Abstract
Significance:
Type 1 diabetes (T1D) is characterized by the autoimmune destruction of the insulin secreting β-cells, with consequent aberrant blood glucose levels. Hyperglycemia is the common denominator for most of the chronic diabetic vascular complications, which represent the main cause of life reduction in T1D patients. For this disease, three interlaced medical needs remain: understanding the underlying mechanisms involved in pancreatic β-cell loss; identifying biomarkers able to predict T1D progression and its related complications; recognizing novel therapeutic targets.
Recent Advances:
Extracellular vesicles (EVs), released by most cell types, were discovered to contain a plethora of different molecules (including microRNAs) with regulatory properties, which are emerging as mediators of cell-to-cell communication at the paracrine and endocrine level. Recent knowledge suggests that EVs may act as pathogenic factors, and be developed into disease biomarkers and therapeutic targets in the context of several human diseases.
Critical Issues:
EVs have been recently shown to sustain a dysregulated cellular crosstalk able to exacerbate the autoimmune response in the pancreatic islets of T1D; moreover, EVs were shown to be able to monitor and/or predict the progression of T1D and the insurgence of vasculopathies.
Future Directions:
More mechanistic studies are needed to investigate whether the dysregulation of EVs in T1D patients is solely reflecting the progression of diabetes and related complications, or EVs also directly participate in the disease process, thus pointing to a potential use of EVs as therapeutic targets/tools in T1D. Antioxid. Redox Signal. 36, 631–651.
Introduction
Type 1
The lymphocytic infiltration of the islets of Langerhans, first observed in a 10-year old who died with ketoacidosis in 1902, is composed of CD8+ and CD4+ T cells, B lymphocytes, and macrophages and consistently parallels the presence of circulating autoantibodies directed against islet cell antigens (102, 114, 209). This infiltration, termed “insulitis,” stands at the core of the pathogenic process that leads to loss of insulin secretion capability. To either slow down or reverse T1D-related insulitis progression, three fundamental medical needs remain. On the one hand, we necessitate to understand the underlying cellular and molecular mechanisms involved in pancreatic β-cell loss: in particular, the characterization of the dysregulated crosstalk between β-cells and the infiltrating immune cells is crucial to design novel models of disease prevention and/or cure. On the other hand, the management of the patient with T1D may sensibly profit from the identification of tracable biomarkers able to predict the rate of disease progression and the discovery of efficacious therapeutic approaches.
Extracellular vesicles (EVs) have recently been investigated for possessing a paramount potential in both these areas of investigation: they are emerging as mediators of the crosstalk among different organs and tissues at the paracrine and endocrine level, and as potentially useful biological entities able to reflect pathological processes (210). EVs can be classified based on their size: small EVs (20–100 nm) are formed by the inward budding and subsequent fusion to the plasma membrane of multivesicular endosomes; larger EVs (usually ranging from 200 nm to 1 μm) can directly bud from the plasma membrane, and similar biogenesis is described for apoptotic bodies (0.5–2 μm) (1, 64, 162, 185). Since no consensus has been reached on specific markers for EV subtypes, such as endosome-originated “exosomes” and plasma membrane-derived “ectosomes” (or microparticles/microvesicles), the International Society for Extracellular Vesicles (ISEV) keeps endorsing the generic name “EVs” for all the particles with the following characteristics: (i) naturally released from the cell; (ii) delimited by a lipid bilayer; (iii) unable to replicate (i.e., devoid of a functional nucleus); and (iv) uncertain biogenesis (i.e., all analyzed vesicles, unless caught during release by live imaging techniques) (194).
In this review, we have thus uniformed the nomenclature from the different studies by the term EVs and summarized the major advancements suggesting EVs and their associated molecules as pathogenic factors, promising disease biomarkers and therapeutic targets in the context of T1D and its complications.
T1D and Related Complications
The analysis of rodent models of T1D has been instrumental in identifying the cellular and molecular pathways that lead to pancreatic β-cell destruction; yet we are uncertain in defining the initiating momentum for the human disease (165). The triggers of the inflammatory phenotype are still matter of investigation, although both pancreas-resident antigen presenting cells (APCs, macrophages, and/or dendritic cells [DCs]) and T cells are required for T1D initiation (Fig. 1A) (93, 192). In particular, APCs are held to release chemokines and proinflammatory cytokines able to recruit lymphocytes (CD4+, CD8+ T, and B cells) into the islets (5). Oxidative stress and altered redox pathways have been suggested to contribute to T1D pathogenesis: superoxide anion (O2 •−) production by T cells and/or macrophages is key in causing β-cell death, and its clearance is able to delay diabetes onset, as shown in nonobese diabetic (NOD) mice (141, 192). In addition, the production of ROS by macrophages in the islet microenvironment drives the proinflammatory M1 macrophage phenotype, and promotes the recruitment of CD4+ and CD8+ T cells (192). The vascular endothelium, in response to increased levels of ROS and inflammatory cytokines, upregulates both intracellular adhesion molecule 1 (ICAM-1) and P-selectin, which allow T cell adhesion and extravasation (125). The abnormal production of interferon (IFN)-γ by infiltrating T cells is also associated with an increase in mitochondrial ROS and a decrease in their respiratory capacity (29, 30). Moreover, ROS production is required to preserve optimal lysosome conditions for autoantigen processing and presentation (30, 140). ROS levels were found elevated in cytotoxic T cells from T1D patients as a consequence of the dysfunction of a T regulatory (Treg) cell population (189). Once activated, CD8+ T cells can cause β-cell loss through the direct linkage of FAS–FAS ligand and the release of perforin and granzymes, which mediate the cleavage of BH3 Interacting Domain Death Agonist (BID) in β-cells (40, 52, 58, 205). Similarly, CD4+ T cells can also augment pancreatic tissue deterioration through the secretion of multiple cytokines, for example, tumor necrosis factor-α (TNF-α), interleukin (IL)-6, and IFN-γ, finally promoting β-cell apoptosis (187).

Pancreatic β-cells themselves are capable of producing proinflammatory cytokines, especially in response to nutrient overload and the consequent O2 •− generation, thus amplifying the inflammatory cascade (133).
Microvascular complications
Microvascular complications, that is, nephropathy, retinopathy, and neuropathy, represent a considerable source of morbidity in people with T1D (Fig. 1B). Recent estimates suggest that ∼70% of the patients with longstanding T1D will experience at least one microvascular complication during the course of their life (2). Although involving different organs, chronic hyperglycemia is the major trigger of all microvascular complications, as demonstrated by multiple follow-ups of the Diabetes Control and Complications Trial (DCCT)/Epidemiology of Diabetes Interventions and Complications (EDIC) study (130). Initial mechanistic experiments pointed at a role for glucose overload in causing mitochondrial dysregulation and ROS accumulation, which in turn promote cellular damage either directly or by fostering a number of pathways, especially in those cells unable to regulate glucose influx such as endothelial cells (ECs) (83, 207). Indeed, the increase in intracellular ROS, and in particular O2 •−, is an early event able to activate several downstream pathways: (i) different isoforms of protein kinase C (PKC), which are in turn involved in cell proliferation, activation of nuclear factor-κB (NF-κB), and alteration of EC permeability (70); (ii) augmented influx of sugars belonging to polyol pathway, such as sorbitol, taking away NADPH from the cytosol, necessary to keep reduced glutathione (GSH), the key antioxidant factor (8); (iii) hexosamine pathway, involved in reduced pericyte proliferation and EC apoptosis (129); (iv) formation of advanced glycation end (AGE) products, which further promote molecular and cellular damage (217). Of note, many of these alterations are long lasting, providing a rationale to the concept of metabolic memory; that is, the benefit provided by an early and intensive glycemic control, which is particularly relevant for microvascular complications (20, 26, 149).
More recently, epigenetic mechanisms have emerged as additional mediators of microvascular complications (154, 191). Alterations in chromatin remodeling are observed days after exposure to hyperglycemia and return to normoglycemic conditions, as shown in both animal models and patient-derived samples. Of note, many of the genes resulting as chronically upregulated encode proinflammatory molecules, for example, heme oxygenase 1 (HMOX1), IL-8, matrix metalloproteinases-10 (MMP-10), TNF-α, and cyclo-oxygenase-2 (COX-2) (38). As a result, initial exposure to glucose overload is able to chronically promote low-grade inflammation (LGI) through a complex pro-oxidant response altering a number of chromatin-regulating enzymes (38, 153). LGI has been reported in different tissues in response to hyperglycemic stimuli, and it can be mediated by a number of different cell types (152). Beyond epigenetic alterations, hyperglycemia has been suggested to directly activate proinflammatory gene expression through: (i) the indirect activation of AGE receptors (RAGE), which promotes NF-κB translocation; (ii) the upregulated expression of toll-like receptor (TLR) 4 and other pattern recognition receptors (PRRs); and (iii) the accumulation of senescent cells (41, 150, 160, 178). Many of these mechanisms are at the basis of microvascular complications, that is, retinopathy (160, 178), neuropathy (18, 23, 99), and nephropathy (50, 51, 92, 131, 135, 188, 204, 216). Of note, both the antioxidant and the anti-inflammatory approaches for the prevention of microvascular complications have been translated into clinical settings (46, 226).
Macrovascular complications
The major causes of morbidity and premature mortality in T1D subjects are CVDs and accelerated atherosclerosis, a chronic condition characterized by the accumulation of lipid-rich plaques in large vessels, which is the common pathogenic phenomenon in the development of coronary artery disease, cerebrovascular disease, and peripheral arterial disease (Fig. 1B) (72, 106, 116). Different cell types take part in plaque formation, among which the most relevant are ECs, vascular smooth muscle cells (VSMCs), and immune cells. LGI and oxidative stress are key phenomena driving all stages of plaque formation, development, and eventually rupture. Endothelial dysfunction, recruitment and infiltration of macrophages, T cells, and of other immune cells, lipid oxidation and accumulation, foam cell formation, smooth muscle cell proliferation, intima-media thickening, and plaque destabilization are all promoted by soluble and cellular inflammatory mediators and pro-oxidant triggers (110). While the control of multiple risk factors, for example, blood pressure and lipids, is the main approach to ameliorate atherosclerosis in virtually all subjects, a tight glycemic control is the most powerful approach to reduce the incidence of CVDs in patients with T1D (14, 186). However, optimal glycemic control (i.e., HbA1c <7%) reduces, but does not suppress, the CVD risk for patients with T1D (111). Additional canonical and noncanonical risk factors might be considered (16). The mechanisms described above for microvascular diseases are likely relevant also for the development of macrovascular complications of T1D, especially considering the chronic inflammatory and pro-oxidant status characterizing patients with this condition (47). However, pro-oxidant and proinflammatory pathways are only marginally targeted by available therapies, likely due to the complexity and redundancy of the enduring alterations underlying the development of complications (25, 148).
EVs: Content and Biological Role
At the beginning of the 1980s, during studies on transferrin receptor turnover in reticulocytes, researchers from two independent laboratories stumbled upon very small vesicles released by those cells during maturation (80, 143). Initially, it was hypothesized that vesicle release univocally represented a mechanism of molecule disposal and, consequently, EVs were initially regarded as mere garbage bins. More than a decade later, though, Raposo et al. revealed that EVs released by B cells contain peptide–MHC class II complexes and have thus the ability to perform antigen presentation, opening the road to the discovery of EV biological functions (161).
The stability of EV structure is ensured by a double layer phospholipid membrane, enriched in lipid rafts; EV lipid composition closely resembles that of the donor cell and is characterized by a variety of molecules, including sphingomyelin, cholesterol, ganglioside GM3, unsaturated lipids and ceramide, prostaglandins and leukotrienes (113, 164, 197). EV-associated proteins comprise those involved in biogenesis and release machinery, signal transduction, the endosomal compartment and tetraspanins (CD81, CD9, and CD63), while proteins associated with the endoplasmic reticulum (ER), Golgi apparatus, or nucleus are depleted in EVs (193). In 2007, EVs were discovered to also contain RNA, including messenger RNAs (mRNAs) and small RNAs such as microRNAs (miRNAs) but also long noncoding RNA, circular RNA, Y RNA, piwi interacting RNA, and vault RNA, while EVs seem to be devoid of ribosomal RNA and DNA (85, 90, 201). The main molecules associated with EVs are depicted in Figure 2A.

In particular, miRNAs are small (22 nucleotides length) highly conserved noncoding RNAs with a very well-characterized intracellular biogenesis (Fig. 2B) (82). They inhibit gene expression in a post-transcriptional manner, by recognizing a specific complementary sequence usually located in the 3′ untranslated region (UTR) of a target mRNA and decreasing its translation rate, and are regarded as key players of gene expression regulation in eukaryotic cells (9, 10, 19, 62, 82, 199, 214). Relevant to the present topic, miRNAs were described to regulate not only the differentiation, but also the survival, the proliferation, and the insulin secretion capability of β-cells (56, 76).
Upon EV-mediated transfer from one cell to another, miRNAs can regulate gene expression, revealing the potential of this “EV shuttle RNA” to deliver specific signals in an autocrine, paracrine, and also endocrine manner (Fig. 2C) (201). Elegant in vivo studies have provided with proof of distant tissue regulation of gene expression by EV-associated miRNAs, with detectable effects on diversified processes such as tumor metastatic progression and metabolic homeostasis (195, 218, 227).
In conclusion, after 25 years of EV investigation, we now acknowledge that most cells release EVs, and that they play several relevant and diversified roles of intercellular communication in the immune system and beyond (107, 117). EV release into the extracellular milieu represents an additional mechanism of plasticity to fine tune specific physiological cellular responses, but it may also become dysregulated and concur to promote and/or enhance pathological conditions (45, 79).
Immune Regulatory Roles of EVs
Starting from the milestone discovery of the EV-mediated antigen presentation (161), a significant amount of work has been devoted to characterize the biological functions of EVs in the immune system. The release of EVs by B cells can be activated by multiple stimuli, such as B cell receptor (BCR) priming and CD40/IL-4 signaling, and in turn stimulates primed CD4+ T cells (3). Upon the formation of the immunological synapse, T cells at the interface with the APCs release EVs, containing an activation-competent T cell receptor (TCR) able to keep the T-APC cellular crosstalk in a contact-independent manner (17, 34). DCs constitutively secrete EVs, but this phenomenon significantly increases upon cell activation; moreover, EVs from mature DCs can be internalized by bystander immature cells and promote their antigen presentation capability (127, 137). In particular, DC-derived EVs carry several costimulatory and adhesion molecules, including ICAM-1, important for antigen-specific T cell priming and antibody responses (37, 177). Similarly, EVs released by TCR-stimulated T cells promoted proliferation in autologous resting CD3+ T cells (206). If on the one side EVs can propagate immune activation, on the other, they may also curb the response by down-modulating immune cell function, suggesting their involvement in the regulation of immune tolerance. For example, EVs derived by CD8+ T cells are able to induce DC apoptosis by exposing FAS ligand on the surface and suppress monocyte activation, by transfer of anti-inflammatory miR-150 (132, 215). EVs have the ability to inhibit effector T cell responses, establishing antigen-specific and allograft tolerance and delaying rejection (97, 144). One potential mechanism of immune suppression is connected to CD4+ Treg cell-derived EVs, containing CD73, which hydrolyzes exogenous 5′-adenosine monophosphate (AMP) and produces adenosine, activates intracellular cAMP production and hampers effector T cell responses (167, 182). In addition, Treg cell-derived EVs contain high quantities of miR-146a-5p able to down modulate target genes crucial for cellular activation, such as signal transducer and activator of transcription (STAT) and IL-1 receptor-associated kinase (IRAK), in effector T cells (122, 196). Moreover, EVs have been shown to participate in the shuttling of cytokines, conferring them higher stability and thus sensibly changing their biodistribution and biological effect (22, 78, 96, 118, 158, 219, 221).
In all, these results have pointed to an active participation of EVs and their molecular cargo in influencing the immune response and the inflammatory milieu. Accumulating evidence has revealed that the participation of EVs in the formation of immune complexes and autoantigen presentation, their contribution to inflammatory and thrombotic phenomena and vascular dysfunction, all concur to implicate EVs in the pathophysiology of autoimmune diseases, including T1D (200). EV potential as cell-to-cell communication mediators and potential drug targets to modulate still unrecognized pathways driving diabetes and its related micro- and macrovascular complications is actually believed to be conspicuous.
EVs and Associated miRNAs: Involvement in T1D Pathogenesis and Use as Biomarkers
In NOD mice, T cells start to infiltrate the pancreatic islets at a disease stage in which there is still a normal glucose control (∼2 months of life): at this very moment, β-cells were found to significantly upregulate several miRNAs that are normally present at extremely low levels, that is, miR-142-3p/-5p and miR-155, and instead are known to be highly expressed in T cells and their released EVs (43, 77, 166). Notably, this miRNA increment was not caused by endogenous novel transcription of the miRNA-containing genes, but by the actual passage of the mature miRNAs from the T- to the β-cells, as elegantly demonstrated by Guay et al. (77). The specific transfer of miR-142-3p/-5p and miR-155 into β-cells did not modify the secretory capacity of β-cells but altered NF-κB activity, the cytokine and chemokine signaling (including release of C-C motif chemokine ligand 2 and 7, CCL-2 and -7 and C-X-C motif chemokine ligand 10, CXCL10); eventually causing β-cell death (77). Furthermore, NOD mice injected with an adeno-associated virus containing an insulin promoter-driven miRNA sponge for miR-142-3p/-5p and miR-155 displayed low grade of insulitis, increased Treg cell infiltration into the islets and, compared with control injected mice, were protected from developing the disease (77). These results have provided experimental evidence that once the pancreatic islets are infiltrated, T cells damage the tissue and promote β-cell loss also by the previously uncharacterized release of EVs with a specific miRNA content, possessing pathophysiological relevance. Immunohistological analyses of NOD mice pancreata showed that endoglin (CD105)+ mesenchymal stem cells (MSCs) penetrating into the β-cell area as lymphocyte infiltration occurs start releasing highly immune-stimulatory EVs, which accelerated the effector T cell-mediated destruction of islets, functioning as a further autoimmune trigger in NOD mice (159).
EVs released by β-cells have been shown to play an additional pathogenetic function in the context of T1D. The characterization of EVs released by insulin secreting rodent cell lines (insulinomas) in response to either high glucose or cytokine stimulation led to the identification of a plethora of EV-associated molecules, including chaperones, cytoskeletal constituents, membrane transporters/ion channels, signaling molecules, nucleic acid binding proteins, and two inflammatory mediators, TNF receptor 1 (TNFR1) and ICAM-1 (105, 142). The strong innate stimuli associated with insulinoma-derived EVs were shown to elicit TLR signal-mediated proliferation of splenic B cells and T cell reactivity in pancreatic lymph nodes from prediabetic NOD, but not diabetic-resistant mice, suggesting that β-cell-derived EVs may contribute to the development of autoimmunity in NOD mice by augmenting autoreactivity of both B and T cells (13, 180). In the human setting, β-cell-derived EVs, differentially represented in the blood of T1D patients compared with healthy controls, showed the ability to activate the TLR7/8 signaling cascade and increase activation as well as cytotoxicity of the blood circulating immune cells, in turn promoting their infiltration in the pancreatic tissue and enhancing local inflammation (190).
The capacity of β-cell-derived EVs to trigger the insulitis may in part depend on becoming passive autoimmune targets. The detrimental ability of EVs to protect self-antigens from degradation and spread them also at long distances was actually described in other autoimmune diseases, such as systemic lupus erythematosus (53, 139). Rat and human pancreatic islets were indeed described to release EVs containing intracellular β-cell autoantigens, such as GAD65, IA-2, and proinsulin, ZNT8, and β-cell-resident glucose transporter 2 (Glut2) (36, 81). These EVs can be internalized by DCs and activate them; importantly, cytokine-induced endoplasmic reticulum stress is able to enhance the secretion of autoantigen-containing EVs by β-cells, suggesting the existence of a positive loop exacerbating local inflammation (36). Human islet EVs were also internalized by peripheral blood mononuclear cells from subjects with T1D and (i) led to proinflammatory cytokine expression by CD14+ monocytes; (ii) triggered T cell proliferation and activation; and (iii) elicited GAD65 antibody production, suggesting important EV implications in both innate and antigen-specific immunity (168). EV pathogenic roles in T1D progression are summarized in Figure 3 (upper left panel) and the hypothesized EV-associated miRNA paracrine functions in Table 1.

Examples of MicroRNAs Released by the Indicated Cell Types, Their Target Cell Type, or Tissue and Possible Biological Paracrine Function
EV, extracellular vesicle; IRAK, IL-1 receptor-associated kinase; miRNA, microRNA; NF-κB, nuclear factor-κB; ROS, reactive oxygen species; STAT, signal transducer and activator of transcription; Treg, T regulatory.
Independently of the pathophysiological relevance that EVs may display at the tissue paracrine level, their presence in blood has turned EVs and EV-associated miRNAs into promising biomarkers to monitor human diseases; dysregulation of blood circulating extracellular miRNAs has been indeed found to associate with T1D and related complications (4, 6, 11, 57, 68, 103, 104, 112, 119, 120, 123, 124, 128, 134, 138, 145, 171, 173, 174, 220) (Table 2). One reason for this dysregulation may be the passive release in the extracellular space of intracellular miRNAs upon cell death: as a relevant example, the uptick of blood circulating miR-375-3p—specifically expressed in pancreatic cells—has been incontrovertibly associated with islet damage (63). Another reason for blood circulating miRNA dysregulation does not refer to cell rupture but instead to a differential EV-associated molecular cargo. Cytokine-treated β-cell lines in vitro showed a specific EV enrichment of miR-21-5p not correlated with a parallel increase of miR-21-5p expression at the intracellular level; consistently, EV-associated miR-21-5p, but not that present in total serum, was significantly increased in blood of T1D subjects compared with nondiabetic individuals (104).
Potential MicroRNA Biomarkers, Associated with Type 1 Diabetes Pathogenesis and/or Diabetic Complication
T1D, type 1 diabetes.
Beside EV-miRNAs originating from β-cells, an additional source of EV-associated miRNAs in blood is represented by the aberrantly activated lymphocytes. MiRNAs enriched in EVs released by proinflammatory T helper (Th)1/Th17 cells—such as miR-106a-5p—were found elevated in serum of patients with psoriasis and returned to control level upon effective treatment with etanercept, a biological drug targeting the TNF pathway and suppressing inflammation; on the contrary, miRNAs enriched in Treg cell-derived EVs, such as miR-146a-5p, were instead increased after the same treatment, suggesting that EV-miRNAs may monitor the inflammatory state of autoimmune patients and the response to anti-inflammatory drugs (196). The contemporary monitoring of T1D progression through blood EV-miRNA quantification may have only a very limited clinical advantage, though the capability to predict disease trajectory at onset may substantially help the clinical management for the single patient. After metabolic stabilization, the preservation over time of endogenous insulin secretion—although typically low—is key since it is associated with less severe diabetes-related complications at later disease stages (184). The quantity of blood circulating members of the miR-23 ∼ 27 ∼ 24 clusters was found to possess the capability to predict the rate of insulin secretion capability loss in children at onset with T1D (67). Moreover, these miRNAs have been previously associated with pancreatic β-cell dysregulation, the unrestricted expansion of diabetogenic cytotoxic T cells and T helper cytokine secretion and the development of diabetic complications (33, 44, 61, 75, 98, 147, 157, 196, 202, 220, 223). Hence, understanding whether overexpressed plasmatic miR-23 ∼ 27 ∼ 24 clusters in T1D subjects are simply a mirror of the T1D inflammatory network, or instead directly contribute, upon uptake, to pathogenic dysregulation of β-cells and immune cells alike is warranted.
Finally, EVs have been proposed as novel therapeutics in the amelioration of the autoimmune attack occurring in T1D pancreatic islets. Although regeneration of β-cells is scanty, bone marrow (BM) transplantation is known to promote their regeneration through still uncharacterized mechanisms. Upon BM transplantation, specific miRNAs were found increased in murine serum circulating small EVs: two of these miRNAs (miR-106b-5p and miR-222-3p) were released by BM cells and increased in islet cells and relevant for transplantation-induced β-cell regeneration. Intravenous administration of the corresponding miRNA mimics was shown to indeed promote growth of β-cells through down modulation of cyclin-dependent kinase inhibitors, thereby ameliorating insulin-deficient diabetes (198). Furthermore, DCs obtained from T1D patients acquired an immature phenotype with reduced levels of activation markers and increased anti-inflammatory cytokine secretion upon treatment with human BM MSC-derived EVs in vitro, thus potentially acquiring the ability to hamper the priming and amplification of autoreactive T cells in tissue inflammation (60). The potential exploitation of EVs as biomarkers and/or therapeutic targets in T1D is schematized in Figure 3 (right panels).
In summary, EVs released by immune cells on the one side and β-cells on the other are now believed to sustain a dysregulated cellular crosstalk able to exacerbate the autoimmune response and the inflammatory reaction at the tissue level: in particular, the exposure of intracellular antigens in the extracellular space upon EV release suggests a potential mechanism of autoimmune response induction in the absence of β-cell necrotic destruction. On the contrary, though, EVs and their associated molecular cargo may become allies to monitor and/or predict T1D disease progression and positively affect pathological pathways.
Participation of EVs in Diabetic Complications: The miRNA Effect
The vascular dysregulation is a multifactorial process involving ECs, VSMCs, macrophages, and other immune cell types; the involvement of EVs in the communication among these cells in response to environmental cues is object of intense investigation, and several regulatory pathways have been already delineated. Indeed, EVs are emerging as critical mediators of oxidative stress and LGI in a range of models related to diabetes complications (151, 155). A common denominator is the detrimental capability of EVs—either released at the local level or circulating in blood—to foster proinflammatory pathway in target cells, thus exacerbating tissue inflammation.
Retinopathy
In vitreous samples of diabetic subjects, EVs were found increased compared with healthy controls as a result of local formation of endothelial-derived EVs from retinal vessels and permeation of platelet EVs from plasma, and hypothesized to contribute to retinopathy (27). Blood circulating EVs were found greatly increased in diabetic rats, and were able to induce the oxidative stress and activate the proinflammatory pathways in human retinal ECs in vitro, suggesting their detrimental role in mediating hyperglycemia-induced cellular injury (222). One hypothesized pathogenic mechanism of retinopathy was ascribed to the presence of IgGs associated with EVs in plasma of diabetic mice, which would lead to activation of the classical complement pathway and the upregulation of proinflammatory cytokines and chemokines, resulting in vascular damage (84). Since microvascular dysfunctions due to aberrant interactions between ECs and pericytes are crucial in the onset of retinopathy, several investigators have attempted to functionally associate blood circulating miRNA dysregulation registered in complicated patients with alteration of this cellular interaction. Compared with EVs from healthy control plasma, EVs from subjects with diabetic retinopathy were indeed able to cause pericyte detachment and pericyte/EC migration, increased the permeability of pericyte/EC bilayers and the formation of vessel-like structures in in vitro experimental system of retinal microvasculature (123). In particular, a role for EV-associated miR-150-5p, miR-21-3p, and miR-30b-5p was hypothesized in abnormal angiogenesis, since these miRNAs regulate genes involved in the hypoxia-induced retinal damage characteristic of the disease (124). Of note, EV-associated miR-15a is associated with neuronal damage in the retina in patients with impaired glucose tolerance, an observation not reproduced with freely circulating miR-15a (172). Shaker et al. identified miR-20b and miR-17-3p as able to discriminate between patients with nonprogressive and progressive diabetic retinopathy, and correlated their levels with the expression of two genes involved in retinal endothelial dysfunction: homeobox antisense intergenic RNA (HOTAIR) and metastasis-associated lung adenocarcinoma transcript 1 (MALAT1) (179). Other miRNAs found upregulated in plasma of progressive retinopathy patients, such as miR-21-5p and miR-106a-3p, are linked with inflammation, angiogenesis, and ischemia response (121). The use of plasma EVs from patients with retinopathy caused pericyte detachment and augmented pericyte and/or endothelial permeability in vitro, a relevant phenomenon in early-stage retinopathy (121).
Nephropathy
The interaction between glomerular endothelial cells (GECs) and glomerular mesangial cells (GMCs) is an essential aspect of diabetic nephropathy. In hyperglycemic conditions, GECs secrete a higher number of EVs with high quantities of transforming growth factor (TGF)-β1 mRNA compared with cells in normal glucose. These TGF-β1 mRNA-enriched EVs are able to activate α-smooth muscle actin (α-SMA) expression, proliferation, and aberrant deposition of extracellular matrix by GMCs through the TGF-β1/Smad3 signaling pathway. Moreover, they can be internalized by podocytes, and mediate their epithelial-to-mesenchymal transition and vascular barrier dysfunction, through the canonical Wnt/β-catenin signaling, suggesting a detrimental role of EC-derived EVs in promoting renal fibrosis (211, 212).
Although the presence of microalbuminuria is indicative of initial diabetic nephropathy, no marker is able to clinically predict its development. In an animal model of early experimental diabetic nephropathy, the glomeruli were described to overexpress miR-145, with consequent increase of urinary EV-associated miR-145 levels; consistently, miR-145 was enriched in urinary exosomes from T1D patients with microalbuminuria. In vitro, mesangial cells exposed to high glucose upregulated miR-145 at both the intra- and extracellular levels, providing a potential mechanism for diabetes-induced miR-145 overexpression and nominating miR-145 as a candidate biomarker/player of incipient diabetic nephropathy (12). In T1D patients with proteinuria, the urinary level of EVs containing Wilm's tumor-1 (WT1) protein was found to be higher compared with patients without proteinuria, possibly due to shedding renal epithelial cells, a phenomenon believed to be correlated with renal function decline, and thus proposed as an additional possible early noninvasive marker for diabetic nephropathy (94).
A small RNA-seq analysis of EVs collected from urines from individuals with T1D and various degrees of diabetic nephropathy identified a number of EV-associated miRNAs with quantitative changes associated with nephropathy occurrence and severity, suggesting them as potential biomarkers of this complication. Furthermore, many of these miRNAs are known to directly target pathways associated with renal fibrosis (including TGF-β and PTEN), suggesting also a potential pathogenic role (71).
One study evaluated the potential therapeutic effect of EVs from conditioned medium of urine-derived stem cells, which were able to in vitro protect podocytes from apoptosis in a high-glucose medium condition. Intravenous injections of these EVs in vivo reduced urinary microalbumin excretion, prevented podocyte and tubular epithelial cell apoptosis, suppressed the caspase-3 overexpression and increased GEC proliferation in streptozotocin-induced Sprague-Dawley diabetic rats, suggesting EV potential as promoters of vascular regeneration and prevention of diabetes-dependent kidney injury (91). Similar findings were obtained by using EVs derived from human mesenchymal stromal cells in a NOD/SCID/IL2Rγ KO (NSG) murine model of diabetic nephropathy, an effect likely mediated by the suppression of TGF-β and related profibrotic genes (73).
Neuropathy
An EV therapeutic potential has also been highlighted in neuropathy. MSC-derived EVs indeed displayed a plethora of positive effects on T1D-dependent neuropathic pathology in vitro and in vivo. Injection of cultured mouse MSC-derived EVs was able to increase the number of intraepidermal nerve fibers, myelin thickness, and axonal diameters of sciatic nerves in a murine model of diabetic peripheral neuropathy. This significant improvement of neurological symptoms and the favorable neurovascular remodeling were traced back to the ability of MSC-derived EVs to increase the M2/M1 macrophage phenotype ratio and shut off inflammation (59). The mixed lymphocyte reaction assay indeed indicated that MSC-derived EVs specifically suppressed development of Th1 and Th17 cells (181). Nontoxic and nonimmunogenic nanoparticles used as artificial carriers of anti-inflammatory miR-146a-5p showed protective effects, by inhibiting the expression of the inflammatory cytokines, improving nerve conduction velocity, and alleviating the morphological damage and demyelination of the sciatic nerve in rats with diabetic peripheral neuropathy (115).
Macrovascular complications
Plasma of T1D patients contained markedly elevated levels of EVs of both endothelial and platelet origin compared with matched healthy controls: these EVs carry both procoagulant phosphatidylserine (PS) and proinflammatory high mobility group box-1 protein (HMGB1), an alarmin able to drive endothelial dysfunction, and they were hypothesized to play a role in the development of vascular complications in T1D (15). The progression of vasculopathies in T1D patients is a multifactorial process involving a plethora of different cells; hence, beside a differential expression in terms of EV-associated miRNAs, specific patients may display a differential representation of EVs from the various cellular origins. An increase of EVs released by ECs or platelets is not only observable in patients with overt plaque calcification or coronary artery disease (35, 89), but also in subjects with cardiovascular risk factors, such as hypercholesterolemia, metabolic syndrome, diabetes mellitus, or hypertension (48, 146, 156, 169). Moreover, the augmented circulating levels of EVs released by activated leukocytes in patients with subclinical atherosclerosis suggest that it may be possible to detect early atherosclerosis burden (31). A receiver operating characteristic curve analysis demonstrated that platelets-derived EVs may differentiate patients with and without vascular complications with high sensitivity and specificity, promoting their role as early markers of vascular complications and subclinical atherosclerosis, able to prognosticate a thrombotic event (170). Interestingly, meal intake itself in T1D patients was found to increase circulating EVs of platelet, endothelial and monocyte origin, with a procoagulant potential and augmented thrombin generating capability (183). Ten days of aspirin administration was able to significantly reduce blood circulating tissue factor-positive EVs in T1D patients, showing that aspirin therapy may inhibit microparticle shedding and vascular wall cell activation (32).
At the level of vessels, the crosstalk via EVs can be very efficacious given the strict contact between immune cells, ECs, and VSMCs. EV-mediated exchange of miR-126 supports endothelial growth, a mechanism that is blunted in hyperglycemic conditions (88). Similarly, another study showed that high glucose promotes the shedding of endothelial EVs carrying miR-503, whose transfer to vascular pericytes blunts their migration and proliferation potential (24). Furthermore, EV-associated miR-92a released by ECs in response to proatherogenic stimuli suppressed the expression of Krüppel-like factor 4 in macrophages, regulating their inflammatory phenotype and, in turn, atherosclerotic lesion formation (28). In addition, VSMCs release EVs containing the proinflammatory miR-155, which is internalized in the ECs thus increasing endothelial permeability and destroying the integrity of endothelial barriers, with detrimental consequences on atherosclerotic progression (224). Macrophage foam cells were shown to release EVs able to transfer integrins to the surface of VSMCs and increase the phosphorylation of ERK and Akt in the same cells, affecting focal adhesion pathways and promoting their migration and adhesion at the site of the atherosclerotic lesion (136). Finally, murine DC-derived EVs present TNF-α on their surface, and cause endothelial inflammation through the activation of the NF-κB pathway and contribute to atherosclerotic progression (66). Some of the proposed pathogenic roles of EVs in diabetic complications are depicted in Figure 3 (lower left panel).
In summary, EVs circulating in blood and/or present in the urine of patients with T1D and suffering from diabetic complications have shown the capability (especially when analyzing their miRNA cargo) to monitor or even predict the progression of complications. In most of the studies, investigators have immediately suggested that their identified potential biomarkers may also have a direct pathogenic role in the insurgence of vasculopathies. As a consequence, the diagnostic/prognostic benefit of these molecules is interlaced with the enticing hypothesis that they are directly involved in cell–cell communication, thus pointing to a potential use as therapeutic targets/tools.
EVs as Novel Mediators of Oxidative Stress
More recently, an additional EV-mediated effect through cell-to-cell crosstalk in diabetes involves the regulation of oxidative stress at the paracrine level but also at long range. As a relevant example, miR-15a, which regulates insulin production in pancreatic β-cells, is elevated in diabetic patients' plasma and correlates with disease severity: its pathogenic role was hypothesized by Kamalden et al., who demonstrated that glial cells of the retina overexpressed miR-15a, which resulted in ROS increase and cell death, upon treatment with rat pancreatic β-cell-derived EVs, suggesting a regulation of oxidative stress by EVs at distant sites and consequent EV contribution to retinal injury (95).
In relation with diabetic conditions, hyperglycemia has been shown to modify the content of EVs released by different types of cells in a way that impacts the oxidative homeostasis of EV-receiving cells. Murine macrophages treated with high glucose were described to release EVs with higher levels of miR-21-5p, which promoted ROS production and activation of podocytes through the inhibition of miR-21 target tumor necrosis factor α-induced protein 3 (TNFAIP3), a key enzyme in NF-κB regulation. This biological effect was actually blunted by miR-21-5p blockade, demonstrating the direct functional intervention of EV-associated miR-21 in podocyte injury both in vitro and in a mouse model of diabetic nephropathy (49). In vitro, ECs exposed to high-glucose concentration release a higher number of EVs, which possess increased procoagulant activity, lead to significantly enhanced ROS and O2 •− production in EC target cells, and exert greater impairment of endothelial vasorelaxation compared with EVs released in normal glucose (21). Similarly, EVs derived from human coronary ECs exposed to hyperglycemia were also demonstrated to impair endothelial function, promoting increased macrophage infiltration in the atherosclerotic lesions of ApoE (−/−) mice, an effect mediated by the NADPH oxidase (NOX) activity within EVs (87). Also EVs derived from activated platelets were demonstrated to contain NOX, which, in turn, further potentiated platelet activation by enhancing superoxide generation and directly intervening in ROS generation (69). The demonstrated capability to abrogate this positive loop of platelet activation by NOX inhibition makes it a feasible strategy to treat and prevent thrombotic diseases, also in diabetic subjects (69).
The interception of these phenomena and/or the use of EVs as carriers of antioxidant properties have indeed been tested as potential novel strategies to hamper diabetes-dependent tissue injury. In particular, EVs demonstrated a preclinical potential to halt the progression of cardiac alterations in a mouse model of streptozotocin-induced diabetes; in these animals, heat-shock protein-20 (HSP-20) transgenic cardiomyocytes showed increased secretion of small EVs enriched with superoxide dismutase (SOD)-1, which protected against in vivo diabetes-induced cardiac adverse remodeling (208). A therapeutic role of EVs secreted by adipose-derived stem cells was also demonstrated in a diabetic rat model in which they delayed stress-mediated senescence of endothelial precursor cells (EPCs) induced by high glucose, by promoting EPC proliferation and angiogenesis. Furthermore, when the nuclear factor erythroid 2-related factor 2 (NRF2), necessary to secure cell redox homeostasis, was overexpressed in EVs, their in vivo protective effect increased. In detail, high NRF2-EVs increased granulation tissue formation and vascularization, while decreasing levels of oxidative-stress-related proteins and inflammation, thus accelerating wound healing and reducing ulcerated area in the feet of diabetic rats (108). The described mechanisms showing a direct role of EVs in oxidative stress and regulation are summarized in Figure 4.

In conclusion, we need to acquire more knowledge about the detrimental oxidant and the protective antioxidant properties of EVs to either block or enhance them and ameliorate the progression of specific diabetic complications.
Conclusions
Recent knowledge that EVs released by immune and pancreatic cells may have a substantial effect on the immune system/autoimmunity progression but also on β-cell dysfunction and death may revolutionize T1D pathogenic paradigm. Compared with soluble factors directly secreted into the extracellular space, EVs confer high stability to their molecular cargo through the protective lipid bilayer, thus significantly impacting the biodistribution of functional molecules at the paracrine and endocrine level. Moreover, EVs carry self-antigens, potentially promoting the autoimmune response to these molecules, and thus exacerbating T1D pathology. EV detection in the totality of biological fluids tested so far suggests their capability to affect biological pathways systemically. On the contrary, since an efficient transfer of miRNAs among pancreatic and immune cells, as described above, is presumed to occur thanks to the proximity of the different cell types, it remains very unlikely that this communication represents the initial trigger of the autoimmune reaction. Moreover, since EVs carry multiple RNA, protein and lipid moieties, each with a plethora of possible molecular targets in EV-receiving cells, EV biological effect is possibly the result of the synergistic action of all the cargo molecules; hence the identification of the crucial potential therapeutic targets will not be an easy task. Notwithstanding the long road still ahead of us, a better understanding of EV-dependent communication and the fate and effect of delivered cargo will not only shed new light on the mechanisms promoting the progression of insulin-dependent diabetes and related complications but may also help designing novel therapeutic approaches for T1D and other autoimmune diseases.
Footnotes
Authors' Contributions
S.G. wrote part of the original draft, helped collecting the literature to be cited, prepared the figures, and revised the final draft. F.P. wrote part of the initial draft and substantially contributed to finalize the article. A.C. provided with useful discussions and reviewed the final article. M.G. gave key advices and reviewed the final article.
Author Disclosure Statement
The authors have no conflicts of interest to declare.
Funding Information
P.d.C. is funded by Fondazione Italiana Sclerosi Multipla (FISM No. 2018/R/4) and M.G. by the European Foundation for the Study of Diabetes (EFSD)/Novo Nordisk Programme for Diabetes Research in Europe 2020. This work has also been supported by the Italian Ministry of Health Ricerca Corrente—IRCCS MultiMedica. Schematic figures were created with images adapted from Smart Servier Medical Art (
