The immunotherapeutic potential of dendritic cells in type 1 diabetes

Abstract

Type 1 diabetes is an autoimmune disease characterized by destruction of the pancreatic islet beta cells that is mediated primarily by T cells specific for beta cell antigens. Insulin administration prolongs the life of affected individuals, but often fails to prevent the serious complications that decrease quality of life and result in significant morbidity and mortality. Thus, new strategies for the prevention and treatment of this disease are warranted. Given the important role of dendritic cells (DCs) in the establishment of peripheral T cell tolerance, DC-based strategies are a rational and exciting avenue of exploration. DCs employ a diverse arsenal to maintain tolerance, including the induction of T cell deletion or anergy and the generation and expansion of regulatory T cell populations. Here we review DC-based immunotherapeutic approaches to type 1 diabetes, most of which have been employed in non-obese diabetic (NOD) mice or other murine models of the disease. These strategies include administration of in vitro-generated DCs, deliberate exposure of DCs to antigens before transfer and the targeting of antigens to DCs in vivo. Although remarkable results have often been obtained in these model systems, the challenge now is to translate DC-based immunotherapeutic strategies to humans, while at the same time minimizing the potential for global immunosuppression or exacerbation of autoimmune responses. In this review, we have devoted considerable attention to antigen-specific DC-based approaches, as results from murine models suggest that they have the potential to result in regulatory T cell populations capable of both preventing and reversing type 1 diabetes.

Type 1 diabetes

Type 1 diabetes is an organ-specific autoimmune disease characterized by progressive loss of the insulin-producing beta cells that reside within the pancreatic islets [1]. This loss is primarily T cell-mediated and orchestrated by CD4⁺ and CD8⁺ T cells specific for beta cell antigens [2]. Present in both type 1 diabetes patients and in non-obese diabetic (NOD) mice, a well-studied model of the disease, these T cells employ a variety of mechanisms to bring about beta cell elimination [3]. These include Fas/FasL interactions and perforin- and cytokine-mediated cell killing. Although systemic pharmacological immunosuppression can halt the autoimmune attack [4], its side effects render it unacceptable for routine use in type 1 diabetes patients. Insulin injections prolong life but are often unable to prevent the serious diabetic complications that are associated with significant morbidity and mortality. Thus, there is an ongoing worldwide effort to develop new strategies for the prevention and treatment of this disease.

Nearly two decades ago, Clare-Salzler and colleagues reported that dendritic cells (DCs) isolated from the pancreatic lymph nodes of NOD mice could prevent diabetes development when transferred adoptively to young recipients [5]. These findings spurred efforts to develop DC-based interventions for type 1 diabetes. The overall favourable safety profile of DC-based therapies revealed by cancer immunotherapy trials has provided further inspiration for such work [6–15]. Here we will discuss the progress that has been made in the area of DC-based therapeutics for type 1 diabetes, with a special emphasis on antigen-specific approaches. We will limit our discussion to ‘conventional’ DCs, as the therapeutic promise of plasmacytoid DCs in type 1 diabetes has been reviewed recently [16].

An introduction to DCs

The identification of DCs was reported by Steinman and Cohn in 1973 [17], a discovery that was driven by a desire to ‘understand immunogenicity’[18]. One of the initial demonstrations of the immunogenic role of DCs was the finding that isolated murine lymphoid organ DCs were potent stimulators of the mixed leucocyte reaction [19]. However, two decades later, when an antigen was delivered specifically to a subset of murine DCs in vivo (i.e. those expressing the endocytic receptor DEC-205), the predicted outcome of a robust immune response did not occur [20]. Antigen-specific tolerance was observed instead, as cognate T cells were largely deleted or rendered unresponsive. It is now understood that in the steady state (i.e. in the absence of infection), DCs are largely immature and present antigens to T cells in a tolerogenic manner, an activity that is important for the establishment of peripheral tolerance [21]. Such DCs are characterized by low expression of CD40 and the T cell co-stimulatory molecules CD80 and CD86. In contrast, in the case of host exposure to a pathogen, DCs undergo a maturation process, e.g. in response to microbial-derived products, that leads to increased antigen presentation and expression of T cell co-stimulatory molecules and T cell responses of a type appropriate to combat the offending pathogen [22]. It should be noted that CD4⁺CD25⁺ regulatory T cells (T_regs) are best expanded, at least in vitro, when CD86^high DCs are used as antigen-presenting cells [23,24]. Thus, it is important to keep in mind that a certain level of DC maturity may be important for the generation of T_regs capable of inhibiting autoimmune disease [25].

The development of conventional lymphoid organ DCs in mice has been clarified recently [26]. The macrophage and DC precursor gives rise to the common DC precursor (the source of both conventional and plasmacytoid DCs). The next developmental stage for the conventional lymphoid organ DC is the pre-DC. The pre-DCs expand in the bone marrow and differentiate to conventional DCs within the spleen and lymph nodes, where they proliferate in response to Flt3L [27]. A number of DC subsets have been described phenotypically in both mice and humans [28]. Some of these are known to be functionally specialized [29]. For example, in mice, the DC subset expressing CD8 and DEC-205 is specialized for capture of dying cells [30] and cross-presentation of antigens on class I major histocompatibility complex (MHC) molecules [31–33], while CD8^-DCIR2⁺ DCs are proficient at presentation of peptides on class II MHC [32].

Strategies adopted by DCs to induce peripheral T cell tolerance

In addition to their well-established role in central tolerance [34], DCs employ a variety of diverse strategies and pathways to maintain T cell tolerance in the periphery (Fig. 1). Apart from induction of deletional tolerance of peripheral T cells [20,35], DCs in the steady state can also render them anergized [20] as a result of antigen recognition without sufficient co-stimulation [36]. T cell co-inhibitory molecules that transduce negative signals, such as cytotoxic T lymphocyte antigen-4 (CTLA-4) [37] or programmed death-1 (PD-1) [38,39], also participate in these processes. For example, steady-state DCs utilize both the PD-1 and CTLA-4 pathways to induce peripheral tolerance of CD8⁺ T cells [40]. In addition to induction of deletion or anergy, DCs can induce increased expression of CD5 on activated T cells that leads to hyporesponsiveness, at least in the setting of the induced autoimmune disease, experimental acute encephalomyelitis [41]. Expression of Fas on antigen-presenting cells is also important for the maintenance of peripheral tolerance and the avoidance of autoimmunity [42], while the production of indoleamine 2,3-dioxygenase (IDO) by DCs is involved in peripheral tolerance in certain specialized settings [43,44]. Finally, DCs are involved in the in vivo expansion of thymic-derived natural CD4⁺CD25⁺ T_regs[45] as well as the induction of adaptive forkhead box P3 (FoxP3⁺) T_regs[45–48] and CD8⁺ T_regs[49], and interleukin (IL)-7 produced by immature DCs appears to function as a CD4⁺CD25⁺ T_reg survival factor [50].

Fig. 1.

Different strategies of induction of peripheral tolerance by dendritic cells (DCs). Steady-state DCs can induce deletion or anergy of cognate autoreactive T cells either through lack of co-stimulation or by recruitment of co-inhibitory molecules. Another efficient method of tolerance induction is the generation and expansion of antigen-specific regulatory T cells. Apart from deletion of DCs by Fas expression, DCs can induce tolerance directly in the periphery by production of indoleamine 2,3-dioxygenase.

The role of DCs during the natural history of type 1 diabetes

Multiple lines of investigation indicate that priming of pathogenic beta cell-specific T cells occurs in the pancreatic lymph nodes. For example, adoptive transfer of 5,6-carboxy-succinimidyl-fluorescein-ester (CFSE)-labelled transgenic CD4⁺ BDC2.5 T cells (specific for a beta cell antigen) revealed proliferation of the transferred cells selectively in the pancreatic lymph nodes and before the onset of insulitis [51]. Removal of the pancreatic lymph nodes of 3-week-old NOD mice prevented diabetes development [52], again suggesting that autoreactive T cell priming occurs at this site. While DCs are responsible for this presentation of beta cell antigens [53–55], it is important to realize that the outcome of this can be T cell deletion or regulation instead of pathogenic T cell priming [53,54], even in the diabetes-prone NOD mouse [56]. Serreze and colleagues found that a significant proportion of transferred islet-reactive CD8⁺ AI4 T cells underwent apoptosis in the pancreatic lymph nodes of NOD mice, but not in other sites such as the mesenteric lymph nodes [56]. In addition, pancreatic lymph node-residing AI4 T cells were less responsive to antigen when compared to cells isolated from the mesenteric lymph nodes [56]. These observations are consistent with the finding that transfer of pancreatic lymph node DCs to young (4-week-old) NOD mice could prevent diabetes development [5]. Such results serve as the foundation for current efforts to explore the immunotherapeutic potential of DCs in type 1 diabetes.

DC-based therapeutic strategies that do not incorporate islet antigens

Morel's group showed that DCs generated from the bone marrow of NOD mice by culture in granulocyte–macrophage colony-stimulating factor (GM-CSF), IL-4 and fetal bovine serum (FBS) could prevent diabetes in some recipients when administered as 3-weekly intravenous injections to young (5-week-old) NOD mice [57]. These bone marrow-derived DCs (BMDCs) expressed class II MHC, CD80, CD86 and CD40 in vitro, although CD40 expression was subsequently diminished upon in vivo administration. Pulsing of the DCs with a mixture of defined beta cell peptides [heat shock protein 60 (HSP60_437–460), glutamic acid decarboxylase 65 (GAD65_509–528) and GAD65_524–543] before transfer did not augment their ability to prevent disease. Mice receiving DCs (pulsed with beta cell peptides or not) exhibited an increased immunoglobulin G₁ (IgG₁) response to GAD65_509–528. As IL-4 facilitates class-switching to this isotype, the investigators speculated, and showed later [58], that DC administration leads to the stimulation of regulatory T helper type 2 (Th2) T cell responses, as determined by cytokine production in response to anti-T cell receptor (anti-TCR) stimulation.

Subsequent to these studies, von Herrath demonstrated that murine BMDCs generated in FBS caused systemic immune deviation in recipients due to a Th2 cell response to FBS-derived proteins [59]. This resulted in impaired clearance of a lymphocytic choriomeningitis virus (LCMV) infection, which normally relies on a Th1 response and interferon (IFN)-γ-producing cytotoxic CD8⁺ T cells. This important study urged investigators to avoid DC exposure to FBS in their preclinical studies, in order to more effectively mimic future clinical trials where FBS would not be used. It also demonstrated the need to monitor possible systemic immunosuppressive effects of DC administration that might have undesirable clinical consequences, e.g. impaired viral clearance.

Genetically modified DCs have also been employed in preclinical models of type 1 diabetes. BMDCs transduced with a lentiviral vector encoding IL-4 were able to prevent disease in old (12-week-old) NOD recipients, i.e. well after the onset of insulitis, whereas unmodified DCs could not [60]. BMDCs engineered to express galectin-1 by transduction with a recombinant adenovirus were capable of delaying the onset of diabetes induced in immunodeficient NOD recipients by transfer of splenocytes from diabetic NOD females [61]. This is consistent with the recent finding that stimuli that induce tolerogenic DCs, such as IL-10 and 1,25-dihydroxyvitamin D₃, also increase their expression of galectin-1 [62].

In addition to viral vectors, treatment with anti-sense oligonucleotides has been used to engineer DCs having a tolerogenic phenotype. Giannoukakis and Trucco used anti-sense oligonucleotides targeting the CD40, CD80 and CD86 messages to treat BMDCs from NOD mice in order to engineer phenotypically immature DCs [63]. When these DCs were administered intraperitoneally to 5–8-week-old NOD mice, a single injection was able to prolong the time to diabetes onset. The therapeutic effect correlated with an increased percentage of splenic CD4⁺CD25⁺ (presumably regulatory) T cells. Systemic immunosuppression was not observed, as splenocytes from DC-treated mice were able to respond to alloantigens in vitro. These investigators showed subsequently that four weekly injections of anti-sense oligonucleotide-treated DCs, beginning at 8 weeks of age, resulted in prevention of disease in all recipients [50].

BMDCs from NOD mice have also been manipulated by treatment with decoy double-stranded oligonucleotides containing nuclear factor-kappa B (NF-κB) binding sites [64]. The treated DCs exhibited reduced NF-κB activity and suppression of co-stimulatory molecule expression and IL-12 production. When administered as a single intravenous injection to NOD mice at 6–7 weeks of age, NF-κB-deficient DCs had a dramatic disease-preventive effect, while untreated DCs or those treated with control oligonucleotides were only modestly beneficial.

When contemplating therapeutic administration of DCs, it is important to consider the in vivo trafficking patterns of the administered cells. Creusot and Fathman showed that BMDCs administered intraperitoneally to mice accumulated preferentially in the pancreatic lymph nodes as opposed to other lymph nodes or the spleen [65]. This was the case even in non-diabetes-prone mouse strains. This could explain why intraperitoneal administration of anti-sense oligonucleotide-treated DCs delayed diabetes onset but did not result in systemic immunosuppression [63]. Intravenously administered BMDCs accumulated predominantly in the spleen and the pancreatic and tracheal lymph nodes [60,65]. This finding is consistent with the speculation [57] that intravenously administered DCs can acquire islet antigens in vivo (a process that would take place in the pancreatic lymph nodes) and, thus, can modulate effector and regulatory T cell responses to diabetes-relevant antigens even without deliberate prior antigen treatment.

Exposure of therapeutic DCs to antigen ex vivo

The original observation that DCs from the pancreatic lymph nodes could prevent diabetes when transferred to NOD mice, while those from other sites could not, suggested the potential importance of the incorporation of beta cell antigens into DC-based therapeutics for this disease [5]. As reviewed recently [66], a variety of immunosuppressive and anti-inflammatory compounds, e.g. vitamin D₃ and mycophenolate mofetil, can endow DCs with a tolerogenic functional phenotype. Cytokines such as IL-10 can behave similarly [67]. This suggests a therapeutic strategy for type 1 diabetes in which tolerogenic DCs would be generated in vitro and then exposed to beta cell antigens prior to administration. Such an approach was employed recently by the von Herrath group [59], who utilized the rat insulin promoter (RIP)-LCMV model of type 1 diabetes in which disease is induced upon LCMV infection. BMDCs were generated in the presence of GM-CSF, IL-10 and normal mouse serum, and then pulsed with a viral peptide recognized by CD8⁺ T cells. When the pulsed DCs were administered intraperitoneally to mice 10 and 3 days prior to LCMV infection, only 45% of the animals developed diabetes, whereas 80% of those treated with unpulsed DCs became diabetic. A reduced expansion of viral-specific T cells in response to viral infection was also observed in mice treated with peptide-pulsed DCs. This study supports the idea that ex vivo-generated tolerogenic DCs, when exposed to disease-relevant antigens, can deliver therapeutic benefit in type 1 diabetes.

Targeting of antigens to DCs in vivo

In a recent thoughtful review of DC-based immunotherapeutic strategies for human diseases, the disadvantages of ex vivo antigen loading of DCs were discussed [68]. These include a requirement for leukapheresis, the inability to manipulate DCs within their natural milieu and a requirement for a tailor-made ‘product’ for each patient, resulting in labour-intensive procedures and high costs. It is for reasons such as these that we [69] and others [70] are exploring the utility of in vivo delivery of beta cell antigens to DCs in the prevention and treatment of type 1 diabetes.

DEC-205: an endocytic receptor involved in antigen processing and presentation

DCs employ a variety of molecules, such as the Fc receptors, the macrophage mannose receptor (MMR) and DEC-205 [71], to execute receptor-mediated endocytosis of antigens. Of these, DEC-205 (Ly75/CD205) has the special ability to uptake and subsequently present antigen via both class I [35] and class II MHC pathways [72]. DEC-205 is a type 1 transmembrane protein homologous to MMR and phospholipase A2 [71]. Being a member of this family, this 205 kDa protein has a characteristic short cytoplasmic domain that mediates adsorptive endocytosis. It is made of a single polypeptide chain and contains 10 lectin domains, which is a deviation from the eight lectin domains of the MMR family. Named from the first observations of its abundant expression on DCs and thymic epithelial cells in mice using the rat monoclonal antibody non-lymphoid dendritic cells (NLDC)-145 [73,74], DEC-205 has a more diverse distribution. B cells from various sources such as spleen, lymph node and peritoneal exudates express DEC-205, but at a much lower level than on DCs [75]. Immunohistochemical staining showed expression of DEC-205 on the follicular B cells, bone marrow stroma and pulmonary airway epithelium [76].

Although found predominantly on DCs, DEC-205 is not expressed ubiquitously on all DC subsets. In the mouse thymus, all DC show DEC-205 expression, the majority of which are CD8⁺[77]. In contrast, murine spleen shows three subsets of DC: CD4⁺CD8^-DEC205^-CD11b⁺, CD4^-CD8^-DEC205^-CD11b⁺ and CD4^-CD8⁺DEC205⁺CD11b^-[77]. Two additional populations can be traced in the lymph nodes, which show lower expression of CD8 but high to moderate expression of DEC-205 [78]. Moreover, non-lymphoid DCs such as the Langerhans cells of the skin and also BMDCs generated in the presence of GM-CSF show high expression of DEC-205 [75,79]. While humans do not have a DC subset that is CD8⁺, most DCs in the T cell areas of human spleen and lymph nodes co-express CD11c and DEC-205 [80,81].

DC-induced tolerance upon delivery of antigen via DEC-205

In a pioneering study, Hawiger and colleagues fused an immunogenic peptide from hen egg lysozyme (HEL) to the carboxyl terminus of the heavy chain of NLDC-145 [20]. They injected mice with the hybrid antibody and found that the anti-DEC-205/HEL could deliver antigen to DCs leading to CD4⁺ T cell activation and proliferation. However, further investigation showed that this treatment led ultimately to the deletion of many of the antigen-specific T cells and that the remaining T cells were unresponsive and could not mount an immune response to subsequent challenge with HEL administered with complete Freund's adjuvant (CFA), showing the induction of HEL-specific tolerance. However, when the same treatment was performed in conjunction with an agonistic anti-CD40 antibody, the outcome was prolonged T cell activation and immunity. Thus, it could be inferred that DCs in the steady state, i.e. in the absence of additional stimuli, act as inducers of antigen-specific peripheral tolerance.

Subsequently, ovalbumin (OVA) was coupled chemically to anti-DEC-205 and was found to permit DCs to present a cognate peptide to OVA-specific CD8⁺ T cells [35]. The antibody-mediated delivery was much more efficient than administration of soluble OVA alone. In agreement with the earlier study that utilized anti-DEC-205/HEL and CD4⁺ T cells [20], when anti-DEC-205/OVA was targeted to DCs in the steady state in vivo there was an initial burst of proliferation of the OVA-specific CD8⁺ T cells which was followed by their deletion. The treated mice also showed tolerance to subsequent rechallenge with OVA along with CFA, which was, however, nullified when anti-CD40 was co-administered with the anti-DEC-205/OVA reagent. Therefore, the DEC-205 receptor can deliver antigen to DCs for presentation to both CD4⁺ and CD8⁺ T cells, and when that is performed in the steady state it leads to deletional tolerance or anergy of the antigen-specific T cells.

Targeting steady-state immature DCs with antigen-linked anti-DEC-205 antibody, apart from inducing anergy and deletion of cognate T cells [20,35], can also lead to the induction and/or expansion of T_regs[47,82]. Anti-DEC-205/OVA drove short-lived proliferation of OVA-specific CD4⁺ T cells in vivo and led to the induction of CD25⁺/CTLA-4⁺ T cells with regulatory properties which could suppress proliferation and IL-2 production of conventional CD4⁺ T cells in a dose-dependent manner [82]. This phenomenon was corroborated further in CD4⁺ and CD8⁺ T cell-driven hypersensitivity models, where suppression of immune responses could be achieved in vivo by the induction of CD4⁺CD25⁺ T_regs by antigen-linked anti-DEC-205. To investigate further whether DCs are able to induce T_regs from truly naive FoxP3^- CD4⁺ T cells, peptide ligands were targeted to DCs through DEC-205 and FoxP3 expression was analysed at the single-cell level [47]. In this study, which used T cells from Rag2^−/− TCR transgenic mice to exclude pre-existing FoxP3⁺ cells, it was shown that the converted T_regs expressed FoxP3 just as do natural T_regs. It was also demonstrated that minute antigen doses with suboptimal DC activation were necessary for T_reg induction, which was enhanced further by the addition of transforming growth factor (TGF)-β or in the absence of IL-2. Importantly, these FoxP3⁺ T_regs could be expanded by immunogenic presentation of antigen and also retained their surface phenotype and suppressor activity.

Recently, Yamazaki and Steinman reported that CD8⁺DEC205⁺ splenic DCs are particularly well equipped to induce FoxP3⁺ T_regs from FoxP3^- precursors [45]. This occurs in the presence of low doses of antigen and requires TGF-β expressed by the DEC-205⁺ DCs themselves. This may explain partially why, in some cases, DC targeting by antigen-linked anti-DEC-205 antibody led to the conversion of conventional CD4⁺ T cells to CD25⁺CD4⁺ T_regs[47,82].

DEC-205-mediated antigen delivery: application to type 1 diabetes

The therapeutic potential of DEC-205-mediated antigen delivery has begun to be explored in mouse models of type 1 diabetes [69,70]. The first such study utilized a CD4⁺ T cell-driven model in which mice express haemagglutinin (HA) under the control of the rat insulin promoter (INS) and an I-E^d-restricted TCR specific for HA_110–120. These mice have a diabetes incidence of 40%. When treated with HA peptide-linked anti-DEC-205 repeatedly from birth until 12–16 weeks of age, diabetes was prevented in most animals. While extensive deletion of the HA-specific T cells was not observed, they exhibited increased expression of FoxP3, CTLA-4, TGF-β and IL-10, which the authors presumed was responsible for the disease-protective effect of anti-DEC-205/HA.

In addition to CD4⁺ T cells, the involvement of cytotoxic CD8⁺ T cells in the pathogenesis of type 1 diabetes is well established in NOD mice [83]. Furthermore, deletion of a single CD8⁺ T cell specificity by soluble peptide therapy has shown some therapeutic benefit in this model [84,85]. Therefore, beta cell antigenic epitopes targeted by CD8⁺ T cells are potential candidates for antigen-based tolerogenic strategies. Keeping this in mind, in our laboratory a superagonist mimotope peptide recognized by the AI4 CD8⁺ T cell clone was delivered to DCs in NOD mice using peptide-linked anti-DEC-205 [69]. Transferred antigen-specific T cells were found to undergo initial proliferation, only to be deleted later. When the treated mice were rechallenged with the mimotope, along with CFA, no immune response could be induced, indicative of antigen-specific tolerance. These findings demonstrated that targeting of DCs with a beta cell antigen, even in the context of the ongoing autoimmune activity present in NOD mice, could lead to deletion of autoreactive CD8⁺ T cells and subsequent tolerance induction.

The wide variety of antigens and T cell epitopes targeted in type 1 diabetes in both NOD mice and humans [2] suggests that simple deletion of a single antigenic specificity, or even several, may be unable to provide durable clinical benefit. However, we believe that targeting of antigens to DEC-205⁺ DCs holds promise due to its additional potential to facilitate the expansion and/or induction of T_regs[45,47,70,82]. The importance of FoxP3⁺ T_regs in type 1 diabetes is demonstrated by the fact that children with a congenital defect in FoxP3 expression rapidly develop a variety of autoimmune diseases, including type 1 diabetes [86,87]. CD4⁺CD25⁺ T_regs have also been shown to prevent or reverse diabetes in NOD mice [23,88–90]. Importantly, DCs from NOD mice were found to be capable of expanding CD4⁺CD25⁺ BDC2.5 T cells in vitro[23]. These islet-specific T_regs were a potent inhibitor of diabetes development in NOD mice, even though multiple antigenic specificities participate in beta cell demise in this model [2]. These DC-expanded islet-specific T_regs, when administered to NOD mice, could also block diabetes long after the initiation of insulitis and caused long-lasting reversal of hyperglycaemia even after development of overt disease [90].

Antigen choice

When developing DEC-205-mediated therapeutic strategies for type 1 diabetes, the choice of antigen is not a straightforward one. As mentioned, multiple antigens are targeted by T cells in both NOD mice and type 1 diabetes patients [2]. Particularly in humans, it is unclear which of these are the most ‘important’, i.e. critical for disease initiation and/or progression. However, accumulating evidence suggests that insulin is a key target of pathogenic T cells in type 1 diabetes in both NOD mice and humans and that it may even be the long-sought-after ‘initiating antigen’ in this disease [91–103]. In NOD mice, establishment of tolerance to insulin can lead to prevention of diabetes [95,100,101] as well as remission of established disease [93]. Importantly, CD8⁺ and CD4⁺ T cell responses to insulin have also been reported in type 1 diabetes patients [91,94,96,99,102]. Furthermore, in humans, the non-MHC locus that confers the strongest susceptibility to type 1 diabetes is the insulin gene variable number of tandem repeats (VNTR) regulatory region [104], and disease-associated alleles are correlated with reduced thymic expression of the insulin gene [105]. We are exploring the feasibility of DEC-205-mediated delivery of the entire preproinsulin molecule, rather than only the known epitopes targeted by effector T cells. This strategy should facilitate translation to patients expressing diverse MHC molecules. In addition, the epitopes recognized by insulin-specific regulatory T cells are largely uncharacterized and could differ from those targeted by pathogenic effector T cells [106]. The finding that DC-expanded T_regs of a single specificity can both prevent and reverse type 1 diabetes in NOD mice [23,90] provides critical support for this approach.

Insuring against harm

We found that peptide-linked anti-DEC-205 could induce tolerance even in NOD mice with ongoing islet inflammation [69]. However, when contemplating the translation of such a strategy to humans, there is a concern that antigen delivery to DCs in the context of an inflammatory environment could lead to exacerbation of a pathogenic autoimmune response rather than tolerance induction. One potential remedy to be considered is the simultaneous use of siRNA specific for co-stimulatory molecules which could be targeted to the DCs in vivo through either DEC-205 or another DC receptor. In vivo siRNA delivery, although difficult to achieve, has been conducted through cell surface receptors by other groups [107–110]. Another possible strategy would be to use microsphere carriers of anti-sense oligonucleotides that can down-modulate co-stimulatory molecules on DCs in vivo[111].

Considerations of disease stage

DC-based therapeutics for type 1 diabetes should be considered at all stages of the disease, including prediabetes, new-onset diabetes and the setting of islet transplantation. In general, it has been easier to prevent diabetes in the NOD mouse model than it has been to reverse it [112]. For this and other reasons, it has been argued that prevention should be the goal [106]. However, given the more favourable risk to benefit ratio represented by new-onset diabetes patients, it may be easier to conduct clinical trials in such individuals, and there are examples of successful reversal of type 1 diabetes in NOD mice (e.g. by transfer of DC-expanded T_regs[90] or in vivo delivery of anti-sense oligonucleotides for CD80, CD86 and CD40 [111]). Studies in animal models suggest that DC-based immunotherapeutic strategies might also be utilized to facilitate islet transplant tolerance [113–115].

DC-based therapies in humans

DC-based therapeutic approaches designed to stimulate immune responses to tumours have been employed in patients with advanced cancers for nearly 15 years, with one of the earliest reports appearing in 1996 [10]. Such studies utilize DCs pulsed with tumour antigens [10], tumour antigen-derived peptides [6,7,11,12,15] or tumour lysates [9], or DCs transfected with tumour antigen cDNA (e.g. Muc1) [13], total tumour RNA [14] or RNA encoding tumour antigens (e.g. prostate-specific antigen) [8]. As reviewed, such therapies are safe, and tumour regression has been observed in some patients [22]. Multiple studies have revealed that mature DCs are optimal for stimulation of anti-tumour immune responses [7,11]. In contrast, and of clear relevance for type 1 diabetes therapeutics, when immature DCs pulsed with an influenza matrix peptide were administered to healthy controls [49,116] the outcome was inhibition of the function of peptide-specific effector CD8⁺ T cells and the appearance of peptide-specific IL-10-producing CD8⁺ T cells [116], as well as regulatory CD8⁺ T cells that required cell–cell contact to exert their suppressive effects [49].

At this time, the use of DCs in humans is being extended slowly beyond cancer immunotherapy to treatment of infectious diseases [117] and autoimmune diseases including type 1 diabetes [118] and rheumatoid arthritis [119]. As discussed in an earlier section of this review, the administration of DCs rendered phenotypically immature by treatment with anti-sense oligonucleotides for CD80, CD86 and CD40 can prevent diabetes development in NOD mice [50,63]. The safety of this strategy is currently being evaluated in a Phase I clinical trial of long-standing adult type 1 diabetes patients in which autologous DCs are being generated from blood precursors after leukapheresis and treated with anti-sense oligonucleotides in vitro[118]. In this study, which began in 2007, the DCs are injected intradermally at a site proximal to the pancreas where they are expected to migrate to the nearest lymph nodes, including those of the pancreas. This same group reported that in vivo administration of microspheres incorporating the anti-sense oligonucleotides is capable of preventing and reversing type 1 diabetes development in NOD mice [111], and they anticipate human trials in the near future [118]. If approved, this strategy would greatly simplify the therapeutic protocol, as it would eliminate the need for leukapheresis and in vitro DC generation and treatment with oligonucleotides.

Conclusion

Despite the DC defects that have been reported in NOD mice [120–123], a variety of DC-based immunotherapeutic strategies have shown great promise in this model, as we have summarized here (Fig. 2). Now the challenge will be to translate these approaches to patients. The ongoing investigation of the safety of phenotypically immature autologous DCs administered to type 1 diabetes patients represents a giant step forward in this regard [118]. In our view, efforts should be made to minimize the possibility of global immunosuppression by utilizing strategies that confer antigen specificity, either by deliberate incorporation of beta cell antigens or by exploiting the migration patterns of administered DCs [60,65]. Continued evaluation of such strategies, particularly in humanized models of the disease [124], should help to allay translational fears and facilitate the transit of DC-based therapies to patients.

Fig. 2.

Dendritic cell (DC)-based therapeutic strategies in the fight against diabetes. DCs have shown promising results in the treatment of type 1 diabetes when used alone or in conjunction with various other molecules (for details, see text). Targeting of DCs with islet antigens has also proved successful. However, most of these studies have been performed in non-obese diabetic (NOD) mice or other mouse models for type 1 diabetes. Although they serve as excellent models, they show significant differences from the disease in humans. Therefore, the ultimate goal, as well as challenge, is to translate these successful studies to human patients. It is hoped that the ongoing Phase I clinical trial with autologous immature DCs (iDC) (see Ref. [118]) will establish the safety of DC administration in type 1 diabetes patients.

Disclosure

The authors declare no conflicts of interest.