Energetic Stress-induced Metabolic Regulation by Extracellular Vesicles

Abstract

Recent studies have demonstrated that extracellular vesicles (EVs) serve powerful and complex functions in metabolic regulation and metabolic-associated disease, although this field of research is still in its infancy. EVs are released into the extracellular space from all cells and carry a wide range of cargo including miRNAs, mRNA, DNA, proteins and metabolites that have robust signaling effects in receiving cells. EV production is stimulated by all major stress pathways and, as such, has a role in both restoring homeostasis during stress and perpetuating disease. In metabolic regulation the dominant stress signal is a lack of energy due either nutrient deficits or damaged mitochondria from nutrient excess. This stress signal is termed “energetic stress”, which triggers a robust and evolutionarily conserved response that engages major cellular stress pathways, the ER unfolded protein response, the hypoxia response, the antioxidant response and autophagy. This review proposes the model that energetic stress is the dominant stimulator of EV release with a focus on metabolically important cells such has hepatocytes, adipocyte, myocytes and pancreatic β-cells. Furthermore, this review will discuss how the cargo in stress-stimulated EVs regulate metabolism in receiving cells in both beneficial and detrimental ways.

Introduction

Recognizing the complexity of inter-cellular regulation of metabolism

Metabolism in multicellular organisms is tightly regulated through the concerted actions of multiple specialized cells. In vertebrates these specialized cells make up organ systems that preform specific functions that benefit the organism. For example, adipocytes store excess nutrients for future release in times of sparsity, hepatocytes package lipids from the diet into a usable form for other tissues and produce glucose to sustain the energetic requirements of the organism during fasting, and collections of neurons in the hypothalamus regulate feeding behavior and energy expenditure. Although the simplistic goal of metabolism-regulating mechanisms is to provide all cells in the organism with adequate fuel to carry out their functions, accomplishing this goal is a highly complex endeavor. Energy supplying organs like adipose tissue, liver and hypothalamus must be able to match energy output with the energy demand in the system. This includes sensing shifts in energy expenditure caused by energetically expensive processes like exercise, reproduction, and illness and responding by altering feeding behavior, blood nutrient availability and basal metabolic rate. The inter-organ regulation of metabolism is traditionally described by the release of hormones from multiple organs with receptor-mediated effects on metabolism in hormone-sensitive cells. The classic endocrine organs that secrete metabolism-regulating hormones are the thyroid gland, adrenal gland and pancreas, but other organs such as the adipose tissue and gastrointestinal tract also secrete an impressive array of hormones. Adipose tissue releases adipokines that span protein hormones, bioactive lipids and metabolites (1). The most well studied are leptin and adiponectin. Leptin is released from adipocytes in the fed state when adipocytes are actively storing energy. Its major action is on the brain to suppress food intake (1). Adiponectin is a strong insulin-sensitizing and anti-lipotoxic hormone that decreases gluconeogenesis in the liver and promotes fatty acid oxidation in skeletal muscle (1). Furthermore, in recent years the gastrointestinal tract has been shown to release over 20 peptide hormones such as ghrelin, glucose-dependent insulinotropic peptide (GIP) and glucagon-like peptides (GLPs; (2)). The release of these hormones is dictated by the gut nutrient composition and can act on various cell types to regulate gut motility, appetite control and maintenance of glucose homeostasis (2). Therefore, under any given physiological state there is a diverse composite of signaling molecules from various organs, that facilitate the tight regulation of systemic metabolism. Dysregulation of this inter-cellular communication can lead to obesity and metabolic syndrome or exacerbate already established metabolic disease (2–4).

The Immergence of Extracellular Vesicles in Metabolic Regulation: Compounding Complexity

Extracellular vesicles (EVs) are highly heterogeneous membranous vesicles released into the extracellular space by most cells. EVs carry signaling proteins, mRNA, DNA, miRNAs, bioactive lipids, and metabolites (5). These EVs can function both to release unwanted or damaged material from the cell, and to initiate communication with other cells (6). As would be expected from such a diverse collection of cargo, EVs can have a more powerful and global effect on cellular signaling than hormone-mediated signaling from a single receptor. To date, several types of EVs have been described and classified by their intracellular origins. The initial discoveries of EVs identified exosomes, microvesicles and apoptotic bodies (5). Exosomes are produced in the multivesicular endosome (MVE) by the budding of the limiting membrane to form intraluminal vesicles (ILVs). Fusion of the MVE with the plasma membrane releases these 30–150 nm ILVs into the extracellular space, which are then termed exosomes (5, 7). Microvesicles, also called ectosomes, are formed by the direct budding of the plasma membrane (ectocytosis; (8)) and tend to be much larger than exosomes (50nm-1000nm; (5)). Whereas exosomes and microvesicles are released from healthy cells, apoptotic bodies are produced at the plasma membrane by dying cells. Additional classes of EVs have recently been discovered to originate from the endolysosomal system including those produced by lysosomal exocytosis, and secretory autophagy (9–11). In recent years EVs have been demonstrated to regulate metabolism, opening a new field of research. EV signaling has been called the next frontier in endocrinology (12), representing an unexplored network of metabolic regulation that may act synergistically or independently of classic hormones.

Recent studies have demonstrated a link between soluble hormone action and EV production. The adipokines adiponectin and resistin have been detected in circulating EVs at trace amounts (13). There is some evidence that EV-associated resistin can have signaling effects despite being a small fraction of the total resistin levels in circulation (14). However, the adipokines adiponectin and leptin have also been found to directly regulation EV production. Obata et. al. demonstrated that circulating adiponectin binds to T-cadherin on the surface of endothelial cells and stimulates EV production and release (15). These EVs contain endocytosed adiponectin-T-cadherin complexes and are highly enriched for ceramides. The authors suggest this is a mechanism by which adiponectin acts to clear the endothelial cell of toxic ceramide species. Leptin has also been shown to stimulate sEV production in breast cancer cell lines (16). This occurs through the post-transcription upregulation of TSG101, a critical member of the endosomal sorting complex (ESCRT-1). Glucagon is one of few examples outside of adipokines, of the crosstalk between classic hormones and EVs. Glucagon stimulates endothelial cell endocytosis of BSA-bound cargo from circulation, incorporates these molecules into sEVs, and enhances sEV production (17). These studies open the exciting possibility of robust and intricate cross-talked between soluble hormones and EV-mediated signaling.

Although little is known about how EVs signal to regulate metabolism under physiological conditions, EVs clearly influence the pathogenesis of obesity, metabolic syndrome, and associated comorbidities (18, 19). In fact, EVs can modulate the pathology of a wide variety of diseases including cardiovascular disease, cancer, mental disorders, and neurodegeneration, each of which have a metabolic component (20–22). The reason that EVs are so active in disease likely stems from the inherent nature of EVs as stress-stimulated signaling mechanisms.

Stress-stimulated EV Production

Many cells release EVs basally, however, various cellular stresses result in higher EV release with altered cargo. For example, stresses such as inflammation, fatty acids, hypoxia, thermal stress, osmotic stress, oxidative stress, genotoxic stress, pH and nutrient deprivation all result in enhanced EV release and altered EV cargo in many cell types (23–31). Well defined cellular stress responses are induced by the above-mentioned stimuli: the unfolded protein response (UPR) to ER stress, the antioxidant response to oxidative stress, and hypoxia response to low oxygen tension. These stress pathways have been linked directly to altered EV production and/or composition. For example, endothelial cells treated with thapsigargin or dithiothreitol to directly induce ER stress did not display changes in microvesicle production, but microvesicles were able to induce ER stress in untreated endothelial cells (32). In cancer cells, chemical induction of ER stress results in higher EV release that carry pro-inflammatory signals (33, 34). Likewise, hepatocytes treated with palmitate, or thapsigargin released more EVs, which was dependent on key proteins in the ER stress signaling pathway, IRE1α and XPB1s (35). In the context of oxidative stress, pro-oxidant conditions result in enhanced small EV release in HEK293 cells and senescent cells altering the cargo to be pro-inflammatory and pro-proliferative respectively (36, 37). These effects could be prevented with the antioxidant molecule N-acetyl-cysteine (NAC). In adipocytes, oxidative stress induced by mitochondrial electron transport chain inhibitors significantly increased small EV release (38). This effect was not seen when adipocytes were treated with buthionine sulphoximine (BSO), a compound that causes oxidative stress by depleting cellular glutathione (38). This suggests that ROS-mediated small EV release is likely due to mitochondria specific oxidative stress. Lastly, the key protein mediator of the hypoxia stress response, HIF1α, is activated by ROS to induce EV secretion in HIV-1 infected CD4⁺ T cells (39). Activation of HIF1α during ischemia in kidney tubular epithelial cells or HEK293 cells also results in increased shedding of EVs with altered miRNA composition (40, 41). Therefore, increased production and altered packaging of EVs is tied to major stress responses.

Stress-induced EV signaling is by far the most well studied in cancer research. Cancer cells undergo all the above-mentioned stressors to various degrees depending on the stage and whether they are associated with a solid tumor. Not much is known about stress stimulated EVs in metabolic regulation, however, many of these same stress states can be found in multiple organs under obese and diabetic conditions. This review will discuss energetic stress as underlying most stress responses that are linked to EV release and explore the functions of these stress associated EVs in receiving cells.

An Energetic Perspective of Stress

Hijacking old pathways for modern life

Nutrient acquisition is one of the most formidable and sustained environmental pressures throughout the evolutionary history of all organisms. Chemical energy from consumed nutrients is the principal requirement for life. As such, starvation elicits one of the strongest survival responses by slowing basal metabolism to preserve energy, promoting systemic reliance on adipose-derived lipids for energy to spare glucose for the brain, and initiating intense food-seeking behavior. Nutrient deprivation can be the result of starvation, but also from intense energy consumption, as in the case of exercise, or from a physical restriction of nutrient delivery to the tissue, as with expanding tissues that outpace angiogenesis. At the cellular level, nutrient deprivation activates multiple stress response pathways in many cell types, the ER UPR, the antioxidant response, hypoxia response and autophagy (Figure 1A) (42–48). Although these stress responses may be temporally regulated in the cell, they are highly interconnected. For example, protein folding in the ER is highly sensitive to ATP and calcium levels and reliant on a strict redox environment to allow for disulfide bonds to form (49). As the ER and mitochondria are physically connected, mitochondrial dysfunction can induce the ER stress response by altering ATP levels and generating ROS (50, 51). The reverse is also true, ER dysfunction can cause mitochondrial dysfunction and oxidative stress (50). The ER can generate hydrogen peroxide through oxidative protein folding (52). This ER-generated ROS can trigger the ER stress response, which in turn can activate antioxidant pathways involved in the oxidative stress response (53). Furthermore, ROS is also generated under hypoxic conditions and is known to activate the integrated stress response, which includes the UPR (54). Lastly, downstream of all these stress pathways is the activation of autophagy, a degradative process requiring specialized machinery to engulf damaged cytoplasmic components, including whole organelles. The autophagy machinery delivers this captured material to the lysosome, which digests and recycles cellular components. It is generally accepted that autophagy is a highly conserved process to maintain homeostasis during nutrient depletion by salvaging fuel for ATP generation and building blocks for anabolic reactions (55). These interacting stress responses have the purpose of restoring the cell to homeostasis and repairing any damage that occurred during nutrient stress.

Figure 1.

Convergent energetic stress responses of obesity, starvation, and exercise. A. Energetic stress induced by obesity, starvation and exercise results in low amino acid levels, low energy charge, high NAD⁺ levels, high ROS and increased circulating lipids. Each of these stress signals activates the respective sensor enzyme as indicated or amplifies other stress signals. B. The stress signals and sensor pathways activate highly interconnected stress response pathways: antioxidant, hypoxia, ER UPR and autophagy. These pathways are known regulators of EV release, however, how they interact to accomplish this is not understood. C. The convergent regulation of the endolysosomal system by stress pathways may account for the net increase in EV release during energetic stress.

In modern Western society there is no restraint on calorie procurement so humans rarely, if ever, must exercise starvation-induced cellular responses. Instead, a significant portion of humans have become obese. Interestingly, the same stress responses are activated by nutrient deprivation and nutrient excess in metabolic tissues (Figure 1A). Obesity stimulates the ER UPR, antioxidant pathways and hypoxia response in the liver, adipose tissue, and muscle in mice (56–67). This can be mimicked by simply providing mice with an infusion of fatty acids (68–71), suggesting these stress responses can be directly activated by excess fatty acids. In both fasting and obesity free fatty acids rise in circulation due to adipocyte lipolysis or adipose tissue dysfunction respectively. Tissues are forced to rely on fatty acids for ATP synthesis, in part because there is either no insulin signal, as in the case of fasting, or the tissues are insulin resistant, as in obesity. Fatty acids are energy-rich and require oxygen to be oxidized in the mitochondria to yield ATP. However, excess fatty acids can lead to significant mitochondrial dysfunction and reduce the cell’s ATP-generating capacity. Mitochondria respiring on fatty acids produce more free radicals than those utilizing glucose (72). Fatty acids can be oxidatively damaged forming highly reactive lipid species that are toxic to most cells. Furthermore, fatty acids can uncouple the mitochondria and inhibit oxidative phosphorylation (73, 74). In addition, the oxygen consumed during enhanced oxidation of fatty acids in adipocytes have been shown to produce a local “pseudohypoxia”, activating the major hypoxia stress response, HIF1α (66, 75). Lastly, long chain fatty acids in cell culture or obesity in vivo have been shown to activate autophagy in pancreatic β cells (76, 77). Palmitate was also found to stimulate autophagy in MEFs, which seems to be a common finding in cultured cells (78, 79). However, the effect of obesity on autophagy is highly dependent on the effected tissue. Autophagy has been reported to be increased in adipose tissue but decreased in other organs like the heart and liver (79). Therefore, fatty acids and their actions in mitochondria are at least one link between the convergent cellular stress responses of starvation and obesity: the ER UPR, antioxidant response, hypoxic response and autophagy (Figure 1A). As obesity is a new problem in our evolutionary history, it would seem that cells are hijacking ancient starvation-induced stress pathways to adapt to the modern stress of over-nutrition. However, this re-purposing of signaling is not ideal. Starvation is generally short lived, on the order of days, so the collateral damage of strict fatty acid utilization for ATP production is an acceptable trade-off. In the case of obesity, excess lipids can circulate throughout the body for years to a lifetime resulting in compounding oxidative stress and inflammation in multiple organs, leading to chronic disease.

Energetic stress: one stress to rule them all

From this perspective, the same stress pathways organisms have evolved to combat nutrient deprivation can be harnessed imperfectly to adapt to overnutrition. Every cellular process needs energy and disruption of the cells ability to acquire it, due to a lack of energetic substrates or overwhelming the mitochondrial capacity, elicits a stress response. Therefore, it can be said that both the nutrient stress of deficiency, or surplus, can be characterized generally as “energetic stress”. Energetic stress can be identified by low cellular energy status (low ATP/ADP and NADH/NAD⁺ ratios), limiting biosynthetic metabolites and high mitochondrial ROS generation due to dysfunctional mitochondria, or limiting oxygen. Induction of an “energetic stress response” activates a network of discrete, but interconnected stress responses that include all major stress pathways, as mentioned above. On the other hand, energy demand is significantly increased during stress to mount, sustain, and recover from stress responses (80). Thus, if a stress response did not originate from energic defects it is still likely accompanied by some degree of energetic stress. For example, DNA damage stimulates mitochondrial oxygen consumption to produce the required ATP for repair but this, in turn, results in oxidative stress-induced mitochondrial damage, inducing the mitochondrial stress response, mitophagy (81, 82). Induction of mitophagy following DNA damage was found to be essential for maintaining mitochondrial ATP production to support DNA repair (82). In addition, AMP-activated protein kinase (AMPK) activation is integral to the DNA damage repair pathways (83). AMPK is a key regulator of cellular bioenergetics being directly activated by AMP and ADP under conditions of low energy charge (84). The activity of AMPK stimulates the restoration of cellular ATP by enhancing glucose and fatty acid catabolism, activating autophagy, and inhibiting energy-consuming processes like cell growth and biosynthesis (83). Phosphorylation of Thr172 by LKB1 is required for AMPK activation (84). Conformational changes induced by AMP or ADP binding to the regulatory subunits of AMPK enhances phosphorylation and suppresses de-phosphorylation, allowing for sustained activation (85). These studies suggest that energetic stress is a consequence of the DNA damage response and the adaptation to which rewires cellular bioenergetics to support repair processes. This is likely a phenomenon that contributes other energy demanding stress responses.

Energetic Stress and EV Production

EV release is significantly regulated by energetic stress in metabolic cells. In starving mice, the total circulating EVs and adipose tissue EVs are increased (17, 27, 86). A similar phenomenon was observed after a bout of intense exercise in humans (87–89). Likewise, in both mice and human obesity total circulating EVs and adipose tissue EVs are increased (38, 86, 90, 91). Although, there are no reliable techniques to accurately identify the cell type origin of circulating EVs, it is estimated that skeletal muscle contributes a significant amount of EVs to the circulating pool (92). In stress conditions, the cells that release more EVs into the blood are likely those that are the most effected by, or the most sensitive to, the given energetic stress. In obesity this is adipocytes, myocytes and hepatocytes, and in exercise it is myocytes, which all display increased EV production under the defined energetic stress (23, 38, 93–99). It is interesting to note that EV release in response to energetic stress is a seemingly dichotomous process. Much more energy is required to restore the biomass lost to the cell in EVs in addition to that required for the stress response itself. Because EV release accompanies all forms of stress, we can speculate the benefit of doing so is worth the energetic cost. For example, one function of EVs is thought to be disposal of unwanted material. Therefore, under energetic stress the cell could extrude damaged cellular components to lessen the toxicity and burden of repair. Additionally, as discussed later in this review, EVs can act as “help me” signals, suggesting the recruitment of resources is worth the loss in biomass.

Signals and Sensors for EV Release During Energetic Stress

Whether EV release is triggered by the energetic stress perse (low ATP/ADP, low NADH/NAD⁺ or high ROS) or the multiple stress pathways activated under energetic stress is still unknown. However, there is evidence that EV production is regulated by proteins that directly sense the energetic state of the cell and function to match the metabolic activity of the cell with nutrient and energy availability: the mechanistic target if rapamycin complex 1 (MTORC1) and AMPK. MTORC1 has been described as a “metabolic rheostat”, sensing nutrient levels and growth signals, and modulating growth pathways and anabolic processes accordingly (100). This sensing function of MTORC1 is downstream of multiple pathways that are modulated by amino acids, ATP, oxygen, glucose, nucleotides and other nutrients (100). When these pathways signal adequate building block availability, MTORC1 is activated to promote anabolism, cell growth and proliferation and simultaneously attenuating catabolic processes like autophagy (100). AMPK and MTORC1 antagonize each other in function by direct or indirect inhibition (101). Under conditions where energy is limiting (high (AMP+ADP)/ATP ratio), AMPK is activated to enhance ATP generating pathways and recycling pathways like autophagy and inhibit anabolic pathways, which includes the direct phosphorylation and inhibition of MTORC1(84). The reciprocal regulation of these kinases situates them at the center of cellular metabolic decisions. Of high importance during metabolic stress is the judgment of whether to induce the “self-eating” pathway of autophagy. Both AMPK and MTORC1 are considered master regulators of this processes either by inducing or inhibiting autophagic flux respectively (84, 100).

MTORC1 inhibition through rapamycin treatment has been shown to significantly enhance EV release in mouse embryonic fibroblasts (MEFs; (27)). Likewise, activation of MTORC1 by silencing the tuberous sclerosis complex 1 and 2 genes (TSC1/TSC2) results in suppression of EV release (27). This data is consistent with the notion that energetic stress, where MTORC1 is inhibited, stimulates EV release. However, Yan et. al. reported a conflicting finding, that AMPK inhibition in adipocytes stimulates EV release and AMPK activation reduces EV release (102). A similar finding was reported in cancer cell lines and immortalized endothelial cells (103). Under a given physiological condition these kinases should have the opposite activation state. For example, starvation is a strong activating stimulus for AMPK and inhibitory for MTORC1. However, based on the available data, in this scenario activated AMPK would be working to suppress EV production while, simultaneously, inhibited MTORC1 would stimulate EV release, resulting in a net EV release of zero or an amount skewed to the stronger signal. While these data clearly demonstrate a role for enzymes that are directly regulated by energetic stress in EV production and release, it is unclear how these pathways interact to do so. It is possible these seemingly incompatible results can be explained by differences in the cell types used between the studies. Additionally, there may be dose-dependent or time-dependent effects that are not accounted for by the experimental design of these studies. Finally, these pathways may be independently regulating sub-populations of EVs.

Interestingly, inhibition of MTORC1 in HeLa results in increased EV release but does not result in changes to the proteome or miRNA profile of EVs (27). In fact, 97.48% of the EV proteome was unaltered in response to rapamycin treatment. Likewise, the miRNA profile of EVs from rapamycin treated cells was indistinguishable from those of control cells (27). Zou et. al. concluded that MTORC1 may regulate EV release but not cargo selection. A similar data set has not been published for AMPK knockout cells. It would be interesting to determine if AMPK regulates EV packaging in addition to secretion.

Another indicator of low energetic status of a cell has been linked to EV release, the NADH/NAD⁺ ratio. The quintessential sensors of NAD status are the NAD⁺-dependent deacetylases, sirtuins (SIRTs). This is a family of 7 proteins in mammals that play a substantial role in many metabolic processes and diseases (104). SIRT1 is the most well studied sirtuin protein. It is localized in the nucleus where it deacetylates various proteins such as histones, transcription factors, DNA repair proteins and metabolic enzymes (104). SIRT1, like all sirtuins, is activated by NAD⁺ accumulation, which occurs in muscle, liver and white adipose tissue during, nutrient deprivation, caloric restriction or exercise (104). Activated SIRT1 mounts a strong adaptive pathway that has been shown and protect the cell from stressors, enhance oxidative metabolism and to extend lifespan (104). As with AMPK, the sirtuins are important positive regulators of autophagy during nutrient stress (105). Adipocyte-specific knockout of SIRT1 results in increased circulated sEVs, which contributes to the whole-body phenotype of insulin resistance and increased fat mass (106). Likewise, in HEK293 cells, knockout of SIRT1 or SIRT2 resulted in enhanced sEV release and altered cargo loading through distinct mechanisms (107). Knockdown of SIRT1 but not SIRT6 and 7 resulted in a significant increase in sEV release in breast cancer cells (108). These data suggest that SIRT1 and 2 may act as negative regulators of EV release. Manipulation of sirtuin expression also effects the cargo loading of EVs. SIRT2 interacts with L-type lectin, LMAN2, which regulates the trafficking of proteins between the trans-Golgi Network (TGN) and the endosome. Knockdown of SIRT1 or SIRT2 in HEK293 cells resulted in release of LMAN2, a Golgi-resident protein, in exosomes as well as another marker protein commonly found in exosomes, GPRC5B (107). Likewise, SIRT1 knockdown in breast cancer cells resulted in a highly altered EV proteome that was enriched with ubiquitinated proteins (108). Therefore, SIRT1 may regulate the recruitment of proteins to EVs via altering their subcellular trafficking.

A drop in cellular energy charge may modulate EV production through MTORC1, AMPK and the sirtuins but ROS production is also a candidate for initiating EV release during energetic stress. The mitochondria are the main cellular source of free radicals, which is a natural biproduct of oxidative phosphorylation. However, dysfunction of mitochondria results in ROS production that exceeds the cell’s antioxidant capacity and leads to oxidative stress. ROS has been reported to stimulate EV release in several cell types. Treatment of adipocytes with non-toxic doses of mitochondrial electron transport chain inhibitors or palmitate, which are all known stimulators of ROS generation, significantly enhance sEV release (38). This was also shown in vivo where mitochondrial dysfunction was induced by overexpression of mitochondrial ferritin (FtMT), overexpression of mitochondrially targeted amyloid precursor protein β (APPβ), or knockout out SOD2 specifically in adipocytes in a doxycycline-inducible system (38). In each condition, circulating sEVs was increased. In another study, the treatment of HEK293 cells with the calcium ionophore A23187 enhances sEV release by generating free radicals (36). Polycyclic aromatic hydrocarbon-induced ROS has also been shown to stimulate EV release in hepatocytes (109). Additionally, mitochondrial ROS is known to activate AMPK indirectly through its effect on ATP production, potentially effecting EV release through AMPK (110). MTORC1 is also activated by low levels of ROS, but inhibited by high levels of ROS, or long-term ROS exposure (111). Li et. al. found that this biphasic effect was dependent on cell type and mediated through the activities of AMPK and protein phosphatase 2A (PP2A) on the MTORC1 regulatory subunit raptor (111). Lastly, ROS activates HIF1α by stabilizing the protein through inhibiting its upstream negative regulator prolyl hydroxylase (PHD), directly activating HIF1α by nitrosylation, or promoting HIF1α transcription (112). Activation of HIF1α is associated with increased EV release (39, 41).

Energetic stress clearly results in increased EV release from multiple cell types, but based on the above-mentioned studies, the molecular mechanism for this phenomenon is unclear. The main initiating signals of energetic stress are low energy charge (high AMP, ADP, NAD⁺) and high ROS production. These are sensed by AMPK, MTORC1, sirtuins and HIF1α, which are all, in turn, activated in response to energic stress. Although each of these pathways are associated with EV release, their direction of regulation is inconsistent with the general observation of increased EV release under energetic stress. For example, AMPK and sirtuins are activated by low energy charge, but negatively regulate EV release (102, 103, 106, 108). In contrast, MTORC1 inhibition and HIF1α activation, as occurs during nutrient deprivation and high ROS levels, promotes EV release (27, 39). Therefore, under any given condition the pathways that are the most dominant regulators of EV release are likely dependent on the cell type, the specific energetic stress stimulus, the extent of energetic stress, and the amount of time the cell has remained with un-resolved energetic stress. Therefore, much work is still required to understand the interactions between these energetic stress sensing pathways that modulate EV production (Figure 1B).

Effector Processes for EV Release During Energetic Stress: Autophagy and Lysosomal Activity

All signaling pathways that lead to EV production and release must end up in the endolysosomal system, except for those that enhance ectosome production at the plasma membrane. The endolysosomal system is made up of a collection of molecularly defined, but dynamic membrane compartments that regulate the degradation or release of various cellular components. The cellular components that end up in this degrative pathway originate from outside the cell through endocytosis, the plasma membrane as materials that escape recycling vesicles, and damaged cellular macromolecules and organelles that enter through autophagy (113). In recent years, EVs have been shown to be released from multiple compartments in the endolysosomal system. The first are exosomes, which are released after fusion of the MVE with the plasma membrane, an alternative pathway to fusion of the MVE with the lysosome for degradation of its contents (9). Additionally, EVs and soluble proteins can be released from autophagic vesicles through what has been termed secretory autophagy. Autophagosomes are doubled membraned structures that are formed during autophagic engulfment of cellular components. During degrative autophagy, the autophagosome fuses with the lysosome forming a degrading compartment. During secretory autophagy, the autophagosome can fuse with the plasma membrane, releasing its contents into the extracellular space (114, 115). This is considered to happen due to failure of the degrative pathway (9). However, it is important to mention this autophagosome-dependent mechanism of unconventional secretion is it not well understood and likely much more complex than simple fusion of the autophagosome with the plasma membrane. For example, the secretion of IL-1β has been found to occur through secretory autophagy (114), however, this cytoplasmic protein, devoid of a signal peptide, translocates across a membrane into the lumen of a vesicle intermediate that is thought to become the autophagosome(116). Interestingly, IL-1β does not enter the cytoplasmic interior that is formed during autophagosome engulfment of cytoplasmic components. Instead, it localizes to the intermembrane space of the double-membraned autophagosome (116). Zhang et. al. found that, because of this topology, release of soluble IL-1β is guaranteed to occur by either autophagosome fusion with the plasma membrane or through autophagosome fusion with MVEs and then subsequent fusion with the plasma membrane. Amphisome is the term given to this hybrid structure produced by fusion of autophagosomes with the late endosomes or MVEs. Amphisome contents can be degraded by fusion with the lysosome or released into the extracellular space (117, 118). Lastly, the lysosome itself can fuse with the plasma membrane and secrete its contents (9). It has been suggested that lysosomal exocytosis and MVE fusion with the plasma membrane is triggered by impaired acidification of these compartments (119, 120). Therefore, specific forms of EVs arise by these different secretory pathways, although we are far from being able to experimentally separate and characterize these populations. However, it seems to be consistent that the degradative and secretory functions of these pathways antagonize each other. For example, signaling pathways that enhance autophagy and lysosomal activity, which includes adequate acidification of the various compartments, results in reduced EV release. Whereas stimuli that disrupt the full execution of the degrative pathway results in enhanced EV release (Figure 1C).

One of the early studies identifying this tug of war was that published by Hessvik et. al. who demonstrated that inhibition or knockdown of Pikfyve increased exosome release via secretory autophagy (121). Pikfyve is a lipid kinase that generates PI(3,5)P₂, a lipid species required for proper MVE fusion with the lysosome and lysosomal function (121, 122). Likewise, depletion of PI2P by disrupting the function of the PI3P-synthesizing kinase Vps34 resulted in impaired lysosomal degradation and autophagy, which lead to enhanced exosome release (123). In the context of energetic stress, ROS has been shown to disrupt lysosomal activity. Centrosome amplification causes ROS production in pancreatic cancer cells, which directly inhibits lysosomal activity, causing increased exosome secretion (124). In addition, several sensors of energetic stress are known to modulate autophagy or lysosomal activity. For example, the inhibition of EV release in SIRT1 depleted cells is thought to occur by suppression of autophagy in adipocytes or reduced lysosomal function in breast cancer cells (106, 108). In addition, AMPK is well known to activate autophagy through phosphorylation of ULK1 and MTORC1 to inhibit autophagy by phosphorylation of ULK1. Therefore, the combined effect of energetic stress pathways on both the initiation of autophagy and lysosomal degradation and full execution of these processes (eg. ensuring sufficient acidification of the compartment takes place) may be a large determinant of EV release. Furthermore, convergence of multiple stress response pathways on the regulation of the endolysosomal system may explain why such a variety of stresses induce EV release.

The Effect of Energetic Stress-induced EVs on Receiving Cells

EVs can act on receiving cells at two levels. EVs can interact with the plasma membrane to induce cell receptor signal transduction without entry into the cell (5). Additionally, EVs can enter the cell and offload cargo to induce functional cellular changes. EV uptake can be accomplished by receptor-mediated endocytosis, micropinocytosis, phagocytosis, caveolin or clathrin-dependent uptake, or direct fusion with the plasma membrane (5). The signaling effects of energetic-stress induced EVs from metabolic cells on target cells can be characterized as 1) a warning, 2) collateral damage, 3) protective, or 4) recruitment. Warning EVs originate in a sensor cell that is undergoing energetic stress and provides a chemical signal to receiving cells to initiate preparation for the potential spread of the stress stimuli. EVs induce collateral damage when they spread disease from one cell to another. EVs can also, carry protective molecules that alleviate stress in receiving cells. Lastly, EVs can be released by stressed cells that recruit resources from surrounding cells to aid in stress resolution.

Warning Signal: the benefits of the by-stander effect

Threats to a single cell are likely potential threats to all cells in a multicellular organism. The cell types that are the most sensitive to a given stress can act as “sensors”, relaying information to other cells about rises in stress stimuli (Figure 2A). The adipocyte is a prime example of a sensor cell. The function of the adipocyte is to sense incoming calories and modulate systemic metabolism and feeding behavior accordingly through secretion of adipokines (1). The adipocyte converts excess calories to neutral lipid for storage in the adipocyte lipid droplet. This storage function is essential to protect other organs from lipotoxicity. This is the most evident in lipodystrophy, where adipocytes do not develop, resulting in hyperlipidemia, organ inflammation and oxidative stress. Lipodystrophy is associated significant comorbidities including non-alcoholic fatty liver disease and cardiovascular disease (125). In obesity, adipocytes become dysfunctional, resulting in many of the same comorbidities. Interestingly, within 1 day of high fat feeding in mice, ROS production in adipose tissue is significantly increased (38). During the first 3 days of high fat feeding adipocyte mitochondria become uncoupled, which enhances oxygen consumption and stimulates a pseudo-hypoxia state where HIF1α is activated (66, 75). Under both acute and chronic high fat feeding adipose tissue releases more sEVs (38). This suggests adipocytes are particularly sensitive to the effects of nutrient overload and respond by releasing EVs. There are potentially cell populations in each tissue that play this sensing function and release EVs that transfer stress traits, including damage and stress responses, to surrounding cells. This effect has been termed “the bystander effect” (126). Based on available data, the bystander effect can have beneficial effects by enhancing cell defenses to foster resilience, or it can cause excessive damage and apoptosis. The former effect can be thought of as a warning signal, a mild stress to get the receiving cell ready for a potentially lethal insult.

Figure 2.

Categories describing the effect of energetic stress-induced EVs on receiving cells. A. Cells that are sensitive to a specific stress, such as adipocytes, can send out warning signals via EVs that enhance the resilience of the receiving cell. B. Stressed cells can also release EVs contain damage and pathology-spreading molecules. C. EVs from support cells, particularly stem cells, can produce EVs that contain inherently protective molecules such as antioxidant enzymes. These EVs protect organs such as the kidney, brain, and liver from oxidative stress and aging. D. EVs from stressed cells have a capacity to recruit needed resources from other cells. This is exemplified by cancer cells or myocytes using EVs to request fatty acids from adipocytes as a source of energy. In addition, these EVs signal to attract immune cells, in the case of obese adipose tissue, and endothelial cells, in the case of a tumor, obese adipose tissue and exercising muscle.

One of the most surprising stress signals is the packaging of oxidatively damaged, but respiration competent mitochondrial fragments into EVs. Obesity or palmitate treatment results in increased incorporation of damaged mitochondria into white adipocyte sEVs, which travel to the heart and induce a burst of ROS (38). This results in compensatory antioxidant signaling in the heart that protects cardiomyocytes from acute oxidative stress. A single injection of sEVs from energetically stressed adipocytes was demonstrated to limit cardiac ischemia/reperfusion injury in mice (38). These mitochondrial fragments have been detected in isolated sEVs from the serum or plasma of obese mice and humans (38). This suggests, that in the obese condition, where adipocytes become dysfunctional, sEV-associated mitochondria are relaying a pro-oxidant signal to other organs with the outcome of stimulating cell defense pathway preparation for a future insult. Such an insult could be an ischemic event in the heart, as is common in obese individuals, but could also be a warning of the lipotoxicity to come as adipocyte function continues to deteriorate over the progression of obesity. Although the outcome of adipocyte-derived sEV-associated mitochondria is a warning signal, it likely originates as a cell-autonomous protective mechanism. The “mitochondrial particles” where found to be mitochondrial derived vesicles (MDVs). MDV formation is a mitochondrial quality control mechanism that is independent of autophagy and mitophagy machinery yet is reliant on the actions of Pink and Parkin proteins (127). MDVs are formed in response to oxidative stress, allowing for the sequestration of oxidatively damaged mitochondrial components into a budding structure that is released as an MDV (127). MDVs are transported to the late endosome and eventually to the lysosome, which completes this quality control pathway (127). However, escape of MDVs from the endolysosomal system as EVs has been shown (38, 128). It appears that most of the mitochondria released from the adipocyte is taken up by macrophages in the tissue (129). Brown adipocytes also release damaged mitochondria in large EVs in response to cold-induced mitochondrial stress or treatment with FCCP, a mitochondrial uncoupling compound (130). These extruded mitochondria are also taken up by macrophages in the tissue. If the macrophages fail to clear these EVs brown adipocytes can take up their own damaged mitochondria, which impairs the thermogenic capcity of the cell (130). Furthermore, a similar but distinct mechanism for the extrusion of damaged mitochondria has been reported in cardiomyocytes. Damaged mitochondria are released in the autophagy-dependent EV subtype, exophers, which are taken up and degraded by cardiac macrophages (10). This “outsourcing of mitophagy” to macrophages by EVs has also been described in in mesenchymal stem cell (MSC) mitochondrial maintenance (131). Therefore, the mitochondria that enter circulation were likely destined for degradation by tissue resident macrophages but escaped this fate through unknown mechanisms. Interestingly, in the obese state mitochondria (naked or EV-enclosed) seem to display a preference for leaving white adipose tissue instead of being taken up by cells in the tissue (132). A major cause of this is the long-chain fatty acids found in a lard-based high fat diets inhibit the uptake of extracellular mitochondria by tissue-resident macrophages and so mitochondria are diverted into the blood (132). This is not true in brown adipose tissue where lard-based HFD increased the transfer of adipocyte mitochondria to tissue macrophages(132). As such, the origin of a mitochondria-mediated systemic warning signal may be tissue specific and a beneficial biproduct of cell autonomous quality control mechanisms. Although, this is likely not an obesity-specific stress response as mitochondrial protein content is also increased in circulating sEVs of melanoma patients (133).

Other stresses are well known to cause a bystander effect through EVs. Cells exposed to thermal stress release EVs that are sufficient to induce DNA damage in surrounding cells (134). Although this does result in some cell death, the remaining cells are more resistant to subsequent stresses (134). ER stress is also potentially transmitted between cells through the transfer of spliced XBP1 mRNA, a major initiating transcription factor in the ER stress response (135). A similar effect can be seen in response to oxidative stress. Oxidative stress during dorsal root ganglia damage induces macrophages EV production that transfers the ROS-generating enzyme NADPH oxidase 2 (NOX2) to the injured neuron (136). NOX2 stimulates P13K/AKT signaling in neurons through oxidative inhibition of PTEN, resulting in axonal regeneration (136).

There is no warning signal more apt than that of a dying cell. Apoptotic bodies or apoptotic EVs (ApoEVs) are produced by membrane blebbing during apoptosis and carry bioactive molecules (137). ApoEVs can be considered a warning signal, but it does not seem to be through perpetuating damage. Instead of causing toxicity, as may be expected, ApoEVs strongly promote growth, proliferation, and survival of surrounding cells. For example, dying endothelial cells release ApoEVs that are taken up by endothelial progenitor cells, where their contents stimulate proliferation and differentiation (138). Macrophage LPS-stimulated ApoEV release had a similar proliferative effect on epithelial cells through the transfer of miR-221 and miR-222 (139). Likewise, ApoEVs from dying epithelial cells also promote expansion of stem cells through the transfer of Wnt8a (140). Bone marrow mesenchymal stem cells (MSCs) have also been shown to engulf ApoEVs from circulation resulting in activation of the Wnt/β-catenin pathway, which plays a critical role in self-renewal and differentiation of MSCs (141). Liu et. al. demonstrated that dying transplanted MSCs in a myocardial infarction model released ApoEVs that induced lysosomal activity and autophagy in endothelial cells (142). This resulted in survival and proliferation of endothelial cells to improve cardiac recovery (142). The final example comes from the bone remodeling literature. Osteoclast ApoEvs induce osteoblast differentiation by activating the PI3K/AKT/MTORC1 pathway (143). Although the studies on ApoEVs are not extensive, the available results are consistently pointing toward the inherent nature of these EVs as warning signals to stimulate cell survival and tissue regeneration.

Collateral damage: the dark side of the bystander effect

It is likely that all warning signals may become pathogenic over time or throughout disease progression, particularly those that originate as damaged cellular components (Figure 2B). Furthermore, EVs have been shown to pass pathology-promoting traits between cells (144). Adipose tissue EVs are one of the best examples of a warning signal that can turn pathogenic. As mentioned above, mitochondrial stress simulates adipocytes to release sEVs that act as a beneficial warning signal to the heart (38). In contrast, EVs from adipocytes harvested from diabetic mice, on a chronic high fat diet, were shown to exacerbate myocardial ischemia/reperfusion injury in lean, non-diabetic mice (145). Mechanistically, adipocyte sEVs carried miR-130b-3p, which targeted AMPK, Birc6 and UCP3 in cardiomyocytes, thereby blunting protective pathways. Adipose tissue EVs also target the hippocampus, and if derived from diet-induced obese mice or humans with type 2 diabetes, these EVs promote synaptic damage and cognitive impairment though transfer of miRNAs (146). Adipocyte EVs from obese mice or adipocytes treated with high glucose and fatty acids to stimulate insulin resistance also have a significantly negative impact on surrounding adipose tissue cells. Insulin resistant adipocyte sEVs and large EVs can induce monocyte migration and differentiation into pro-inflammatory M1-like macrophages (86, 147–150). Concomitantly, adipocyte sEVs suppress the polarization of macrophages into M2-like, anti-inflammatory macrophages via miR-34a (151). Insulin resistant adipocyte or obese adipose tissue EVs can also provoke pathologic angiogenesis and increase atherosclerotic plaque burden (150, 152). Lastly, metabolically stressed adipocytes or obese adipose tissue EVs can induce inflammation and insulin resistance in muscle and liver cells (153–155), as well as dysregulate fibrogenic signaling in hepatocytes (155, 156). This is also true in humans, where both plasma and adipose tissue-derived sEVs from obese humans with nonalcoholic fatty liver disease induce insulin resistance in myotubes and hepatocytes (157). The insulin resistance phenotype is generated by direct effects of adipocyte EVs on myocytes and hepatocytes but may also be caused indirectly through effects on pancreatic β-cell insulin secretion. Gesmundo et. al. reported that healthy 3T3-L1 adipocyte EVs enhanced the proliferation, survival, and insulin secretion activity of the INS-1E β-cell line and human islets (158). In contrast, EVs from inflamed adipocytes promoted β-cell death and dysfunction in the presence or absence of pro-inflammatory cytokines (158). Finally, induction of ER stress in adipocytes stimulates the release of EVs that contain also-keto-reductase 1B7 (Akr1b7), a mediator of lipid metabolism is the liver (159). These adipocyte Ark1b7- containing EVs cause hepatic steatosis, inflammation in fibrosis in the liver, all of which lead to NASH in mice (159).

Adipocytes are not the only cells that spread pathology through EVs in obesity. M1-like macrophages from obese, inflamed adipose tissue can produce EVs that induce whole body insulin resistance, and blunt insulin-stimulated AKT phosphorylation in myocytes and hepatocytes at least partially through transfer of miR-155 (160). MiR29a was also found to be a major insulin resistance-promoting RNA from obese adipose tissue macrophage EVs (161). These effects are thought to be dependent on the miRNA cargo of adipocyte EVs. Therefore, the effects of EVs from adipose tissue cells in obesity and type 2 diabetes contribute to the severe insulin resistance and inflammation that is characteristic of the disease.

Hepatocyte EV actions also fall into the category of a beneficial warning signal that turns pathologic over the progression of obesity. Early during the onset of obesity, hepatocytes release miR-3075-containing EVs that promote insulin sensitivity in metabolic cells by downregulating fatty acid 2-hyroxylase (96). However, in chronic obesity hepatocyte EVs promote insulin resistance through activation of pro-inflammatory macrophages (96). Likewise, EVs produced by cultured hepatocytes in lipotoxic conditions to mimic obesity attract macrophages, activate inflammatory pathways in macrophages and endothelial cells and promote angiogenesis (99, 162–164). In addition, HepG2 cells treated with fatty acids and the HIF1α-stabilizing compound cobalt chloride released EVs that stimulated a fibrosis-related transcriptional profile in hepatic stellate cells (165). Therefore, the data suggests that during chronic obesity lipid laden hepatocytes produce EVs that promote local inflammation and fibrosis, key features of NAFLD.

Protection: guardian EVs

In some cases, EVs can be powerfully protective for receiving cells under stress, not through hormesis, as with warning signals, but through the transfer of inherently protective molecules like antioxidants and pro-survival signals (Figure 2C). These guardian EVs generally originate from stem cells, a type of cell that is integral to tissue maintenance and regeneration following injury. MSCs and adipose tissue-derived stem cells (ADSCs) are important mediators of protective EV-mediated signaling. Aging is a physiological process where regeneration of tissues is impaired due to a loss of stem cell function, metabolic disturbances and increased cellular senescence (166). Senescent cells are post-mitotic, resistant to apoptosis, metabolically active and secrete an array of molecules that promote inflammation, tissue damage, and impede stem cell activity (166). Interestingly, treatment of aged, senescent stem cells with young human MSC-derived EVs rescued stem cell function, and dramatically increase the lifespan and healthspan of the short-lived ERCC1-deficient mice (166). This effect was attributed to a reduction in the number of senescent cells allowing for improved tissue recovery from stress and augmented regeneration. MSC-sEVs also carry miR-146a, which suppresses Src phosphorylation in endothelial cells, resulting in reduced oxidative stress-induced senescence and enhanced angiogenesis (167). EVs can also correct the metabolic defects of aging. Yoshida et. al. demonstrated that circulating eNAMPT levels decrease with age, an enzyme that, in the blood, is exclusively associated with EVs. eNAMPT produces NAD⁺, a coenzyme that is essential for ATP production, and the levels of which are reduced with age. Injection of eNAMPT-containing EVs from young mice or from eNAMPT-overexpressing adipocytes significantly increased the lifespan of mice, corrected metabolic defects associated with aging, improved exercise tolerance and increased NAD⁺ biosynthesis in hypothalamic neurons (168). Any beneficial effect of endogenous EV during aging is likely hindered as EV levels decline with age and carry pathological, not protective cargo (169, 170).

MSCs and ADSCs EVs have been shown to have potent antioxidant characteristics. MSC-derived EVs protect hippocampal neurons from the oxidative stress associated with amyloid-β oligomers in Alzheimer’s disease (171). These protective EVs contain catalase, an antioxidant enzyme that converts the pro-oxidant hydrogen peroxide to oxygen and water. MSC-EVs are also enriched with antioxidant miRNAs that increase catalase, superoxide dismutase, and glutathione peroxidase activities in cultured cells to reduce ROS generation and protect hippocampal neurons from lasting effects of seizure damage (172). Furthermore, EVs from MSCs have been shown to reduced inflammation, oxidative stress, and apoptosis in a mouse model of colitis via suppression of ROS production and increased antioxidant capacity in the colon (173). MSC-sEVs protect the kidney from ischemic injury by stimulating the antioxidant transcription factor NRF2 (174). Similarly, MSC-EVs protect the heart during ischemia by promoting angiogenesis, reducing ROS, attenuating fibrosis, correcting cellular bioenergetics and modulating inflammation (175–179) ADSC-derived EVs have also been shown to attenuate cardiomyocyte death in response to oxidative stress (180). In the lung, ADSC EVs have been shown to attenuate the ROS produced by environmental toxin PM_2.5 by upregulating NRF2 (181). These antioxidant features of ADSC EVs have also been implicated in protection against ultraviolet B-induced damage of the skin, and attenuation of inflammation and oxidative stress in macrophages (182, 183).

It is important to note that most of the studies that demonstrate a protective effect of EVs were conducted using healthy, unstimulated, or pharmacologically treated stem cells. These studies provide interesting ideas for therapeutic approaches and may provide evidence that stem cells in distal organs, that are not affected by a specific stress, may provide a beneficial signal to affected organs. However, stem cells are important for local maintenance of tissue homeostasis and thus, the actions of EVs from MSCs under the same stress as the rest of the cells in a microenvironment are important to consider. For example, an ischemic event may affect large portions of an organ. Several studies have determined that stem cells subjected to hypoxia in vitro still have protective effects on in vivo ischemic disease. Uptake of EVs from hypoxia conditioned ADSCs or MSCs attenuates epithelial oxidative stress and promotes angiogenesis respectively (184, 185). A similar result was reported with in vivo studies. ADSCs preconditioned with anoxia limited myocardial ischemia/reperfusion injury when injected into mice (186). Lastly, EVs from hypoxia-treated ADSCs promoted recovery of renal function following ischemic injury by promoting angiogenesis and mitigating oxidative stress (187). These studies suggest that either healthy or stressed stem cells release EVs that have protective effects on receiving cells.

Stem cells and other support cells can also transfer healthy mitochondria to surrounding cells that may be deficient. This occurs tunneling nanotubules, cell fusion, GAP junction and EVs (188). Although there are few examples of healthy mitochondria transfer between cells via EVs, those that we have suggest this process can be powerfully protective for the receiving cell with mitochondrial deficiencies. In the lungs, EV-associated mitochondria are transferred from MSCs to alveolar cells during LPS stress (189). The result is increased ATP production in alveolar cells, increasing survival rates in mice following LPS lung instillation. Similarly, cancer-associated fibroblasts release EVs that contain mitochondria. These mitochondria are taken up by dormant breast cancer cells resulting in stimulation oxidative phosphorylation and escape from dormancy (190). A recent study has demonstrated the presence of what seems to be healthy, respiration competent mitochondria in human plasma (191). By electron microscopy, this study demonstrated that this population of mitochondria was not enclosed in a phospholipid membrane and therefore not associated with EVs. It would be interesting to identify any function differences between extracellular naked mitochondria and those in EVs. Further work is required to determine which conditions or which cell types are prone to releasing healthy mitochondria, which may have protective outcomes, vs damaged mitochondria, which may act as a warning signal or pathology-spreading process as discussed above.

EVs released by muscle during exercise have also been shown to display guardian functions. The systemic effects of exercise have long been considered potently beneficial to the health of rodents and humans (192). Although the mechanisms are not understood, both the release of hormones called mitokines and EVs from muscle during exercise is thought to play a role (192). Several studies have now demonstrated increased circulating EVs following exercise that are enriched with skeletal muscle-specific miRNAs and proteins in humans and mice (87, 89, 193–195). Furthermore, exercise intensity positively correlates with total circulating EVs in rats (196). However, it is important to note that not all exercise-induced EVs in circulation are from muscle cells. In addition to myocytes, platelets, endothelial progenitors, endothelial cells, and leukocytes also contribute EVs to circulation during exercise (197). Therefore, it is challenging to ascribe a beneficial function to EVs from any one cell type, so many studies focus on the total blood EV population. Vechetti et. al. compiled a list of mRNAs identified in the literature to be associated with exercise induced EVs and ran a prediction analysis to identify target genes (192). Gene ontology analysis showed that the two most significant biological processes were response to reactive oxygen species and insulin secretion (192). Exercise has been reported to improve β-cell function and increase antioxidant enzyme activity in circulation, suggesting EVs may mediate these protective effects (192). Furthermore, exercise-induced EV cargo is thought to promote angiogenesis, modulate immune function, enhance insulin sensitivity and stimulate regeneration (192, 195). The overall beneficial effects of circulating EVs during exercise is exemplified by a study, which demonstrates that an intramyocardial injection of EVs isolated from the plasma of exercised mice significantly protects mice from myocardial ischemia/reperfusion injury (198). This occurs through activation of the ERK1/2 pathway and Hsp27, by mitigating oxidative stress in cardiomyocytes (198).

EVs released from the liver following acute high fat feeding can also have a have a protective function. Zhao et. al. demonstrated that liver triglyceride content is increased within 6 hours of high fat feeding, whereas adipose tissue lipid storage lags to 12 hours (95). During those 6 hours of lipid overload hepatocytes produce more EVs that stimulate differentiation of preadipocytes and lipogenesis resulting in increased fat pad mass. This form of adipose tissue expansion, where new adipocytes are recruited, is considered healthy adipose tissue remodeling that protects other organs from lipotoxicity (199). Recently, Jung et. al. demonstrated hepatocytes treated with high glucose release EVs containing transmembrane 4 L six family member 5 (TM4SF5), which stimulates brown adipose tissue gluocose uptake (200). Therefore, hepatocyte EVs may not directly protect the receiving cell, but the signaling outcome in white and brown adipocytes results in protection of the whole organism from both lipotoxicity and glucose toxicity.

Recruitment: calling the calvary

The function of EVs in the recruitment of resources is greatly understudied, but there are examples in the literature. Under physiological and pathological conditions, cells request energy from adipocytes via EVs (Figure 2D). During resistance training, the mechanical load stimulates skeletal muscle to release EVs into circulation, which are preferentially taken up by the visceral adipose tissue (201). miR-1 in these muscle EVs stimulates lipolysis in adipocytes by repressing Tfap2α expression and thereby stimulating β adrenergic signaling. The fatty acids liberated through this process are utilized by the muscle tissue but also systemically for energy. Exercise may be the only physiological example of this phenomenon, however, cancer cells that originate in a wide range of tissues display an aptitude for energy recruitment from adipocytes. As cancer cells are under intense energetic stress because of the energetic burden of proliferation, adipocyte-derived lipids are considered highly advantageous to cancer cell growth and survival (202). Lewis lung carcinoma cells (LLC) shed EVs that stimulate adipocyte lipolysis in vitro and in vivo (203–205). Mechanistically, these EVs carry parathyroid hormone-related protein (PTHrP), which stimulates lipolysis by activating the protein kinase A (PKA) signaling pathway (203). LLC-derived EVs can also carry IL-6, which promotes lipolysis through activation of the STAT3 pathway (205). Pancreatic cancer cells stimulate adipocyte lipolysis through EVs carrying adrenomedullin (AM; (206)). EV-associated AM functions through its receptor, ADMR, to activate hormone sensitive lipase (HSL) via p38/ERK1/2 signaling. Likewise, miRNAs in EVs from breast cancer cells rewire adipocyte metabolism, which results in increased release of fatty acids, pyruvate, lactate, and ketone bodies (207). Therefore, EVs can signal to recruit energy from adipocytes during nutrient stress.

EVs can also draft support cells into areas in the tissue where they are required. The immunomodulatory effect of EVs can contribute to disease, however, the purpose of this function is to recruit immune cells to clean up toxic material in damaged tissue or fight off infection. In obesity, there is substantial infiltration of pro-inflammatory macrophages into adipose tissue, which attempt to repair the tissue following the death of adipocytes. Although this can have detrimental effects, inflammatory processes are essential for proper adaptation of the adipose tissue to nutrient overload (208). EVs from palmitate-stressed adipocyte have been described as “find me” signals which recruit monocytes and macrophages to the adipose tissue (94). This effect has also been demonstrated in several other studies (86, 147–150). Immune cells are not the only support cells that are responsive to EVs. EVs produced by hypoxic MSCs stimulates new blood vessel formation which is required for re-supplying the effected tissue with nutrients and oxygen (185). This is a key process in regeneration following ischemic injury. Likewise, cardiomyocytes release EVs during ischemia that promote angiogenesis partially though the transfer of miR-222 and miR-143 to endothelial cells (209). Hypoxia is also a common condition in expanding tumors. Cancer cells release EVs that stimulate angiogenesis to alleviate hypoxia and supply the tumor with energy (210). The peudo hypoxia and energetic stress induced in muscle during exercise is thought to stimulate EVs that have angiogenic potential (211). Interestingly, the liver produces EVs during exercise that promote angiogenesis in skeletal muscle by miR-122–5p (212). Therefore, EVs released under energetic stress can function as a beacon to recruit energetic resources or support cells.

Conclusion and Future Directions

The true complexity of intercellular metabolic regulation has now become apparent in light of the burgeoning field of EV research. EV release is stimulated by multiple stress pathways and as such, EVs are key mediators of pathogenesis in many disease states. A connecting feature of all major stress pathways is energetic stress, which is either as a cause of stress signaling, or the consequence. The molecular characteristics of energetic stress include high AMP, ADP, NAD⁺, and ROS levels. Each of these stress signals directly and robustly regulate the activity of enzymes that sense the energetic status of the cell: AMPK, MTORC1, SIRT1 and HIF1α. These sensors, in turn modulate EV production and release through mechanisms that are not well understood. A likely candidate effector pathway for stress induced EV release is regulation of the endolysosomal system, a pathway that is downstream of all the mentioned sensor enzymes and major cellular stress pathways. Once released, stress induced EVs signal to regulate the metabolism of surrounding cells or distal organs through surprisingly complex mechanisms. Energetic stress induced EVs function to 1) send a chemical warning to prepare the metabolic infrastructure of the receiving cell for impending insults, 2) pass on disease traits to surrounding cells, 3) protect surrounding cells from stress, or 4) recruit energy and support cells to the EV-producing cell.

There are multiple areas where we lack knowledge of how EVs signal in metabolism. The first is how EV release is regulated by energetic stress. Specifically, how stress signals, sensor enzymes and effector pathways interact to produce a net positive release of EVs. This will likely have to be studied in a cell-type and stress-type specific way. In addition, we have to understand how stress-stimulated EVs signal to regulate metabolism in both physiological and pathologic conditions. This will mean defining networks of cells that communicate via EVs, profile cell type-specific EVs, and determine the functional outcome of these signaling axis in health and disease. Examining these novel EV-mediated signals may provide new possibilities for targeted treatment of metabolic disorders in the years to come.

Didactic Synopsis.

• Extracellular Vesicles (EVs) are nano-sized vesicles released by all cells that carry signaling molecules like proteins, lipids, metabolites, and RNA species. EVs can have a strong effect on the function of a receiving cell.

• EVs from metabolically important cells can regulate metabolism in target cells, however, little is known mechanistically. Of particular importance is the need to understand the role EVs have in promoting pathology in obesity and type 2 diabetes.

• EV release is stimulated by major forms of cellular stress that all coincide with or are driven by energetic stress. Energetic stress is characterized by the inability of the cell to maintain adequate ATP levels due to reduced availability of oxidizable substrates or mitochondrial dysfunction.

• Major nutrient sensing proteins like AMPK, MTORC1, sirtuins and HIF1α have been shown to influence EV release. Downstream of most of these pathways is autophagy, which too regulates stress-induced EV release.

• Stress-induced EVs can act as a warning signal to other cells, propagate disease, protect the receiving cell from stress or recruit support cells to the area.

• Understanding the regulation of EV production under stress and the cell type specific signaling that stress-induced EVs elicit may open opportunities for new therapeutic targets to combat the pathologies associated with obesity.