Extracellular vesicles are a general term for a variety of nano-scale vesicles that are actively released by cells, which can be divided into microvesicles budding from the plasma membrane, apoptotic bodies shed from dying and disintegrating cells, and exosomes derived from the endolysosomal pathway (14–16) ( Figure 1 ). Plasma membrane funnel-like internalization of cell surface proteins and soluble proteins associated with the extracellular environment led to the formation of early endosomes (17, 18). Incoming endocytic cargo matures into late endosomes through clathrin-dependent or clathrin-independent pathways and ultimately forms MVBs. This process is accompanied by the recruitment of partially soluble molecules (cytosolic proteins and RNA) into ILVs(intraluminal vesicles) (19–21). Different MVBs have different fates and can either fuse with lysosomes or autophagosomes to be degraded, or fuse with the plasma membrane of the parent cell to release the contained ILVs as exosomes (17, 22, 23). Rab family (such as Rab27A, Rab27B) are key mediators of exosome release (23), SNARE (soluble N- ethylmaleimide- sensitive factor attachment protein receptor) can drive membrane fusion to promote exosome secretion (24). The exosomes released outside the cell can fuse with the receptor cytoplasmic membrane to release the cargo directly, or the exosomal transmembrane protein can directly interact with the signaling receptor of the target cell (25–27). The third way in which exosomes released into the extracellular space communicate with recipient cells is that exosomes enter recipient cells by phagocytosis/endocytosis (28, 29), some phagocytosed exosomes will merge into endosomes by endocytosis and be released into adjacent cells, and others will mature into lysosomes and degrade after fusion with endosomes (30).

Tetraspanins, ESCRT (endosomal sorting complex required for transport) protein, phospholipids, TSG101 (tumor susceptibility gene 101), ALIX (ALG-2-interacting protein X), ceramide and SNARE, etc. are involved in exosome biogenesis. Tetraspanins are not only one of the most commonly used exosome marker proteins, but also promote the transport of other membrane proteins and improve their stability (19). As an exosomal scaffold protein, Alix binds TSG101 and ESCRT. Studies have shown that the ESCRT-related cytoplasmic protein ALIX is not only recruited to endosomes through its interaction with LBPA (lysobisphosphatidic acid) (22), but also regulates the formation of ILVs in acidic late endosomes and thus affects exosome generation (27). Trajkovic et al. found that purified exosomes were rich in ceramides, and inhibition of neutral sphingomyelinase reduced exosome release (31). Ceramides can stimulate exosome production and the content of ceramides is positively correlated with the increase of exosome biosynthesis.

The size of exosomes is arranged from 40nm to 160 nm. Exosomes contain mRNA, microRNA (miRNA), ribosomal RNA, long non-coding RNA (lncRNA) and DNA, proteins and lipids (32). Exosomes from different sources contain different structural proteins and lipids. Although exosomes are formed by the invagination of the plasma membrane, the exosome membrane also includes lipids from the Golgi complex resulting in a different membrane lipid content than the plasma membrane (21). The most common proteins include tetraspanins (like CD63, CD9, CD81), heat shock proteins (like Hsp70, Hsp90) and endosomal markers (like Alix). It is worth noting that in addition to the rest of the common cytoskeletal proteins, albumin, etc., major histocompatibility complex class I (MHC I) is expressed on all exosomes, while major histocompatibility complex class II (MHC II) is only expressed on antigen-presenting cells derived exosomes (33). Microvesicles and apoptotic bodies are extracellular vesicles of 50nm -5μm and 1μm -5μm in size, respectively. Microvesicles were originally studied for their role in blood coagulation and were formerly known as “platelet dust” (34). Unlike exosomes, microvesicle biogenesis involves changes in lipid composition, protein composition, and Ca2⁺ levels. Microvesicles contain a large amount of cholesterol and calpain, which facilitates membrane budding and the formation of microvesicles (35). Compared to exosomes, microvesicles, apoptotic bodies are relatively large extracellular vesicles that contain remnants of encapsulated dying cells. Phosphatidylserine is the only marker of apoptotic bodies found so far (36). Recent studies on apoptotic bodies have found that apoptotic bodies can not only activate the immune system, but also transmit genetic information. Samos et al. detected apoptotic bodies in the blood of tumor xenograft-bearing mice (37). Due to cross-running of exosomes with the biogenesis pathways of the remaining extracellular vesicles, the precise rate-limiting roles and functions of these molecules in exosome biogenesis need to be further explored.

The most important feature of obesity is the excessive accumulation of adipose tissue (42). There is increasing evidence that adipose tissue-derived exosomes not only function in paracrine and endocrine modes, but also act as intercellular communicators to facilitate the transition of adipocytes towards maturity (43). In studies of the role of exosomes in systemic metabolism, adipocyte-derived exosomes were found to be mediators linking obesity and insulin resistance in surrounding tissues such as the liver, interacting with adjacent sites to promote lipid esterification (44, 45). Connolly et al. showed that exosomes per adipocyte have higher yields in the preadipocyte stage and are rich in pre-signal fatty acids (46). Exosomes contain different fatty acids with different stages of differentiation. Interestingly, ATM (adipose tissue macrophage)-derived exosomes not only modulate adipose tissue function and insulin sensitivity, but also promote the activation of monocytes to macrophages (M1 pro-inflammatory phenotype) after uptake by peripheral blood (47). Wei et al. found that miRNAs contained in adipose tissue macrophage-derived exosomes from obese mice promoted insulin resistance, while ATM-derived exosomes from lean mice attenuated insulin resistance in obese mice (48). In addition, it has been demonstrated obesity has an impact on the cargo carried by adipocyte-derived exosomes. Carlos et al. used exosomes from obese mice to treat lean mice with glucose intolerance and insulin resistance, and further found that obesity alters the distribution of exosomal miRNAs in mice (49). Studies on the 3T3-L1 adipocyte model showed that these adipocytes increased the content of exosomal miR-802-5p, which promotes insulin resistance in cardiomyocytes by downregulating HSP60 (50). Clinical studies have shown that subcutaneous adipocyte-derived exosomes from obese patients are enriched in proteins related to fatty acid oxidation, leading to pathophysiological changes (51). Although many researches have showed that macrophages in adipose tissue are the main pool of exosomes, new studies have showed that it is adipocytes that released large quantities of exosomes.

The effect of adipocyte-derived exosomes on body metabolism has been widely recognized. However, the functional impact of adipocyte-derived exosomes on the control of adipogenesis needs to be further explored to clarify its profound impact in cardiovascular and metabolic diseases. On the other hand, the effect of obesity on the cargo contained in adipocyte-derived exosomes has been intensively studied and explored (52). However, since there is no specific cargo selection process for cellular uptake of exosomes, the relationship between the effect of exosomes on recipient cells and specific exosome contents needs to be further clarified to facilitate design exosomes for disease treatment.

Diabetes is a group of metabolic disorders characterized by chronically high levels of blood sugar due to insufficient insulin production (T1DM) or poor response of receptor cells to insulin (T2DM) (4, 10). Exosomes have recently been investigated as a potential candidate for biomarkers involved in obesity-related diabetes progression and prognosis. Existing evidence suggests that adipose tissue may coordinate systemic metabolic homeostasis through exosomes, but the quantity, quality, and cargo of exosomes produced by adipose tissue are abnormal under pathological conditions (26, 53). Adipose tissue is divided into white adipose tissue (WAT) and brown adipose tissue (BAT). The number of exosomes extracted from white adipose tissue in patients with inflammation and systemic insulin resistance is increased, the uptake of exosomes by leukocytes is increased, and the cargo carried by exosomes may signal the conversion of endothelial cells with a normal phenotype to a diabetic phenotype (54). Clair et al. found that adipose tissue-derived exosomes were more responsive to glucagon compared to the exosomes derived from other tissues (55). Glucagon not only accelerated the uptake of BSA and cargo into exosomes by adipose tissue endothelial cells, but also enhanced the secretion of exosomes (56, 57). Because endothelial cell exosomes contain extracellular/serum signaling molecules, it can be speculated that the type of signaling in the blood can affect endothelial cell exosomes and produce functional responses to adipocytes and metabolic systems (58, 59).

Adipocyte-derived exosomes and adipose tissue macrophage-derived exosomes drive this intra-organ crosstalk of adipocytes and macrophages in type 2 diabetes. In type 2 diabetes, sonic hedgehog containing exosomes can mediate pro-inflammatory macrophage (M1 pro-inflammatory phenotype) activation (60). Macrophages (M1 pro-inflammatory phenotype) affect the metabolic status of adipocytes by releasing cytokines (like IL-6, TNF-α), leading to systemic glucose intolerance and insulin resistance (61, 62). Ying et al. found that injection of ATM (Adipose Tissue Macrophage)-derived exosomes from obese mice into lean mice reduced the expression levels of peroxisome proliferator-activated receptor γ (PPARγ), glucose transporter type 4 (GLUT4) and adipocyte sensitivity to insulin (41, 63). miRNA-155, an inhibitor of PPARγ, has been shown to be the most critical factor in ATM exosomes affecting metabolic disorders (64). Surprisingly, in one study, brown adipocyte-derived exosomes from normal mice alleviated systemic insulin resistance in ADiceKO mice after being taken up by the liver (65). Existing evidence suggest that adipose tissue, especially brown adipose tissue-derived exosomes, maintain metabolic homeostasis under physiological conditions (66, 67), while under pathological conditions, such as T2DM, adipose tissue-derived exosomes cause intra-organ crosstalk synergy to aggravate glucose intolerance and insulin resistance (68, 69).

In the progression of obesity, chronic overnutrition often causes unhealthy expansion of adipose tissue leading to immune cell infiltration, which in turn triggers acute and chronic inflammation of the tissue (70, 71). The most important factor causing a series of inflammation is the inflammation of adipose tissue. Macrophages are the main immune cells in adipose tissue and exhibit different phenotypes depending on the changing environment in the body (72). ATM undergoes a transition from an anti-inflammatory M2 phenotype to an M1 pro-inflammatory phenotype during the progression of obesity and produces pro-inflammatory factors to exacerbate adipose inflammation (73). In addition to macrophages, obese adipose tissue recruits additional immune cells. Studies have shown that CD4+ and CD8+ T cells secreting proinflammatory mediators are increased in WAT (74). Studies have shown that injection of M2 macrophage-derived exosomes into obese mice can significantly improve insulin-glucose homeostasis (75, 76). The exosomes secreted by adipocytes carrying miRNA-34a inhibited the polarization of M2 macrophages to promote the occurrence and development of adipose inflammation, and further found that miRNA-34a KO mice had higher levels of adiponectin than WT mice (77). Adiponectin, an adipokine with insulin-sensitizing activity, abolished macrophage-involved extracellular matrix remodeling, such as collagen formation and fibrosis (78). Adiponectin, Omentin-1 and Secreted Frizzled-Related Protein 5 (SFRP5) are anti-inflammatory adipokines (1, 79, 80). Omentin-1 has anti-inflammatory, anti-obesity, and anti-diabetic properties (81, 82). Omentin-1 enhances insulin stimulation to enhance glucose uptake and activate insulin receptor substrate (IRS) by inhibiting mTOR signaling pathway (83, 84). SFRP5 is thought to be a negative regulator of adipose tissue-related chronic inflammation. SFRP5 inhibits Wnt5a-mediated inflammation, obesity and insulin resistance by binding to Wnt proteins (85). Serum SFRP5 levels were lower in obese and T2DM patients, and SFRP5 depletion increased macrophage numbers and pro-inflammatory proteins in mouse adipose tissue (86–88).

miRNA can be packaged into exosomes, loaded into high-density lipoprotein (HDL), and bound to AGO2 protein outside the vesicle to avoid miRNA degradation (94). The proportion of miRNAs in exosomes is higher than in parental cells. Studies have confirmed that miRNAs do not enter exosomes randomly, and some miRNAs (e.g. miRNA-150, miRNA-451) preferentially enter the exosome lumen (95). Several studies have shown that AGO2 and other RNA-binding proteins are involved in regulating the process of miRNA entry into exosomes (94). According to current researches, several potential approaches have been put forward to show the exosomal miRNA sorting modes. In 2013s, Kosaka et al. firstly found that overexpression of sphingomyelinase 2 (nSMase2) in cancer cells can increase the number of exosomal miRNAs and promote angiogenesis as well as metastasis in tumor microenvironment (96). miRNA motif and sumoylated heterogeneous nuclear ribonucleoproteins (hnRNPs)-dependent pathway have also been reported to regulate exosomal miRNA sorting. Both Villarroya Beltri and Gao et al. have found that hnRNPA2B1 could recognize the GGAG motif in the 3’ sides of miRNA and result in specific miRNAs to be packed into exosomes (97, 98). Similarly, Koppers-Lalic et al. have revealed that 3’ sides of uridylated endogenous miRNAs were mainly sorted in exosomes derived from B cells or urine, in the meanwhile, the 3’ sides of adenylated endogenous miRNAs were mainly concentrated in B cells (99). Also, recent studies have found that miRNA-induced silencing complex (miRISC)-related pathway might be correlated with exomal miRNA sorting (100). For instance, Guduric Fuchs et al. have discovered that knockout of AGO2 in 293T cells can downregulate the bundance of the specific miRNAs such as miR-451 and miR-142-3p in exosomes (95). In the past year, Xiao-Man Liu et al. have found that phase separation-mediated local enrichment of cytosolic RNA-binding proteins enables their targeting and packaging by exosomes (101). They found that YBX1 protein efficiently forms liquid-like droplets in cells and miR-223 could be efficiently partitioned into YBX1 droplets (101). Although miRNA sorting to exosomes might be a passive mechanism to deliver miRNAs even in excess of their acceptor cells, emerging researches suggest that exosomes can be actively uptake by other cells. These results lead to specific miRNA transfer among different cells. In addition, certain miRNA signatures in exosomes might represent biomarkers of diseases. For example, tumor cells have altered transcriptomic profiles, which in turn will significantly influence miRNA sorting to exosomes (102). A great number of previous reviews have summarized the exosomal miRNA can act as biomarkers. Not only miRNA can function as biomarkers, but also miRNAs can reach neighboring and distant cells through exosome circulation, mediate cell-to-cell communication by targeting mRNAs and confer characteristic changes in the expression levels of target genes (103). Thomas Thomou ei al. have found that adipose tissues are major source of circulating exosomal miRNAs and function as gene regulator in distant tissues thereby serving as a novel adipokine (104). To be specific, they found that mice with adipose-specific knockout of the miRNA-processing enzyme Dicer (ADicerKO), as well as humans with lipodystrophy, have less circulating exosomal miRNAs compared with normal samples (104). Similarly, C. Ronald Kahn et al. also have found that different cell types released different amounts of exosomes and differentiated 3T3-L1 cells (white adipocytes) having the highest production and release rates per cell (102). Furthermore, they found that miRNAs possess sorting sequences which determine their secretion in exosomes or cellular retention and that different cell types make preferential use of specific sorting sequences, thus defining the exosomal miRNA profile of that cell type (105).

Adipose tissue is a particularly important contributor to the circulating exosomal miRNA pool, with adipocytes and stem cells expressing a wide range of miRNAs. miRNAs in adipose tissue-derived exosomes promote metabolic homeostasis in an endocrine manner. Thomas et al. used ADicerKO mice to significantly reduce exosomal miRNAs, reduce WAT, and whiten BAT (104). After transplantation of adipose tissue into ADicerKO mice, the glucose tolerance of the mice improved, and the content of most exosomal miRNAs returned to normal levels (104). Interestingly, serum FGF21 (fibroblast growth factor-21) and liver FGF21 mRNA remained unchanged in ADicerKO mice after WAT transplantation (104). In contrast, FGF21 mRNA in the liver of ADicerKO mice transplanted with BAT was reduced by about 50% (104). To further clarify which miRNAs might regulate FGF21, AML-12 hepatocytes were transfected with adenoviral pacAd5-FGF213’-UTR luciferase reporter gene and found that only miR-99b resulted in a significant decrease in FGF21 luciferase activity (104). Other studies have shown that miRNAs in adipose tissue-derived exosomes may also regulate body metabolism in a paracrine manner. Exosomes containing miR-16, miR-27a, miR-146b and miR-222 released from adipocytes can enter small adipocytes to stimulate adipogenesis (49). By comparing exosomes released by adipose tissue-derived stem cells (ADSC-Exos) with exosomes released by adipose tissue (AT-Exos), Zhang et al. found that compared with ADSC-Exos, AT-Exos contained 7 times more miR-450a, While miR-450a increases adipogenesis by inhibiting WISP2 (106). This study demonstrated that exosomal miRNAs also have autocrine functions. Adipose tissue macrophage-derived exosomes also overexpressed miR-155. miR-155 inhibits insulin action by downregulating PPARγ mRNA (41, 65). Diabetic patients had lower levels of miRNA-126 and miRNA-26a compared to normal individuals, which are characteristic markers of endothelial cell-derived exosomes (107, 108). Studies have shown that the release of exosomes is up-regulated by physiological (such as insulin) and pharmacological (such as glimepiride) stimuli (109). Fang et al. reported that after administration of rosiglitazone, adipocytes secreted exosomes containing a large amount of miR-200 and were taken up by cardiomyocytes resulting in hypertrophic changes (110). Interorgan crosstalk mediated by exosomal miR-200 may be the molecular mechanism of the adverse effects of rosiglitazone. However, studies on the effects of metabolized drugs on the production, release, and transport of exosomes on the body are limited, but research to explore the therapeutic application of exosomes still holds attractive prospects. In addition to adipose tissue-derived exosomal miRNAs, circulating miRNAs from other sources have increasingly been found to be strongly linked to metabolic disorders (111). Continued in-depth exploration of miRNAs will expand our understanding of the specific relationship between miRNAs and body metabolism.

Besides miRNAs, adipose tissue-derived exosomes or other cargoes are also involved in metabolic homeostasis. Liu et al. transferred adipose tissue-derived exosomes to mouse macrophages by melatonin to reduce metabolic inflammation and increase α-ketoglutarate (αKG) levels, and found that αKG is a melatonin inhibitor of adipose inflammation ‘s target (112). Melatonin attenuates adipocyte inflammation by promoting TET-mediated DNA methylation by transporting adipose tissue-derived exosomal αKG to macrophages (112). Studies have isolated exosomes derived from adipose-derived mesenchymal stem cells, and demonstrated that glyoxalase-1 (GLO-1) is highly expressed in the exosomes and efficiently targeted to deliver adipose-derived exosomes loaded with GLO-1 protein bygenetic engineering to protect endothelial cells and enhance angiogenesis in type 2 diabetic mice (113). Regarding the study of proteins and other cargoes, current evidence suggests that adipose-derived exosomes play an active role in metabolic homeostasis.