Transplantation of bone marrow genetically engineered to express proinsulin II protects against autoimmune insulitis in NOD mice

THE JOURNAL OF GENE MEDICINE R E S E A R C H A R T I C L EJ Gene Med 2006; 8: 1281–1290.Published online 21 September 2006 in Wiley InterScience (www.interscience.wiley.com) DOI: 10.1002/jgm.968

Transplantation of bone marrow geneticallyengineered to express proinsulin II protects againstautoimmune insulitis in NOD mice

James Chan1†

Warren Clements1

Judith Field1†

Zeyad Nasa1†

Peter Lock2

Felicia Yap3

Ban-Hock Toh1†

Frank Alderuccio1,*

1Department of Immunology, MonashUniversity, Commercial Road,Prahran, Victoria 3181, Australia2Department of Surgery, University ofMelbourne, Level 5 Clinical SciencesBuilding, Royal Melbourne Hospital,Victoria 3050, Australia3Danielle Alberti Memorial Centre forDiabetes Complications, Baker HeartResearch Institute, P.O. Box 6492, St.Kilda Rd. Central, Melbourne8008, Australia

*Correspondence to:Frank Alderuccio, Department ofImmunology, Monash University,Central and Eastern Clinical School,Commercial Road, Prahran, Victoria3181, Australia.E-mail: [email protected]

†Present address: AutoimmunityLaboratory, Centre for InflammatoryDiseases, Monash Institute ofMedical Research, MonashUniversity, Clayton, Victoria 3168,Australia.

Received: 30 May 2006Revised: 23 July 2006Accepted: 24 July 2006

Abstract

Background Type 1 diabetes (T1D) is a T-cell-dependent autoimmunedisease resulting from destructive inflammation (insulitis) of the insulin-producing pancreatic β-cells. Transgenic expression of proinsulin II by a MHCclass II promoter or transfer of bone marrow from these transgenic miceprotects NOD mice from insulitis and diabetes. We assessed the feasibility ofgene therapy in the NOD mouse as an approach to treat T1D by ex vivo geneticmanipulation of normal hematopoietic stem cells (HSCs) with proinsulin IIfollowed by transfer to recipient mice.

Methods HSCs were isolated from 6–8-week-old NOD female mice andtransduced in vitro with retrovirus encoding enhanced green fluorescentprotein (EGFP) and either proinsulin II or control autoantigen. Additionalcontrol groups included mice transferred with non-manipulated bone marrowand mice which did not receive bone marrow transfer. EGFP-sorted or non-sorted HSCs were transferred into pre-conditioned 3–4-week-old female NODmice and insulitis was assessed 8 weeks post-transfer.

Results Chimerism was established in all major lymphoid tissues, rangingfrom 5–15% in non-sorted bone marrow transplants to 20–45% in EGFP-sorted bone marrow transplants. The incidence and degree of insulitis wassignificantly reduced in mice receiving proinsulin II bone marrow comparedto controls. However, the incidence of sialitis in mice receiving proinsulin IIbone marrow and control mice was not altered, indicating protection frominsulitis was antigen specific.

Conclusions We show for the first time that ex vivo genetic manipulationof HSCs to express proinsulin II followed by transplantation to NOD micecan establish molecular chimerism and protect from destructive insulitis inan antigen-specific manner. Copyright 2006 John Wiley & Sons, Ltd.

Keywords bone marrow transplantation; gene therapy; autoimmunity;hematopoietic stem cells; insulitis; proinsulin II

Introduction

Our immune system is designed to protect us from foreign pathogens andnot adversely react to the many self-antigens that make up our own tis-sues. However, since 5–6% of the population develop autoimmune diseasessuch as type 1 diabetes (T1D), multiple sclerosis and rheumatoid arthritis[1], it is apparent that the mechanisms that eliminate self-reactivity are not

Copyright 2006 John Wiley & Sons, Ltd.

1282 J. Chan et al.

fail-safe. During development of the immune system,central and peripheral tolerance mechanisms establishand maintain tolerance to self-antigens. Central toleranceis dominated by deletion of T cells in the thymus driven byinteraction between T cells and antigen-presenting cells(APCs), of which bone marrow (BM)-derived dendriticcells (DCs) and thymic epithelium play major roles [2–6].Studies of Aire, a molecule ascribed with transcriptionalregulatory function that controls thymic expression ofperipheral self-antigens, have highlighted how subtlechanges in ectopic expression of these antigens canmarkedly influence the immune repertoire and expressionof autoimmunity [7,8].

In autoimmune diseases, specific self-antigens aretargeted by the immune system with a resultant pathologyin the target organ that can lead to clinical symptoms.While many autoimmune diseases can be controlled bynon-specific immunosuppression, they remain incurable.Type 1 diabetes (T1D) is a chronic autoimmune diseaseassociated with progressive inflammation of the islets ofLangerhans (insulitis) that leads to destruction of islets,and resulting in diabetes [9]. In the NOD mouse modelof T1D, the infiltrate is initially ‘benign’ and limitedto a peri-islet distribution that eventually invades theislets to become ‘malignant’ with accompanying β-celldestruction and the clinical manifestation of diabetes.While a number of β-cell antigens have been implicatedin T1D [10], compelling observations identify insulin asa key causative self-antigen driving destructive insulitisin humans and mice. Mice deficient in proinsulinII expression have increased insulitis, implicating aprotective role for thymic proinsulin II expression inT1D development [11]. Peripheral T cells reactive toinsulin are present in some subjects at risk of T1D witha greater proportion showing reactivity to a peptidespanning the insulin B-chain and the connecting (C-)peptide of proinsulin [12,13]. T cell clones from NODislets have reactivity to the insulin B-chain (aminoacids 9–23) [14] and when transferred to young orlymphopenic NOD mice initiate diabetes [15,16]. FemaleNOD mice with a mutant insulin (in which a changeof residue 16 on the B chain from tyrosine to alanineabrogated the response of insulin-B-chain-9-23-reactiveT cell clones) did not develop insulitis or T1D [17]. Inhuman T1D, a high proportion of T cell clones isolatedfrom pancreatic draining lymph nodes react with insulinpeptides [18].

Our present study is part of our long-term interest indeveloping strategies to cure autoimmune disease. Wewere the first to show that transgenic expression of anautoimmune disease-associated self-antigen, the gastricH/K ATPase β-subunit, in MHC class II positive cells,including those in the thymus, prevented the developmentof autoimmune gastritis [19]. Furthermore, and ofclinical relevance, tolerance to the self-antigen could betransferred by the bone marrow (BM) compartment oftransgenic animals, supporting the role of BM-derivedcells in tolerance induction [20]. Using the same MHCclass II promoter and strategy, French et al. [21] showed

that transgenic expression of proinsulin II was alsosufficient to establish tolerance and prevent insulitisand T1D and that tolerance can also be transferred bytransplantation of BM cells from these transgenic mice[22]. The strategy of using bone marrow transplantation(BMT) to treat autoimmune disease is not new. However,allogeneic BM has proven unacceptable with its significantmortality rate due to graft-versus-host disease [23] andautologous BM can be associated with high relapse rates[24], likely due to re-emergent self-reactive T cells.Given this a strategy is clearly required which canpromote the deletion of autoreactive cells within thethymus that may be associated with inducing relapse.Unlike the transgenic mouse studies, transgenic BM isnot available for transplantation in humans, and wesuggest that coupling BMT with ex vivo manipulationof autologous BM to express autoantigen can providea strategy for the treatment of patients with autoimmunedisease without the risk of relapse [25]. Here, we extendon the observation of transplantation with transgenicBMT to show that transduction of normal NOD BMcells to express proinsulin II before transfer into youngNOD mice can establish levels of molecular chimerismsufficient to dramatically reduce the incidence andseverity of insulitis. Our data point the way towards agene therapy approach to control insulitis and the ensuingdiabetes.

Materials and methods

Mice

NOD/Lt mice were purchased from Animal ResourcesCentre (Western Australia). As bone marrow (BM) donors6–8-week-old female mice were used, and 3–4-week-oldfemale mice were used for BM recipients. All experi-ments were performed in accordance with institutionalguidelines.

Vector construction

The mouse proinsulin II (proinsulin II) gene was clonedby polymerase chain reaction (PCR) from NOD genomicDNA (a gift from Dr G. Morahan) using the primer set:sense: 5′-GTAGGCTGGGTAGTGGTGGGTCTAGTTG-3′ ;anti-sense: 5′-CTATCCTCAACCCAGCCTATCTTCCAGG-3′. Restriction enzyme sites Xho1 and Not1 were includedin the sense and anti-sense oligonucleotides respectivelyfor subcloning. The proinsulin II gene sequence wasconfirmed with published sequence and subcloned intothe retrovirus vector pMYs-IG [26] (Figure 1A) to createthe vector pMYs-proinsulin II-IRES-EGFP (pMYs-PIG). Aretroviral vector pMYs-BIG encoding the cDNA of thegastric H/K ATPase β-subunit was used as control.

Copyright 2006 John Wiley & Sons, Ltd. J Gene Med 2006; 8: 1281–1290.DOI: 10.1002/jgm

Manipulating stem cells to treat autoimmunity 1283

Figure 1. (A) Structure of bicistronic retroviral vector. Modified pMYs-IG retroviral vector, flanked by its long terminal repeats (LTR)carrying the proinsulin II gene. Proinsulin II (909 bp) was subcloned upstream of an internal ribosomal entry site (IRES) and EGFPgene. The proinsulin gene comprises exons 1 and 2 separated by an intervening intron. On transcription, removal of the interveningintron by splicing is predicted to generate a 460 bp mRNA. (B) Proinsulin II expression. Reverse transcription polymerase chainreaction (RT-PCR) analysis of pMYs-PIG-transfected BOSC23 cells shows expression of a ∼460 bp band, corresponding to proinsulinII mRNA. Non-transfected cells and transfected cells without reverse transcriptase did not produce a PCR product. (C) Detectionof (pro)insulin. Sorted EGFP+ NIH 3T3 cells transduced with either pMYs-PIG or pMYs-BIG were cultured for 7 days, after whichsupernatants (s/n) and cell lysates were collected and subjected to radioimmunoassay. (Pro)insulin was detected in supernatantand cell lysate of pMYs-PIG- but not pMYs-BIG-transduced NIH 3T3 cells. ND: not detected

Production of recombinantretroviruses and determination of viraltiters

Recombinant retrovirus was generated by transfection ofBOSC23 cells [27] with pMYs-PIG and pMYs-BIG andsupernatants were collected at 48 and 72 h. In brief,2.5 × 106 BOSC23 cells were seeded onto 60 mm dishes in4 ml Dulbecco’s modified Eagle’s medium (DMEM)/10%fetal calf serum (FCS) overnight before transfection.Chloroquine to a final concentration of 25 µM wasadded to the medium followed by addition of DNA(15 µg)/CaCl2 (0.25 M)/Hepes (10 mM) buffered salinesolution (pH 7.1). Cells were incubated for 7–12 h at37 ◦C (10% CO2), washed with Tris-buffered saline andfresh medium added. After 48 and 72 h, supernatantswere collected, filtered through a 0.45 µm filter and storedat −70 ◦C.

Retroviral titers were determined using NIH 3T3 cells.In brief, 50 000 cells were seeded in each well of a 6-wellplate overnight in DMEM/10% FCS. Culture medium wasaspirated and 2 ml of supernatant or serial 1/5 dilution(together with polybrene, final concentration 5 µg/ml)

and HEPES buffered solution (final concentration 10 mM)were added to cells. Cells were spin-infected at roomtemperature for 45 min at 310 g. Supernatant wasreplaced with fresh DMEM/10% FCS medium and cellsincubated at 37 ◦C/5% CO2. After 48 h, the percentageof infected cells (enhanced green fluorescent protein(EGFP) positive) was determined by flow cytometry(FACScan, Becton Dickinson, USA). A concentrationversus percentage EGFP+ curve was plotted and viraltiters determined using the linear portion of the curve.The retrovirus titer for each retrovirus determined in thisway was ∼1.25 × 106 infectious units (IU)/ml.

Bone marrow harvest, retroviraltransduction and transfer

Donor NOD mice were treated with 5-fluorouracil(200 mg/kg body weight) 3.5 days before BM harvest.Under sterile conditions, BM was extracted from thefemur and tibia using Hank’s balanced salt solution(HBSS; Invitrogen) with 1% FCS. Cells were collected,washed, red cells lysed, and plated onto 24-well plates at2 × 106 cells/well in DMEM/10% FCS supplemented with


1284 J. Chan et al.

recombinant cytokines; rmIL-6 (10 ng/ml, R&D Systems)and rmSCF (50 ng/ml, R&D Systems). After 24 h, cellswere spin-infected with 2 ml of viral supernatant for1.5 h at 680 g. Viral supernatant was replaced withDMEM/10% FCS and rmIL-6 and rmSCF as describedabove. Cells were cultured for 3–4 days with a change ofmedia and cytokines every 2 days. Recipient mice weresubjected to 650–800 cGy total body irradiation at least4 h prior to transfer of hematopoietic stem cells (HSCs).Bulk transduced cells (1 × 106) or sorted EGFP-expressingcells (3 × 103) (FACStar Plus, Beckton Dickinson, USA)were injected into the lateral tail vein. On days 0, 7, 14 and21, 0.25 mg of purified anti-CD4 (clone GK1.5) antibodywas injected intraperitoneally into BMT recipients anduntreated control groups to deplete any residual CD4 Tcells. Mice were killed at 8 weeks post-BMT for analysis.

mRNA analysis

Total mRNA from BOSC23 cells transfected with pMYs-PIG was prepared using an mRNA isolation kit (Qiagen,USA). Messenger RNA (2 µg) was treated with DNaseI (Roche Diagnostics, USA), reverse-transcribed andamplified by PCR (one-step PCR kit; Invitrogen, USA)using the proinsulin II primer set as mentioned above.Identification of the 460 bp proinsulin II mRNA productcould be easily distinguished from the predicted larger909 bp genomic DNA product (Figure 1A). PCR productswere separated by agarose gel electrophoresis.

Radioimmunoassay

(Pro)insulin concentration (ng/ml) was determined usinga radioimmunoassay kit (sensitivity of 0.02 ng/ml)supplied by Linco (Linco Research, Missouri, USA).Simply, sorted EGFP+ NIH 3T3 cells transduced witheither pMYs-PIG or pMYs-BIG were cultured for 7 days.Culture supernatants and cell lysates were collected forradioimmunoassay.

Histology

Pancreases and salivary glands were fixed in 10%buffered formalin, embedded in paraffin, and sections(5 µm) were collected at least 50 µm apart, stainedwith haematoxylin and eosin (H&E) and viewed by lightmicroscopy. Insulitis was graded according to the level ofmononuclear cell infiltration of pancreatic islets. Grade0, normal islets; grade 1, mononuclear cell infiltratepredominantly peri-islet in less than 25% of the islet;grade 2, infiltrate observed in 25–50% of islet; grade3, infiltrate in 50–75% of the islet; grade 4, >75% ofislet containing mononuclear cell infiltrate. A minimumof 14 islets were scored from different levels of eachpancreas. Salivary glands were scored by the presence ofpockets of mononuclear cell infiltrates in tissue sections.

Scoring of islets and salivary glands was performed bytwo independent investigators.

Thymic frozen sections of pMYs-PIG chimeras were cutat 5 µm thick, air-dried and acetone fixed. Sections wereblocked with 5% horse serum for 20 min, then incubatedwith rabbit anti-keratin (Dako, Denmark) followed byAlexa (627 nm) anti-rabbit (Molecular Probes, USA) andanti-CD11c-APC (Pharmingen, USA). Each antibody wasincubated for 1 h followed by extensive washing. Sectionswere viewed under a Meta confocal microscope (Zeiss,Germany).

Antibodies and flow cytometry

Single cell suspensions of spleen, thymus and inguinallymph node were prepared and stained with monoclonalantibodies, and acquired using a FACScan flow cytometerwith CellQuest software (Becton Dickinson, USA).The following monoclonal antibodies (purchased fromPharmingen, USA) were used: anti-CD4-APC, anti-CD8-PerCp, anti-CD11b-PerCp, anti-CD11c-APC, anti-CD19-PE, and anti-MHC II (IAg7)-PE.

Statistical analyses

Mann-Whitney, two-tailed, data are expressed as mean ±standard error of the mean (SEM). A p value of less than0.05 was considered significant.

Results

Retroviral vectors encoding proinsulinII express proinsulin in vitro

NOD mouse genomic DNA encoding the proinsulin IIgene was PCR-amplified and subcloned into the multiplecloning site of the pMYs-IG vector under the constitutivetranscriptional control of the viral promoter/enhancerlong terminal repeat (LTR) to produce pMYs-PIG(Figure 1A). As a control, we generated retrovirusencoding an irrelevant self-antigen, the gastric H/KATPase β-subunit mini-gene (pMYs-BIG) [19]. Generationof the proinsulin II mRNA transcript in transfectedBOSC23 cells was confirmed by reverse-transcription(RT)-PCR with a 460 bp product corresponding tothe expected size of proinsulin II mRNA (Figure 1B).Messenger RNA from non-transfected BOSC23 cells ornot reverse-transcribed did not produce a PCR product.Attempts to detect retrovirus-driven proinsulin II proteinby antibody staining in these cells were unsuccessful(data not shown) and may be due to the level of antigenexpressed being insufficient or, more likely, indicatingrelease of proinsulin from the cells. Consequently, culturesupernatants and cell lysates of sorted EGFP+ NIH3T3 cells transduced with either pMYs-PIG or pMYs-BIG(over 30 passages) were analyzed by radioimmunoassay



for (pro)insulin expression. The results showed that(pro)insulin was detectable in the supernatant and celllysates of pMYs-PIG- but not pMYs-BIG-transduced cells(Figure 1C). Since EGFP and proinsulin II are encodedby a single transcript and EGFP mRNA expression intransduced NIH 3T3 correlates with proinsulin II mRNAexpression (data not shown), we suggest that the presenceof EGFP expression reflects proinsulin II expression. EGFPexpression in transduced NIH 3T3 cells was stable anddetectable after more than 30 passages.

Transfer of EGFP-sorted, transducedBM cells enhances molecularchimerism in recipient mice

Haematopoietic stem cell (HSC)-enriched bone marrowfrom 5′-fluorouracil-treated donor mice was transducedwith pMYs-PIG or pMYs-BIG retrovirus. In our hands, thisprotocol routinely results in a bulk transduction efficiencyof approximately 20% as determined by EGFP expression(data not shown). The ability to identify and isolate EGFP-expressing cells allowed us to compare the influence onchimerism and insulitis when EGFP+ transduced cells

were purified before transfer. Preconditioned 3–4-week-old female NOD mice received either 1 × 106 non-sortedor 3 × 103 sorted EGFP+ transduced cells.

Eight weeks after BMT, mice were killed and EGFPexpression was determined in lymphoid organs. EGFPexpression could readily be detected in lymphoid tissues(Figure 2A) including thymus (∼14%), spleen (∼14%)and inguinal lymph nodes (∼5%) of mice that receivednon-sorted, transduced BM (Figure 2B). Even morestriking was the significant increase in the levels ofchimerism achieved (thymus ∼20%, spleen ∼45% andinguinal lymph node ∼25%) in mice that received asfew as 3 × 103 cells sorted for EGFP expression priorto transfer (Figure 2B) (p < 0.05). Analysis of individualcell populations indicated that the degree of chimerismfrom sorted BMT groups, as compared to non-sortedBMT groups, varied and ranged from hardly any increasein chimerism in CD19+ cells to a 7-fold increase inchimerism of CD11c + MHC class II+ DCs in the thymus(Table 1). Levels of chimerism achieved in a range ofcell types including CD4, CD8, MHC II and CD19 cellswere similar for both non-sorted pMYs-PIG and pMYs-BIG BMT groups (Table 1), indicating that the levels ofchimerism achieved were not determined by the antigenencoded. Interestingly, chimerism levels of CD19+ cells,in particular in the thymus, were higher than in the

Figure 2. Chimerism levels in thymus, spleen and inguinal lymph nodes of recipient mice. (A) Representative histogram ofthymocytes from a NOD mouse 8 weeks after receiving sorted pMYs-PIG-transduced bone marrow (BM) showing 35% chimerismbased on EGFP expression. Grey indicates background EGFP signals. (B) Comparisons of chimerism levels in lymphoid organs ofNOD mice transferred with retrovirally transduced BM cells. Compared are the levels of chimerism achieved with non-sorted orsorted pMYs-PIG-transduced and non-sorted pMYs-BIG-transduced BM cells. Transplantation with sorted BM resulted in higheroverall EGFP levels in thymus, spleen and inguinal lymph node than non-sorted recipient mice. Values in (B) represent p values

Table 1. Percentage (mean ± SEM) of EGFP+ cells in various cell populations in thymus and inguinal lymph node (LN) of mice thatreceived non-sorted pMYs-PIG (n = 3), pMYs-BIG (n = 5) or sorted pMYs-PIG (n = 4) transduced bone marrow

Non-sorted Sorted

pMYs-PIG (n = 3) pMYs-BIG (n = 5) pMYs-PIG (n = 4)

Cell marker Thymus Inguinal LN Thymus Inguinal LN Thymus Inguinal LN

CD4 3 ± 0.5 2.7 ± 1.1 4.8 ± 2.9 2.4 ± 1.6 14.2 ± 2.9 12.5 ± 1.8CD8 4.4 ± 1.7 5.9 ± 2.3 5.3 ± 4.2 3.1 ± 1.2 15.6 ± 2.6 12.7 ± 1.4MHC II 11.5 ± 4.5 11.0 ± 2.6 8.9 ± 3.3 11.5 ± 2.3 19.8 ± 2.7 18.3 ± 1.6CD11c + MHC II 2.3 ± 1.3 5.9 ± 1.6 3.6 ± 1.9 7.4 ± 1.9 16.3 ± 2.3 15.8 ± 1.5CD19 38.4 ± 23.3 20.6 ± 8.9 20.9 ± 16.9 14.5 ± 8.9 45.5 ± 9 23.5 ± 1


1286 J. Chan et al.

Figure 3. Indirect immunofluoroscence staining of CD11c+ (dendritic) cells expressing EGFP in thymus of NOD mouse transplantedwith pMYs-PIG-transduced bone marrow. (A) EGFP+ cells within the thymus that have been derived from transduced bone marrowcells. Regions of the thymus can be identified based on the concentration of EGFP+ thymocytes with the cortical (cor) regioncontaining a higher concentration of thymocytes than the medullary (med) region; (B) CD11c+ cells identified by antibody stainingare shown in red and indicated by arrows for reference in (A) and (C); (C) overlay of (A) and (B) showing dual expression of EGFPand CD11c in the same cells (arrowed) indicates that the EGFP+ CD11c+ cells are of donor origin. The band of dendritic cellstaining is found at the junction of the cortex and medulla which is known to be rich in dendritic cells. Blue staining in (C) (Alexa627 nm) represents keratin-positive epithelial cells

other cell populations, an observation that has also beenreported elsewhere [28].

BM-derived DCs are a major cell type associated withinduction of thymic tolerance, by presenting antigensto, and deleting, developing T cells that have high T cellreceptor avidity for self-antigens. Therefore, we examinedfor the presence of EGFP + CD11c + DCs in the thymii ofchimeric mice by indirect immunofluorescence (Figure 3).We found CD11c staining predominantly at the cortico-medullary junction (Figure 3B), a region known to be richin BM-derived DCs [29], and that the majority of these DCsexpressed EGFP (Figure 3C), and, by inference, proinsulinII. In addition to EGFP+ DCs, and as expected, thymocyteswithin the thymus are also EGFP+ (Figure 3A). Thedata indicate that chimerism was achieved within thethymus including the APC compartment comprising theCD11c + MHC II + DCs.

Transplantation of transduced BM cellsexpressing proinsulin II protect againstdestructive insulitis but not sialitis

Our first study using non-sorted BM cells contained fourgroups of animals. Three-to-four-week-old female NODmice received 1 × 106 non-sorted pMYs-PIG-transduced,pMYs-BIG-transduced- or non-transduced BM cells permouse. A fourth age-matched non-treated group did notreceive any BM but only the course of anti-CD4-depletingantibody.

Histological assessment of islets (counted islets, n =105) from pMYs-PIG NOD mice showed a significantreduction in the average insulitis score (0.076 ± 0.026)compared to pMYs-BIG (0.39 ± 0.076; n = 121), BMTalone (0.44 ± 0.086; n = 140) and non-treated (0.83 ±0.098; n = 174) groups (Figure 4A). Insulitis was also

significantly reduced in the BM control groups (pMYs-BIG-and non-transduced BM) compared to the non-treatedage-matched NOD mice (Figure 4A), demonstrating aretarding effect of the BMT procedure on insulitisprogression. There was no significant difference betweenthe level of insulitis in the pMYs-BIG- and non-transduced(normal) BM control groups (p = 0.96) indicating thatthe process of transduction of BM cells with retroviruswas not itself associated with the reduction of insulitis.The majority of islets from the pMYs-PIG -treatedgroup remained free of inflammation (Figure 5A). Inthe minority of islets in which insulitis was evident,insulitis was restricted to a peri-islet ‘benign’ distribution(Figure 5B). In contrast, islets from pMYs-BIG NODmice or NOD mice that received non-transduced BMdeveloped insulitis with a greater incidence and moreextensive infiltration into the islets that included complete‘malignant’ penetrance throughout the islet (Figures 4Aand 5C).

Similarly, NOD mice that received only 3 × 103

sorted EGFP+ pMYs-PIG-transduced BM cells displayeda reduced level and incidence of insulitis compared tothe non-BMT control group (Figure 4B). The averagescore of insulitis in pMYs-PIG-treated mice (0.18 ± 0.05,counted islets, n = 85) was significantly lower than thenon-BMT control group (0.78 ± 0.14, counted islets,n = 90; p < 0.029). In addition, there was no significantdifference in insulitis score of mice who had receivedsorted (n = 85 islets) or non-sorted (n = 105 islets)pMYs-PIG-transduced cells (p = 0.40). Therefore, whilepre-sorting transduced BM cells before transfer increasedchimerism levels, this did not appear to influence thedegree or level of insulitis in NOD recipients. Likewise,there was no significant difference (p = 0.50) in averageinsulitis scores of the non-treated control groups that



Figure 4. Reduction of insulitis development in NOD mice transplanted with bone marrow (BM) cells transduced with retrovirusencoding proinsulin II. All mice were analysed 8 weeks post-transplantation. Pancreatic sections stained by H&E were scoredfor insulitis from 0–4 as described in Materials and Methods with a minimum of 14 islets scored for each animal. The overallincidence and degree of insulitis for each group is displayed. (A) NOD mice received non-sorted BM cells transduced with retrovirusencoding proinsulin II (pMYs-PIG, n = 3), control autoantigen, H/K ATPase β-subunit (pMYs-BIG, n = 5) or non-manipulated BM(normal BMT, n = 6). One non-treated group did not receive any BMT (n = 7). All groups were subjected to a course of anti-CD4T cell-depleting antibody as detailed in Materials and Methods. (B) NOD mice received EGFP-sorted BM cells transduced withretrovirus encoding proinsulin II (pMYs-PIG, n = 4) or mice which did not receive any BMT (n = 4). All mice were subjected to acourse of anti-CD4 T cell-depleting antibody. Statistical significance between the groups is indicated

Figure 5. Examples of insulitis in NOD mice. (A) Insulitis score of 0 representing normal islet morphology; (B) insulitis score of 1representing peri-islet infiltration with inflammation concentrated around the islet; and (C) insulitis score of 4 in which greaterthan >75% of islet is penetrated by inflammatory infiltrate. In contrast to islets, salivary gland inflammation was identical in micetransplanted with bone marrow cells transduced with retrovirus encoding proinsulin II (D) or mice that did not receive any BMT(E)


1288 J. Chan et al.

received non-sorted BM (counted = 174) or sorted BM(counted = 90).

To assess whether the observed effect was antigen-specific, the incidence and degree of sialitis wasdetermined in mice which received sorted pMYs-PIG-transduced BM. The incidence of sialitis was the samein the pMYs-PIG (4/4) and control (4/4) group, withall mice displaying severe mononuclear cell infiltrateswithin the salivary glands (Figures 5D and 5E). Thesefindings indicate that the effect of BMT with proinsulin IItransduced cells in reducing insulitis was antigen specificand did not influence the development of an independentautoimmune response in the NOD mouse which is notassociated with proinsulin II.

Discussion

The ability to isolate, genetically manipulate and transferbone marrow (BM) cells into recipient hosts has openedthe way of exploiting this pathway for treating manydiseases. In this study, we have targeted the expressionof the T1D-associated autoantigen, proinsulin II, toBM-derived cells using a retroviral system. We showthat significant levels of molecular chimerism can beachieved by transplantation of BM cells that havebeen transduced ex vivo with a modified retrovirusincorporating proinsulin II. Chimerism was observed inBM-derived cells including CD4+ and CD8+ T cells, Bcells and CD11c + MHC II + DCs in the thymus. Pre-sorting cells for EGFP expression before transfer increasedthe chimerism level, supporting an earlier observation[30] that enriching for transduced cells before transferincreases chimerism levels, presumably by removingcompetitive non-transduced BM cells. Our findingssuggest that relatively small numbers of transduced cells(3 × 103 in this instance) are sufficient to induce asignificant level of chimerism in recipient mice. However,the increased level of chimerism did not further reducethe incidence of insulitis. While some minimal levelof chimerism is probably required to induce tolerance,it is possible the levels achieved in both studies mayhave exceeded such minimal level such that no obviousdifference was observed. Studies in transplantationtolerance suggested that chimerism levels of 10–15% aresufficient for tolerance [31] and, more recently, Bonillaand colleagues demonstrated in an allogeneic studythat levels of chimerism less than 1% (microchimerism)were sufficient to maintain tolerance in the cytotoxicT lymphocyte compartment [32]. With experimentalautoimmune diseases, Steptoe and colleagues reportedthat 5% chimerism following BMT from transgenic micewas sufficient to prevent insulitis and diabetes in theNOD mouse [22]. Taken together with our findings, theseobservations support the concept that only a low levelof chimerism may be required to induce autoantigen-specific tolerance and that sufficient chimerism may beachieved with a relatively small number of genetically

modified HSCs. The ability to sort for transduced cellsand thus eliminate non-transduced competitive cells mayhave relevance for clinical translation in which sufficientchimerism levels may be achieved through the use of lesstoxic pre-conditioning regimes.

The readout in our study was the level of insulitisfollowing transfer of HSCs which allowed us to evaluatethe different degrees of cellular infiltration in the pancreasduring the early phase of the disease. Mice receivingsorted or non-sorted HSCs transduced with retrovirusexpressing proinsulin II had significantly lower levels ofinsulitis compared to control mice that received non-manipulated HSCs or HSCs expressing an irrelevantautoantigen. We noted a significant difference in insulitisscore in mice that received non-manipulated BM cellscompared to non-treated mice, indicating that BMT itselfhas a retarding effect, as has been shown in otherdisease models [33]. However, within the BMT groups,insulitis was significantly reduced in mice receivingHSCs transduced with proinsulin II. There was nosignificant difference in insulitis between the H/K β-subunit-transduced and non-transduced BM groups. Notonly was the overall incidence of insulitis significantlylower in proinsulin II expressing mice, the severityof insulitis was also markedly reduced in that theinsulitis was mainly limited to a peri-islet, non-invasivedistribution with preservation of islets. These observationsindicate that expression of proinsulin II arrested mice ina state of ‘benign’ non-destructive autoimmunity. Theobservation that sialitis was found equally in proinsulin IIand non-BMT control mice supports the suggestion thatthe procedure is antigen specific and not associated withsome general perturbation of the immune system. Thestrategy of ex vivo genetic manipulation of HSCs to induceautoantigen-specific tolerance is relatively new [25] withfew experimental studies thus far published. A recentstudy by Xu and colleagues, using a strategy similar to thatdescribed in our study, has shown that the developmentof experimental autoimmune encephalomyelitis can beprevented in mice transferred with HSCs transducedwith retrovirus encoding phospholipid protein (PLP)[34]. As more studies which utilize autoantigens in thismanner are conducted, it will reinforce the concept thatautoantigen-specific tolerance can be induced throughgenetic manipulation of HSCs and that this can preventend-organ damage.

The mechanism(s) associated with immune toleranceinduced by transfer of retroviral transduced HSCs is notfully understood. Dendritic cells have been attributed akey role in deletional tolerance of self-reactive T cells [3].Our observation of chimerism in this population withinthe thymus supports the notion that this mechanismmay be responsible for tolerance induction, a suggestionsupported by the finding of Xu and colleagues [34].Previous studies in which neo- or alloantigen has beenexpressed in HSCs by retroviral transduction have alsoshown deletion of antigen-specific T cells [35,36]. Whilethis assumes direct expression of antigens by thymic DCs,passive uptake of antigen by these cells within the thymus



is another possibility. More than 90% of T cells diewithin the thymus during development and introductionof peripheral antigens by intrathymic delivery inducestolerance [37,38]. Therefore, in the system describedin this study, it is possible the delivery of antigen tothe thymus by developing T cells of donor origin couldpromote its release and uptake by thymic APCs withsubsequent tolerance induction. Other possible tolerancemechanisms include the generation of regulatory T cells.However, this seems less likely as transgenic expressionof proinsulin or the β-subunit of gastric H/K ATPase, andretroviral expression of PLP, does not appear to generatea regulatory population (unpublished data) [19,21,34].Apart from BM-derived thymic DCs, other BM-derivedcells that can express autoantigen following retroviraltransduction of HSCs may contribute to tolerance. Thesepopulations include peripheral APCs such as resting DCsor B cells [39], and T cells [37,38,40]. Mature peripheralT cells have been shown to re-enter the thymus to mediatenegative selection [41], and the potential of peripheralDCs and B cells as tolerogenic APCs has also been explored[42–45]. Indeed, in our study, high levels of chimerismwere detected in these cell populations in peripheral andcentral lymphoid organs and a potential role of these intolerance mechanisms cannot be excluded and is worthyof further examination.

In summary, our study extends the literature demon-strating that ectopic expression of autoantigens can beused to tolerize the immune system to self-antigens andarrest end-organ damage. This is the first study to demon-strate that ex vivo transduction of normal HSCs to expressproinsulin II reduces the incidence and the severity ofinsulitis in a disease-prone mouse model. This strategymay be applicable to other autoimmune diseases in whichcausative autoantigens are known.

Acknowledgements

The authors thank Ian MacKay and Robyn Slattery forcritically reading the manuscript, and Josephine Forbes forthe (pro)insulin radioimmunoassay. This investigation wassupported by the National Health and Medical Research Councilof Australia and the Australian Stem Cell Centre.

References

1. Davidson A, Diamond B. Autoimmune diseases. N Engl J Med2001; 345: 340–350.

2. Kyewski B, Derbinski J. Self-representation in the thymus: anextended view. Nat Rev Immunol 2004; 4: 688–698.

3. Brocker T, Riedinger M, Karjalainen K. Targeted expression ofmajor histocompatibility complex (MHC) class II moleculesdemonstrates that dendritic cells can induce negative but notpositive selection of thymocytes in vivo. J Exp Med 1997; 185:541–550.

4. Goodnow CC, Sprent J, de St Groth BF, et al. Cellular andgenetic mechanisms of self tolerance and autoimmunity. Nature2005; 435: 590–597.

5. Kubach J, Becker C, Schmitt E, et al. Dendritic cells: sentinels ofimmunity and tolerance. Int J Hematol 2005; 81: 197–203.

6. Sprent J, Webb SR. Intrathymic and extrathymic clonal deletionof T cells. Curr Opin Immunol 1995; 7: 196–205.

7. Ramsey C, Winqvist O, Puhakka L, et al. Aire deficientmice develop multiple features of APECED phenotype andshow altered immune response. Hum Mol Genet 2002; 11:397–409.

8. Betterle C, Zanchetta R. Update on autoimmune polyendocrinesyndromes (APS). Acta Biomed Ateneo Parmense 2003; 74:9–33.

9. Andre I, Gonzalez A, Wang B, et al. Checkpoints in theprogression of autoimmune disease: lessons from diabetesmodels. Proc Natl Acad Sci U S A 1996; 93: 2260–2263.

10. Nepom G, Quinn A, Sercarz E, et al. How important is GAD inthe etiology of spontaneous disease in human and murine type1 diabetes? J Autoimmun 2003; 20: 193–194.

11. Dubois-Lafforgue D, Mogenet L, Thebault K, et al. Proinsulin 2knockout NOD mice: a model for genetic variation of insulingene expression in type 1 diabetes. Diabetes 2002; 51(Suppl 3):S489–493.

12. Harrison LC, Chu SX, DeAizpurua HJ, et al. Islet-reactive T cellsare a marker of preclinical insulin-dependent diabetes. J ClinInvest 1992; 89: 1161–1165.

13. Rudy G, Stone N, Harrison LC, et al. Similar peptides fromtwo beta cell autoantigens, proinsulin and glutamic aciddecarboxylase, stimulate T cells of individuals at risk for insulin-dependent diabetes. Mol Med 1995; 1: 625–633.

14. Daniel D, Gill RG, Schloot N, et al. Epitope specificity, cytokineproduction profile and diabetogenic activity of insulin-specific Tcell clones isolated from NOD mice. Eur J Immunol 1995; 25:1056–1062.

15. Wegmann DR, Norbury-Glaser M, Daniel D. Insulin-specific Tcells are a predominant component of islet infiltrates in pre-diabetic NOD mice. Eur J Immunol 1994; 24: 1853–1857.

16. Daniel D, Wegmann DR. Protection of nonobese diabetic micefrom diabetes by intranasal or subcutaneous administration ofinsulin peptide B-(9–23). Proc Natl Acad Sci U S A 1996; 93:956–960.

17. Nakayama M, Abiru N, Moriyama H, et al. Prime role for aninsulin epitope in the development of type 1 diabetes in NODmice. Nature 2005; 435: 220–223.

18. Kent SC, Chen Y, Bregoli L, et al. Expanded T cells frompancreatic lymph nodes of type 1 diabetic subjects recognizean insulin epitope. Nature 2005; 435: 224–228.

19. Alderuccio F, Toh BH, Tan SS, et al. An autoimmune diseasewith multiple molecular targets abrogated by the transgenicexpression of a single autoantigen in the thymus. J Exp Med1993; 178: 419–426.

20. Murphy K, Biondo M, Toh BH, et al. Tolerance established inautoimmune disease by mating or bone marrow transplantationthat target autoantigen to thymus. Int Immunol 2003; 15:269–277.

21. French MB, Allison J, Cram DS, et al. Transgenic expression ofmouse proinsulin II prevents diabetes in nonobese diabetic mice.Diabetes 1997; 46: 34–39.

22. Steptoe RJ, Ritchie JM, Harrison LC. Transfer of hematopoieticstem cells encoding autoantigen prevents autoimmune diabetes.J Clin Invest 2003; 111: 1357–1363.

23. Shizuru JA, Negrin RS, Weissman IL. Hematopoietic stem andprogenitor cells: clinical and preclinical regeneration of thehematolymphoid system. Annu Rev Med 2005; 56: 509–538.

24. Tyndall A, Saccardi R. Haematopoietic stem cell transplantationin the treatment of severe autoimmune disease: results fromphase I/II studies, prospective randomized trials and futuredirections. Clin Exp Immunol 2005; 141: 1–9.

25. Alderuccio F, Murphy K, Toh BH. Stem cells engineered toexpress self-antigen to treat autoimmunity. Trends Immunol2003; 24: 176–180.

26. Kitamura T, Koshino Y, Shibata F, et al. Retrovirus-mediatedgene transfer and expression cloning: powerful tools infunctional genomics. Exp Hematol 2003; 31: 1007–1014.

27. Pear WS, Nolan GP, Scott ML, et al. Production of high-titerhelper-free retroviruses by transient transfection. Proc Natl AcadSci U S A 1993; 90: 8392–8396.

28. Tian C, Bagley J, Cretin N, et al. Prevention of type 1 diabetesby gene therapy. J Clin Invest 2004; 114: 969–978.

29. Janeway C Jr, Travers P, Mark W, et al. The development andsurvival of lymphocytes. In Immunobiology, the Immune Systemin Health and Disease (6th edn), Lawrence E (ed). GarlandScience Publishing: 2005; 241–316.

30. Bagley J, Aboody-Guterman K, Breakefield X, et al. Long-termexpression of the gene encoding green fluorescent protein


1290 J. Chan et al.

in murine hematopoietic cells using retroviral gene transfer.Transplantation 1998; 65: 1233–1240.

31. Wekerle T, Sykes M. Mixed chimerism and transplantationtolerance. Annu Rev Med 2001; 52: 353–370.

32. Bonilla WV, Geuking MB, Aichele P, et al. Microchimerismmaintains deletion of the donor cell-specific CD8+ T cellrepertoire. J Clin Invest 2006; 116: 156–162.

33. Burt RK, Slavin S, Burns WH, et al. Induction of tolerancein autoimmune diseases by hematopoietic stem celltransplantation: getting closer to a cure? Blood 2002; 99:768–784.

34. Xu B, Haviernik P, Wolfraim LA, et al. Bone marrowtransplantation combined with gene therapy to induce antigen-specific tolerance and ameliorate EAE. Mol Ther 2006; 13:42–48.

35. Ally BA, Hawley TS, McKall-Faienza KJ, et al. Prevention ofautoimmune disease by retroviral-mediated gene therapy. JImmunol 1995; 155: 5404–5408.

36. Kang ES, Iacomini J. Induction of central deletional Tcell tolerance by gene therapy. J Immunol 2002; 169:1930–1935.

37. Wilson DB, Wilson DH, Schroder K. Acquired thymic toleranceand experimental allergic encephalomyelitis in the rat. I.Parameters and analysis of possible mechanisms. Eur J Immunol1998; 28: 2770–2779.

38. Posselt AM, Barker CF, Friedman AL, et al. Prevention ofautoimmune diabetes in the BB rat by intrathymic islettransplantation at birth. Science 1992; 256: 1321–1324.

39. Xu B, Scott DW. A novel retroviral gene therapy approach toinhibit specific antibody production and suppress experimentalautoimmune encephalomyelitis induced by MOG and MBP. ClinImmunol 2004; 111: 47–52.

40. Tian C, Bagley J, Iacomini J. Expression of antigen on maturelymphocytes is required to induce T cell tolerance by genetherapy. J Immunol 2002; 169: 3771–3776.

41. Tian C, Bagley J, Forman D, et al. Induction of central toleranceby mature T cells. J Immunol 2004; 173: 7217–7222.

42. Lutz MB, Schuler G. Immature, semi-mature and fully maturedendritic cells: which signals induce tolerance or immunity?Trends Immunol 2002; 23: 445–449.

43. Finkelman FD, Lees A, Birnbaum R, et al. Dendritic cells canpresent antigen in vivo in a tolerogenic or immunogenic fashion.J Immunol 1996; 157: 1406–1414.

44. Agarwal RK, Kang Y, Zambidis E, et al. Retroviral gene therapywith an immunoglobulin-antigen fusion construct protects fromexperimental autoimmune uveitis. J Clin Invest 2000; 106:245–252.

45. Melo ME, Qian J, El-Amine M, et al. Gene transfer of Ig-fusionproteins into B cells prevents and treats autoimmune diseases. JImmunol 2002; 168: 4788–4795.


Documents

Transplantation of bone marrow genetically engineered to express proinsulin II protects against autoimmune insulitis in NOD mice