Type 1 diabetes (T1D) is a persistent autoimmune disorder where immune cells attack and destroy the insulin-producing beta cells of the pancreas. Eventually, once enough beta cell mass is lost, individuals will begin to experience loss of natural blood glucose regulation and become reliant on exogenous administration of insulin. Numerous studies have characterized genetic variance and single nucleotide polymorphisms associated with T1D, which can explain why some individuals are more predisposed than others (1–6). Genome-wide association studies have found that Human leukocyte antigen (HLA) and insulin genes are responsible for a significant portion of the genetic risk for T1D. Additionally, many polymorphisms have been identified within immune-related genes including PTPN22, IFIH1, CTLA4, and IL2RA (5, 6). However, genetic make-up only accounts for part of the equation. After all, the immune system is shaped to an incredible extent by non-heritable forces and instead moulded largely by environmental exposures (7).

An array of exogenous stressors have been associated with precipitating autoimmunity (8). However, understanding exactly how environmental factors contribute to disease pathogenesis is a messy ordeal. Dysbiosis, infection, exposure to dietary antigen, and vitamin D deficiency have all been significantly implicated in altering susceptibility to T1D (9, 10). With such complicated etiology, incorporation of multi-faceted approaches, which take into account the extensive amount of cross-talk that occurs between each of these influences on the host, should be strongly considered in future studies.

Virus infections may be an instigating factor for the gut pathology and dysbiosis that is observed in patients leading up to islet autoimmunity and/or T1D onset ( Figure 1 ). Clinical evidence suggests that diabetic patients experience prolonged enterovirus infections associated with the gut mucosa, resulting in persistent inflammation. Furthermore, patients with islet autoimmunity have increased intestinal permeability, low-grade enteropathy, and a dysbiotic microbiome. Seasonal patterns observed in T1D and other autoimmune disease diagnosis could, at least partially, be explained by seasonal variations in infection (11, 12). In this review, we will examine the known effects of virus infection on the microbiome and gastrointestinal (GI) physiology, and how this modulation may relate to T1D pathogenesis.

Numerous viruses, particularly those associated with the gut, have been connected with T1D pathogenesis including enterovirus, rotavirus, cytomegalovirus, and norovirus (13–17). The enterovirus, coxsackievirus B (CVB), has been the virus most frequently associated with T1D. So much so, that recently there has been movement and discussion towards the necessity to develop a vaccine specific for coxsackievirus to help mitigate the globally increasing rates of T1D (18–21). CVB binds to the coxsackie and adenovirus receptor (CAR), which is highly expressed on the insulin-secreting beta cells in the pancreatic islets (22). Variance in CAR expression has been correlated with increased predisposition for T1D (23). In both human populations and experimental mouse models, infection with enteroviruses has been identified to precede onset of islet autoimmunity (24–26). A recent large-scale study looking at virus shedding in the stool of children found that while those with islet-autoantibodies had no difference in total incidence of infection, they did experience a higher rate of sustained enterovirus B (particularly of CVB serotype) infections, which may be contributing to islet autoimmunity (23). Interestingly, this study also found association of other mammalian viruses including human masadenovirus-C (HAdV-C), which actually correlates with reduced incidence of auto-reactivity. The authors suggest that this may be due to HAdV-C competitively inhibiting CAR engagement or through sustained activation of innate immunity resulting in protection from other strains of virus including enterovirus. Children who developed T1D and islet-specific autoantibodies also have a history of increased incidence of respiratory infections in early adolescence (27). It is unclear, however, if there may be underlying immune differences that cause these populations to have increased susceptibility to both these types of infections and T1D autoimmunity. But, children who experience early loss of B cell tolerance to insulin exhibit weak humoral protection against CVB, whereas those with autoantibodies to the T1D biomarker, glutamic acid decarboxylase (GAD), have competent CVB responses - signifying viral clearance may be altered in individuals with T1D-related autoimmunity (28).

Rotavirus infection in children with a genetic predisposition to T1D is associated with increased islet autoimmunity, signifying that infection may exacerbate autoimmunity and diabetes (14). In non-obese diabetic (NOD) mice, rotavirus infection has also been shown to accelerate onset of T1D (29). However, pre-existing autoimmunity is necessary to accelerate disease onset (29). Thus, rotavirus may likely promote pathogenic events rather than serving as a trigger of diabetes.

Antiviral responses to viruses including CVB can likely have direct effects within the pancreas in precipitating T1D (30–32). While CVB has been shown to impair beta cell function in vitro, evidence suggests that the virus itself does not destroy beta cells through cytopathy (30, 33). Antiviral responses are largely mediated through expression of the three classes of interferon (IFN): type I (IFN-α and IFN-β), type II (IFN-γ) and type III (IFN-λ). Innate viral receptor engagement and ensuing immune pathway activation can have a significant role in T1D initiation and pathogenesis (34, 35). A transient type I IFN signature has been observed preceding islet autoantibody development in genetically-susceptible children, but is lost by the time of diabetes diagnosis (36–38). This IFN signalling may be a significant contributor to the hyperexpression of major histocompatibility complex (MHC) class I, endoplasmic reticulum stress, epigenetic and transcriptional/translational modifications observed in the islet microenvironment prior to T1D development. Recently, researchers were able to detect viral signatures (enteroviral protein and dsRNA) in the islets of autoantibody-positive and recent-onset T1D donors along with increased interferon and microbial stress markers (39). There has also been some suggestion that terminally-deleted viral genomes are able to persist in the islet microenvironment causing inflammation and increased immune cell recruitment (40). While there are existing direct links to virus-causing inflammation and modulation of the immune system within the islet microenvironment, there are likely secondary effects of infection, which are also long-term contributors to disease pathogenesis including microbial dysbiosis.

Comprising of a rich diversity of bacteria, archaea, viruses, fungi, and helminths – the microbiome is a dynamic environment that is constantly shifting. This review primarily focuses on the impact of commensal bacterial communities and later the collective virome. The microbiome has a substantial role in shaping peripheral immune tolerance, activation, migration, and differentiation, as well as local inflammatory responses (41). In response, the immune system is in constant communication to respond to these fluctuations in stimuli (42). Alterations in the microbiome have been heavily implicated in the pathogenesis of T1D (43–45) and genetic risk for T1D autoimmunity even confers differences in the bacterial microbiome (46–48). The intestinal microbiota can exert potent influence on immune homeostasis through the production of various metabolites and particularly short chain fatty acids (SCFAs). Both clinical studies and mouse models have established SCFAs including butyrate, propionate, and acetate as significant factors affecting immune regulation in T1D pathogenesis (43, 49). Metabolite-related dietary patterns have been shown to influence T1D susceptibility and metabolomic alterations precede the development of islet autoantibodies in children (50, 51).

While the human microbiome can be quite heterogenous and studies examining the relationship between the microbiome and diabetes have produced highly variable results, there are some notable microbial hallmarks which have been often identified in individuals with T1D and islet-autoimmunity including: a marked decrease in the diversity of bacteria colonizing the gut, increased abundance of bacteria within the Bacteroides phylum, the loss of Firmicutes, and decreased production of SCFAs among other variances (52–55). However, understanding the effects of perturbations in humans while also controlling for extraneous factors is incredibly difficult. The use of NOD mice as a model for spontaneously developing diabetes has given significant insight into understanding the disease pathogenesis of T1D. While the autoimmunity experienced by NOD mice is not the same as that experienced by humans, it allows the use of environmental and genetic interventions in order to understand how they may impact diabetes development (56). Dysbiosis occurs in both humans and NOD mice prior to disease onset and diabetes incidence can even be predicted in these mice based on sampling from various mucosal microbiomes (43, 57, 58). A “diabetogenic” microbiome from a diabetes-prone NOD mouse is sufficient to promote insulitis when transferred to a non-obese diabetes-resistant (NOR) mouse (59). Typically, female NOD mice are more susceptible to developing autoimmune diabetes than male mice; however, this difference is not observed in germ-free mice (60). This discrepancy in sex bias can at least partially be explained by microbial stimulation of testosterone (60).

Pattern recognition receptors (PRRs) including toll-like receptors (TLRs), RIG-I-like receptors, and caspase activation and recruitment domains (CARDs) are innate sensors that can detect viral and microbial pathogen-associated molecular patterns (PAMPs). Signalling through these receptors can be detrimental for modifying susceptibility to T1D development (34). There are 13 total types of TLRs, each of which is specific for various bacterial (LPS, flagellin, peptidoglycan, etc.) and/or viral (dsRNA, CpG DNA, viral protein, etc.) antigens. Microbiota can regulate T1D through TLR signaling differences (61, 62). For instance, imbalance between TLR2 vs. TLR4 stimulation can determine T1D susceptibility where TLR2 provides a pro-diabetic signal whereas TLR4 provides microbiota-induced tolerization (61). This overlap in bacterial and virus infection immune signaling may signify a role between commensal microbes and virus infection in host immune regulation.

In the first few years of life, colonization of the GI tract plays an indispensable role in shaping host immune development, regulation, and maintenance (63, 64). With age, the microbiome experiences decreasing plasticity and tolerance for new antigen exposure and environmental disruptions (63, 65). Following infancy, the microbiome seems to stabilize with relatively established communities that continue to shape mucosal and systemic immune homeostasis into adulthood (66). Thus, timing of environmental perturbations is likely an important factor for producing dysbiosis, which impacts disease susceptibility. The “Hygiene Hypothesis” proposes that exposure to antigens in early life during immune development can have profound effects for the development of autoimmune and allergic disease later on. Evidence suggests that instigating factors leading to T1D occur early in life – especially since a majority of early-onset individuals who progress to overt T1D before adolescence develop autoantibodies by 3 years of age (1, 67). However, most individuals are diagnosed with T1D in adulthood, hinting that tolerance for environmental stressors may not necessarily be limited to a defined age or that triggering events can occur long before disease onset (68).

Infections that are relatively mild later in life, may have the ability to be quite detrimental early in life at promoting T1D, as the immune system is not yet fully developed and may lack the ability to properly defend the host (69). Viruses cause dysbiosis (70–72), potentially signifying lasting consequences whereby individuals may develop a more autoimmune-skewed microbiome that might be characterized by decreased diversity and less beneficial bacteria (e.g., less butyrate producers). Studies in NOD mice have shown early life exposure to a “diabetogenic microbiome” through fecal microbiome transfers (FMT) can regulate B cell activation and promote T1D onset later on (69). However, when mice are given this same microbiome composition post-adolescence they do not experience the same modulation of the immune system and increased incidence of diabetes autoimmunity. Thus, there may be a unique window, particularly early in life, whereby disruptions in the microbiome from exogenous stressors like infection can have much larger implications on future health.

Studies have found that detection of enteric infection precedes islet autoimmunity by 6 months or more (24, 73, 74). The existing confluence between intestinal inflammation and T1D maintain the gut as an important site for immune modulation that has implications for islet autoreactivity. While some viruses may have deleterious effects on the microbiome, others may actually promote tolerance. For example, norovirus infection was shown to protect from T1D through modulation of the microbiome (75). In this study, Pearson et al. found that NOD mice infected with murine norovirus (MNV) had significantly lower diabetes incidence, less immune infiltration into the islets, increased bacterial diversity, and an increased regulatory rather than inflammatory T cell profile.

Islet-autoreactive CD8⁺ T cells circulate in the blood in approximately the same quantities between healthy and diabetic patients – suggesting that these cells are a normal part of the T cell repertoire (76). However these cells are more abundant in the pancreata of T1D patients, indicating that they must home to the pancreas due to altered immunoregulatory signalling, proinflammatory islet environment, and/or peripheral activation (76). The GI tract plays a fundamental part in communicating between the host and microbiota. Even at healthy steady-state conditions, there is significant T cell trafficking between the gut and pancreatic tissues (77). Existing within this gut-pancreas axis, the pancreatic lymph nodes (pLN), which drain from the duodenum and pancreas, are sampling antigen heavily from both organs. The pLN resides at a critical and significant confluence whereby intestinal stress can alter the presentation of pancreatic self-antigens (77). It has even been suggested that this may be the portal connecting celiac disease with T1D, where GI inflammation due to gluten-sensitivity potentially stimulates anti-islet immune activation in the pancreas (77, 78). Diabetic patients experience prolonged enterovirus infections associated with the gut mucosa, resulting in persistent inflammation (79). This sustained inflammation may be sufficient to result in loss of self-tolerance and T1D development.

Containment of commensal bacteria and dietary antigens within the intestinal lumen relies on several physiological and molecular barriers. The first line of defense is a layer of mucus created by O-linked glycoproteins (mucins) secreted from intestinal goblet cells combined with luminal saccharides. In the colon, a double layer of mucus serves as a physical barrier. The apical layer is typically colonized with various mucus-degrading microbes including those within the Akkermansia family. The innermost mucus layer, however, is predominately uncolonized and creates a largely impenetrable barrier for bacteria. A single cell layer of epithelial cells (IEB) is joined through tight junctions to create a continuous cellular barrier throughout the GI tract. This IEB can be maintained by cytokines including IL-22 produced by group 3 innate lymphoid cells and IL-17A from Th17 lamina propria T cells. Epithelial cells and resident lamina propria immune cells constantly sample the mucosal environment and respond to changes in microbial and viral stimuli. Commensal bacteria populations are regulated through production of antimicrobial peptides (AMPs) and by secreted IgA antibodies. AMPs are bactericidal for specific bugs, particularly within the small intestine where the mucus barrier can be more discontinuous. Colonization of bacteria within the GI tract is also highly regulated by IgA antibodies, which can coat bacteria for neutralization and opsonization. Most secreted IgA is polyreactive and holds an innate specificity to multiple strains of bacteria, but can also undergo somatic hypermutation to produce highly specific IgA against particular bacteria (91, 108).

Autoimmune disorders including T1D, rheumatoid arthritis, multiple sclerosis, and systemic lupus erythematosus (SLE) have all been associated with increases in intestinal permeability - or a so-called “leaky gut” (109–111). Clinical studies have found that individuals with islet autoimmunity experience increased intestinal permeability and low-grade enteropathy (112–116). Loss of integrity occurs prior to T1D development in both human and mouse models, indicating that it may be a significant trigger – rather than a result – of autoimmunity (112, 117). In fact, Sorini et al. found that breaking the intestinal barrier using low-dose dextran-sulfate sodium (DSS) treatments in NOD mice was sufficient to increase onset of autoimmune diabetes (109). This subsequent loss of intestinal integrity can induce activation of islet-specific immune cells in the gut to travel to the pancreas and promote onset of diabetes in T cell receptor-transgenic BDC2.5 crossed NOD mice. Activation of these T cells also appeared to be dependent on the presence of the gut microbiome; however, microbial dysbiosis caused by the DSS-treatment alone was not sufficient to promote autoimmunity.

The intestinal virome is made up of rich and diverse prokaryotic and eukaryotic viral communities, which are shaped by numerous factors including diet, genetics, disease, and geography (140). While a vast majority of the viruses in the body are bacteria-infecting phages, the human virome is also made up of: genomically-integrated human endogenous retroviruses (HERVs); latently-infecting viruses, such as human herpes viruses (HHVs); and potentially persistent/chronic infections – including common enteric viruses previously discussed in this review (CVB, norovirus, rotavirus, etc.) (23, 141). With the GI tract being the most abundant site of viral colonization, the intestinal virome is crucial for maintaining homeostasis and regulating disease pathogenesis through interaction with both commensal bacteria as well as the host (142). Typically germ-free and antibiotic-treated mice face immune dysfunction and altered intestinal morphology. However, infecting these mice with MNV mitigates these aberrations in the intestinal environment (143). Norovirus is therefore sufficient to preserve gut homeostasis and intestinal immunity in a manner that is typically served by microbiota. With potential for such an influential impact, it should be no surprise that alterations and dysbiosis in the viral composition have been associated with several diseases and can alter host immune homeostasis, particularly within mucosal environments (142, 144).

There is a significant degree of bidirectional influence between the microbiome and antiviral response. Not only does infection alter the microbial homeostasis, but the microbiome can also have a significant impact on the outcome of virus infection and the ensuing immunological responses (144, 166–168). The microbiome has been shown to determine severity of viral infection and promote resistance to enteric infection (169–171). Certain species of commensal bacteria can colonize intestinal lymphoid tissues including the Peyer’s patches and mLNs to modify antigen-presenting cell cytokine expression even under healthy homeostatic conditions (172). There is some evidence that microbial antigens may even share sufficient homology to induce cross-reactive T cells against pancreatic targets (76, 122). For instance, an integrase protein expressed by many bacteria within the Bacteroides genus is capable of serving as a low-avidity mimotope of pancreatic autoantigen (173).

Commensal bacteria can aid or limit virus infection through enhanced viral genetic recombination, stabilization of virus particles, promotion of virion dissemination to permissive cells, and modification to immune homeostasis (174). Surface bacterial polysaccharides, such as peptidoglycan and LPS, have been shown to promote virion stability and receptor engagement to increase poliovirus and reovirus infectivity in mice (170). Certain Bifidobacteria and Lactobacillus species have even exhibited an inhibitory potential of CVB4 in vitro (175, 176). Additionally, depleting microbiota through use of antibiotics is able to reduce rotavirus infection by promoting virus-specific humoral responses (177). Infection with H3N2 and H1N1 influenza strains in mice causes intestinal dysbiosis and results in reduced SCFA production and diminished immune responses to secondary infections (135). Conversely, commensal bacteria LPS and extracellular matrix-binding proteins have also been shown to destabilize influenza virions and block infection at mucosal sites, respectively (178, 179).

Understanding how enteric viruses contribute to homeostatic regulation of immunity and may contribute to autoimmune disorders is of great importance. Consequences of virus exposure within the intestinal environment are difficult to determine due to a lack of established animal models and confounding variables including commensal microbes commonly found in murine colonies (e.g., SFB, astrovirus), which may limit viral infection and skew results (150, 186). Ultimately, mice also exhibit differences in viral susceptibility, tropism, and pathogenesis when compared to humans.

Changes in the microbiota have been observed to occur prior to autoimmunity development, which suggests that dysbiosis has a causative role in T1D rather than a result of autoreactive or metabolic pathophysiological responses (58). Intervention studies in humans modulating the microbiome through dietary means or FMT have shown some success in improving T1D outcomes and prevention (187). However, conclusive results in these studies may be limited and require further work.

While there remains much controversy with regards to the precise role and importance of virus infection and the microbiome in determining whether a genetically susceptible individual will lose self-tolerance, significant efforts are being made to understand the patterns and commonalities that are able break through the heterogeneity of human data, background noise, and experimental limitations currently impeding understanding of these issues. Mouse models certainly provide a great deal of potential mechanistic insight into T1D; however, longitudinal human studies integrating clinical data for microbiome differences, infection history, and susceptibility to T1D-related autoimmunity are absolutely necessary to dissect the complicated etiology leading to diabetes development. Blood and stool samples from these large cohort studies can shed light on changes in the microbial, viral, and immunological landscape prior to disease onset. Furthermore, intestinal inflammation and potential increases in gut permeability can be identified by determining abundance of blood markers, clinical tests, and presence of translocated bacterial antigen (188).

Communication between the intestinal microbiota and resident immune populations likely have a profound role in dictating susceptibility and immune system response to virus infection. The intimate inter-relatedness of genetic susceptibility, viral responses, dysbiosis, and host immune state produces an incredibly complex web whereby perturbation can cause a myriad of effects. Understanding the experimental complexity in host-virus-microbe interactions is a monumental challenge. It is difficult to determine which factors and pathways are active contributors to, rather than incidental by-products of, disease. Though challenging, exploring this relationship further is necessary to inform the ultimate prevention, detection, and treatment of autoimmunity.

ZM and MH conceptualized, wrote, and edited the manuscript. ZM created the figures. All authors contributed to the article and approved the submitted version.

The work was funded by grants from the Juvenile Diabetes Research Foundation and GenomeBC.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.