Introduction

Extracellular vesicles (EVs) are nanosized, membrane-bound vesicles released from eukaryotic and prokaryotic cells that can transport cargo including DNA, RNA, lipids, and proteins, between cells as a form of intercellular communication (Fevrier and Raposo 2004; Mathivanan et al. 2010; Raposo and Stoorvogel 2013; Colombo et al. 2014; Zaborowski et al. 2015). EVs have been found in various body fluids such as amniotic fluid, ascites, bile, blood, breast milk, cerebrospinal fluid, saliva, semen, and urine. In the literature, EV is the general name for various cell-derived vesicles, such as microparticles, microvesicles, nanovesicles, nanoparticles, calcifying matrix vesicles, argosomes, tolerosomes, oncosomes, prostasomes, secretomes, exosomes, exovesicles, exosome-like vesicles, and ectosomes (Colombo et al. 2014; Thery et al. 2018). Though the nomenclature of EVs is still a matter of debate (Gould and Raposo 2013; Thery et al. 2018; Chiang and Chen 2019), the terms ectosomes, microparticles, and microvesicles mainly refer to vesicles ranging in size from 150 to 1000 nm that are released from the cell membrane by the budding process. The term “exosome” was initially used to tag vesicles whose size ranged from 40 to 1000 nm with a 5’-nucleotidase activity (Trams et al. 1981). In the late 1980s, however, the use of this term was restricted to only include vesicles of the endosomal origin ranging in size from 30 to 100 nm (Johnstone et al. 1987; Fevrier and Raposo 2004).

Almost three decades ago EVs were considered to be cellular debris that was deprived of any essential biological function(s). However, emerging literature strongly implicates critical roles of EVs in the context of intercellular signaling, maintenance of tissue homeostasis, modulation of immune responses, inflammation, cancer progression, angiogenesis, and coagulation, under both physiological and pathological states (Hu et al. 2012; Andaloussi et al. 2013; Yanez-Mo et al. 2015; Di Rocco et al. 2016; Iraci et al. 2016; Kalra et al. 2016; Hu et al. 2017a, b). Based on the ability of EVs to shuttle proteins, lipids, carbohydrates, mRNAs, long non-coding RNAs (lncRNAs), microRNAs, chromosomal DNA, and mitochondrial DNA into target cells, the presence and content of EVs in biofluids has been exploited for biomarker research in the context of diagnosis, prognosis and treatment strategies (Walker et al. 1988; Valadi et al. 2007; Lin et al. 2015; Ratajczak and Ratajczak 2016; Wang et al. 2017a; Samanta et al. 2018; Abdel-Haq 2019). EVs are comprised of lipid bilayer membranes coated with various ligands, which, in turn, can interact with receptors on target cells, thereby making these vesicles promising candidates for targeted delivery (Agrahari et al. 2019). Due to their increased stability in circulation and biocompatibility, as well as low immunogenicity and toxicity, EVs are attractive systems for transport and delivery of therapeutics. EVs are increasingly being exploited as conduits for delivery of therapeutics for anticancer strategies, immunomodulation, targeted drug delivery, tissue regeneration, and vaccination (Gyorgy et al. 2015; Ohno et al. 2016). In this review, we highlight and discuss the various strategies employed for the use of EVs as delivery vehicles, including the potential advantages and challenges involved.

Basic and Therapeutic Implications of EVs

EVs can be secreted in vitro by a variety of cells including adipocytes, fibroblasts, glial cells, hematopoietic cells (B cells, T cells, dendritic cells, mast cells, and platelets), intestinal epithelial cells, neuronal cells, Schwann cells and numerous tumor cell lines (Yanez-Mo et al. 2015; Hu et al. 2016a). Additionally, in vivo EVs exist in various biological fluids including blood, urine, saliva, epididymal fluid, amniotic liquid, malignant and pleural effusions or ascites, bronchoalveolar lavage fluid, synovial fluid and breast milk. Within these fluid compartments, EVs serve as mediators for cellular communication and cargo transportation, thereby regulating various physiological processes (Schorey and Bhatnagar 2008; Kooijmans et al. 2012; Vlassov et al. 2012; Hu et al. 2013; Antimisiaris et al. 2018; Bunggulawa et al. 2018; Yang et al. 2018a). EVs bear combinations of ligands that engage different cell surface receptors simultaneuosly and can communicate without the need of direct cell-to-cell contact. For example, EVs can transfer MHC II / peptide complex from antigen presenting cells to T cells and subsequently, antigen presentation to secondary T lymphocytes, thereby facilitating antigen-specific communication between nonadjacent APC and T cells (Arnold and Mannie 1999). Additionaly, EVs stimulated from dendritic cells in response to IL10 treatment, suppressed inflammation and collagen-induced arthritis in mice, thereby underscoring the use of EVs as a better therapeutic approach compared with DCs for the treatment of autoimmune diseases such as rheumatoid arthritis (Kim et al. 2005; Schorey and Bhatnagar 2008). In addition to the role of EVs in antigen-specific communication, EVs released from epithelial cell origin are known to carry antimicrobial peptides such as cathelicidin-37 and beta-defensin 2, which during the infection by a protozoan parasite Cryptosporidium parvum, leads to increased release of EV, thereby resulting in protection of epithilial cells (Hu et al. 2013). Additionally, EV-cargo such as miRNAs and lncRNAs are known to be stabilized in circulation via protection of the vesicular structure, and are subsequently transferred to target cells to inhibit the expression of target genes. EV-miRNAs have also been shown to trigger malignancy by entering the tumor microenvironment. For example, Felicetti et al., demonstrated that vesicles released from miR-222-overexpressing cells were able to transfer miR-222-dependent malignancy to recipient primary melanomas (Felicetti et al. 2016). EVs are also involved in a variety of physiological events such as the cross talk among glial cells. As an example, EV-miR-9 released from HIV Tat protein stimulated astrocytes can be taken up by microglia resulting in increased migration of the latter cells (Yang et al. 2018a). A study by Hu et al. has also demonstrated that lincRNA-Cox2 expression was increased in EVs derived from astrocytes exposed to morphine, in turn, leading to impaired phagocytosis in microglial cells (Hu et al. 2018).

During erythrocyte maturation, EV secretion serves an excretory function by which the unwanted proteins and RNA are cleared from the cells. However, in cells that lack efficient degradation capability or are located in close proximity to a drainage system such as the tubules of the kidney or the gut, it is EV release rather than lysosomal processing that is beneficial for the cells (Johnstone et al. 1987; Johnstone 2006; Vlassov et al. 2012).

The composition of the EV is primarily governed by the physiological state of its environment as well as the type of producer cell. While the membranes of all EVs are enriched with cholesterol (Morelli et al. 2004; Llorente et al. 2013), glycosphingolipids (Llorente et al. 2013), and phosphatidylserine (Laulagnier et al. 2004; Morelli et al. 2004), the exact lipid profile of specific EVs tends to be similar to, yet distinguishable from that of its cell of origin (Vidal et al. 1989). The proteomic content of the EV is multifactorial; some proteins are present in most EVs, including HSP70, Alix, CD6, CD81, CD9, and major histocompatibility complex class II proteins (Simpson et al. 2009; Mathivanan et al. 2012; Pleet et al. 2018). Other proteins are associated with specific EV subsets, including receptors and other membrane proteins, that confer various functions to the EV. As with lipids and nucleic acids, the proteins incorporated into EVs are related to, but distinct from, the overall protein pool in the cell of origin, suggesting the existence of an intracellular sorting mechanism that helps to determine the EV protein content. The nucleic acid content of EVs is also variable, including various types and quantities of DNA, ribosomal RNA, mRNA, and non-coding RNAs such as miRNAs and lncRNAs.

The mechanism(s) by which EVs interact with their recipient cells still remain elusive. EVs are proposed to interact primarily via the docking of the ligand on the vesicle surface to the receptor(s) on the recipient cells. This docking elicits a signaling response, followed by the transfer of membrane proteins from the vesicle to the cell membrane, fusion of the vesicle with the recipient cell membrane, vesicle uptake through endocytotic processes (clathrin-coated pits, pinocytosis, caveolae, macropinocytosis, and phagocytosis), and ultimately extrusion through a vesicle-cell channel (de Curtis and Meldolesi 2012; Mittelbrunn and Sanchez-Madrid 2012). The fate of vesicular components in recipient cells could depend on the mode of uptake, with processing through the endosomal pathway potentially leading to degradation of EV contents (Tian et al. 2013). Although the mechanism(s) of cargo transfer remains to be elucidated, it is well-recognized that endogenous EVs can exert diverse and potent effects on recipient cells. The diversity of mechanisms by which EVs are generated and can confer functional effects provides a platform for both opportunities and challenges for developing EV-based therapeutics.

In recent years, various novel EV functions have been elucidated, with much of the diversity of the functions ascribed to their cell of origin (Vlassov et al. 2012). For instance, EVs have been investigated as immune response mediators with roles specifically in antigen presentation (Thery et al. 2002, 2009). Furthermore, the role of EVs in angiogenesis, apoptosis, coagulation, and inflammation has now been well-established (Janowska-Wieczorek et al. 2005; Becker et al. 2016; Todorova et al. 2017; Fu et al. 2018; Fujita et al. 2018; Deng et al. 2019; Silachev et al. 2019). Emerging literature has also demonstrated that distinctive properties of EVs make them suitable carriers and vehicles for delivery of various drugs and biomolecules, thereby underscoring their use in therapeutic applications (Table 1) (Srivastava et al. 2016b, a). EVs generated from various cell types, including but not restricted to stem cells, stromal cells, progenitor cells, neuronal cells, cancer cells, and circulating cells, have been tested for their therpaeutic efficacy involving delivery via intraperitoneal, intranasal, intrathecal and intravenous routes in various in vivo model systems of disease pathogeneis. Role of administration routes for EV drug delivery in animal models will be further discussed in section 5 of this review.

Table 1 Origin and therapeutic application of EVs
Full size table
Table 2 Examples of EV engineering
Full size table

Isolation and Characterization of Blood-Derived EVs

EVs can be derived from various sources, including blood, and have been shown to exhibit a change in their composition as well as numbers, under various pathological conditions. It has been shown that blood-derived EVs from healthy individuals can be derived from endothelial cells, erythrocytes, leukocytes, megakaryocytes and/or platelets. Under diseased states however, the numbers and composition of these EVs has been shown to be altered (Zara et al. 2019). As an example, there are reports demonstrating increased numbers of EVs derived from endothelial cells of patients with systemic lupus erythematosus and cardiac failure and this was shown to positively correlate with increased risk for cardiovascular problems (Nozaki et al. 2010; Parker et al. 2014). It has also been shown that increased platelet-specific EVs are a biological marker for cerebral dysfunction(s) in patients with malaria and furhter, that platelet-derived EV numbers are directly associated with coma depth and thrombocytopenia (Pankoui Mfonkeu et al. 2010; Sierro and Grau 2019). Several excellent review articles have described the biology and role of EVs in various disease pathogenesis (Brites and Fernandes 2015; Withrow et al. 2016; Zhang et al. 2016c; Gopal et al. 2017; Huang-Doran et al. 2017; Kinoshita et al. 2017; Castro-Marrero et al. 2018; Fujita et al. 2018; Konoshenko et al. 2018; Letsiou and Bauer 2018; Murphy et al. 2018; Yang et al. 2018b; Aghabozorgi et al. 2019; Pegtel and Gould 2019; Watson et al. 2019; Wu et al. 2019; Zara et al. 2019; Zhu et al. 2019), and the isolation and characterization of EVs from various cells including blood (Aatonen et al. 2014; Nguyen et al. 2016; Xu et al. 2016; Menck et al. 2017; Mushahary et al. 2018; Gorgun et al. 2019; Kim et al. 2019; Pulliam et al. 2019; Richter et al. 2019; Rossi et al. 2019; Skalnikova et al. 2019; Weber et al. 2019).

Bioengineering of EVs

In order to boost their therapeutic potential, EVs can be bioengineered through modifications such as the loading of drugs or attachment of molecules to their surface. Another type of bioengineered EVs relies on the development of artificial exosomes, exosome-based semisynthetic vesicles, exosome-like nanovesicles, and exosome-mimetic nanovesicles (De La Pena et al. 2009; Bryniarski et al. 2013; Jang et al. 2013; Forterre et al. 2014; Jeong et al. 2014; Yoon et al. 2015). These two main categories of EV bioengineering will be referred to in the following sections as engineered EVs and EV mimetics.

Engineered EVs

As mentioned previously, engineered EVs are primarily modified through the loading of drugs as well as via alteration or attachment of molecules on their surface, to enhance delivery efficacy of the therapeutic contents. The in vivo clearance of unmodified EVs is very rapid following their administration. Thus, these engineered surface modifications are meant to extend the biodistribution, stability, and pharmacokinetic profiles of the EVs, thereby facilitating the proposed drug delivery. Several examples of successful engineering of EVs exist in the literature, with the most recent examples listed in Table 2.

EV mimetics

A possible substitute for naturally derived or purified EVs in the development of drug delivery systems and therapeutics is synthetically designed EV mimetics. Synthesis of EV mimetics permits scalable production for use in the clinical setting and provides additional advantages over naturally occurring EVs, in that EV mimetics are sterile and uniform in size and content. EV mimetics, however, do not always function in the same way as endogenous EVs due to the lack of several additional components that are essential for the primary functions of the EVs. Furthermore, the process of screening the core component, extraction, and incorporation of the screened core components into the liposomal complex is cumbersome and labor intensive. Despite these limitations, several studies have begun to evaluate the use of EV mimetics for therapeutic applications. To date, three primary sub-types of EV mimetics have emerged: artificial EV mimetics, physical-origin EV mimetics, and hybrid EV mimetics (Antimisiaris et al. 2018). Recent discussions on nanoparticles for drug delivery include an elegant review by DeMarino et al. covering various aspects of nanoparticle formulation and their applications in improving the delivery efficiency of drugs (DeMarino et al. 2017).

Artificial EV mimetics

Artificial EVs are generated through the assembly of lipids into a bilayer model to resemble the membranes of natural EVs, followed by functionalizing the vesicle surface with proteins or other modifications, thereby allowing the surface to have direct contact with the receptors of the target cells. In some cases, artificial EVs are also tagged with hydrophilic molecules to decrease their elimination and extend their time in circulation. One limitation of artificial EV mimetics is that it is based on the premise that EV function does not require all the components of natural EVs for target-specific, efficient drug delivery. Examples of artificial EV mimetics that have been produced for drug discovery and therapeutics are listed in Table 3. It is important to note that most of the artificial EV mimetics investigated to date are primarily liposomes.

Table 3 Examples of artificial EV mimetics
Full size table

Physical-Origin EV Mimetics

In this category of EV-mimetics, the starting material is not artificial, but rather is derived from other non-EV cellular components. These includeEV mimetics derived from whole cells (termed “cellular vesicles” or “cell-derived vesicles”). Nanovesicles can be generated from whole cells using a variety of techniques, including extrusion through nanopores or cutting the cells with blade-lined microchannels (Jo et al. 2014; Yoon et al. 2015). These physical-origin EV mimetics are able to overcome some of the limitations of other types of EVs such as the low-yield of EVs isolation from cell media or other sources and the lack of a true physiological cell membrane in artificial EV mimetics. Cell-derived vesicles make up the majority of the physical-origin EVs that have been investigated to date (Table 4), demonstrating promising features that could augment the efficiencies of drug targeting (Fuhrmann et al. 2015a).

Table 4 EV mimetics and their therapeutic applications
Full size table

Hybrid EV Mimetics

Other types of EV mimetics have also been described in the literature; the most common other type of EV mimetic is a hybrid model. Hybrid EV mimetics link EVs to another biological messaging system in order to take advantage of the characteristics of both systems. For example, exosomes have been fused with liposomes, thereby altering the cellular uptake of the EV through changes in the lipid composition or the properties of the lipids making up the liposome (Sato et al. 2016). Another example is the fusion of non-enveloped viruses with EVs to create virus-EV particles (Feng et al. 2013). This natural defense of the virus allows it to escape neutralizing antibodies using the EV-like membrane as camouflage.

Methods of Preparation and Engineering of Engineered EVs and EV-Mimetics

EV Loading

On the basis of EVs biogenesis, the methods of EV loading have been primarily categorized as follows: (a) strategies including site of exosomal functional entities, e.g. transmembrane proteins and the use of their natural tropism to co-localize the exogenous components; (b) strategies involving the use of molecular mechanisms for the effective incorporation of exogenous molecules into EVs for their cytosolic delivery; and (c) strategies involving enrichment of the quantity of molecules into the cellular plasma membrane to be encapsulated by passive mechanism during multivesicular body formation. The techniques for the production of bioengineered EVs are generally classified by the presence or timing of EV isolation, including: (a) pre-loading modifications, (b) post-loading modifications, and (c) creation of artificial mimetic structures of the natural exosomes.

Pre-loading Modifications

The mechanisms of pre-isolation engineering can be grouped into the following three primary categories: (a) use of “exosome display” to engineer the expression of transmembrane proteins for co-localization to exogenous entities; (b) use of molecular mechanisms to directly incorporate exogenous molecules into EVs for their cytosolic delivery; and (c) enriching the quantity of molecules into the origin cell’s plasma membrane to be encapsulated by passive mechanism during multivesicular body formation (Garcia-Manrique et al. 2018). In each of these cases, the drug of interest is loaded directly into or onto the surface of the parental source cells and, as a result, the EVs are released or isolated from the source cells pre-loaded with the drug of interest. This approach is often employed when specific oligonucleotides or proteins of interest are to be loaded in the EVs wherein the parental source cells are designed to release the EVs that are pre-loaded with either the specific oligonucleotides or the protein of interest. Pre-loading modifications of EVs can be achieved by the treatment of parental source cells with drugs of interest or engineering of parental source cells (Luan et al. 2017).

In a simple example of pre-loading modification, the parental source cells are exposed to the drug of interest for a stipulated time, resulting in the drug-exposed cells secreting EVs that are pre-loaded with the drug of interest. While this method is simple, it is limited by a lack of control over the loading efficiency of the drug into the secreted EVs. Despite the limitations, several studies have successfully used this approach. One such study exposed the murine mesenchymal stromal cell line, SR4987 treated with paclitaxel for 24 h, and found significant anti-proliferative effects on CFPAC-1 cells (a paclitaxel-sensitive, human pancreatic cell line) when compared with cells that were treated with conditioned medium from untreated mesenchymal stromal cells (Pascucci et al. 2014). In another study, human adipose-derived mesenchymal stem cells that were incubated in p5 (a peptide derived from p35) for 24 h were able to release biologically functional p5 to inhibit p35 cleavage, CDK5 phosphorylation and calpain-mediated p53 upregulation in bovine aortic endothelial cells. The p5-incubated cells protected the aortic endothelial cells from stress like hypoxia/ischemia, oxidative stress, and inflammation (Fang et al. 2016).

The effect of drug exposure is not limited to cellular uptake of the drug, as the exposure can also cause reactive changes within the cell that are reflected in the secreted EVs. For example, exposure of human hepatocellular carcinoma cells to heat shock or anticancer drugs such as paclitaxel, carboplatin, etoposide, or irinotecan hydrochloride result in the release of EVs that are loaded with heat shock proteins on their membrane surface. These heat shock protein-bearing EVs can elicit anti-tumor effects in natural killer cells, in vitro (Lv et al. 2012). In another study, human hepatocellular carcinoma cells were exposed to the histone deacetylase inhibitor MS-275 for 72 h, and following exposure, EVs that were isolated from the culture medium demonstrated increased cytotoxicity of natural killer cells and increased proliferation of peripheral blood mononuclear cells, thereby suggesting a promising therapeutic strategy against hepatocellular carcinoma (Xiao et al. 2013). Recently, Yuan et al. have also demonstrated the anti-cancer potential of EVs that were released from human tumor necrosis factor-related apoptosis-inducing ligand (TRAIL)-transduced mesenchymal stromal cells in 11 different cancer cell lines, suggesting the efficacy of these EVs in inducing selective apoptosis in various cancer cells (Yuan et al. 2017).

Extrusion

Extrusion is a process by which EVs are derived from cells through filters of reducing pore sizes (Jang et al. 2013). The vesicles are produced artificially by breaking up the cells and reforming the contents in the exosome mimetics. This technique has shown to produce higher quantities of EVs when compared to the EVs released by the cells (Jang et al. 2013; Jang and Gho 2014; Lunavat et al. 2016). Investigators have used this method to develop the exosome mimetic nanovesicles to effectively deliver chemotherapeutics such as doxorubicin, 5-fluorouracil, gemcitabine, and carboplatin and study their effects on tumor growth (Jang et al. 2013). EVs were also harvested from the same cells to compare the efficacy of EVs to that of the exosome mimetic nanovesicles. Both of the vesicles have similar efficacy in reducing tumor growth, however, when compared to free drug exosome mimetics were more efficient compared to natural EVs. Another interesting finding was that when the exosome mimetics were isolated from the two cell lines containing cancer drugs and injected into an immunocompetent mouse tumor model, they both exhibited similar anti-tumor effects with no reported systemic side effects. Other investigators have shown that loading of RNAi in the exosome-mimetic nanovesicles was therapeutically active (Lunavat et al. 2016). From this study, it was reported that both exogenous and endogenous loading methods were efficient to cause a reduction in the expression levels of c-Myc. The positive findings of the study imply that the exosome-mimetic nanovesicles could, in fact, be used to overcome some of the scale-up issues currently associated with the development of EV therapeutics.

Microfluidic Method

This method has been recently used for purification of vesicles from cell media or biological fluids (Wang et al. 2013; Liang et al. 2017; Liu et al. 2017; Wu et al. 2017) and whole cells (Yoon et al. 2015). This method has been used in the preparation of liposomes and other types of nanoparticles for drug delivery. In fact, it has been established that one-step, the fully automated, and the scalable microfluidic system can be used for ligand-targeted liposomes (Ran et al. 2016; Rosenblum et al. 2018). This technique can also be used to prepare exosome mimetics for efficient drug delivery, however, this remains unexplored until now.

Post-Isolation Modifications

Incubation with Drugs

Through a similar process, as is used pre-isolation, drug loading of EVs can also be performed by incubating the EVs post-isolation with the drug. Liposomes have long been used to improve the therapeutic and pharmacokinetic profiles of therapeutic drugs through increased bioavailability and retention in the target tissues, although opsonization and rapid clearance continue to be a significant hurdle for some of these nanoparticles (Zhang et al. 2005).

Recent reports have shown that curcumin, doxorubicin, and paclitaxel can be passively loaded within EVs to improve their therapeutic efficacy (Sun et al. 2010; Zhuang et al. 2011; Yang et al. 2015). Doxorubicin- and paclitaxel-loaded EVs have been demonstrated to cross the blood-brain barrier in zebrafish (Yang et al. 2015), with paclitaxel-loaded EVs demonstrating anti-tumorigenic effects (Pascucci et al. 2014; Rani et al. 2015). Curcumin, on the other hand, interacts with the lipid membrane of the EV to form a complex which, upon administration to macrophages, exhibits better anti-inflammatory efficacy than curcumin delivered alone (Sun et al. 2010). Curcumin complex has also been delivered in vivo in the lipopolysaccharide (LPS)-induced mouse model of shock. This study demonstrated the stability of the curcumin complex over a longer period, and also showed that administration of curcumin-loaded EVs intranasally protected mice from LPS-induced brain inflammation and autoimmune encephalomyelitis, and delayed tumor growth (Zhuang et al. 2011). Based on these findings, a phase I clinical trial (NCT01294072) is currently ongoing to evaluate the efficacy of plant exosomes to deliver curcumin drug to colon cancer patients.

Sonication

Several techniques have now been developed to increase the efficiency of transferring drug into EVs, including sonication. In this process, EVs are mixed with the drugs or proteins of interest followed by sonication with a homogenizer probe. The sonicator induces mechanical shear forces that affect the EV membrane integrity and allow increased drug entry into the EV (Kim et al. 2016). Although the membrane integrity is affected, the sonication process does not appear to alter other contents within the EVs significantly, and the membrane integrity is restored within an hour of incubation. In addition to being encapsulated inside the EV following sonication, drugs could also adhere to the outer surface of the EV membrane, resulting in two phases of drug release. The first burst release phase results from the release of the drug adhered to the outer membrane of the EVs, followed by the slow release of the drug encapsulated inside the EV (Kim et al. 2016).

Electroporation

A popular method to load cargo into EVs is electroporation, a process by which transient pores are made into the membranes of the EVs. In this method, purified EVs and the therapeutic cargo are mixed together in a buffer followed by electroporation and incubation (Shtam et al. 2013; Tian et al. 2014; Lamichhane et al. 2015). After incubation, the EVs are washed with PBS to remove unloaded cargo followed by ultracentrifugation. A study by Alvarez-Erviti et al. successfully engineered bone marrow dendritic cells to express rabies virus glycoprotein peptide that was fused to an EV membrane expressing Lamp2b. Intravenous injection of these EVs to mice resulted in neuron-specific gene silencing (Alvarez-Erviti et al. 2011). Similar studies have been carried out with RAD51, luciferase and MAPK1 siRNAs loaded into EVs through electroporation and delivered to HeLa cells, endothelial cells, monocytes and lymphocytes respectively (Shtam et al. 2013; Banizs et al. 2014). Other investigators have also loaded dsDNAs and chemotherapeutic drugs in EVs using this technique (Tian et al. 2014; Lamichhane et al. 2015). Overall, electroporation is one of the most useful techniques for delivery of siRNA, DNA, chemotherapeutic agents as well as miRNA, mRNA and proteins into EVs. Although this technique results in a minimal effect on the EV components, it may produce aggregation (Hertzberg and Wolff 1990; Weaver 1993; Hood et al. 2014; Johnsen et al. 2016) and lacks significant scalability, that would be necessary for large clinical investigations.

Saponin Assisted Loading

Another method for EV loading is the permeabilization of the EV membrane through the use of saponin. Saponin is a detergent-like molecule that interacts with cholesterol in the EV membrane resulting in pore formation (Jacob et al. 1991; Jamur and Oliver 2010). This technique was used in a study assessing the use of catalase-loaded EVs derived from macrophages for drug delivery in Parkinson’s disease, which resulted in protection against oxidative stress and neurodegeneration (Haney et al. 2015). The authors compared different loading techniques and showed that EVs loaded by saponin permeabilization showed no alterations in EV size or morphology and had similar loading efficiencies and sustained release compared to sonication and extrusion methods. The EVs that underwent sonication appeared to have more non-spherical morphology than those undergoing saponin permeabilization (Haney et al. 2015). Others have shown similar success with saponin in preparing a porphyrin-EV complex, which was shown to be taken up by MDA cells (Fuhrmann et al. 2015b). Although saponin permeabilization is a simple and easy procedure for loading therapeutic proteins, it has not yet been well-studied. Additionally, it is important to ensure removal of the saponins after use, as prolonged exposure may affect the EV morphology, uptake, and stability.

Freezing and Thawing

Another simple method of instilling drug within the EVs is through freeze and thaw cycles. In this method, drugs are incubated with the EVs at 37°C followed by rapid freezing at -80 °C and then thawed to room temperature. This process is repeated for a minimum of 3 cycles for drug encapsulation (Sato et al. 2016). The major drawback of this procedure is that it can induce aggregation, and it tends to result in lower drug loading than many of the other methods. Of note, however, this method can be used for fusion of exosomes with liposomes to develop exosome mimetic particles (Sato et al. 2016).

Surface Modification Method

The proteins located on the surface of EVs are significantly associated with the biodistribution characteristics of the EV. Modification of the surface proteins through gene transfer vectors can, therefore, improve the targeting efficiency of the exosomes (Sato et al. 2016). Some of the transmembrane proteins that are most commonly altered include tetraspanins, Lamp-2b, glycosyl-phosphatidyl-inositol, platelet-derived growth-factor receptors, and lactadherin (Mentkowski et al. 2018). For example, fusing rabies viral glycoprotein with Lamp-2b on EVs results in specific delivery of the EVs to neurons and glia (Alvarez-Erviti et al. 2011; Liu et al. 2015). Similarly, immature dendritic cells have been modified to express Lamp-2b fused with αv integrin-specific iRGD peptide to target tumor cells (Tian et al. 2014). Several methods have been adapted for surface modification of EVs while ensuring that functionality is retained. One such method is copper-catalyzed azide-alkyne cycloaddition (CCAAC) (Smyth et al. 2014; Oude Blenke et al. 2015; Wang et al. 2015) which has been used successfully for delivering chemotherapeutics (Lee et al. 2016). Recently, modified methods have been developed for loading and surface modification of exosomes. Anchor peptides such as CP05 have been demonstrated to aid in targeting, loading and purification of diverse-origin-exosomes through binding to CD63- exosomal surface protein. Furthermore, it was shown that exosomal anchor peptide could be used as a tool for exosomal engineering, probing gene function in vivo, as well as targeted therapeutic drug delivery (Gao et al. 2018).

Methods of EV Delivery

Therapeutic efficacy and toxicity of EVs are critically influenced by their biodistribution (Wiklander et al. 2015). For relevance in a clinical setting, EVs must be stable and capable of delivering their cargo through the commonly used (preferably non-invasive) administration routes. Here we compare various administration routes currently used for effective EV delivery of therapeutics in vivo (Johnsen et al. 2014; Lener et al. 2015).

It has been well-established that systemic administration of EVs results in accumulation in the liver, kidneys and spleen resulting in the rapid removal of the EVs from blood circulation. Multimodal imaging of systemically administered luciferase-loaded EVs in vivo revealed that the half-life of EVs was less than 30 minutes in most tissues and the EVs were completely cleared from the animals by 6 hours (Lai et al. 2014). Likewise, a pharmacokinetic analysis revealed that the half-life of EVs loaded with luciferase-lactadherin fusion protein in the circulation is approximately 2 minutes and only weakly detectable after 4 hours, indicating rapid clearance in vivo (Takahashi et al. 2013). These results are in line with previous studies demonstrating that EVs can be detected in the liver and/or spleen, but not in circulation, at 24 h after systemic administration (Peinado et al. 2012; Ohno et al. 2013).

Despite the advanced development of drugs for the treatment of various brain disorders, delivery of these drugs to the brain, however, remains a significant challenge because of difficulty in penetrating the blood-brain barrier (Gabathuler 2009). Intranasal delivery provides a practical, noninvasive method for delivering therapeutic agents to the brain, the quantities of drug administered nasally and that are transported directly from nose to brain, however, is small (Johnson et al. 2010). Zhuang and colleagues showed that intranasal delivery of EV-encapsulated curcumin or Stat3 inhibitor, JSI-124 (cucurbitacin I), resulted in the compounds reaching microglial cells. Additionally, administration of the curcumin or JSI-124 EVs inhibited LPS-induced microglial cell activation, delayed experimental autoimmune encephalomyelitis (EAE) disease, and inhibited tumor growth in vivo (Zhuang et al. 2011). Recently, intranasal delivery of lincRNA-Cox2 siRNA loaded EVs showed decrease of LPS-Induced microglial proliferation in mice (Liao et al. 2019). Table 5 summarizes the administration routes used for EV drug delivery in the studies published to date.

Table 5 Administration routes used for EV drug delivery
Full size table

Therapeutic Applications of EVs

With increased promising preclinical and early clinical evidence, exploring the potential of EVs as therapeutic agents have attracted a lot of attention and have made its way into the clinics. As shown in Table 6, EVs are currently being tested as drug delivery vehicles in several different trials. Given the ability of EVs to modulate various responses in recipient cells and strong candidacy as a biomarker for a variety of diseases, there is a growing interest in using them as therapeutic entities (Fais et al. 2016). In non-small cell lung cancer (NSCLC), tumor-associated antigen (TAA) loaded dendritic cell-derived EVs (Dex) showed immunomodulatory response and underwent phase I clinical trial (Escudier et al. 2005; Morse et al. 2005). In this study, MAGE antigen-loaded Dex therapy in 15 MAGE3+ advanced melanoma patients resulted in no detectable MAGE-specific T cell responses in peripheral blood, although enhanced NK cell effector functions were observed in 8 out of 13 patients (Escudier et al. 2005). In another study, 3 out of 9 patients with advanced MAGE+ NSCLC who received MAGE3 A1-loaded Dex successfully developed MAGE3 A1-specific systemic immune responses as determined by delayed type hypersensitivity (DTH) reactivity, although only minimal increases in peptide-specific T cell activity were detected (Morse et al. 2005). The demonstration of Dex administration safety profile, the feasibility of therapy and success in some patients resulted in the clinical phase II trial for the treatment of non-small-cell lung cancer patients (Besse et al. 2016). In this study, to overcome the minimal peptide-specific activity, TLR4L- or interferon (IFN)-γ-maturated Dex was used to induce greater T cell stimulation compared to Dex from immature DCs (Segura et al. 2005; Viaud et al. 2011).

Table 6 Clinical trials involving EVs as therapeutic agents
Full size table

In another series of cancer trials, the ascites-derived exosomes in combination with the granulocyte-macrophage-colony-stimulating factor (GM-CSF) in the immunotherapy of colorectal cancer (CRC) were proven safe and well tolerated in phase I clinical trial. The ascites-derived exosomes isolated by sucrose/D(2)O density gradient ultracentrifugation is 60-90-nm vesicles that contain the diverse immunomodulatory markers of exosomes and tumor-associated carcinoembryonic antigen (CEA) (Dai et al. 2008). The above studies indicate that this therapeutic concept is safe and feasible, thus reinforcing the use of EVs as a new therapeutic approach against diseases (Figs. 1 and 2).

Fig. 1
figure 1

Bioengineered EVs. a Schematic showing classification of bioengineering of EVs for effective drug delivery systems; b Schematic showing various route of administration of EVs in the in vivo model systems

Full size image
Fig. 2
figure 2

Examples of organ systems targeted by EV administration. The organ disease-specific studies have utilized various EV-related interventions, including: a umbilical cord mesenchymal stem cell (MSC)-derived, cardiac progenitor cell-derived, hypoxic cardiosphere-derived, or embryonic stem cell-derived exosomes, or curcumin-loaded exosomes; b hypoxia preconditioned MSC-derived exosomes or bone marrow-derived A1 exosomes; c MSC-derived exosomes; d umbilical cord MSC-derived or urine-derived stem cell exosomes; e MSC-derived exosomes; f iron oxide exosomes, tumor associated antigen containing exosomes, paclitaxel containing exosomes, trastuzmab-emtansine containing exosomes, or GM-CSF expressing embryonic stem cell-derived exosomes; g adipose-derived, induced pluripotent stem cell-derived, or umbilical cord MSC-derived exosomes; h MSC-derived or adipose-derived stem cell exosomes; i miR-223 containing exosomes; j LDL-stimulated macrophage-derived exosomes

Full size image

Conclusions and Future Perspectives

The significant advancements in the knowledge surrounding the biology of EVs over the past several years have opened new avenues in the field of life sciences, especially in medicine. Not the least of these is the work that has been done in investigating the role of EVs in both health and diseases, resulting in novel prospects for the advancement of enriched therapeutic EVs. These interventions could help in the synthesis of new cargos inspired by natural vesicles or conventional synthetic alternatives (liposomes, polymersomes, inorganic nanoparticles, and so on) without serious inconveniences. Furthermore, the development of EVs for drug delivery has generated significant excitement in the field; however, the main limitations/challenges of EVs as genuine therapeutic agents include developing methods for efficient, large-scale clinical grade production, isolation, storage, modification, purification as well as target delivery. For example, storage and retrieval conditions of EVs and EV mimetics can can alter their characteristics (Thery et al. 2018). Although efforts have been made in this regard (Zhou et al. 2006; Yuana et al. 2015; Reiner et al. 2017; Leiferman et al. 2019), currently there are no standard operating procedures for long term storage of various types of EVs. Additionally, organ- or cell- specific delivery of therapeutics with EVs poses yet another challenge. Indeed, targeted delivery of EVs has gained increasing attention in the field (Table 5). Recent investigations are focused on overcoming these limitations by establishing artificial EV mimetics or by generating vesicles from membrane fragments created by the extrusion or slicing of cells. Ultimately, regardless of the methods used, the development of multidisciplinary teams with skills in applied biology, pharmacology, chemical engineering, material sciences, and medicine will be required to translate EV-based therapy to clinical practices successfully.