Skip to main content
NCBI home page
Clinical and Experimental Immunology logoLink to Clinical and Experimental Immunology
. 2004 Aug;137(2):253–262. doi: 10.1111/j.1365-2249.2004.02546.x

DNA vaccination encoding glutamic acid decarboxylase can enhance insulitis and diabetes in correlation with a specific Th2/3 CD4 T cell response in non-obese diabetic mice

A Gauvrit 1, M Debailleul 1, A-T Vu 1, P Saï 1, J-M Bach 1
PMCID: PMC1809102  PMID: 15270841

Abstract

DNA vaccination encoding β cell autoantigens has been shown very recently to prevent type I diabetes in non-obese diabetic (NOD) mice. However, DNA vaccination encoding microbial or reporter antigens is known to induce specific long-lasting CD4 Th1 and strong cytolytic CD8 T cell responses. As this immune phenotype is associated strongly with β cell destruction leading to diabetes, we have chosen to study the effects of plasmids encoding glutamic acid decarboxylase (GAD), a crucial β cell autoantigen, in female NOD mice that developed a ‘moderate’ diabetes incidence. In the present study, 3-week-old female NOD mice were vaccinated twice in tibialis muscles with plasmid-DNA encoding 65-kDa GAD or βgalactosidase. In GAD-DNA immunized mice, diabetes cumulative incidence (P < 3·10−3) and insulitis (P < 7·10−3) increased significantly. Simultaneously, DNA immunization induced GAD-specific CD4 T cells secreting interleukin (IL)-4 (P < 0·05) and transforming growth factor (TGF)-β (P = 0·03). These cells were detected in spleen and in pancreatic lymph nodes. Furthermore, vaccination produced high amounts of Th2 cytokine-related IgG1 (P < 3·10−3) and TGF-β-related IgG2b to GAD (P = 0·015). Surprisingly, diabetes onset was correlated positively with Th2-related GAD-specific IgG1 (P < 10−4) and TGF-β-related IgG2b (P < 3·10−3). Moreover, pancreatic lesions resembled Th2-related allergic inflammation. These results indicate, for the first time, that GAD-DNA vaccination could increase insulitis and diabetes in NOD mice. In addition, our study suggests that Th2/3 cells may have potentiated β cell injury.

Keywords: DNA vaccination, glutamate decarboxylase, insulitis, Th2/3 T cells, type I diabetes

INTRODUCTION

Type I diabetes results from autoimmune destruction of islet β cells. In the NOD (non-obese diabetic) mouse model, this β cell injury is thought to involve autoantigen-specific CD4 T cells with a Th1 phenotype and specific CD8 T lymphocytes [1]. Diabetogenic CD4 T cell clones usually produce Th1 cytokines [2] and many Th1 cells derived from islet infiltration of NOD mice and reactive to insulin can transfer diabetes [3]. However, both CD4 and CD8 subsets are required to transfer diabetes to neonate [4] or irradiated adult mice [5]. Mice lacking MHC class I molecules (NOD-β2m–/–) do not develop diabetes [6], and some islet-specific CD8 T cell clones transfer diabetes in the absence of CD4 T cells [7].

On the other hand, regulatory T cells could control diabetes development in NOD mouse [8]. Th2 or Th2/3 CD4 T cells may play a suppressive role in diabetes [1]. Th2 lymphocytes typically secrete interleukin (IL)-4, IL-5 and IL-10, whereas Th2/3 cells produce IL-10 and transforming growth factor (TGF)-β. Systemic administration of IL-4 prevents insulitis and diabetes in NOD mice [9] and pancreatic IL-4 expression promotes antigen-specific Th2 cells that inhibit diabetogenic T cells in the pancreas [10]. Th2/3 cells may also be involved in diabetes regulation, as diabetes is reduced in NOD mice expressing TGF-β in periphery [11] or in transgenic pancreatic α-cells [12].

During the past decade, several strategies have been elaborated to modulate the autoimmune process and to prevent diabetes by activation of regulatory T cells in an autoantigen-specific manner. Consistently, systemic [13–15] or nasal [16] administrations of autoantigens such as glutamic acid decarboxylase (GAD), insulin or derived peptides induce specific Th2 functions and reduce diabetes.

Very recently, DNA vaccination encoding GAD [17,18], insulin [19–22] or HSP70 [23] has been shown to be an effective approach to prevent type I diabetes in the NOD mouse. Such therapeutic trials tested in the mouse are considered as future possible specific immune therapy of type I diabetes in humans. Curiously, genetic immunization is known to allow direct intracellular synthesis of the antigen and leads generally to long-lasting CD4 Th1 and strong specific cytolytic CD8 responses (reviewed in [24]). In this context, one could have expected that DNA vaccination encoding a β cell autoantigen may enhance type I diabetes. Simultaneously with other studies, we decided to investigate the effect of injection of plasmid-DNA encoding 65 kDa GAD (GAD65) on diabetes and insulitis course in the NOD mouse. GAD65 is a crucial early target autoantigen involved in diabetes in the mouse [25,26]. Diabetes can be prevented by intrathymic [26] or intravenous [15,25] injections of recombinant GAD65 and by suppression of GAD expression in β cells in antisense GAD-transgenic NOD mice [27].

Our approach was different from that of others. Given the known characteristics of the immune response induced by DNA vaccination against microbial or reporter antigens and to have conditions related closely to humans, we have chosen to study the effects of GAD65-plasmids in female NOD mice which developed a ‘moderate’ diabetes incidence. Our study shows for the first time that GAD65-DNA vaccination can increase insulitis and diabetes in NOD mice. Surprisingly, GAD65-DNA immunization also induces a specific Th2/3 CD4 T cell response that may be implicated in diabetes worsening as GAD65-specific TGF-β-related IgG2b and Th2 cytokine-related IgG1 secretions were correlated positively with diabetes onset and pc-GAD NOD mice showed Th2-related allergy-like islet lesions.

MATERIALS AND METHODS

Mice and diabetes

Given the immune characteristics induced by intramuscular DNA vaccination, NOD mice, which developed a ‘moderate’ diabetes incidence, were used to study the effects of plasmids encoding GAD65. NOD mice obtained from Clea (Tokyo, Japan) were maintained in controlled and clean conventional conditions at the Veterinary School of Nantes (France). Diabetes incidence at 300 days in females and males was, respectively, 25% and 5%. Diabetes was diagnosed when glycaemia was ≥13·5 mmol/l for 2 consecutive weeks.

Plasmid construction

Full-length human islet GAD65 cDNA was amplified by polymerase chain reaction (PCR) (Taq DNA polymerase, Promega, Madison, WI, USA) from GAD-pVEGT (kindly provided by T. Dyrberg) using a 5′ oligonucleotide containing the EcoRI digestion site (5′GCGAATTCTAACATGGCATCTCCGGGCTC3′, MWG, Ebersberg, Germany) and a 3′ primer containing the XbaI site (5′GCTCTAGAGCTAAATCTTGTCCAAGGCG3′). Pc-GAD was obtained by cloning EcoRI-XbaI-digested amplicons into pcDNA3·1/B (Invitrogen, Groningen, the Netherlands). The construct was controlled by sequencing (MWG). Recombinant protein produced by transient transfected mammalian cells (Hela-S3 and Nor-10) was detected by immunoblotting using GAD65-specific monoclonal antibodies (MoAb) GAD1 and GAD6 (data not shown).

Plasmid and protein injections

Anaesthetized female NOD mice were allocated randomly and injected at 3 and 8 weeks of age in each tibialis anterior muscle with 50 µg [50 µl in phosphate buffered saline (PBS)] of endotoxin-free pc-GAD, pc-LacZ (pcDNA3·1/LacZ encoding β galactosidase (βGal), Invitrogen) or PBS. Injections were performed 5 days after administration of 100 µl of 10 µm cardiotoxin (Latoxan, Rosans, France) per leg at the same muscle site. In some experiments, mice previously treated or not with pc-GAD or pc-LacZ were injected intraperitoneally with 30 µg of recombinant human GAD65 (rhGAD65, Diamyd, Stockholm, Sweden) or recombinant βGal (rβGal, Sigma, St Louis, MO, USA) with Complete Freund Adjuyant (CFA).

Detection of GAD or βGal expression in injected muscles

Following cardiotoxin treatment, 3-week-old female NOD mice were injected with pc-GAD or pc-LacZ in one tibialis anterior muscle as indicated above. The tibialis anterior muscle of the other leg received only cardiotoxin (negative control). Muscles were removed 12 h after DNA injection. Endogenous biotins of 5-µm cryo-sections were blocked (Biotin-Blocking System, Dako, Glostrup, Denmark) and non-specific staining were inhibited by incubating sections 30 min in the presence of rabbit serum [1/5 diluted in TBS + 1% bovine serum albumin (BSA)] and unconjugated rabbit antimouse IgG antobodies (5 µg/section, Sigma). GAD and βGal expressions were then detected by indirect immunostaining with 4 µg/section primary MoAbs specific to GAD (Clone GAD6, mouse IgG2a, BD Biosciences, San Diego, CA, USA) or to βGal (clone GAL-13, mouse IgG1, Sigma) in combination with biotin-conjugated rabbit antimouse IgG antibodies (5 µg/section, Sigma) and alkaline phosphatase-conjugated streptavidine (Sigma). Sections were then incubated with substrate (Naphtol AS-TR Phosphate, Sigma), which produced a red colour when activated. Uninjected muscles were treated identically. Adjacent sections were stained identically but without primary MoAb. Complete muscle samples were analysed (every five couples of 5 µm sections).

Pancreatic histopathology

Five-µm frozen pancreas sections were fixed in acetone and stained with haematoxylin–eosin to analyse insulitis or with toluidin blue to detect specific infiltration of mast cells in pancreatic islets. To quantify islet infiltration, at least 30 islets from three sections cut 150 µm apart were blindly examined for each pancreas by two independent observers. Degrees of insulitis were evaluated using a semiquantitative scale: (0) islet devoid of insulitis; (1) peri-insulitis or insulitis occupying up to 10% of the islet; (2) insulitis involving 10–50% of the islet; and (3) severe insulitis occupying more than 50% of the islet. The percentage of islets affected by each degree of insulitis was calculated for each mouse.

T cell proliferation assays

Spleen cells (5·105) were cultured in triplicate in flat-bottomed 96-well plates (Nunc, Roskilde, Denmark) in 0·2 ml/well serum-free AIMV“ medium (Life Technologies, Paisley, Scotland, UK) in the presence or absence of antigen at 37°C, 5% C02. rhGAD65 dialysed against 50 mm Hepes (pH 7·4) containing 10% glycerol and 1 mm pyridoxal 5′-phosphate (PLP, Sigma) or rβGal were used at a final concentration of 20 µg/ml. After 96 h, proliferations were quantified by incubation with 1 µCi/well of [3H]-thymidine (Amersham, Saclay, France) for an additional 16 h. Proliferations were expressed as stimulation indices [SI = mean counts per minute (cpm) with antigen/mean cpm without antigen]. Proliferations with SI < 1·5 were considered as negative.

Cytokine assays

Spleen or lymph node (LN) cells (2·5 × 106/ml/well) were incubated in AIMV in flat-bottomed 24-well plates with or without 20 µg of dialysed rhGAD65 or rβGAL. After 48 h, supernatants were tested by a sandwich enzyme-linked immunosorbent assay (ELISA) method. Interferon (IFN)-γ, Il-4 and Il-10 levels were measured using Duoset kits from R&D Systems (MN, USA). TGF-β1 was measured using antibodies purchased from BD Biosciences.

ELISA for quantification of anti-GAD and anti-βGal antibodies

Maxisorb plates (Nunc) were coated in duplicate overnight at + 4°C with 0·5 µg/well (for IgG1) or 1 µg/well (for IgG2a/2c or IgG2b) of antigen, or with PBS. Plates were blocked for 120 min at room temperature (RT) with 200 µl of PBS 0·5% fetal calf serum (FCS) 0·05% Tween/well. Sera were diluted in PBS 30-fold for IgG1 detection and 20-fold for IgG2a/2c or IgG2b, and incubated (50 µl/well) for 150 min at RT. Fifty µl of 1 : 2000 diluted horseradish peroxidase conjugated goat antimouse IgG1, IgG2a, or IgG2b antibodies (Caltag, Burlingham, CA, USA) were added for 120 min. Plates were then incubated with o-phenylenediamine dihydrochloride substrate (Sigma). For each serum, specific OD450nm (OD with antigen less OD without antigen) was evaluated. Antibody titres (µg/ml) were calculated based on calibration curves obtained with various dilutions of standard sera included in each plate and previously quantified for antibody to GAD or to βGal, respectively. Standards were pooled sera from four protein-injected mice. Standard antibody titres were calculated according to the dilution that gave a specific OD450nm equal to the cut-offs of detection of IgG1, IgG2a and IgG2b, respectively, in a commercial quantified serum (ICN Biomedicals, CA, USA).

NOD and C57BL/6 mice expressing the IgC1-b haplotype produce IgG2c (IgG2a/2c) instead of IgG2a. As commercial antimouse IgG2a antibodies recognize IgG2a/2c differently [28], we determined beforehand that conjugated goat antimouse IgG2a antibody obtained from Caltag recognized IgG2a/2c efficiently. Antimouse IgG2a/2c (BD Biosciences) was used (0·2 µg/well) for coating plates overnight at + 4°C. ELISA was then performed as indicated above, using 100-, 1000- and 10 000-fold NOD, C57BL/6 and BALB/c serum dilutions. IgG2a/2c antibodies were significantly detectable in up to 1000-fold dilution of sera from NOD and C57BL/6 mice, but never in serum from BALB/c mice (data not shown).

Statistical analysis

Statistical significance was evaluated using Student's t-test, the Mann–Whitney U-test, log-rank test or χ2 test. Independent contributions of serum antibody titres to the delay of diabetes onset (days) were evaluated using a multiple stepwise regression procedure including parameters related to diabetes onset delay with a P-value below 0·5. P < 0·05 was considered as significant.

RESULTS

GAD and βGal expression in injected muscles

GAD and βGal expression were clearly detected in muscle cells of mice injected, respectively, with pc-GAD or pc-LacZ (Fig. 1a and d, respectively). In both cases, recombinant protein expression was observed in a disorganized muscle area, which was in regeneration following cardiotoxin injection. Adjacent sections, treated identically but without primary MoAb, were not stained positively (Fig. 1b,e). Finally, uninjected muscle from pc-GAD or pc-LacZ mouse, incubated with anti-GAD or anti-βGal primary MoAbs, respectively, did not have red coloration in muscle cells (Fig. 1c,f)

Fig. 1.

Fig. 1

Open in a new tab

Detection of GAD and βGal in injected muscles. (a) injected muscle from pc-GAD mouse incubated with anti-GAD primary MoAb; (b) injected muscle from pc-GAD mouse [same portion as (a)] without anti-GAD primary MoAb; (c) uninjected muscle incubated with anti-GAD primary MoAb; (d) injected muscle of pc-LacZ mouse incubated with anti-βGal primary MoAb; (e) injected muscle from pc-LacZ mouse in absence of anti-βGal primary MoAb; [same portion as (d)]; and (f) uninjected muscle incubated with anti-βGal primary MoAb. Original magnifications: ×600 for (a, b and c); ×400 for (d, e and f).

DNA immunization encoding GAD65-enhanced pancreatic islet cell injury and diabetes

Diabetes was accelerated and the incidence increased specifically in pc-GAD female NOD mice (P < 3·10−3, Fig. 2a). At 42 weeks of age, diabetes incidence was 52% in pc-GAD mice versus 25%, 24% and 33% in PBS, unmanipulated and pc-LacZ groups, respectively. Concomitantly with diabetes aggravation, insulitis severity increased before diabetes onset in pc-GAD mice (P < 7·10−3, Fig. 2b). More than 50% of the islets from pc-GAD mice were severely infiltrated 6 weeks after the second DNA injection, while severe insulitis affected only 25% and 10% of islets from pc-LacZ and PBS mice, respectively. After the second DNA injection, a slight increase in insulitis was observed in the pc-LacZ group (P < 0·05). However, the course of insulitis in pc-LacZ mice did not lead to significant disease acceleration and mice injected with pc-LacZ displayed a diabetes incidence similar to that of PBS mice at the end-point of the study (Fig. 2a).

Fig. 2.

Fig. 2

Open in a new tab

Diabetes incidence and insulitis severity in GAD-DNA-vaccinated NOD mice. (a) Female NOD mice were injected at 3 (I1) and 8 (I2) weeks of age with 50 µg of pc-GAD (n = 29), pc-LacZ (n = 21), or PBS (n = 24). Unmanipulated mice (n = 22) are also indicated. *P < 0·02 (log-rank test), **P < 3·10−3 (χ2 test). (b) Pancreata from PBS, pc-LacZ and PC-GAD mice were removed 6 weeks after the second injection (n = 4, 4 and 11, respectively), i.e. before diabetes onset. Insulitis was evaluated using scores as indicated in Materials and methods. Results are expressed as mean percentage of islets affected by each degree of insulitis. *P < 0·05, **P < 0·02, ***P < 7·10−3.

T cell proliferations to GAD were altered in pc-GAD mice

Spleen cell proliferations to rhGAD65 were significantly higher in pc-GAD mice than in pc-LacZ- (P < 0·04, Fig. 3a) or unvaccinated mice (P = 6·10−3). Splenocytes from more than 85% of pc-GAD mice, but none from pc-LacZ mice, exhibited SI above 1·5 in response to GAD. Surprisingly, T cell proliferations obtained in vitro to whole rhGAD65 were weak and lower than, first, those usually obtained in our laboratory after GAD65 protein intraperitoneal injections, and secondly, the strong T cell reactivities usually produced by DNA vaccination (often stronger than with protein injections). Therefore, we tested whether injections of pc-GAD in NOD mice (priming) could modify GAD65-specific T cell proliferative responses following a subsequent intraperitoneal boost with GAD65-protein (Fig. 3b). Proliferation to GAD was significantly increased in NOD mice immunized with GAD protein compared to untreated mice (P < 0·02). Pc-LacZ injection did not modify anti-GAD proliferation after GAD protein injection. Also, DNA vaccination encoding GAD (pc-GAD) strongly decreased specific T cell responses in response to the following GAD protein immunization (P < 0·02). No significant difference was observed between proliferations of spleen cells from mice injected with pc-GAD and GAD protein and proliferations of spleen cells from mice receiving pc-GAD only. On the other hand, pc-LacZ mice exhibited strong βGal-specific proliferations (P < 0·02).

Fig. 3.

Fig. 3

Open in a new tab

T cell responses to GAD in DNA-immunized mice. (a) Spleen T cells from mice immunized twice (n = 10 with pc-GAD, n = 6 with pc-LacZ; and n = 5 with PBS) were tested for proliferative responses to dialysed rhGAD65 protein. Results are expressed as stimulation indices (SI) ± s.e.m. [3H]-thymidine incorporations in the absence of antigen were 3344 ± 821, 2192 ± 178 and 1840 ± 413 cpm for pc-GAD-, pc-LacZ- and PBS-mice, respectively. *P < 0·04, **P = 6·10−3. (b) Mice were injected twice with pc-GAD (n = 4) or pc-LacZ (n = 4) and received intraperitoneal administration of rhGAD65 1 week after the last DNA injection. Control mice were injected with only rhGAD65 (n = 5) or PBS (n = 5). Spleen cells were then tested for proliferative responses to rhGAD65 or rβGal. [3H]-thymidine incorporations in the absence of antigen were, respectively, 2398 ± 409, 3284 ± 644 and 3211 ± 473 cpm for mice injected with pc-GAD and rhGAD65, pc-LacZ and rhGAD65 and rhGAD65 alone. *P < 0·02.

DNA vaccination encoding GAD65 induced a biased Th2/3 specific T cell response

Cytokine secretion patterns of spleen cells from twice DNA-vaccinated mice were analysed in response to rhGAD65 (Fig. 4a). Surprisingly, Th2/3 responses specific to GAD were detected in pc-GAD mice, as IL-10 (P < 0·02), IL-4 (P < 0·05) and TGF-β (P = 0·03) were produced in response to rhGAD65. Spleen cells from pc-GAD mice did not secrete specific IFN-γ. Conversely, as described previously in BALB/c and C57BL/6 mice [29], spleen cells from pc-LacZ mice produced IFN-γ strongly in response to rβGal (P < 6·10−3, Fig. 4b). Spleen cells from pc-LacZ mice also secreted TGF-β (P < 0·05), but not IL-10 and IL-4, in response to rβGal.

Fig. 4.

Fig. 4

Open in a new tab

GAD65- and βGal-induced cytokine production by spleen cells from NOD mice after two DNA injections. Spleen cells from mice injected with pc-GAD (n = 8) or pc-LacZ (n = 5) were cultured in vitro with rhGAD65 (a) or rβGal (b), 2 weeks after the second DNA injection. Two days after the second DNA injection, cells from popliteal/inguinal LN (c) or pancreatic LN (d) of pc-GAD mice (n = 4) or pc-LacZ mice (n = 5) were cultured with rhGAD65. IFN-γ, IL-10, TGF-β and IL-4 secretions in supernatants were analysed by ELISA. Results are expressed as mean cytokine concentrations ± s.e.m. *P < 0·05, **P = 0·03, ***P < 0·02, ****P < 6·10−3.

In pc-GAD mice, GAD-specific Th2/3 lymphocytes secreting IL-4 (P = 0·03), IL-10 (P < 0·05) and TGF-β (P < 0·05) were identified in lymph nodes (LN) draining the injection site 2 days after the second DNA administration (Fig. 4c). Such cells secreting IL-4 and TGF-β were also detected in pancreatic LN, 2 days (P < 0·02, Fig. 4d) and 2 weeks (P < 0·02, data not shown) after the second DNA injection, but without secreting IL-10 in response to GAD. In pc-LacZ mice, IFN-γ-secreting T cells in response to rβGal were detected in inguinal LN (P < 0·05) but not in pancreatic LN (data not shown).

GAD65-specific antibody production was consistent with the Th2/3 cellular profile of pc-GAD mice

Two weeks after the second DNA injection, IgG1, IgG2a/2c and IgG2b specific to GAD increased significantly in the sera of pc-GAD mice (Fig. 5a) compared to uninjected (P < 2·10−4) or pc-LacZ mice (P < 10−4). GAD protein-injected mice also showed higher levels of IgG1, IgG2a/2c and IgG2b specific to GAD than control mice (P < 3·10−2). According to the predominant Th2/3 cellular profile detected previously, pc-GAD, as well as GAD protein-injected mice, produced higher titres of anti-GAD IgG1 (P < 3·10−3) and IgG2b (P = 0·015) than anti-GAD IgG2a/2c.

Fig. 5.

Fig. 5

Open in a new tab

GAD65- and βGal antibody production in DNA-injected mice. Results are given as mean titres ± s.e.m. Sera of pc-GAD (a, n = 29) and pc-LacZ mice (b, n = 20) were tested 2 weeks after the second DNA injection. Serum from four GAD65- or four βGal protein-immunized mice were tested 2 weeks after a single protein injection. 1P < 3·10−3 and P = 0·015 for anti-GAD IgG2a/2c versus IgG1 and IgG2b, respectively, in pc-GAD mice, 2Absence of significant differences for anti-βGal IgG2a/2c versus IgG1 and IgG2b in pc-LacZ mice. *P < 6·10−4, **P < 10−4.

Anti-βGal IgG1, IgG2a/2c and IgG2b increased in sera of pc-LacZ mice (P < 6·10−4, Fig. 5b). Anti-βGal antibody titres in pc-LacZ mice were approximatively 10-fold higher than anti-GAD antibody levels in pc-GAD mice, which is consistent with the known strong immunogenicity of βGal. Pc-LacZ mice displayed anti-βGal IgG2a/2c titres comparable to those of other isotypes tested consistently with the strong activation of βGal-specific IFN-γ-secreting T cells described above.

Detection of high levels of anti-GAD or anti-βGal IgG2b in pc-GAD and pc-LacZ mice, respectively (Fig. 5a,b) was consistent with the strong specific secretions of TGF-β described above.

GAD65-specific TGF-β-related IgG2b and Th2 cytokine-related IgG1 secretions correlated positively with diabetes onset

Two weeks after the second DNA injection, GAD-specific IgG2b levels were associated strongly and negatively with the delay of subsequent diabetes onset in GAD-DNA vaccinated mice (P < 3·10−3, Fig. 6a). In addition, serum anti-GAD IgG1 titres were negatively correlated with the delay of diabetes onset after the first injection (P < 10−4, Fig. 6a). After a single injection of pc-GAD, GAD-specific IgG2b were predominantly produced compared to IgG2a/2c (P < 2·10−3, Fig. 6b). GAD-IgG1 reached a maximum level 2 weeks after the second DNA injection, while GAD-IgG2a/2c increased up to 28 weeks of age (Fig. 6b). However, rises in IgG1 or IgG2a/2c titres were never correlated with diabetes onset.

Fig. 6.

Fig. 6

Open in a new tab

Kinetics of GAD65 antibodies in pc-GAD mice and correlation with diabetes onset. (a) Antibody levels were tested at 6 and 10 weeks of age, i.e. after the first and the second DNA injections, respectively. β: standardized partial regression coefficient. (b) GAD65-specific anibody levels were analysed in sera of pc-GAD-injected mice at 6, 10 and 28 weeks of age. Analyses at 6 weeks were performed between the two DNA injections (I1 and I2). *P < 0·05, **P < 6·10−3, ***P < 2·10−3, ****P = 4·10−4.

Islet infiltration in pc-GAD mice resembled Th2-related allergic inflammation

Six weeks after the second DNA injection pancreatic islets from pc-GAD mice showed an allergy-like inflammation, as suggested strongly by the histopathological differences observed in pancreas of pc-GAD mice comparative to pc-LacZ mice (Fig. 7). Pancreatic islets of pc-GAD mice were more disorganized than those from pc-LacZ mice (Fig. 7a and c, respectively). Pc-GAD islet infiltrations were diffuse and very heterogeneous in more than 50% of the analysed infiltrated islets and contained polymorphonuclear cells. Moreover, some unexpected typical mast cells characterized by metachromatic purple bodies in the presence of toluidine blue were detected in pc-GAD insulitis but not in pc-LacZ islet infiltrations (Fig. 7b and d, respectively).

Fig. 7.

Fig. 7

Open in a new tab

Histopathology of pancreatic infiltration in pc-GAD mice. Pancreata from pc-GAD (n = 5) or pc-LacZ (n = 2) mice were analysed 6 weeks after the second DNA injection. Five-µm frozen pancreatic sections of pc-GAD (a and b) and pc-LacZ (c and d) mice were stained with haematoxylin and eosin (a and c) or with toluidine blue (b and d). Original photomicrograph magnification was ×600.

DISCUSSION

While plasmids encoding insulin or fragments [19–22] or HSP70 [23] could prevent diabetes in NOD mice, genetic vaccination encoding GAD has been described recently to lead to no effect [20,22,30, 31] or to prevention [17,18]. Other studies have shown that injections of plasmids encoding insulin could also aggravate diabetes [18,32]. Here, we demonstrate for the first time that intracellular expression of GAD65 by DNA vaccination could accelerate and increase diabetes incidence and insulitis severity in NOD mice.

The apparent discrepancy of these results emphasizes the importance of the choice of the model used to study prevention of type I diabetes. Differences in the incidence of spontaneous diabetes between NOD colonies, in the number of DNA injections and in the use or not of pretreatment may lead to the divergent effects observed. Given the immune characteristics usually induced by intramuscular DNA vaccination against microbial or reporter antigens, we decided to test GAD65-DNA vaccination in NOD mice with a ‘moderate’ diabetes incidence, house-bred in controlled and clean conventional conditions. These conditions seem to be related more closely to human diabetes. In mouse colonies with a higher incidence (75–95%), acceleration or worsening of disease may be more difficult to observe. Other studies have investigated effects of GAD-DNA vaccination in NOD mice which develop a high diabetes incidence [17,18,20,22, 31]. In another study [30], a single immunization of 6-week-old mice with GAD plasmid provided a low diabetes enhancement during the first stage after DNA vaccination but was not durable, possibly because of the lack of a second DNA injection.

In contrast with our study and those of Wiest Ladenburger et al. [30] and Balasa et al. [17], other studies [20,31] did not include pretreatment with cardiotoxin, which leads to muscle regeneration and improves DNA immunization efficiency [33]. Consistently, we observed a lower diabetes aggravation only in NOD mice treated twice with pc-GAD without cardiotoxin injection (data not shown).

In published DNA vaccination studies, comparisons between experimental and control plasmids do not indicate and do not take into account possible variations either in the expression level of different antigens (even though the same plasmid or promoter is used) or in the numbers and types of CpG motifs. Number and localization of CpG motifs present in DNA vaccines, as well as their opposite immune effects (stimulation by CpG-A and -B motifs, or neutralization by CpG-N motifs) [34], makes interpretation of CpG repeats very difficult. In our study, the best CpG motif (GACGTT) known to stimulate murine cells [34] was present twice in both pc-GAD and pc-LacZ, in the pcDNA3·1-vector part only. Our study, like that of Tisch et al. [31], did not detect an effect of control plasmid on diabetes incidence, while CpG sequences were shown to induce IgG2b specific to HSP60, but not to GAD or insulin [23]. However, in our study, after the second DNA injection, a slight increase in insulitis was observed in the pc-LacZ group. These observations suggest that DNA or particular plasmid sequences, such as CpG motifs, could have modulated native anti-β cell autoreactivity, but without detectable effect on diabetes. However, the course of insulitis in pc-LacZ mice did not lead to a significant disease acceleration, and mice injected with pc-LacZ displayed a diabetes incidence similar to that of PBS mice and unmanipulated mice at the end-point of the study.

In the case of insulin- or HSP70-DNA vaccination, disease prevention was attributed to the stimulation of autoantigen-specific regulatory Th2 CD4 T cells [19,20,23] or to the down-regulation of aggressive IFN-γ secretions [21]. GAD-DNA immunization seems to elicit different immune responses and to implicate other immune mechanisms. Indeed, diabetes prevention by GAD-DNA administration is not associated with the stimulation of specific immunoregulatory Th2 cells in vivo[17]. In some cases, this approach even produces GAD-specific Th1 reactivities [20,31] but, surprisingly, without modifying the diabetes course. To obtain efficient protection with GAD-DNA administration, Tisch et al. have performed immunization of 4-week-old mice three times with plasmid encoding IL-4 and a secreted form of GAD65 nt 656–1070, which led to a specific Th2 response [31,32].

GAD-specific CD4 T cells detected in our protocol could have been stimulated after endocytosis of GAD released by transfected myoblasts by antigen-presenting cells (APC). As indicated in the Results, recombinant GAD and βGal were detectable in injected muscle. By homology with the known immune response induced by DNA vaccination encoding microbial antigens, release of recombinant GAD by transfected myoblasts may be attributed to specific CD8 T lymphocyte-mediated lysis [24] or to the result of toxic superproduction of the encoded protein itself [35]. In fact, GAD-specific CD4 T cells were detected especially after the second DNA injection and were stimulated in vitro by soluble GAD protein. Endocytosed protein could be expected to promote presentation by MHC class II-expressing APC and CD4 T cell reactivities. These CD4 T cells were characterized by an unexpected GAD65-specific secretion of Th2/3 cytokines (IL-4, IL-10 and TGF-β). This response is comparable to that detected after in vivo GAD protein administration, which requires endocytosis of the autoantigen by APC [15, 16, 25]. GAD-specific Th2/3 cells may regulate diabetes negatively and inhibit Th1 reactivities to GAD, providing an explanation for the weakness of proliferation detected in vitro to this autoantigen. As we found that GAD-IgG2b and -IgG1 levels were correlated strongly with disease onset in a multiple regression test, it seemed likely that Th2/3 CD4 T lymphocytes have participated directly or not in diabetes and insulitis aggravation. This hypothesis is supported by the observation that TGF-β and IL-4 together can induce and stimulate strong specific cytolytic CD8 T cell responses, although cytotoxicity is inhibited by TGF-β or IL-4 separately [36]. While separate expression of TGF-β[11] or IL-4 [10] prevents autoimmune diabetes in mice, co-secretion of TGF-β and IL-4 to GAD may activate cytotoxic T lymphocytes to this autoantigen and to β cells. Similarly, some autoreactive Th2 cell clones were unable to protect mice [2,3] or even elicited diabetes [3,37]. As described previously [38], no mast cells were known to infiltrate NOD pancreatic islets during spontaneous β cell aggression. In our study, pancreatic islet infiltration of pc-GAD NOD mouse exhibited an unexpected infiltration of polymorphonuclear cells and mast cells usually related to Th2-allergy inflammations. This result clearly supported a possible role for Th2/3 T cells specific to GAD in pc-GAD mouse diabetes aggravation. Such an observation has already been shown by others: autoimmune Th2 CD4 T cells transferring diabetes [37] or experimental autoimmune encephalomyelitis (EAE) [39] induced allergy-like lesions, respectively, in the pancreas and central nervous system of recipient mice. Additional experiments of diabetes adoptive transfer would permit elucidation of the potential role of GAD-specific Th2/3 cells in diabetes aggravation in our model.

As IL-4- and TGF-β-secreting lymphocytes were also detected in our study in pancreatic LN in response to GAD, GAD-specific Th2/3 lymphocytes were able to migrate to the target tissue and may thus have played a role not only in the periphery but also in pancreatic LN, which are required for the priming of β cell-reactive T cells in NOD mice [40]. However, surprisingly, Th2/3 T lymphocytes detected in pancreatic LN did not secrete IL-10 in the presence of GAD, suggesting that this cytokine does not take a prominent part in the target tissue and in diabetes worsening, in contrast to Pakala et al., who described a local role of IL-10 in diabetes transfer by autoimmune Th2 CD4 T cells [37].

In conclusion, our study indicates for the first time that GAD65-DNA vaccination could increase insulitis and diabetes in NOD mice in some conditions, indicating that GAD-DNA vaccination as well as insulin-DNA administration may not be a safe approach for type I diabetes prevention in humans. We also describe a surprising GAD-specific Th2/3 CD4 T cell response that could have been implicated in diabetes aggravation.

Acknowledgments

The authors are most grateful to Mrs M. Ouary, Mrs M. Allard and Mr C. Chevalier for their excellent collaboration and Dr K. Bach-Ngohou for critical reading of this paper. This work was supported by a grant from the ‘Association de Langue Française pour l’Etude du Diabète et des Maladies Métaboliques’ (Lilly/Roche Diagnostics grant).

REFERENCES

  • 1.Delovitch TL, Singh B. The nonobese diabetic mouse as a model of autoimmune diabetes: immune dysregulation gets the NOD. Immunity. 1997;7:727–38. doi: 10.1016/s1074-7613(00)80392-1. [DOI] [PubMed] [Google Scholar]
  • 2.Katz JD, Benoist C, Mathis D. T helper cell subsets in insulin-dependent diabetes. Science. 1995;268:1185–8. doi: 10.1126/science.7761837. [DOI] [PubMed] [Google Scholar]
  • 3.Haskins K, Wegmann D. Diabetogenic T cell clones. Diabetes. 1996;45:1299–305. doi: 10.2337/diab.45.10.1299. [DOI] [PubMed] [Google Scholar]
  • 4.Bendelac A, Carnaud C, Boitard C, et al. Syngeneic transfer of autoimmune diabetes from diabetic NOD mice to healthy neonates. Requirement for both L3T4+ and Lyt 2+ T cells. J Exp Med. 1987;166:823–32. doi: 10.1084/jem.166.4.823. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 5.Miller BJ, Appel MC, O'Neil JJ, et al. Both the Lyt 2+ and L3T4+ T cell subsets are required for the transfer of diabetes in non obese diabetic mice. J Immunol. 1988;140:52–8. [PubMed] [Google Scholar]
  • 6.Wicker LS, Leiter EH, Todd JA, et al. Beta 2-microglobulin-deficient NOD mice do not develop insulitis or diabetes. Diabetes. 1994;43:500–4. doi: 10.2337/diab.43.3.500. [DOI] [PubMed] [Google Scholar]
  • 7.Wong FS, Visitin I, Wen L, et al. CD8 T cell clones from young nonobese diabetic (NOD) islets can transfer rapid onset of diabetes in NOD mice in the absence of CD4 cells. J Exp Med. 1996;183:67–76. doi: 10.1084/jem.183.1.67. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 8.Boitard C, Yasunami R, Dardenne M, et al. T cell mediated inhibition of the transfer of autoimmune diabetes in NOD mice. J Exp Med. 1989;169:1669–80. doi: 10.1084/jem.169.5.1669. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 9.Rapoport MJ, Jaramillo A, Zipris D, et al. Interleukin 4 reverses T cell proliferation unresponsiveness and prevents the onset of diabetes in nonobese diabetic mice. J Exp Med. 1993;178:87–99. doi: 10.1084/jem.178.1.87. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 10.Gallichan WS, Balasa B, Davies JD, et al. Pancreatic IL-4 expression results in islet-reactive Th2 cells that inhibit diabetogenic lymphocytes in nonobese diabetic mouse. J Immunol. 1999;163:1696–703. [PubMed] [Google Scholar]
  • 11.Piccirillo CA, Chang Y, Prud’homme GJ. TGF-β1 somatic gene therapy prevents autoimmune disease in non obese diabetic mice. J Immunol. 1998;161:3950–6. [PubMed] [Google Scholar]
  • 12.Moritani M, Yoshimoto K, Wong SF, et al. Abrogation of autoimmune diabetes in nonobese diabetic mice and protection against effector lymphocytes by transgenic paracrine TGF-β1. J Clin Invest. 1998;102:499–506. doi: 10.1172/JCI2992. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 13.Muir A, Peck A, Clare-Sazler M, et al. Insulin immunization of nonbese diabetic mice induces a protective insulitis characterized by diminished intra-islet interferon-γ transcription. J Clin Invest. 1995;95:628–34. doi: 10.1172/JCI117707. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 14.Saï P, Rivereau AS, Granier C, et al. Immunization of non-obese diabetic (NOD) mice with glutamic acid decarboxylase-derived peptide 524–543 reduces cyclophosphamide-accelerated diabetes. Clin Exp Immunol. 1996;105:330–7. doi: 10.1046/j.1365-2249.1996.d01-751.x. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 15.Tisch R, Liblau RS, Yang XD, et al. Induction of GAD65-specific regulatory T cells inhibits ongoing autoimmune diabetes in nonobese diabetic mice. Diabetes. 1998;47:894–9. doi: 10.2337/diabetes.47.6.894. [DOI] [PubMed] [Google Scholar]
  • 16.Tian J, Atkinson MA, Clare-Salzer M, et al. Nasal administration of glutamate decarboxylase (GAD65) peptides induces Th2 responses and prevents murine insulin-dependent diabetes. J Exp Med. 1996;183:1561–7. doi: 10.1084/jem.183.4.1561. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 17.Balasa B, Boehm BO, Fortnagel A, et al. Vaccination with glutamic acid decarboxylase plasmid DNA protects mice from spontaneous autoimmune diabetes and B7/CD28 costimulation circumvents that protection. Clin Immunol. 2001;99:241–52. doi: 10.1006/clim.2001.5012. [DOI] [PubMed] [Google Scholar]
  • 18.Karges W, Pechhold K, Al Dahouk S, et al. Induction of autoimmune diabetes through insulin (but not GAD65) DNA vaccination in nonobese diabetic mice and in RIP-B7.1 mice. Diabetes. 2002;51:3237–44. doi: 10.2337/diabetes.51.11.3237. [DOI] [PubMed] [Google Scholar]
  • 19.Coon B, An LL, Whitton JL, et al. DNA immunization to prevent autoimmune diabetes. J Clin Invest. 1999;104:189–94. doi: 10.1172/JCI7209. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 20.Bot A, Smith D, Bot S, et al. Plasmid vaccination with insulin B chain prevents autoimmune diabetes in nonobese diabetic mice. J Immunol. 2001;167:2950–5. doi: 10.4049/jimmunol.167.5.2950. [DOI] [PubMed] [Google Scholar]
  • 21.Urbanek-Ruiz I, Ruiz PJ, Paragas V, et al. Immunization with DNA encoding an immunodominant peptide of insulin prevents diabetes in NOD mice. Clin Immunol. 2001;100:164–71. doi: 10.1006/clim.2001.5055. [DOI] [PubMed] [Google Scholar]
  • 22.Wolfe T, Bot A, Hughes A, et al. Endogenous expression levels of autoantigens influence success or failure of DNA immunuzations to prevent type 1 diabetes: addition of Il-4 increases safety. Eur J Immunol. 2002;32:113–21. doi: 10.1002/1521-4141(200201)32:1<113::AID-IMMU113>3.0.CO;2-C. [DOI] [PubMed] [Google Scholar]
  • 23.Quintana FJ, Rotem A, Carmi P, et al. Vaccination with empty plasmid DNA or CpG oligonucleotide inhibits diabetes in nonobese diabetic mice: modulation of spontaneous 60-kDa heat shock protein autoimmunity. J Immunol. 2000;165:6148–55. doi: 10.4049/jimmunol.165.11.6148. [DOI] [PubMed] [Google Scholar]
  • 24.Gurunathan S, Klinman DM, Seder RADNA. vaccines: immunology, application, and optimization. Ann Rev Immunol. 2000;18:927–74. doi: 10.1146/annurev.immunol.18.1.927. [DOI] [PubMed] [Google Scholar]
  • 25.Kaufman DL, Clare-Salzler M, Tian J, et al. Spontaneous loss of T cell tolerance to glutamic acid decarboxylase in murine insulin-dependent diabetes. Nature. 1993;366:69–72. doi: 10.1038/366069a0. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 26.Tisch R, Yang XD, Singer SM, et al. Immune response to glutamic acid decarboxylase correlates with insulitis in non-obese diabetic mice. Nature. 1993;366:72–5. doi: 10.1038/366072a0. [DOI] [PubMed] [Google Scholar]
  • 27.Yoon JW, Yoon CS, Lim HW, et al. Control of autoimmune diabetes in NOD mice by GAD expression or suppression in β cells. Science. 1999;284:1183–7. doi: 10.1126/science.284.5417.1183. [DOI] [PubMed] [Google Scholar]
  • 28.Martin R, Brady JL, Lew AM. The need for IgG2c specific antiserum when isotyping antibodies from C57BL/6 and NOD mice. J Immunol Meth. 1998;212:187–92. doi: 10.1016/s0022-1759(98)00015-5. [DOI] [PubMed] [Google Scholar]
  • 29.Tighe H, Corr M, Roman M, et al. Gene vaccination: plasmid DNA is more than just a blueprint. Immunol Today. 1998;19:89–97. doi: 10.1016/s0167-5699(97)01201-2. [DOI] [PubMed] [Google Scholar]
  • 30.Wiest Ladenburger U, Fortnagel A, Richter W, et al. DNA vaccination with glutamic acid decarboxylase (GAD) generates a strong humoral immune response in BALB/c, C56BL/6, and in diabetes-prone NOD mice. Horm Metab Res. 1998;30:605–9. doi: 10.1055/s-2007-978942. [DOI] [PubMed] [Google Scholar]
  • 31.Tisch R, Wang B, Weaver DJ, et al. Antigen-specific mediated suppression of β cell autoimmunity by plasmid DNA vaccination. J Immunol. 2001;166:2122–32. doi: 10.4049/jimmunol.166.3.2122. [DOI] [PubMed] [Google Scholar]
  • 32.Weaver DJ, Jr, Liu B, Tisch R. Plasmid DNAs encoding insulin and glutamic acid decarboxylase 65 have distinct effects on the progression of autoimmune diabetes in nonobese diabetic mice. J Immunol. 2001;167:586–92. doi: 10.4049/jimmunol.167.1.586. [DOI] [PubMed] [Google Scholar]
  • 33.Danko I, Fritz JD, Jiao S, et al. Pharmacological enhancement of in vivo foreign gene expression in muscle. Gene Ther. 1994;1:114–21. [PubMed] [Google Scholar]
  • 34.Krieg AM. CpG motifs in bacterial DNA and their immune effects. Annu Rev Immunol. 2002;20:709–60. doi: 10.1146/annurev.immunol.20.100301.064842. [DOI] [PubMed] [Google Scholar]
  • 35.Hassett DE, Zhang J, Whitton JL. Induction of antiviral antibodies by DNA immunization requires neither perforin-mediated nor CD8(+)-T cell-mediated lysis of antigen-expressing cells. J Virol. 1999;73:7870–3. doi: 10.1128/jvi.73.9.7870-7873.1999. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 36.Erard F, Garcia-Sanz JA, Moriggl R, et al. Presence or absence of TGF-β determines IL-4-induced generation of type 1 or type 2 CD8 T cell subsets. J Immunol. 1999;162:209–14. [PubMed] [Google Scholar]
  • 37.Pakala SV, Kurrer MO, Katz JD. T helper (Th2) T cells induce acute pancreatitis and diabetes in immune-compromised nonobese diabetic (NOD) mice. J Exp Med. 1997;186:299–306. doi: 10.1084/jem.186.2.299. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 38.Mandel TE, Kovarik J, Koulmanda M. A comparison of organ cultured fetal pancreas allo-, iso-, and xenografts (pig) in non-immunosuppressed non-obese diabetic mice. Am J Pathol. 1995;147:834–44. [PMC free article] [PubMed] [Google Scholar]
  • 39.Lafaille JJ, Van de Keere F, Hsu AL, et al. Myelin protein-specific T helper (Th2) cells cause experimental autoimmune encephalomyelitis in immunodeficient hosts rather than protect them from disease. J Exp Med. 1997;186:307–12. doi: 10.1084/jem.186.2.307. [DOI] [PMC free article] [PubMed] [Google Scholar]
  • 40.Gagnerault MC, Luan JJ, Lotton C, et al. Pancreatic lymph nodes are required for priming of beta cell reactive T cells in NOD mice. J Exp Med. 2002;196:369–77. doi: 10.1084/jem.20011353. [DOI] [PMC free article] [PubMed] [Google Scholar]

Articles from Clinical and Experimental Immunology are provided here courtesy of British Society for Immunology

RESOURCES