Is Type 1 Diabetes “Going Viral”?

Although genetic predisposition to type 1 diabetes (T1D) is a key risk factor, it does not explain the recent and rapid rise in incidence of the disease, particularly among children under the age of 5 (1). Changes in the environment and/or how individuals respond to these changes are now widely believed to be responsible for the recent increases in T1D. While viruses have been hypothesized to have an etiological role in T1D for more than five decades, definitive proof has not yet materialized. Nonetheless, a number of candidate viruses have been proposed, including enteroviruses, rotaviruses, herpesviruses, cytomegalovirus, and endogenous retroviruses (2). The strongest and most clinically significant associations point to enterovirus infection. A recent meta-analysis of 26 studies that assessed enteroviral infection using both molecular and immunological assays showed that compared with nondiabetic control subjects, the likelihood of finding evidence of enterovirus is 10-fold higher in T1D patients and fourfold higher in individuals with diabetes-related autoimmunity (3).

Despite this evidence, important questions remain. These include whether a viral infection is causal in the development of T1D, whether it acts to accelerate the onset of disease in individuals in whom the process of β-cell destruction has already been initiated, or whether it is simply present because individuals with diabetes are more susceptible to infection. Two articles published in this issue of Diabetes help to address these questions. Ferreira et al. (4) and Kallionpää et al. (5) used peripheral blood samples from two longitudinal birth cohorts—the BABYDIET cohort (6) and the Finnish Type 1 Diabetes Prediction and Prevention (DIPP) study (7), respectively. Both studies monitored development of diabetes-related autoantibodies as well as the onset of T1D in “at-risk” populations, thereby enabling access to samples from children before and after development of autoantibodies and before and after disease diagnosis.

Using targeted (4) and genome-wide transcriptomics (5), each group independently identified an interferon (IFN) signature in individuals prior to development of the first clinical signs of diabetes. Ferreira et al. found that the IFN signature correlates with two or more episodes of upper respiratory tract infections, suggesting that the IFN signature is driven by a viral infection (4). Of particular interest was their discovery that the IFN signature is strongest just prior to seroconversion, that it is least evident in control subjects, and that the strength of the signature is intermediate in samples taken postseroconversion and in children who already have clinical T1D. Is this observation indicative of an inefficiently resolved infection in these individuals, which might have spread to the pancreas? If so, it would be consistent with emerging data showing that low-level, persistent enteroviral infections can be detected in the peripheral blood of other recently diagnosed patients (8). Importantly, when IFN signature and low-level infection are considered together with evidence that a low-level, noncytolytic, persistent enteroviral infection of β-cells occurs in pancreata from cohorts in Italy (9), the U.K. (10,11), and the U.S. (network of pancreatic organ donors, nPOD) (11), one can envision a scenario where a primary acute viral infection leads to a secondary long-term persistent pancreatic infection that could ultimately drive the development of disease.

Supporting the case for viral infection, genome-wide association studies have associated variants in the IFIH1 gene with susceptibility and resistance to T1D (12). IFIH1 encodes MDA5, an intracellular sensor of dsRNA, a product of most virus replication cycles including enteroviruses. Activation of MDA5 potently induces the antiviral type I IFN response. Protective variants in IFIH1 in patients who are otherwise genetically susceptible to diabetes exhibit reduced MDA5 expression and function (12). Further, recent data have also shown that mice with elevated expression of MDA5 exhibit increased systemic autoimmune disease (13,14). However, although identification of IFN I transcriptional signatures and increased IFN I expression has implicated a role for IFN I in diabetes in mice, loss of IFN I signaling did not affect diabetes induction (15). This implies that the IFN I signature itself is not causal to the disease process; rather, the signature reflects an ongoing and underlying causal process, i.e., virus infection.

Following recognition of a virus infection, the host senses and responds by inducing the antiviral IFN response (Fig. 1). The normal outcome of this response is activation of a strong effector T-cell population and inhibition of the regulatory T-cell response. Within this T-cell response are T cells specific to both the virus and the infected host cell(s). The transient expression and signaling of the IFN I pathway are at the heart of this defense mechanism because they stimulate the adaptive response by facilitating activation of T cells and upregulating major histocompatibility complex (MHC) molecules. Presentation of antigens by MHC on antigen-presenting cells (APCs) is required to engage and activate appropriate immune cells and, on infected cells, targeting them for destruction by effector cells. Further control over the host-specific or autoimmune cells is tasked in part to the regulatory T-cell response. As the virus is eliminated, less virus is sensed, and the balance of effector and regulatory cells swings back to minimize any overreaction. However, individuals with susceptibility to T1D have genetic variants that include key signaling components involved in the response to virus (e.g., IFIH1 [MDA5], PTPN2) as well as the activation/regulation of immune cells (CTLA4, PTPN22, IL-2R). In these individuals, genetic variability likely contributes to an altered response to a viral infection and a subsequent loss of control of the immune cell balance.

In the pancreata of T1D patients, type I IFN, in particular IFNα, is produced by pancreatic β-cells and correlates with hyperexpression of class I MHC (16). Both observations are likely driven by an inappropriate response to a viral infection—lending support to this scenario. Importantly, hyperexpression of class I MHC, which precedes immune cell infiltration (17), can facilitate enhanced presentation of β-cell antigens to infiltrating immune cells. For those immune cells that are β-cell specific, recognition of their targets in an environment characterized by an unbalanced immune cell response will likely lead to inappropriate activation and regulation. Thus, alterations in virus sensing and subsequent host responses could ultimately trigger β-cell–specific autoimmunity. However, there are still many outstanding questions, including whether the IFN signature in the peripheral blood and pancreas—likely a consequence of circulating APCs—is from the same infection, or if several “viral hits” are required? These considerations also raise the question as to why the incidence of T1D is increasing. Are there more diabetogenic viruses in the environment? Is there a change in the timing of exposure to a particular virus, or could variability in maternal or individual viral exposure histories play a role?

The studies by Ferreira et al. (4) and Kallionpää et al. (5) are the first to demonstrate activation of the innate IFN pathway prior to initiation of islet autoimmunity, and they support previous studies demonstrating an increased frequency of infections prior to development of autoantibodies (18,19). Together, they argue that against the backdrop of a genetic predisposition to autoimmunity viral infections can, in some individuals, be causal in development of T1D.