In the last decade there has been a major increase in interest in EVs in their role in cell-to-cell communication, as biomarkers and as therapeutics (7, 8). Many terms and definitions are used to describe EVs, and in this review we use the term EVs to refer to all extracellular, lipid bilayer, sub-cellular particles and their functional contents with sizes ranging from several nm to several μm (9, 10). This umbrella term includes the widely recognized major subgroups termed exosomes, microvesicles, and apoptotic bodies, which are currently distinguishable only by their theorized origin and size but not by experimental means (11). EVs are engaged in cellular communication in both health and disease as transporters of molecular signals in the form of nucleic acids (e.g., DNA, mRNA, microRNA/miRs, long-coding RNA/lcRNA and circular RNA/circRNA), proteins and lipids (11).

When tissue environments are perturbed or cells become damaged as occurs during a viral infection, EV content changes based on cellular reprogramming in response to pathological stress (11, 12). EVs can either activate or inhibit innate and adaptive immune cell responses based on their content (13–15). EVs have been demonstrated to express major histocompatibility complex (MHC) class I or II and directly activate innate antigen presenting cells (APCs) or adaptive T and B cells in an antigen/self-antigen-specific manner (16, 17). Tetraspanins like CD9, CD63 and CD81, which are commonly used to characterize EVs, bind factors on innate immune cells like integrins (i.e., CD11b) that are important in activating and modulating immune responses (12, 18).

Importantly, many viruses use EV cellular machinery (i.e., exosome endosomal sorting complexes required for the transport/ESCRT pathway) for viral transmission such as cytomegalovirus, coxsackievirus, SARS-CoV-2, human immunodeficiency virus 1 (HIV-1), hepatitis viruses B, C and E (HBV, HCV, and HEV), and multiple members of the human herpesvirus (HHV) family (reviewed in (19–22). As a result, EVs can contain infectious virus, viral particles and/or viral proteins following infection that can subvert the immune response to promote viral replication. Important from an autoimmune disease context, a ‘mix’ of self and foreign antigens in/on EVs that may occur after viral infection may be presented to APCs and drive the immune response to target not only the infectious agent but also host antigens resulting in an autoimmune response.

The role of EVs in the development of autoimmune disease has been studied in patients and animal models. A review article by Tian et al. recently examined the role of EVs in a number of autoimmune diseases including thyroiditis, systemic lupus erythematosus (SLE), multiple sclerosis (MS), rheumatoid arthritis (RA), anti-phospholipid syndrome and type I diabetes (23). Many investigators have reported that the number of circulating EVs are elevated in patients with autoimmune disease compared to controls (24–26); however, the wide variety of methods and procedures for isolating EVs as well as differences in storage conditions makes it difficult to interpret these findings.

Studies examining changes in EV content and function may provide a clearer picture of their effects in patients with autoimmune disease. MicroRNA (miRs) content in blood EVs (exosomes) were identified as biomarkers that distinguished patients with relapsing-remitting MS (i.e., miR-15b-5p, miR-451a, miR-30b-5p, miR-342-3p) from those with progressive MS or healthy controls (27, 28). Eight out of the nine miRs that were identified in the study were confirmed in a separate group of patients indicating that the miRs/EVs could serve as biomarkers to predict MS type. Similar results have been found for other autoimmune diseases like type I diabetes (29). EVs have also been found to either promote inflammation/remodeling or to inhibit harmful immune responses for a number of autoimmune diseases including RA (30–33), Hashimoto’s thyroiditis (34), type I diabetes (35), SLE (36), and myocarditis (15). Additionally, EVs have been found with immunoglobulins (Ig) on their surface including IgG or internally in the form of self-antigen-complement-Ig immune complexes (ICs) (37, 38) suggesting that EVs may initiate and/or promote autoimmune damage and inflammation via ICs (39–42). Understanding the role of EVs in autoimmune disease is an emerging field with many questions still to be answered.

For decades, small non-enveloped RNA viruses like CVB3 were thought to cause host cell lysis as the primary method of viral dissemination, but recent evidence has demonstrated that infectious CVB3 and viral particles are released in mitochondrial EVs (47, 56). The first evidence that CVB3 infection disrupts cardiac mitochondria was published in 1964 using young Swiss white mice (Webster strain) inoculated with tissue culture-derived virus (57). Investigators utilized microscopy to assess subcellular changes to the myocardium during viral infection. Notably, they identified an increase in mitochondrial fission, disruption of mitochondrial cristae, and smaller mitochondria with additional membranes that enclosed them that likely depict mitophagosomes (57). A later study from 1997 identified CVB3 localization around and within cardiac mitochondria during myocarditis in mice ( Figure 2 ) (58).

In 2017, Roberta Gottlieb’s laboratory published a study demonstrating CVB3 viral transmission via EVs containing mitochondrial components (47). Using an in vitro neural progenitor stem cell model of viral infection, Robinson et al. demonstrated that CVB3 localizes to the mitochondrial compartment of infected cells and is later ejected from cells in vesicles containing virus, inner and outer mitochondrial membrane components, and autophagy machinery (i.e., microtubule-associated protein light chain 3/LC3-II) (47). Using electron microscopy, they found these particles ranged in size from 100-200 nm in diameter and contained single or multiple virions ( Figure 3 ) (47). They also observed that the ejected particles were infectious to adjacent uninfected host cells.

The protein dynamin-related protein 1 (Drp1) is required for mitochondria to undergo fission. Drs. Gottlieb and Sin showed that CVB3 infection led to Drp1-induced mitochondrial fission resulting in damaged mitochondria being processed into mitophagosomes via mitophagy and released from host HL-1 cardiomyocytes in culture as mitochondrial EVs (48). The role of fission in the production of mitochondrial vesicles was confirmed by inhibition/blocking mitochondrial fission machinery using mitochondrial division inhibitor-1 (Mdivi-1) or direct inhibition of Drp1 with siRNA which resulted in less viral replication and fewer/no virus containing EVs in the culture supernatant (48). Dr. Sin’s group additionally showed that Tank binding kinase I (TBK1) increased phosphorylation of GABA type A receptor associated protein-like (GABARAPL) proteins leading to EVs that contain mitochondria being released from the cells (59). CVB3 infection has also been shown to perturb syntaxin-17 facilitated mitophagosome-lysosomal fusion, which may lead to build up and release of formed mitophagosomes from the cell (60). Thus, these studies confirm that CVB3 localizes to mitochondria and is released in mitochondrial vesicles. Further research is needed to better understand the molecular mechanisms of intracellular mitophagosome formation in the context of viral infections to determine how viruses take advantage of mitochondrial compartments and evade intracellular degradation by targeted autophagy. A summary of our current understanding of CVB3-mitochondria interaction and the development of EV populations containing mitochondria and virus is illustrated in Figure 4 .

Dr. Peter Schultheiss, a major contributor to the fields of cardiology and myocarditis, began identifying and characterizing autoimmune antibodies in patients with myocarditis in the 1970s. In 1978, Bolte and Schultheiss reported that 76% of 17 patients with viral myocarditis had autoantibodies in sera and 41% of these were anti-nuclear antibodies (72). They went on to show that autoantibodies against the adenine nucleotide transporter (ANT), which is a component of the inner mitochondrial membrane, were elevated in patients with myocarditis and dilated cardiomyopathy (DCM) (73, 74). Myocarditis progresses to DCM in susceptible patients and animal models (75). Patients with suspected or confirmed viral myocarditis or cardiomyopathy/DCM had the highest reactivity to anti-mitochondrial antigen and highest expression of anti-mitochondrial antibodies (74). Further analysis found that the sera had uniquely specific reactivity toward cardiac mitochondrial antigen compared to liver or kidney mitochondrial antigen (76).

Another group independently reported that patients with various cardiomyopathies including myocarditis had autoantibodies reactive against mitochondrial proteins (77). They found that 13% with acute myocarditis, 31% of patients with DCM, and 33% with hypertrophic cardiomyopathy generated antibody responses specifically to the M7 antigen of the mitochondrial enzyme sarcosine dehydrogenase, and 25% of these reacted against the cardiac-specific form of the mitochondrial antigen (77). Another group observed autoantibodies against ANT in patients diagnosed with myocarditis or DCM (78). They also observed cardiac-specific reactivity and suggested mitochondrial autoimmune activity as a potential mechanism for the development of DCM following acute myocarditis (78). This agrees with our current understanding of the development of DCM following acute myocarditis (75, 79, 80).

Although viral-induced myocarditis is often categorized as a distinct condition from autoimmune myocarditis clinically and in animal models, the distinction between the two conditions is not clear-cut because patients with viral myocarditis and mouse models of viral myocarditis have been demonstrated to develop autoantibodies and autoreactive T and B cells against cardiac myosin and other self-antigens including mitochondria (2, 39, 79, 81–83).

Importantly, a study examining autoantibody levels that compared experimental autoimmune myocarditis (EAM) to CVB3-induced myocarditis in mice found that ANT was only produced after viral infection but not in EAM suggesting that viral infection was necessary for the production of mitochondrial autoantibodies whereas both models produced autoantibodies against cardiac myosin (84). Additionally, Lin et al. showed that depletion of Drp1 (required for fission) in mice using the mitochondrial fission inhibitor Mdivi-1 reduced CVB3 myocarditis and restored mitochondrial function in the heart (71) suggesting that mitochondrial EVs containing virus may increase myocarditis, although they did not examine this in the study. Overall, these findings suggest that viral infection may be an important mechanism to produce mitochondrial autoantibodies found in patients with autoimmune diseases.

Anti-mitochondrial antibodies (for example, antibodies that target cardiolipin, mitofusin 1, mitochondrial DNA or mitochondrial RNA) are commonly found in patients with rheumatic autoimmune diseases such as RA, SLE, and anti-phospholipid syndrome (85–87). Mobarrez et al. reports most larger EVs (0.7 - 3.0 μm) found in SLE patients contain functional mitochondrial components, as indicated by the presence of the translocase of outer mitochondrial membrane 20 (TOMM20) and hexokinase1 (25). Elevated levels of these type of EVs containing mitochondria are positively associated with increased SLE disease activity, proinflammatory cytokines, and anti-dsDNA antibodies, suggesting that these EVs may be involved in disease pathogenesis (25). Becker et al. recently reviewed the mechanism of immune activation leading to autoimmune disease by mitochondria in these rheumatic conditions but does not discuss the potential role of viral infections in the process or whether the mitochondrial EVs also contain virus or viral components (85). These findings suggest that damage to mitochondria resulting in autoimmune responses may be a common mechanism in the pathogenesis of many autoimmune diseases.

One possible mechanism where myocarditis and other autoimmune diseases could be induced and/or exacerbated by mitochondrial EVs is by activation of Toll-like receptor (TLR)4, interleukin (IL)-1β and leucin-rich repeat (LRR)-containing protein (NLRP)3, which is a pathway that has been demonstrated to increase myocarditis and viral replication in CVB3 models of myocarditis (88, 89). Mitochondria are known to be immunogenic and the TLR4/NLRP3 signaling pathway can be activated by mitochondrial components such as cytochrome c, mitochondrial transcription factor A (TFAM), ATP and cardiolipin, which can all be found in mitochondrial EVs, to initiate a proinflammatory and profibrotic immune response (90–94). Mitochondrial and viral antigens may be expressed on the cell surface or interior of EVs and activate APCs via MHC class II presentation, TLRs or be processed for presentation after the EV lipid membrane has ‘merged’ with an APC (23).

However, not all mitochondria found in EVs stimulate the innate immune response. In some cases, healthy mitochondria within EVs are found to fuse with recipient mitochondria in cultured cardiomyocytes and in a mouse model of myocardial infarction where they improve mitochondrial function and disease (95). These investigators showed a similar improvement in doxorubicin-induced toxicity in cultured cardiomyocytes (96). Thus, transfer of healthy intact mitochondria within EVs represents a novel and potentially viable therapy for patients with mitochondrial damage or dysfunction (97, 98).

AIRE is a transcriptional regulator that protects against self-reactivity by inducing the production of tissue-specific antigens normally not expressed in the thymus, a process that occurs in medullary thymic epithelial cells (mTECs) (99). Resident dendritic cells of the thymus take up self-proteins and present them to T cells. If reactivity to self-antigen is too strong, dendritic cells undergo cytokine signalling programs that destroy autoreactive T cells (99). Migratory and peripheral dendritic cell populations further contribute to negative selection of autoreactive T cells by selecting against cells reactive to peripheral antigen from other tissue microenvironments. Migratory dendritic cells take up antigen in their respective resident tissues and travel to the thymus whereas peripheral dendritic cells test autoreactive T cells in the periphery (i.e., their tissue of origin) thereby inducing, depending on the conditions, cell deletion, anergy or polarization toward a regulatory phenotype (100). It is estimated that there are 1,140 murine genes that interact with or localize to mitochondrial compartments (101), so AIRE should protect the host from developing autoimmunity against these mitochondrial antigens. One important question is whether mitochondrial genes represent a gap in the normal negative selection criteria in the thymus.

To our knowledge, no studies examining the function of AIRE describe its ability to produce mitochondrial self-antigen. Two major studies exist with publically available data of AIRE genomic binding and related expression (102, 103). A study by Bansal et al. reported murine transcriptomic data that examined AIRE binding using chromatin immunoprecipitation (ChIP) sequencing (ChIPseq). We examined whether any of the antigens that they reported for AIRE were directed against mitochondrial antigens using their published transcript-level data. They did not have protein/proteomic level data available to assess this question. They estimated AIREs coverage of genes by assessing transcription among AIRE knockout (KO) mice versus wild type (WT) controls in data derived from mouse mTECs (103). A negative log fold-change (LogFC) indicated downregulation of the transcript in the AIRE KO mice, suggesting that in WT conditions, AIRE may be responsible, in part, for regulating the transcription of a respective gene. To determine AIRE regulation of mitochondrial related genes, we performed a keyword search for “mitochondri” (which yielded results for mitochondrial, mitochondria, and mitochondrion) in the gene description column of their dataset, which yielded 315 genes. Among these, 20 were significantly downregulated at an FDR p ≤ 0.05 comparing AIRE WT to KO indicating that AIRE may regulate only 6.3% of the 315 mitochondrial related genes. We also assessed the potential role of AIRE to regulate 85 murine nuclear-encoded mitochondrial respiratory chain genes using keyword searches in the gene name column for “nduf,” “sdh,” “cox,” “uqcc,” and “atp,” which are the prefixes for gene names among components of respiratory complexes I-V, respectively. Their data showed a significant downregulation of 2 of 85 (2.3%) nuclear encoded mitochondrial respiratory chain genes in AIRE KO vs WT samples. Mitochondrial genes that were regulated by AIRE are listed in Table 1 . Thus, only a small percentage of potential mitochondrial genes were regulated by AIRE using this method. More research is needed to better understand whether AIRE contributes to tolerance against mitochondrial antigens. Thus, a lack of mitochondrial tolerance may be one possible explanation for the development of mitochondrial targeted autoimmune responses in myocarditis and other autoimmune diseases.