Engineering Therapeutical Extracellular Vesicles for Clinical Translation

Abstract

Cell-based therapies are revolutionizing medicine by replacing or modifying dysfunctional cells with healthy cells or engineered derivatives, offering disease reversal and cure. One promising approach is using cell-derived extracellular vesicles (EVs), which offer therapeutic benefits similar to cell transplants without the biosafety risks. Although EV applications face challenges like limited production, inadequate therapeutic loading, and poor targeting efficiency, recent advances in bioengineering have enhanced their effectiveness. Herein, we summarize technological breakthroughs in EV bioengineering over the past 5 years, highlighting their improved therapeutic functionalities and potential clinical prospects. We also discuss biomanufacturing processes, regulation, and safety considerations for bioengineered EV therapies, emphasizing the significance of establishing robust frameworks to ensure translation capability, safety, and therapeutic effectiveness for successful clinical adoption.

Extracellular Vesicles: A New Dawn in Therapy

Extracellular vesicles (EVs) are small cargo-bearing vesicles secreted by cells into the extracellular matrix. Research on EVs has grown exponentially over the past decade (Figure 1A) with the realization that EVs are not merely cellular waste disposal systems or simple mechanisms for cell-to-cell communication [1]; indeed, EVs have been shown to shuttle functional cargos from cells and can have diagnostic and therapeutic functions [2, 3]. Preclinical and early clinical studies have capitalized on these features, especially in biomedical fields such as vaccines, diagnosis and disease treatment (Figure 1B) [4-8]. Noteworthy clinical trials from 2017 to date involving native and bioengineered EVs are listed in Table 1.

Figure 1. Milestones of bioengineered EV development and potential applications.

(A) Milestones in developing bioengineered EVs and their clinical applications span three phases: early discovery (pre-1980s), identification and exploration (1980-2014), and clinical applications (2014-present). Early studies laid the groundwork for understanding EVs [111, 112]. The 1980s brought detailed exploration and provided compelling evidence of intraluminal vesicles' release from cells. Johnstone and coworkers’ work further defined exosomes, marking a new era of EV research [113, 114]. Inserted figure: i. Illustration of multivesicular endosome within a cell; ii. an adherent membrane vesicle with associated labeling particles. Both figures represent EV exocytosis process [114, 115]. (Bar: 100 nm). From 1990 to 2000, EVs' biological functions were unveiled, highlighting their roles beyond cellular debris [116, 117]. In 2005, the first-in-human clinical trials using autologous dendritic cell-derived EVs for cancer treatment signify EVs' transition from laboratory curiosities to therapeutic agents [118, 119]. Inserted figure: Workflow of the clinical study on the vaccination of metastatic melanoma [118]. Between 2006 and 2008, the discovery of EVs' ability to transfer nucleic acids emphasized their importance in intercellular communication [120-122]. The significance of vesicle-mediated transport in cellular processes was further recognized in 2013, when the Nobel Prize in Physiology or Medicine was awarded for research in this area. The introduction of the Minimal Information for Studies of Extracellular Vesicles (MISEV) guideline in 2014 standardized EV research for bioengineering and clinical applications [123]. Post-2010, EVs emerged as tools for disease diagnosis, notably in cancer and Parkinson's disease, with several ongoing clinical trials. In 2017, EVs were first bioengineered for CRISPR/Cas9 gene editing delivery [124]. Next year, the second MISEV guidelines (MISEV2018) was released, offering updated standards for the evolving field of EV research [9]. The COVID-19 pandemic in 2020 accelerated the clinical application of EVs for diagnostics and potential therapeutics. In the same year, CODIAK Biosciences initiated groundbreaking first-in-human trials for bioengineered EVs. Preclinical studies post-2020 focused more on bioengineering EVs to load large nucleic acids, enhancing gene therapy potential [32, 60, 61, 70]. The latest MISEV2023 guidelines, further reflecting advances in EV characterization, underscore the commitment to enhancing their therapeutic capabilities [10]. (B) Applications of EVs in preclinical and early clinical studies across a broad spectrum of biomedical fields, including vaccines, diagnostics, and the treatment of diseases affecting the nervous, cardiovascular, respiratory, gastrointestinal, hepatic, pancreatic, skeletal systems, as well as skin and muscle disorders.

Table 1.

Selected clinical trials using native and bioengineered EVs, 2017-2023 ^a

EV Types	Program or Drug Code	Year	Study Phase (Status)	EV Sources	Therapeutical Cargos	Administration	Applications and Targeted Diseases	Sponsor Institution/Country	Reference (NCT Identifier)
Native EVs	ExoDx Prostate (IntelliScore)	2017/2021/2022	Observational study (Partially completed, still active)	Urine	Native	N.A.	Diagnostic test for male patients with prostate cancer	Exosome Diagnostics, Inc./United States	[125, 126] (NCT04720599, NCT05572099, NCT03031418, NCT03235687)
	AGLE-102	2019	Phase I/II(Active)	Allogeneic BM-MSCs	Native	On wounds	Treatment of patients with deep second degree burns of the skin and lesions in subjects with EB	Aegle Therapeutics/United States	NCT04173650, NCT05078385
	CBMG-AD-01	2020	Phase I/II (Completed)	Allogenic ADMSCs	Native	Nasal drip	Treatment of mild to moderate dementia due to Alzheimer's disease.	Ruijin Hospital/China	[127] (NCT04388982)
	ExoFlo	2020	Phase II (Completed)/Phase III (Recruiting)	BM-MSCs	Native	Intravenous	Patients with ARDS and severe COVID-19	Direct Biologics, LLC/United States	[128] (NCT04493242, NCT05354141)
	MEXVT/MEX COVID	2020	Phase I (Completed)	Allogenic ADMSCs	Native	Aerosol inhalation	A tolerance clinical study in healthy volunteers and treatment of patients with severe COVID-19 pneumonia	Ruijin Hospital/China	[129, 130] (NCT04313647, NCT04276987)
	ExoOA-1	2021	Phase I (Not yet recruiting)	Allogeneic MSCs	Native	Intra-articular	Patients with mild to moderate symptomatic osteoarthritis	University of the Andes/Chile	NCT05060107
	ExoVerita™	2022	Observational study (Recruiting)	Blood	Native	N.A.	Early detection of PDAC	Biological Dynamics/United States	[131, 132] (NCT05625529)
	GD-iExo	2023	Early Phase I (Recruiting)	IPSCs	Native	Nasal drip or on skin	Treatment of refractory focal epilepsy and atopic dermatitis	Peking Union Medical College Hospital/China	NCT05886205, NCT05969717
Bioengineered-EVs	BCC-GI-10 Curcumin	2011/2021	Early trial (Completed)/Phase I (Recruiting)	Ginger	Curcumin	Oral	Treatment of IBD and colon cancer	University of Louisville/United States	NCT01294072, NCT04879810
	N.A.	2017	Phase I/II (Unknown)	MSCs	miR-124	Intravenous	Treatment of disable patients with acute ischemic stroke	Isfahan University of Medical Sciences/Iran	[133] (NCT03384433)
	iExosomes	2018	Phase I (Active)	MSCs	KrasG12D siRNA	Intratumoral	Treatment of metastatic pancreas cancer harboring KRASG12D Mutation	M.D. Anderson Cancer Center/United States	NCT03608631
	ENDFH	2021	Phase I (Not yet recruiting)	Normal donor BM-MSCs	LDLR mRNA	Abdominal puncture under ultrasound guidance	Treatment of HoFH	Tang-Du Hospital/China	NCT05043181
	EXO-CD24/Coven D24	2021/2023	Phase II (Active and recruiting)	T-REx™-293	CD24 overexpression	Inhalation	Prevent clinical deterioration in patients with ARDS and with COVID-19 infection	Tel-Aviv Sourasky Medical Center/Israel, Athens Medical Society/Greece	[134] (NCT05947747, NCT04747574, NCT04969172, NCT04902183)
	CDK-002 (exoSTING)	2020	Phase I/II (Completed)	Undisclosed	STING agonist	Intratumoral	Patients with advanced/met astatic, recurrent, injectable solid tumors	Codiak Biosciences/United States	NCT04592484
	CDK-003	2021	Phase I/IIa (Terminated)		IL-12	Intralesional	Patients with CTCL		NCT05156229
	CDK-004 (exoASO-STAT6)	2022	Phase I (Terminated)		ASO	Intravenous	Patients with advanced HCC, liver metastases from either primary gastric cancer or CRC		NCT05375604

^aAbbreviations: BM-MSCs, bone marrow-derived mesenchymal stem cells; ARDS, acute respiratory distress syndrome; COVID-19, coronavirus disease 2019; ADMSCs, adipose mesenchymal stem cells; ADSCs, adipose-derived stem cells; iPSCs, induced pluripotent stem cells; EB, epidermolysis bullosa; MSCs, mesenchymal stem cells; PDAC, pancreatic ductal adenocarcinoma; HoFH, homozygous familial hypercholesterolaemia; LDLR, low-density lipoprotein receptor; IBD, inflammatory bowel disease; STING, stimulator of interferon genes; HNSCC, squamous cell carcinoma of the head and neck; TNBC, triple negative breast cancer; ATC, anaplastic thyroid carcinoma; cSCC, cutaneous squamous cell carcinoma; IL-12, interleukin 12; CTCL, cutaneous T-cell lymphoma; ASO, antisense oligonucleotide; HCC, hepatocellular carcinoma; CRC, colorectal cancer.

EVs are heterogeneous vesicles that vary in size from nano to microscale. According to the guidelines from the International Society for Extracellular Vesicles (ISEV), they can be grouped by size as medium/large (>200 nm in diameter) or small (<200 nm) and further classified into various subpopulations based on their biogenesis, sources, or functions [9, 10]. Notably, EVs often share overlapping physicochemical properties, biological functions, and origins, which reveal their complexity and potential barriers for their practical application [11]. Consequently, a thorough understanding of these attributes is crucial to unlock the full therapeutic potential of EVs and to harness them as biotherapeutics. (Box 1).

Box 1. Pathways to EV Biogenesis: Unlocking Therapeutic Potential and Bioengineering Horizons.

EVs are released from cells primarily via exocytosis of the plasma membrane, forming through mechanisms such as the trans-Golgi network or plasma membrane invagination, which create early endosomes. These early endosomes mature into late endosomes, generating intraluminal vesicles within multivesicular endosomes. These multivesicular endosomes can then either release the intraluminal vesicles into the extracellular space as exosomes or become degraded by fusing with lysosomes/autophagosomes. Compared with exosomes, microvesicles or ectosomes are simpler structures formed through direct outward budding and fission of the plasma membrane, which resembles a reverse version of endocytosis.

Selective Cargo-sorting Mechanisms

The capacity of EVs to transport a wide range of biomolecules for therapeutic purposes, along with their selective cargo-sorting mechanisms, has been extensively researched and reviewed [101]. These selective sorting processes are intricate, involving several key proteins, particularly the 'endosomal-sorting complex required for transport' (ESCRT), in which components included in the vesicle bind specifically to ubiquitylated transmembrane proteins and non-ubiquitination proteins on the endosome membrane, creating patches of membrane with high cargo concentrations that are then incorporated into budding EVs. Two other ESCRT-independent mechanisms, that is, lipid-dependent pathways (e.g., ALIX–Syntenin1–Syndecan1 machinery, ceramide transporter) and tetraspanins in the EV membrane, are also important in cargo sorting, as does the post-translational modification of cargo molecules [102-105]. The enrichment of RNA within EVs is another crucial critical factor that contributes to their therapeutic functionality and is significantly governed by RNA-binding proteins. miRNAs have been extensively investigated as RNA cargo within EVs, and numerous binding motifs (e.g., hnRNPA2B1 [GGAG, AGG or UAG]) have been identified as key regulators in loading various miRNAs [101, 106]. Conversely, the presence of full-length exosomal mRNAs in EVs is still low and their role in trafficking is not well understood. Clarifying the various cargo-sorting mechanisms is crucial for facilitating the design of strategies for functionalizing EVs.

Potential pathways to enhance EVs biogenesis

Aside from EV potential as cargo carriers, exploring the upstream and downstream pathways interacting with its biogenesis can advance the mass production of functional EVs. After maturation, exosome precursors within late endosomes/multivesicular endosomes face two divergent pathways: fusing with lysosomes or docking to plasma membranes. Because fusion of late/endosomal/multivesicular endosomes/lysosomes leads to their degradation, one strategy to increase EV secretion is by inhibiting lysosomes or endo-lysosomal fusion with alkaline agents [107, 108]. On the other hand, exosomes can reside within lysosomes, where they are protected from degradation and subsequently released through lysosomal exocytosis [109]. Despite the contradictions in these findings and the need for further exploration, they highlight the endo-lysosomal pathway as a viable mechanism for regulating the production of therapeutic EVs in donor cells, offering potential for EV bioengineering. Another pathway that interacts with EV biogenesis is autophagy, which is thought to augment the lysosomal degradation of EVs [109]. Intriguingly, the cellular response to external stimuli such as growth factors, nutrients, and stress can affect this dynamic. For example, treatment with rapamycin or deprivation of nutrients and growth factors can stimulate exosome release via changing cellular metabolic pathways, particularly through the mechanistic target of rapamycin complex (mTORC) [110]. In another example, electrical field stimulation was reported to activate mTORC1-autophagy in donor cells, leading to the production of more functional EVs carrying full-length mRNA transcripts [66]. Overall, better understanding of these pathways and their interactions with EV biogenesis and cargo-sorting mechanisms is essential for unlocking the full therapeutic potential of EVs. Such an understanding also provides a theoretical foundation for developing more effective bioengineering EV strategies.

Compared with traditional synthetic or viral carriers, EVs are natural origin, and present low immunogenicity, high biocompatibility, and the unique ability to transport a wide range of native nucleic acids and proteins from their parental cells [6, 12]. Despite these advantages, the practical deployment of EVs has faced significant challenges, notably in terms of limited production yields and suboptimal loading efficiency with desired bioagents. To overcome these limitations, exogenous EV modification and endogenous cell bioengineering have been developed to boost EV yields and enable the incorporation of therapeutic factors to augment their functional capabilities. Unfortunately, these bioengineering processes introduce additional challenges in EV bioproduction that extend beyond natural batch-to-batch variations, and further include the potential for byproducts and side effects associated with the bioengineering procedures. These and other challenges highlight the need for advanced strategies and the establishment of robust production standards and clear definitions in EV bioengineering to optimize their therapeutic efficacy while minimizing potential complications, thereby ensuring their safe and effective use.

Bioengineering Strategies for Mass-producing Therapeutic EVs

Various bioengineering strategies have been developed to scale up EV production and efficiently incorporate substantial quantities of desired cargos. These strategies generally fall into two categories: exogenous EV modification and endogenous cell bioengineering (Figure 2). Each approach comes with its advantages and challenges, and potential for scaling up EV bioproduction depends on several factors, including the type of biocomponent encapsulated, EV yield, modification efficiency and the overall scalability of the process (Table 2). Examples of EVs produced using bioengineering strategies during the past 5 years are given in Table 3.

Figure 2. A schematic of EV bioengineering strategies that is categorized into two groups: exogenous EV modification and endogenous cell bioengineering.

The exogenous EV modification is a straightforward method to augment cargo loading or modify the surface properties post EV isolation. EV surface modification mainly leverages the EV membrane’s properties, such as hydrophobicity, functional groups, and surface proteins, enabling non-covalent and covalent binding of specific moieties to the EV surface to facilitate targeting, internalization, and cargo-carrying capacity. Means of perturbing the EV membranes to carry the cargo of interest in the EV lumen include passive approaches like freeze-thaw cycles and surfactant treatments, as well as active methods like electroporation and sonication. EV membranes can also be used to encapsulate synthetic cargo-carrying monomers or nanoparticles, or alternatively, EVs can be engineered to fuse with lipid-based structures that carry or graft specific cargos. Although pre-isolation and post-EV modification approaches are relatively simple, they often suffer from reduced yields, potential contamination risks, and limited engineering efficiency. Endogenous cell bioengineering, by contrast, offers a promising alternative for the mass production of artificial EVs. This strategy involves initially manipulating parental cells through environmental variations, including hypoxia nutrient deprivation, pH variation, dimensionality, and metabolic regulation, can significantly boost EV yield. Compared to this, applying physical and biochemical stimuli can lead to a more substantial increase in both the quantity of EVs produced and the efficiency of cargo encapsulation. Genetical manipulation, particularly through transfection methods, enables the stable encapsulation of genetic cargos, utilizing both random and active capture mechanisms.

Table 2.

Comparison of exogenous EV modification and endogenous cell bioengineering on cargos types, EV yield, modification efficiency, scalability, and the associated challenges currently faced ^a,b

Bioengineering Strategies		Cargos Types	EV Yield	Modification Efficiency	Scalability	Main Challenges in Bioproduction
Exogenous EV modification	Direct EV surface functionalization	Small nucleic acids and compounds	+	+	++	Cargo stability, loading efficiency
	EV membrane perturbation and cargo encapsulation	Compounds, moderate-sized proteins and nucleic acids	+	+++	++	EV Recovery, contamination
	Hybridizing EVs with other nanoplatforms	Lipophilic moieties, nanoparticles with cargos loaded	+	++	++	EV Recovery, cargo leakage
Endogenous cell bioengineering	Environmental factors	Endogenous alternation of proteins and nucleic acids	++	+	+++	Need further engineering steps
	Physical and biochemical stimulation	Compounds, nanoparticles, proteins, nucleic acids	+++	++	+	Contamination, cost-effectiveness
	Genetical manipulation	Proteins, nucleic acids	++	++	++	Time cost, contamination, limited option of donor parental cells

^aCargo types, main biocomponents involved in EV bioengineering; EV yield, quantity of EVs after bioengineering; Modification efficiency; and payloads; Scalability, feasibility to scale-up regarding biomanufacturing time and cost.

^bArbitrary units listed as + (low), ++ (intermediate), +++ (high).

Table 3.

Representative cases of bioengineered EVs utilizing exogenous EV modification and endogenous cell bioengineering for various applications since 2018 ^a

Bioengineering Strategies		Donor Cell	Bioengineering Principle	Introduced Biological Components	Modification Efficiency	Applications	Reference
Exogenous EV Modification	Direct EV surface functionalization	Murine C2C12	Non-covalent binding through CD63	Exosomal anchor peptide (CP05), PMO	Exosome/CP05 binding efficiency was up to 88.7%; 108.98 ± 7.82 μg exosome per milliliter of serum when using CP05 dynabeads.	Targeted delivery and functional improvement for DMD	[21]
		PANC-1, B16-F10 and HEK293	Copper-free click chemistry	AF488 dye	Dye molecule per exosome is 6.4 ± 1.2, 9.5 ± 0.8 and 6.1 ± 1.4 in PANC-1, B16-F10 and HEK293, respectively	Systematic investigation of the uptake efficiency of exosomes from various sources by pancreatic cancer cells	[20]
		HEK293T	Thiol–maleimide reaction for targeting modification; electroporation for drug loading	E3-Aptamer, SIRT6 siRNA	No quantified results. Efficiency was determined by the downstream signaling of recipient cells.	Targeted delivery of siRNA-exosomes for treatment of metastatic castration-resistant prostate cancer	[135]
		HepG2	EDC/NHS-mediated amide reaction	Polyarginine peptide R9, ASO G3139	~3000 R9 polypeptides and 3.1 × 104 G3139-cholesterol per exosome	Improved intracellular delivery of nucleic acids for potential anti-cancer therapy	[136]
		M2 phenotype microglia	Copper-free click chemistry	Vascular targeting peptide (DA7R), SDF-1	1 mg/mL modified EVs were conjugated with about 45.4 μg/mL DA7R and 9.46 μg/mL SDF-1	Improved NSCs performance (recruiting and differentiation) for treatment of ischemic stroke	[18]
	EV membrane perturbation and cargos encapsulation	ESCs	Co-incubation for drug therapeutic loading; postinsertion for targeting	PTX, c(RGDyK)	32 μg PTX per 1010 particles	Targeted delivery of chemotherapeutics for glioblastoma treatment	[137]
		MDA-MB-231 and HT29	Modified freeze-thawing process consisting of additional incubation and ultrasonication steps	Nanoamorphous aspirin	No quantified results. Efficacy was evaluated by the uptake efficiency and therapeutic effect	Efficient delivery of aspirin for anticancer treatment through enhanced apoptosis and autophagy	[138]
		Macrophages (IC21)	Sonication and saponin permeabilization	Lysosomal enzyme TPP1	Around 70 μg TPP1 and 50 μg TPP1 in 1011 EVs loaded by sonication and saponin permeabilization, respectively	Enhanced delivery of active enzymes across the BBB for the treatment of LSDs	[28]
		HSCs (LX-2)	Electroporation	Cas9 RNP complexes	Encapsulation efficiency of exosomal Cas 9 RNP complexes is about 20% calculated by Western blot analysis	Delivery of Cas9 RNP complexes for treatment of liver disorders	[31]
		HEK293T	Electroporation	IL-12-encoding mRNA	Encapsulation efficiency level for IL-12 mRNA in exosome is 27.6%	Inhalation of exosome/IL-12 mRNA promoted IFNγ-mediated immune activation, thereby enhancing anti-lung tumor efficacy.	[32]
	Hybridizing EVs with other nanoplatforms	MEFs	Freeze-thaw cycles for hybrid; endogenous genetic modifications for enhanced protein cargo expression	Thermosensitiv e liposomes, GM-CSF, DTX, CD47 overexpression	Fusion efficiency is 95.7%; GM-CSF and DTX are 7.2 pg μg−1 and 4.3% in the hybrid nanoparticles	Combination of locoregional delivery of HIPEC and systemic delivery of chemoimmunotherap y for the treatment of metastatic peritoneal carcinoma	[139]
		RAW264.7	Co-incubation at 37°C	10B carbon dots	Drug loading and encapsulation efficiency are 17.90 ± 0.27% and 85.24 ± 1.12%	In-situ precise boron neutron capture therapy for glioma treatment	[39]
		HUVECs, MSCs, MDCKCs	PEG fusion	Liposomes, antitumor photosensitizer mTHPC, rhodamine	Fusion rate was determined by fluorescence. Maximal encapsulation efficiency of mTHPC at ~90% and rhodamine at 43%.	Enhanced loading and cellular delivery efficiency for both lipophilic and hydrophilic compounds	[140]
		MIA-PaCa-2	Sonication	Chlorin e6 photosensitizer	No quantified results. Encapsulation was confirmed by fluorescence and absorbance spectrum.	Photoacoustic imaging-guided photodynamic and immune-combination therapy for the treatment of cancers	[141]
		LSCs	Insertion of DSPE-PEG-cargo	Recombinant SARS-CoV-2 RBD	0.52 μg RBD per 1010 exosomes	Inhalable COVID-19 vaccine attenuated severe pneumonia and reduced inflammatory infiltration	[142]
Endogenous cell bioengineering	Environmental factors	MEFs	Metabolic regulation (rapamycin, amino acids) and serum starvation through mTOR pathway	Not specific	Exosome secretion increased by 1.5 to 3-fold according to in vitro and in vivo analyses.	Regulation of cellular metabolic activities to stimulate exosome production	[110]
		BMSCs		Endogenous miRNAs (specific on miR-126)	Exosome release was enhanced by approximately 1.5-fold, while miR-126 expression increased 15-fold according to sequencing data.	Promoted proliferation, angiogenesis and migration in HUVECs for bone regeneration	[143]
		ADSCs		Endogenous circRNA (specific on circ-Snhg11)	Circ-Snhg11 exhibited a roughly 5-fold increase as indicated by sequencing data.	Promoted angiogenesis for diabetes wound healing	[144]
		MSCs	3D culture based on hollow fiber bioreactor	Not specific	The yield of 3D-EVs was about 8.21-fold greater than that of 2D-EVs.	Enhanced cardioprotective effect for heart repair	[145]
	Physical and biochemical stimulation	Human hepatocarcinom a (Bel7402 cells)	Nanoparticles incubation	DOX-loaded PSiNPs	Drug loading was 300 ng DOX μg−1 exosome protein; loading efficiency was 0.8% determined by HPLC.	Exosome-sheathed nanoparticles for targeted cancer chemotherapy	[56]
		Tissues seeded with DPSCs and MSCs	Flow stimulation or cyclic stretching	Not specific	The yield of 3D dynamic EVs in DPSC and MSCs was approximately 150-fold higher and 40.7-fold higher, respectively, than that of 2D production	Boost EV production	[51]
		MEFs and BMDCs	Porous silicone chip-based nanoporation	PTEN mRNA, modified CD47 protein overexpression	50-fold more exosomes production and over 1000-fold increase in exosomal mRNA transcripts than BEP	Targeted delivery of large nucleic acids for treatment of glioma	[60]
		nHDFs		COL1A1 mRNA	10-fold higher EV number per cell and 200-fold higher COL1A1 mRNA when compared to BEP	Protein-replacement therapy for the treatment of photoaged skin	[61]
		MEF and BMSCs	Porous polymeric membrane-based nanoporation	TP53 mRNA, siKRASG12D, modified CD64 protein overexpression	Over 30-fold EVs, with around 4 copies of TP53 mRNA and over 1000 copies of siKRASG12D per EV	Targeted delivery of RNAs for the treatment of advanced pancreatic cancer	[67]
		ADSCs		BMP-2 and VEGFA mRNAs	10-fold higher EV number per cell, over 1000-fold (~1.3 copies/exosome) and over1500-fold (~1.8 copies/exosome) of BMP-2 and VEGFA mRNA	Enhanced angiogenic–osteogenic effects for bone regeneration	[66]
	Genetical methodology	HEK293T	Lentivirus transfection	Overexpression of modified CD9 proteins, dCas9 mRNA, miR-155	Over 8000-fold miR-155 expression when compared to control groups; 22.3 ± 8.5 copies of Cas9 were in 100 CD9-modified exosomes	Efficient loading of RNAs for gene therapy	[71]
		HEK293T	Lipofectamine transfection	Overexpression of CD2-scFv protein, Cas9–single-guide-RNA complexes	No quantified results. Encapsulation was confirmed by fluorescence	Targeted delivery of gene editing Biologics to T cells	[75]
		HEK293, human monocytes (Thp1 cells), BMDCs	Lipofectamine or PEI transfection	Arc protein capsid, reporter mRNAs	No quantified results. Encapsulation was confirmed by fluorescence of recipient cells	Targeted delivery of mRNA to neurons for neurodegenerative diseases	[76]

^aAbbreviations: PMO, phosphorodiamidate morpholino oligomer; DMD, Duchenne muscular dystrophy; HEK293T, human embryonic kidney 293T; SIRT6, Sirtuin 6; ASO, antisense oligonucleotide; SDF-1, stromal cell-derived factor 1; NSCs, neural stem cells; EDC/NHS, 1-ethyl-3-(3-dimethylaminopropyl)carbodiimide/ N-hydroxysuccinimide; AF488, AlexaFluor^®488; HEK293, human embryonic kidney 293; ESCs, embryonic stem cells; PTX, paclitaxel; c(RGDyK), Cyclo (Arg-Gly-Asp-D-Tyr-Lys); TPP1, tripeptidyl peptidase 1; BBB, blood-brain barrier; LSDs, lysosomal storage disorders; Cas9 RNP, clustered regularly interspaced short palindromic repeats (CRISPR)/CRISPR-associated protein 9 ribonucleoprotein; HSCs, hepatic stellate cells; IL-12, interleukin 12; IFNγ, interferon gamma; LSCs, lung spheroid cells; RBD, receptor-binding domain; COVID-19, coronavirus disease 2019; MEFs, mouse embryonic fibroblasts; GM-CSF, granulocyte-macrophage colony-stimulating factor; DTX, docetaxel; HIPEC, hyperthermic intraperitoneal chemotherapy; HUVECs, human umbilical vein endothelial cells; MSCs, mesenchymal stem cells; MDCKCs, Madin-Darby canine kidney cells; mTHPC, meta-tetra(hydroxyphenyl)chlorin; BMSCs, bone marrow-derived stem cells; ADSCs, adipose-derived stem cells; HUVECs, human umbilical vein endothelial cells; mTOR, mammalian target of rapamycin; DPSCs, dental pulp stem cells; BMDCs, bone marrow-derived dendritic cells; PTEN, phosphatase and tensin homolog; BEP, bulk electroporation; nHDFs, Neonatal human dermal fibroblasts; COL1A1, collagen type I alpha 1; TP 53, tumor protein P53; BMP-2, bone morphogenetic protein-2; VEGFA, vascular endothelial growth factor A; DOX, doxorubicin; PSiNPs, porous silicon nanoparticles; HPLC, high performance liquid chromatography; PEI, polyethylenimine.

Exogenous EV Modification

Direct EV Surface Functionalization.

The native EVs are rich in membrane lipids, proteins, and glycans that contribute to their preferential interaction with target cells, but often face in vivo inefficiencies due to rapid clearance (e.g., through phagocytosis) and nonspecific uptake, notably by the liver and spleen [13, 14]. Recent advances in engineering EV surfaces post-isolation include adding targeting moieties and fluorescent dyes to improve their specificity, internalization, and traceability. Non-covalent approaches, such as multivalent electrostatic interactions and protein-ligand recognition, often rely on the EVs' negative charge and surface proteins for attachment of various biomolecules, but they raise concerns about stability and specificity loss in circulation [15-17]. Given these issues, covalent strategies offer more durable solutions by utilizing the EV membrane's functional groups for more stable and effective modification, with copper-free azide-alkyne cycloaddition (click chemistry) being preferred for its specificity and mild reaction conditions that minimize potential EV disruption or aggregation caused by issues related to buffer solutions, osmotic stress, and variation in reaction temperatures [18-20].

Beyond improving targeting and tracking, EV membrane functionalization is also utilized for carting extramembrane contexts for therapeutic purposes, particularly small compounds and nucleic acids [21, 22]. While readily accessible, positioning these cargo molecules on the EV surface raises concerns due to the instability of exposed therapeutic cargos during circulation and endocytosis; luminal loading, in which cargos are encapsulated within the EVs, is generally more preferred. Moreover, challenges include potential damage to the vesicles, residues of byproducts and intermediates, and unknowns in the recovery rate of EVs after a series of surface modifications, all of which present barriers to its clinical application. The development of mild click chemistry has mitigated some of these concerns but needs systematic investigation for optimized scale-up [18]. More importantly, limitation in reactive site availability on EVs leads to suboptimal grafting and loading efficiency, prompting exploration into genetical manipulation at the cellular level, particularly for surface protein enrichment as described below.

EV Membrane Perturbation and Cargo Encapsulation.

Like cell membranes, EV membranes consist of a lipid bilayer that can be permeable and temporarily porous under certain conditions, thereby allowing biomolecules to passively diffuse into the vesicle’s lumen. One of facile physical procedures enhancing EV membrane accessibility includes direct incubation for small compounds and nucleic acids, although encapsulation efficacy remains low due to passive loading, as well as uncertainties about cargo association with the EVs, particularly regarding luminal loading or membrane attachment [23]. Improvement strategies like rapid freeze-thaw cycles, extrusion or microfluidics, and the use of surfactants transiently destabilize membranes, enhancing loading efficiency for small molecules while making surfactants preferable for protein loading [24-27]. Active encapsulation methods like electroporation and sonication can further increase post-loading efficiency by actively inducing cargo diffused across porous membrane of the EVs [23, 28, 29]. One such method, electroporation, has proven highly efficient for loading nucleic acids and charged compounds, and has progressed to clinical trials [30].

Physical procedures or technologies used to perturb EV membranes can significantly improve cargo loading efficiency. However, the ability to encapsulate larger molecules, such as proteins and mRNA, remains limited and questionable, even with methods like saponin pretreatment or electroporation, which recently claim to accommodate large enzymes of >200 kDa or full-length synthetic mRNAs >2000 nt [31, 32]. The efficacy of these loading techniques often relies on aggressive membrane disruption, raising concerns about the recovery yield of the engineered EVs after treatment that remain to be addressed. Moreover, processes like surfactant treatment and electroporation introduce risks of contamination with residual surfactant or electroporation buffers, a concern that becomes more pronounced during scaling up for industrial production. These cargo-loading strategies may also compromise the integrity or bioactivity of desired biomolecules such as, for example, nucleic acid aggregation or degradation under high electrical fields or ultrasound frequency [33]. Despite these challenges, the maneuverability and facile rationales behind membrane-permeabilization techniques render them as leading candidates for future industrial adoption in EV bioengineering.

Hybridizing EVs with Other Nanoplatforms.

Structurally, EVs are comparable to lipid-based nanoparticles (LNPs), as both are based on phospholipids. Similarities between these two particle types has led to modification of EVs by inserting lipophilic moieties into their membranes and or by fusing them with LNPs to enhance molecule adsorption or encapsulation. Cholesterol and its derivatives are often preferred for membrane insertion owing to their lower phase-transition temperature, which makes their insertion easier while still maintaining the integrity of the EVs [16]. However, concerns remain regarding the bioactivity of post-inserted biomolecules, particularly the need for optimal temperature control during modification to ensure effective delivery, and the exposure of these biomolecules could reduce their in vivo stability, limiting the efficacy of targeted delivery. Unlike the direct insertion, modifying EVs with LNPs through fusion, particularly with liposomes, creates a semi-synthetic hybrid system. Among various fusion strategies, PEG-induced fusion is more favorable due to its invasive feature, time-independent fusion efficiency and low risk of cargo leakage [34-36]. These hybrids can significantly increase EV payload; nevertheless, differences in the mechanisms of endocytosis and endosomal escape between EVs and LNPs could alter the route of cargo delivery to the cytosol and potentially reduce the effectiveness of delivery, necessitating systematic investigation [37, 38]. In addition to EV hybrid LNP-based systems, EV also can be hybridized with other delivery systems, such as poly(lactic-co-glycolic acid), metal nanoparticles and inorganic nanoparticles [39-42]. Unlike LNP-based systems that rely on fusion, these hybrid systems are formed primarily by physical procedures (e.g., sonication, microfluidics and extrusion) that transiently disrupt the lipid layers of EVs and often form a “core–shell” structure that increases the particle size [43, 44]. Such a structure is comparable to that of cell membrane–coated mimics designed to cross specific barriers and evade detection by the immune system. Yet, challenges remain in EV recovery and the characterization of these hybrids due to their similarity to free nanoparticles.

The goal in designing EV-based hybrid systems is to merge the advantages of natural EVs with those of synthetic platforms to optimize the therapeutic cargo payload, immuno-evasive properties, and the ability to cross biological barriers [35]. Notably, integration with synthetic platforms mitigates EV heterogeneity and may clarify specific biological components, which may be beneficial in terms of regulatory compliance. However, challenges with fusion or encapsulation efficiency and potential leakage represent significant hurdles to clinical applications.

Endogenous Cell Bioengineering

Environmental Factors.

Simply changing the parental cell culture environment through hypoxia, nutrient deprivation, pH variation, culture dimensionality and metabolic regulation, can significantly boost EV yield. While manipulations of these that benefit scaled-up production and cost reduction do not basic EV characteristics, they may lead to changes in the molecular constituents like lipids, RNAs, and proteins of EVs [45]. These variations in vesicle composition, along with increased EV secretion, rely not only on the environmental stresses applied but on the type of parental cells used. Use of tumor cells, which are typically more aggressive than normal cells, can increase EV secretion rates by ~70 times when cells are also exposed to lower pH conditions, and these EVs may harbor higher numbers of oncogenic molecules (e.g., hypoxia-inducible factors and related miRNAs), which would be expected to influence cancer progression and thus present another potential avenue for developing anti-cancer therapy [46-48]. Stem-cell–derived EVs, especially when preconditioned by hypoxic conditions and grown in 3D cultures, are rich with angiogenic factors used for regenerative medicines, but their potential tumor-promoting activity raises safety concerns, necessitating further investigation and monitoring [49].

Overall, EV bioengineering through environmental regulation is straightforward and cost-effective, without the need to further purify EVs and without undermining their structural integrity, rendering it an appealing option for large-scale EV production. Despite its merits, this approach has limited cargo loading capacity, often requiring further exogenous modifications to enhance their therapeutic functions.

Physical and Biochemical Stimulation.

To further boost large-scale EV production, more potent forms of physical stimulation have been used, including microfluidics, heat shock, mechanical stressors, acoustic stimulation, and electroporation [45, 50]. These methods facilitate large-scale EV production; notably, mechanical stressors can enhance EV secretion yields by up to ~150-fold over traditional 2D culturing methods [51]. However, intense stimulation can have negative effects on cell behavior and lead to production of unwanted substances along with the secreted EVs, such as heat-shock proteins, stress granules, autophagy components, and inflammatory cytokines [52-55]. These byproducts pose safety concerns for bioengineered EVs, and thus the loading intensity must be balanced to meet the necessary threshold for optimal secretion while minimizing the risk of excessive cell stress and toxicity.

Further efforts in bioengineering extend to efficient loading of therapeutic cargos into the EV lumen. Biochemical stimulants, particularly chemotherapeutic agents, have been widely used to stimulate donor cells to facilitate their incorporation within the released EVs. However, biochemical stimulants such as these often result in inefficient payload delivery because of their variable endocytosis efficiency and the active involvement of endosome-lysosome degradation or autophagy pathways, both of which are heavily influenced by the physicochemical properties of the compounds, which ultimately affect their exocytosis in EV form [56]. Alternatively, donor cells can generate nanoparticle-induced EV hybrids through regulating exocytosis processes, which improve yield and facilitate their use in advanced therapies like photothermal therapy, radiotherapy, and other combinatory treatments [50, 56-58]. Additionally, physical transfection methods are being explored to boost the endogenous trafficking capacity for nucleic acids, although efficient incorporation of full-length mRNA remains a challenge [59]. Our group has developed a cellular-nanoporation technique that enables the large-scale production of bioengineered EVs, capable of achieving an over 1000-fold increase in endogenous mRNA transcription (including phosphatase and tensin homolog [PTEN], extracellular-matrix α1 type-I collagen [COL1A1], and Interferon [IFN]-γ), with EV yield increases peaking at roughly 50 fold [60-62]. Unfortunately, the complexity and high cost of these sophisticated device-based transfection technologies pose barriers to widespread industrial adoption [63-65]. In response, a more facile and commercially available porous membrane-based cellular electroporation system has been developed, offering similar enhancements in EV yield and mRNA loading while improving the cost-effectiveness and scalability for industrial purposes [66, 67].

While physical and biochemical stimulants can produce significant numbers of functional EVs, they also lead to extraneous substances from the donor cells, as well as contamination from the biochemicals used, which complicate purification and raise safety considerations [68]. Balancing effective production of bioengineered EVs with the purity and safety of the final product remains crucial in their bioengineering.

Genetical Manipulation.

To stably enhance the loading efficiency of EVs, genetical methodologies, particularly transfection are commonly used to modify donor cells to enrich EVs with large nucleic acids and protein drugs. This approach leverages the EV biogenesis pathways that govern drug accessibility within EVs, and facilitates desired cargo trafficking within EVs. Inspired by the sorting capacity of tetraspanins, strategies have harnessed their features to further boost the yield and capture of endogenous RNAs or proteins [69]. Unlike conventional approaches where surface proteins randomly capture cargos into EVs, co-transfection and engineered tetraspanins allow for selective cargo capture. Specifically, transfecting donor cells with CD63 plasmids, along with overexpression constructs for selected encoding candidates and cytosolic facilitators, can increase exosome production by more than 40-fold and enhance the encapsulation of mRNA within the EVs, respectively [70]. Further advancements have included the modification of other tetraspanins, particularly CD9, which has been modified with fusion proteins to specifically bind to small RNAs and large CRISPR/dCas9 mRNA [70-72]. Beyond these efforts, researchers have also explored the overexpression of other surface proteins on EVs, such as lysosome-associated membrane protein 2 isoforms and CD47 with peptide designs, to actively capture endogenous cargos and evade phagocytosis for enhanced distribution and treatment efficacy, respectively [73, 74]. Moreover, the development of single-chain fragment variable-based fusion proteins and the use of viral-like proteins have been pursued to further enable the efficient capture of large biomolecules in EV lumen, such cas9 ribonucleoproteins and mRNAs [75, 76].

Although genetic engineering approaches such as these offer promising avenues for EV production and large-size cargo loading, they also encounter significant challenges in bioproduction and regulatory compliance. The use of viral vectors restricts donor cell selection to those with high transfection efficiency and tolerance. Additionally, the process of cloning and selecting stable expression cell lines, along with verifying their EV functions, can be more time-consuming compared to other bioengineering strategies. These methods also raise concerns about potential contamination in EV preparations and the complexities of production scaling and regulatory compliance compared with conventional vesicle isolation from naïve cells or post-isolation engineering techniques [77].

Biomanufacturing and Clinical Translation of Bioengineered EVs

Native EV bioproduction and clinical translation stand to benefit from developments in classical biologics (i.e., antibodies and proteins) and cell therapy at two major manufacturing stages: upstream and downstream processes [78, 79]. Upstream processes include the selection of EV sources EV sources, expansion, and incubation and harvesting; downstream processes focus on the isolation and purification of EVs, their storage and characterization, quality assurance and control and ultimately, their administration. The bioengineering of EVs follows a path similar to the traditional workflow for native EV production but introduces additional manufacturing steps that bridge the upstream and downstream processes. These introduced bioengineering interventions require adjustments to the developed workflow and therefore deserve specific attention for bioengineered EV bioproduction (Figure 3). Below we note four key points regarding the distinctions in workflows between bioengineering of EVs and native EV production, with the goal of offering a well-rounded perspective on progress and obstacles in the field of bioengineered EV biomanufacturing toward clinical translation.

Figure 3. A comparison of the manufacturing process workflow between traditional native EVs and bioengineered EVs.

In native EV production, upstream processes include the selection of EV sources, expansion, and incubation and harvesting; downstream processes then focus on the isolation and purification of EVs, their storage and characterization, quality assurance and control and ultimately, their administration. The bioengineering of EVs introduces additional steps into the manufacturing process, where endogenous cell engineering is mainly at upstream flow, while exogenous cell modifications typically occur downstream. These bioengineering steps require adjustments to the developed workflow for native EV manufacture, therefore highlighting the need for specific considerations during the bioproduction of bioengineered EVs in future clinical translation.

Key Point 1: Parental Cell Source and Bioengineering

The selection of source cells significantly influences the upstream manufacturing process of EVs, with safety regarding immunogenicity and oncogenicity being paramount. Specific therapeutic attributes of native EVs (e.g., their tissue-homing ability, innate therapeutic biomolecules, and others), along with EV yields and host-cell impurities, are crucial for the selection of parent cells [80]. Current clinical trials primarily utilize autologous or allogeneic mesenchymal stem cells, induced pluripotent stem cell derivatives, or dendritic cells for native EV production (Table 1). Other promising sources under clinical investigation include HEK293 cells, and plant- or bovine milk-derived EVs because of their scalability and low costs, [81, 82], and bacterial EVs, particularly outer membrane vesicles from pathogenic bacteria, which are being explored for use in vaccines [83]. For bioengineered EVs, similar considerations apply in donor cell selection, especially the selection criteria for the cells tailored for exogenous modifications of EVs. However, the production of EVs through endogenous cell bioengineering entails further complexities, particularly when genetically engineering EV-producing cells by physical or chemical methods. For instance, primary immune cells are challenging to engineer due to their low transfection tolerance and inherent resistance [84]. Therefore, the robustness of donor cells is crucial for withstanding stimulations that could make them fragile and genetically unstable, emphasizing the need for appropriate donor cell selection tailored to the bioengineering process.

The most significant difference between conventional EV production and bioengineered EV production lies in their manufacturing processes. Conventional EV production uses methods like microfluidics and 3D culture to increase yields; production of bioengineered EVs can also involve these approaches but requires further processing for exogenous modifications. These modifications often involve straightforward techniques for encapsulating cargos within the lumen or anchoring moieties on the membrane (see more details in Exogenous EV Modification). However, the aspects, particularly recovery and stability following these modifications receive relatively little attention and are seldom addressed in current preclinical studies. For endogenous cell bioengineering, the focus shifts towards direct modification of parental cells, to potentially vary or extend the harvesting window of therapeutic EVs and increase the heterogeneity of EVs. These effects deserve further attention in the production of bioengineered EVs. Also, an enhanced focus on quality assurance and quality control for the donor cells after the bioengineering process is essential to monitor genetic drift and contamination risks. Cost efficiency remains crucial; although cutting-edge technologies have some advantages, their complexity and cost may limit their scalability. Consequently, more commercially viable and scalable engineering methods are often preferred for producing bioengineered EVs [63]. Overall, modification efficiency remains a focal point, with researchers striving to optimize both bioengineering approaches, yet there remains compared to synthetic platforms currently used in clinical settings, highlighting the need for ongoing development and evaluation.

Key Point 2: Isolation, Purification, and Shelf Life

The isolation and purification of EVs are critical downstream steps in their bioproduction and clinical application. Among current mainstream methods for EV isolation and purification, ultracentrifugation-based strategies and tangential flow filtration (TFF) have shown potential for scalable Good Manufacturing Practice (GMP) processes [85-87]. However, an optimal large-scale EV isolation method remains elusive due to concerns over EV integrity and quality compromise, with generally low purity levels reported [63, 88]. Moreover, the isolation process may selectively enrich specific EV subpopulations, which could affect their biological activity. For bioengineered EVs, a significant downstream challenge is the introduction of "extra" or unwanted substances (free therapeutic biomolecules and contaminants) during the bioengineering process. Most of the currently used purification techniques were originally developed for the purification of viruses and their related vectors; their standard operating procedures can be adapted for removing free drugs and contaminants from bioengineered EVs and for evaluating clearance [89]. However, the intrinsic heterogeneity of EV complicates therapeutic factor distribution across subpopulations, necessitating further characterization before isolation. For drugs encapsulated within the EV lumen, the subsets rich in therapeutic cargo encapsulation can usually be distinguished on the basis of size or density. In the case of surface-engineered EVs, preference is given to biochemical capture or affinity-based technologies, although concerns remain regarding their low throughput and challenges associated with dissociation, which may impair the stability of the EVs and potentially lead to off-target effects [90]. This diversity in EVs, together with the additional bioengineering steps needed, highlights the need for extensive multistep downstream processing, which may further reduce EV recovery and biological activity while increasing manufacturing costs.

Shelf life is another critical factor influencing both the quantity and quality of EVs. Standard storage at −80°C minimizes changes to native EV characteristics, while storage at 4°C can lead to degradation and aggregation [91]. These storage characteristics are largely applicable to bioengineered EVs as well, with most biomolecules retaining their stability at −80°C, although storage under these conditions has logistical and financial challenges. Lyophilization offers a potential solution for long-term storage, especially with the use of cryoprotectants like trehalose, which are promising to preserve bioengineered EV integrity and extend shelf life [92, 93]. Notably, moieties non-covalently bonded to the bioengineered EV membrane and sensitive luminal cargos (e.g., endogenous mRNAs) require further investigation because of their temperature sensitivity, particularly during freeze and thaw cycles [94].

Key Point 3: Characterization and Quality Assurance/Quality Control

Systematic characterization of EVs requires determination of both their identity and purity. Table 4 summarizes key updates of characteristics requiring evaluation when working with EVs as per the latest Minimal Information for Studies of Extracellular Vesicles (MISEV2023) guideline [10]. For native EVs, routine processes for characterization and quality assurance and quality control for clinical applications include assessment of: (1) parental cell properties (e.g., morphology, viability, and phenotype); (2) basic EV characteristics (e.g., quantification, size, zeta potential, and surface marker); (3) purity and potential contamination (e.g., mycoplasma detection); and (4) application-specific functional activities [95]. Measurement techniques employed in the routine characterization of EVs include flow cytometry, nanoparticle tracking analysis, resistive pulse sensing, multi-angle light scattering, and dynamic light scattering, while microscopy-based methods such as scanning electron microscopy, scanning-probe microscopy, and cryogenic electron microscopy are utilized to confirm the morphology of EVs [9]. In terms of the components within EVs, beyond traditional Western blotting for identifying EV surface markers, techniques such as nucleic acid profiling, mass spectrometry-based proteomics (e.g., liquid chromatography–mass spectrometry), and genetic protein tagging are increasingly used to detail the various constituents of EVs [10]. Recently, characterization technologies have shifted towards the analysis of individual vesicles, incorporating methods like single EV flow cytometry and total internal reflection fluorescence microscopy [10]. These advancements provide potential solutions for addressing the heterogeneity of EVs.

Table 4.

Summary of the latest updates of EV characterization in the MISEV guidelines (MISEV 2023) released in early 2024 versus previous versions ^a

Characterization Items		MISEV 2018	MISEV 2023
Basic EV characterization	Quantification	Emphasizes global quantification of EVs using at least two methods (e.g., protein amount, particle number, lipid amount).	Introduces additional methods for EV characterization including quantification of total RNA, alongside previously recommended protein and lipid quantifications. Emphasis on utilizing advanced technologies for more accurate quantification and characterization.
	Protein composition	Requires the assessment of transmembrane or GPI-anchored proteins, cytosolic proteins, and assessment of contaminants.	Continues the five-component framework introduced in previous MISEV versions for reporting protein content, with categories assessing EV features, purity, and potential intracellular origins. It encourages showing enrichment or depletion of markers versus unfractionated source material but refrains from limiting which EV proteins should be analyzed, suggesting use of databases like Uniprot for categorization.
	EV morphology	Encourages the use of SEM, TEM, and SPM (e.g., AFM) to assess EV morphology.	Expands on imaging techniques particularly high-resolution and cryo-based methods (e.g., cryo-EM) for better visualization of smaller EVs (≤200nm).
Advanced technology integration		N.A.	Extends the list of characterization techniques, highlighting the role of advanced methodologies, particularly flow cytometry-based methods (e.g., bead-based and single-EV flow cytometry), genetic protein tagging and mass spectrometry proteomics, microscopy-based methods and nucleic acid characterization, and Raman spectroscopy and resistive pulse sensing.
Functional activities		Highlights the importance of detailed reporting to ensure reproducibility, including the use of EV-TRACK for protocol registration.	Continues to stress detailed reporting for reproducibility but also introduces the concept of community feedback and updating protocols based on new evidence and techniques

^aAbbreviations: GPI, glycosylphosphatidylinositol; SEM, scanning electron microscopy; TEM, transmission electron microscopy; SPM, scanning-probe microscopy; AFM, atomic force microscopy; cryo-EM, cryogenic electron microscopy

Similar approaches for evaluating bioengineered EVs remain relatively unexplored and require further development. Significant challenges in assessing parent cell properties result from variations in bioengineering processes that can affect reproducibility between batches. Continuous monitoring and analysis of genetic alterations in EV-producing cells, as well as measurement of drug loading (or transfection) levels after cellular bioengineering, are essential for ensuring reproducibility and safety of bioengineered EVs. As discussed in Key Point 2, detecting contamination in bioengineered EV products presents additional complexities requiring more rigorous operating and monitoring procedures to remove free drugs and other contaminants introduced during bioengineering process. Moreover, accurately quantifying active substances within EVs and determining their therapeutic potency for specific applications deserve special attention. Evaluation of the potency of bioengineered EV may be aligned with functional activity of these introduced active substances toward their intended application. Currently, both native EVs and bioengineered EVs often undergo batch-by-batch testing, with different laboratories and companies using different assays and detection variables, including non-standardized timeframes for assays and inconsistent criteria for assessing therapeutic potency [34]. This variability highlights the urgent need for standardized assessments to ensure the consistency, reproducibility, and safety of bioengineered EV–based therapies.

Key Point 4: Regulation and Safety

As EVs continue to emerge as promising therapeutics and delivery platforms, a lack of consensus remains regarding regulatory compliance and safety. According to a review by the ISEV, native EVs can be classified as biological medicines or pharmaceutical class of biologicals and are considered non-active components necessary in the final formulation of a drug (termed as ‘excipients’) [96]. In this case, only the safety profile, not the mode of action, is typically requested for the native EVs. The scenario is more complex for bioengineered EVs, which can fall into three categories: (1) Bioengineered EVs modified exogenously with chemical drugs or encapsulation of molecular components. If EVs merely serve as a delivery system without explicit therapeutic functions, they may not be deemed part of the active substance. If therapeutic moieties are engineered on the EV membrane, the EVs could be considered a part of the active substance. (2) Bioengineered EVs derived from parental cells subjected to external stimuli or genetic modifications primarily to enhance EV production without involving the loading of active substances (e.g., transgene products), can be considered as biological medicines. (3) Bioengineered EVs derived from physically or genetically manipulated cells that contain active substances may be categorized as an independent subcategory of biologicals (e.g., advanced therapy medicinal products) depending on whether the therapeutic effect is explicitly attributed to the introduced products or the EVs themselves. Although this preliminary categorization for the regulation of bioengineered EVs for clinical translation is helpful, the reality is far more complicated, as the therapeutic function of EVs themselves cannot be readily excluded. Moreover, the bioengineering process may affect the biogenesis of EVs, leading to the generation of endogenous byproducts that can complicate the categorization of biologicals. Therefore, the ISEV community and global drug regulatory agencies are urged to establish more detailed and consensus regulations that address the complexity and safe applications of bioengineered EVs.

Concluding Remarks and Future Perspectives

EVs are continuing to emerge as promising therapeutic agents in regenerative medicine and as versatile platforms for targeted drug delivery across various diseases. The interest in EVs is reflected by over 340 EV-related clinical trials listed on ClinicalTrials.gov (https://classic.clinicaltrials.gov/, using the keywords ‘exosomes’ and ‘extracellular vesicles’), the primary focus of which is on diagnosis, observational studies, and safety assessments of native EVs. Among these records, only 16 cases (around 5%) involved bioengineered EVs as therapeutic agents, indicating the need for more clinical trials to evaluate these functional EVs given the explosion in preclinical studies. Despite the rapid growth in recent years, translating these EVs into practical applications poses multifaceted challenges, from the technical complexities of bioengineering to regulatory compliance (see more details in Outstanding Questions). Compared with native EVs, the transition of bioengineered EVs into clinical trials requires adherence to GMP-compliant quality systems to address a range of crucial factors: (1) selection of optimal cell sources for bioengineering; (2) quality assurance and control during the cell bioengineering process; (3) heterogeneity in EVs generated by bioengineered cells; (4) EV recovery after bioengineering, isolation, purification and storage; (5) industry standards for the molecular characterization of EV content and “unwanted” luminal cargo; and (6) standard protocols for assessing the safety and therapeutic potency of bioengineered EV therapeutics. The successful translation also hinges on the availability of economical large-scale manufacturing processes and downstream procedures [50]. Although bioengineered EVs can build on the advantages of native EVs while addressing their shortcomings, they currently show lower encapsulation efficiency relative to established and commercial platforms [97, 98]. The complexities of regulatory compliance and manufacturing costs for bioengineered EVs are also greater than those for these synthetic platforms. EV-based hybrid systems could harness the benefits of both natural and synthetic platforms and offer improved formulations. Additional insight into the aspects of EV biogenesis that further enhance therapeutic cargo release from donor cells is also crucial for unlocking the full therapeutic potential of bioengineered EVs. Moreover, optimizing the administration of EVs (because systemic administration leads to accumulation in undesired organs) represents another avenue for enhancing their performance. A combination of targeted and localized delivery of bioengineered EVs may prove to be the best approach to improving efficacy and achieving desired therapeutic outcomes [99, 100].

Outstanding Questions.

• How can we stimulate parental cells in a mild manner to efficiently produce large quantities of therapeutic EVs, while minimizing the risk of genetic instability and the inclusion of unwanted cargos?

• What methods or technologies can be developed to minimize the loss of bioengineered EVs during the multiple isolation and purification steps necessary to remove contamination?

• How can we mitigate the heterogeneity of EVs generated by bioengineered cells?

• What parameters and technologies should we use to identify and remove non-functional EVs to ensure a consistent and predictable therapeutic product?

• In what ways does bioengineering influence regulatory compliance and safety considerations for bioengineered EVs, particularly in the context of clinical translations?

• How can bioengineering strategies be further optimized to increase the abundance of therapeutic EVs, and in what ways can the natural process of EV biogenesis be leveraged to unlock the full therapeutic potential of bioengineered EVs?

• Is there any way to simplify the regulatory compliance and reduce manufacturing costs associated with bioengineered EVs?

• Beyond the scope of bioengineering itself, are there alternative aspects that can be explored to further enhance the performance and therapeutic efficacy of bioengineered EVs?

Figure in Box 1. Mechanisms of EVs biogenesis and related pathways reflect its therapeutic potential and bioengineering horizons.

Highlights.

Extracellular vesicles (EVs) have been widely explored both as inherent therapeutics in regenerative medicine and as drug delivery systems targeting multiple diseases.

Primary challenges in developing EV therapy include low production yields and limited capacity for carrying therapeutic agents. Bioengineering strategies have been developed to enhance EV production and maximize their potential for carrying therapeutic payloads.

Despite the bloom of bioengineering strategies for therapeutic EV aplenty, the biomanufacturing of bioengineered EVs toward clinical translation faces complex challenges that need to be addressed.