Practical Considerations for Translating Mesenchymal Stromal Cell-Derived Extracellular Vesicles from Bench to Bed

Extracellular vesicles (EVs) derived from mesenchymal stromal cells (MSCs) have shown potential for the treatment of tendon and ligament injuries. This approach can eliminate the need to transplant live cells to the human body, thereby reducing issues related to the maintenance of cell viability and stability and potential erroneous differentiation of transplanted cells to bone or tumor. Despite these advantages, there are practical issues that need to be considered for successful clinical application of MSC-EV-based products in the treatment of tendon and ligament injuries. This review aims to discuss the general and tissue-specific considerations for manufacturing MSC-EVs for clinical translation. Specifically, we will discuss Good Manufacturing Practice (GMP)-compliant manufacturing and quality control (parent cell source, culture conditions, concentration method, quantity, identity, purity and impurities, sterility, potency, reproducibility, storage and formulation), as well as safety and efficacy issues. Special considerations for applying MSC-EVs, such as their compatibility with arthroscopy for the treatment of tendon and ligament injuries, are also highlighted.


MSCs and MSC-EVs for Tissue Repair and Disease Treatment
Mesenchymal stromal cells (MSCs) are pluripotent, non-hematopoietic stem cells with self-renewal capability. They have immunomodulatory, pro-angiogenic and growth promoting effects and are capable of homing to injured sites. Consequently, there has been an extensive interest in their therapeutic use for tissue repair due to conditions such as neurological disorders [1], liver diseases [2], cardiovascular diseases [3], immune-mediated disorders [4], bone fractures [5] and osteoarthritis [6].
Mounting evidence suggests that MSCs enhanced tissue repair via paracrine factors rather than by direct differentiation [7]. As a consequence, researchers are primarily focusing on MSC secretome, which include soluble proteins, lipids and extracellular vesicles (EVs). EVs are membranous, nano-sized vesicles secreted by all cells. They participate in intercellular communication and possess intrinsic therapeutic activity via addressing their vesicular content of DNA, RNA, proteins and other cellular components from a producer cell to a recipient cell. They can be classified into different types, i.e., exosomes (50-150 nm in diameter), microvesicles (100-1000 nm in diameter) and apoptotic bodies (1-5 µm in diameter), depending on their size and biogenesis. EVs have been shown to modulate the immune response [8,9], potentiate tissue regeneration [10] and function as a potential alternative to stem cell-based therapies [11]. Indeed, MSC-EV-enriched preparations have been shown to be as therapeutically effective as their parent cells in different pre-clinical models in head-to-head comparisons [12][13][14]. Genetically modified EVs can also act as delivery vehicles for therapeutic agents [15]. The capacity of EVs has hence spurred interest in their use both as a delivery system and as a drug for the treatment of various disorders.

Advantages of MSC-EV-Based Therapeutics
EV-based therapies in tissue repair and regeneration offer numerous advantages, i.e., lower cost, easier storage and improved safety and patient compliance. First, EVs lack self-replicating ability, thereby reducing the risk of malignant transformation and ectopic tissue formation. Second, while MSCs generally have low immunogenicity and are suitable for allogeneic transplantation [16][17][18][19][20][21][22], the lack of immunogenic cell surface proteins in exosomes likely further reduces immunogenic response after allogeneic transplantation and allows them to be used as an off-the-shelf product. Third, the lipid bilayer membrane also enhances the stability and reduces the toxicity of the EV cargo during in vivo delivery. Unlike soluble proteins in the MSC secretome, it is easier to concentrate EVs, though there may be a loss of some potency of the secretome. Moreover, EVs have the intrinsic ability to cross the epithelium and blood-brain barrier, and hence, may be useful for the delivery of drugs with low oral bioavailability and poor epithelial penetration [23][24][25][26]. In addition, the ability to engineer EVs to improve their specific cell targeting capacity or potentiate their therapeutic properties also makes them good drug carriers and therapeutics for tissue repair. EVs derived from MSCs therefore have the potential to serve as an alternative to MSCs in tissue repair. Early clinical trials investigating MSC-EV-based therapeutics have already begun [27][28][29]. Table 1 summarizes completed and ongoing clinical trials of MSC-EVs registered at www.clinicaltrials.gov (accessed on 30 June 2022). No clinical studies on MSC-EV were found in clinicaltrialsregister.eu (accessed on 30 June 2022), with the keyword of "Extracellular vesicle", "exosome", "EV", "micro-vesicle" and "nano-vesicle". Abbreviations: bone marrow-derived stromal cells (BMSCs), adipose-derived stromal cells (ADSCs), mesenchymal stromal cells (MSCs), human umbilical cord mesenchymal stromal cells (HUMSCs), extracellular vesicles (EVs), exosomes (Exos), phosphate buffered saline (PBS), intravenous (IV), intraperitoneal infusion (IP); Acute Respiratory Distress Syndrome (ARDS).

MSC-and MSC-EV-Based Therapies for Tendon and Ligament Repair
Tendons and ligaments are subject to high tensile loads and are easily torn as a result of overuse or trauma, resulting in significant pain and disability. Together, tendon and ligament injuries account for 30% of all musculoskeletal consultations [30]. More than 32 million acute and chronic tendon and ligament injuries occur annually in the United States [31]. The outcomes of both conservative treatments and surgical repair of tendon are not satisfactory due to the long healing time, scar tissue formation, adhesion, occasional bone formation and high re-rupture rate. Similar to other tissues, the administration of MSCs is a promising approach for tendon, ligament and tendon-to-bone junction repair [32][33][34]. We have shown that the transplantation of tendon-derived stem cells (TDSCs) promotes tendon regeneration and graft healing after anterior cruciate ligament reconstruction (ACLR) [35][36][37]. Recently, there has been considerable interest in studying the effect of MSC-EVs on tendon and ligament repair, with encouraging results ( Table 2) . However, to date, no clinical trials have been undertaken. As such, the effects of MSC-EVs need to be evaluated in a clinical setting.
The administration of MSC-EVs was generally found to promote cell proliferation and migration [43,44,[47][48][49][53][54][55]58,59], suppress tissue inflammation and apoptosis [40,44,46,48,52,55,58,59], modulate inflammatory response of macrophages [38,39,46,48,52,55], increase collagen deposition [43,45,49], reduce fatty infiltration [50] and promote angiogenesis [52] during tendon and ligament repair. The active molecules in MSC-EVs that contribute to tendon and ligament repair are not entirely known. However, numerous studies have shown that MSC-EVs are rich in miRNA, which contribute to tendon and ligament repair. miRNA sequencing showed that human umbilical cord MSC-derived exosomes (HUMSC-Exos) expressed an antagonist to miR-21a-3p, as miR-21a-3p was under expressed in the exosomes compared to the parent cells. The inhibition of exosomal miR-21a-3p in HUMSC-Exos inactivated RelA/p65 (a core element in the NF-kB pathway involved in inflammation and fibrosis) and reduced the protein expression of Cox2 and α-SMA in rat fibroblasts [40]. Additionally, endoplasmic reticulum stress (ERS)-associated proteins (GRP78, CHOP) and pro-apoptotic protein (Bax) were presented in EVs derived from HUMSCs, potentially explaining their anti-adhesive effect on traumatic tendon injury [41]. In addition, HUMSC-Exos we have been shown to promote tendon healing via miR-27b-3p-mediated suppression of ARHGAP5, resulting in RhoA activation and increased proliferation and migration of primary injured tenocytes [42]. In another study, rat EVs derived from bone marrow-derived stromal cells (BMSC-EVs) were reported to express pro-collagen1A2 and MMP14 proteins, which are important factors for tendon extracellular matrix remodeling. Pro-collagen1A2 was expressed on the membrane surface of BMSC-EVs. Pretreatment of BMSC-EVs with trypsin abrogated their effects on tendon cell proliferation and migration and the expression of collagen type I, suggesting that the biological effects of EVs depended on the interaction of membrane-bound proteins with the recipient tendon cells [43]. Furthermore, miRNA sequencing indicated a significantly higher level of miR-29a-3p in HUMSC-Exos compared to HUMSCs [45]. The level of miR-29a-3p in HUMSC-Exos-treated Achilles tendons was also significantly elevated, and HUMSC-Exos overexpressing miR-29a-3p was found to amplify the effects of HUMSC-Exos on tendon healing in vivo [45]. In addition, exosomes derived from TDSCs (TDSC-Exos) have been found to contain miR-144-3p and enhance tendon repair through miR-144-3p-regulated tenocyte proliferation and migration [49]. Finally, exosomes derived from Scx overexpressing PDGFRα(+) BMSCs reduced osteoclastogenesis and improved tendon-bone healing strength via exosomal miR-6924-5p [56]. Achilles tendon transection and repair in nude mouse -BMSC-EVs educated macrophages (EEM) (3 × 10 9 human BMSC-derived EVs were used to educate 75 cm 2 flask of human macrophages to M2 phenotype for 3 days) -1 × 10 6 of EEM in 20 µL of saline; same amount of BMSC-EVs used to educate macrophage was also tested but exact dosage not reported -Immediately after repair, once -Injection in the surgical pouch around the injured tendon 7 and 14 days post-injury EEM treatment substantially improved the biomechanical properties of the healing tendon but showed no improvement in collagen fiber organization. The EV or BMSC treatment showed biological responses but no effects on the biomechanical properties of tendon or collagen fiber organization. Both EEM and EV treatment reduced the MI/M2 ratio. EV, but not EEM, further increased the number of endothelial cells compared to injury only and EEM treatments. Although both EEM and EV treatments reduced the protein expression of collagen type I, no treatment effects were observed with the expression of type III collagen, type I/type III collagen ratio or collagen organization. [38] Mouse Achilles tendon two-third partial transection and repair -Mouse naïve and IFNγ-primed ADSC-EVs laden on collagen sheet -5-6 × 10 9 of particles derived from 5 × 105 of ADSCs -Once -EV-laden collagen sheet was wrapped around injury site

1, 3 and 7 days post-injury
Compared with the untreated control group, primed ADSC-EVs, but not their unprimed counterparts, further reduced the rate of post-repair tendon gap formation and rupture and promoted collagen formation at the injury site. Primed ADSC-EVs, but not unprimed EVs, attenuated the early tendon inflammatory response after injury via modulation of the macrophage inflammatory response. [39] Rat mid-Achilles tendon transection and suture repair Both HCPT-EVs and unprimed EVs reduced tendon adhesion. However, only HCPT-EVs significantly improved the histological healing score. There was no significant improvement in the maximal tensile strengths of the healing tendon after treatment with HCPT-EVs or unprimed EVs. HCPT-EVs contained more endoplasmic reticulum stress (ERS)-associated protein compared to unprimed EVs and activated the ERS pathway in fibroblast to counteract myofibroblast differentiation. [41] Rat mid-Achilles tendon transection and suture repair -HUMSC-Exos -100 µg of particles dissolved in 50 µL of PBS -Once -Subcutaneous injection 7 days post-injury HUMSC-Exos promoted tendon repair via exosomal miR-27b-3p, which increased cell proliferation, invasion and RhoA activity of primary injured tenocytes.
[42]   Exos wrapped with "BMP-2 supplemented PASP-PLA microcapsules" promoted tendon and bone interface healing after rotator cuff injury via the Smad/RUNX2 signaling pathway. The expressions of tendon regeneration-and cartilage differentiation-related proteins were significantly upregulated. [51] Rat rotator cuff repair The ADSC-Exos-hydrogel group promoted rotator cuff repair compared to the control group, with improved histology and biomechanical properties. [53] Rat supraspinatus tendon transection and repair -Blood derived purified exosome product (i.e., PEP) -3.8 × 10 10 of particles per mL in 3 × 3 × 3 mm of TISSEEL -Once -Local application 6 weeks post-injury PEP in TISSEEL increased the mRNA expression of Col1, Col3, Scx, Tnmd, Tnc, Dcn and IGF compared to the control group. It also promoted remodeling of collagen fibers and new cartilage-like tissue formation at the tendon-bone interface after 6 weeks. [54] Mouse Achilles tendon-bone reconstruction model [55] Mouse extra-articular Achilles tendon-bone tunnel model -Exosomes derived from mouse Scx overexpressing PDGFRα(+) BMSCs -1 × 10 10 of particles (direct injection to the bone tunnel with tendon graft) -Once -Local injection into bone tunnel and incubated for 3-5 min for absorption before graft insertion 1, 2 and 3 weeks post-injury Exosomes derived from Scx overexpressing PDGFRα(+) BMSCs reduced osteoclastogenesis and improved tendon-bone healing strength via exosomal miR-6924-5p.
[56] Both BMSC-EVs and EEMs. BMSC-EVs promoted ligament repair with improvement in collagen (type I and III) production and collagen organization, as well as reduction of scar formation. EMMs improved the mechanical properties of healing ligament and reduced the M1/M2 macrophage ratio. [60] A case study of a horse suffering from suspensory ligament injury -Allogeneic microvesicles (MVs) derived from 5-azacytidine (AZA) and resveratrol (RES)-treated ADSCs isolated from horse with metabolic syndrome -Dosage not reported -Twice (7 days after injury and 9 months after first injection) -Ultrasound-guided injection into injury site 10 months and 12 months after first injection MVs improved the lesion filling, angiogenesis and elasticity of injured tissue. [61] Abbreviation: extracellular vesicle (EVs), exosomes (Exos), bone marrow-derived stromal cells (BMSCs), adipose-derived stromal cells (ADSCs), human umbilical cord mesenchymal stem cells (HUMSCs), phosphate buffered saline (PBS), tendon-derived stem cells (TDSCs).

Translational Gap and Aim of Review
Unlike single active substances, MSCs and their EVs are comprised of pleiotropic bioactive ingredients, representing a major obstacle for the manufacture of reproducible drug products with consistently high efficacy and safety. The content of EVs is affected by the heterogenicity of the donor MSCs. The phenotypes and biological activities of MSCs depend on their origin (biological niche) or the conditions of potential donors (age, diseases, obesity or unknown factors) [62]. The artificial microenvironment of MSCs such as O 2 tension, substrate and extracellular matrix cues, culture media, inflammatory stimuli and genetic manipulations can also influence the resulting phenotypes and, hence, paracrine activities [62]. HUMSC-Exos cultured in two different cell culture media showed different surface compositions and cytokine contents [63]. Furthermore, the passage number of MSCs influences the biological activities of EVs. Exosomes isolated from early passage of rat BMSCs exhibited higher neuroprotective potential compared to exosomes derived from later passages [64]. Moreover, the mechanisms of the biogenesis of EVs also influence their content. At least three major subpopulations of EVs produced by different mechanisms, namely exosomes, microvesicles and apoptotic bodies, have been reported. Immortalized E1-MYC 16.3 human embryonic stem cell-derived MSCs were reported to produce at least three distinct 100 nm EV types by different biogenesis pathways that could be distinguished by their membrane lipid composition, proteome and RNA cargos [65]. In one study, BMSC-EVs were fractionated into different density fractions [66]. The different EV gradient fractions were heterogeneous in terms of the quantity and expression of classical exosomal markers. The miRNA and protein profiles of these EV fractions were different, and they also showed differential effects on renal tubular cells in terms of degree of internationalization, stimulation of cell proliferation and inhibition of apoptosis. Therefore, MSC-EVs are heterogeneous, with specific signatures accounting for the biological activity of different subpopulations. EVs produced by a given population of cells cultured under identical conditions are not identical.
Gene therapy medicinal products, somatic cell therapy medicinal products and tissueengineered products are regulated as advanced therapy products (ATPs) in the authors' homeland [67] and other places, including Europe (Regulation EC No 1394/2007) [68]. Native cell-free EVs without transgene products are not categorized as ATPs and fall within a regulatory gap. However, EVs are categorized as pharmaceutical products/biological medicinal products, and hence, require clinical trials as investigational new drugs (INDs) before getting marketing approval. To receive an IND, investigators have to demonstrate that their EV-based products fulfill the requirements regarding quality, reproducibility, safety and efficacy. This is especially a challenge for EV-based products due to the heterogeneity and complexity of their composition. While fractionation or sorting may help to identify more homogeneous subpopulations of EVs, careful control of the donor cell source, the cell culture conditions and the EV enrichment process are crucial for clinical translation of EV-based therapies, as they would standardize the EV composition and content.
In this review, we will discuss the general and tendon/ligament-specific considerations of manufacturing MSC-EV-based products for clinical translation. Specially, we will discuss Good Manufacturing Practice (GMP)-compliant manufacturing and quality control (QC), safety and efficacy issues. Special considerations for applying MSC-EVs for the treatment of tendon and ligament injuries are also discussed.

Considerations for the Manufacturing, Quality Control, Safety and Efficacy of MSC-EVs
The process is the product. A scalable, reproducible and GMP-compliant manufacturing protocol should be followed to produce MSC-EV therapeutics for the promotion of tendon and ligament healing. Standard operating procedures (SOPs) for MSC-EV manufacturing should be followed and the production process should be documented to ensure high batch-to-batch consistency. Newly produced batches should be compared to previous batches regarding physiochemical properties and biological activity.
Besides controlling the manufacturing process, it is important to specify the release criteria of the physio-chemical-biological properties of MSC-EVs for clinical trials or applications. This is vital for ensuring the efficacy, safety and consistency of MSC-EVs as pharmaceutical products. The release criteria are highly product-specific and depend on the cell source and clinical application. Each MSC-EV preparation for a clinical application has its own specifications. The requirements to examine the physiochemical characteristics and sterility are common to different EV products. However, the requirements for testing the identity, purity, impurities, potency, reproducibility, storage and formulation vary with different MSC-EV preparations. The issues that need to be considered for the manufacturing, QC, safety and efficacy of MSC-EVs are discussed below. The following factors should be considered for the isolation of EV-producing MSCs for therapeutic applications [69].
Age, medication, and medical history of donor 3.
Any priming of MSCs 5.
Any genetic modification of MSCs 6.
MSC storage and recovery conditions The source of MSCs for EV production determines the manufacturing and QC strategy of the process and the final product. The primary factors to consider include the tissue origin of the cells, the age and medical history of the donor, whether the cells are autologous or allogeneic and whether the cells have been primed or genetically modified. The plasticity of MSCs isolated from different tissues varies, affecting the properties of the isolated EVs [70][71][72]. The proliferation capacity and yield of MSCs also vary based on the tissue of origin [73,74]. Both proliferation capacity and MSCs yield are important factors for large-scale and affordable production of MSCs and EVs for therapeutic applications. The use of cells isolated from tissue similar to the site of transplantation may have an advantage due to the priming of the cells by the local tissue microenvironment. For instance, TDSCs were reported to show higher proliferation, colony-forming ability and multi-lineage differentiation potential compared to paired BMSCs [73] and to form tendon-like tissue after subcutaneous transplantation to a nude mouse model [75]. MSCs isolated from young and healthy subjects are preferred, as aged MSCs and MSCs derived from diabetic or obese patients have been reported to show reduced activities [76][77][78][79][80][81][82]. Clinical examinations are needed to check for signs or symptoms of communicable diseases among donors. In addition, serological tests of at least the human immunodeficiency virus (types 1 and 2), hepatitis virus (B and C) and Treponema pallidum should be performed [83]. The inclusion and exclusion criteria for MSC donors need to be carefully defined and followed.
MSCs are generally reported to have low immunogenicity, and the use of MSC-EVs further reduces safety concerns. As EVs cannot replicate, the safety concerns of using EVs derived from autologous and allogeneic cells for tissue repair are similar. The use of allogeneic cells allows for the large-scale production of EV-based products as off-the-shelf materials and is, hence, preferred. However, the ease, feasibility and ethical concern of getting allogeneic MSCs from patients not suffering from injuries vary with the tissue of origin of MSCs. The identification of a practical MSC source with high proliferative and multi-lineage differential potential is crucial for large-scale manufacturing of MSC-EVs for clinical application. To use allogeneic MSCs for EV production, a two-tier cell banking system, consisting of a fully characterized master cell bank (MCB) and partially characterized working cell banks (WCB), should be set up [84]. A four-tier cell banking system has also been suggested [85]. However, at the early stages, as a first-in man clinical trial, an approach based on only a well-characterized MCB is acceptable (may add a Post Production Cell Bank (PPCB) as discussed below). The criteria for releasing MSCs for EV production should be clearly defined.

Culture Conditions
In additional to the MSC source, the culture conditions should also be considered. How the cells are manipulated (i.e., any stimulation or pre-treatment), cultured and stored, as well as the experimental conditions (i.e., culture medium composition and lot numbers, passage number, days in culture, seeding density, density/confluence at harvest, culture volume, culture containers, surface coatings, oxygen or other gas tensions and frequency and intervals of harvest) can affect the recovery of MSC-EVs, and, hence, should be documented and remain consistent [86].
For the culture of EV-producing MSCs, a GMP-grade culture system should be considered. Both static systems, like flasks, and dynamic systems, such as bioreactors, can be used for MSC culture. For static systems, both standard tissue culture flasks and special-coated flasks such as CellBIND ® flasks have been used to produce GMP-compliant EVs [87]. For large-scale production of MSCs, bioreactors are preferred. The metabolic (glucose and lactate) and physical condition (temperature, pH, oxygen, carbon dioxide) readouts during culturing in bioreactors allows for better control and monitoring of the activities of MSCs, which is beneficial for meeting the regulatory requirements for clinical-grade production. Cell culture conditions such as pH, temperature, pO 2 , cell culture duration and metabolic activity (such as glucose and lactose), when applicable, should be defined, recorded and compared to the QC acceptance criteria at the manufacturing phase. Regular cell counting and checking of cell viability should be done. The growth rate of MSCs should be monitored by population doubling time and level.
The use of animal origin-free dissociation enzymes and a culture medium of defined composition is preferred for MSC culture and EV enrichment. The use of a xeno-free medium for cell culture has been reported to reduce the doubling time compared to that of cells cultured in research-grade conditions, to increase exosome yield and to remove 97% of the contaminating proteins [88]. EV-depleted medium should be used for EV isolation. The collection of EVs under ideal, serum-free conditions reduces contamination by non-EV proteins and co-isolation of exogenous EVs from serum. However, starvation may affect the physiology of the EV-producing MSCs, and, hence, EV quantity and quality. Human platelet lysate (HPL)-based EV-depleted medium was shown to be suitable for GMP-compliant MSC propagation and MSC-EV enrichment with retained characteristic of MSC surface marker expression, cell morphology, viability and in vitro differentiation potential [89]. In such cases, we recommend including a non-conditioned medium control to assess the contribution of the medium itself to EV production. If priming agents are added to MSCs to improve the quantity and quality of EV production, the concentration of residual priming molecules in the medium after washing and EV yield per cell equivalent (CE) should be documented and compared to the acceptance criteria. The percentage of dead cells at the time of MSC-EV harvest should be indicated, since even a small percentage of dead cells can release apoptotic bodies exceeding the amount of EVs [86].
A cell passage limit for harvesting EVs should be set, as MSCs show cellular senescence and reduced activity during sub-culturing [90]. To test the stability of MSCs during in vitro passaging, a PPCB, consisting of cells at the limit of in vitro passaging for EV production, can be included.
The cryopreservation and thawing methods, as well as the maximum cryopreservation time, should be validated and defined in the manufacturing process. The cell identity, viability and growth rate should be checked after recovery of MSCs from freezing. According to the Food and Drug Administration (FDA) and European Medicines Agency (EMA) guidelines, the minimum acceptance criteria for the viability of MSC-based products and after cell thawing are ≥80% and ≥70%, respectively [91,92].
The identity of MSC culture needs to be checked. The International Society for Cellular Therapy has published a set of minimal criteria for defining MSCs (plastic-adherent, expression of phenotypic markers and ability to differentiate) [93]. In reality, it is wellknown that the ISCT criteria are rarely clinically useful, as the markers are not specific to MSCs [94]. For MSCs isolated from specific tissue, additional tissue-specific markers may be included to confirm the cell source. For instance, TDSCs are known to express high levels of scleraxis (Scx) and tendomodulin (Tnmd), which could be added to the marker list for cell identification [94]. In addition, positive expression of major histocompatibility complex (MHC) class I and low/negative expression of MHC II molecules are frequently evaluated due to their roles in immune cell recognition and, thus, their relationship with potential immune response after allogeneic MSC transplantation.
A potency test mimicking the in vivo effects of MSCs is recommended. Mixed lymphocyte reaction (MLR) and similar approaches (peripheral blood mononuclear cell, PBMC stimulation assay) are commonly used by most groups studying MSCs [95]. However, there are concerns about their robustness as potency release tests. Several variations of the standard assays have been proposed to increase robustness, such as measuring MSCstimulated induction of regulatory T cells (Treg) and assessing markers indicating the immunosuppressive properties of MSCs, including CD200, TNF-αR, IDO and PD-L1, after IFN-γ stimulation [96][97][98][99][100]. Further validation of these tests is needed.
The manufacture and QC of EVs derived from unmodified cells are simpler compared to those for EVs isolated from apparently more complicated, genetically modified cells. More stringent regulations by health authorities are anticipated. The potential risks due to the vector and the specific transgene used for cell transformation should be identified and controlled in the manufacturing and QC procedures. Furthermore, endotoxin, bacterial, fungal, mycoplasma, adventitious viral, genomic stability (such as karyotyping) and in vitro tumorigenicity tests (such as soft agar colony formation assay) of MSC culture should be done.
In summary, throughout the in vitro expansion process, parent MSCs should be characterized at different checkpoints. The QC panel, including cell culture conditions, cryopreservation and thawing procedures, identity, viability, growth rate, purity, impurities, sterility, stability, safety, and potency of parent MSC banking, is shown in Table 3. MSCs are released for EV isolation only when all the QC results comply with the specifications. As with EV preparations, some MSC samples should be retained for analytical (reference samples) and identification purposes (final product).

Enrichment Method
There is no single optimal method for the concentration or enrichment of MSC-EVs. Different enrichment methods produce different populations [72]. Enrichment methods differ in terms of cost, ease of use, throughput, requirements in time and instrumentation, vesicle loss and the purity of the final product. The MISEV2018 guideline suggests arranging each method on a "recovery vs. specificity" grid [86]. By choosing a highly scalable and automatable concentration method as early as in the pre-clinical stage for MSC-EV isolation, changes in the EV specifications due to the scaling up of the isolation method for clinical trials can be avoided.
Ultracentrifugation (UC) is the most commonly used method for EV enrichment. However, this approach can co-isolate non-EV components, especially for biological fluids such as serum and urine. Due to high shear forces, EVs may aggregate or break, leading to reduced biological activity [106,107]. Ultracentrifugation is time-consuming and has low throughput, limiting its use to small-scale studies. Density gradient centrifugation (DGC) is more efficient for EV isolation which concentrates EVs in a discrete density band. However, the presence of residual gradient material may affect the purity of the finished product. Tangential flow filtration (TFF) is a gentle, size-based fractionation method for EV concentration. It is time-efficient and suitable for large-scale concentration. TFF provides higher EV yields and more effective removal of soluble proteins than enrichment by UC. EVs thus isolated are morphologically intact and display superior batch-to-batch consistency [72,108]. An MSC secretome concentrated by TFF was reported to yield higher protein, lipid, cytokine, and exosome contents than those concentrated by UC [109]. Sizeexclusion chromatography (SEC) produces purer EVs. A previous study showed a 100-fold reduction in ferritin, a contaminant, in SEC-concentrated exosomes [110,111].
The enrichment of EVs by precipitation with polymers such as polyethylene glycol (PEG) inevitably leaves residual polymers in the preparation, interfering with downstream characterization and biological activities [111]. EVs prepared by the precipitation method have been shown to have lower biological activities than those enriched by DGC; this might be due to the masking of important EV surface molecules by residual precipitation reagents [111,112] and may render the EVs unusable for therapeutic applications. Moreover, this method is highly non-specific and can easily co-precipitate non-EV components. A combination of different enrichment methods may be used to enhance EV integrity and activities while reducing contamination. For example, SEC has been used to remove gradient material in DGC [113]. TFF combined with SEC has been reported to produce highly pure and functional EVs from large volumes of samples with minimal vesicle loss [114]. Details of the advantages and limitations of different enrichment methods can be found in a recent review [115].
The production and enrichment process, once standardized, should not be easily changed. Any modifications made to any step of the enrichment process need to be validated to confirm that the properties of the MSC-EVs remain unchanged.

MSC-EV Quality Control
Once the manufacturing procedures of MSC-EVs have been established, a list of QC standards to ensure the quality of the resulting EV preparations produced in different batches should be devised. While the characterization of MSCs and MSC-EVs should be as exhaustive as possible during the product development phase, the tests chosen for routine in-process control, recovery from storage and final product analysis should be technically relevant and feasible in terms of cost, labor and time. The quantity and identity, purity and impurities, sterility, potency, reproducibility, storage and formulation are the most commonly assessed parameters of MSC-EVs.

Quality and Identity Physicochemical Properties
The physical appearance of MSC-EVs, including the color, uniformity of mass and the presence of visible particles, needs to be noted and reported. General tests, such as pH and osmolality, of MSC-EV solutions are required. For freeze-dried MSC-EVs, the appearance of lyophilizate as well as the dissolution time, color, moisture content and clarity of the reconstituted solution should be recorded in the product development, in-process manufacturing and recovery phases, as well as in the final product.
Particle Size and Concentration MSC-EV size and concentration, and optionally, zeta potential (ZP), should be documented in the product development, in-process manufacturing and recovery phases, as well as in the final product. The particle size of MSC-EVs can be determined by tunable resistive pulse sensing (TRPS), microfluidic resistive pulse sensing (MRPS), nanoparticle tracking analysis (NTA) or dynamic light scattering (DLS). Additionally, NTA, MRPS and TRPS make it possible to measure particle concentrations. Since these techniques cannot differentiate between EVs and non-EV particles and tend to overestimate EV concentration, unless antibodies staining is allowed, additional tests to validate the successful isolation of MSC-EV particles are needed. Many NTA devices can be used to analyze vesicles labelled with fluorescence dyes, which can increase the specificity of MSC-EV measurements. However, the sensitivity of such devices is limited for small particles. Moreover, only a subset of MSC-EVs can be measured, as there is no universal EV marker. Therefore, it is important to supplement this approach with examinations of MSC-EV morphologies by electron microscopy (EM) in the product development stage. The expression of MSC-EV quantities in cell or EV component (proteins, lipids, RNA) equivalents is helpful in dose-response studies and in gauging the yield of EVs in different batches of MSCs.

Zeta Potential
Zeta potential (ZP) describes the net electrical charge of molecules on the EV surface and is a measure of colloidal stability and aggregation of EVs. The particle environment (presence of charged or uncharged molecules that can adsorb on the particle surface, particle concentration and the pH and ionic strength of the solution) affects the ZP [116]. The ZP of MSC-EVs can be monitored using TRPS/MRPS/NTA/DLS and is expected to be slightly negative [117][118][119]. A stability study based on evaluating the colloidal behavior of MSC-EVs may be useful for identifying the optimal storage conditions.

Morphology
To differentiate EVs from non-EV particles, EM analysis is needed. Transmission electronic microscopy (TEM) shows the typical lipid bilayer of EV as a cup-shaped structure in the resulting image. It can show non-EV substances in the mixture, but it cannot quantify the amount of contaminating soluble factors. EM analysis may not be feasible as a routine test in the in-process manufacturing and recovery phases or in the final product but is recommended at least during the product development phase.

EV Phenotyping
EVs should be phenotyped by the expression of EV protein markers, as suggested by International Society of Extracellular Vesicles [86]. The minimal information for studies of EVs 2018 (MISEV2018) has suggested some criteria for defining EVs [86]. They recommended that at least one positive transmembrane/glycosylphosphatidylinositol (GPI)-anchored protein (e.g., tetraspanins (CD9, CD63, CD81)), one cytosolic protein recovered from EVs (e.g., ESCRT-I/II/III (TSG101) and the accessory protein ALIX, as well as heat shock proteins HSC70 and HSP84) and a negative non-EV protein marker in a specific system (e.g., apolipoprotein A1/2, apolipoprotein B, albumin for plasma) should be assessed. Quantitative analysis of potential contamination based on the presence of proteins in subcellular compartments other than the plasma membrane and endosomes, such as endoplasmic reticulum markers calnexin and GRP94/GP96, can also provide information about the purity of the exosomes. A list of potential contaminants for testing needs to be created in advance, based on the tissue source used for MSC-EV isolation.
Unlike the minimal criteria for defining multipotent MSCs proposed by the International Society for Cellular Therapy [93], the threshold percentage of particles expressing EV markers is not defined by MISEV [86]. Hence, Western blotting of EV markers is still commonly used to confirm EV identity. The result is semi-quantitative and requires large sample volumes, and, hence, is poorly adapted to high throughput analysis. Techniques such as enzyme-linked immunosorbent assay (ELISA), MACSPlex Exosome kit [120], small particle flow cytometry and nano-flow cytometry [121], which enable multiplexing and require only small amounts of EVs for analysis, are preferred for QC of MSC-EVs.

MSC-EV Content
As MSC-EVs carry inherently complex cell-specific cargos of proteins, lipids and genetic materials, standardized characterizations are challenging. In addition to the core signature of highly enriched vesicular proteins, other proteins reflecting the specific parent cell-origin and biogenesis pathway are present. From a therapeutic perspective, this complexity needs to be understood through comprehensive (multi)omic studies [122]. Mass spectrometry (MS)-based proteomics, metabolomics and lipidomics and next-generation sequencing transcriptomics (NGS/RNA-seq) allow high throughput and quantitative studies of EV-derived proteins, metabolites, lipids and RNA species, respectively [123]. These molecules can provide insights into unique biomarkers and the underlying biological mechanisms of MSC-EV-based therapeutics. This information is useful for the development of biomarkers for QC of MSC-EVs of different tissue origins and with different biogenesis pathways.
For routine QC in GMP facilities, it may not be suitable to use the non-targeted profiling approach described above for the measurement of proteins, metabolites, lipids and RNA species in MSC-EVs. As a proxy for the standardization of MSC-EV preparations, the use of simple, reliable and high throughput methods, such as bicinchoninic acid (BCA) assay for total proteins, sulfovanilin assay [124] or Nile red assay [109] for lipids and RiboGreen assay or UV-Vis spectrophotometry (e.g., NanoDrop) for RNA species, may be more appropriate, despite the fact that the sensitivities of these detection methods are limited. It is important to keep in mind that the total protein content is valid only if the MSC-EVs are isolated from cells cultured in a serum-free, chemically defined medium. Some RNA and DNA species may be presented in the MSC-EV preparation as un-encapsulated particles. We recommend measuring the RNA/DNA content in MSC-EV preparations with and without DNase/RNase treatment to differentiate RNA/DNA content in MSC-EVs from impurities in the development, manufacturing and recovery phases, as well as in the final product. For MSC-EV-based products with known RNA molecules, microarrays and real time quantitative reverse transcription polymerase chain reaction (qRT-PCR) can be used to test for the presence of specific RNA species in the development, manufacturing and recovery phase and in the final product.

Purity and Impurities Purity
Consensus on the definitions and quality metrics of EV purity remain to be established. The particle-to-protein ratio (P/µg protein) [125,126], protein-to-lipid ratio [124,127] and particle-to-RNA ratio [128] have been suggested as indicators of EV purity. Among them, the particle-to-protein ratio is commonly used, because impure preparations are expected to contain non-EV proteins, reducing the particle-to-protein ratio.
Based on the study of cell lines and biological fluids, Webber and Clayton [125] proposed the concentrations >3 × 10 10 P/µg, 2 × 10 9 -2 × 10 10 P/µg and <1.5 × 10 9 P/µg to represent highly pure, less pure and impure EV preparations, respectively. However, it was demonstrated that it was difficult to reach the high purity ratio for biological fluids by simple ultracentrifugation and washing. The simple ultracentrifugation and washing of EVs derived from culture cells only reached the less pure ratio. More complex enrichment strategies or additional steps such as further concentration of EVs by DGC are required to achieve high purity; however, these methods may not be suitable for achieving the medium/high throughput required in clinical trials. Affinity capture-based approaches increase the purity of EVs, but the presence of antibodies may affect purity assessments. With the caveats that the particle-to-protein, protein-to-lipid and particle-to-RNA ratios are specific to a given biofluid or type of EV-producing cell and depend heavily on the respective enrichment method, they are useful in assessing batch-to-batch purity in a defined production system. These ratios should therefore be determined in-house for each specific MSC-EV product. Since the particle count obtained by nanoparticle tracking does not differentiate EV from non-EV particles, measuring the ratio of fluorescent EV particles expressing specific EV markers (CD9/CD63/CD81) to protein concentration may improve the specificity, although this method has not been validated.
It is understandable that a less pure MSC-EV preparation may raise concerns about efficacy and safety during clinical applications. While a certain degree of purity of MSC-EVs is definitely required for therapeutic applications, an extraordinarily pure MSC-EV preparation is not necessarily more effective than a less pure one. Additionally, highly pure MSC-EVs are not stable. Moreover, elevated concentration may damage the EVs and remove the loosely associated factors that act in conjunction with them, resulting in a loss of biological effects. Indeed, a recent study showed that the biological effects of MSC-EVs were actually attributable to the contaminating soluble factors present in the EV samples collected using a low-purity method [129]. Therefore, from the cost perspective of manufacturing MSC-EVs and where function is paramount, a less pure MSC-EV preparation may be more efficacious than a highly pure one. Some impurities, such as albumin and fibrinogen, are sometimes added to increase product stability. In cases where the impurities are not expected to have harmful effects on patients, batch-to-batch consistency is more important than reducing the level of impurities.

Impurities
Impurities can be process-or product-related. Process-related impurities include materials from the manufacturing steps, substrates or cell culture supplements. Product-related impurities include MSC-EVs degradation products which appear during manufacturing and storage. An upper limit of impurities should be defined for MSC-EV-based products. Measuring particle concentration and particle size, the concentration of biomarker-positive EVs and the total protein, total lipid and total RNA/DNA contents and their ratios will make it possible to indirectly monitor MSC-EVs and their degradation.

Sterility
Standard microbiological tests determining endotoxin levels, microorganisms, mycoplasma and adventitious viruses, performed by certified laboratories, are mandatory for GMP-compliant MSC-EV manufacturing.

Potency
In vitro biological assays that reflect the proposed/hypothesized mechanisms of action (MoA) of MSC-EVs in a quantitative manner should be developed for QC of the potency of different batches of MSC-EV preparations. Each therapeutic application of MSC-EVs should have its own in vitro potency assays to predict the intended therapeutic effects. Potency assays should be valid, reproducible, specific, sensitive, robust and cost-effective and have well-defined pass/fail criteria to meet the GMP-compliant release guidelines [130]. For instance, MSC-EVs have been shown to promote tendon and ligament healing by promoting tenogenesis, immunosuppression, proliferation and/or angiogenesis [131]. Hence, quantitative analysis of RNA or protein cargos that mediate these therapeutic effects may be one approach for quality control. Due to the complex MoA of MSC-EVs, the use of multiple assays to measure several MSC-EV attributes related to the intended use may be needed. Each individual potency assay should have its own reference standard, i.e., either a validated international or national one for established biochemical assays or one based results of an internally validated reference of MSC-EV preparations for cell-based assays.

Reproducibility
Batch-to-batch reproducibility in the manufacturing process is a key parameter of every pharmaceutical drug. A previous study showed that the batch of secretome, among other factors, including lyophilization, the concentration of excipient and total amount of excipients, is the primary source of variability of the lipid and protein contents and the anti-elastase activity in the production of EVs derived from human adipose-derived stromal cells (ADSC-EVs) [132]. To reduce variability, MSC-EVs isolated from MSCs of different donors or different cell passages can be pooled to compensate for inter-donor differences. The quantity, identity, purity, impurities and potency of different batches of MSC-EVs can then be compared. The pool size is determined by the variability of samples from different donors or passages. Reductions in variability are theoretically the reciprocal of the square root of the number of samples pooled. This equation was confirmed in a previous study after pooling EVs harvested from peripheral blood mononuclear cells (PBMSCs) isolated from different donors [133]. The % deviation was calculated as the standard deviation (SD)/mean × 100%; this can give information about the variability of different batches. In a previous study, a % deviation of less than 10 was regarded as demonstrating low variation [133]. There is a potential safety concern regarding combining EVs from different donors. Strict donor selection criteria and in-process controls such as traceability of the donor should be implemented in the manufacturing process. To enhance traceability, the pooling of MSC-EVs derived from different cell stocks is preferred over the pooling of MSCs from different donors. As there are currently no standards for MSC-EV-based products, a batch developed in-house, shown to be stable and suitable for clinical trials, should be used as an internal reference by which to calibrate the results of different batches and analytical instruments. Potency relative to the in-house reference batch should be reported.

Storage and Formulation
The storage and recovery conditions of isolated MSC-EVs can affect the EV characteristics, including stability, number of particles, aggregation and functions. The stability of biological products decreases over time during storage. The physiochemical and biological properties of MSC-EVs during storage for various times (e.g., 1, 3, 6 months) under accelerated and stress conditions (e.g., temperatures, relative humidity, pH) need to be determined to understand the degradation patterns and to define the limits of stability. EVs are sensitive to changes in pH, which can cause EV aggregation and loss of function. Isotonic buffers are recommended for storing EVs to prevent pH shifts during storage, freezing and thawing. Highly pure EVs may adhere to the surface of the storage container, making them impossible to recover. Low-protein binding synthetic materials should be used as storage containers. Common storage methods of MSC-EVs include freezing at −80 • C and lyophilization. The storage and retrieval conditions of MSC-EVs should be documented. If MSC-EV products are frozen, the number of repeated freeze-thaw cycles should be minimized to prevent EV aggregation, and QC measures should be built in after thawing in the manufacturing process. One or more cryoprotectants are usually added during cryopreservation [134]. Freeze-drying is expected to increase the ease of handling and shelf-life of MSC-EVs compared to suspensions. However, freeze-drying was reported to reduce the amount of proteins and lipids in ADSC-EV preparations [132]. The addition of lyoprotectants such as mannitol, trehalose or sucrose may improve the preservation of EVs under freeze-drying [134]. The choice of excipients can affect the protein and lipid contents, as well as the anti-elastase activity, of MSC-EV preparations [132,135,136] and should be tested. Since excipients affect QC procedures, they should be confirmed early in the manufacturing process.
The route of administration also influences the formulation, efficacy and safety of MSC-EVs. Biomaterials can be added to aid MSC-EV delivery. The kinetics of EV release from biomaterials should be determined. Local administration of MSC-EVs, ideally ultrasoundguided, is preferred, as tendon and ligament injuries are localized, allowing better control and fewer side effects of MSC-EV treatment. Surgeries such as ACLR and rotator cuff repair are assisted with arthroscopy to minimize damage to the surrounding healthy tissue and scarring, shorten recovery time and lower the risk of complications compared to open surgeries. Continuous flow of the irrigation buffer is required to visualize the operated site via an arthroscope. For arthroscopic-assisted tendon and ligament repair, MSC-EVs loaded in pre-formed biomaterial ex vivo are required. This avoids dispersion of MSC-EVs by the irrigation buffer and ensures their successful delivery to the operated site.

Safety
Prior to application in clinical trials, the effect of dose and frequency of administration of MSC-EVs on toxicity and immunogenicity need to be evaluated in animals. Phase I clinical trials of pharmacodynamics, pharmacokinetics and potential off-target effects of MSC-EVs at the proposed route of administration are also needed. While MSC-EVs share the complexity of ATP products, they do not fulfill the definition of an ATP unless they contain the transgene-products from genetically modified cells. Despite this, the general scientific principles regarding quality (characterization, potency, reproducibility) and non-clinical and clinical requirements for pharmaceutical products are applicable to MSC-EVs [68]. Autologous MSC-EVs naturally occur in the body, and the release of MSC-EVs is a natural physiological process. Substances in MSC-EVs are also physiological body constituents. There is compelling evidence of a good safety profile of allogeneic MSCs in animal studies [140,141] and clinical trials [142,143]. Derived EVs are, therefore, expected to be safe. There is no evidence that allogeneic EVs cause any adverse events after transplantation in immunocompetent animals [144,145] (Table 2). EVs derived from ADSCs and BMSCs have been found to be non-immunogenic in an in vitro immunogenicity assay [109]. The transplantation of MSC-EVs, if derived from the same origin of target tissue, would further lower safety concerns. Recent clinical reports on the application of allogeneic MSC-EVs for the treatment of a human graft-versus-host disease patient [146], chronic kidney disease patients [147] and patients with COVID-19 [29] showed no adverse side effects, supporting their general safety.

Efficacy
The availability of animal data demonstrating the efficacy of MSC-EVs in tendon and ligament repair is crucial for subsequent clinical trials. MSC-EVs isolated from different tissues have been shown to promote healing of acute tendon injuries, acute ligament injuries, a failed healing model of tendinopathy and tendon-to-bone junction repair (Table 2). Further research is needed due to the different tissue origins, species, dose and route of administration of MSC-EVs, as well as the different animal models used in previous studies. Data on the dose-frequency response relationship of MSC-EVs are needed to determine the optimal dose for the treatment of specific tendon or ligament injuries. There is no consensus on the best normalization strategy for EV dose (e.g., number of cells or tissue mass), but a rationale should be provided, according to MISEV2018 [86]. As discussed, local injections of MSC-EVs appear to be more suitable for the treatment of local tendon and ligament injuries, and may reduce the dose and manufacturing cost of MSC-EVs in future clinical trials. One pre-clinical study reported systematic intravenous injection of 200 µg protein of BMSC-EVs weekly for the promotion of rotator cuff repair in a rat model [52]. The dose used was high and prohibited future clinical application. Figure 1 outlines the manufacturing considerations and QC program for the development of GMP-grade MSC-EVs.

Conclusions
In recent years, there has been a boom in research on the effects of MSC-EVs for the treatment of various tendon and ligament injuries. However, none of this research has reached the clinical phase yet. Several key technical challenges need to be overcome for the clinical application of MSC-EVs. First, the biochemical composition of MSC-EVs remains unclear. The production or uptake mechanisms are poorly described, and GMP standards for clinical grade production, storage and recovery are lacking. Second, the drug loading efficiency of EVs is relatively lower than that of liposomes. Third, engineered MSC-EVs containing transgene products are categorized as ATPs, and hence, need to meet more stringent regulatory requirements. Fourth, as the half-life of MSC-EVs is short and they cannot replicate themselves like MSCs, large amounts of MSC-EVs are required for clinical application. The large scale and efficient production of MSC-EVs remains difficult at present. Finally, the physiochemical properties, particle size and concentration, as well as the MSC-EV content, have been the focus of numerous investigations. The MoA or potency of MSC-EVs remains relatively unexplored, resulting in a lack of appropriate functional assays. To overcome the bottleneck for MSC-EV-based therapies, clinical-grade MSCs meeting the requirements for therapeutic use and transplantation are needed for MSC-EV production. Conditions to ensure the sustainable release of MSC-EVs should be established. The specifications of MSC-EVs vary according to the enrichment method. Therefore, it is important to choose a cost-effective, scalable and automatable concentration method as early as possible to avoid changing the method and, hence, MSC-EV specification due to a subsequent scaling up of production. While there is some consensus regarding the transportation and storage of MSC-EVs at −80 • C, this poses challenges and is a costineffective approach. Alternative methods, such as lyophilization, may improve MSC-EV stability during storage and transportation. The route of administration greatly influences the therapeutic efficacy and safety of MSC-EVs and should be determined prior to clinical application. Efforts to develop high-throughput and precise quantification methods for assessing MoA or potency will greatly facilitate the clinical translation of MSC-EV-based therapies. The product is the process. Any slight change in the MSC-EV production process, formulation or storage conditions can influence the composition and biological activity of the finished product. It is, hence, important to build in scalability, reproducibility and GMP concepts, along with QC measures and release criteria to MSC-EV-based products from the time of deciding to manufacture them into pharmaceutical products, or even earlier, at the pre-clinical stage. Researchers have to consider practical issues associated with manufacturing procedures and QC measures, as described in this review, in order to translate the pre-clinical results of MSC-EVs into clinical practice for the treatment of tendon and ligament injuries.