Extracellular vesicles as potential biomarkers for alcohol- and drug-induced liver injury and their therapeutic applications

Abstract

Extracellular vesicles (EVs) are small membranous vesicles originating from various cells and tissues, including the liver parenchymal hepatocytes and nonparenchymal cells such as Kupffer and stellate cells. Recently, the pathophysiological role of EVs, such as exosomes and microvesicles, has been increasingly recognized based on intercellular communications. These EVs travel through the circulating blood and interact with specific cells and then deliver their cargos such as nucleic acids and proteins into recipient cells. In addition, based on their stabilities, circulating EVs from body fluids such as blood, cerebrospinal fluid, urine, saliva, semen, breast milk and amniotic fluids are being studied as a valuable source of potential biomarkers for providing information about the physiological status of original cells or tissues. In addition, EVs are considered potential therapeutic agents due to their property for intercellular communications between different cell types within the liver and between various organs through transfer of their cargos. In this review, we have briefly described recent advances in the characteristics and pathophysiological roles of EVs in alcoholic liver disease (ALD) or drug-induced liver injury (DILI) and discuss their advantages in the discovery of potential biomarkers and therapeutic agents.

1. Introduction

In the last two decades, numerous laboratories reported that intercellular communication is an essential hallmark of multicellular organisms and can be mediated through direct cell–cell interaction with transfer of the cellular components in circulating extracellular vesicles (EVs) that are secreted from various cells (Colombo, Raposo, & Thery, 2014; Raposo & Stoorvogel, 2013; Yanez-Mo, et al., 2015). EVs are enclosed with a lipid bilayer and contain various types of functional molecules, such as distinct proteins, RNA species (including mRNA, miRNA, IncRNA, and other RNA), DNAs (mtDNA, ssDNA, dsDNA), and lipids (Colombo, et al., 2014; Xu, Greening, Zhu, Takahashi, & Simpson, 2016; Zaborowski, Balaj, Breakefield, & Lai, 2015) that are likely to reflect the origin and pathophysiological status of different cell types. EV-mediated intercellular communications are observed in a variety of cellular processes as exemplified in tumor development (J. E. Lee, et al., 2016; Moon, Lee, Cho, Lee, Chae, et al., 2016; Moon, Lee, Cho, Lee, Jung, et al., 2016) and progression through immune suppression, angiogenesis, and metastasis (Costa-Silva, et al., 2015; Hoshino, et al., 2015; Peinado, et al., 2012). The surface receptor proteins of EVs, such as CD63 and Integrin isoforms, allow their targeting and capture by the recipient cells, which can then receive and incorporate extra proteins, lipids, and even genetic cargos carried by these EVs, resulting in alterations of their physiological states (Buzas, Gyorgy, Nagy, Falus, & Gay, 2014; van der Pol, Boing, Harrison, Sturk, & Nieuwland, 2012). Consequently, the possibilities of using these EVs as a source of potential biomarkers and therapeutic agents in various liver diseases are currently being evaluated.

Alcoholic liver disease (ALD), caused by long-term consumption of excessive amounts of alcohol (ethanol), is a major cause of morbidity and mortality globally and represents one of the most preventable common diseases (O’Shea, Dasarathy, McCullough, Practice Guideline Committee of the American Association for the Study of Liver, & Practice Parameters Committee of the American College of, 2010). The pathogenesis of ALD is characterized by a progressive process of hepatic accumulation of lipids (steatosis), inflammation (steatohepatitis), and in some individuals, fibrosis and cirrhosis before liver failure or hepatocarcinoma (Mills & Harrison, 2005; Stewart, Jones, & Day, 2001). ALD with hepatitis can be caused by direct toxic effects of alcohol on the liver cells as well as indirectly through increased gut leakiness and impaired function of adipose tissues which can release bacterial product endotoxin such as lipopolysaccharide (LPS) and free fatty acids with elevated proinflammatory adipokines, respectively, into the blood and ultimately affect the liver. These molecules can cause inflammatory liver injury with increased endoplasmic reticulum (ER) stress and mitochondrial dysfunction. In fact, it is known that the development of advanced liver disease such as fibrosis/cirrhosis can be accelerated by the elevated levels of circulating endotoxin through gut leakiness (Bode, Kugler, & Bode, 1987).

Drug-induced liver injury (DILI) can be caused by many clinically-used drugs, such as acetaminophen (APAP), isoniazid, and halothane in humans (W. M. Lee, 1995, 2003), as well as potentially hepatotoxic agents, including LPS, carbon tetrachloride (CCl₄) and d-galatosamine (DGAL) in experimental models. Incidences of DILI depend on the doses of the potentially hepatotoxic agents and physiological states of the hosts such as poor nutrition, obesity, prediabetic conditions, and chronic alcohol drinking, all of which are known to decrease cellular antioxidants, including glutathione (GSH). Acute liver injury, either by binge alcohol exposure or DILI, is usually accompanied with necroapoptosis of hepatocytes through stimulation of the cell death associated stress-activated protein kinases such as c-Jun N-terminal protein kinase (JNK), p38K and mitochondrial dysfunction through various post-translational modifications of mitochondrial proteins (Abdelmegeed & Song, 2014; Jang, et al., 2015; Song, et al., 2014). In addition, DILI is one of the major reasons for withdrawal of drugs from the market or during clinical testings, and thus it is of prime concern to both the regulatory agencies, including the US Food and Drug Administration (FDA), and pharmaceutical companies as well as consumers (W. M. Lee, 2003; Navarro & Senior, 2006). Recently, the emerging field of omics technology has been greatly advanced by quantitative proteomics technologies and integrative metabolomics along with various gene expression arrays and advanced (deep) DNA/RNA sequencing technologies. Several reports demonstrated that omics technologies have been used to find the specific diagnostic and prognostic biomarkers and to understand toxic signaling pathways for prevention and treatment of DILI (Cho, Kim, & Baek, 2012; Cho, et al., 2013; Cho, Singh, et al., 2012) and other liver diseases, including ALD and nonalcoholic fatty liver disease or inflammatory hepatitis (NAFLD/NASH) (Ganz & Szabo, 2013; Szabo, 2015; Szabo & Petrasek, 2015).

Currently, plasma or serum alanine and aspartate aminotransferases (ALT and AST, respectively) are being frequently used as surrogate liver disease markers to determine the degree of liver injury by alcohol or many potentially hepatotoxic drugs/agents. The plasma ALT/AST levels positively correlate with liver histology data, although a few exceptions exist. However, it is known that the serum ALT/AST levels also increase in other organ injury, such as skeletal muscle, and that their assessments require fresh blood samples due to relatively short half-lives (Nathwani, Pais, Reynolds, & Kaplowitz, 2005), depending on the species (Ramaiah, 2007). Recently, new approaches have been introduced to identify more specific and sensitive biomarkers for various liver diseases. These potential biomarkers include certain genes (Cui & Paules, 2010; Li, Doiron, Patterson, Gonzalez, & Fornace, 2013), microRNAs (Bala, et al., 2012; Cho, Kim, Lee, & Baek, 2017; Starkey Lewis, et al., 2011), and protein profiles (Bell, et al., 2012; Cho, Mezey, & Song, 2017; Saha, et al., 2017; Torrente, Freeman, & Vrana, 2012) in human patients and experimental models of ALD and DILI. In this review, we highlight and discuss the latest research results with liver-derived plasma EVs as potentially non- or minimally-invasive biomarkers for ALD and DILI.

2. Definition and size of EVs

The circulating EVs display a range of different sizes (40 – 1,000 nm in diameter) and are generally known as exosomes, microvesicles, ectosomes, or microparticles (Kalra, et al., 2012; Lotvall, et al., 2014; Momen-Heravi, et al., 2013; Raposo & Stoorvogel, 2013; Stoorvogel, Kleijmeer, Geuze, & Raposo, 2002; Yanez-Mo, et al., 2015). Currently, EVs are classified into three major groups based on their cellular biogenesis: exosomes, microvesicles, and apoptotic bodies (Fig. 1). Exosomes represent small vesicles of endosomal pathway secreted from various cells and their sizes usually range from 40 nm to 150 nm in diameter (Colombo, et al., 2014). Microvesicles are demarcated vesicles generated by budding of the plasma membrane. The size of microvesicles is estimated to range between 50 nm and 1,000 nm in diameter (Lemoinne, et al., 2014). Apoptotic bodies or other vesicles, formed by large-scale plasma membrane blebbing during apoptosis, also fall into the category of EVs. Apoptotic bodies are usually greater than 500 nm in diameter as they represent cell fragments (Gould & Raposo, 2013; Kowal, Tkach, & Thery, 2014). Recently, unbiased proteomics studies highlighted the heterogeneity of isolated EV populations and commonality of shared protein markers among EVs isolated on the basis of size alone and thus improved their classification (Kowal, et al., 2016). However, the heterogeneity of isolated EVs reflects the lack of standardized isolation techniques mainly because different terminologies have been used in the past. Consequently, the distinction of subtypes of those vesicles has become difficult (Szabo & Momen-Heravi, 2017). The positional reports on standardization procedures for EV purification and minimal requirements for EV definition and function have been recently described (Lotvall, et al., 2014; Witwer, et al., 2013).

Fig. 1. Release of exosomes, microvesicles, and apoptotic bodies.

Exosomes originate from the MVBs, which later fuse with the cellular membrane to release exosomes into the extracellular milieu. Microvesicles and apoptotic bodies bud directly from the plasma membrane.

3. The composition and biogenesis of EVs

3.1. Composition of EVs

EVs contain common exosomal marker proteins, cell-type specific proteins and nucleic acids, including mRNAs, microRNAs (miRNAs) and other non-coding RNAs, and lipids where the specific composition of the EVs depends on the functional state of the cells (Bang & Thum, 2012; Sun, et al., 2013; Vlassov, Magdaleno, Setterquist, & Conrad, 2012). In fact, various databases, such as ExoCarta (Mathivanan & Simpson, 2009), Vesiclepedia (Kalra, et al., 2012), and EVpedia (D. K. Kim, et al., 2013) include published compositional data of both exosomes and microvesicles using different isolation and characterization methods. For instance, the protein contents of EVs from different cells were analyzed by immunoblotting and/or mass-spectral analysis-based proteomics approaches. EVs should be devoid of cellular contaminants, such as serum proteins and protein components of intracellular compartments, since they are not in contact with EVs (Raposo & Stoorvogel, 2013). However, EVs from different cell types may contain endosome-associated proteins that are involved in the biogenesis of multi-vesicular bodies (MVBs) (e.g., Alix and TSG101) and cytoplasmic heat shock proteins (Hsc70 and Hsp90). Exosomes derived from B cells are enriched for all tetraspanins analyzed (CD37, CD53, CD63, CD81, and CD82) by 7- to 124-fold (Escola, et al., 1998). This protein signature can be considered as a genuine marker for both the plasma membrane and early endosomes (Escola, et al., 1998). Other reports showed that tetraspanins are abundantly present in tumor-derived small vesicles that aid intercellular communication (Zoller, 2009). Although tetraspanins enriched with membrane domains display some lipid raft-like properties, they show resistance to detergent solubilization (Hemler, 2008). EVs are also enriched with proteins that are associated with lipid rafts, including glycosylphosphatidylinositol-anchored proteins and flotillin (Thery, et al., 1999; Wubbolts, et al., 2003).

EV subpopulations can be distinguished based on their physiological properties such as size and density, despite some heterogeneities. Therefore, specific markers are needed to identify the different subtypes of EVs. Kowal et al., described that large-, medium-, and small (exosomes)-sized EVs can be isolated by different ultracentrifugation such as low, medium, and high speed (Kowal, et al., 2016). Among the small-EVs (exosomes), four subcategories can be defined depending on the enriched amounts of CD63, CD9, and CD81 tetraspanins and endosome markers. In detail, major histocompatibility complex (MHC) class I, class II, and heat-shock proteins 70 (HSC70/HSP73 and HSP70/HSP72) were presented all EV subgroups while GP96 and possible other ER-stress-associated proteins are mainly present in large-EVs. In addition, actin-4, mitofilin, and other mitochondria proteins were identified in both large- and medium-EVs but not in small-EVs. Interestingly, synteinin-1, TSG101, ADA10, and EHD4 were only identified in small-EVs.

Electron microscopy (EM) is the only method to visualize exosomes, since their size (30–100 nm) is below the limit of optical resolution and that individual EVs can be demonstrated by super resolution microscopy (Lo Cicero, Stahl, & Raposo, 2015). A major limiting factor in the characterization of EVs has been the absence of trustworthy methods for quantifying vesicle release. A reliable method to analyze small amounts of isolated EVs is an optimized FACS procedure based on the labelling of isolated vesicles with fluorescent probes (Nolte-‘t Hoen, et al., 2013; van der Vlist, Nolte-‘t Hoen, Stoorvogel, Arkesteijn, & Wauben, 2012). As a complement tool, the Nanoparticle Tracking Analysis (NTA) can be used based on automatical trackings of light scattering and determination of the nanoparticle sizes on an individual basis, although it is difficult to discriminate between small aggregates and vesicles (Jeppesen, et al., 2014).

3.2. Biogenesis and secretion of EVs

It is well recognized that MVs and exosomes arise from different biogenesis mechanisms, with microvesicles directly budding from plasma membranes. While exosomes are formed as intraluminal vesicles in the lumen of multi-vesicular endosomes (MVEs) formed by the invagination of the limiting membrane of late endosomes, a process that involves the segregation of cargo and pinching of a vesicle into the lumen. The microvesicles subsequently fuse with the plasma membrane and release their internal vesicle contents into the extracellular environment as exosomes. In fact, the endosomal sorting complex responsible for transport (ESCRTs) (Henne, Buchkovich, & Emr, 2011) or other components such as ceramide lipids (Trajkovic, et al., 2008) and tetraspanins (van Niel, et al., 2011) appear to be involved in exosome biogenesis. Exosome secretion involves the transport of MVBs to the cell periphery in preparation for fusion with the plasma membrane. Rab GTPases are involved in exosome secretion but the requirements for specific Rabs may differ depending on the cell type. A medium high-throughput screen using a selected shRNA library targeting Rab proteins revealed that loss of expression of Rab27a and Rab27b resulted in a 50% reduction in exosome secretion (Ostrowski, et al., 2010).

The microvesicles are defined as plasma membrane-derived vesicles protruding into the extracellular space and are clipped, similar to the final stage of cytokinesis (Muralidharan-Chari, Clancy, Sedgwick, & D’Souza-Schorey, 2010). Release of microvesicles from plasma membrane is currently correlated with the activation of acidic sphingomyelinase and ceramide generation as seen in glial cells (Tetta, Ghigo, Silengo, Deregibus, & Camussi, 2013). Other mechanisms for microvesicle shedding include: (i) lipid aggregation and asymmetry with phosphatidylserine; (ii) membrane curvature proteins to force membrane bending, and (iii) intracellular calcium signaling (Muralidharan-Chari, et al., 2010). Previously, adenosine diphosphate-ribosylation factor 6 was implicated in microvesicle shedding by tumor cells via extracellular signal-regulated kinase (ERK)-mediated phosphorylation of myosin light chain kinase (MLCK) (Muralidharan-Chari, et al., 2009). ESCRT proteins are reported to involve in microvesicle release as observed by the direct association between the arrestin-related protein and ESCRT-I, which led to plasma membrane shedding (Hurley & Odorizzi, 2012; Nabhan, Hu, Oh, Cohen, & Lu, 2012). The importance of plasma membrane protein aggregation in microvesicle biogenesis has been demonstrated by using the overexpressed candidate proteins. Overexpression of those proteins leads to higher-order oligomerization in the plasma membrane, resulting in increased protein targeting to the microvesicles (Fang, et al., 2007). In addition, stress responses in cells and intracellular Ca²⁺ levels can activate the release of microvesicles in some cells while in other cells, release is dependent on the pre-apoptotic activation of pro-caspase 3, which stimulates the Rho-associated protein kinase 1, resulting in the actin-myosin contraction and microvesicle discharge (Hirsova, et al., 2016; Povero, et al., 2013).

4. Usage of EVs as potential biomarkers of DILI or ALD

As a standard clarification, the NIH Biomarkers Definitions Working Group has defined a biomarker strictly as “a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacological responses to a therapeutic intervention” (Biomarkers Definitions Working, 2001). The Food and Drug Administration has already proposed rigorous standards for the term biomarker and criteria for validating a biomarker. In general, a biomarker may be the analyte or validate in a diagnostic test (Katz, 2004). EVs are considered as non-invasive (e.g., urine and saliva) and minimally-invasive (e.g., blood) sources of potential molecular biomarkers for the early detection, monitoring and evaluation of drug or alcohol response in various liver diseases, as well as a potential treatment tool (Vlassov, et al., 2012). This approach is based on the observations that EVs present in blood and urine contain specific proteins, mRNAs, and miRNAs derived from multiple organs, including the liver, and that the EV cargos may reflect the pathophysiological state of a given liver at the time of sample collection (Bala, et al., 2012; Cho, Im, et al., 2017; Wetmore, et al., 2010). The recent reports on EVs as promising diagnostic and prognostic biomarkers in acute liver diseases have been summarized in Table 1. The current gold-standard biomarkers for liver injury state is based on the measurements of plasma or serum levels of hepatic enzymes, including ALT, AST, alkaline phosphatase and γ-glutamyl-transpeptidase (GGT) (W. R. Kim, Flamm, Di Bisceglie, Bodenheimer, & Public Policy Committee of the American Association for the Study of Liver, 2008). However, altered serum activities of the hepatic enzymes are not specifical within a spectrum of different liver diseases and these traditional markers do not always correlate with the various stages of liver disease (W. R. Kim, et al., 2008) due to a relative short half-life of ALT (Giannini, Testa, & Savarino, 2005). Therefore, recent reports indicated that the specific components in circulating EVs may have a great utility as a non- or minimally-invasive biomarker for the diagnosis before and during treatment of liver disease. In fact, several EV components, such as specific miRNAs, mRNA, and proteins alone or in combination, have been introduced as potential biomarkers for detecting the severity and progress of various liver diseases, including non-alcoholic fatty liver disease (NAFLD) (Kato, et al., 1985; Mezey, Cherrick, & Holt, 1968) and virus-mediated liver diseases. However, based on our experience and recent publications, we only focus on the potential markers of ALD and DILI, as described below.

Table 1.

Summary of reported EV biomarker in ALD or DILI.

Alcohol or Drug	Species/Sample source	Isolation method	Biomarker	References
Drug
D-(+)-Galactosamine	Rat/Plasma	Ultracentrifugation (210k × g)	mRNAs: Alb, Fgn, Hp, and Actb	Wetmore, et al., 2010
D-(+)-Galactosamine	Rat/Urine	Ultracentrifugation (100k × g)	Proteins: Cd26, Slc3a1, Cd81, LimpII and Cd10	Conde-Vancells, et al., 2010
D-(+)-Galactosamine, LPS, and Acetaminophen	Mouse/Plasma	ExoQuick kit	miRNAs: miR-122 and miR-155	Bala, et al., 2012
D-(+)-Galactosamine, diclofenac, and Acetaminophen	Rat primary hepatocytes/Medium	Ultracentrifugation (100k × g)	mRNAs: Alb, Gnb21, and Rbp4	Royo, et al., 2013
D-(+)-Galactosamine	Rat/Serum	Ultracentrifugation (100k × g)	mRNAs: Alb, Gnb21, and Rbp4	Royo, et al., 2013
D-(+)-Galactosamine, LPS	Rat primary hepatocytes/Medium	Ultracentrifugation (100k × g)	Proteins: Ces3, Slc27a2, Sult1, Hsp90, Hsp70, Fril1, Clusterin, Alix, Cps1, Mat1a, and Comt	Rodríguez-Suárez E, et al, 2014
D-(+)-Galactosamine	Rat/Serum	Ultracentrifugation (100k × g)	Proteins: Ces3, Slc27a2, Sult1, Hsp90, Hsp70, Fril1, Clusterin, Alix, Cps1, Mat1a, and Comt	Rodríguez-Suárez E et al, 2014
Acetaminophen	Rat/Serum	ExoQuick kit	miRNA and protein: miR-122 and albumin	Holman, et al., 2016
Acetaminophen	Mouse/Plasma	ExoQuick kit/Ultracentrifugation (100k × g)	Proteins: Ass1, Prx1, Cs3, Ces1, Adg1, Gst, Apoa1, Alb, Hp, and Fgb	Cho, Im, et al., 2017
Thioacetamide and D-Galactosamine	Mouse/Plasma	ExoQuick kit/Ultracentrifugation (100k × g)	Number of EVs Proteins: Alb, Hp, and Fgb	Cho, Im, et al., 2017
Acetaminophen	Mouse/Plasma	ExoQuick kit	miRNAs: miR-122, 192, and miR-155	Cho, Kim, et al, 2017
Acetaminophen	Rat hepatocytes/Medium	Ultracentrifugation (100k × g)	Number of EVs 67 Metabolites	Royo, et al., 2017
Acetaminophen	Rat/Plasma	Total exosomes isolation solution (Life technologies)	miR-122	Thulin, et al., 2017
Alcohol
Chronic alcohol	Mouse/Plasma	ExoQuick kit	miRNAs: miR-122 and miR-155	Bala, et al., 2012
Chronic alcohol	Mouse/Plasma	ExoQuick kit	Number of EVs miRNAs: miR-122 and miR-155	Momen-Heravi, Saha, et al., 2015
Chronic alcohol	Human/Plasma	ExoQuick kit	miRNAs: miR-122	Momen-Heravi, Saha, et al., 2015
Binge alcohol	Mouse/Plasma	ExoQuick kit/Ultracentrifugation (100k × g)	Number of EVs Proteins: ALB, HP, and FGB	Cho, Im, et al, 2017
Binge alcohol and human alcoholic	Rodents and human alcoholics/Plasma	ExoQuick kit	Number of EVs Proteins: CYP2E1 and other isoforms	Cho, Mezey, et al, 2017
Binge alcohol and human alcoholic	Mouse/Plasma	Ultracentrifugation (100k × g)	Number of EVs miRNAs: m let7f, miR-29a, and miR-340	Eguchi, et al., 2017

4.1. Potential EV biomarkers for DILI

4.1.1. Circulating miRNAs, and mRNAs

The current Guidance for Industry (FDA guideline) includes the standard recommendations where new methods of liver safety assessment are required (https://www.fda.gov/downloads/Drugs/Guidances/UCM174090.pdf). However, few, acceptable biomarkers for DILI and/or acute liver injury have been identified, despite the extensive research on the pathological mechanisms in these diseases (Kullak-Ublick, et al., 2017). Circulating liver-derived miRNAs have been emerged as new, potential biomarkers for identifying DILI or acute liver injury (Starkey Lewis, et al., 2011; Ward, et al., 2014). The potential usage of EV components as potential biomarkers was demonstrated after assessing the altered levels of EV miRNAs in hepatocyte injury in ALD, DILI, and inflammatory liver disease (Bala, et al., 2012). In a mouse model of APAP-induced acute liver injury, the EV levels of hepatocyte-specific miR-122 were markedly elevated while its hepatic levels were unchanged and the EV level of inflammation-associated miR-155 was moderately increased (Bala, et al., 2012). Increased serum miR-122 levels were found both in the exosome-rich and protein-rich (remainder of plasma) of the serum fractions (Bala, et al., 2012). Exosomal miRNAs including miR-122 are more stable than the non-vesicle-associated plasma miRNAs when assessed ex vivo (Koberle, et al., 2013). These results strongly indicate that exosomal miRNAs can become a better, stable biomarker(s) for detecting DILI than the plasma miRNAs.

In a rat model of APAP-induced liver injury with elevated serum ALT and hepatocyte necrosis in the pericentral regions, increased numbers of circulating EVs were observed and these EVs showed elevated levels of miR-122 and albumin mRNA after exposure to a subtoxic dose of APAP (500 mg/kg by oral gavage) and a toxic dose of APAP (1400 mg/kg) (Holman, Mosedale, Wolf, LeCluyse, & Watkins, 2016). In addition, in primary human hepatocytes, treatment with 10 mM APAP for 24 h increased the levels of both albumin mRNA and miR-122 in EVs, but not in cells or protein-rich fractions (Holman, et al., 2016). In another study, a single i.p. injection of APAP at 300 mg/kg into Balb C mice not only caused centrilobular necrosis, but also elevated the numbers of circulating EVs with increased amounts of liver-specific miR-122, miR-192, and miR-155 (Cho, Kim, et al., 2017). However, little changes in the muscle-specific miR-206 or kidney-specific miR-146a were observed in APAP-exposed mice. In addition, the elevated levels of the circulating EV miR-122, miR-192, and miR-155 following APAP-treatment were significantly decreased and returned to basal levels after pre-treatment with an antioxidant N-acetylcysteine (NAC), which fully prevented the hepatotoxicity (Cho, Kim, et al., 2017). Increased serum miR-122 levels were also observed in patients with APAP-induced liver injury and the profiles of elevated serum miRNAs in DILI were different compared with the individuals with ischemic hepatitis, indicating some specificities between the two different types of liver injury (Ward, et al., 2014). Indeed, most of circulating microRNAs, including miR-122, were decreased to normal levels in patients following NAC treatment (Ward, et al., 2014). The plasma levels of hepatocyte-derived miR-122, miR-148a, and miR-194 were elevated in patients with liver injury from acute rejection after liver transplantation and more importantly, their levels appeared to rise earlier than the elevation of serum ALT levels (Farid, et al., 2012). In this study cellular miRNAs could be released into microvesicels, including exosomes, and only distinct sets of miRNAs were sensitively packaged into microvesicles (Farid, et al., 2012). Consistently, several studies showed that liver-specific miRNAs in circulating EVs can serve as sensitive diagnostic and prognostic tools for detecting severe liver injury and could be useful for future drug discovery by monitoring the early signs of DILI.

Recently, Thulin et al. determined the amount and distribution of miR-122 in exosomes and the ‘protein-rich’ supernatant (the remainder) from the plasma of rats exposed to oral administration of APAP (Thulin, et al., 2017). The quantitative analysis revealed that much greater amounts of miR-122 were found in the protein-rich supernatant than the exosomes in APAP-induced acute DILI rats. Similarly, Bala et al. also showed that the levels of miR-122 in the plasma protein-rich fraction were higher than those in exosomes in APAP-induced acute liver injury (Bala, et al., 2012). Additionaly, plama miR-122 was increased in exosomesrich fraction compared to the protein-rich fraction in primary human hepatocytes to a subtoxic APAP concentration (10 mM) for 24 h (Holman, et al., 2016). These results suggest that acute APAP-induced liver injury with massive hepatocytes necrosis were increased the necrotic protein-rich miR-122. By comparing the pattern of time- and dose-dependent elevation of many other DILI markers, Thulin et al. indicated that miR-122 and glutamate dehydrogenase (GLDH) are more readily-detectable biomarkers than ALT mainly due to their elevated levels at earlier time points. However, the elevated miR-122 in plasma returned to basal levels earlier than other markers. It would have been better if the time-dependent changes in exosomal miR-122 following APAP administration were analyzed since exosomal components, including miRNAs, are known to be more stable (Koberle et al., 2013; Wetmore et al., 2010).

On the other hand, He et al. reported that neutrophil specific miR-223 is likely to play a critical role in preventing APAP-induced liver injury by attenuating the inflammatory responses of neutrophils (He, et al., 2017). Deletion of miR-223 significantly exacerbated APAP-induced neutrophil infiltration, oxidative/nitrateive stress and hepatotoxicity while overexpression of miR-223 attenuated the liver injury. APAP overexposure markedly elevated serum levels of mtDNA, which was found in microparticles (MPs) and exosomes although much greater amounts were present in MPs (He, et al., 2017). Treatment with MPs from APAP-exposed hepatocytes significantly up-regulated the expression of miR-223 in WT neutrophils but not in Toll-like receptor 9 (TLR9)-knockout neutrophils, suggesting an important role of TLR9 in APAP-induced neutrophil infiltration and liver injury (He, et al., 2017).

In the rat models of DGAL- or APAP-induced acute liver injury, the amounts of exosome associated liver-specific mRNAs in circulating serum were increased and their increments positively correlated with the elevation of traditional liver injury markers (i.e., serum ALT and AST levels) (Wetmore, et al., 2010). However, the comparative analysis of circulating liver mRNAs with ALT and histopathological lesions indicated that circulating liver mRNAs in EVs were more specific and sensitive biomarkers for recognizing liver injury, possibly due to their stabilities in circulating EVs and the earlier appearance (Wetmore, et al., 2010). Thus, the mRNAs contained in EVs of hepatic origin could be added to the repertoire of biomarkers for detection of acute liver injuries caused by many different compounds APAP, DGAL, and diclofenac (Royo, et al., 2013). For instance, the elevated EV mRNAs, such as Alb, Gnb2l and Rbp4 genes, were observed DGAL-induced liver injury in rats (Royo, et al., 2013). Consistently, the levels of liver-specific miRNAs such as miR-122, miR-192, and miR-155 in circulating exosomes were markedly elevated in APAP-exposed mice with hepatic necrosis (Cho, Kim, et al., 2017). However, the circulating levels of kidney-specific miR-146a and muscle-specific miR-206 were not elevated in the APAP-induced acute liver injury models, while their levels were selectively elevated in cisplatin-induced nephrotoxicity and bupivacaine-induced myotoxicity, respectively. These results stongly support the utilities of liver-specific miRNAs as potential biomarkers in detecting acute DILI.

4.1.2. Circulating proteins and metabolites

The number and amount of circulating EVs were markedly increased before or during the manifestation of full-blown DILI, which also affects the release and the amounts of hepatocyte-derived proteins and metabolites (Cho, Im, et al., 2017; Royo, Palomo, et al., 2017). Primary hepatocytes secrete EV proteins that included the exosomal marker proteins (e.g., Tsg101, Cd63, and Cd81), hepatic-specific proteins, likes the asialoglycoprotein receptor, and the proteins associated with metabolic diseases, such as annexin A2, paraoxonase-1, apolipoprotein-E, catechol O-methyltransferase, and insulin receptor substrate 1 (Conde-Vancells, et al., 2008). The presence of several proteins (e.g., cytochromes P450, UDP-glucuronyltransferases, glutathione-S-transferase [GST], etc.) involved in detoxification of xenobiotic compounds suggests that these hepatocyte-derived vesicles are not only important in biomarker discovery, but also relevant to the pharmaceutical industry since these proteins could participate in drug metabolism and clearance processes (Conde-Vancells, et al., 2008; Conde-Vancells, Rodriguez-Suarez, et al., 2010). Proteomics analysis of urinary EVs isolated from DGAL or LPS-administered rats showed the changes in the abundance and content of liver-derived proteins. In particular, in DGAL-exposed rats, the amounts of Cd26, Cd81, Slc3A1, and Cd10-positive urinary EVs were dramatically reduced, supporting the potential use of these four proteins as candidate urinary indicators of acute liver damage (Conde-Vancells, Rodriguez-Suarez, et al., 2010). Proteomic analysis of serum EVs isolated from the rats exposed to DGAL or LPS showed a characteristic protein composition that correlated with liver toxicity, indicating the EV protein cargo as a potential biomarker for DILI (Rodriguez-Suarez, et al., 2014). In a mouse model of APAP-induced liver injury, mass-spectral based proteomics analysis of plasma EVs revealed that the elevated numbers of circulating EVs were observed and these EVs showed increased amounts of liver specific proteins such as alcohol dehydrogenase-1 (ADH1), GST, albumin (ALB), haptoglobin (HP), and fibrinogen (FGB) (Cho, Im, et al., 2017). In particular, the number and the amounts of EV liver-specific proteins, such as liver carboxylesterase-1 (CES1), apolipoprotein A-1 (APOA1), ADH1, GST, ALB, HP, and FGB, were increased by more than 2 folds in APAP-exposed mice compared to those of vehicle controls. The increased EV ADH1 is consistent with the previous reports with the elevated amounts of serum ADH that positively correlated with intrahepatic cholestasis (Mezey, et al., 1968) and acute alcohol-induced centrilobular necrosis despite the normal serum ALT and GLDH activities (Kato, et al., 1985). The changes in EV protein amounts also showed greater correlations with APAP dosage and exposure time than those with serum ALT changes (Cho, Im, et al., 2017). Moreover, the increased level of EV albumin protein was detected in very early disease stages and was consistently observed over the time course of hepatic disease progression (Cho, Im, et al., 2017), strongly indicating that EV proteins could be used as better biomarkers for DILI than the plasma or serum ALT levels.

The comprehensive proteomic analysis of hepatocyte-derived EVs revealed the presence of proteins involved in the metabolism of lipoproteins, endogenous compounds, and xenobiotics. These results showed that hepatocytes secrete EVs, containing a high number of enzymes, to extracellular compartments, suggesting a role of EVs in the metabolisms of lipids and xenobiotics. Some of these proteins include various cytochromes P450 (CYPs), UDP-glucuronosyltransferase, GSTs and other liver proteins (Conde-Vancells, Gonzalez, Lu, Mato, & Falcon-Perez, 2010; Conde-Vancells, et al., 2008). An ultra-high performance liquid chromatography-mass spectrometry based targeted metabolomics analysis of serum samples identified 67 serum metabolites whose levels were altered after incubation with the EVs released from hepatocytes exposed to APAP or diclofenac compared to control EVs. These metabolites included various amino acids and different species of phosphatidylcholines and phosphatidylethanolamines (Royo, Palomo, et al., 2017). The serum levels of l-glutamic acid, alpha-linolenic acid, pentadecanoic acid, 9,10-dihydroxy-octadecanoic acid and the monounsaturated fatty acid species (17:1 n-x) were increased after incubation with EVs from hepatocytes treated with APAP or diclofenac compared to control EVs. Interestingly, this report showed that the 9,10-dihydroxy-octadecanoic acid is a product of the linoleic acid metabolism by cytochromes CYP2C, CYP2E1 and CYP3A4, supporting that it likely represents a hepatocyte-derived EV component (Royo, Palomo, et al., 2017). Alterations in the levels of some other serum metabolites might cause deleterious consequences. For example, decreased l-arginine may negatively affect the physiological maintenance of vascular endothelial function in APAP-exposed rodents, since l-arginine is the precursor of nitric oxide (NO), which is a key regulator of vascular function. These results with a targeted metabolomics study showed that hepatocyte-derived EVs are metabolically active and can modify the levels of blood metabolites associated with the energy supply, redox metabolisms and endothelial regulation (Royo, Palomo, et al., 2017). It is of interest that the arginase activity was also detected in serum derived-EVs and that this vesicular arginase activity was significantly increased under liver-damaging conditions. Functionally, the arginase contained in circulating EVs are likely to be involved in the regulation of the endothelial function, contributing to pulmonary vascular hypertension (Royo, Moreno, et al., 2017). Thus, the hepatocytes-derived EVs are metabolically active and might play a role in the pathophysiological change in DILI through regulating the metabolisms of endogenous substrates.

4.2. EV biomarkers for ALD

4.2.1. Circulating miRNAs and mRNAs

The National Institute on Alcohol Abuse and Alcoholism (NIAAA) has been a leading sponsor of research grants and service contracts related to the development of biomarkers for alcohol consumption and alcohol-induced tissue injury for over three decades (Freeman & Vrana, 2010). In addition to the alcohol-induced tissue injury including ALD occurred after long-term chronic alcohol consumption, binge alcohol intake (e.g., 4~5 drinks within 2-h) represents a pattern of alcohol drinking responsible for more than 75% alcohol consumed in the USA (Esser, et al., 2014). Binge alcohol administration is also known to cause acute hypoxic tissue injury in experimental animals and human alcoholics (Yun, et al., 2014). Despite the extensive research on the pathological mechanisms, reliable biomarkers for specifically recognizing alcohol-induced liver injury have not been clearly established.

Therefore, many laboratories have recently sought to identify potential biomarkers specific for alcohol-induced liver injury by evaluating the characteristics of circulating EVs and their components isolated from rodents or human specimens following binge and chronic alcohol exposure. Multiple studies have recently shown that the number and amount of circulating blood EVs are elevated in alcoholic hepatitis patients and mice after chronic alcohol feeding or binge alcohol exposure (Cho, Im, et al., 2017; Momen-Heravi, Bala, Kodys, & Szabo, 2015; Saha, et al., 2017; Saha, Momen-Heravi, Kodys, & Szabo, 2016). Further studies will be required to ascertain whether increased EV number or their components can serve as a specific biomarker for diagnosis and prognosis of alcoholic hepatitis only, without detecting DILI.

In a mouse model of chronic alcohol-induced liver damage, the number of circulating EVs was increased compared with the pair-fed controls, and elevated serum/plasma miR-122 and miR-155 were predominantly associated with the EVs-rich fraction (Bala, et al., 2012). The elevated miR-122 and miR-155 were found in the exosomes-rich fraction compared to the protein-rich fraction (supernant) in alcohol-exposed mice (Bala, et al., 2012). These results suggest the important role of the sources of miRNAs in identifying disease specificity (e.g., distinction between ALD and DILI), since greater levels of miR-122 were observed in the protein-rich fraction in DILI. The number of circulating EVs and amounts of miR-122, miR-192 and miR-30a mostly associated with small-size vesicles (i.e., exosomes) were increased in a mouse model of alcohol-induced steatohepatitis compared to controls (Momen-Heravi, Saha, et al., 2015). Furthermore, the elevated levels of miR-192 and miR-30a were also observed in human alcoholics with hepatitis, validating the animal data with human specimens. In addition, the number and amounts of EVs were significantly increased in the sera of healthy individuals after alcohol binge drinking. By performing receiver operating characteristic curve analyses to determine the sensitivity and specificity of the candidate miRNAs in both mouse and human specimens, miR-192 turned out to be one of the best potential biomarkers for diagnosis of alcoholic hepatitis (Momen-Heravi, Saha, et al., 2015).

Recently, Feldstein group reported that stressed or damaged hepatocytes release EVs with a miRNA signature that can be detected in circulating EVs in mice with alcoholic steatohepatitis (ASH) caused by constant intragastric ethanol infusion (Eguchi, et al., 2017). Three miRNAs (let-7f, miR-29a, and miR-340) were increased in blood EVs from the ASH mice, but not in blood EVs from different liver disease mouse models, including bile duct ligation, NASH, and obese mice, as well as EVs released from hepatocytes exposed to ethanol. Furthermore, the number of EVs and levels of three miRNAs in blood EVs such as let-7f, miR-29a, and miR-340 were significantly increased in patients with ALD compared to nonalcoholics (Eguchi, et al., 2017). These experimental and human results clearly indicated the specificity of these three miRNAs in circulating EVs as potential biomarkers for detecting ASH.

4.2.2. Circulating proteins and metabolites

Recent advancements in proteomics technologies have greatly increased the prospect of discovering specific biomarkers for alcohol abuse and tissue injury (Torrente, et al., 2012). Proteomic studies performed with blood samples have prompted the requirement for diminishing both the complexity and the heterogeneity of the samples to analyze. The tissue-specific EVs contained in blood samples provides an important source to obtain information on the pathophysiological status of a specific organ or tissue with relatively low complexity and minimal invasiveness (Falcon-Perez, Lu, & Mato, 2010). Mass-spectral based proteomic identification of specific proteins in the secreted EVsunambiguously allows tracing of the origin of specific cell types (Simpson, Jensen, & Lim, 2008). Falcon-Perez group has reported that primary hepatocytes secrete EVs to the extracellular space and the characterization of their protein composition has revealed the presence of EV protein markers (e.g., Tsg101, Cd63, Cd81) and liver-specific proteins, including the asialoglycoprotein receptor (Conde-Vancells, et al., 2008), which could be useful to immunoisolate hepatocyte-specific EVs from blood samples (Falcon-Perez, et al., 2010).

In addition to elevated EV miRNAs, the number and protein contents in circulating EVs were significantly increased in alcohol-exposed rodents and alcoholic hepatitis patients, compared to those in healthy controls (Cho, Im, et al., 2017; Momen-Heravi, Bala, et al., 2015; Verma, et al., 2016). Moreover, the increased number and contents of EV proteins in alcohol-exposed mice and rats as well as patients with alcoholic hepatitis suggest conserved commonality among different species. In addition, binge alcohol exposure, elevated the number of EVs and their protein amounts in the ethanol-inducible cytochrome P450-2E1 (CYP2E1)-dependent manner. Further mechanistic studies conducted with primary hepatocytes demonstrated that the amounts of CYP2E1 and other P450 isoforms such as CYP2A, CYP1A1/2, and CYP4B proteins in culture medium EVs were increased through alcohol-mediated oxidative and ER stress. However, their elevated levels in EVs produced from alcohol-exposed hepatocytes were decreased by co-treatment with a CYP2E1 inhibitor chlormethiazole (CMZ), an anti-oxidant NAC, or an ER stress suppressor 4-phenylbutyrate. These results indicate that elevated CYP2E1 and other isoforms in circulating EVs not only correlated with the levels of their hepatic counterparts but, also depended on the status of increased oxidative and ER stress in the liver (Cho, Mezey, et al., 2017).

Protein tyrosine phosphatase receptor type gamma (PTPRG) is a hepatic, transmembrane protein, which removes the phosphate group from phosphotyrosine residues of many proteins, but released into cell-derived vesicles (EVs) during hepatotoxicity. For instance, the release of sPTPRG from HepG2 human hepatocellular hepatoma cells could be stimulated by ethanol exposure and this release seems sensitive to metalloproteinase but insensitive to Furin inhibitors. The increased levels of the plasma ~120 kDa isoform of sPTRPG were associated with the occurrence of liver damage and exhibited a positive correlation with high plasma levels of biomarkers associated with alcohol-induced hepatic injury. These results suggested that EV PTPRG can become a novel candidate biomarker protein in plasma whose increased expression is associated with hepatocyte damage (Moratti, Vezzalini, Tomasello, Giavarina, & Sorio, 2015).

Metabolomics is a rapidly emerging field that aims to identify and quantify the concentration changes of all the small molecule metabolites during chronic alcohol consumption (Manna, et al., 2010). Some reports suggested that the application of metabolomics to understand the effects of ALD represents a powerful means not only to identify the earliest biomarkers, but also to unravel the molecular mechanism of its pathogenesis (Javors, Hill-Kapturczak, Roache, Karns-Wright, & Dougherty, 2016; Manna, et al., 2011; Manna, et al., 2010). For instance, indole-3-lactic acid and phenyllactic acid are potential candidates for conserved and pathology-specific high-throughput noninvasive biomarkers for early stages of ALD (Manna, et al., 2011). In addition, phosphatidylethanol (PEth) 16:0/18:1 and 16:0/18:2 species in human blood could be a sensitive biomarker for the identification of relatively low levels of alcohol consumption (Javors, et al., 2016). The half-life of combined PEth species was 4.6 ± 3.5 (SD) days, which allowed them to be detectable even after 2 weeks of confirmed abstinence in most participants. However, EV metabolomics approach in specifically detecting the ALD need further studies.

4.3. Advantages and disadvantages of EVs components as potential biomarkers

The ALT and AST enzymes are normally involved in amino acid metabolism and biosynthesis but can be released from damaged or dying hepatocytes following exposure to various hepatotoxic agents such as ethanol, APAP and DGAL (Amacher, Schomaker, & Aubrecht, 2013). Although these markers have been used as the gold standard, they suffer from a lack of liver specificity. For instance, the activities of these enzymes are known to be elevated in some other clinical disorders besides liver disease (W. R. Kim, et al., 2008; Nathwani, et al., 2005; Shabaneh Al-Tamimi & McDonald, 2008). Additionally, these biomarkers do not always correlate well with histomorphologic data in clinical applications (Ozer, Ratner, Shaw, Bailey, & Schomaker, 2008). ALT exists in two isoforms (ALT1 and ALT2) and are highly expressed in kidney, liver, fat, and muscle tissues (R. Z. Yang, Blaileanu, Hansen, Shuldiner, & Gong, 2002). DILI is a major concern both in the clinic and in the development of safer drugs. To aid the diagnosis of DILI in the clinic and to improve the development of safe new drugs, sensitive and specific biomarkers in relatively easily accessible biofluids are required. The current gold-standard biomarkers are hampered by either non-specific tissue expression or not detectible early enough during the course of a disease progression to be prognostic (Hornby, Starkey Lewis, Dear, Goldring, & Park, 2014).

Pharmacogenetic testing, next-generation deep sequencing, proteomics, metabolomics and mechanistic markers can help to preselect susceptible patient populations for enrichment purpose and tailor drug therapy to individual patients. To find new DILI indicators, several studies are under investigation for evaluating various biological candidates, including microRNAs, cytokeratin-18 (CK18), high mobility group box protein 1 (HMGB-1), and several other biomarkers. Recently, tissue-specific miRNAs have emerged as a promising class of biomarkers for detecting various diseases and organ damage. For instance, elevated miR-122, a liver-enriched miRNA, has been demonstrated for its use as a DILI biomarker in both animal and human studies (Hornby, et al., 2014; Thulin, et al., 2017).

There are several advantages in using EV-based DILI or ALD biomarkers. First, it may be possible to identify liver specific biomarkers in circulating EVs because EVs reflect the original cells’ status (X. Yang, Weng, Mendrick, & Shi, 2014). Second, EV components are protected from degradation because of vesicle status (Raposo & Stoorvogel, 2013; van der Pol, et al., 2012) with longer half-lives, providing a wider time window for injury detection. Third, efforts to directly use serum and plasma to discover DILI or ALD biomarkers are often hindered by the highly abundant blood constituents. Finally, the association of specific EV components (miRNAs, mRNA, mtDNA, and proteins) offers the utilities of EVs as potential biomarkers to identify toxic compounds and/or predict treatment response. Although the concept of gene expression patterns in EVs being specific to the toxicant or injury type is not necessarily new, the added dimension of comparing expressions of mRNAs and miRNAs inside or outside EVs may provide novel insights into the mechanisms of DILI and additional evidence in identifying the drug causing DILI (Bala, et al., 2012; Wetmore, et al., 2010).

The limitations of biomarker discovery using EVs could be resulted from the lack of generally accepted standardization in EV isolation methods and guidelines related to sample collection and handling. The standard differential centrifugation, density gradients, polymer-based precipitation, microfiltration and size-exclusion-based methods have been developed for the EV isolation (Lotvall, et al., 2014). In fact, all these isolation methods can significantly influence the amount, type and purity of EVs recovered (Van Deun, et al., 2014). The International Society for Extracellular Vesicles (ISEV) has attempted to address some of these critical issues through the publication of position papers, EV RNA analysis (Hill, et al., 2013) and EV-based therapeutics (Lener, et al., 2015), and the minimal experimental requirements for the definition of EVs and their function (MISEV) (Lotvall, et al., 2014).

Conditions of sample storage, EV isolation methods and subsequent RNA isolation methods can affect downstream RNA and protein expression profiles. For example, Zhou et al. reported that recovery after freezing at −80 °C was almost complete. Extensive vortexing after thawing markedly increased EVs recovery in urine specimens frozen at −20 °C or −80 °C, even after storage for 7 months (Zhou, et al., 2006). Increasing time between venipuncture and centrifugation has also been reported to induce degradation and rupture of EVs (Ayers, et al., 2011). Another example of a pre-analytical parameter that should be taken into account is the effect of circadian rhythm on EV cargos in urine specimens. For instance, the expression of NaCl cotransporter and prostasin in urinary EVs has been reported to be affected by circadian rhythm (Castagna, et al., 2015).

4.4. EV biomarkers as organ-specific proteins or miRNAs

Blood plasma contains a number of potential biomarkers that reflect the pathophysiological states of certain organs and tissues. Specific blood proteins or miRNAs, produced primarily in one or a few tissues and then secreted into the blood, can be a powerful and specific tool to assess pathophysiological states in their cognate organs. The organ-specific blood proteins or miRNAs in the EVs are likely to offer a new and possibly better strategy in finding potential biomarkers for diagnosis and treatment assessment of DILI and ALD, as illustrated in Fig. 2. We investigated whether organ-specific circulating EVs were secreted after injury to drug-sensitive target organs. Mice were treated with a specific drug such as cisplatin (CIS), cyclophosphamide (CPP), and 1-methyl-4-phenyl-1,2,3,6-tetrahydropyridine (MPTP) to promote injury in kidney, heart, and brain, respectively (Cho, Im, et al., 2017). EVs, from CIS-treated mice with severe kidney damage, showed significantly increased levels of kidney-specific proteins, such as kidney injury molecule-1 (KIM-1) and neutrophil gelatinase-associated lipocalin (NGAL). EVs, derived from CPP-treated mice with cardiac abnormality, exhibited elevated levels of heart-specific protein cardiac troponin I (cTnI), while EVs, obtained from MPTP-treated mice with neuronal damage, showed increased levels of neuron-specific protein S100 calcium binding protein B (SB). Therefore, it is likely that circulating EVs may contain potential markers for a specific organ damage caused by a particular drug or other pathophysiological reasons. These results indicate that characterization of EV proteins is also important in identifying potential biomarkers for detecting organ-specific toxicity.

Fig. 2. Organ-specific EVs are circulated in the blood after drug-induced organ injury.

Wild-type male BALB/c mice (6 weeks old) were treated with PBS, MPTP (30 mg/kg), APAP (200 mg/kg), CIS (10 mg/kg), or CPP (200 mg/kg) for 24 h (n = 10/group). (A) Western blot analysis of liver-specific proteins (ALB, HP, and FGB), kidney-specific proteins (Kim-1 and NGAL), a heart-specific protein (cTnI), and a brain-specific protein (SB) in circulating EVs isolated from plasma of each group. (B) Organ-specific proteins in EVs could be used as potential biomarkers for drug-induced tissue injury.

In addition, many miRNAs expressed in hepatocytes and other liver cells are known to be released during hepatic injury. However, only the most abundant and liver-specific miRNAs in serum are likely to be detected (Ward, et al., 2014). There may be sequence specificity or selectivity for the release of miRNAs rather than just a general leakage of all miRNAs from the injured cells. The cellular miRNAs can be secreted from the originating hepatocytes, Kupffer cells, or stellate cells as microvesicles, including exosomes, and only distinct sets of miRNAs are selectively packaged into microvesicles (Kosaka, et al., 2010; van der Vlist, et al., 2012). Taken together, further studies are needed to find out additional, potential biomarker proteins and/or miRNAs in EVs in the circulating blood for early detection of organ injury caused by certain drugs or heavy alcohol intake.

4.5. EV biomarkers from mitochondria derived vesicles?

Mitochondria are critically important in providing cellular energy ATP as well as their involvement in anti-oxidant defense, oxidation of fatty acids, intermediary metabolism and cell death processes when living cells or mammals are exposed to potentially toxic agents including alcohol, n-6 fatty acid-containing high fat diets, and certain drugs including APAP in large quantities (Song, et al., 2014). Mitochondria-derived vesicles (MDVs) can be generated through the selective incorporation of protein cargos consisting of the outer membrane, inner membrane, and matrix contents (Andrade-Navarro, Sanchez-Pulido, & McBride, 2009; Cadete, et al., 2016; Shutt & McBride, 2013; Soubannier, Rippstein, Kaufman, Shoubridge, & McBride, 2012; Sugiura, McLelland, Fon, & McBride, 2014). Soubannier et al. reported that MDVs were produced following exposure to various forms of mitochondrial stress (components of OXPHOS complexes II, III and IV), while MDVs did not contain complex I, complex V protein, and any nucleus-like nucleoids, demonstrating the specificity of cargo incorporation (Soubannier, et al., 2012). Stress-induced MDVs were selectively enriched with oxidized proteins, suggesting that conformational changes induced by oxidative modifications might initiate their incorporation into the MDVs (Soubannier, et al., 2012). Another recent report showed that MDV production was increased together with mitophagy in doxorubicin-induced mitochondrial and cardiac dysfunction, implying that MDV formation may be a mitochondrial quality control system operating in the heart and functionally active as the first line of defense against stress (Cadete, et al., 2016). However, it is relatively unknown whether similar types of MDV exist in the cases of DILI or alcohol-induced liver damage, where mitochondrial dysfunction plays an important role in pathogenesis (Abdelmegeed & Song, 2014; Song, et al., 2014). Thus, additional studies are required to further demonstrate the roles of MDVs that are released due to mitochondria dysfunction in the drug- or alcohol-induced liver injury.

5. The biological effects of secreted EVs in or during DILI or ALD

It is well-established that secreted EVs can play an important role in cell-cell communication by transferring protein, RNA and miRNA cargos from the source cells to the recipient target cells. Such intercellular communications allow one type of cell to influence neighboring cells or those at remote sites, regardless of cell type, and alter their microenvironments and cell fates. Thus, it is highly likely that various components contained in circulating EVs produced from DILI or ALD play a functional role in many biological processes (Fig. 3).

Fig. 3. Exogenous EVs from alcohol- or drug-exposed cells can promote apoptosis or inflammation of the recipient cells.

EVs produced after alcohol exposures can be taken up by recipient hepatocytes, leading to inhibition of cell proliferation and enhancement of cell death signals in recipient cells. The EVs secreted from alcohol-exposed rodents and patients with ALD contain increased amounts of miR-122 and miR-155, that could stimulate hepatic inflammation. CD40L- and Hsp90-enriched EVs activated the hepatic monocytes/macrophages. The EVs from alcohol or acetaminophen-exposed rodents are functional and can promote cell death by activating the apoptosis signaling pathway, including phospho-JNK, proapoptotic Bax and activated, cleaved caspase-3 in recipient hepatocytes.

For instance, several recent studies have focused on the properties and roles of EVs isolated from alcoholic hepatitis patients. In fact, multiple studies have shown that the number and amounts of EVs are increased in patients with alcoholic hepatitis and even in healthy people consuming excess amounts of ethanol. Both hepatocyte-derived and monocyte-derived EVs in the models of alcoholic hepatitis have been shown to stimulate macrophages, thereby promoting inflammation. In this case, several miRNAs such as miR-122 and miR-155 have been proposed to be responsible for EV-mediated cell-to-cell signaling (Momen-Heravi, Bala, et al., 2015). In addition, EV protein cargos from ethanol-exposed hepatocytes activated macrophages, resulting in elevated production of proinflammatory cytokines. CD40 ligand was proposed as an EV cargo that could promote macrophage activation in in vitro and in vivo experimental models of alcoholic hepatitis (Verma, et al., 2016).

Increased number and amount of EVs were released into culture media when primary rat hepatocytes and HepG2 human hepatoma cells were treated with alcohol (Cho, Mezey, et al., 2017). In alcoholic liver disease, hepatocyte-derived EVs enriched in miR-122 were taken up by macrophages, resulting in functional changes and sensitization of these cells to LPS-induced proinflammatory responses (Momen-Heravi, Bala, et al., 2015). Alcohol treatment also promoted EV release from macrophages, suggesting that EV production and release occurs in both parenchymal and immune cells in ALD (Saha, et al., 2016). These observations suggest an important but an unexpected role for hepatocyte-derived EVs in regulating greater inflammation in ALD. Verma et al., reported that hepatocytes released CD40-ligand containing EVs in a caspase-dependent manner in response to alcohol exposure, and that these EVs promoted macrophage activation, thus contributing to the inflammatory profile of alcoholic hepatitis (Verma, et al., 2016). Furthermore, Saha et al. recently reported that EVs isolated from a mouse model of ALD were taken up by the recipient hepatocytes and macrophages/Kupffer cells to produce proinflammatory cytokines such as MCP-1 and TNF-α with decreased amounts of the anti-inflammatory cytokines compared to those exposed to EVs from control mice (Saha, et al., 2017). These investigators further demonstrated that Hsp90 was the main protein which activated the Kupffer cells/macrophages in the recipient mice, showing the biological function of the exogenous EV cargos (Saha, et al., 2017). Furthermore, the elevated CYP2E1-mediated oxidative and/or ER stress in alcohol-exposed hepatocytes, rodents and human alcoholics can promote the release of EV proteins and many other components that can be toxic to the recipient hepatocytes by activating the cell death pathway with elevated levels of reactive oxygen species, phospho-JNK, Bax, and activated cleaved caspase-3 (Cho, Mezey, et al., 2017). Taken together, these results suggest that dynamic, EV-mediated interceullar communications between hepatocytes and macrophages may play an important role in further promoting the ALD.

A recent study showed that the levels of circulating EVs and microparticles (MPs) were markedly elevated in individuals with a history of recent excessive drinking compared with those in chronic heavy alcohol drinkers or healthy controls. The levels of EVs positively correlated with serum levels of aminotransferases, circulating neutrophils, and mitochondria DNA (mtDNA) contents in serum MPs (Cai, et al., 2017). Mice exposed to chronic-plus-binge (E10d + 1B) ethanol treatment also showed markedly elevated serum levels of circulating neutrophils and mtDNA-enriched MPs. These mtDNA-enriched MPs were secondary to the activation of ER stress and its activation of the hepatic inflammasome (Garcia-Martinez, et al., 2016). Circulating mtDNA levels were elevated and contributed to the inflammatory liver disease through activating macrophages. It is possible that the elevation of mtDNA-enriched MPs may activate macrophages, leading to hepatic inflammation.

6. Therapeutic applications of EVs in preventing DILI and ALD

Potential therapies of EVs from a variety of cells have been investigated for future clinical applications such as treating different types of cancer or liver disease (Vlassov, et al., 2012). Especially, the dynamic stem cell regulation of specific organ injury and recovery may be the result of microvesicle-mediated interactions between differentiated cells and stem cells (Quesenberry & Aliotta, 2008). Administration of mesenchymal stem cells (MSCs)-derived exosomes has yielded beneficial effects in a variety of animal models of liver disease, including DILI and liver fibrosis (Lou, Chen, Zheng, & Liu, 2017). The recent reports related to MSCs-derived EVs as potential therapeutic agents in preventing or treating various models of liver disease have been summarized (Table 2).

Table 2.

Summary of reported EV therapeutics in preventing or treating liver disease.

MSCs	Isolation method	EVs term	Disease model	Functions	References
hucMSCs	Ultracentrifugation with 30%Sucorse/D2O cushions (100k × g)	Exosomes	CCl4-induecd liver fibrosis	Inhibit hepatic inflammation, collagen production, AST activity, and EMT associated markers	Li, et al., 2013
MSCs	Ultracentrifugation (100k × g)	Exosomes	APAP-, H2O2-, and CCl4-induecd liver injury	Increase hepatocyte proliferation by upregulating proliferation proteins and inhibition of apoptosis gene	Tan, et al., 2014
CPMSCs	Ultracentrifugation (100k × g)	Exosomes	CCl4-induced liver fibrosis	Inhibit the expression of Hedgehog and profibrotic genes	Hyun, Wang, Kim, Kim, & Jung, 2015
hucMSCs	Ultracentrifugation (100k × g)	Exosomes	H2O2-, and CCl4-induecd liver injury	Promote the recovery of hepatic oxidant injury through the delivery of GPX1	Yan, et al., 2017
ADMSCs	Ultracentrifugation (100k × g)	Exosomes	CCl4-induecd liver fibrosis	Decrease the levels of collagen I, vimentin, α-smooth muscle actin, and fibronectin proteins	Qu, et al., 2017

hucMSC, Human Umbilical Cord Mesenchymal Stem cells; EMT, epithelial-to-mesenchymal transit; CCL₄, carbon tetrachloride; MSCs, Mesenchymal Stem cells; APAP, acetaminophen; AST, aspartate aminotransferase; ADMSCs, adipose-derived mesenchymal stem cells; CPMSCs, chorionic plate-derived MSCs; GPX1, glutathione peroxidase1

EVs derived from human umbilical code MSCs can reduce myocardial ischemia/reperfusion damage and protect against acute tubular injury (T. Li, et al., 2013). In addition, these MSC-derived EVs also reduced the surface fibrous capsules, alleviated hepatic inflammation and collagen deposition in carbon tetrachloride (CCl₄)-induced fibrotic mouse liver. Furthermore, MSCs-derived EVs significantly restored AST activity and decreased the levels of fibrosis markers collagen types I and III, transforming growth factor (TGF)-β1, and phosphorylated Smad2 expression. These changes in the fibrosis markers and the restored levels of serum AST suggested a positive therapeutic effect of exosomes on the fibrotic liver (T. Li, et al., 2013). In addition, MSC-derived exosomes could elicit hepatoprotective effects against drug- or chemical-induced liver injury, mainly through activation of proliferative and regenerative responses (Tan, et al., 2014). The higher survival rate of hepatocytes was associated with upregulation of the priming-phase genes during liver regeneration, which subsequently led to higher expression of proliferation proteins (PCNA and cyclin D1) in the EVs-treated group. EVs also inhibited the APAP- and H₂O₂-induced hepatocyte apoptosis through upregulation of Bcl-xL protein expression (Tan, et al., 2014). Similarly, miR-125b prepared from chorionic plate-derived mesenchymal stem cells (CP-MSCs) suppressed the activation of Hh signaling, and reduced fibrosis, suggesting that a CP-MSCs-derived microRNA can promote liver regeneration in an in vivo model (Hyun, Wang, Kim, Kim, & Jung, 2015).

A single systemic administration of human MSC-EVs effectively rescued the recipient mice from CCl₄-induced liver failure (Yan, et al., 2017). Interestingly, human MSC-EVs-derived glutathione peroxidase1 (GPX1), which detoxifies CCl₄ and H₂O₂, reduced oxidative stress and apoptosis. Knockdown of GPX1 in human MSCs abrogated antioxidant and anti-apoptotic abilities of hucMSC-EVs and diminished the hepatoprotective effects of human MSC-EVs in in vitro and in vivo models. Thus, hucMSC-EVs can prevent hepatic oxidant injury through the delivery of GPX1 (Yan, et al., 2017). Human menstrual blood-derived stem cell-derived EVs (MenSC-EVs) expressed cytokines, including ICAM-1, angiopoietin-2, Axl, angiogenin, IGFBP-6, osteoprotegerin, IL-6, and IL-8. MenSC-EVs markedly improved liver function, enhanced survival rates, and inhibited hepatocyte apoptosis at 6 h after transplantation. MenSC-EVs prevented liver injury in DGAL/LPS-exposed mice by decreasing the number of liver mononuclear cells and the amount of the active, cleaved caspase-3 (Chen, Xiang, Wang, & Xiang, 2017). Administration of MenSC-EVs attenuated liver injury and significantly down-regulated collagen I, vimentin, α-SMA and fibronectin in mice. Furthermore, transferring miR-181-5p via exosomes to damaged liver cells showed the anti-fibrotic function and would be used for therapeutic delivery of miRNA for targeting liver disease (Qu, et al., 2017). All these results clearly demonstrated the potential usages of stem-cell derived EVs (either via proteins and/or miRNAs) in treating acute liver injury and liver fibrosis.

Recent studies have revealed that small molecule compounds can regulate the biogenesis and release of EVs. Several studies have demonstrated that drugs such as amiloride, fasudil, or inhibitors of neutral sphingomyelinase 2 can inhibit EV release in in vitro and in vivo models to attenuate disease pathogenesis and progression (Trajkovic, et al., 2008). Multiple proteins belonging to the RAB guanosine triphosphatase family may also be targeted to prevent EV release. Although several critical components in the EV biogenesis and release are known, specific inhibition strategies remain largely unexplored. Treatment with cytochalacin-D significantly suppressed the increased amounts of CYP2E1 protein expression and EV release from primary hepatocytes after ethanol exposure. Similar results were also observed with other EV secretion inhibitors such as GW4869 and dimethyl amiloride (Cho, Mezey, et al., 2017), suggesting the possibility of other small molecule compounds in regulating the number and amounts of EV components. Particularly, CYP2E1 inhibitor (CMZ) efficiently blocked the elevated EV release from primary hepatocytes in response to ethanol (Cho, Mezey, et al., 2017). The real challenge of these strategies would be to find very specific targets or small molecule compounds that would not interfere with the normal, physiological functions of EVs derived from normal cells.

Furthermore, EVs can be an ideal vector for delivering of both macromolecules and small-molecule drugs (Luan, et al., 2017) because EVs are composed of a lipid membrane bilayer structure containing the surface ligands and receptors from the source cells. Therefore, EVs could be loaded with macromolecular drugs or chemotherapeutic agents and engineered to deliver these EVs to specifically targeted recipient cells. This would be of a great potential for cancer therapy with minimal toxicity. These carrier vesicles could be obtained from diverse sources ranging from healthy patient-derived cultured cells to milk or plant-derived EVs. For example, drug-loaded EVs from bovine milk showed significant higher efficacy, compared to that observed with a free drug in cell culture studies and against lung tumor xenograft in in vivo models. Moreover, anti-cancer agents, including folate can be included in cancer-cell targeting EVs for enhanced tumor reduction (Munagala, Aqil, Jeyabalan, & Gupta, 2016). Encapsulation of Anthocyanidins onto/into milk EVs can enhance higher drug efficacy with no toxic side effects. Therefore, EVs can provide an effective alternative for oral delivery of Anthocyanidins that is efficacious, cost-effective and safe in treating multiple cancers (Munagala, et al., 2017).

Taken together, EVs have a great potential to be translated into clinical practice as preventive applications or alternative vectors for targeted drug therapies.

7. Future directions

Patient with severe alcoholic hepatitis defined have a 1-month mortality rate as high as 20%-50%, so the 30-day survival was an endpoint for most trials (Lucey, Mathurin, & Morgan, 2009). Assessing predictors of mortality beyond one year are difficult, as the main determinants are the presence of cirrhosis and persistent drinking (Potts, Goubet, Heneghan, & Verma, 2013). However, there were no histologic features of liver biopsy including the presence of bridging fibrosis or cirrhosis that correlated with long-term prognosis (Masson, et al., 2014). Therefore, characterizations of circulating EVs can be a useful option in confirming the prognosis or diagnosis of alcoholic hepatitis instead of liver biopsy.

The heterogeneity of isolated EVs reflects the lack of standardized isolation techniques and different terminologies have been used in the past. Improved methods of specialized isolation and functional characterization of selected EVs have been recently summarized (Lotvall, et al., 2014). However, robust and reproducible methods for the detection, isolation, and characterization of molecularly defined populations of EVs from various organs are lacking. The current methods for isolating hepatic EVs rely on relatively non-specific enrichment strategies such as immune-absorption while characterization of EV-derived cargos are highly diverse and lack standardization of experimental and analytical approaches. It is, therefore, necessary to modify current strategies to develop new methodologies to identify the EVs, specifically derived from different cell or organs, in plasma and saliva, and other bio-fluids. Development of such reliable technologies and reproducible methods would potentially lead to new strategies in prognosis, diagnosis, and interventions. Furthermore, it is unclear yet whether exosomes or different subtype of EVs have the same composition (protein, mRNA, and miRNA population) and functional properties. Potential therapies of using EVs from a variety of cells also need to be investigated for functional mechanisms. Consequently, proper characterizations of biological fluids containing these EVs may have prognostic, diagnostic or therapeutics value in treating alcoholic hepatitis and DILI.

8. Conclusions

In this review article, we have briefly summarized recent findings about the properties and functional roles of circulating EVs in the development and progression of ALD and DILI, their clinical applications, as potential biomarkers and therapeutic agents. The expression and extensive signaling potential of the regulatory miRNAs and non-coding RNAs in the liver and other organs are complex and the researchers are now beginning to understand the complex roles of EVs carrying various miRNAs and protein cargos. EVs contain various biological macromolecules, including proteins, mRNA, microRNA, and lipids, and play important roles in conveying critical information between different cell types in the liver. Recent in vitro and in vivo studies have clearly indicated that circulating EVs and their cargos may have been used as potential biomarkers for identifying certain liver disease and in monitoring host responses to various treatments. EVs can also modulate cellular functions and activate different pathways in recipient cells, by which they can contribute to the initiation, progression and pathogenesis of different liver diseases, as summarized in Fig. 4. Furthermore, exogenous EVs, especially from stem cells, could be used for the potential treatment of various liver diseases by delivering nucleic acids and other drug cargos to hepatic or other target cells and this concept might allow EVs to emerge as a novel therapeutic approach. However, a critical challenge in EVs research has been the poor reproducibility and normalization in different experimental settings, conditions and assay methods among different laboratories. To the present date, various existing methods for EVs isolation are either time consuming or may not produce good quality EVs. However, the recently improved isolation techniques of EVs with reproducible results could significantly aid the future development of potential diagnostic platforms and treatment paradigms for individualized patient care for those suffering from ALD, DILI and other liver diseases.

Fig. 4. Potential EV research in DILI and ALD.

The in vitro and in vivo studies have clearly demonstrated that circulating EVs and their cargos may have been used as potential biomarkers for identifying liver disease and in monitoring host responses to various treatments. EVs can also modulate cellular functions and activate different pathways in recipient cells, which then contribute to the initiation, progression and pathogenesis of different liver diseases. In contrast, EVs particularly from stem cells can be exploited for their therapeutic potentials as novel treatments against various liver diseases.

Abbreviations

EVs

extracellular vesicles

ALD

Alcohol liver disease

ALT

alanine aminotransferase

CYP2E1

ethanol-inducible cytochrome P450-2E1

DILI

Drug-induced liver injury

MSC

mesenchymal stem cells

CCl₄

carbon tetrachloride

AST

aspartate aminotransferase

hucMSC

Human Umbilical Cord Mesenchymal Stem cells

EMT

epithelial-to-mesenchymal transit

ER

endoplasmic reticulum

MSCs

Mesenchymal stem cells

MVBs

multi-vesicular bodies

MVEs

multi-vesicular endosomes

APAP

acetaminophen

ADMSCs

adipose-derived mesenchymal stem cells

CPMSCs

chorionic plate-derived MSCs

GPX1

glutathione peroxidase1

MDVs

Mitochondrial-derived vesicles

MPs

microparticles

mtDNA

mitochondria DNA

LPS

lipopolysaccharide