Gene Therapy (2005) 12, 1679–1685 & 2005 Nature Publishing Group All rights reserved 0969-7128/05 $30.00 www.nature.com/gt RESEARCH ARTICLE DNA vaccination with an insulin construct and a chimeric protein binding to both CTLA4 and CD40 ameliorates type 1 in NOD mice

Y Chang1,2,SYap1,XGe1, J Piganelli1,2, S Bertera1, N Giannokakis2, C Mathews1,2, G Prud’homme3 and M Trucco1,2 1Division of Immunogenetics, Department of Pediatrics, University of Pittsburgh, School of Medicine, Rangos Research Center, Children’s Hospital of Pittsburgh, Pittsburgh, PA, USA; 2Diabetes Institute of University of Pittsburgh, Children’s Hospital of Pittsburgh, Pittsburgh, PA, USA; and 3Department of Laboratory Medicine and Pathobiology, St Michael’s Hospital and University of Toronto, Toronto, Ontario, Canada

Type 1 diabetes (T1D), a T-cell-mediated autoimmune vector encoding mutant B7.1/CD40L (mB7.1/CD40L) fusion disease, could be attributed to many defects in nonobese protein. This mutant B7.1 binds CTLA4 but not CD28. We diabetic (NOD) mice, including deficient expressions of report that young NOD mice immunized with mbPPI along costimulatory molecules that impair antigen presentation. with mB7.1/CD40L DNA vectors significantly reduced dia- Thus, this deficient antigen presentation may result in a betes incidence while treatment with CTLA4/IgG1 exacer- reduced ability to induce a tolerogenic response through bated diabetes. In conclusion, the combination of mbPPI and negative selection/regulation of autoreactive T cells. Impro- mB7.1/CD40L was able to protect against and perly activated T cells seem to be able to induce autoimmune diabetes in NOD mice possibly by promoting a more efficient responses causing diabetes. To re-establish tolerance to presentation of autoantigen PPI and inducing specific autoantigens by modulating costimulation, we constructed tolerance to PPI by negatively regulating autoreactive T cells. and tested a new type of DNA vaccine encoding a Gene Therapy (2005) 12, 1679–1685. doi:10.1038/ membrane-bound preproinsulin (mbPPI) and a chimeric gene sj.gt.3302578; published online 18 August 2005

Keywords: (T1D); mB7.1/CD40L; autoimmunity; nonobese diabetic (NOD) mouse, tolerance

Introduction Different defects of the immune system in NOD mice may play a role in the loss of tolerance to pancreatic The nonobese diabetic (NOD) mouse, the most widely autoantigens. Low levels of CD86 (B7.2) expression seem used animal model for studying type 1 diabetes (T1D), to contribute to a defective regulation of autoreactive T spontaneously develop autoimmunity characterized by a cells by preventing their full activation and the upregu- T-cell mediated inflammatory process, which ultimately lation of CTLA-4 (CD152).11 It has been reported that on leads to the destruction of insulin-producing b cells.1–4 cell surface of NOD bone marrow dendritic cells (DCs), Prevention of T1D could be obtained by arresting the there is deficient expression of MHC class II, CD80 (B7.1), chronic destruction of pancreatic b cells. This can be CD86 (B7.2) and CD40.12 Also, a developmental defect in achieved by inducing specific tolerance to autoantigens splenic macrophages of NOD mouse was found respon- while islet cell mass is still able to produce sufficient sible for their inability to process and/or present insulin for maintaining normal levels of blood glucose. autoantigens in a tolerogenic fashion, while allowing Proinsulin is normally expressed in b cells and has them to retain the capacity to activate autoreactive T cells been identified as a possible autoantigen.5–8 It has been in the periphery.13 These defects in costimulatory indicated that NOD transgenic mice expressing proinsu- molecules might result in insufficient presenting of lin II did not develop diabetes.6 In contrast, in NOD antigen signals contributing to deficiently activated mice, insufficient proinsulin II expression (ins2À/À) (suboptimal) T cells unable to undergo negative selection within both thymus and peripheral immune competent but sufficient to mount autoimmunity.14–17 It has been organs accelerated diabetes.9,10 indicated that in NOD mice, the islet-reactive T cells were refractory to CTLA-4 and their ability to escape peripheral regulation might be greatly increased as well.18 Importantly, CD28À/À ko NOD mice, in which Correspondence: Dr Y Chang, Division of Immunogenetics, Children’s CD28/B7s interactions are null, developed an earlier Hospital of Pittsburgh, Rangos Medical Center, 3460 Fifth Avenue, Pittsburgh, PA 15213, USA onset of diabetes, increased incidence and severity of 19–21 Received 30 November 2004; accepted 18 May 2005; published autoimmunity. Recently, the genetic mapping of T1D online 18 August 2005 has shown that susceptibility is associated with a Induction of tolerance to autoantigen preproinsulin Y Chang et al 1680 variation in CTLA-4 gene splicing. A reduced production the antigen and expression sites can be chosen at will. of a spliced form encoding a molecule lacking the Moreover, recombinant DNA techniques allow for CD80/CD86 ligand-binding domain contributed to the sequence modifications to improve efficacy, safety and autoimmune tissue destruction in human and NOD the addition of supplemental elements. Our previous mice22 (Figures 1 and 2). studies have shown that induction of tolerance can be DNA vaccination by nonviral techniques, such as the achieved after immunizing recipient NOD mice with intramuscular (i.m.) injection of naked plasmid DNA, plasmid DNA encoding islet antigens by i.m. injection.26 permits local or systemic delivery of immunomodulatory In this study, we have constructed a new type of molecules along with the autoantigens to modulate autoantigen-carrying DNA vector encoding membrane- immunoresponses.23–26 DNA vaccines have the advan- bound Preproinsulin (mbPPI), which is not secreted tage that they can be very easily formulated and also when i.m. delivered into recipients. In order to compare the effect on the tolerance induction to PPI by either promoting or blocking antigen presentation, we have also constructed immuneregulatory vectors encoding mB7.1/CD40L or CTLA4/Fc for DNA vaccination study. DNA vaccination using mbPPI and mB7.1/CD40L DNA vaccines has been evaluated for the induction of tolerance to PPI in NOD mice models.

Results Confirmation of mbPPI expression by RT-PCR and insulin ELISA To determine if mbPPI is membrane-bound protein, COS-7 cells were transiently transfected with mbPPI expression vector for 72 h. From transfected super- Figure 1 Construction of membrane-bound preproinsulin gene (mbPPI). Mouse PPI and a transmembrane sequence from mouseCD80 cDNA were natants, insulin protein was not detectable by insulin amplified by PCR separately. Two PCR products were used as templates for ELISA. This result showed that mbPPI was not expressed second round (overlapping) PCR to generate mbPPI. mbPPI was cloned as soluble protein. However, the transcripts of mbPPI into expression vector VR1255. could be detected either from transiently transfected COS-7 cells (Figure 3a) or from DNA injected muscles (Figure 3b) by RT-PCR with pair of primers that specifically amplify mbPPI gene, a 540-bp PCR product as described in Materials and methods.

Expression of mB7.1/CD40L and CTLA4/Fc constructs in vitro and in vivo COS-7 cells were transiently transfected with purified mB7.1/CD40L and CTLA4/Fc DNA vectors and cul- tured in either DMEM with 10% FBS or serum-free Opti- MEM. The expressions of these soluble fusion proteins from 72 h-transfected supernatants were detected by ELISA (Figure 4a and b), in which mB7.1/CD40L and CTLA4/Fc transfected supernatants (1:40 dilution) showed much higher OD readings compared with control supernatants that confirmed these fusion proteins were expressed as soluble proteins. Western blot showed that these fusion proteins have homodimer conformation (Figure 5).

ab

500bp 500bp

Figure 3 Detection of mbPPI transcripts from transfected cells and DNA vector injected muscles. (a) Confirmation of mbPPI transcripts in transfected COS-7 cells. RNA isolated from transfected cells was used for RT-PCR. (b) Detection of mbPPI transcripts in TA muscles of mice injected with mbPPI vector. Each TA muscle was injected with 50 mg Figure 2 (a) Construction of modified chimeric costimulatory molecule vector DNA. RNA was extracted from muscles 15 days following DNA mutant B7.1-Fc/CD40L. Mutant B7.1/Fc without stop codon and injection. mbPPI transcripts were evaluated by RT-PCR. In (a) and (b), extracellular sequence of CD40L was fused together by overlapping PCR mbPPI transcripts were confirmed by a specific 540-bp PCR product and clone into VR1255 expression vector. (b) CTLA4/IgG1 was generated stained with ethidium bromide in 1% DNA agarose gel. In (b), each lane is by overlapping PCR and cloned into VR1255 vector. from a separate mouse.

Gene Therapy Induction of tolerance to autoantigen preproinsulin Y Chang et al 1681 a 1.4 a 0.6

1.2 0.5 m µ m µ 1 0.4

0.8 0.3

0.6 0.2 0.4 OD Reading At 450 0.1 OD Reading At 450 0.2 0 Blank Vector mB7.1/CD40L 0 COS-7 mB7.1/CD40L (1:3) b 0.7 b 1 0.6 m 0.8 µ

m 0.5 µ 0.4 0.6 0.3

0.4 0.2 OD Reading At 450 0.1

OD Reading At 450 0.2 0 Blank Vector CTLA4/Fc 0 (1:3) COS-7 CTLA4/Fc Figure 6 In vivo transgenic expression of fusion proteins. Serum protein Figure 4 In vitro expressions of mB7.1/CD40L and CTLA4/IgG1. levels were detectable by ELISA 15 days after DNA plasmid vectors were Expressions of soluble proteins (a) mB7.1/CD40L and (b) CTLA4/Fc from i.m. injected. (a) mB7.1/CD40L (serum, 1:3 dilution). (b) CTLA4/IgG1 transiently transfected COS-7 cells were confirmed by ELISA. (serum, 1:3 dilution).

mPPI alone only slightly reduced diabetes incidence, but when mPPI vaccine was codelivered with a DNA vector encoding a fusion protein mB7.1/CD40L, it prevented insulitis and autoimmune diabetes effectively (Figure 7a). Also, we found that CTLA4/Fc could antagonize mPPI protective effects on insulitis and diabetes when Figure 5 Western blot assay of mB7.1/CD40L and CTLA4/Fc proteins. codelivered with mbPPI into young NOD mice (Figure Western blot showed that fusion protein has a bivalent form. 7b). Histology has shown that islets from protected mice mostly remained insulitis-free compared with islets from diabetic mice that showed evident insulitis (Figure 8). For in vivo transgenic expression, 6–8-week-old female NOD mice were i.m. injected with DNA vectors encoding mB7.1/CD40L or CTLA4/Fc. Mice were sacrificed 15 Discussion days after DNA vector injection and sera were tested for protein expression. ELISA results using specific anti- Administration of insulin, or its B:9-23 peptide, to NOD bodies shown that sera (1:3 dilution) from mB7.1/CD40L mice through different routes has been proven to protect or CTLA4/Fc injected mice had much higher OD reading mice from diabetes.27–31 Also, a peptide spanning the compared with sera from mice injected with blank vector B–C junction of proinsulin I (p24–33 epitope) has shown that confirmed detectable levels of proteins from to be an early autoantigen for T1D pathogenesis and was transfected genes in circulation (Figure 6). protective against TID when NOD mice were immunized with this peptide.32,33 T cells responses to PPI epitopes have been observed in the NOD mice model.34 Protection of autoimmune diabetes DNA immunization against insulin or its peptide To test whether PPI could protect from autoimmune constructs has produced controversial results. NOD mice diabetes in NOD mice, we had delivered this expression immunized with a plasmid encoding the B chain of DNA vector into young female NOD mice by i.m. insulin by i.m. injection had a lower disease incidence injection combined with electroporation. We found that and delayed onset of disease.35,36 However, when

Gene Therapy Induction of tolerance to autoantigen preproinsulin Y Chang et al 1682 a Blank Vector mbPPI mbPPI+mB7.1/CD40L 100 (18/19) (15/20) (3/19) 90 80 70 60 50 40 30 20 Diabetes Incidence (%) 10 0 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 Weeks (Age)

b Blank Vector mbPPI mbPPI+CTLA4/Fc 100 (8/10) (6/10) (12/14) 90 80 Figure 8 Histology of pancreatic islets. (a) Islet of prediabetic mouse (6 weeks of age) revealed no insulitis. (b) Islet of mouse treated with blank 70 vector showed a typical insuliitis. (c) Islet of mouse treated with 60 mbPPI+CTLA4/IgG1 showed insulitis. (d) Islet of mouse treated with 50 mbPPI and mB7.1/CD40L (35 weeks of age) showed no insulitis. 40 30 20 of causing hypoglycemia regardless to routes of admin- Diabetes Incidence (%) 10 istration particularly when insulin is used as a means for the intervention in T1D. mbPPI should reduce the risk of 0 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 low blood glucose generally associated with native Weeks (Age) vaccines like insulin or its encoding DNA vaccines. Also, we constructed DNA vectors encoding soluble CD40L/ Figure 7 Diabetes incidence following DNA vaccinations. (a) DNA vaccination with mbPPI and mB7.1/CD40L prevents diabetes development IgG1 or mB7.1/CD40L fusion proteins. CD40L engage- in NOD mice. Female NOD mice were immunized with mbPPI and ment with CD40 on APCs upregulates CD80 (B7.1), mB7.1/CD40L plasmid DNA beginning at 4 weeks of age 3 weeks apart for CD86 (B7.2) and other costimulatory molecules that a total five injections or in the absence of mB7.1/CD40L as well as blank drive T cell further activation. Also, CD40L could play an vector as control. In all, 95% mice (n ¼ 19) immunized with blank vector important tolerogenic role and was needed as tolerogenic developed diabetes at 32 weeks of age. In all, 75% mice (n ¼ 20) immunized signal for B cells to increase expression of Fas (CD95) that with mbPPI DNA alone developed diabetes. Only 16% (n ¼ 19) mice immunized with mbPPI and mB7.1/CD40L DNA developed diabetes. triggered the death and deletion of self-reactive B and 39–42 (Po0.01 compared with control group). (b) DNA vaccination with mbPPI T cells through Fas–FasL interaction. Furthermore, and CTLA4/IgG1 did not prevent diabetes development in NOD mice. studies have shown that the expression of CD40L was Female NOD mice were immunized with mbPPI and CTLA4/IgG1 required for efficient thymic-negative selection, in which plasmid DNA beginning at 4 weeks of age 3 weeks apart for a total five CD5 and CD28 were involved.43 Our results had injections or in the absence of CTLA4/IgG1 as well as blank vector as demonstrated that splenic cells treated in vitro with control. In all, 86% mice (n ¼ 14) of control group developed diabetes at 32 weeks of age. In all, 60% mice (n ¼ 10) immunized with mbPPI DNA soluble CD40L or mB7.1/CD40L protein could increase + alone developed diabetes. In all, 80% (n ¼ 10) mice immunized with the CD86 cell population. Similarly, cell population mbPPI and CTLA4/IgG1 DNA developed diabetes. CTLA4/IgG1 anta- detected with mAb to B220, a marker for B lymphocytes, gonized protective effect of mbPPI by accelerating the development of were increased (unpublished data). We hypothesized diabetes. that these CD86+ APCs, perhaps in a large percentage activated B cells, played their role in presenting antigen signals to T cells. Also the soluble CD86 molecule recombinant DNA encoding murine insulin A and B produced by intramuscularly delivered DNA vector chains fused to IgG–Fc (insulin B chain/Fc) was i.m. could participate in T-cell-negative regulation via injected into young NOD mice, early onset and acceler- CTLA4/CD86 interaction. ated progression of diabetes were observed.37 Some In this study, we found that mbPPI could not be an study has also shown that in female NOD mice, diabetes effective tolerogenic antigen per se. This was consistent development was enhanced after PPI DNA treatment.38 with our previous finding that DNA vector encoding Thus, it is critically important to choose the correct target a soluble PPI could not protect against autoimmune antigens and antigen form that can deliver b-cell diabetes.26 A combination of mbPPI and soluble CD40L antigens in tolerogenic fashion. also could not protect against autoimmune diabetes (data In order to regulate immune response to PPI by not shown). We reasoned that it was possible that presenting tolerogenic signals to T cells, we constructed a although detectable after i.m. injection, protein level of new DNA vaccine vector encoding a nonsecreted mbPPI. CD40L in blood circulation was not sufficient to induce Hyperglycemia seems to be the basis of many complica- enough CD86+ cells, so that activated T cells could not be tions in T1D patients. However, there always is the risk negatively regulated through CTLA4/B7s. Our results

Gene Therapy Induction of tolerance to autoantigen preproinsulin Y Chang et al 1683 also showed that mB7.1/CD40L itself did not offer any Materials and methods protection against autoimmune diabetes. However, when combined with mbPPI, mB7.1/CD40L were able to Animals prevent T1D in NOD mice. Thus, mutant B7.1 portion Female NOD mice of 4 weeks old were purchased from of mB7.1/CD40L was definitely necessary and played Jackson Laboratory (Bar Harbor, ME, USA) and housed a crucial role in this induction of tolerance to antigen in specific pathogen-free facilities at Rangos Medical PPI. We hypothesized that due to its binding specificity, Research Center in Children’s Hospital of Pittsburgh. All mB7.1/CD40L can directly bind to CTLA4 on circulated animal protocols were approved by Animal Research fresh antigen-activated autoreactive T cells and resting and Care Committees (ARCC) of Children’s Hospital of autoreactive T cells while released right after i.m. Pittsburgh. Blood glucose of NOD mice was monitored administration and therefore induce negative regulation starting at 10–12 weeks of age. to these reactive T cells, which could be both suboptimal and already activated T cells. Also, this protein is a Construction of autoantigen DNA vaccine and homodimer that might have higher affinity to bind immunoregulatory molecular vectors CTLA4 than its wild-type form. We assumed that both Murine PPI was cloned and was fused to a transmem- soluble mutant CD80 (mB7.1/CD40L) and CD86 on brane sequence of CD80 (B7.1) to generate mbPPI by CD40L-expanded cells participated in negative regula- overlapping PCR as previously described.26,46,47 Briefly, tion. APCs of mice immunized with mbPPI and mB7.1/ PPI was amplified by PCR (50 primer A: TAGTAGGAA CD40L could present effectively PPI antigen signal to TTCCTGCCTATCTTTCAGGTCATTGTT and 30 primer T cells. As a result, these antigen-activated T cells could B: TAGTTGGCTACTAATGGTGTACAAGTTGCAGTAG undergo tolerance through negative regulation. More- TTCTCCAGCTGGTA) and transmembrane sequence over, It has been indicated that CD40L could function in from mouse B7.1 (kindly provided from Dr Lieping a non-cell-antonomous manner during negative selec- Chen, John Hopkins University) was amplified by PCR tion, which means that CD40–CD40L-dependent nega- (50 primer C: TACCAGCTGGAGAACTACTGCAACTT tive selection does not require CD40L to be expressed on GTACACCATAG-TAGCCAACTA and 30 primer D: those thymocytes that are undergoing negative selec- TATTATGGATCCCTAGCATTTGATGATGACAACGAT tion.42 We hypothesized that central selection may also GAC). Fragment amplified by primer A, B and fragment play a role for tolerance induction in this model. amplified by primer C, D being used as templates, a However, this assumption needs further experimental second round PCR was performed by using primer A evidence. Consistently with others, our results also and D to generate fusion gene encoding mbPPI (Figure showed that blockage of costimulatory CD28/B7s and/ 1). For constructing the immunoregulatory molecular or CTLA4/B7s by delivering soluble CTLA4/Fc exacer- vector, a mutant B7.1/Fc fragment without stop codon bated diabetes. CD80 and CD86 are required to nega- was amplified from a mutant B7.1/Fc gene by PCR (50 tively regulate autoreactive T cells. This supported that primer A: TATTATGCGGCCGCATGGCTTGCAATTGT the deficiency in autoantigen presentation and the CAGTTGATG and 30 primer B: TTCCTCTTCGACCT defective activation of T cells contributed to autoimmu- TATCCAATCTACCAGGAGAGTGGGAGAGGCTCTT). nity and diabetes in NOD mice. This mutant B7.1 (a gift from Dr Yang Liu, Ohio We have observed that syngeneic splenic cells treated University) binds to CTLA4 but not to CD28.48,49 The with PPI and mB7.1/CD40L could delay and reduce extracellular sequence of CD40L (mouse CD40L cDNA diabetes incidence when adoptively transferred into from ATCC, Manassas, VA, USA) was amplified by PCD young NOD mice. Our in vitro results have shown that (50 primer C: AAGAGCCTCTCCCACTCTCCTGGTAG the prediabetic NOD spleen cells treated with PPI and ATTGGATAAGGTCGAAGAGGAA and 30 primer D: mB7.1/CD40L protein could decrease INFg production TAGTAGGAATTCACTGTTCAGAGTTTGAGTAAGCCA). (data submitted for publication). It has been demon- Two fragments were used as templates for second round strated that LPS-activated B cells could downregulate PCR using primer A and D as described above to pathogenic T-cell responses by inducing apoptosis of generate fusion gene mB7.1-Fc-CD40L (mB7.1/CD40L) both diabetogenic T cells and activated B cells. Those (Figure 2). CTLA4/Fc was generated by overlapping activated B cells contributed to the maintenance of PCR to fuse CTLA4 gene (from Dr Lieping Chen, John peripheral self-tolerance and prevented autoimmune Hopkins University) to murine Fc sequence of IgG1. Both diabetes in NOD mice.44 A recent study has demon- fusion genes were cloned into VR1255 plasmid vector strated that syngeneic transplantation of hematopoietic (Vical Inc, San Diego, CA, USA). stem cells (HSC), most likely by generating transgenic APCs, encoding proinsulin totally prevented the deve- lopment of spontaneous autoimmune diabetes in Detection of in vitro expression of mbPPI NOD.45 and immunoregulatory molecules In summary, codelivery of a membrane bound PPI Plasmid DNA vectors were amplified and purified by DNA vaccine with an mB7.1/CD40 encoding vector QIAGEN Endofree plasmid Giga kit (QIAGEN Inc., could prevent autoimmunity and T1D in NOD mouse. Valencia, CA, USA), according to manufacture’s instruc- In contrast, when codelivered with mbPPI, CTLA4/Fc tion. Purified DNA was dissolved in 0.85% saline. exacerbated diabetes. These studies supported working To confirm mbPPI expression, COS-7 cells were hypothesis that deficient T-cell activation might be transfected with plasmid DNA by Fugen-6 (Roche responsible for failure of negative selection or tolerance diagnostics Corp., Indianapolis, IN, USA). Insulin pro- induction to autoantigens, leading to T1D. More im- tein was tested from supernatants 3 days after transfec- portantly, our studies suggest that this new type of tion by using mouse insulin ELISA kit (ALPCO vaccine can be well suited for future clinical applications. Diagnostics, Windham, NH, USA). mbPPI transcription

Gene Therapy Induction of tolerance to autoantigen preproinsulin Y Chang et al 1684 from transfected cells was analysed by RT-PCR. Total glucose levels 4300 mg/dl. For analysis of insulitis, RNA was isolated from transfected cells by trizol mice were killed and pancreas was fixed in 10% buffered procedure and cDNA were synthesized and amplified formalin and histological slides were prepared and by PCR by using one-step RT-PCR system (Invitrogen, stained with hematoxylin & eosin. Insulitis was deter- Carlsbad, CA, USA) with a unique primer set (50 primer mined as previously described.26,46,47 ATGGCCCTGTTGGTGCACTTC and 30 primer TTTGAT GATGACAACGATGAC) that specifically amplified Statistical analysis 0 a 540-bp amplifier from mbPPI, since its 5 primer w2 0 The Chi-square ( ) statistical analysis was performed. matches PPI and its 3 primer matches transmembrane Statistical significance is defined as Po0.05. sequence. To confirm secretion of soluble immunoregulatory molecules, COS-7 cells were transiently transfected with plasmid DNA vectors encoding mB7.1/CD40L and Acknowledgements CTLA4/Fc and cultured either in DMEM with 10% FBS This study was supported by Research Advisory or serum-free Opti-MEM. The 72 h-transfected super- Committee of Children’s Hospital of Pittsburgh and natants were analysed for protein secretions by ELISA. Juvenile Diabetes Research Foundation International. An anti-mouse CD40L as capture Ab and an anti-mouse CD80 as detection Ab (BD Pharmingen, San Diego, CA, USA) were used for ELISA to detect mB7.1/CD40L References fusion protein. An anti-mouse CTLA4 as capture Ab and an anti-mouse IgG1 as detection Ab (BD Pharmingen, 1 Adorini L, Gregori S, Harrison LC. Understanding autoimmune San Diego, CA, USA) were used for ELISA to detect diabetes: insights from mouse models. Trends Mol Med 2002; 8: CTLA4/Fc fusion protein. These fusion proteins were 31–38. further analysed by Western blot. Fusion proteins from 2 Serreze DV, Leiter EH. Genes and cellular requirements for serum-free medium were separated on 12% nondena- autoimmune diabetes susceptibility in nonobese diabetic mice. tured PAGE gel and transferred into PVDF membrane. Curr Dir Autoimmun 2001; 4: 31–67. The proteins were probed with anti-mouse IgG antibody 3 Suarez-Pinzon WL, Rabinovitch A. Approaches to type 1 (BD Pharmingen, San Diego, CA, USA). diabetes prevention by intervention in cytokine immunoregula- tory circuits. Int J Exp Diabetes Res 2001; 2: 3–17. Confirmation of in vivo transgenic expression 4 Gottlieb PA, Hayward AR. Cytokine and immunosuppressive mbPPI, mB7.1/CD40L and CTLA4/Fc vectors were therapies of type 1 diabetes mellitus. Endocrinol Metab Clin North i.m. administrated into tibialis anterior muscles with Am 2002; 31: 477–495. 50 mg/muscle of plasmid. After DNA injection, electro- 5 Rudy G et al. Similar peptides from two beta cell autoantigens, poration was immediately applied to DNA injected proinsulin and glutamic acid decarboxylase, stimulate T cells of muscles by caliper-type electrodes of ECM830 (Gene- individuals at risk for insulin-dependent diabetes. Mol Med 1995; 1: 625–633. tronics Inc., San Diego, CA, USA) as we have previously 26,50 6 French MB et al. Transgenic expression of mouse proinsulin II described. prevents diabetes in nonobese diabetic mice (published erratum For mbPPI in vivo expression, mbPPI transcripts were appears in Diabetes 1997 May 46:924). Diabetes 1997; 46: 34–39. detected from DNA injected muscles by RT-PCR. Briefly, 7 Chen W et al. Evidence that a peptide spanning the B–C junction mice were killed and DNA vector-injected TA muscles of Proinsulin is an early autoantigen epitope in the pathogenesis were excised and homogenized. Total mRNA was of type 1 diabetes. J Immunol 2001; 167: 4926–4935. isolated by trizol procedure and transcripts were 8 Ivana D-B, Bernhard OB, Ziegler A-G. Predominantly recog- analysed by RT-PCR following the same protocol as nized proinsulin epitopes in individuals with and described for mbPPI-transfected COS-7 cells. without islet cell autoimmunity. J Autoimmun 2002; 18: 55–66. In vivo expressions of immunoregulatory molecule 9 Dubois-Lafforgue D et al. Proinsulin 2 knockout NOD mice: fusion proteins mB7.1/CD40L and CTLA4/Fc were a model for genetic variation of insulin gene expression in type 1 confirmed by protein levels in the sera of mice receiving diabetes. Diabetes 2002; 51 (Suppl 3): S489–S499. DNA injections by ELISA. 10 Thebault-Baumont K et al. Acceleration of type 1 diabetes mellitus in proinsulin 2-deficient NOD mice. J Clin Invest 2003; DNA vaccination 111: 851–857. For DNA vaccination, 4-week-old female NOD mice 11 Dahlen E, Hedlund G, Dawe K. Low CD86 expression in received i.m. DNA injection (100 mg/muscle) for a total the nonobese diabetic mouse results in the impairment of both 500 mg DNA in five injections. For these five DNA T cell activation and CTLA-4 up-regulation. J Immunol 2000; 164: injections, tibialis anterior muscles of both sides were 2444–2456. alternatively injected with plasmid DNA in 3 weeks 12 Strid J et al. A defect in bone marrow derived dendritic cell maturation in the nonobesediabetic mouse. Clin Exp Immunol apart. When two plasmids were codelivered in the same 2001; 123: 375–381. m m site, they were mixed in equal amounts (50 g:50 g) 13 Piganelli JD, Martin T, Haskins K. Splenic macrophages from the prior to i.m. injections. The same total amount of blank NOD mouse are defective in the ability to present antigen. was always injected as for the experimental groups. Diabetes 1998; 47: 1212–1218. 14 Rapoport MJ et al. Thymic T cell anergy in autoimmune Diagnosis of diabetes and histology of insulitis nonobese diabetic mice is mediated by deficient T cell receptor Blood glucose levels were monitored once weekly with a regulation of thr pathway of p21 ras activation. J Exp Med 1993; Precision QID Meter (MediSense, Incorporated, Bedford, 177: 1221–1226. MA, USA) when mice were at 12 weeks of age. Diabetes 15 Serreze DV. Autoimmune diabetes results from genetic defects was defined by two consecutive nonfasting blood manifest by antigen presenting cells. FASEB J 1993; 7: 1092–1096.

Gene Therapy Induction of tolerance to autoantigen preproinsulin Y Chang et al 1685 16 Salojin K et al. Impaired plasma membrane targeting of Grb2- 33 Martinez NR et al. Disabling an integral CTL epitope allows murine son of sevenless (mSOS) complex and differential suppression of autoimmune diabetes by intranasal proinsulin activation of the Fyn-T cell receptor (TCR)-z-Cb1 pathway peptide. J Clin Invest 2003; 111: 1365–1371. mediate T cell hyporesponsiveness in autoimmune nonobese 34 Halbout P et al. T cell response to preproinsulin I and II in the diabetic mice. J Exp Med 1997; 186: 887–897. nonobese diabetic mouse. J Immunol 2002; 169: 2436–2443. 17 Serreze DV, Gaskins HR, Leiter EH. Defects in the differentiation 35 Urbanek-Ruiz I et al. Immunization with DNA encoding an and function of antigen presenting cells in NOD/Lt mice. immunodominant peptide of insulin prevents diabetes in NOD J Immunol 1993; 150: 2534–2543. mice. Clin Immunol 2001; 100: 164–171. 18 Piganelli JD et al. Cytotoxic T lymphocyte antigen 4 (CD152) 36 Bot A et al. Plasmid vaccination with insulin B chain prevents regulates self-reactive T cells in BALB/c but not in the autoimmune diabetes in nonobese diabetic mice. J Immunol 2001; autoimmune NOD mouse. J Autoimmun 2000; 14: 123–131. 167: 2950–2955. 19 Lenschow DJ et al. CD28/B7 regulation of Th1 and Th2 subsets 37 Weaver Jr DJ, Liu B, Tisch R. Plasmid DNAs encoding insulin in the development of autoimmune diabetes. Immunity 1996; 5: and glutamic acid decarboxylase 65 have distinct effects on the 285–293. progression of autoimmune diabetes in nonobese diabetic mice. 20 Arreaza GA et al. Neonatal activation of CD28 signaling J Immunol 2001; 167: 586–592. overcomes T cell anergy and prevents autoimmune diabetes by 38 Karges W et al. Induction of autoimmune diabetes through an IL-4-dependent mechanism. J Clin Invest 1997; 100: 2243–2253. insulin (but not GAD65) DNA vaccination in nonobese diabetic 21 Lenschow DJ et al. Differential effects of anti-B7-1 and anti-B7-2 and in RIP-B7.1 mice. Diabetes 2002; 51: 3237–3244. monoclonal antibody treatment on the development of diabetes 39 van Essen D, Kikutani H, Gray D. CD40 ligand-transduced in the nonobese diabetic mouse. J Exp Med 1995; 181: 1145–1155. co-stimulation of T cells in the development of helper function. 22 Ueda H et al. Association of the T-cell regulatory gene CTLA4 Nature 1995; 378: 620–623. with susceptibility to autoimmune disease. Nature 2003; 423: 40 Goodnow CC. Pathways for self-tolerance and treatment of 506–511. autoimmune diseases. The Lancet 2001; 357: 2115–2121. 23 Prud’homme GJ. Gene therapy of autoimmune diseases with 41 Lederman S. The role of CD154 (CD40-ligand) in costimulation. vectors encoding regulatory cytokines or inflammatory cytokine Transplantation Proceedings 2001; 33: 202–206. inhibitors. J Gene Med 2000; 2: 222–232. 42 Williams JA, Sharrow SO, Adams AJ, Hodes RJ. CD40 ligand 24 Garren H, Steinman L. DNA vaccination in the treatment of functions non-cell autonomously to promote deletion of self- autoimmune diseases. Curr Dir Autoimmun 2000; 2: 203–216. reactive thymocytes. J Immunol 2002; 168: 2759–2765. 25 Scheerlinck JY. Genetic adjuvants for DNA vaccines. Vaccine 43 Li R, Page DM. Requirement for a complex array of costimu- 2001; 19: 2647–2656. lators in the negative selection of autoreactive thymocytes 26 Prud’homme GJ, Chang Y, Li X. Immunoinhibitory DNA vaccine in vivo. J Immunol 2001; 166: 6050–6056. protects against autoimmune diabetes through cDNA encoding 44 Tian J et al. Lipopolysaccharide-activated B cells down-regulate a selective CTLA-4 (CD152) ligand. Human Gene Ther 2002; 13: Th1 immunity and prevent autoimmune diabetes in nonobese 395–406. diabetic mice. J Immunol 2001; 167: 1081–1089. 27 Polanski M, Melican NS, Zhang J, Weiner HL. Oral administra- 45 Steptoe RJ, Ritchie JM, Harrison LC. Transfer of hematopoietic tion of the immunodominant B-chain of insulin reduces diabetes stem cells encoding autoantigen prevents autoimmune diabetes. in a co-transfer modes of diabetes in the NOD mouse and is J Clin Invest 2003; 111: 1357–1363. associated with a switch from Th1 to Th2 cytokines. J Autoimmun 46 Prud’homme GJ, Chang Y. Prevention of autoimmune diabetes 1997; 10: 339–346. by intramuscular gene therapy with a nonviral vector encoding 28 Harrison LC, Dempsey-Collier M, Kramer DR, Takihashi K. an interferon-g receptor/IgG1 fusion protein. Gene Therapy 1999; Aerosol insulin induces regulatory CD8 gamma/delta T cells 6: 771–777. that prevent murine insulin-dependent diabetes. J Exp Med 1996; 47 Chang Y, Prud’homme GJ. Intramuscular administration of 184: 2167–2174. expression plasmids encoding interferon-gamma-receptor/IgG1 29 Ramiya VK, Shang XZ, Wasserfall CH, Maclaren NK. Effect of or IL-4/IgG1 chimeric proteins protects from autoimmunity. oral and intravenous insulin and glutamic acid decarboxylase in J Gene Med 1999; 1: 415–423. NOD mice. Autoimmunity 1997; 26: 139–151. 48 Guo Y, Wu Y, Kong X, Liu Y. Identification of conserved 30 Hutchings P, Cooke A. Comparative study of the protective amino acids in murine B7-1IgV domain critical for CTLA4/ effect afforded by intravenous administration of bovine or ovine CD28:B7 interaction by site-directed mutagenesis: a novel insulin to young NOD mice. Diabetes 1995; 44: 906–910. structural model of the binding site. Mol Immunol 1998; 35: 31 Daniel D, Wegmann DR. Protection of nonobese diabetic mice 215–225. from diabetes by intranasal or subcutaneous administration of 49 Guo Y et al. Mutational analysis of an alternatively spliced insulin B-(9-23). Proc Natl Acad Sci USA 1996; 93: 956–960. product of B7 defines its CD28/CTLA4-binding site on 32 Chen W et al. Evidence that a peptide spanning the B–C junction immunoglobulin C domain. J Exp Med 1995; 181: 1345–1355. of proinsulin I is an early autoantigen epitope in the pathogen- 50 Lawson BR et al. Treatment of murine lupus with cDNA esis of type 1 diabetes. J Immunol 2001; 167: 4926–4935. encoding IFN-gammaR/Fc. J Clin Invest 2000; 106: 207–215.

Gene Therapy