Clinical applications for exosomes: Are we there yet?

Exosomes are a subset of extracellular vesicles essential for cell–cell communication in health and disease with the ability to transport nucleic acids, functional proteins and other metabolites. Their clinical use as diagnostic biomarkers and therapeutic carriers has become a major field of research over recent years, generating rapidly expanding scientific interest and financial investment. Their reduced immunogenicity compared to liposomes or viral vectors and their ability to cross major physiological barriers like the blood–brain barrier make them an appealing and innovative option as biomarkers and therapeutic agents. Here, we review the latest clinical developments of exosome biotechnology for diagnostic and therapeutic purposes, including the most recent COVID‐19‐related exosome‐based clinical trials. We present current exosome engineering strategies for optimal clinical safety and efficacy, and assess the technology developed for good manufacturing practice compliant scaling up and storage approaches along with their limitations in pharmaceutical industry.

. These components are a signature for their cell origin and their composition can alter their pharmacokinetic attributes (Kalluri & LeBleu, 2020). Exosomes are extracellular vesicles (Pan & Johnstone, 1983) generated in the cytoplasm in the multivesicular body, which is defined by the presence of intraluminal vesicles originating from inward budding into the endosomal lumen. Multivesicular bodies get transported to the plasma membrane leading to fusion with the cell surface whereby the intraluminal vesicles get secreted as exosomes (Colombo et al., 2014).
Extracellular vesicles are classified into three groups based on their sizes:-exosomes (40-160 nm) (Kalluri & LeBleu, 2020), microvesicles (100-1000 nm) and apoptotic bodies (>1000 nm) (Andaloussi et al., 2013;Pegtel & Gould, 2019). The size and marker overlap between exosomes and microvesicles, and make their differentiation complex (Corso et al., 2017;Doyle & Wang, 2019). However, the International Society for Extracellular Vesicles (ISEV) proposes a series of references to characterise exosomes based on their function or composition (Théry et al., 2018). Their reduced immunogenicity compared to liposomes or viral vectors, and their ability to cross major physiological barriers like the blood-brain barrier make them an attractive and innovative option as biomarkers and therapeutic agents ( Figure 1). Indeed exosomes, which are present in a wide range of body fluids, could be ideal biomarkers particularly for early detection of diseases. To support translation and ever-expanding portfolio of clinical trials, various strategies are being developed to optimise cargo loading, improve cell type targeting and increase exosome uptake.
Good manufacturing practice of exosomes requires robust scaling up and storage processes (Whitford & Guterstam, 2019).
Here, we review the latest clinical developments of exosome biotechnology for diagnostic and therapeutic purposes, including the most recent COVID-19-related exosome-based clinical trials. We present current exosome engineering strategies for optimal clinical safety and efficacy and assess the technology developed for good manufacturing practice compliant scaling up and storage approaches along with their limitations.
Some of the tetraspanins, such as CD9, CD63, CD81 and CD151, often found at the exosomal membranes (Hoshino et al., 2015;Thery et al., 2002) can be used as screening tools in health and disease. For example, increased CD81 correlates with fibrosis progression and has been proposed as a diagnostic biomarker for complications in viral hepatitis C Welker et al., 2012).
The secretion of extracellular vesicles from different types of tumour cells is a significant method of conditioning and altering the tumour microenvironment by malignant cells (Harris et al., 2015).
Although exosomes are produced by most cell types, observations support the increased secretion of exosomes under pathological conditions, such as cancer (Ohno et al., 2013). Proteomic analysis of those exosomes secreted under various physiological and pathological conditions has shown significant changes in protein expression F I G U R E 1 From mode of actions to therapies. Exosomes are promising players where their polymorph uses (orange) can influence their purpose (green) in clinical settings  (Duijvesz et al., 2013;Lee et al., 2016). Proteins from various cancers are also involved in similar biological processes and functions. A study performed gene ontology analysis on a variety of differentially expressed proteins derived from exosomes. Results suggested that those proteins were involved in similar biological mechanisms, such as cell adhesion, migration and transport (Sherman & Lempicki, 2009).
Contribution of exosomal proteins to angiogenesis, metastasis, tumour formation and disease makes them appealing biomarkers for diagnosis or prognosis applications in cancer, for example, lysosomalassociated membrane protein 3 (LAMP-3; also known as CD63) in ovarian cancer, lung cancer and melanoma (Hurwitz et al., 2016;Pols & Klumperman, 2009), epidermal growth factor receptor (EGFR) in glioblastoma (Skog et al., 2008), proteoglycan glypican-1 in early stage of pancreatic cancer (Melo et al., 2015) and proteins associated with the EGFR pathway, such as retinoic acid-induced protein 3, Gs-α subunit and resistin in bladder cancer and kidney disorders (Smalley et al., 2008;Zhou et al., 2008).
Exosomes also incorporate nucleic acids such as messenger RNA (mRNA), micro-RNA (miRNA), small interfering RNA (siRNA) and p-element-induced wimpy testis (PIWI)-interacting RNA (piRNA) that display mutations in the host cell (Han et al., 2017;Shtam et al., 2013;Thakur et al., 2014;Valadi et al., 2007). The early detection of a tumour is crucial for successful treatment. In addition to tissue biopsy-  pathophysiology (Sharma, 2018). Several studies have focused their analysis on circulating miRNA from enriched exosomes. A recent study showed that exosomal miR-141 was progressively increased in prostatic hypertrophy, localised prostate cancer and metastatic disease, suggesting a potential diagnostic or prognostic role (Li et al., 2016). A different project compared the performance of whole blood miRNA with exosomal miRNA analysis. The results presented that miR-375 derived from whole plasma could differentiate pancreatic cancer from benign prostatic hyperplasia patients, while exosomal miR-200c-3p and miR-21-5p were better discriminators and Let-7a-5p miRNA in exosomes could distinguish pancreatic cancer patients with different severity scores (Endzeliņš et al., 2017). miRNA-375 levels in exosomes from squamous cells of carcinoma patients determine the progression from local inflammation to carcinoma and are potential early-stage biomarkers for oral carcinoma . Exosomal miRNA-20a-5p, miRNA-24-3p, miRNA106a-5p, miRNA-891a and miRNA-1908 in nasopharyngeal carcinoma affect both cell differentiation and expansion by down-regulating the mitogen-activated protein kinase-1 (MARK1) signalling pathway, hence their prognostic application (Ye et al., 2014). Similarly, exosomal mRNAs can be used as diagnostic biomarkers in body fluids reducing the impact and cost of surgical biopsies (Rabinowits et al., 2009;Taylor & Gercel-Taylor, 2008).
For example, saliva-derived exosomes (Lau et al., 2013), human amniotic fluid-derived exosomes and urine-derived exosomes (Keller et al., 2007)  and invasion and could be used as prognostic and diagnostic markers in cancer development (Martinez et al., 2015). Various studies have also showed that piRNAs could be valuable markers for cancer metastasis, for example, piR-4987 in lymph node metastasis (Huang et al., 2013), piR-932 and PIWIL2 in metastasis of breast cancer  and piR-32051, piR-39894, and piR-43607 in clear cell renal cell carcinoma metastasis, late clinical stage and poor cancer-specific survival (Fu et al., 2015). Recent publications have questioned whether some RNA detected in exosome preparations could be derived from culture media, particularly when using sera, following the most accepted purification method based on ultracentrifugation (Tosar et al., 2017;Wei et al., 2016).

| EXOSOMES AS THERAPEUTICS
While the vast majority of ongoing exosome-based clinical trials aims at identifying diagnostic or prognostic biomarkers (Table 1) (Table 2).

| Cancer therapies
Different exosome-based approaches can be used for cancer therapies like oncogene inhibition. For example, the phase I trial  (Andrews et al., 2001). No results have been disclosed yet.

| Anti-inflammation/immunomodulation therapies
Exosomes are promising delivery agents to treat inflammatory disorders due to their low immunogenicity, intrinsic anti-inflammatory  immune balance (Nassar et al., 2016;Zhao et al., 2012). No results have been disclosed yet.

| Neurological diseases
Exosomes have the advantages of being able to cross the blood-brain barrier, a critical step for brain-targeted therapies. This property has been exploited in diseases such as Parkinson (Haney et al., 2015) or ischemic stroke. Recently, studies have shown that nanoparticles Lv et al., 2018) but also exosomes can be used as a therapeutic agent for the treatment of ischemic stroke where an engineered c(RGDyK)-conjugated exosomes were able to target the lesion of the ischemic brain following intravenous injections (Tian et al., 2018). Reperfusion can be the only option to reverse brain damage following a stroke, but this can induce an inflammatory reaction potentially causing further damage. This is how Dong et al. (2019) suggested the use of neutrophil membrane-derived vesicles loaded with resolvin D2, acting as anti-inflammatory agent and specifically delivered to the brain and in particular to a stroke lesion.

| Infectious diseases
Exosome-based technologies to generate vaccines have been exploited for years (Devhare & Ray, 2017). Exosomes have been explored as a platform for vaccination, by delivering diseaseassociated antigens. This approach was tested by delivering hepatitis C-associated antigens (Desjardins et al., 2009) and extended to the patients.

| EXOSOME ENGINEERING
Exosomes are being widely investigated as immunomodulators or therapeutic cargo vehicles. This intense research field is continuously refining engineering strategies to optimise efficacy and/or delivery. A specific targeting of the recipient cell is paramount to adequately deliver exosome content (Horibe et al., 2018). This is mediated by the  Incorporation of pH-sensitive peptides such as the alphagalactosidase A protein can be used to enhance membrane fusion.
Similarly, exosome membrane while coupled with 3-(diethylamino) propylamine results in membrane disruption in an acidic environment resulting in enhanced cytosolic release of the cargo (Lee, Park, et al., 2019).
Extracellular vesicles including exosomes are being targeted by plasma proteins making them prone to rapid clearance during circulation. Polyethylene glycol (PEG) adjunction to the outer membrane of exosomes can act as a shield to protect vesicles from clearance (Antimisiaris et al., 2018). However, this can hinder interactions between target cells and exosomes. To counteract this, PEG-conjugated antibodies or "nanobodies" against EGFR are added onto the exosomal membrane increasing the circulation time and binding to EGFR overexpressing tumour cells (Kooijmans et al., 2016).

| Optimising cargoes
Based on their natural capacity to transfer cargo such as proteins or nucleic acids, exosomes can be engineered as a therapeutic vehicle.
However, mechanisms for loading such cargoes into exosomes are F I G U R E 2 Engineering strategies to refine a specific function and/or tropism to exosomes. Depending on their parent cell line, exosomes can express intrinsic ligands or can be engineered to express specific targeting ligands, stimuli response peptide, fusion protein, immune-evasive components, or viral glycoproteins. Their cargos can vary from small-sized genetic material such as noncoding RNAs to components as large as AAV. Exosome engineering can enable specific targeting of the central nervous system, systemic organs, or tumours. AAV: adeno-associated vector, HEK: human embryonic kidney, MSC: mesenchymal stem cells, PEG: polyethylene glycol, TRAIL: TNF-related apoptosis-inducing ligand challenging. This is why recent efforts are being developed to maximise exosomal cargo loading and cargo refinement for increased therapeutic benefit. This can be achieved by creating exosome scaffolds like the prostaglandin F2 receptor negative regulator protein and the brain acid soluble protein 1 (Codiak Ltd., 2021)  Similarly, genetic materials such as miRNAs are naturally present in exosomes but at levels unlikely to exert detectable effects (Chevillet et al., 2014). This is why strategies have been developed to overexpress them in exosomes to mediate posttranscriptional regulation in recipient cells. For example, this involves miRNA-126 acting as miRNA mimic to supress the PTEN/PI3K/AKT signalling pathway involved in cancer proliferation and migration (Nie et al., 2020). In another application in cancer, exosomes are used as specific drug vehicle delivering doxorubicin, paclitaxel or curcumin with a better efficacy and reduced off-target toxicity compared to treatment with drug alone (György et al., 2015;Mentkowski et al., 2018). To tackle chemotherapy resistance, exosomes are used for co-delivery of chemotherapeutic drug and chemoresistance inhibitors using miRNA inhibitors to modulate the expression of tumour suppressor genes.
For instance, exosomes loaded with 5-fluorouracil (5-FU) against colorectal cancer and miR-21 inhibitor (miR-21-i) successfully reduced miR-21-mediated 5-FU resistance in 5-FU-resistant colorectal cancer cell line (Liang et al., 2020). Similar technologies can be developed to deliver other genetic materials such as siRNAs which have therapeutic applications in gene silencing or the CRISPR-Cas gene editing systems. This is of particular interest for the treatment of neurological diseases due to the exosome's ability to cross the blood-brain barrier.
The ability of exosomes in protecting their content against degradation is another advantage for some cargoes such as adeno-associated vector-mediated vectors. This system protects adeno-associated vectors against specific capsid humoral immunity in preimmunised patients, an increasing issue in adeno-associated vector gene therapy (Perocheau et al., 2019). A subpopulation of adeno-associated vectors originally from the media of manufacturing cells is physiologically encapsulated into exosomes creating exosomes-associated adeno-associated vectors (exo-adeno-associated vectors) (György & Maguire, 2018). Exosomes-associated adeno-associated vectors enable successful transduction in vivo despite low titres of neutralising antibodies, where wild-type adenoassociated vectors are neutralised by humoral immunity.

| Selecting the specific exosome parent cell
The origin of the exosome-producing cells influence the exosome biodistribution and its therapeutic effect. Tumour-derived exosomes have preferential tropism towards their parent cell type allowing a direct application in their native form to target these tumour cells (Sun et al., 2018).  (Yuan et al., 2017). Exosomes derived from human umbilical cordmesenchymal stem cells improve inflammation in carbon tetrachoide (CCl4)-induced fibrosis in mouse livers (Li et al., 2013). Mesenchymal stem cell-derived exosomes overexpressing GATA4 alleviate sequelae of ischemic heart disease via the anti-apoptotic miRNA, miR19a, which promotes mesenchymal stem cell differentiation in cardiomyocytes, reduces cardiomyocyte apoptosis and enhances angiogenesis (Yu et al., 2015).
As part of their physiological function, exosomes are involved in the regulation of the immune system. Such exosomes derived from macrophages, dendritic cells and natural killer (NK) cells exhibit therapeutic effect by activating T cell responses and displaying anti-tumour effects in vitro (Andaloussi et al., 2013;Quah & O'Neill, 2005).

| MANUFACTURING STRATEGIES AND LIMITATIONS
Exosome are defined by their cells of origin and their engineered properties. However, their efficient and reliable production is controlled by their microenvironment, culture and dissociation systems. Further downstream steps in their manufacturing are also required involving isolation and characterisation. This process is generally divided into three steps:-(i) removing of cells and cell debris, (ii) concentration of condition medium and (iii) purification Gao et al., 2020).
Various types of cells have been used for good manufacturing practice exosome manufacturing such as HEK293 (Watson et al., 2018), mesenchymal stem cells (Pachler et al., 2017), dendritic cells (Lamparski et al., 2002), adipose tissue-derived stem cells (Bari et al., 2019) and human cardiac progenitors cells (Andriolo et al., 2018). Those cell types might require specific culture parameters such as the need for growth factors, oxygen requirements, cell density, cell passage and cell differentiation (Panchalingam et al., 2015;Patel et al., 2018;Sart et al., 2010). The cultivation medium is usually defined into animal-free (Andriolo et al., 2018;Lamparski et al., 2002) or animal-derived (Mendt et al., 2018) components and is based on the type of cells used. However, for ethical issues and risks of contamination, animal-derived components should be avoided. The use of xeno-free conditions in good manufacturing practice production should be favoured as alternatives, xeno-free media has also been shown to increase yield (Andriolo et al., 2018). The careful selection of a clone for further expansion is an option of choice: the clonal cell line will enable a homogeneous production and simplify the downstream characterisation process.
Culture systems include either static systems such as flasks, but scaling up processes will include dynamic systems like bioreactors which offer the advantage of improved controls of parameters such as CO 2 , O 2 and pH. Microcarriers in stirred-tanked bioreactors provide maximum surface area (Sart et al., 2010) and hollow fibre perfusion bioreactors (Yan & Wu, 2020) are also showing promising results.
Hollow fibre bioreactors (Watson et al., 2016) tend to be used for conditioned medium harvest (Mendt et al., 2018;Watson et al., 2018). They also allow a more efficient production of exosomes compared with a static system. Dissociation enzymes if considered should be animal free (Andriolo et al., 2018).
Downstream processing involves differential centrifugation, the most common method for media concentration. Exosome purification methods rely on density centrifugation, precipitation, chromatography, membrane filtration and size exclusion. Each method will influence the amount, type and purity of exosomes (Yang et al., 2019).
Tangential flow filtration is another method recently developed for media concentration and exosome isolation and is proven to be more gentle, efficient and scalable than ultracentrifugation and resulting in higher yield. It combines membrane filtration and a tangential flow across the surface to avoid filter cake formation (Busatto et al., 2018).
As clinical applications of exosomes expand, manufacturing strategies initially developed in academic settings need to be adapted for scaling up (Figure 3) to meet adequate Chemistry, Manufacturing and Controls (CMC) timelines and good manufacturing practice requirements. However, this scale-up process might alter the cell line phenotype and the downstream biological function of exosomes. For instance, this could impact the cellular microenvironment by modifying the cellular physiology, pH, oxygen supply, media composition and supply of growth factors (Chen et al., 2019). Also, the use of primary or stem cell lines adds further complexity to the process (Cheng et al., 2017;Sart et al., 2010) as some cell lines like mesenchymal stem cells have slower growth and altered biology during the scale-up process (Chen et al., 2011). Cell density, apoptotic blebs and impurities from cells undergoing cell death can further impact the exosome functionality (Bollini et al., 2013;Crescitelli et al., 2013;Patel et al., 2018).
Scaling up using bioreactors can also have mechanical consequences.
mesenchymal stem cells, which have accrued potential for exosome production, can modify their cellular phenotype when exposed to shear stress in a bioreactor (Brindley et al., 2011;Panchalingam et al., 2015). Cyclic stretch combined with shear stress inhibits their smooth muscle actin formation and induces a switch of mesenchymal stem cell phenotype towards endothelial cells (Patel et al., 2018). An alternative to bioreactor is to use multiple flasks or stacked multilayer culture flasks, but this may not be a compatible option for industrial scale-up. One popular and cost-effective system is the Integra CELLine™ system, which allows high yield of exosomes compared to flasks without altering the exosome morphology, phenotype and function (Mitchell et al., 2008). The CELLine™ system uses different compartments for cells and media allowing a continuous flow of nutrients while reducing waste products and allowing optimal cell proliferation.
Another limiting factor for scaling up is the need of serum and growth factors in large volumes (Shelke et al., 2014). Sera like fetal bovine serum are rich in endogenous exosomes and contaminants, a major pitfall for good manufacturing practice production. Furthermore, the drug-loading efficacy in exosomes is influenced by the production cell line, the cargo and the loading method and can vary from 1.4% to 38% (Walker et al., 2019).
As extracellular vesicles are heterogeneous by nature, the purification process is critical, which limits the downstream scale-up.
Purification is frequently achieved using operator-dependent ultracentrifugation which produces low yields (Lee, Johansson, et al., 2019). However, this challenge is progressively being overcome as illustrated by the production of good manufacturing practicecompliant mesenchymal stem cells-derived exosomes for clinical trials (Chen et al., 2019). Additional purification steps are often required which include exclusion size chromatography and sucrose or Optiprep™ density gradients. Physical modification with the formation of exosome aggregates is also a significant drawback of ultracentrifugation as few reports suggest the exosome original phenotype and morphology to be compromised causing artefacts or undesired effects (Linares et al., 2015). An adaptation of the rotor size and type (Cvjetkovic et al., 2014) can partly alleviate these detrimental effects.
Combined or as an alternative to ultracentrifugation, filtration and size exclusion chromatography methods are also commonly used for isolation. A promising technique combining bind-elution with size exclusion chromatography shows high yields and better reproducibility. This technology enables elution of large particles like exosomes while smaller impurities like nonvesicular proteins and RNAs remain bound to the column. It is highly reproducible generating up to 80% yields even when used with large volumes of media (Corso et al., 2017) while limiting risks of vesicular disruption and aggregation.
A safety regulatory framework with good manufacturing practice standards for exosome characterisation throughout the manufacturing process is being developed by the International Society for Extracellular Vesicles or the European Network on Macrovesicles and Exosomes in Health and Disease (Lener, Gimona, Aigner, Börger, Buzas, Camussi, Chaput, Chatterjee, Court, Portillo, & O'Driscoll, 2015a). This standardisation is paramount for the clinical translation of exosomes (Ilic et al., 2012;Witwer et al., 2013). Exosome characterisation includes (i) protein quantification using techniques such as bicinchoninic acid (BCA) assay, western blotting or liquid chromatography and mass spectrometry for proteomic analysis (Yang, Guo, et al., 2017), size and concentration assessed by nanoparticle tracking analysis (Dragovic et al., 2011), (ii) morphology by transmission electron microscopy (Chuo et al., 2018) and (iii) exosome surface profiling or cargo characterisation by flow cytometry and liquid chromatography or mass spectrometry (Schey et al., 2015).
Finally, standardising storage methods of pure extracellular vesicle fractions is also a crucial step for translation (Mora et al., 2015) (Lener, Gimona, Aigner, Börger, Buzas, Camussi, Chaput, Chatterjee, Court, Portillo, O'Driscoll, Fais, et al., 2015b). Storage at 4 C or −80 C can impact the biological activity and protein content of exosomes (Maroto et al., 2017). Storage at −80 C is advised as the optimal temperature providing the least impact on exosome morphology and content (Yamashita et al., 2018). Other parameters adding to the complexity of the scale-up process are the storage buffer and its pH, the number of freeze-thaw cycles and their effect on exosomal protein content (Cheng et al., 2019).

| CONCLUSION
Exosomes are poised at crossroads for clinical applications. The use of these small vesicles is rapidly expanding for diagnostic and therapeutic purposes following their relative recent discovery as key players in physiology and pathology. The leading clinical applications in cancer and inflammation are exploiting the exosome role in immunomodulation and its use as a vehicle for specific drug delivery. Numerous academic publications as well as high value investments in pharmaceutic development dedicated to the exosome field demonstrate general and cross-sector enthusiasm for these biotechnologies.
Various strategies to optimise the therapeutic efficacy of exosomes are being developed. The regulatory framework is evolving in order to allow safe and successful clinical trials. Improved scaling up strategies F I G U R E 3 Exosomes manufacturing for clinical use: from the laboratory to the industry. Schematic highlighting current differences between academic and industrial scale up. Exosomes can be derived from a cell bank or from the patients themselves. Laboratory set up predominantly uses a cell flask platform while the scaling up process involves bioreactors. While bioreactors increase yields and purity, exosomes need to be adequately characterised. FACS: fluorescence-activated cell sorting, LC/MS: liquid chromatography-mass spectrometry, MSC: mesenchymal stem cells, ME: microenvironment, NTA: nanoparticle tracking analysis are being developed in order to overcome the limitations of manufacturing and characterisation processes. Hence, while still at its infancy, the exosome field is moving fast to maturity for the benefit of the patients.

| Nomenclature of targets and ligands
Key protein targets and ligands in this article are hyperlinked to corresponding entries in the IUPHAR/BPS Guide to PHARMACOL-OGY http://www.guidetopharmacology.org and are permanently archived in the Concise Guide to PHARMACOLOGY 2019/20