Extracellular Vesicles from Mesenchymal Stromal Cells for the Treatment of Inflammation-Related Conditions

Over the past two decades, mesenchymal stromal cells (MSCs) have demonstrated great potential in the treatment of inflammation-related conditions. Numerous early stage clinical trials have suggested that this treatment strategy has potential to lead to significant improvements in clinical outcomes. While promising, there remain substantial regulatory hurdles, safety concerns, and logistical issues that need to be addressed before cell-based treatments can have widespread clinical impact. These drawbacks, along with research aimed at elucidating the mechanisms by which MSCs exert their therapeutic effects, have inspired the development of extracellular vesicles (EVs) as anti-inflammatory therapeutic agents. The use of MSC-derived EVs for treating inflammation-related conditions has shown therapeutic potential in both in vitro and small animal studies. This review will explore the current research landscape pertaining to the use of MSC-derived EVs as anti-inflammatory and pro-regenerative agents in a range of inflammation-related conditions: osteoarthritis, rheumatoid arthritis, Alzheimer’s disease, cardiovascular disease, and preeclampsia. Along with this, the mechanisms by which MSC-derived EVs exert their beneficial effects on the damaged or degenerative tissues will be reviewed, giving insight into their therapeutic potential. Challenges and future perspectives on the use of MSC-derived EVs for the treatment of inflammation-related conditions will be discussed.


Introduction
Inflammation is a crucial mechanism initiated by the body as a first line of defence against harmful stimuli such as pathogens, tissue damage, radiation, and toxic compounds [1]. An acute inflammatory response is normally triggered by immune cells sensing a pathogen or endogenous stress signal, resulting in the release of pro-inflammatory cytokines and chemokines. This reaction can have a multitude of effects, including neutrophil and macrophage activation, vasodilation, hypotension, induction of capillary leakage, and platelet activation [1,2]. These effects typically facilitate tissue regeneration or the clearance of infection, ultimately leading to the removal of the initial harmful stimuli. Once cleared of harmful stimuli, the multifaceted process of inflammation resolution can begin, which involves substantial reprogramming of cells to the anti-inflammatory phenotype [2]. Unfortunately, acute inflammation can often progress into chronic nonresolving inflammation, which may elicit more harm to the body than the initial stimuli that triggered the inflammatory response [3]. Though not the primary cause, non-resolving chronic inflammation has been identified to play an important role in the pathogenesis of a myriad of debilitating diseases including rheumatoid arthritis [2,4], atherosclerosis [2,5], Alzheimer's disease [6], various cancers [2,[7][8][9], asthma [10], type 2 diabetes [11], diabetic nephropathy [12], osteoarthritis [13][14][15], multiple sclerosis [16], depression [17], chronic rhinosinusitis [18], idiopathic pulmonary fibrosis [19], and atrial fibrillation [20]. These diseases share many common pathophysiological mechanisms, including the activation of inflammatory cells, release of soluble inflammatory factors (most notably cytokines and chemokines), and extracellular matrix (ECM) remodelling [21].
With such a long list of conditions in which non-resolving inflammation plays a key role, there is no doubt that it imposes an immense burden on society. Unfortunately, commonly used anti-inflammatory treatments such as non-steroidal anti-inflammatory drugs (NSAIDs) and glucocorticoids appear to merely relieve symptoms of the underlying disease, and there is little evidence to demonstrate that these treatments have any effectiveness in ceasing disease progression [22]. As such, there is an urgent need to develop new therapeutic strategies, which perhaps can act on multiple pathways of disease progression rather than only targeting the inflammatory characteristics.
Mesenchymal stromal cells (MSCs), previously commonly referred to as mesenchymal stem cells [23], are the most widely explored cell type for cell-based therapeutics, and their use in clinical trials to treat a wide range of diseases has increased dramatically over the past two decades [24]. The literature provides ample evidence of studies showing the beneficial effects of MSCs when applied for treating inflammatory diseases in animal models [25,26], with evidence in multiple tissue types including cardiovascular (myocardial infarction, vascular disease, peripheral artery disease, preeclampsia); neural (multiple sclerosis, Parkinson's disease, Alzheimer's disease); and osteochondral (rheumatoid arthritis, osteoarthritis) [26][27][28][29]. As such, there is an ongoing urge within the scientific community to translate these promising findings to humans. It was initially believed that the therapeutic potential of MSCs was a function of injected MSCs engrafting to existing cellular structures, and subsequently differentiating and facilitating the formation of neo-tissue [30]. However, this belief has been subverted in recent years. It has been widely observed that implanted MSCs show very low levels of engraftment (less than 3%) in the target tissue [31], with the vast majority of the population of implanted cells being rapidly cleared [32]. For this reason, other mechanisms have been investigated, and it is now evident that the regenerative, anti-inflammatory, and immunomodulatory capacity of MSCs is exerted through their secretion of paracrine factors [33][34][35].
The MSC secretome accounts for all molecules secreted by the cell. It includes a variety of chemokines, cytokines, immunomodulatory factors, and ECM components, along with a range of other proteins, nucleic acids, and lipids [32]. It is suggested that once MSCs are implanted into damaged or diseased tissue, they secrete a host of anti-inflammatory and regenerative factors that elicit a therapeutic response. Importantly, the secretion profile appears to be a function of the microenvironment around the secreting cell, for instance, MSCs exposed to inflammatory signals can elicit an enhanced secretory profile [36]. However, the majority of investigations surrounding this observation have been in vitro gene expression or proteomic studies and require further in vivo validation [32].
It has been suggested that the apoptosis or phagocytosis of implanted MSCs act as the trigger for the observed immunomodulatory effects elicited by MSCs [32]. There are so far two key observations supporting this mechanism. First, observations in mouse models of graft-versus-host disease have demonstrated that, for MSCs to exert their immunosuppressive effects, they must first undergo natural killer cell/T-cell induced apoptosis [37]. Second, observations in a mouse model showed that injected populations of MSCs were rapidly cleared through monocytic phagocytosis. The monocytes that phagocytosed the MSCs were shown to modulate their phenotype, which changed the course of the immune response [38]. These two observations provide a potential hypothesis for the mechanisms of MSC-mediated immunomodulation, though further studies are required to confirm the details.
Aside from the above two proposed mechanisms underlying the therapeutic effects of MSCs, a third mechanism has gained increasing attention in recent years: extracellular vesicles (EVs) derived from MSCs. This will be the topic of focus in this review that will be discussed in the context of treating inflammation-related conditions.

The Fundamentals of Extracellular Vesicles
EV is an umbrella definition which encompasses all vesicles released or 'shed' by cells [39]. Typically, EVs have a diameter in the range of 30-2000 nm. They consist of a lipid bi-layer membrane encasing an organelle-free cytosole, which contains a combination of various proteins, lipids, and nucleic acids [40,41]. EVs have been recently discovered as a key mechanism of the intercellular communication network. Since EV release was first observed in rat and sheep reticulocytes in the early 1980s [42,43], an ever-growing number of cells have been shown to release EVs as a form of intercellular communication. Almost all mammalian cell types have demonstrated EV secretion including stem cells, neuronal cells, immune cells, and cancer cells [39,44]. EVs have also been isolated from an extensive range of biological fluids including blood, urine, semen, breast milk, cerebrospinal fluid, bile, amniotic fluid, and ascites fluid [44]. Interestingly, EV secretion has been observed in lower eukaryotes and prokaryotes, with speculations that microbial EVs may mediate the host response to infection [44,45].
The exact classification of EVs is still evolving, and the current definition of nomenclature is not consistently used in the literature [46,47]. Presently, EV classifications are based on their size and biogenesis [46], with three widely accepted distinct populations. Exosomes are the most widely studied subpopulation of EVs [48]. Although the size range of exosomes has not been consolidated in the literature, it is generally accepted that they have a diameter in the range of 20-150 nm. The biogenesis of exosomes begins with endocytosis, a process of invagination of the plasma membrane to form an endosome. Within the endocytic pathway, endosomes are classified into three sub-populations: early endosomes, late endosomes, and recycling endosomes [49]. Early endosomes which are not destined for secretion, recycling, or degradation become late endosomes. Late endosomal membrane invagination subsequently forms intraluminal vesicles (ILVs) which contain proteins, lipids, and nucleic acids. At this point, the late endosome now containing a host of small vesicles is deemed a multivesicular body (MVB) [40]. The MVB has two possible routes, either fusing with the lysosome where its contents will be recycled or fusing with the plasma membrane. The latter releases the ILVs into the extracellular space, where they are now referred to as exosomes. This process is visualised in Figure 1. The formation of ILVs is believed to be mainly regulated by two processes. First, the endosomal membrane is enriched for tetraspanins, specifically CD9 and CD63 [50]. Second, the endosomal sorting complexes required for transport (ESCRTs) are present during the process of ILV formation. These two processes regulate the initial inward membrane budding of the late endosome, ILV cargo sorting, and subsequent ILV formation. Although it is generally accepted that the ESCRT pathway is the main mechanism governing exosome formation, there exist supplementary mechanisms of ILV formation such as the syndecan-syntenin-ALIX pathway [40]. Since exosomes arise from endosome membrane invagination, they present common proteins associated with this process across all cell types. These proteins include flotillins, GTPases and annexins (membrane transport and fusion); integrins (adhesion); ALIX and the tetraspanins CD9, CD63, CD81, CD82 (MVB formation); and major histocompatibility complex (MHC) molecules (antigen presentation) [51]. Typically, the lipid composition of exosomes mirrors that of their parent cell. Exosomes are commonly enriched with cholesterol, phosphatidylserine, ceramide, and sphingomyelin [52]. Interestingly, the concentration of diacyl-glycerol and phosphatidyl-choline appear to be lower in exosomes than their parent cells [53]. The nucleic acid content of exosomes typically consists of mRNAs, microRNAs (miRNAs), and other non-coding RNAs [51], although genomic and mitochondrial DNA have also been found in exosomes [54,55]. Figure 1. Extracellular vesicle (EV) biogenesis, secretion, and uptake [56]. Exosomes (20-150 nm) are intraluminal vesicles (ILVs) formed by inward budding of the endosomal membrane during maturation of multivesicular body (MVB), which are secreted upon fusion of the MVBs with the plasma membrane. Microvesicles (50-1000 nm) are a heterogeneous group of vesicles with different membranes depending on their origin and morphology. Apoptotic bodies are shedding vesicles derived from apoptotic cells. After their release into the extracellular space, EVs can bind to cell surface receptors to initiate intracellular signalling pathways. EVs can also be internalised through processes such as macropinocytosis and phagocytosis, or by fusion with the plasma membrane. The cargo of EVs consisting of proteins, nucleic acids and lipids are released in the intracellular space or taken up by the endosomal system of the recipient cell. Reproduced with permission from [56].
The second most widely studied subpopulation of EVs are microvesicles (MVs) [48]. It is generally accepted that MVs have a diameter in the range of 50-1000 nm, meaning that they may have a size overlap with exosomes. This creates challenges for purely size-based EV isolation techniques in distinguishing between exosome and MV populations [39]. In contrast to exosomes, MVs are formed through direct shedding from the plasma membrane of the parent cell. The formation of MVs is regulated by aminophospholipid translocases, which control the phospholipid re-distribution in the plasma membrane and the dynamics of cytoskeletal actin-myosin contractions [57]. As MVs form through direct outward budding of the plasma membrane (Figure 1), they share many of the same membrane markers as their parent cell, which may include integrins, selectins, and CD40 ligand [58]. The variations in membrane markers among MVs is a result of the induced changes which occur during the process of nucleation and budding [51]. The cargo carried by MVs, like exosomes, is not simply representative of the cytoplasmic content. Some loading mechanisms such as ARF6 trafficking of proteins and CSE1L nucleic acid export have been identified [59,60]. However, the exact mechanisms of regulation remain incompletely understood and constitute an area of active research. The protein and nucleic acid content of MVs are dependent on the cell type along with the external physiological conditions experienced by the parent cell [40]. A number of proteins are commonly identified in MVs, such as matrix metalloproteinases (MMPs), cytoskeletal components, and glycoproteins [51]. Like exosomes, MVs generally contain a combination of mRNAs, miRNAs, and other non-coding RNAs [51], as well as possible genomic and mitochondrial DNA [54,55].
Apoptotic bodies are the final widely recognised subpopulation of EVs. They are by far the largest in size, ranging 500-5000 nm in diameter, and are produced by outward membrane blebbing on the surface of cells undergoing apoptosis [58,61]. There is no evidence that apoptotic bodies play a role in intercellular communication or have a potential therapeutic effect, although they do show potential to be used as disease biomarkers [62].
EV-cell communication can occur through several distinct pathways: lysis of EVs in the extracellular space releasing their contents, direct EV-cell binding, membrane fusion and release of EV contents, and EV uptake into the endocytic system [56,63]. Ligandreceptor binding associated with EV extracellular content release and direct EV binding are believed to be the mechanisms behind several of the biological effects exerted by EVs on cells, such as growth and angiogenic factor delivery [63]. For the nucleic acids or proteins suspended in the EV cytosol to act as messengers in the recipient cell, the EVs must fuse either with the plasma membrane after ligand-receptor binding, or with the endosomal membrane after endocytosis [63]. Endocytosis of EVs is thought to be the most common route of uptake [40,41,63], although several questions remain to be answered about this uptake route. Since the endocytic pathway inevitably ends with degradation or expulsion from the cell, the cargo carried by the EVs must exit the endosome somehow and find its way into the cytoplasm if it is to alter cell composition and function [40]. Although this phenomenon of endosomal escape has been widely observed, the underlying mechanisms are still unclear [40,64,65]. EV-cell communication is known to be involved in an extensive range of biological processes, including modulation of the immune system [66,67], neurobiological functions such as synaptic plasticity [68], and stem cell differentiation [69,70].
With the extensive role that EVs play in biological processes, it is unsurprising that they are also heavily involved in the pathogenesis of disease. The most in-depth understanding of this concept is in tumour biology [71]. EVs have been shown to have important roles in promoting tumour cell proliferation [72,73], angiogenesis [73,74], ECM remodelling [75], and metastasis [58,75]. Although beyond the scope of this review, there is a great potential in targeting the phenotype altering mechanisms exerted by EVs in tumour biology to help develop new treatment strategies, as well as to apply stem cell-derived EVs as cancer therapeutics [76]. In the field of regenerative medicine, EVs derived from stem cells are shown to replicate the therapeutic properties of the parent cells, and have demonstrated many beneficial effects such as apoptosis suppression [77], promotion of cellular proliferation [78] and angiogenesis [79], and the ability to modulate the diseased cell phenotype to facilitate tissue regeneration [80]. The precise cargo carried by EVs and the mechanisms which facilitate their regenerative potential are still unclear. However, it is known that EV composition is a function of its cellular origin and physiological conditions [81]. By varying factors such as cellular stress, media composition, and physical stimulation, or by enriching certain miRNAs in the parent cells, it may be possible to optimise the EV composition for specific regenerative applications [82][83][84][85].
Over the past decade, MSC-derived EVs have been increasingly explored in regenerative medicine to treat disease or promote repair through local delivery in a range of tissue types, including cardiovascular, musculoskeletal, neural, renal, hepatic, lung, dermal, and reproductive tissues [56,86]. It is thought that the MSC-derived EVs can deliver the same anti-inflammatory and trophic effects as the parent cells [87]. Compared to injecting live cells into tissues, MSC-derived EVs bypass potential safety concerns of the MSCs exhibiting uncontrollable behaviour or differentiating into problematic tissue at the site of injection [88]. The EVs also have an additional advantage of presenting minimal toxicity and immunogenicity, even when applied xenogenetically as a large dose at high frequency [89]. The rest of this review will summarise the current state of research into MSC-derived EVs as therapeutic agents for treating a number of inflammation-related conditions: osteoarthritis, rheumatoid arthritis, Alzheimer's disease, cardiovascular disease, and preeclampsia ( Figure 2). For each of these conditions, evidence related to the therapeutic effects of MSC-derived EVs has been collected from a range of experimental studies published within the last ten years, as shown in Table 1. These conditions represent examples of diseases with significant societal impact, where pathogenesis is closely linked with inflammation in musculoskeletal, neural, and cardiovascular tissues as three major body systems. MSC-derived EVs have also demonstrated beneficial effects in other conditions and body systems impacted by inflammation, such as graft-versus-host disease [90], kidney disease [91], liver failure [92], and skin wounds [93], although a detailed discussion of these is beyond the scope of this review.

Extracellular Vesicles from Mesenchymal Stem Cells for the Treatment of Osteoarthritis
Osteoarthritis (OA) is the most prevalent joint disease globally, affecting 18% of women and 10% of men over the age of 60 [94]. While OA is generally characterised by the degeneration of articular cartilage, it is a disease affecting the entire joint including the subchondral bone and synovium [95]. Although not the primary defining feature of OA, chronic inflammation forms an important part of the catabolic environment that induces the irreversible progression of joint degeneration [96]. The exact pathogenesis of OA is incompletely understood, but it is generally accepted that antagonistic biomechanics acting on a vulnerable joint is intrinsically linked to disease progression [94]. A number of risk factors including age, obesity, abnormal joint morphology, and prior joint injury are strongly associated with the development of OA. Once the disease progresses, the regeneration of damaged joint structures is unlikely. Non-surgical treatments such as anti-inflammatory medication and intra-articular injections of corticosteroids or hyaluronic acid may help to relieve pain, although these have shown little to no benefit in slowing disease progression [97]. With a lack of viable treatment options, the final destination for most patients after all options have been exhausted is total joint replacement, which surgically removes the diseased joint. Although this can lead to significant pain reduction and overall improved quality of life, the level of activity post-surgery is relatively low compared to pre-replacement levels due to component failure or loosening [98], and the implant may need revision in younger patients due to having a limited lifetime of approximately 20 years [99].
The use of MSCs to treat OA has shown promise over the last decade, with numerous early clinical studies suggesting that this approach is safe and effective, and may lead to significant improvements in clinical outcomes along with some preservation or regeneration of damaged joint tissues [100,101]. However, there remain significant hurdles before MSCs can be scaled up for widespread clinical use, due to limited cell survival following injection, inability to be used as an 'off-the-shelf' therapy, and regulatory issues associated with the injection of live cells [102]. The use of MSC-derived EVs circumvents these issues, and have shown promising preliminary outcomes in both in vitro and small animal models of OA, as described below and in recent reviews on this topic [87,103].
The 16 studies on OA described in Table 1 demonstrate relatively consistent therapeutic effects of MSC-derived EVs. The MSCs used to generate EVs were derived from many different sources, including bone marrow [83,[104][105][106][107], adipose tissue [82,85,108,109], synovial membrane [84,110], embryonic stem cells [111][112][113], and induced pluripotent stem cells [110,114]. The majority of studies used exosomes [83][84][85]105,106,[110][111][112][113][114][115], while others used MVs [107] or a heterogenous population of EVs that likely contained both exosomes and MVs [82,108,109]. One study also compared the effects of exosomes and MVs [104]. There was not a consensus among the studies on the methods of identifying EV populations, with some purely based on size and others based on size and protein markers, although the size and exact protein markers were also not consistent. The terminology used to refer to EV populations varied among studies, with MVs and microparticles being used interchangeably.
The in vitro experiments conducted using MSC-derived EVs in OA cell models showed that the EVs were quickly internalised by the treated cells, usually within 30 min [82,84,106,107]. It was also widely observed that EVs improved the migration and proliferation potential of the treated cells [83,84,110,[113][114][115], together with increased viability and reduced rate of apoptosis, and that the improvements were dose-dependent [85,104,107,[113][114][115]. EV treatment of OA chondrocytes and fibroblasts commonly resulted in the upregulation of anabolic proteins such as aggrecan, and collagen types I and II [85,104,106,109,111]. This was accompanied by the downregulation of catabolic markers such as MMP-13 and ADAMTS5 [83,85,109,115]. EV treatment also showed anti-immunomodulatory and antiinflammatory effects, through the suppression of COX-2, IL-1α, IL-1β, IL-6, IL-8, IL-17, and TNF-α [106,109], as well as the inhibition of macrophage activation [113].
Similarly, in surgically induced OA models, all studies showed that EVs could attenuate OA progression [84,85,109,111], together with downregulation of catabolic markers (MMP-13, ADAMTS5) and upregulation of anabolic markers (collagen type II) [85,109,111]. The enrichment of EVs with miR-140-5p significantly improved the level of osteochondral protection from OA progression [84], and gait abnormalities in the DMM model were found to be alleviated following exosomes treatment [85].
In studies using an osteochondral defect model, EV treatment resulted in complete neo-tissue infilling and the development of hyaline-like cartilage that was integrated with the surrounding tissue [107,[112][113][114]. An increase in PCNA-presenting cells and decrease in CCP3 apoptotic cells was also observed, along with the switching of macrophage phenotype from M1 (pro inflammatory) to M2 (anti-inflammatory) and suppression of inflammatory cytokines [113]. In the study that used a cardiotoxin-induced muscular injury model, EVs were shown to accelerate muscular regeneration, and have an antiinflammatory function that increased M2 anti-inflammatory markers and reduced M1 inflammatory markers [82].

Extracellular Vesicles from Mesenchymal Stem Cells for the Treatment of Rheumatoid Arthritis
Rheumatoid arthritis (RA) is the second most common form of arthritis, after OA. There is no cure for RA, with the only treatments being physical therapy and medication to help relieve symptoms and slow disease progression. RA is an autoimmune disease that is primarily defined by chronic joint inflammation, together with bone erosion and ECM destruction [116]. The pathogenesis of RA is driven by pre-existing genetic disposition coupled with risk factors that increase the likelihood of disease progression, such as smoking, silica dust exposure, vitamin D deficiency, and obesity [117]. The transition to a chronically inflamed synovium in RA is incompletely understood, although it may be triggered by a number of antagonistic stimuli such as local microtrauma, microvascular injury, and complement activation. Additionally, autoantibodies can activate periarticular osteoclasts, a step which initiates bone damage and is associated with the release of TNF-α and IL-8, both of which promote synovitis [118]. Once synovitis is established, the synovial ECM is disrupted, and there is activation of stromal cells together with a myriad of infiltrating cells including T-cells, B-cells, macrophages, mast cells, and plasma cells. Disease progression involves a complex and incompletely understood relationship between cells and soluble immune factors, most notably chemokines and cytokines [4].
Driven by the well-known anti-inflammatory and immunomodulatory functions of MSCs, a handful of studies have explored the use of MSC-derived EVs as a therapeutic for RA. In contrast to the number of publications available for OA, there were only five studies that applied MSC-derived EVs to experimental models of RA or inflammatory arthritis, as shown in Table 1. Of these, four studies sourced MSCs from the bone marrow [119][120][121][122], and one from RA synovial fluid and human neutrophils [123]. All studies applied exosomes as the EV subpopulation of interest, including one that used both exosomes and MVs [123]. The exosomes used in most studies were characterised to be approximately 100 nm in diameter [119,121,122], while the diameter of MVs was within the range of 150-575 nm [120].
For the in vitro experiments, two studies implemented a RA cell model comprising fibroblast-like synoviocytes (FLS) isolated from the diseased tissue of RA patients [119,121]. The invasion of RA FLS into cartilage and angiogenesis are important processes mediating the pathogenesis of RA. Both studies explored the enrichment of exosomes with a specific miRNA, namely miR-150-5p [119] or miR-124a [121]. Both studies showed that the miRNA-enriched exosomes could inhibit the increased migration and invasion associated with RA FLS. In one study, a co-culture of RA FLS with human umbilical vein endothelial cells (HUVECs) that was treated with miR-150-5p-enriched exosomes substantially inhibited tube formation, compared to exosomes that were enriched with a control miRNA (miR-67) [119], suggesting that the former could suppress angiogenesis. When treated with inflammatory cytokines, RA FLS were shown to upregulate MMP-14 and vascular endothelial growth factor (VEGF), but this upregulation was attenuated when the cells were treated with miR-150-5p-enriched exosomes [119]. These findings led to the hypothe-sis that miR-150-5p delivered by exosomes could be one of the mechanisms responsible for suppressing cell migration, invasion, and tube formation in RA. In the other study, the effects of miR-124a-enriched exosomes on RA FLS were compared with unaltered exosomes by evaluating proliferation, cell cycle progression, apoptosis, and 'wound' closure (modelled by a wound scratch assay) [121]. The miR-124a-enriched exosomes were found to induce a more pronounced suppression of proliferation, arrest of the cell cycle in the G0/G1 phase rather than the S phase, stronger inhibition of the wound closure healing rate, and similar levels of apoptosis in RA FLS compared to unaltered exosomes [121]. One other study involving an in vitro investigation explored the immunomodulatory effects of exosomes and MVs derived from both IFN-γ-primed and un-primed murine bone marrow MSCs on murine T-and B-lymphocytes [120]. Both primed and un-primed MSC culture media were found to suppress the proliferation of T-lymphocytes. After centrifugation, the culture medium supernatant lost its ability to suppress proliferation, suggesting that the immunomodulatory agents secreted by MSCs were present in the EV-containing pellet. The EVs were shown to have dose-dependent effects on suppressing T-lymphocyte proliferation, but these suppressive effects were lost when the EVs were subjected to a freeze-thaw cycle. Both exosomes and MVs were shown to have the ability to suppress several types of pro-inflammatory cells (CD8+ IFN-γ+ and CD4+ IFN-γ+ cells) and increase the number of anti-inflammatory cells (CD4+ IL-10+ Tr1 and CD4+ CD25+ Treg).
Three studies performed in vivo experiments using a collagen-induced arthritis (CIA) model in mice or rats. Two explored the enrichment of exosomes with specific miRNAs, miR-150-5p [119] or miR-192-5p [122]. The third used exosomes and MVs derived from IFNγ-primed and un-primed murine bone marrow MSCs [120]. Treating CIA mice with miR-150-5p-enriched exosomes resulted in reduced levels of inflammation, as measured by hind paw thickness and reduced clinical arthritis scores compared to PBS and exosomes enriched with a control miRNA (miR-67) [119]. Mice treated with miR-150-5p-enriched exosomes also showed downregulation of VEGF (angiogenic factor), CD31 (angiogenesis marker), and MMP-14 (promotes the characteristic invasion of FLS into cartilage). When miR-192-5p-enriched exosomes were injected into CIA rats, they were found to migrate from the bloodstream into synovial tissue, where they downregulated the expression of a number of factors that normally show upregulation in the CIA model (RAC2, TRAP, PGE2, IL-1β, TNFα, NO, and iNOS) and upregulated miR-192-5p expression [122]. To identify an optimal dose of EVs for the CIA model, a study comparing exosomes and MVs first employed a delayed type hypersensitivity (DTH) model to measure inflammation through paw swelling [120]. Both exosomes and MVs exerted a dose-dependent anti-inflammatory effect and were the most effective at the maximum dose of 250 ng. Carrying this dose into the CIA model, the exosomes-treated group developed arthritis at a rate of 5% with extremely low clinical arthritis scores, compared to the MVs-treated group at 20% and higher clinical arthritis scores. Both EV groups performed significantly better than the PBS-treated group, which developed arthritis at a rate of 47% with relatively severe clinical arthritis scores. In a separate experiment using the CIA model, a 250-ng dose of exosomes was shown to have a superior anti-inflammatory effect compared to 500 ng of MVs, as assessed through paw swelling, lower rates of arthritis development, and lower clinical arthritis scores [120]. In addition, one study used both K/BxN-and glucose-6-phosphate isomerase (GPI)-induced inflammatory arthritis models [123]. Treatment with a neutrophil-derived EV population that likely contained both exosomes and MVs was found to reduce cartilage degradation and proteoglycan loss, and this chondroprotective effect was shown to require AnxA1. The interactions of AnxA1 in MVs with the FPR2/ALX receptor was suggested to increase TGF-β production by chondrocytes.

Extracellular Vesicles from Mesenchymal Stem Cells for the Treatment of Alzheimer's Disease
Alzheimer's disease (AD) is the most common form of dementia, characterised by a progressive and irreversible degeneration of the central nervous system (CNS). The pathological hallmarks of AD include extracellular amyloid-β plaques, neurofibrillary tangles, and neuronal dysfunction and degeneration [124]. Chronic neuroinflammation is now broadly understood to play an important role in the pathogenesis of AD, which involves activated microglia and the release of numerous cytokines to produce a sustained immune response [125]. Due to its complex pathophysiology, there are currently no effective treatments for AD. The only approved drugs used for AD treatment provide only symptomatic relief rather than slowing the progression of disease.
Due to their anti-inflammatory, pro-regenerative, and immunomodulatory characteristics, MSCs have been recently considered as a possible treatment for AD [126]. However, despite the promising trophic functions of these cells demonstrated in preclinical models of AD, there are concerns surrounding their intracranial transplantation in a clinical setting, which is associated with increased risks of complications and mortality [126]. In this context, the use of MSC-derived EVs may provide a breakthrough strategy for the treatment of AD, that can convey the therapeutic effects of the parent cells while exhibiting the advantages of small size, low immunogenicity, and lack of ability to undergo malignant transformation.
As shown in Table 1, the 11 studies on AD demonstrated a range of therapeutic effects elicited by MSC-derived EVs in both in vitro and in vivo experimental models. The MSCs were derived from various sources, including bone marrow [127][128][129][130][131], adipose tissue [132], umbilical cord [133,134], Wharton's Jelly [135], and commercially obtained MSCs with an unspecified origin [136]. One study also used neural stem cells (NSCs) isolated from the hippocampus of mice as the cell source [137]. Half of the studies used a heterogeneous population of EVs [128][129][130]132,135], while the other half used exosomes [127,131,133,134,136,137]. Different size ranges and protein markers were used to identify EV populations in these studies.
Six of the studies involved in vitro experiments, four of which used amyloid-β oligomers (AβO) to induce disease, since this has been identified as the major toxic species driving neuroinflammation in AD. Two studies used hippocampal cells [128,135], one used primary cortex neurons [132], and one used microglia [133] as AβO-induced AD models. The other two studies used an AD model comprising microglia exposed to inflammatory cytokines (TNF-α and IFN-γ) [130], or neuroblastoma cells transfected with amyloid precursor protein (APP) gene to overexpress Aβ peptides [134]. In the two studies that used hippocampal cells, one used neurons exclusively [128], while the other used a mixture of hippocampal cells including neurons and astrocytes, as well as a neuron-only oxidative stress model [135]. The hippocampal cells treated with AβO showed a much higher level of EV uptake than control cells, although it was also noted that the uptake of EVs was primarily performed by astrocytes rather than the neuronal cells [135]. The EVs were shown to have an anti-oxidant effect, as they protected against synapse damage [128] and prevented the formation of reactive oxygen species (ROS) in hippocampal cells induced by AβO [135]. However, once the catalase (a known anti-oxidant) content of EVs was inactivated, these protective effects of EVs were lost [128,135]. In the two studies that used microglia, one showed that, after AβO treatment, the application of exosomes increased the markers for alternative activation into the anti-inflammatory M2 microglia phenotype, as well as induced higher levels of anti-inflammatory cytokines (IL-10 and TGF-β) and lower levels of pro-inflammatory cytokines (IL-1β and TNF-α) [133]. The other study showed that, after microglial pre-treatment with inflammatory cytokines, the application of EVs significantly reduced the production of the pro-inflammatory cytokines IL-6 and IL-1β and increased the production of the anti-inflammatory cytokine IL-10 [130]. In the remaining AD cell models, the EVs were shown to reverse neuronal toxicity [132] and reduce the levels of secreted and intracellular Aβ [134].
From the nine studies that performed in vivo investigations, eight used transgenic mice pre-determined to develop an early onset of AD [127,[129][130][131][132][133][134]137], and one used an amyloid-β aggregate-induced AD mouse model [136]. Treatment with EVs led to a range of beneficial effects, including reduced plaque deposition in both the cortex and hippocampus [127,129,131]; reduced levels of soluble and insoluble amyloid-β aggregates [127,129,[132][133][134]; improved memory and spatial learning abilities as assessed by the Morris water maze (MWM) test [127,131,133,134,136,137]; improved memory as assessed by the Novel Object Recognition (NOR) test [132,136]; reduced levels of pro-inflammatory markers (including IL-6, IL-β, IL-α, IL-1β, and TNF-α) [127,133,137]; increased levels of anti-inflammatory markers (including IL-13, IL-10, IL-4, and TGF-β) [127,133]; increased levels of the AβO degrading enzymes neprilysin (NEP) [131,133] and insulindegrading enzyme (IDE) [133]; reduced numbers of pro-inflammatory M1 microglia [130,133]; and increased numbers of anti-inflammatory M2 microglia [133]. One study explored the conjugation of EVs with the rabies virus glycoprotein (RVG) to improve targeting of the nerve tissue [127]. Compared to unconjugated EVs, conjugated EVs showed increased migration from the bloodstream into the cortex and hippocampus. This resulted in lower plaque deposition in these brain regions, as well as greater reduction in AβO aggregates, increased levels of anti-inflammatory markers, decreased levels of pro-inflammatory markers, and improved spatial learning and memory ability as assessed by MWM. Another interesting study cultured MSCs on graphene 2D films or 3D scaffolds and compared the effects of the resulting exosomes [134]. The 3D exosomes were found to be much more effective at improving spatial learning and memory, reducing Aβ deposition, and reducing neuroinflammation and oxidative stress in AD mice compared to 2D exosomes. These differences were thought to be related to the topographical structures of the 3D scaffold, which might have led to increased secretion of protective factors by the parent MSCs that were then captured in the exosomes.

Extracellular Vesicles from Mesenchymal Stem Cells for the Treatment of Cardio Vascular Disease
Acute myocardial infarction (AMI) is the leading cause of death in people with cardiovascular disease worldwide. AMI is characterised by insufficient blood supply to the heart leading to cardiac ischemia, decrease in functioning cardiomyocytes, and death of myocardial tissue [29,138]. The irreversible progression of cardiomyocyte and myocardial tissue necrosis negatively impact cardiac function and can result in congestive heart failure [29,138]. Following AMI, inflammatory cells including macrophages, monocytes, and neutrophils migrate to the infarcted myocardium to induce repair. However, the resulting inflammatory response may persist for longer than necessary and lead to further cardiac damage [139]. MSCs can potentially protect the myocardium following AMI by suppressing persistent inflammation, stimulating angiogenesis and the differentiation of fibroblasts within the infarcted region, abrogating apoptosis, and alleviating fibrosis, hence repairing the myocardium and likely preventing further cardiac dysfunction or heart failure [138].
Due to the anti-inflammatory, pro-angiogenic and anti-oxidant effects of MSCs, MSCderived EVs have been recently investigated for their reparative role in cardiovascular disease [140,141]. The immunomodulatory role of MSC-derived exosomes was shown in an in vitro study that used MSCs from healthy human donor bone marrow [140]. The exosomes exhibited prominent anti-inflammatory properties on peripheral blood mononuclear cells (PBMCs), where they attenuated the pro-inflammatory factor TNF-α and elevated the anti-inflammatory factor TGF-β during in vitro culture. At the later stages of tissue repair, TGF-β is also pro-fibrotic with a well-established role in cardiac fibrosis, cardiomyocyte apoptosis, and cardiac hypertrophy through activin receptor-like kinase activity (ALK) [142]. The physiological process of cardiac repair post AMI consists of three major stages: inflammatory, proliferative, and maturation. Fibroblasts are the key 'reparative' cardiac cells during the inflammatory stage, which are capable of degrading ECM components and releasing inflammatory mediators (IL-1β, TNF-α) to enhance the inflammatory response during cardiac rupture [138]. In contrast, myofibroblasts exhibit an anti-inflammatory phenotype important for the maturation stage of cardiac repair by secreting ECM components and anti-inflammatory factors (TGF-β), and enhancing the differentiation of fibroblasts into myofibroblasts [138].
As seen in Table 1, the 14 included studies all showed therapeutic effects of MSCderived EVs in various small animal models of cardiovascular disease. Exosomes from several MSC sources including bone marrow, adipose tissue, and umbilical cord were all shown to have effects in increasing angiogenesis and cardiac function in a left anterior descending (LAD) coronary artery ligation myocardial infarction (MI) rat model [29]. Related to the process of cardiac repair following AMI as described above, a study showed that exosomes derived from human umbilical cord MSCs can attenuate the inflammation induced by cardiac fibroblasts in vitro, and also have cardioprotective effects in vivo in a lipopolysaccharide (LPS)-induced rat model of MI [138]. These findings were supported by a similar study where PBS, MSCs, and MSC-derived exosomes were injected into the peri-infarct zone of rat infarcted myocardium, where the treatment group receiving exosomes showed the most prominent decrease in the infiltration of CD68+ inflammatory cells [141]. The MSC-derived exosomes were shown to exert their beneficial effects through different mechanisms compared to the parent MSCs. In an LAD coronary artery ligation MI rat model, MSCs and MSC-derived exosomes showed notable distinctions in the expression of several miRNAs [141]. For example, the expression levels of miR-21 and miR-15 were significantly lower in exosomes compared to in MSCs. The downregulation of these miRNAs in exosomes might act to suppress hypertrophy and reduce cardiac ischaemic injury, which might explain their greater therapeutic benefits compared to MSCs. Other mechanisms by which MSC-derived exosomes may reduce cardiomyocyte apoptosis in MI is by regulating cell autophagy through AMPK/mTOR and Akt/mTOR signalling [143].
Dilated cardiomyopathy (DCM) and inflammatory cardiomyopathy (ICM) are some of the most multifaceted complications of myocardial inflammation, characterised by ventricular enlargement and subsequent systolic dysfunction [144]. In a doxorubicin-induced model of dilated cardiomyopathy, cardiac dilation was ameliorated by MSC-derived exosomes, which also showed anti-inflammatory effects that decreased the expression of IL-1, IL-6, and TNF-α cytokines. In addition, these exosomes suppressed the pro-inflammatory macrophages (ILY6C high and M1-like F4/80+ CD11c+) and elevated anti-inflammatory macrophages (LY6C low and M2-like F4/80+ CD206+) in the blood and heart of the treated mice, which restored macrophage polarisation through activation of the JAK2-STAT6 pathway [144]. Similar effects were demonstrated in a clodronate liposome-induced myocardial ischaemia-reperfusion (I/R) injury model, where MSC-derived exosomes were capable of preserving cardioprotective efficacy, reducing inflammation of heart tissue, and inducting macrophage polarisation by shuttling the miR-182 gene to downregulate inflammatory toll like receptor 4 (TLR 4) and upregulate the pro-survival PI3k/Akt signalling pathway [145]. The immunomodulatory effects of MSC-derived exosomes were demonstrated by reduced M1 gene expression markers (iNOS, IL-1β, IL-6, TNF-α) and increased M2 gene expression markers (Arg1, IL-10, CD206, TGF-β) both in vitro and in vivo [145]. In a D-galactose aging-induced cardiac dysfunction model, exosomes derived from umbilical cord MSCs were shown to inhibit cardiac inflammation by increasing the expression of the long noncoding RNA (lncRNA) MALAT1, which suppressed NF-kB/TNF-α signalling pathways [146]. Rat bone marrow MSC-derived exosomes carrying miRNA-125b exhibited substantial therapeutic effects on a myocardial I/R injury rat model, leading to a reduction in inflammatory factors (IL-β, IL-6, and TNF-α) by downregulating the SIRT7 gene and thus improving heart function [147]. The miRNA-125b abrogated the infarct area and provided protection from myocardial ischemia reperfusion injury [147]. The exosomes derived from LPS-primed bone marrow-derived MSCs showed a potential role in macrophage polarisation and tissue repair in several inflammatory models [148].
In a mouse MI model, exosomes from bone marrow-derived MSCs, both primed or unprimed with LPS, were shown to reduce inflammation and myocardial injury by strongly attenuating LPS-dependent NF-κB signalling and activating the Akt1/Akt2 pathway [149]. The LPS-primed MSC-derived exosomes were found to have the greatest effects on cardiac inflammation, cell viability, and apoptosis. In rat cardiomyocytes and rats treated with doxorubicin and trastuzumab to induce oxidative stress, exosomes derived from human mesenchymal progenitor cells obtained from right cardiac appendage tissue were found to reduce cell death, and provide protection against cardiac dysfunction and myocardial fibrosis [150]. These exosomes were shown to be enriched in miR-146a-5p, which might act to suppress a number of target genes that encode signalling mediators of inflammation and cell death in cardiomyocytes.
To increase the efficacy of MSC-derived EVs, drug delivery systems have been developed to enable their intramyocardial delivery post-AMI. One approach is to encapsulate the EVs within a self-assembling peptide (SAP) hydrogel modified with a SDKP pattern [151]. The MSC-derived EVs were administered by intramyocardial injection in a rat MI model, either alone or in conjugation with an angiogenic and anti-inflammatory SAP hydrogel based on (RADA) 4 -SDKP [151]. This was found to increase α-smooth muscle actin (SMA) vessel-like structures to promote angiogenesis, as well as reduced inflammation and proinflammatory CD68+ cells, which resulted in improved ventricular remodelling and cardiac function. Similarly, exosomes encapsulated in a functional peptide hydrogel could be released in a prolonged and stable manner in a rat model of MI, resulting in reduced cardiac damage and improved angiogenesis [152]. In another study, lentivirus packaging was used to engineer exosomal-enriched membrane protein (Lamp2b) fused with ischemic myocardium-targeting peptide CSTSMLKAC (IMTP) [153]. The IMTP-exosomes produced by transfected MSCs were shown to induce a greater reduction in mRNA levels of proinflammatory factors (TNF-α, IL-6, and IL-1β) within the ischaemic heart area in an AMI model compared to blank exosomes. The group treated with IMTP-exosomes also showed a reduced percentage of M1 macrophages (TNF-α+ CD68+) and increased percentage of M2 macrophages (CD206+) compared to groups treated with blank exosomes or PBS. Furthermore, the IMTP-exosomes were found to reduce cardiac cell apoptosis and infarct size, and improve vasculogenesis. In an interesting study that combined the delivery of exosomes and MSCs in a rat MI model, MSC-derived exosomes were given through intramyocardial injection, with or without intravenous infusion of atorvastatin-pretreated MSCs on days 1, 3, or 7 after MI [154]. The combined treatment of exosomes followed by MSC transplantation was found to improve cardiac function, reduce infarct size, and increase neovascularisation to a greater extent than controls treated with exosomes or MSCs alone. Furthermore, the highest improvement in heart function was achieved by the group that received exosomes followed by MSCs at day 3 after MI. The delivery of exosomes before MSC transplantation was found to enhance cell survival and reduce cell apoptosis.

Extracellular Vesicles from Mesenchymal Stem Cells for the Treatment of Preeclampsia
Preeclampsia is a life-threatening cardiovascular disorder occurring in the second-half of pregnancy with multifactorial pathogenesis [155,156]. Although a leading cause of maternal and neonatal morbidity and mortality, there are no current treatments for preeclampsia. The exact pathogenic mechanisms of preeclampsia are not well understood, but two phenotypes have been identified based on the time of diagnosis: early-onset (before 34 weeks of gestation) and late-onset (from 34 weeks of gestation) [155,156]. Early onset preeclampsia is often associated with inappropriate placentation due to inadequate trophoblast invasion leading to placental hypoxia. Late-onset preeclampsia is characterised by irregular growth of the placenta and underlying maternal cardiovascular, metabolic, and inflammatory conditions [156][157][158]. Nevertheless, both phenotypes display inflammation, which is often mediated by abnormally upregulated immune responses and activation of the innate immune system and pro-inflammatory factors [158]. Imbalance between ROS and antioxidants due to placental hypoxia can also lead to systemic inflammation and endothelial dysfunction before preeclampsia is manifested. The pathogenic mechanisms underlying these processes appear to involve the secretion of syncytiotrophoblast macrovesicles, inflammatory factors such as TNF-α, and anti-angiogenic factors such as soluble fms-like tyrosine kinase (sFlt-1) and Endoglin (sEng) into the maternal circulation [156]. In vitro: -MSCs cultured in normoxic or hypoxic conditions produced EVs with the same size and morphology -Both EV groups increased the expression of angiogenic molecules and induced epithelial tube formation, with hypoxic EVs being more angiogenic -Both EV groups were internalised by macrophages and increased macrophage proliferation; both demonstrated anti-inflammatory properties and ability to switch macrophages from M1 to M2 phenotype; hypoxic EVs had greater effects -Hypoxic EVs showed downregulation of 48 miRNAs and upregulation of 20 miRNAs; 4 of the upregulated miRNAs were associated with inflammatory processes (miR-223, miR-146b), proliferation and differentiation (miR-126, miR-199a) In vivo: -Both EV groups showed anti-inflammatory properties, with increase in M2 anti-inflammatory markers and decrease in M1 pro-inflammatory markers -Both EV groups showed accelerated regeneration of muscle tissue, although hypoxic EVs had a greater effect  In vitro: -Reduced expression of pro-inflammatory markers IL-6 and IL-1β, and increased expression of the anti-inflammatory marker IL-10 -Completely or partly attenuated the negative effects exerted by pro-inflammatory cytokine treatment In vivo: -Reduced Iba-1+ cell density, together with reduction in microglial cell body size -Reduced CD68 expression associated with the activated microglia phenotype In vitro: -3D-exosomes had greater effect in upregulating α-secretase and downregulating β-secretase to reduce levels of secreted and intracellular Aβ In vivo: -3D exosomes were more effective at improving spatial learning and memory function, as shown through MWM test -Exosomes were mainly concentrated at the site of delivery but also distributed throughout the brain parenchyma; 3D exosomes were more effective at decreasing Aβ deposition by eliminating production and facilitating clearance of Aβ -Exosomes reduced neuroinflammation by attenuating microglial activation, and markedly reduced oxidative stress; 3D exosomes produced more pronounced effects

Study Source of EVs Type of EVs Administration and Dose Model(s) Used Major Findings
Chen et al., 2020 [147] Rat bone marrow MSCs Exosomes (60-100 nm), enriched with miR-125b or control miR-67 Exosomes (   In vitro: Cells were cultured with exosomes (10 µg/mL) for 6, 12, and 24 h In vivo: Exosomes (5 µg in 10 µL PBS) were injected into the anterior and lateral part of the visibly injured region 5 min before reperfusion In vitro: -Increased myofibroblast density and improved collagen contraction -Promoted fibroblast-to-myofibroblast differentiation in inflammatory environments -Reduced cardiomyocyte apoptosis -Decreased fibroblast migration, but no effect on fibroblast proliferation -Decreased expression of IL-1β and TNF-α, and increased expression of TGF-β In vivo: -Suppressed inflammatory response and improved cardioprotective effects   MSCs have gained growing attention as a potential therapeutic strategy to alleviate inflammation, oxidative stress, and restricted angiogenesis in preeclampsia [156]. MSC treatment has been shown to cause deactivation of both innate and adaptive inflammatory immune cells including monocytes, macrophages, dendritic cells (DCs), CD4+ and CD8+ cells, natural killer (NK) cells, and B-cells, as well as activation of a regulatory sub-set of immune cells to resolve inflammation in preeclampsia [156]. MSC-derived EVs have recently been explored as an alternative, safer and more effective option to whole cell-based therapy for the treatment of preeclampsia. A number of studies have investigated the functional role of the miRNA cargo of MSC-derived EVs, which can metabolically induce changes in macrophage phenotype from a glycolysis-conducted M1 pro-inflammatory phenotype to an oxidative phosphorylation-dependent M2 anti-inflammatory phenotype [156].
From the four studies shown in Table 1 on testing MSC-derived EVs in models of preeclampsia, two were in vitro studies performed using HUVECs [159] or trophoblasts [160] as cell models, while the other two were in vivo studies performed in rat models of preeclampsia [161,162]. EVs derived from human decidua MSCs were found to enhance HUVEC proliferation by reducing inflammation and oxidative stress [159]. The addition of EVs to LPS-treated HUVECs also caused a significant reduction in pro-inflammatory IL-6 cytokine concentration. On the other hand, exosomes from human umbilical cord MSCs were found to promote cell proliferation, migration and cell cycle entry, as well as inhibit apoptosis in human trophoblast cell lines [160]. These effects were increased when exosomes were derived from the same MSCs transfected with miR-133b. The exosomal miR-133b was shown to exert its effects on trophoblasts by regulating glucocorticoid-regulated kinase 1 (SGK1).
The two studies that performed in vivo testing to investigate the effects of MSCderived EVs in rat models of preeclampsia both used human umbilical cord MSCs as the cell source for producing EVs. One study using exosomes from the MSCs showed that these significantly lowered blood pressure and proteinuria in the animal model, both key features of preeclampsia [161]. From in vitro experiments performed in the same study, it was noted that the exosomes could promote trophoblast cell migration, invasion and proliferation while reducing apoptosis, possibly by transferring miR-139-5p to the trophoblasts and thereby regulating the ERK/MMP-2 pathway. In the other study, the exosomes were administered at low, medium or high doses to the animals [162]. Their effects included lowering blood pressure and proteinuria, as well as increasing the number and quality of foetuses, quality of the placenta, microvascular density of the placenta, and VEGF expression while decreasing cell apoptosis and the expression of the anti-angiogenic factor sFlt1. It was interesting to note that the effects of exosomes were exerted in a dose-dependent manner.

Limitations and Future Perspectives of Using Extracellular Vesicles from Mesenchymal Stem Cells for the Treatment of Inflammation-Related Conditions
As demonstrated by the studies discussed in this review, the use of MSC-derived EVs has potential to provide substantial therapeutic benefits in a range of inflammationrelated conditions, including osteoarthritis, rheumatoid arthritis, Alzheimer's disease, cardiovascular disease, and preeclampsia. To date, a small number of studies have been published on applying MSC-derived EVs for each of these conditions in experiments involving in vitro and small animal models of disease. The field of using MSC-derived EVs as therapeutics is still in its infancy, but holds significant promise as a viable option in the management and treatment of inflammatory conditions. Compared to injecting live cells, MSC-derived EVs offer a prominent advantage in the ability to be used as off-the-shelf therapeutics, since they can be preserved in cold storage for long periods of time without loss of biological activity upon thawing, for example, at −20 • C for six months [163] or at −80 • C for up to two years [164]. The isolation and purification of EVs from the conditioned medium that has been used to culture MSCs is expected to be conducted under sterile conditions, removing the need to perform additional sterilisation prior to EV application, unlike for locally administered pharmacological agents. Although comprehensive analyses are still emerging in the field, preclinical studies have demonstrated that EVs have minimal toxicity and immunogenicity, even when applied xenogenetically as a large dose at high frequency [89]. There is a potential safety concern with intravascular injection of large quantities of EVs, particularly MVs that are shed from the cell membrane, if they express the same levels of prothrombotic tissue factor (TF/CD142) and collagen type-1 (COL-1) as the parent MSCs, which may trigger coagulation pathways and lead to blood clots [165]. This is an area that warrants further investigation, although it is less concerning in the local delivery of MSC-derived EVs to specific tissues, which is the typical method of administration for regenerative medicine applications. Although MSC-derived EVs show promising characteristics and preclinical evidence, there remain a number of considerable questions to answer and limitations to address before the application of these therapeutic EVs can be translated to a clinical setting.
The exact mechanisms underlying the therapeutic effects of EVs when applied to different diseases remain incompletely understood in current studies, posing an area where significantly more work is needed. In particular, the optimal tissue source of MSCs for generating EVs has been rarely explored. A wide variability in therapeutic efficacy has been seen across the many MSC sources used in different studies, including bone marrow, adipose tissue, synovial membrane, induced pluripotent stem cells (iPSCs), and embryonic stem cells (ESCs). Direct comparison between studies is difficult due to significant variations in the culture media and conditions used to generate EVs, isolation methods, experimental set-up, and EV dosage. Only one study so far has directly compared two different MSC sources, which showed that EVs derived from iPSCs had superior therapeutic efficacy for the treatment of OA compared to those derived from synovial membrane MSCs [110]. In addition, few studies have investigated the influence of environmental conditions and stresses on the molecular cargo and effectiveness of EVs generated by stem cells [34,166]. These observations raise multiple important questions: (i) which MSC source possesses the greatest therapeutic potential for a given condition, (ii) which factors are critical in determining the therapeutic potential of EVs, and (iii) how to optimise the therapeutic potential of EVs for a chosen MSC source. The posing of these open questions is further supported by interesting observations from a study showing that EVs from adipose-derived MSCs cultured in hypoxic conditions exerted a greater regenerative effect than those cultured in normoxic conditions [82], as well as others demonstrating that the therapeutic potential of MSC-derived EVs could be improved through enrichment with specific miRNAs [83][84][85]119,121,145,148]. In addition to the cell source, another aspect of EV-based therapy that has not been sufficiently explored is the dosage and frequency of treatment. Studies which compared variable dose levels of EVs all tend to demonstrate the best therapeutic outcomes in the highest dose group, creating difficulties in finding an 'optimal' level of EV administration. There is also a wide variation in treatment frequency among the studies included in this review. Further investigations are needed to elucidate the most effective treatment method for a given disease that takes into consideration the source of MSCs, dose of EVs and frequency of administration, among other important factors.
The currently available evidence on the therapeutic applications of MSC-derived EVs in inflammatory conditions are limited to in vitro and small animal studies exploring the short-term outcomes of EV treatment. There is a critical need for future studies to investigate the long-term effects. A question remains about whether the regenerative outcomes and disease attenuation induced by EVs are permanent or temporary. It is possible that the short-term positive outcomes of EV treatment observed in current studies are a result of temporarily altered gene expression, and that given sufficient time the diseased phenotype may return. Addressing this question will enable researchers to determine if EV treatment will require ongoing administration, and if so, the frequency that is required to produce a sustained therapeutic effect. Furthermore, it is crucial that research on EVs as therapeutics is translated to large animal studies, using models such as sheep or pigs that may more closely mirror disease progression and treatment response in humans.
For EVs to become a widely accepted and applied treatment option, a number of technical limitations pertaining to the general field of EV research needs to be addressed. Currently, there is no consensus on the optimal methods for EV generation, isolation, purification and characterisation. To upscale the technology surrounding the use of EVs as therapeutics, standardisation of these processes is essential [167,168]. EV isolation is currently achieved through techniques such as centrifugation, immunoaffinity isolation, polymeric precipitation, size exclusion, and microfluidic devices [47]. Unfortunately, all of these methods carry certain limitations which reduce the ability to upscale the EV production, such as insufficient exclusion capacity resulting in contamination, alteration of EV structure resulting in a loss of function, and poor isolation capacity resulting in incomplete EV fraction isolation and low yield [169]. Another significant challenge lies in the generation of a sufficient quantity of EVs for practical applications, such as for preclinical animal models or clinical trials. The current yield of EVs from MSCs in conventional culture conditions is extremely low, with 1 L of culture medium conditioned with a total of approximately 60 million MSCs only producing 1-2 mg (protein content) of EVs [170]. With single injections in mice of up to 500 µg EVs used in some studies [104] and many others conducting weekly injections, it remains a considerable hurdle to upscale these EV quantities to clinically viable treatment options for humans.
One method that has been explored in the hope of increasing the therapeutic efficacy of EVs is to load them within a hydrogel for delivery. When delivered in an aqueous solution (typically saline), EVs are either rapidly absorbed by the tissue at the site of injection or dissipate to other areas, thereby reducing their effectiveness. The rationale for developing EV-loaded hydrogel delivery systems is to enable the sustained release of EVs over an extended period of time. The feasibility of this approach has been reported in a number of studies, where MSC-derived EVs exhibited increased therapeutic potential to treat chronic liver failure in a rat model [171] or renal I/R injury in a mouse model [172], when delivered in vivo within a polymerised hydrogel compared to EVs delivered in an aqueous medium.
The characterisation and quantification of EVs is another area lacking consensus in the field [167,168]. EV quantification is commonly performed through dynamic light scattering or nanoparticle tracking analysis, the characterisation of surface markers through protein analysis by flow cytometry or western blotting, and morphological analysis through electron microscopy [173]. Nevertheless, there remains significant variation among studies regarding the types of methods selected for EV characterisation. In addition, the nomenclature for the classification of EVs is currently not consistently defined across the literature, with different studies implementing varying size ranges and protein markers for distinguishing different EV subpopulations such as exosomes and MVs. The importance of having a standardised set of methods and definitions cannot be overstated, as it is a requirement for clinical application that EVs can be produced in a consistent and accurate manner to ensure that maximum therapeutic potential along with safety outcomes are consistently achieved.
The current research strongly suggests that MSC-derived EVs have potential to provide significant anti-inflammatory and regenerative effects on damaged or diseased tissues in inflammation-related conditions, particularly when locally administered in large doses. While the landscape for using MSC-derived EVs in the future as a new generation of therapeutics appears promising, there is a critical need for enhancing the efficacy and robustness of EVs, and standardising the methods of EV production and isolation.

Conflicts of Interest:
The authors declare no conflict of interest. The funders had no role in the design of the study; in the collection, analyses, or interpretation of data; in the writing of the manuscript, or in the decision to publish the results.