCS for 2–3 h at 37°C. After washing, supernatants were added in triplicate to the plate for 1 h at 37°C. Then after washing, biotin rat anti-mouse IL-4 (BVD6-24G2; 4 μg/ml), biotin rat anti-mouse IL-10 (SXC-1; 4 μg/ml), or biotin rat anti-mouse IFN-γ (XMG1.2; 4 μg/ml) Ab were added to the plate, respectively, and were incubated for another hour at 37°C. Thereafter, unbinding Abs were washed away, followed by the addition of avidin-HRP (1/1000 diluted) (all Abs were obtained from BD Pharmingen). Plates were incubated for 1 h at 37°C. Finally, the color was developed by incubation with o-phenylenediamine. The OD was read at 492 nm with an ELISA reader (Bio-Rad). Standard curves were established to quantitate the amount of the respective cytokines in the culture supernatants. TGF-β was measured by sandwich ELISA using a Quantikine kit (R&D Systems).
ELISPOT kits were all purchased from U-CyTech. Plates were coated overnight at 4°C with anti-mouse IL-4, IFN-γ, or anti-mouse IL-10 Ab. Nonspecific binding site was blocked by incubation with 1% BSA in PBS. Thereafter, splenocytes preincubated with medium control or with different peptide are added to the ELISPOT plate for a further incubation (20 h) to allow spot formation. Then the cells were removed and the residual cells were lysed with ice-cold water for 10 min on melting ice. After extensive washing, 100 μl of biotinylated anti-mouse IL-4, IL-10, or biotinylated anti-mouse IFN-γ detector Ab was added to each well and incubated 1 h at 37°C or overnight at 4°C, followed by extensive washing and the addition of 50 μl of φ–labeled anti-biotin Ab solution for 1 h at 37°C. Finally, after washing, 30 μl of activator solution was added to each well. The plate was incubated in the dark until clear spots developed. The reaction was stopped by rinsing the wells with tap water. The spots were counted using an immunospot image analyzer (Bioreader 4000 PRO-X).
Adoptive transfer of diabetes
Seven-week-old female NOD mice were injected with AAV, rAAV/ GAD500–585, or left untreated. At 20 wk of age, splenocytes (1 × 107 cells) from rAAV- or rAAV/GAD500–585-treated mice were mixed with splenocytes (1 × 107 cells) from uninjected, acutely diabetic NOD mice and given i.v. into the tail veins of 6- to 8-wk-old NOD-scid mice. Age-matched NOD-scid mice receiving only 107 diabetic splenocytes were used as positive control. Recipient mice were monitored for the development of diabetes up to 7 wk.
Mice that had been immunized with saline, saline containing of rAAV or rAAV/GAD500–585, were sacrificed at 12 wk of age, and the pancreas was removed. Each pancreas was fixed with 10% buffered Formalin, embedded in paraffin, sectioned at 4.5 μm, and stained with H&E (25). Insulitis grade was determined as follows: 0, normal islet; 1, mononuclear infiltration, largely in the periphery, in <25% of the islet; 2, 25–50% of the islets showing mononuclear infiltration; 3, >50% of the islet showing mononuclear infiltration; and 4, small, retracted islet with few mononuclear cells.
Statistical analyses were assayed by Student‘s t test or the log-rank test. A p value of 0.05 or less was considered to be statistically significant.
Prevention of autoimmune diabetes and insulitis in NOD mice by injection of rAAV expressing GAD500–585
We cloned the mouse GAD500–585 cDNA into the AAV vector. The expression of GAD500–585 protein was confirmed by sandwich ELISA (Fig. 1). To determine whether rAAV/GAD500–585 immunization could prevent the development of autoimmune diabetes in NOD mice, we i.m. injected female NOD mice at 7 wk of age with saline, saline containing rAAV or rAAV/GAD500–585, and examined the development of overt diabetes. Only 27% (4 of 15) of mice that received rAAV/GAD500–585 developed diabetes, whereas 83% (10 of 12) and 88% (7 of 8) of mice that received rAAV or saline, respectively, developed diabetes by 31 wk of age. These data indicated that rAAV/GAD500–585 administration prevented the onset of type I diabetes in NOD mice (p < 0.01, compared with either saline or AAV vector-alone groups) and the prevention is Ag specific, not simply a nonspecific effect of rAAV injection (Fig. 2).
At the same time, effects of rAAV/GAD500–585 immunization on the progression of insulitis were also examined. We found that a total of 62.4% of the examined islets from rAAV/GAD500–585-injected mice was intact, whereas >50% of the islets from rAAV- or saline-injected mice showed moderate to severe insulitis at 12 wk of age (Fig. 3). The results showed that rAAV/GAD500–585 administration efficiently prevented the progression of insulitis in NOD mice.
rAAV/GAD500–585 immunization of NOD mice tolerizes T cells’ immune response to GAD500–585 and induces Th2 immune response
To determine the mechanisms by which rAAV/GAD500–585 prevented the development of diabetes, we first examined whether rAAV/GAD500–585 administration suppressed, anergied, or deletioned the autoantigen-specific T cells. The T cell recall proliferation assay in response to GAD500–585 Ag was conducted in vitro. Our results showed that splenocytes collected from rAAV/GAD500–585-immunized mice showed lower response to GAD500–585 stimulation. In contrast, GAD500–585 induced strong proliferation of splenocytes collected from rAAV- or saline-injected NOD mice (p < 0.05). Splenocytes from all three groups showed no proliferative response against OVA, an Ag unrelated to β cell proteins. These results imply that GAD500–585-reactive T cells are tolerated in rAAV/GAD500–585-injected mice (Fig. 4), and the tolerance is Ag specific, as splenocytes from all groups respond to Con A stimulation in the same degree (data not shown).
Previous studies have demonstrated that GAD65 or its peptide-specific therapy prevented type I diabetes in NOD mice by inducing regulatory Th2 cells (26, 27). To determine whether the protection provided by rAAV/GAD500–585 immunization was mediated by Th2-regulatory cells in our study, the humoral immune response against GAD500–585 and the cytokine profile of splenocytes in response to GAD500–585 stimulation in vitro were subsequently examined. The amount of total anti-GAD500–585 IgG in the serum of rAAV/GAD500–585-injected mice was increased when compared with those of rAAV-injected controls. When the isotypes of the IgG Abs to GAD500–585 were measured, the amount of IgG1 subtype was specifically increased in the serum of rAAV/GAD500–585-injected mice, whereas the amount of IgG2b subtype did not change compared with those of rAAV-injected mice (Fig. 5), indicating that the rAAV/GAD500–585 immunization enhanced the Th2 immune response. At the same time, we analyzed the level of IFN-γ, IL-4, and IL-10 produced by splenocytes following GAD500–585 stimulation in vitro. We found that the production of IL-4 and IL-10 was significantly up-regulated, while the production of IFN-γ was significantly down-regulated in rAAV/GAD500–585-treated mice when compared with those in rAAV-treated controls. These results indicated that rAAV/GAD500–585 injection induced a Th2 shift in NOD mice (Fig. 6).
In addition to examining the cytokine profile related to Th1/Th2 balance, another potent immunoregulatory cytokine, TGF-β, which is secreted by Th3 or by T regulatory type 1 cells regulators, was also examined. The production of TGF-β was significantly increased in rAAV/GAD500–585-injected mice as compared with that in rAAV-treated controls (p < 0.01). These data indicated that the activity of autoreactive T cells might also be actively suppressed by the up-regulated TGF-β after rAAV/GAD500–585 administration (Fig. 6).
Peptides 509–528, 546–554, and 570–585 play important roles in rAAV/GAD500–585 injection-induced immune tolerance
Our above results have shown that rAAV/GAD500–585 gene therapy prevented insulitis and diabetes in NOD mice, and the protection was marked by the inactivation of GAD500–585-specific T cells and the enhanced Th2 response. However, as GAD500–585 is composed of a panel of determinants and they are known to induce both regulatory and effector cells, the potential mechanisms by which rAAV/GAD500–585 immnization induced immune tolerance remain to be determined. So, we thereafter synthesized all known determinants covering the 500–585 region of GAD and tested whether these determinants differ in their capability in inducing Ag-specific tolerance using proliferation and cytokine assay. We found that, in rAAV/GAD500–585-treated mice, the T cell immune response to peptide 509–528 was significantly diminished, while significant proliferative response was observed to peptide 570–585. It is quite unexpected that peptide 524–538 only induced modest, but not significant proliferative response in rAAV/GAD500–585-treated mice, as this peptide is known to elicit regulatory response. We did not find significant proliferative response to peptides 530–543, 539–553, and 546–558 in both groups; however, T cell immune responses to these three peptides were all slightly diminished in rAAV/GAD500–585-treated NOD mice. No evident proliferative response was detected to the CD8 T cell determinant 546–554 in proliferation assay (Fig. 7).
As we could not detect proliferation to peptide 546–554 in T cell recall assay, an ELISPOT assay was conducted, as described, to find whether this MHC I-restricted GAD peptide could induce an Ag-specific tolerance (22). Spleen cells from both groups were cultured with peptide 546–554 or control peptide (PA) and then were tested for IFN-γ production using the ELISPOT assay. We found that peptide 546–554 was able to incite Ag-specific IFN-γ production, as demonstrated by the numerous IFN-γ spot-forming cells seen upon restimulation of NOD effectors. And we found that the IFN-γ production in rAAV/GAD500–585-treated mouse was significantly decreased when compared with that in AAV-treated control NOD mice. Only few IFN-γ spot-forming cells were found in splenocytes collected from AAV- or rAAV/GAD500–585-treated NOD mice in response to control peptide PA (Fig. 8 A). P815 cells, which express no MHC class II molecules, severed as the APC in IFN-γ ELISPOT assay. We found no significant difference between assays in which the APC were prepulsed with peptide and washed, before being added to the effectors, and those assays in which the peptide remained throughout the culture period (data not shown). These results demonstrated that the p815 cells function as the APC in the ELISPOT assay and, given that the only MHC molecule shared between p815 and NOD mice is H-2Kd, it is clear that the response is mediated by MHC class I-restricted T cells.
rAAV/GAD500–585 administration induces regulatory cells
To determine whether administration of rAAV/GAD500–585 protected NOD mice from the development of diabetes by regulatory cells, we used an adoptive transfer model. Splenocytes (1 × 107 cells) from rAAV- or rAAV/GAD500–585-treated mice were mixed with splenocytes (1 × 107 cells) from uninjected, acutely diabetic NOD mice and given i.v. into the tail veins of 6- to 8-wk-old NOD-scid mice. Age-matched NOD-scid mice receiving only 107 diabetic splenocytes were used as positive control. As expected, 100% NOD-scid mice cotransferred with splenocytes from rAAV-treated mice and diabetic splenocytes developed diabetes 16 days posttransfer, a diabetes incidence not different from that of diabetic-alone group. However, only 50% of the NOD-scid mice cotransferred with splenocytes from rAAV/GAD500–585-treated mice and diabetic splenocytes developed diabetes 40 days posttransfer, and 25% of these NOD-scid mice are still free of diabetes 49 days after transfer. These results showed that splenocytes collected from rAAV/GAD500–585-immunized mice delayed the onset of adoptively transferred diabetes in NOD-scid mice (p < 0.01 by log-rank test), and suggested that administration of rAAV/GAD500–585 generated protective regulatory cells (Fig. 9).
CD4+CD25+ T cells are up-regulated at later stage in rAAV/GAD500–585-immunized mice
rAAV/GAD500–585 immunization induced regulatory T cells. Among different regulatory T cell populations, CD4+CD25+ T cells have been implicated as a highly specialized T cell subset that is pivotal in immunoregulatory response. So, we finally followed the level of CD4+CD25+ T cell population at different time by flow cytometry. Our results showed that there was no significantly increased CD4+CD25+ T population shortly after rAAV/GAD500–585 injection (12 wk of age). However, when they are detected at later stage (30 wk of age), the percentage of CD4+CD25+ T cells was significantly increased in rAAV/GAD500–585-injected mice as compared with that in rAAV-treated controls (Fig. 10). These results indicated that rAAV/GAD500–585 administration promoted the development of CD4+CD25+ T cells at later stage and implied that the up-regulation of these T subtypes might act as an immune tolerance maintainer in NOD mice.
Type I diabetes mellitus results from T cell-mediated autoimmune destruction of pancreatic β cells. Among the β cell autoantigens that have been implicated in triggering β cell-specific autoimmunity, GAD65 and insulin have been identified as major candidate. Tolerization against these autoantigens has been attempted as a method for the prevention of the disease (6, 28). It has been reported that administration of GAD65 or its peptide to NOD mice by a variety of routes can tolerize the T cell-mediated immune response against pancreatic β cells, resulting in the prevention or delay of the development of insulitis and diabetes (11, 26, 29, 30). In many cases, the preventive effect was found to be associated with a Th2 shift (6, 28). However, some other studies gave negative reports (12, 13, 14, 15).
In this study, to induce potent immune tolerance and to obtain persistent therapeutic effects for type I diabetes, we analyzed the epitope character of GAD65 and developed rAAV expressing GAD500–585. Skeletal muscle transduction of 7-wk-old female NOD mice with this vector resulted in significant therapeutic effects. A single injection of this vector quantitatively reduced pancreatic insulitis, and efficiently prevented diabetes in NOD mice. The protection is marked by the tolerance of GAD500–585-specific T cells and the enhanced GAD500–585-specific Th2 response. Peptide involved in vitro proliferation assay, and cytokine assays showed that peptides 509–528, 546–554, and 570–585 played important roles in rAAV/GAD500–585 administration-induced immunotolerance. rAAV/GAD500–585 immunization also increased TGF-β production and produced protective regulatory cells and up-regulated CD4+CD25+ T population at later stage. These results indicated that using rAAV, a vector with advantage for therapeutic gene delivery, to transfer autoantigen GAD500–585 could induce potent immunological tolerance through active suppression of effector T cells and prevent type I diabetes in NOD mice.
The selection of autoantigen is one of the key factors that affect the intervention effects in Ag-specific therapy for autoimmune disease. Although lots of data have shown that GAD65 was an important Ag in triggering autoimmunity at early stage in NOD mice (20), different groups using GAD65 or its peptides as autoantigen to induce Ag-specific tolerance have given inconsistent reports (9, 10, 11, 12, 13, 14). The work of Cetkovic-Cvrlje et al. (15) suggests that retardation or acceleration of disease in NOD mice after GAD65 administration depended on the peptides used. These data indicated that correct peptide selection and subsequently appropriate epitope or epitopes presentation in certain local microenvironment play important roles in inducing immune tolerance. Based on this understanding, to induce potent immune tolerance and prevent type I diabetes in NOD mice, we analyzed the epitope character of GAD65 and selected one peptide, GAD500–585, as an immunotherapy target. GAD500–585 peptide is composed of many important early CD4+ T and CD8+ T cell-inducing determinants, such as previously described I-Ag7-restricted GAD65 determinants 509–528 and 524–543, and Kd-restricted GAD65 determinant 546–554. Young NOD mice pretreated with 509–528, 524–543, or 546–554 determinants all resulted in reduced self reactivity and delayed onset of type I diabetes (21, 22). The role of GAD500–585 peptide in preventing the development of insulitis and diabetes in NOD mice was subsequently examined. As expected, rAAV/GAD500–585 immunization induced potent immune tolerance and enhanced Th2 immune response, thus efficiently inhibiting the progression to overt type I diabetes in NOD mice. These results indicated that GAD500–585 could be used as an ideal immunotherapy target in preventing type I diabetes.
Apart from peptide selection, sustained and appropriate extra protein expression is another important factor in Ag-specific therapy. In this study, we selected AAV as gene delivery system based on the fact that AAV can mediate sustained extra protein expression and that this vector is safe, with no or only limited toxicity. As expected, a single injection of female NOD mice with rAAV/GAD500–585 quantitatively reduced pancreatic insulitis and markedly reduced the development of diabetes without known pathology. The advantages of AAV vectors have made them a promising vector for therapeutic gene delivery: 1) They are safe. One of the important concerns about gene therapy is the probability that recombinant virus vectors may cause or contribute to cancer in animals or humans. However, Kay and Nakai (18) and others recently argue that AAV vectors are safe and this concern is not founded. First, unlike retroviral or adenovirus vectors, AAV-mediated vector integration is relatively uncommon (31). Second, unlike retroviral vectors, AAVs are less likely to activate a gene that they insert next to and they do not contain the protein machinery needed to cause host chromosomal DNA breaks. Third, data from hundreds of normal mice treated with this vector did not show any evidence that the vector caused cancer. And finally, data from clinical trials have shown an excellent safety profile of AAV vector (32). Although additional long-term safety studies are supported, based on the infrequent integration efficiency and the quiescent nature of the target tissues, Kay and Nakai (18) believe that the risk of cancer in current AAV trials is negligible. 2) AAV vectors are much more efficient in gene transferring and expressing than other viral gene delivery systems. First, AAV vectors transduce a variety of somatic tissues, including liver, CNS, and skeletal muscle, and mediate long-term and steady transgene expression. In contrast, transgene expression by adenovirus often gradually declines even when high capacity adenoviral vectors are used (31, 33, 34, 35). Second, unlike retroviral vectors, AAVs can infect both dividing and nondividing cells (36). 3) AAV vectors only cause low toxicity and limited immune response. Rabinowitz and Samulski (19) reported that AAV vectors did not cause evident immune response or toxicity. Recent data showed that on certain conditions, neutralizing Abs to the wild-type AAV2 capsid may have developed. However, data suggest that transduction is not blocked by these neutralizing Abs, because subjects with high titer pretreatment-neutralizing Abs all had evidence on muscle biopsy for gene transfer or expression or both (32). In current study, only humoral immune response was induced by AAV2 vector, a different pattern from adenovirus (37). The advantages of AAV vector and the therapeutic value of rAAV/GAD500–585 in this study make our AAV recombinant-expressing GAD500–585 a good candidate for preventing type I diabetes in clinical trials.
Following the selection of autoantigen peptide and the gene delivery system, the time point of intervention and the dosage of rAAV/GAD500–585 vector are the other two questions to be discussed. Considering the phenomenon of epitope spreading in autoimmune diabetes, intervention before the onset of autoimmune response usually has better therapeutic results. Lots of data have shown that type I diabetes could be inhibited in NOD mice immunized with GAD65 or GAD65-derived peptides at an age when insulitis had not yet developed (3–4 wk of age). However, in clinical settings, because it is not easy to find the potential patients before the onset of insulitis, it is necessary to explore effective ways to inhibit type I diabetes at various preclinical stages. In this study, we examined the therapeutic effects of rAAV/GAD500–585 at 7 wk of age, and our results showed that intervention at this point also prevented type I diabetes in NOD mice. As insulitis is already present in NOD mice at 7 wk of age, the mechanisms by which intervention at this time point prevented diabetes can be explained as follows: 1) rAAV/GAD500–585 administration tolerated autoreactive T cells in the peripheral immune system and inhibited the further invasion of T cell into the pancreas. In the absence of T cell reactivity to GAD500–585, autoimmnity and damage to β cell do not develop, thus preventing diabetes in NOD mice; 2) rAAV/GAD500–585 administration induced Th2 regulatory cells. These Th2 cells inhibited the further damage to β cells by suppressing the pathogenic role of Th1 cells. The protective role of Th2 regulatory cells in NOD mice at a late preclinical stage has also been reported by many other groups. For example, the data of Tisch et al. (30) showed that GAD65-specific peptide immunization could block the progression to overt diabetes in NOD mice at a late preclinical stage by inducing regulatory Th2 cells. 3) As Th2 cytokines have been shown to suppress immune response to neighboring Ags that arise by epitope spreading (38), we postulate that rAAV/GAD500–585 administration at this stage might also offer protective effects through blocking the spreading of autoimmunity to other β cell autoantigens.
In consideration of the influence of rAAV/GAD500–585 dosage on disease expression, three different dosages (1, 2, 3 × 1011 IU/mouse) were used in this study. Low incidence of diabetes, 27% (4 of 15), was obtained only in high dose (3 × 1011 IU/mouse) groups, indicating that administration of rAAV/GAD500–585 with this dosage can result in significant preventive effects.
To know the mechanisms by which type I diabetes was prevented in this assay, we first determined whether the immunization of NOD mice with rAAV/GAD500–585 especially affected β cell-specific autoimmunity. Our result of in vitro T cell recall proliferation assay showed that splenocytes collected from rAAV/GAD500–585-treated mice showed diminished respond to GAD500–585, while GAD500–585 protein stimulated the proliferation of splenocytes collected from saline- or AAV vector-treated mice. These data indicated that GAD500–585-reactive T cells are tolerated rather than anergied, or deletioned in rAAV/GAD500–585-injected mice (Fig. 4).
Lots of data have shown that autoimmune response in NOD mice resulted from the loss of immunoregulation with a dominance of pathogenic Th1 cells over preventive Th2 cells (39, 40), while autoantigen-based immune therapy can delay or prevent the onset of diabetes by inducing a Th2 response and inhibiting the Th1-mediated β cell destruction in a general bystander manner (26, 41, 42). To test whether rAAV/GAD500–585 immunization also enhanced Th2 response in our study, the humoral immune response against GAD500–585 and the cytokine profiles of splenocytes in response to GAD500–585 were examined in NOD mice. There were increased total IgG Abs to GAD500–585, especially increased anti-GAD500–585 IgG1 subtype, and relatively unchanged anti-GAD500–585 IgG2b subtype in rAAV/GAD500–585-injected mice as compared with those in rAAV- or saline-injected mice (Fig. 5). These results demonstrate that rAAV/GAD500–585 injection enhanced Th2 response in NOD mice. The increase in anti-GAD Abs may synergize the preventive effects, as administration of anti-GAD Ab has been shown to prevent the development of diabetes in NOD mice (43). At the same time, results of cytokine assays showed that rAAV/GAD500–585 administration up-regulated IL-4 and IL-10 production, whereas it down-regulated the production of IFN-γ in rAAV/GAD500–585-injected mice (Fig. 6). These results indicated that rAAV/GAD500–585 injection induced a Th2 cell function. These data indicated that the protection conveyed to NOD mice by rAAV/GAD500–585 injection resulted from active suppression by Th2 response rather than from deletion or anergy of GAD500–585-specific diabetogenic T cells because in recall proliferation assays, the T cells proliferated against GAD500–585. In such a scenario, early immunization with a secreted protein might lead to a predominantly MHC class II context for the presentation of the Ag and subsequently elicit primarily a Th2 response. Because the progression to diabetes depends upon a Th1 predominance, this early drive toward a Th2 predominance could offer some protective effects.
TGF-β is another cytokine that is secreted by a different type of regulator T cells, Th3, and has been shown to have potent immunoregulatory effects. The importance of TGF-β in immunoregulation and tolerance maintenance has been increasingly recognized (44). In this study, we found that secretion of TGF-β in rAAV/GAD500–585-treated mice significantly increased (Fig. 6). The suppressive properties for TGF-β may partially explain the persistent preventive role of rAAV/GAD500–585 administration in this study.
rAAV/GAD500–585 gene therapy induced active immune tolerance in NOD mice. However, the precise mechanisms remain unclear. So, we thereafter synthesized all known determinants covering the 500–585 region of GAD and tested whether these determinants differ in their capability in inducing Ag-specific tolerance using proliferation and cytokine assay. Our results showed that peptide 509–528- and peptide 546–554-specific T cells were tolerated, while peptide 570–585-specific T cells were expanded in rAAV/GAD500–585-immunized NOD mice. It is quite unexpected that peptide 524–538-specific T cells were not significantly expanded in rAAV/GAD500–585-immunized NOD mice as this peptide was known to induce regulatory CD4 T cell response (Fig. 7). Quite interestingly, we found that peptides 509–528 and 570–585, which were especially tolerated or especially expanded in rAAV/GAD500–585-immunized NOD mice, were very effective in inducing Th2 cell function (Fig. 8,B). These results implied that peptide 570–585 might have induced a regulatory CD4+ T cell population, and indicated that there was a correlation between the capacity of GAD500–585-specific peptides to induce Th2 shift and hyporesponsiveness of GAD500–585-specific Th1 cells. We were unable to detect evident proliferation response to MHC class I-restricted peptide 546–554 in T cell recall assay. However, results of ELISPOT assay showed that the production of IFN-γ of splenocytes in response to peptide 546–554 was significantly decreased in rAAV/GAD500–585-immunized NOD mice as compared with that in AAV-treated NOD mice (Fig. 8 A). These data suggested that 546–554-reactive T cells were also especially tolerated in rAAV/GAD500–585-immunized NOD mice. Based on these results, the mechanisms by which rAAV/GAD500–585 immunization induced potent immune tolerance could be postulated as follows: 1) the immunodominant CD4+ T cell determinant 509–528 and CD8+ T cell determinant 546–554 were captured and processed by resting APC cells after expression; epitope presentation this way tolerated rather than activated peptide-specific T cells. As peptides 509–528 and 546–554 play important roles in early pathogenic process of diabetes, tolerance induction to these peptides might contribute mostly to the prevention of insulitis and diabetes in NOD mice. 2) After the peptide 509–528- and peptide 546–554-specific T cells being tolerated or being actively suppressed, peptide 570–585, but not peptide 524–538 or other determinant became immunodominant, immune response to peptide 570–585 induced the expanding of T regulatory cells, and these cells provided protective effects by suppressing GAD500–585-specific Th1 cells in a bystander manner. 3) Although no significant proliferative response was found to peptides 530–543, 539–553, and 546–558, T cell response to these peptides was slightly diminished in rAAV/GAD500–585-treated NOD mice, indicating that T cell response to these peptides was also suppressed. However, the precise mechanisms by which peptides 509–528 and 546–554 induced tolerance and peptide 570–585 elicited Th2 response remain to be determined.
Finally, we examined whether rAAV/GAD500–585 treatment induced the generation of regulatory T cells that could suppress diabetogenic effector T cells. We adoptively cotransferred splenocytes from rAAV/GAD500–585-immunized mice with splenocytes from untreated, acutely diabetic NOD mice into NOD-scid mice. The splenocytes from the rAAV/GAD500–585-treated mice were found to inhibit the transfer of diabetes by splenocytes from acutely diabetic NOD mice. This result suggested that immunization with rAAV/GAD500–585 generated a regulatory population. Although the precise mechanism involved in the generation of regulatory cells in rAAV/GAD500–585-treated mice remains to be determined, we postulate that peptide 570–585 might play an important role in this process. The presentation of GAD65 epitope 570–585 by APC cells in a Th2 cytokine microenviroment may partially contribute to the generation of regulatory T cell populations.
As the population of CD4+CD25+ T regulatory cell has been implicated as a highly specialized T cell subset that is pivotal in immunoregulatory response, and several studies have shown that CD4+CD25+ T cells can be induced by autoantigen administration, we finally followed the changes of CD4+CD25+ T cell population in rAAV/GAD500–585-injected mice (45, 46, 47, 48, 49). Our results showed that CD4+CD25+ T cell population was up-regulated in later stage in rAAV/GAD500–585-treated mice. The reasons that this up-regulation could not be detected in the splenocytes of 12-wk-old NOD mice might be explained as follows. First, rAAV/GAD500–585 immunization did up-regulate CD4+CD25+ T cells; however, the up-regulation of CD4+CD25+ T cells was too weak to be detected shortly after immunization in the peripheral immune system. This kind of phenomenon has also been reported by other groups whose data showed that CD3-induced CD4+CD25+ T cell up-regulation could be detected in the pancreas, but not in the spleen or in the lymph node (50). Second, as rAAV/GAD500–585 immunization mainly expanded peptide 570–585-specific T cells, we postulated that these T regulatory cell populations had different phenotype from CD4+CD25+ T cells. Although the precise mechanisms involved in the significant up-regulation of CD4+CD25+ T cells in rAAV/GAD500–585-treated mice at later stage remain to be determined, we postulated that the continued expression of GAD500–585 protein and systemically up-regulated IL-10 and TGF-β production later after rAAV/GAD500–585 administration might partially explain this (51) (Fig. 10).
In conclusion, we have analyzed the epitope character of GAD65 Ag and developed a rAAV expressing GAD500–585, a peptide including mutiple immunodominant epitopes. Immunogene therapy using this recombinant vector to 7-wk-old female NOD mice prevented the development of insulitis and type I diabetes. We found that protection from diabetes by rAAV/GAD500–585 injection was due to the induction of active immunological tolerance rather than clonal deletion or clonal anergy of effector T cells. CD4+ T cell determinants 509–528 and 570–585 and CD8+ T cell determinant 546–554 played important roles in rAAV/GAD500–585 immunization-induced tolerance. Because rAAV vector holds some superiority to other viral and nonviral gene delivery systems with regard to safety, efficiency, lack of need for repeated viral administration, and duration of action without known pathology, a rAAV-expressing GAD500–585 might have therapeutic value for clinical use in preventing type I diabetes.
The authors have no financial conflict of interest.
We gratefully acknowledge Dr. Feng Jiannan for useful advice.
The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked advertisement in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.
This work was supported by National “863” Fund Grant 2003AA216082 and “973” Fund Grant 2001CB510005 of China.
Abbreviations used in this paper: GAD, glutamic acid decarboxylase; AAV, adenoassociated virus; BHK, baby hamster kidney.