|Home | About | Journals | Submit | Contact Us | Français|
Type 1 Diabetes (T1D), also called juvenile diabetes because of its classically early onset, is considered an autoimmune disease targeting the insulin-producing β cells in the pancreatic islets of Langerhans. T1D reflects a loss of tolerance to tissue self-antigens caused by defects in both central tolerance, which aims at eliminating potentially autoreactive lymphocytes developing in the thymus, and peripheral tolerance, which normally controls autoreactive T cells that escaped the thymus. Like in other autoimmune diseases, the mechanisms leading to T1D are multifactorial and depend on a complex combination of genetic, epigenetic, molecular, and cellular elements that result in the breakdown of peripheral tolerance. In this article, we discuss the contribution of these factors in the development of the autoimmune response targeting pancreatic islets in T1D and the therapeutic strategies currently being explored to correct these defects.
The immune system is a finely tuned network of intertwined pathways that work to recognize, respond to, and eliminate infectious agents while refraining from self-inflicted tissue destruction. The fundamental self versus nonself discrimination starts in the earliest days of development when the white blood T-cell system first forms in the thymus. T-cell development is based on an inherent selection system wherein maturing thymocytes encounter small self peptides bound to cell surface molecules, termed major histocompatibility complex (MHC) proteins, that simultaneously skew the repertoire toward self-recognition whereas eliminating mature T cells expressing a T-cell receptor (TCR) with too high an affinity for self-peptides bound to the MHC selecting molecules. However, this process of negative selection, termed central tolerance, is not complete as some autoreactive T cells escape selection, either because of the absence of certain peptide-MHC complexes in the thymus, or genetic defects or variants that compromise the thymic selection process. Fortunately, in most individuals, back-up processes control potentially autoreactive T cells in the periphery. These processes include the active elimination of self-reactive T cells through inappropriate peptide-MHC recognition leading to cell death or inactivation as well as dominant regulatory processes that shut down the potentially autoreactive T cells. Unfortunately, in some individuals, genetic predisposition, combined with environmental stresses can lead to a breakdown in peripheral tolerance leading to autoimmunity. Type 1 Diabetes (T1D) is an example of one such autoimmune disease wherein the breakdown in tolerance leads to the initiation and progressive destruction of the insulin-producing β cells. In this article, we describe the key players in this saga. We highlight the multifactorial events that lead to the breakdown of peripheral tolerance and development of the disease. Finally, we discuss ongoing clinical efforts to develop therapeutic approaches to repair and reinstate immune homeostasis to treat and prevent this devastating disease.
The pathogenesis of T1D remains unclear, but clues to its origins and etiology can be gathered from studies of both nonobese diabetic (NOD) mice and humans (Fig. 1). For instance, it is widely believed that pathogenic autoreactive T cells, which infiltrate (so-called “insulitis”) and destroy the pancreatic islets, mediate T1D. Both CD4+ and CD8+ T cells can transfer disease in the NOD mouse model of T1D (Anderson and Bluestone 2005). Additionally, T cells specific for islet self-antigens, including insulin, glutamic acid decarboxylase 65 (GAD65), insulinoma-associated protein 2 (IA2), zinc transporter 8 (ZnT8), and islet-specific glucose-6-phosphatase catalytic subunit-related protein (IGRP), have been found in the islets and peripheral blood of NOD mice and T1D patients, respectively (Lieberman and DiLorenzo 2003). Autoantibodies targeting some of these islet antigens are used as diagnostic and prognostic tools in the clinic. However, these autoantibodies are not believed to be directly pathogenic and the important role of B cells in diabetes is instead related to their efficient presentation of self-antigens to autoreactive T cells (Wong et al. 2004; Bour-Jordan et al. 2007). The activation of autoreactive T cells in NOD mice is initiated by dendritic cells (DCs) presenting islet self-antigens in the draining pancreatic lymph nodes (LN) following a wave of β-cell death of unexplained etiology early in life (Turley et al. 2003). Further immune-mediated tissue damage results in additional shedding of islet antigens and epitope spreading in the autoreactive T-cell response, leading to infiltration of the tissue by a diverse population of autoreactive T cells that nevertheless are predominantly tissue-specific rather than recruited by bystander mechanisms (Lennon et al. 2009). Thus, after an unknown initial hit, priming of autoreactive T cells in the pancreatic LN represents the first step toward autoimmune diabetes.
For autoreactive CD8+ T cells, tolerance is achieved by immunological ignorance if the level of self-antigen presentation in the draining LN (dLN) is low, and by anergy or death, which is mediated by Bim, the proapoptotic protein when expressed at high levels. The ability to break tolerance depends both on the avidity of TCR-self-antigen interactions and the status of DCs presenting self-antigens in the pancreatic LN (Dissanayake et al. 2011). Indeed, the phenotypic and functional status of DCs and the DC subset involved in presentation are critical to determine whether DCs will promote tolerance or present antigen in an immunogenic manner. Additionally, the avidity of interactions between autoreactive TCR and their cognate antigen presented by DCs must reach a certain threshold to trigger the activation of autoreactive CD8+ T cells in pancreatic LN and their pathogenic potential in the tissue (Gronski et al. 2004). In this regard, it has been shown that progression from insulitis to diabetes was driven by “avidity maturation” of islet-specific CD8+ T cells (Amrani et al. 2000). Additionally, among many known inhibitory molecules, CTLA-4 and Cbl-b are important to prevent T1D as they can increase the threshold for TCR signaling and represent crucial cell-intrinsic brakes for the activation of autoreactive T cells in pancreatic LN (Ausubel et al. 2002; Walker et al. 2002; Eggena et al. 2004; Gronski et al. 2004; Hoyne et al. 2011). Of note, the influence of negative regulators of TCR signaling on tolerance can be complex and is not necessarily limited to the periphery. For example, polymorphisms in the T1D susceptibility gene PTPN22 encoding the lymphoid protein tyrosine phosphatase (LYP), which negatively regulates TCR signaling, have been characterized as gain-of-function mutations that result in lower T-cell activation and IL-2 production (Vang et al. 2005). This, in turn, may lead to both thymic selection of autoreactive cells caused by defects in TCR signal strength and negative selection, and suboptimal IL-2 production by effector T cells in the islets that could locally compromise immunoregulation by Tregs, as discussed later in this article.
Importantly, it is becoming clear that recognition of self-antigen on antigen-presenting cells (APCs) other than tolerogenic DCs may play a role in the maintenance of peripheral tolerance. Stromal cells have been recently shown to present tissue-specific antigens (TSAs) in LNs, through mechanisms involving the transcription factors Autoimmune Regulator (Aire) or Deformed Epidermal Autoregulatory Factor 1 (Deaf1) (Lee et al. 2007; Gardner et al. 2009). Presentation of islet antigens by stromal cells in pancreatic LN may participate in tolerizing autoreactive T cells by inducing their deletion and may thus be involved in preventing diabetes in mice and humans (Lee et al. 2007; Gardner et al. 2008; Yip et al. 2009). In this regard, mutations in genes coding for Aire and PTPN22 have been associated with T1D (Bottini et al. 2004, 2006). Additionally, polymorphisms in the promoter region of the insulin gene contribute to disease susceptibility (Pugliese et al. 1997), possibly by affecting the ability of Aire to mediate thymic epithelial expression of insulin as a self-antigen during negative selection. Taken together, these findings reflect the qualitative as well as quantitative role of the presentation of islet antigens and the level of TCR signaling in the thymus and the periphery in the maintenance of peripheral tolerance (see Fig. 1 in Gardner et al. 2009).
Finally, the subset of T cells activated by self-antigens and, in particular, their cytokine production may be important for the outcome of the immune response. There has long been a general consensus that proinflammatory Th1 cells (producing IFN-γ) tend to be pathogenic whereas Th2 cells (producing IL-4) can protect from diabetes in the NOD mouse. The diabetogenic role of Th1 cells was suggested by numerous studies, including prevention of diabetes by anti-IFN-γ monoclonal antibodies (mAbs) in NOD mice, abrogation of disease in IFN-γ deficient mice in the RIP-LCMV TCR-transgenic (TCR-Tg) model of diabetes, and protection from diabetes in NOD mice deficient for the Th1 lineage-specific transcription factor T-bet (Debray-Sachs et al. 1991; von Herrath and Oldstone 1997; Esensten et al. 2009). The protective role of IL-4 in autoimmune diabetes has been suggested by a converging array of evidence including decreased diabetes incidence in NOD mice treated with exogenous IL-4 (Rapoport et al. 1993) or expressing IL-4 in pancreatic islets (Mueller et al. 1996), decreased diabetes after injection of islet-specific T cells expressing high levels of IL-4 (Tian et al. 1996; Ploix et al. 1998), the association of tolerogenic therapies or regulatory cell types such as NKT cells with IL-4 production (Elias et al. 1997; Hammond et al. 1998) and the inverse correlation of disease severity with IL-4 (Lenschow et al. 1996; Fox and Danska 1997). However, the incidence of diabetes was not dramatically affected in NOD mice genetically deficient for IFN-γ (Hultgren et al. 1996; Serreze et al. 2001) or for IL-4 (Wang et al. 1998), suggesting that deficiency in either cytokine alone was not sufficient to modify disease progression. Rather, the pathogenic versus tolerogenic role of IFN-γ versus IL-4 likely reflects a more complex pattern of cell populations and cellular interactions involved in promoting autoimmunity or favoring tolerance. Furthermore, this paradigm has to be reexamined in view of the recent realization that there is a certain degree of plasticity in T-cell subsets (Zhou et al. 2009a; O’Shea and Paul 2010), which implies that the contribution and stability of a given cell subset during the autoimmune response may vary depending on epigenetic factors and the immunological environment. In this regard, it was recently shown that proinflammatory Th17 cells, which have been implicated in many autoimmune diseases (Korn et al. 2009), only induce diabetes after conversion into IFN-γ-producing Th1-like cells (Bending et al. 2009; Martin-Orozco et al. 2009).
Successful activation of T cells to promote autoimmunity, including T1D, requires two signals, the TCR signal 1 and costimulatory signal 2. This concept has become a paradigm for understanding the positive activation events needed to initiate immunity. However, over the last 15 years, it has become increasingly apparent that there are negative regulatory receptors that control immunity. The cell surface receptor/ligand pairs are quintessential for maintaining homeostasis by tempering normal immunity and shutting down unwanted immune responses. Two of the best-characterized negative regulators are, Cytotoxic T-lymphocyte antigen-4 (CTLA-4) and Programmed death-1 (PD-1) (Bour-Jordan and Bluestone 2009). These molecules are related members of the CD28 family but are selectively expressed on activated T cells and when engaged shut down immune activation to promote T-cell homeostasis and peripheral tolerance. CTLA-4 and PD-1 are both important to maintain peripheral tolerance and control autoimmune diabetes in the NOD strain. Although NOD mice deficient for CTLA-4 die at 3–4 weeks of age of a massive lymphoproliferative disease (LPD), as in other genetic backgrounds, CTLA-4KO islet-antigen specific BDC2.5 TCR-Tg mice develop autoimmune diabetes with higher incidence and earlier onset compared with CTLA-4 sufficient animals (Luhder et al. 2000). Conversely, triggering of CTLA-4 signaling by a single-chain membrane-bound anti-CTLA-4 Ab transgene (scαCTLA-4 Tg) expressed on B cells altered the activation of islet-specific T cells and reduced the incidence of diabetes in NOD mice (Griffin et al. 2000; Fife et al. 2006a). Of note, CTLA-4 is also critical for the function of regulatory T cells (Tregs) (Fife and Bluestone 2008; Wing et al. 2008; Bour-Jordan and Bluestone 2009), but the scαCTLA-4 Tg protected Treg-deficient NOD-B7-1/2KO mice from diabetes as well, showing an autoreactive conventional T-cell intrinsic effect (Fife et al. 2006a). The absence of PD-1 signals in NOD mice deficient for PD-1 or its ligands, PD-L1 and PD-L2, results in accelerated diabetes (Fife et al. 2006b). Importantly, CTLA-4 and PD-1 appear to have complementary but nonoverlapping roles in diabetes tolerance (Fife and Bluestone 2008). Indeed, CTLA-4 blockade is most efficient in early stages of the autoimmune response in the pancreatic LN whereas expression of PD-L1 in the peripheral tissue is critical for the tolerogenic role of PD-1 (Luhder et al. 1998; Walker et al. 2002; Keir et al. 2006). In a model of tolerance induced by antigen-coupled ethylcarbodiimide (ECDI)-fixed spleen cells, both CTLA-4 and PD-1 were important in induction of tolerance, but only PD-1 controlled maintenance of the tolerant state (Fife et al. 2006b). This paradigm was reinforced by the predominant role of PD-L1, rather than PD-L2, in the tolerogenic role of PD-1 in diabetes, which correlates with the widespread expression of PD-L1, including in pancreatic islets, as compared with the limited expression of PD-L2 on DCs and monocytes (Fife et al. 2006b; Keir et al. 2006).
Potential defects in the CTLA-4 and/or PD-1 pathways in T1D patients are suggested by the association of polymorphisms in both genes with susceptibility to disease (Nielsen et al. 2003; Ueda et al. 2003). Additionally, lower expression levels of PD-1 was found on peripheral blood-derived CD4+ T cells from a small group of T1D patients as compared with controls (Tsutsumi et al. 2006). The CTLA-4 gene has several splice variants, most notably a ligand-independent isoform (liCTLA-4) and a soluble secreted isoform (sCTLA-4) found in mice and humans. Several autoimmune diseases, including T1D, have been associated with the differential expression of splice variants. In NOD mice, lower expression of the liCTLA-4 isoform caused by a single nucleotide polymorphism in exon 2 has been identified as the basis for one of the genetic loci, Idd5.1, that has been mapped, genetically to be associated with disease susceptibility in mice (Araki et al. 2009). In fact, increasing expression of liCTLA-4 in Tg mice reduced the incidence of diabetes in NOD mice (Araki et al. 2009). The respective role and mechanisms of full-length and ligand-independent CTLA-4 isoforms in peripheral tolerance are still under discussion, but it is conceivable that they differentially control homeostatic proliferation versus tissue-specific T-cell responses (Bour-Jordan et al. 2011).
Finally, other negative regulators of T-cell function have been described and could be involved in maintaining tolerance to pancreatic islets. B- and T-lymphocyte attenuator (BTLA) is a negative regulator of T-cell activation that is predominantly expressed on recently activated T cells and differentiated Th1 cells (Watanabe et al. 2003). Islet-specific TCR-Tg CD8+ T cells deficient for BTLA showed increased accumulation in pancreatic LN and induced diabetes more efficiently than wild-type (WT) T cells (Liu et al. 2009). Additionally, treatment with a nonblocking anti-BTLA mAb resulted in CD4+ T-cell and B-cell depletion and decreased diabetes incidence in NOD mice (Truong et al. 2009). Tim-3 (T-cell immunoglobulin domain, mucin domain-3) is another inhibitory molecule of T-cell function preferentially expressed on Th1 cells (Monney et al. 2002). Blockade of interactions between Tim-3 and Tim-3L accelerated diabetes in NOD mice (Sanchez-Fueyo et al. 2003). Although their mode of action remains to be fully elucidated, both BTLA and Tim-3 appear to affect tolerance by altering the relative number and/or functional efficiency of Th1, Th2, and Treg subsets (Sanchez-Fueyo et al. 2003; Liu et al. 2009; Truong et al. 2009).
During the last decade, a wealth of data has identified CD4+ CD25+ Foxp3+ regulatory T cells (Tregs) as a critical cell population for the extrinsic control of peripheral tolerance. In scurfy mice and immune dysregulation, polyendocrinopathy enteropathy X-linked (IPEX) syndrome IPEX patients, genetic deficiency in the Treg lineage marker Foxp3 rapidly results in widespread autoimmunity and a fatal LPD (Bennett et al. 2001; Fontenot et al. 2003; Hori et al. 2003; Khattri et al. 2003). Foxp3 remains critical throughout life to maintain the Treg population and prevent autoimmunity (Kim et al. 2007). In NOD mice, depletion of CD4+ CD25+ Tregs greatly accelerates the development of diabetes, and T1D is a hallmark of the clinical triad in infants with IPEX syndrome (Salomon et al. 2000). Similarly, affecting the homeostasis of Tregs by targeting signals necessary for their development or survival, such as IL-2 or CD28, exacerbates diabetes and other underlying autoimmune diseases in NOD mice (Salomon et al. 2000; Setoguchi et al. 2005; Meagher et al. 2008). NOD mice deficient for Foxp3 develop the same LPD as in other strains (Chen et al. 2005), but conditional depletion of Foxp3-expressing Treg in BDC2.5 TCR-Tg mice resulted in fulminant diabetes within 3 days (Feuerer et al. 2009). Thus, peripheral tolerance in NOD mice is dependent on the balance of effector and regulatory T cells (Fig. 2) (Bour-Jordan et al. 2004a).
Pathogenic T cells undergo qualitative changes during the progression to diabetes that render them less susceptible to regulation by Tregs (You et al. 2005; D’Alise et al. 2008). Counterintuitively, the percentage of Tregs in pancreatic LN increased with age in NOD mice (Tang et al. 2008). However, in the pancreatic tissue, there was an inverse correlation between the frequency of Tregs and the size of the infiltrate in NOD mice, suggesting a local deficiency in Treg-mediated immunoregulation (Tang et al. 2008). Additionally, islet-infiltrating Tregs displayed low levels of the Interleukin-2 receptor alpha (CD25), Foxp3, and the survival factor Bcl-2, which levels of expression are all promoted by IL-2 signaling. This, in turn, could affect the survival and suppressive function of Tregs in pancreatic islets (Malek et al. 2002; Fontenot et al. 2005; Barron et al. 2010). Importantly, treatment with low-dose IL-2 preferentially targets regulatory rather than effector T cells, restores Treg expression of CD25 and Foxp3, and can prevent or even reverse diabetes in NOD mice (Tang et al. 2008; Grinberg-Bleyer et al. 2010). In this regard, the IL-2 gene was identified as the likely candidate for susceptibility locus Idd3 in NOD mice, and polymorphisms in the IL-2 gene result in lower levels of IL-2 production (Wicker et al. 1994; Lyons et al. 2000; Yamanouchi et al. 2007). Thus, defective IL-2 production by Teff in pancreatic islets could be responsible for the low CD25 expression and poor survival of Tregs in the tissue.
Of note, genome wide association studies have identified polymorphisms in the IL-2, CD25 and PTPN22 genes associated with susceptibility to T1D (Bottini et al. 2004; Vella et al. 2005; Todd et al. 2007). Tregs from T1D patients express normal levels of CD25 but present defects in IL-2 receptor signaling, including inferior phosphorylation of STAT5, which may result in impaired Foxp3 expression (Long et al. 2010). The percentage of Tregs in the peripheral blood is normal in T1D patients (Liu et al. 2006) and Tregs from T1D patients and control subjects were found to be equally capable of suppressing T-cell proliferation in vitro (Putnam et al. 2009). However, like in NOD mice, polyclonal Treg numbers at the site of inflammation might be dysregulated even when Treg frequencies in peripheral blood are normal. In addition, a reduction of the most potent antigen-specific Treg might be missed by measuring Treg numbers solely based on CD25 and Foxp3. Quantitation of Treg to Teff ratios in human pancreas is challenging not only because of difficulties in getting access to recent onset T1D pancreata but also because islet destruction can be limited to certain lobules whereas others remain intact (Bluestone et al. 2010). Other studies have described defects in in vitro suppression that may relate to an increased resistance of Teff to suppression (Schneider et al. 2008), as in NOD mice. Moreover, the function of human Tregs is commonly assessed by in vitro suppression assays. However, genetic studies in mice have clearly established that Treg dysfunction that leads to very severe spontaneous LPD in vivo can go along with normal in vitro suppressive capacity (Liston et al. 2008; Zhou et al. 2008; Lu et al. 2010). Thus, while impaired in vitro suppression suggests a Treg defect, normal in vitro suppressive capacity does not equal normal Treg function. It therefore remains to be determined if T1D Tregs are fully functional and alternative human Treg assays are needed.
Recently, the plasticity of nTregs as it relates to the stability of the Foxp3 lineage marker has been evaluated in elaborate reporter mouse models. A significant subset of CD4+ T cells was found to show unstable Foxp3 expression (exFoxp3 cells) and produce IFN-γ and IL-17 (Zhou et al. 2009b). exFoxp3 cells were enriched in inflammatory conditions and/or autoimmunity-prone backgrounds such as the NOD mouse. Importantly, self-reactive exFoxp3 cells were pathogenic and could induce autoimmunity on adoptive transfer in vivo. These cells were likely derived both from unstable Tregs and from Teff that had transiently up-regulated Foxp3, a result that has implications in humans in which most Teff transiently express Foxp3 on activation. In this regard, the frequency of Tregs producing proinflammatory cytokines such as IFN-γ or IL-17 is elevated in T1D patients (McClymont et al. 2011). These Foxp3+IFN-γ+ Tregs are suppressive in vitro but display phenotypic and epigenetic characteristics of adaptive Tregs, such as low expression of Helios and overall methylation of the TSDR at the Foxp3 locus, raising the possibility that they may not be as stable as natural (e.g., thymus-derived) Tregs. Thus, although the stability of Tregs is still controversial (Rubtsov et al. 2010; Bailey-Bucktrout et al. 2011), the prospect of unstable Tregs is clearly a topic that must be analyzed further in view of the current development of Treg-based clinical trials in autoimmunity and transplantation.
Despite all the advances in Treg biology, the suppressive mechanisms employed by Tregs are still a matter of debate. Among the mechanisms that have been proposed, Tregs can directly suppress Teff cells in a contact-dependent manner, especially in vitro, produce immunoregulatory cytokines such as IL-10 and TGF-β, and alter antigen presentation by dendritic cells and/or drive them to produce suppressive factors such IDO (Tang and Bluestone 2008). It is possible that the identification of these different pathways reflects the suppression by Tregs of immune responses taking place in different tissues with distinct dynamics and unique immunological environments. In this regard, it has been postulated that the transcriptional program of Tregs, and conceivably their suppression mechanism, can be tailored to the nature of the effector response they regulate (Chaudhry et al. 2009; Koch et al. 2009; Zheng et al. 2009). Another confounding factor is the fact that the “regulatory T cell” label can be comprised of functionally distinct subsets. For example, besides thymically-derived CD4+ Foxp3+ natural Tregs (nTregs), CD4+ Foxp3+ adaptive Tregs (aTregs) (induced from CD4+ CD25− cells), TGF-β-dependent Th3 cells, IL-10-dependent Tr1 cells, and CD8+ Tregs have been described in mice and humans and could be involved in different models of tolerance (Bisikirska et al. 2005; Roncarolo et al. 2006; Weiner et al. 2011). In the NOD mouse, Tregs have been shown to interact with DCs rather than autoreactive T cells and alter presentation of self-antigens in pancreatic LNs (Tang et al. 2006). Ablation of Tregs acutely triggered diabetes in BDC2.5 TCR-Tg mice. Intriguingly, the immunoregulatory control exerted by Tregs in the pancreatic islets appeared to largely target natural killer (NK) cells and their production of IFN-γ, which subsequently licensed autoreactive CD4+ Teff for tissue destruction (Feuerer et al. 2009). Different subsets of regulatory T cells producing IL-10 or TGF-β have been found in unmanipulated NOD mice or after tolerogenic treatments such as anti-CD3 therapy (You et al. 2004, 2007; Bresson et al. 2006; Weiner et al. 2011). In humans, T cells stimulated with self-peptides isolated from a susceptibility MHC allele were skewed toward proinflammatory cytokine production in T1D patients versus immunoregulatory IL-10 in healthy controls (Arif et al. 2004). Additionally, treatment of new-onset diabetic patients with FcR-nonbinding anti-CD3 mAbs, which led to insulin independence for over 5 years in some cases, was associated with enrichment in IL-10-producing CD4+ T cells and in CD8+ CD25+ Foxp3+ regulatory T cells (Herold et al. 2003, 2009; Bisikirska et al. 2005). Thus, Treg play a crucial role preventing T1D through a multitude of effector mechanisms and subtypes.
Despite having a well-documented genetic component, it has also been recognized that the environment plays an important role in the triggering of or protection from T1D. The exact nature of environmental factors influencing the development of T1D in humans is largely speculative, but it is generally accepted that frequent exposure to pathogens inversely correlates with the incidence of disease, a phenomenon known as the “hygiene hypothesis.” Similarly, diabetes in NOD mice is influenced by the cleanliness of the facility housing the mouse colony, with the incidence of disease higher in NOD mice housed in SPF facilities by comparison with “dirty” facilities (Pozzilli et al. 1993; Okada et al. 2010). Diabetes in NOD mice is also prevented by administration of complete Freund’s adjuvant, which is essentially an emulsion of mycobacteria (Sadelain et al. 1990; McInerney et al. 1991). The mechanisms involved are complex, but among pathways that sense and recognize foreign pathogens, receptors for pathogen-associated molecular patterns (PAMPs), most notably toll-like receptors (TLRs), are likely to play a significant role.
Most TLRs use the adapter protein MyD88 for signal transduction. MyD88 affects antigen presentation by both DCs and B cells and promotes T-cell proliferation and survival in a T cell-intrinsic manner (Gelman et al. 2006; Hou et al. 2008, 2011; Rahman et al. 2008). Thus, deficiency in MyD88 leads to a profound defect in T cell and T cell-dependent responses. It was recently shown that MyD88 influences the composition of the commensal flora in the gut of NOD mice, and that this in turn affects the development of diabetes (Wen et al. 2008). Indeed, NOD mice deficient for MyD88 (MyD88KO) were protected from diabetes in SPF facilities. Intriguingly, NOD-MyD88KO mice housed in germ-free facilities developed diabetes at a normal rate, but infiltration in the islets was less severe if these mice were colonized with a representative sample of gut flora from NOD mice. Protection from diabetes in SPF NOD-MyD88KO mice was accompanied by reduced islet infiltration. Additionally, T-cell responses to islet antigens were lower in pancreatic LN only and spleen cells from NOD-MyD88KO mice transferred diabetes in immunodeficient recipients as efficiently as NOD cells, demonstrating a local but not global tolerance to islet antigens (Wen et al. 2008). The underlying mechanisms are still unknown, but NOD-MyD88KO mice had a normal frequency of Tregs in pancreatic LN, and tolerance induction in this model may involve alterations of APCs and presentation of self-antigens. In this regard, the gut mucosal tissue has long been known to be a hot spot for T-cell tolerance resulting from a complex network of interactions between various cell-types, including tolerogenic CD103+ DCs and various populations of regulatory T cells (Weiner 2001; Coombes et al. 2007). The oral route is a prevalent path of pathogen exposure and increased intestinal permeability has been associated with diabetes in mice and humans (Vaarala et al. 2008). Additionally, gut antigens have been shown to be presented in pancreatic LN and the immune environment in the gut can affect the presentation of islet antigens (Turley et al. 2005). Finally, oral administration of anti-CD3 mAbs ameliorates diabetes in NOD mice (Weiner et al. 2011). Taken together, these data suggest that gut antigens and the gut flora can influence the development of diabetes.
MicroRNAs (miRNAs) are a class of nonprotein-coding RNAs that repress other genes posttranscriptionally by binding complementary mRNA sequences leading to translational inhibition and/or mRNA destabilization (Lim et al. 2005; Fabian et al. 2010; Guo et al. 2010). miRNAs are considered fine-tuners (Bartel 2009) because inhibition is moderate for most individual proteins, often less than 50% (Baek et al. 2008; Selbach et al. 2008). Genetic loss of one allele can, however, impair immune homeostasis: PTEN or proapoptotic protein Bim heterozygous mice develop spontaneous systemic autoimmune disease (Bouillet et al. 1999; Di Cristofano et al. 1999). Importantly, B7-1 and B7-2 heterozygous NOD mice display reduced Treg numbers (Salomon et al. 2000; Bour-Jordan and Bluestone 2009). In B7-1 heterozygous mice this leads to accelerated diabetes onset compared to control mice because of a change in CD28 and/or CTLA-4 signaling (Salomon et al. 2000) in T cells. In contrast, B7-2 heterozygous NOD mice are protected from disease, despite the reduced Treg numbers. Further analysis revealed that reduction of B7-1 preferentially affected Treg numbers whereas reduction of B7-2 preferentially reduced Teff numbers (Bour-Jordan et al. 2004). Thus, gradual effects selectively affect the Teff/Treg balance (Bour-Jordan and Bluestone 2009). These examples illustrate how sensitive the immune system can be to minor changes in protein concentrations. miRNAs are likely involved in the tight regulation of immunologically relevant protein concentrations and haploinsufficient genes are likely key miRNA targets (Xiao and Rajewsky 2009; Jeker and Bluestone 2010). Finally, miRNAs target hundreds of genes simultaneously (Lim et al. 2005) and often several members of the same biochemical pathway are targeted by the same miRNA resulting in a cumulative inhibition. In summary, minor changes in protein concentrations can have dramatic effects and can predispose to autoimmune disease.
Cell-type-specific ablation of key proteins required for proper miRNA biogenesis (Dicer, Drosha or DGCR8) has shown a critical role for miRNAs in all immune cell types tested so far (Cobb et al. 2005, 2006; Chong et al. 2008; Koralov et al. 2008; Liston et al. 2008; Zhou et al. 2008; Bezman et al. 2010). Decreased proliferation and survival are commonly found. Deletion of Dicer in all T cells leads to a spontaneous multiorgan autoimmune disease in about half of the affected mice (Cobb et al. 2006). CD4+ Foxp3+ regulatory T-cell-specific miRNA ablation, leads to a much more aggressive disease closely resembling Foxp3-deficient scurfy disease in which Treg are entirely dysfunctional (Chong et al. 2008; Liston et al. 2008; Zhou et al. 2008).
Despite these clear demonstrations of an essential nonredundant role for miRNAs in immune regulation our understanding of the role of individual miRNAs in immune regulation is fairly limited. Some of the best studied miRNAs to date are miR-155 and miR-146. Mice deficient for miR-155 display a complex phenotype of immunodeficiency and likely increased autoimmunity caused by dysfunction of DC, B, and T cells (Rodriguez et al. 2007; Thai et al. 2007). Simultaneous immunodeficiency and autoimmune disease might be a consequence of miR-155 having cell-type-dependent opposing functions in immune regulation. Surprisingly, despite bic/miR-155 being a direct Foxp3 target leading to significant enrichment in Treg expression (Gavin et al. 2007; Marson et al. 2007; Zheng et al. 2007) Treg function is largely preserved in the absence of miR-155 (Kohlhaas et al. 2009; Lu et al. 2009). Nevertheless, miR-155 is required for normal numbers of Treg under homeostatic conditions, possibly in part because of impaired proliferation and IL-2 signaling in the absence of miR-155 (Lu et al. 2009). Of note, Dicer-deficient Treg lose their lineage defining transcription factor Foxp3 (Zhou et al. 2008). Given the important role of IL-2 for the maintenance of high Foxp3 and CD25 expression and Treg function and stability in inflamed pancreatic islets (Tang et al. 2008; Zhou et al. 2009b) it is conceivable that miR-155 might play a role in the prevention of T1D through stabilizing Treg. On the other hand, miR-155-deficient mice are resistant to experimental autoimmune encephalitis (EAE), a model of multiple sclerosis (O’Connell et al. 2010) caused by defects in T cells and DCs. Thus, miR-155 plays opposing functions in immune regulation.
In contrast, miR-146a was dampening immune responses in all immune cell types analyzed to date. miR-146a-deficient mice spontaneously develop an inflammatory syndrome of myeloproliferation and autoimmunity likely due to a reduced negative feedback loop in various myeloid cells and T cells (Taganov et al. 2006). In addition, miR-146a is enriched in Treg and is required under stress conditions for proper Treg suppression of Th1 responses (Lu et al. 2010). Thus, miR-146a seems to dampen immune responses through negative regulation of proinflammatory cell types and enhancement of suppressive Treg. If, and to what degree, miR-146a prevents T1D by repressing IFN-γ secretion by autoreactive T effector cells and/or enhancing Treg function to suppress pancreatic inflammation will need careful analysis. Such studies might shed new light onto the controversial role of IFN-γ in the pathogenesis of T1D.
With miRNAs predicted to be targeting 50% of all protein coding genes, there is no doubt that miRNAs are involved in the prevention of T1D. Studies investigating a causative involvement of miRNAs in T1D are currently lacking. Treg from T1D patients display altered miRNA expression compared to Treg from healthy controls (Hezova et al. 2010) and miR10a has been shown to be differentially expressed in regulatory T cells from diabetes-prone NOD mice (LT Jeker, Y Zhou, and K Gershberg, submitted). The importance of miRNAs for the control of immune regulation suggests that dysregulation of miRNAs or mutations in binding sites of miRNA targets may shift the balance between immunity and immune regulation. This may happen in various cell types and may influence multiple immunologic checkpoints of tolerance. A better understanding of miRNA function in the immune system in general and in T1D in particular are needed to exploit this relatively new class of genes for diagnostics and/or therapeutic approaches.
Many therapeutic approaches aimed at correcting some of the immunological defects described above are currently being tested in clinical trials in T1D (Bluestone et al. 2010; van Belle et al. 2011). Among these, we will focus on a few selected immunotherapies, which attempt restoring tolerance rather than controlling inflammation and/or improving islet function. A few preventive trials have been or are being performed in at-risk individuals, usually identified as first-degree relatives of T1D patients positive for at least two autoantibody specificities against insulin, GAD or IA-2. Some prevention trials attempt to induce antigen-specific tolerance using oral or nasal administration of insulin, which is believed to be a major and early auto-antigen in T1D (Zhang et al. 2008). So far, preventive approaches have yielded disappointing results with little to no efficacy in delaying or preventing T1D (Chaillous et al. 2000; DPT-1-Diabetes-Study-Group 2002; Barker et al. 2007; Nanto-Salonen et al. 2008). Many intervention clinical trials are performed in new-onset diabetic patients during the first few months after clinical diagnosis. This phase of reduced insulin dependency reflects a certain level of islet preservation and can last from a few months to a couple of years in some patients. Other trials enroll patients completely dependent on insulin therapy but must be accompanied by islet replacement strategies such as allogeneic pancreatic islet transplantation (Alejandro et al. 2008; Posselt et al. 2010).
Recently, clinical trials of Fc receptor nonbinding (FNB) anti-CD3 mAbs have proved to be one of the most successful immunotherapies in T1D to date (Herold et al. 2002; Keymeulen et al. 2005). FNB anti-CD3 reagents provide a suboptimal TCR signal to T cells and have powerful tolerogenic properties in animal models (Smith et al. 1997), as evidenced by the reversal of new-onset diabetes in NOD mice (Chatenoud et al. 1994, 1997). Remarkably, a short-course treatment with FNB anti-CD3 mAbs (teplizumab) led to preserved islet function for up to five years in new-onset T1D patients (Herold et al. 2005, 2009). In mice, the tolerogenic effects of FNB anti-CD3 mAbs may rely on the preferential depletion of pathogenic effector T cells whereas Tregs are relatively resistant to depletion (Smith et al. 1998; Penaranda et al. 2011). As a result, the balance of Teff to Tregs is reset to favor immunoregulation because of a higher frequency (if not absolute number) of Tregs (Belghith et al. 2003; Bresson et al. 2006). In humans, FNB anti-CD3 mAbs do not induce massive T-cell depletion but treatment in T1D patients induces a population of regulatory T cells (Bisikirska et al. 2005). Another approach targeting T cells at the population level is using antilymphocyte serum (ALS) or antithymocyte globulin (ATG), two related polyclonal anti-T-cell reagents that induce a profound but transient T-cell depletion. ATG/ALS could induce long-term remission in new-onset diabetic NOD mice via mechanisms that may include preferential repopulation of the T-cell compartment by Tregs (Ogawa et al. 2006; Simon et al. 2008). Treatment of T1D patients with ATG yielded promising results in a small study but side effects associated with cytokine release syndrome were observed (Saudek et al. 2004). An immune tolerance network (ITN)-sponsored phase II clinical trial using a different dosing regimen of ATG has been initiated in a larger number of patients. The efficacy of both ATG and FNB anti-CD3 mAbs in new-onset diabetic NOD mice was enhanced in combination with exendin-4, a reagent that improved the functional recovery of residual islets (Ogawa et al. 2004; Sherry et al. 2007). Concomitant targeting of the autoimmune response and the islet tissue may be important in achieving long-term tolerance and restoration of β-cell function in T1D patients. Alternatively, combination therapies that synergistically alter immune responses are being considered to maximize the tolerogenic outcome (Matthews et al 2010). For example, a phase I trial supported by the ITN and combining IL-2 and rapamycin is currently underway in T1D patients, with the rationale that both reagents will synergize to delete autoreactive effector T cells and favor the expansion or survival of Tregs.
Antigen-specific therapies are an attractive alternative because the expectation is that they could help restore tolerance to islet antigens without inducing global immunosuppression. Antigen-specific clinical trials have been targeting T-cell responses either to known major auto-antigens in T1D (GAD, insulin) or to islet-derived peptides eluted from human MHC class II molecules (Ergun-Longmire et al. 2004; Alleva et al. 2006; Ludvigsson et al. 2008, 2011; Agardh et al. 2009; Thrower et al. 2009; Walter et al. 2009; Gottlieb et al. 2010; Orban et al. 2010). Different approaches have been used, including, theoretically, favoring presentation of auto-antigens by tolerogenic APCs, altering the TCR recognition of its cognate antigen in autoreactive cells and/or the skewing of the T-cell response from pathogenic Th1 cells to protective Th2 cells. So far, some phase I and II clinical trials have shown efficacy in maintaining C-peptide levels and inducing favorable immune parameters such as decreased Th1/Th2 ratio and increased Tregs and TGF-β (Agardh et al. 2005; Alleva et al. 2006; Ludvigsson 2009; Orban et al. 2010; Ludvigsson et al. 2011). Currently available data on the outcome on disease progression have shown either no effect or encouraging results such as improved C-peptide and HbA1C levels and decreased insulin need, and the clinical benefits were but short-lived. Although these results are clearly less dramatic than in animal models, additional studies with higher numbers of patients and different reagents, doses and/or timing of administration will determine whether antigen-specific therapies could be part of the therapeutic arsenal against T1D, alone or in combination with other immunomodulating agents or treatment targeting β-cell survival and function.
Cellular therapy is currently being actively pursued. In preclinical animal models, Tregs have shown a great degree of efficacy in both preventive and curative approaches in many models of autoimmune disease and transplantation. A JDRF-sponsored phase I clinical trial was recently initiated to evaluate autologous Tregs expanded in vitro (Putnam et al. 2009). The rationale for this approach is based on animal models demonstrating an imbalance between Teff and Tregs in autoimmune diabetes and the demonstration that expanded Tregs can suppress disease in NOD mice (Tang et al. 2004). Treg therapy holds great promises in many human diseases but significant challenges are present and will need to be dealt with, including the absence of lineage-specific surface marker that can be used to isolate Tregs in humans; the possibility that effective therapy will require either high numbers of polyclonal Tregs or antigen-specific Tregs, which will both require significant technical prowess from the clinical trial team; and, as mentioned above, the possible occurrence of unstable Tregs that may affect the efficacy of Tregs as immunotherapy and/or the initial disease. Recent trials of autologous nonmyeloablative hematopoietic stem cell transplantation have shown promising results in a small number of new-onset T1D patients with preservation of C-peptide levels and prolonged insulin independence in many patients (Voltarelli et al. 2007; Couri et al. 2009). Finally, spleen cells incubated with a self-antigen or peptide and chemically fixed with ECDI have potent tolerogenic properties in mouse models of autoimmunity (Miller et al. 2007). Antigen-coupled ECDI-treated spleen cells can reverse diabetes in NOD mice and prevent allogeneic islet rejection through intrinsic control of autoreactive T cells by the PD-1 pathway and extrinsic regulation by Tregs (Fife et al. 2006b; Luo et al. 2008). A clinical trial of insulin-coupled autologous PBMCs is currently being developed by the ITN. Finally, other cell therapies are being tested including cord blood and mesenchymal stem cells (Haller et al. 2010).
Briefly, many other clinical trials are being conducted to restore tolerance in T1D patients. Costimulation blockade using CTLA-4Ig (abatacept or belatacept) was recently shown in a phase II clinical trial in T1D patients and after islet transplantation to slow progression of the disease and block rejection, respectively (Orban et al. 2011; Posselt et al. 2010). Of note, like other immunotherapies targeting effector T cells, CTLA-4Ig was already under clinical development at the time Tregs became prominent. The discovery that CD28 signals were critical for Treg homeostasis (Salomon et al. 2000) raised the concern that CTLA-4Ig treatment may deplete the Treg population and have unintended consequences on disease incidence. A recent trial in renal transplantation reported that belatacept did not reduce the frequency of Tregs in the peripheral blood or kidney transplant (Bluestone et al. 2008). However, it will be important to perform immune in addition to clinical monitoring in these clinical trials to determine the outcome of a given reagent at a given dose on effector versus regulatory T cells. In view of the “hygiene hypothesis” and the dramatic tolerogenic effect of CFA in NOD mice, several studies have been carried out in humans using agents that activate the innate immune system. Although previous trials of BCG (Bacillus Calmette-Guerin) administration had no effect on T1D (Elliott et al. 1998; Allen et al. 1999), there are ongoing clinical efforts underway to test the administration of BCG and other toll-like receptor ligand to activate this innate arm of the immune response in the hopes of moderating disease progression.
Considerable strides have been made during the last two decades in the understanding of the mechanisms of peripheral tolerance that are affected in autoimmune diabetes. As described in this article, many steps in the stimulation, function and regulation of autoreactive T cells can be impaired in genetically-susceptible individuals and as a consequence of a given environment. The recent and ongoing discovery of the role of microRNAs and other innovative strategies in fine-tuning immune responses, cellular interactions and decision-taking in the immune system will likely provide important insights in the breakdown of peripheral tolerance in T1D and other autoimmune diseases in the near future. In fact, while tolerogenic immunotherapies have been and are currently being developed with sometimes disappointing results and/or unacceptable side effects, it is conceivable that the microRNA field will produce the next breakthroughs in therapeutic approaches in T1D and bring forward innovative strategies aiming at restoring tolerance by producing incremental or concomitant minute changes at the system level.
The authors wish to thank the members (past and present) of the Bluestone laboratory who have contributed to the studies highlighted in this article. In addition, the authors thank the funders, including NIAID, NIDDK, JDRF, the Swiss National Science Foundation, the Brehm Coalition, and the ADA as well as several companies (Pfizer and Becton-Dickenson), who have made this research possible.
Editors: Jeffrey A. Bluestone, Mark A. Atkinson, and Peter R. Arvan
Additional Perspectives on Type I Diabetes available at www.perspectivesinmedicine.org