Direct delivery of assay reagents to extracellular vesicles in liquid biopsies for biomarker analysis

Abstract

Extracellular vesicles (EVs), present in blood as well as other biological fluids, encapsulate nucleic acid biomarkers used for diagnosis, prognosis and treatment monitoring of disease via minimally invasive liquid biopsy. EVs are a reliable source of biomarkers because their contents reflect the cells from which they are derived, and their lipid bilayer membranes protect nucleic acids from degradation. Previously, analyzing EVs in blood was difficult because of time-consuming, labor-intensive EV isolation methods. Here, we provide a protocol for an EV detection approach in which reagent-loaded liposomes fuse with EVs directly in patient blood to sensitively detect RNA within the EVs. In this ‘liposome–EV fusion assay’, antibodies capture EVs in blood, and reagent-loaded liposomes initiate liposome–EV fusion and CRISPR-based nucleic acid detection. We originally used this assay to detect EV-encapsulated viral RNA and accurately diagnose infectious diseases from patient plasma. It has since been adopted by many other research groups to detect mRNA, microRNA, DNA, DNA mutations and EV surface proteins in a variety of patient-derived tumor samples, incorporating enzymatic and nonenzymatic detection reagents and different diagnostic readouts. As a clinical and research tool, this approach has great potential for the diagnosis, treatment and study of cancer, infectious diseases and neurological dysfunction.

Introduction

Extracellular vesicles (EVs) are nanoscale membrane-bound particles that are actively secreted by most cells and function as critical mediators of intercellular communication, having pivotal roles in numerous biological processes, including immune modulation, tissue repair and disease progression^1,2. EVs carry cargoes of proteins, lipids and nucleic acids (NAs) that regulate physiological and pathological phenotypic responses of their recipient cells. For instance, mesenchymal stromal cell (MSC)-derived EVs can deliver growth factors and anti-inflammatory molecules to promote tissue regeneration³, while tumor-derived EVs can reprogram cells in their tumor microenvironment to facilitate metastasis^4–7. EVs are abundantly secreted by diseased, infected or injured cells. They carry markers that reflect their cell of origin, making it possible to detect changes in specific cell populations by analyzing EVs with cell-type-specific markers within the overall EV population. This information can be used to distinguish EVs derived from diseased or injured cells and healthy cells nearby that are affected by them, to evaluate distinct aspects of disease progression and associated cell-to-cell interactions. EVs thus represent a valuable resource for the discovery of disease-associated biomarkers and pathways that regulate these processes⁸.

However, most EV isolation methods (for example, ultracentrifugation and size-exclusion chromatography) are labor-intensive, time-consuming and yield heterogeneous EV populations that can obscure functional differences among distinct EV subpopulations^9–12. The sensitivity and specificity of standard EV analysis methods, including quantitative polymerase chain reaction (qPCR), reverse transcription (RT)-qPCR^13,14 and enzyme-linked immunosorbent assay (ELISA)¹⁵, can also be limited by the low abundance and rapid degradation of EV factors, particularly in complex biological specimens¹⁶. Sensitivity and specificity issues are of particular concern in diagnostic applications, as accurate detection and precise quantification of EV biomarkers are often critical when diagnosing a disease and monitoring its treatment response. Sensitive detection of disease-specific EV biomarkers is also valuable for early disease diagnosis, where detection of subtle EV differences could inform clinical decisions necessary to prevent or limit disease progression and pathology to improve patient outcomes¹⁷.

Biological specimens frequently contain heterogeneous EV subpopulations with distinct biogenesis pathways, biological functions and cellular origins, which can complicate their analysis^18–20, and conclusions about their specific roles in physiologic and pathologic processes^21,22. Innovative approaches are needed to address the limitations of traditional EV isolation and analysis methods to allow rapid, sensitive and specific detection of EV-associated biomarkers^20,22. Liposome–EV fusion approaches that permit in situ analysis of EV cargos by leveraging the specific properties of these two vesicle types²³ offer a new means to address these issues. Synthetic liposomes can be engineered to carry reagents that permit sensitive and specific detection of various biomarker types upon their fusion with EVs carrying these factors^17,24–26. Such reagent-loaded liposomes can be used to detect biomarker-positive EVs in the general EV population, or in specific EV subpopulations isolated by cell- or lineage-specific capture antibodies. Specificity for distinct EV subpopulations or disease states can potentially also be further increased by surface-modifying reagent-loaded liposomes with affinity factors that recognize additional biomarkers associated with a cell population or disease condition to favor affinity-mediated vesicle fusion events.

Liposome–EV fusion assays have several technical advantages over existing methods that can enhance their sensitivity and reproducibility¹⁷. For example, our approach directly integrates EV capture into the assay workflow to avoid the additional sample manipulation steps required by assays that use EV preisolation procedures, reducing the analytical variance associated with these steps. This EV capture procedure can also reduce the shear stress and EV contamination effects associated with filtration- and precipitation-based EV enrichment protocols to yield EV isolates with robust integrity and high purity and thus enable sensitive and reproducible detection of low-abundance EV biomarkers. Unlike conventional bead-based EV enrichment approaches that require downstream lysis and RNA/protein extraction, our method directly integrates antibody capture with liposome fusion and clustered regularly interspaced short palindromic repeat (CRISPR)-based detection, thereby eliminating additional processing steps and enabling rapid, in situ molecular readout from captured EVs.

Liposome–EV fusion assays can also have substantial utility for basic EV research and therapeutic EV development studies. Since these assays can detect biomarker signals derived from individual EVs, they can be used to study EV functional heterogeneity and evaluate the roles of distinct EV subpopulations in regulating specific intercellular communication and pathogenic processes^27,28. Most single-EV studies currently use high- or super-resolution microscopy or nanoscale flow cytometry systems, which can be prone to artifacts and require the use of expensive equipment. These methods are primarily used to evaluate EV surface markers, as EV cargo analyses can require permeabilization approaches that reduce EV integrity. Liposome–EV fusion assays could thus streamline single-EV analyses of biomarker cargoes of specific EV populations, while reducing the cost and complexity of such analyses. This could also be valuable when analyzing the distribution of functional biomolecules among single EVs in therapeutic EV isolates, which may be necessary to optimize the reproducible preparation of such EVs. Furthermore, the ability to engineer liposomes for targeted delivery of therapeutic agents to specific EV subpopulations opens new possibilities for the development of EV-based therapies^27,29.

In summary, the liposome–EV fusion approach fills a critical gap in EV research by providing an integrated and streamlined approach for EV capture and analysis that provides high sensitivity and specificity. This innovation has the potential to transform both basic research and clinical applications, advancing our understanding of EV biology and paving the way for new diagnostic and therapeutic strategies. By addressing the limitations of existing methods and overcoming the challenges posed by EV heterogeneity, this approach represents an important step forward in the field, with far-reaching implications for science and medicine.

Development of the protocol

Synthetic liposome, which can be readily produced using a variety of lipid components, represents one of the earliest and most widely utilized drug delivery systems³⁰. These particles can be loaded with a wide array of therapeutic materials, traverse most membrane barriers and exhibit selective delivery of their cargoes to target cell populations after affinity modifications to their membranes³¹. Challenges such as rapid clearance by the immune system, limited cargo loading efficiency and constraints in surface modifications have created developmental bottlenecks for liposome technology. However, EVs are naturally released from cells and thus inherently possess characteristics and targeting properties of their parent cells, allowing them to function as effective delivery vectors for some therapeutic applications³².

With the rise of gene editing tools, the CRISPR–Cas system has also been encapsulated and delivered for gene therapy^33–35. Drawing from these technologies and leveraging our prior expertise in CRISPR-based diagnostics, we developed a novel method for NA detection using liposome–EV fusion that involves encapsulating recombinase polymerase amplification (RPA)/RT-RPA–CRISPR³⁶ reagents into liposomes, capturing EVs using antibodies and facilitating membrane fusion to deliver the CRISPR detection system into EVs^37,38. Upon activation by target NAs, the CRISPR reaction releases fluorescent signals that can be detected through various means, including qPCR, plate readers and fluorescence microscopy (Fig. 1).

Fig. 1 |. A summary of the liposome–EV fusion workflow.

The workflow comprises four main stages: liposome synthesis (Steps 4–26), EV capturing (Steps 57–63), liposome–EV fusion (Steps 64–68) and signal output (Steps 76–79). Functional liposomes are synthesized and loaded with molecular cargo such as oligonucleotides and enzymes. EVs containing target DNA or RNA markers are captured from body fluids (using antibody-based surface protein recognition) and fused with the liposomes, enabling the release and amplification of EV-derived NAs via RPA or RT-RPA. The CRISPR-based detection system provides ultrasensitive signal generation, with outputs measurable by real-time qPCR, fluorescence plate reader or fluorescence microscopy.

Following pathogen infection, both the pathogens themselves and host cells (such as virus-infected cells and immune cells such as macrophages that have engulfed pathogens) produce substantial amounts of EVs³⁹. These EVs carry antigens and NAs of the pathogens, making them excellent targets for infection diagnosis³⁹. To detect pathogen NAs within EVs (using severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) RNA as an example), a highly sensitive detection technology is required, and CRISPR has emerged as our preferred choice¹⁷. By using a liposome–EV fusion approach, it is possible to deliver the CRISPR detection system directly into the EVs, enabling the detection of pathogen NA fragments without the need for NA purification steps.

The primary step involves synthesizing liposomes encapsulating detection reagents. We selected highly compatible dimyristoylphosphatidylcholine (DMPC) and cholesterol⁴⁰ as the membrane carriers. In the initial approach, we achieved rapid encapsulation of reagents through repeated extrusion and membrane filtration. Other methods, such as freeze–thaw cycles, can also synthesize liposomes but are less efficient and more time-consuming⁴¹.

For NA detection using liposome–EV fusion, the detection reagents must meet certain prerequisites. Given the low abundance of NAs in EVs⁴², the RNA content is often minimal⁹. Although CRISPR-based detection without amplification has been reported, it still requires reaching a sufficient copy number⁴³. Therefore, NA amplification is a necessary step to enable one-step CRISPR detection. As high temperatures can inactivate Cas12 and disrupt membrane structures, PCR is unsuitable. Instead, we opted for isothermal amplification as the prestep for CRISPR detection. After evaluating several isothermal amplification methods, we balanced factors such as amplification temperature, primer design, amplification efficiency and CRISPR compatibility, ultimately selecting RPA as the NA amplification method. Our previous work⁴⁴ laid the foundation for one-step RPA–CRISPR detection of viral NAs, achieving the required sensitivity. In addition, continuous incubation at 37 °C does not compromise membrane stability.

Another critical step is the enrichment of EVs from body fluids. Although EVs are abundant in bodily fluids such as blood (up to 10⁹~10¹² particles/mL), their concentration remains relatively low⁹. Without enrichment, the fluorescent signal generated by CRISPR would be diluted and difficult to detect. For SARS-CoV-2 detection, we utilized an EV-specific antibody (anti-CD81) for capture and enrichment⁴⁵. Similar to ELISA, we first coated the antibody onto a 96-well plate, then incubated bodily fluids, such as serum, on the coated plate to enrich EVs. This design eliminates the need for complex EV separation steps such as centrifugation or filtration, notably simplifying the detection process and reducing operational requirements. After adding the RPA–CRISPR reagent-encapsulated liposomes to the EV-enriched plate, liposome–EV fusion can commence. However, this process is inherently slow and may take up to a day²⁹. To address this, we used polyethylene glycol (PEG)⁴⁶ to promote fusion, reducing the detection time to ~2 h (refs. 25,29).

The fluorescent signals generated by RT-RPA/RPA–CRISPR can be detected through various methods. If a 96-well plate format is used, a fluorescence plate reader is the optimal choice. EVs can also be immobilized on glass surfaces, where small-volume droplet arrays enable rapid detection of multiple targets. Fluorescence microscopy or scanners can then be used to acquire a large volume of results. Alternatively, if fusion occurs in an aqueous phase, qPCR can serve as the detection instrument by reading the isothermal amplification and fluorescent signals.

Overview of the procedure

The comprehensive liposome–EV fusion technology solution offers a simple and rapid method for pathogen NA detection. The standout advantage of this approach is its elimination of EV isolation and purification steps, enabling direct detection of NA biomarkers within EVs. In addition, the versatility of CRISPR endows this method with exceptional scalability; targeting different NA sequences merely requires the replacement of primers and guide RNAs (gRNAs). Similar to drug delivery strategies, this method can be further enhanced by modifying the liposome surface with antibodies or other targeting molecules, enabling direct targeting of tissue-specific EVs and analysis of their internal NAs or even non-NA biomarkers^30,47.

The procedure comprises four main stages (Figs. 1 and 2): (1) liposome synthesis (Steps 4–26), (2) EV capturing (Steps 57–63), (3) liposome–EV fusion (Steps 64–68) and (4) signal output (Steps 76–79). The workflow begins with the synthesis of functional liposomes, composed of membrane components and loaded with molecular cargo such as oligonucleotides and enzymes (for example, polymerases, reverse transcriptase and CRISPR effectors). EVs containing target DNA or RNA markers are captured from body fluids (cell culture (Step 3A) or plasma (Step 3B)) using antibody-based surface protein recognition. Fusion between liposomes and EVs is then accelerated by PEG, enabling the release and amplification of EV-derived NAs via RPA (RPA or RT-RPA).

Fig. 2 |. Applications of liposome–EV fusion systems in diagnostics, therapeutics, regenerative medicine and immune modulation.

This schematic illustrates the design of liposomes, their integration with EVs from various biological sources (for example, blood, saliva, cerebrospinal fluid, urine, milk and amniotic fluid) and their applications across multiple disease contexts, including neurological disorders, infections, wounds and cancer. The top illustration shows liposome features (lipid membrane, surface modifications and cargo loading), EV sources and related diseases. The bottom illustration is divided into four application areas: diagnostics, liposomes encapsulating CRISPR systems or nanomaterials that detect pathogen- or host-derived NA from EVs for disease detection and monitoring; therapeutics, drug-loaded liposomes targeting disease-associated EVs for tumor- or brain-specific delivery; regenerative medicine, bioactive cargo and surface markers enhancing fusion with stem cell-derived EVs for tissue repair; immune modulation, liposomes functionalized with immunoregulatory agents or antigens engaging bacterial EVs to activate immune cells, offering platforms for immunotherapy or vaccination. APC, antigen-presenting cell; CSF, cerebrospinal fluid; LPS, lipopolysaccharide; siRNA, small interfering RNA.

The EVs, liposomes and fused EV–liposomes can be characterized by various assays. ELISA (Steps 27–41) is used to quantify EV-specific surface markers and is typically performed after EV isolation and before fusion, serving as a required control to confirm the identity and purity of the EVs. NanoSight analysis (Steps 42–46) is used to determine particle concentration and size distribution, and is essential both pre- and postfusion to assess particle yield and detect aggregation or loss. Morphological assessments using transmission electron microscopy (TEM)/scanning electron microscopy (SEM) (Steps 47–51) and cryo-electron microscopy (cryo-EM) (Steps 52–56) provide high-resolution imaging to evaluate vesicle integrity and structural features; these are recommended for initial characterization and for validating new batches, although they may be optional for routine assays. Finally, Förster resonance energy transfer (FRET) analysis (Steps 69–75) is a critical assay for confirming successful liposome–EV fusion, and is typically performed after the fusion step. While some assays are optional depending on the experimental goals and prior validation, ELISA, NanoSight and FRET are generally considered essential for confirming identity, quantity and fusion efficiency, respectively. The different characterization assays are summarized in Box 1.

BOX 1. Characterization of EVs, liposomes and fused EV–liposomes.

To ensure the quality, identity and functional integrity of EVs, liposomes and their fused counterparts, various characterization assays are incorporated at defined stages of the workflow:

• ELISA (Steps 27–41): used to quantify EV-specific surface markers (for example, CD63, CD81, CD9). This assay is typically performed after EV isolation and before fusion, serving as a required control to confirm EV identity and purity. It is especially important for first-time users or when working with new EV sources

• NanoSight tracking analysis (Steps 42–46): measures particle concentration and size distribution. This assay is recommended both before and after fusion to assess EV/liposome yield, detect aggregation and monitor changes in particle size. It is considered a core QC step

• TEM (Steps 47–51): provides high-resolution imaging of vesicle morphology and membrane integrity. TEM is strongly recommended for initial characterization of EV and liposome preparations, and for validating new batches

• cryo-EM (Steps 52–56): this is optional, but informative assays offer complementary structural insights. cryo-EM, in particular, preserves native vesicle morphology and is useful for high-resolution structural validation

• FRET assay (Steps 69–75): a critical assay for confirming successful fusion between liposomes and EVs. FRET is performed after the fusion step and is required to validate the efficiency and specificity of the fusion process

While some assays (for example, SEM, cryo-EM) may be optional depending on the experimental context, ELISA, NanoSight and FRET are considered essential for confirming EV identity, quantifying particle yield and verifying fusion, respectively. These assays can be adapted for routine use or expanded for in-depth characterization as needed.

The CRISPR-based detection system provides ultrasensitive signal generation, with outputs measurable by real-time qPCR, fluorescence plate reader or fluorescence microscopy. Typically, a fluorescence plate reader can provide high sensitivity and convenience for method development and optimization. However, the fluorescence acquisition algorithm is brand-dependent, causing the scale to be consistent across different reader brands. Real-time qPCR has been broadly used in research and pathology labs. Integrated fluorescence acquisition and analysis software is easy to use but lacks manual optimization options. Thus, real-time qPCR is more suitable for established liposome–EV fusion protocols. In addition, if the fusion reaction is set up on slides, fluorescence microscopy will be the best option (Fig. 1).

Applications of the method

Diagnostics

Liposome–EV fusion technology has emerged as a versatile platform that bridges synthetic and biological systems to enhance biomedical applications across diagnostics, therapeutics, regenerative medicine and immune modulation (Fig. 2). In diagnostics, based on our protocol, the fusion of synthetic liposomes with EVs improves the sensitivity and specificity of liquid biopsy approaches. Hybrid vesicles enable efficient multiplexed detection of miRNAs⁴⁸ and proteins⁴⁹, and NAs directly from plasma or fecal EVs without prior isolation or lysis. Platforms such as ddSEE⁵⁰ and EXTRA-CRISPR⁵¹ demonstrate high-throughput, one-pot detection of microRNAs (miRNAs) and proteins at the single-EV level using Cas systems and liposome-mediated delivery. Similarly, encoded fusion vesicles^52,53 offer classification of tumor subtypes with >95% accuracy. Integration with portable or digital readouts, such as glucose meters⁵⁴ or smartphone apps, broadens accessibility to point-of-care diagnostics. Machine learning-guided systems such as fecal extracellular vesicle miRNA signatures leverage fecal EVs for colorectal cancer screening with >97% accuracy⁵⁵.

Therapeutics

Exosome–liposome fusion systems have enabled highly efficient therapeutic delivery of proteins, drugs and large plasmids across various cell types, and our protocol further improves the efficacy. For example, Sato²⁵ demonstrated that HER2-expressing exosomes fused with liposomes via freeze–thaw cycles retained functional receptor activity and improved cellular uptake, providing a strategy for stable, bioactive therapeutic carriers. Yang developed a coiled-coil lipopeptide-mediated fusion platform that bypassed endocytosis, enabling rapid cytosolic delivery of doxorubicin and fluorescent dyes directly through plasma membrane fusion, reducing lysosomal degradation⁵⁶. Lin et al. showed that liposome–exosome hybrids could encapsulate and deliver CRISPR–Cas9 plasmids into MSCs, overcoming limitations of traditional lipofection and enabling in vitro gene editing³⁷. Gharehchelou et al.⁵⁷ further confirmed the utility of such hybrids in plasmid delivery to HEK293T cells, highlighting the critical influence of liposome preparation and fusion methods on transfection efficiency. Collectively, these strategies offer a nonviral, biocompatible route for precision therapeutic delivery, particularly where conventional vectors fail.

Regenerative medicine

Regenerative medicine benefits from the biocompatibility and cell-specific communication of EVs combined with liposomal stability and scalable engineering. Functionalized hybrids have been used to deliver YAP1 to tendon progenitor cells, promoting tendon regeneration⁵⁸, and WNT3A-decorated vesicles support alveolar epithelial repair following lung injury. The controlled fusion of hybrid vesicles with tissue-resident cells provides spatially directed regenerative signals, making them promising vehicles in cell-free regenerative therapies. Although our protocol has not yet been directly applied in regenerative medicine, its capacity for antibody modification gives us confidence in its potential for future applications.

Immune modulation

In immune modulation, hybrid vesicles enable antigen multiplexing and safe immune activation. Gnopo et al.⁵⁹ demonstrated that outer membrane vesicle (OMV)–OMV or OMV–liposome fusion allows the assembly of multiantigen vaccines without genetic co-expression. Genetically modified Escherichia coli strain ‘ClearColi’-derived OMVs reduce endotoxin toxicity while preserving immunogenicity, creating safer vaccine candidates. Adapting from our protocol, Long et al.⁶⁰ introduced the Underlying liposome-induced Membrane Exchange (LIME) platform, wherein liposomes integrate with bacterial membranes to produce membrane-integrated liposomes (MILs) loaded with redox-active cytochromes. These MILs reprogram electron transfer in host microbes, offering a novel redox-based immune interface and enhancing biocompatible immunotherapeutics. Together, liposome–EV fusion systems enable scalable, multifunctional and tunable delivery systems that advance multiple frontiers in translational biomedicine. Beyond these published applications, liposome–EV fusion remains a modular platform that can be adapted for various secreted biomolecules and therapeutic interventions. The technique can be combined with machine learning to classify disease states on the basis of EV composition, or it can be integrated with single-cell technologies for high-resolution functional analysis. Future directions in this field may include the development of personalized nanomedicine platforms, expansion into neurological disease diagnostics and optimization for precision immunotherapy.

Liposome–EV fusion technology provides a powerful toolkit for early disease detection, targeted therapy and biosensing. By enhancing biomarker detection, drug delivery and functional characterization of EVs, this technology continues to bridge the gap between synthetic nanomedicine and natural EV biology. Representative studies spanning diagnostics, therapeutics, regenerative medicine and immune modulation are summarized in Table 1, with details of experimental approach and key outcomes for each report.

Table 1 |.

Applications of liposome–EV fusion technology

Model/sample	Application	Cargo/target	Readout/key innovation	Key findings (as reported)	Ref.
Exosomal miRNAs (cell lines/plasma)	Virus-mimicking fusogenic vesicles used to deliver detection reagents to EVs	miR-21, other miRNAs	Rapid, fusion-enabled miRNA detection within EVs	Fast assay enabling EV-contained miRNA readout	24
Exosomes + liposomes (in vitro)	Produced hybrid exosomes by membrane fusion with liposomes	HER2 protein on exosomes	Maintained receptor activity; improved uptake	Stable, bioactive hybrids; enhanced cellular uptake	25
Cell delivery model	Coiled-coil lipopeptide-mediated membrane fusion (endocytosis-independent)	Doxorubicin, dyes	Direct cytosolic delivery via fusion	Reduced lysosomal degradation; efficient delivery	56
MSCs (in vitro)	Liposome–exosome hybrids to deliver large plasmids	CRISPR–Cas9 plasmid	Hybrid vesicles overcome lipofection limits	Enabled in vitro gene editing of MSCs	37
HEK293T (in vitro)	Compared fusion methods for plasmid delivery via hybrids	Plasmid DNA	Method-dependent transfection efficiency	Confirmed utility; emphasized prep/fusion effects	57
Breast cancer EVs	‘All-in-one’ fusogenic nanoreactor for rapid EV miRNA detection	miRNAs	One-pot, fast readout with fusogenic system	Rapid breast-cancer-relevant EV miRNA detection	48
Single EVs	Concurrent protein + miRNA detection at single-EV level using digital dual CRISPR	Protein + miRNA	Digital single-EV CRISPR	High-sensitivity single-EV profiling	50
EV miRNAs	One-pot isothermal Cas12 assay for sensitive miRNA detection	miRNAs	Isothermal, programmable Cas12 readout	Sensitive EV miRNA detection	51
Tumor-derived EVs	Encoded fusion-mediated miRNA signature profiling	miRNA panel	Encoded fusion vesicles for classification	Tumor subtype classification	52
Tumor EV subsets	Dual surface-protein orthogonal barcoding; subset tracing + miRNA profiling	miRNAs + subset markers	Orthogonal barcoding of EV subsets	>95% classification accuracy (reported)	53
Urine EVs (concept)	Sucrose-powered liposome nanosensors with glucometer readout	NA/enzymatic reporters	Low-cost glucometer detection	Portable readout concept for EV assays	54
Fecal EVs	Machine learning-aided EV miRNA signatures for colorectal cancer screening	miRNA panel	Machine learning classifier from EV cargo	>97% accuracy (reported)	55
Tendon/lung repair	Hybrid vesicles deliver morphogens to promote repair	YAP1; WNT3A	Localized regenerative signaling via hybrids	Enhanced tissue repair, motor recovery	58
Vaccinology	OMV–OMV/OMV–liposome fusion for multiantigen assembly	Bacterial antigens	Genetic-free antigen multiplexing	Scalable immune activation	59
Redox immunotherapy	Liposomes merge with bacterial membranes to form membrane-integrated liposomes	Cytochromes (MtrCAB)	Membrane-integrated liposomes reprogram electron transfer	Functional protein transfer; biocompatible	60

Comparison with other methods

Substantial interest and effort have been focused on the development of new approaches to detect and analyze EV biomarkers, but current EV detection methods still have obvious limitations. Most methods require independent EV isolation steps to produce the purified EV samples they analyze, and these procedures are usually the most time-consuming and labor-intensive part of these analyses, substantially increasing the time required for the overall analytical process. Some recently introduced EV purification systems (for example, Exodus⁶¹) can be used to streamline the EV isolation procedures required by these assays, although these methods usually do not reduce the sample volume required for such isolations, and can have substantial equipment and operational costs. Subsequent analysis of these purified EV isolates usually requires enzymatic signal amplification approaches to detect targeted biomarkers.

EV NA biomarkers are usually analyzed by PCR- or RT-PCR-based approaches that require separate NA extraction procedures to isolate purified DNA or RNA from purified EV isolates. However, even these PCR-based analyses may require a substantial amount of material when attempting to detect low-abundance targets within the analyzed EV population. Next-generation sequencing (NGS) is also sometimes used to analyze EV NAs, as it offers important advantages when attempting to analyze numerous targets, but this approach is expensive and requires substantial technical expertise during the library preparation, sequencing and data analysis procedures. Liposome–EV fusion assays that use integrated nucleic amplification and CRISPR reactions can, however, address all these issues to permit high-sensitivity detection of EV NA targets using a straightforward assay workflow compatible with a broad array of specimen types.

EV assays often focus on protein biomarkers, particularly surface proteins, using modified ELISA protocols. In these EV-specific ELISAs, surface markers serve as targets for capture and enzyme-linked detection antibodies. To ensure reliable results, especially when the target biomarker is present at low levels, these assays typically require a separate EV isolation step to concentrate the sample, despite the inherent signal amplification of the enzyme-based approach. However, the use of an emerging proximity extension assay approach can mitigate this issue. This assay approach uses two NA-labeled antibodies that recognize two distinct EV surface biomarkers so that dual binding of these antibodies to an EV can permit their NA labels to hybridize and serve as a template for a polymerase-mediated extension and NA amplification reaction. Fluorescent signal produced upon dye or probe hybridization to this amplicon, or by a CRISPR-based reporter cleavage reaction, can be used to sensitively detect specific EV surface proteins. Alternatively, these amplicons can be subjected to NGS for high-throughput, multiplexed detection of targeted EV surface proteins. Notably, this proximity extension assay method could also be adapted for use in liposome–EV fusion assays to sensitively detect specific proteins or pairs of proteins in the cargos of target EV populations. ELISAs have also been used to analyze biomarker targets in protein lysates of purified EV isolates when attempting to analyze soluble non-NA lumen or inner membrane EV factors, instead of transmembrane or outer membrane EV factors, although these lysate samples can also be analyzed by a variety of other methods with varying sensitivity, technical requirements and costs.

Limitations

Liposome size, reagent encapsulation and stability characteristics influence assay performance. Liposome loading capacity is determined by the internal volume of the synthetic vesicle, and by the concentration of reagents that can be loaded into this volume without destabilizing the vesicle or compromising the conditions required for optimal reaction performance upon fusion with a target EV. Liposome diameter (typically 100–200 nm) should be optimized to maximize reagent uptake and retention, while still maintaining fusion efficiency with target EVs to maximize assay sensitivity and target detection^25,41. Insufficient reagent loading can attenuate reaction efficiency and resulting assay signal intensity, whereas use of oversized liposomes may compromise their stability and fusion efficiency⁴⁰. The uniformity of liposome synthesis, achieved by sequential extrusion through membranes with decreasing pore sizes (for example, 0.8 μm, 0.4 μm, to 0.2 or 0.1 μm), has a crucial role in ensuring consistent assay performance⁶².

Selection of antibodies that have high affinity and specificity for the target EV population is critical for robust assay performance, as the EVs analyzed in this assay are directly captured from the biospecimen. Use of antibodies with poor affinity or specificity reduces assay sensitivity by reducing the number of biomarker-positive EVs captured, as well as potentially increasing the assay background by detecting signal from an off-target EV population. Another limitation of our workflow is that, although it avoids bulk EV purification steps such as ultracentrifugation, it still relies on antibody-based capture. This represents a targeted enrichment strategy that inherently selects a subset of EVs and may bias the observed cargo profile toward specific vesicle populations. Specific antibodies for CD81, a pan-specific EV surface marker, are often used to capture EVs reflecting the overall EV population of a specimen. However, specific antibodies to lineage-, tissue- or cell-specific surface biomarkers (for example, proteins, glycoproteins or integrins) can also be used to capture EVs derived from these sources to differentially evaluate their specific disease or injury phenotypes. Such antibodies must be selected with care, however, to minimize off-target recognition of undesired EV populations that may obscure or distort assay results⁸. Capture antibody surface conjugation and blocking procedures must also be optimized to enhance EV binding and attenuate nonspecific interactions that could introduce background noise.

NA amplification and CRISPR reaction conditions must be carefully optimized to enable sensitive detection, as an imbalance favoring the CRISPR reaction can deplete the amplicon needed for both processes, reducing sensitivity and increasing the risk of false negatives. Care must thus be taken when selecting both the NA amplification primers and gRNA to avoid this effect and reduce the potential for false-positive recognition of closely related NA sequences, particularly when testing clinical samples containing diverse RNA species. Our reported liposome–EV fusion assay for SARS-CoV-2 uses a well-validated gRNA sequence with a high binding affinity for a target amplicon derived from the SARS-CoV-2 nucleocapsid (N)-gene RNA sequence⁶³. Sequences selected for the assay NA amplification primers and gRNA confer high specificity if they correspond to regions with minor variations in conserved sequence. Careful selection of the gRNA recognition site can, by itself, permit sequence discrimination at single-nucleotide resolution.

RPA reagents used in our assay can present practical constraints when used at scale, including cost and availability issues. The RPA mix, which includes proprietary reagents, is sourced from specific vendors, and its cost can be a limiting factor for assay scalability⁶⁴. Bulk purchasing or alternative amplification strategies may be necessary to mitigate cost and availability concerns unless RPA reagents become more broadly available in the future.

The purity and yield of reagent-loaded liposome fractions also affect assay performance and reproducibility. Following extrusion, reagent-loaded liposome samples are gel-filtrated on a Sephadex G-25 column to remove free reagents and small vesicles^41,65, although care must be taken during this procedure to maximize liposome recovery. These fractions should also be evaluated by nanoparticle tracking analysis (NTA) or a comparable method to confirm the size distribution and concentration of the isolated liposomes. Liposome incubations with captured EVs must also be carefully controlled to optimize their interaction and fusion. Liposome–EV fusion is mediated by PEG, which facilitates membrane mergers⁴⁶, and signal intensities detected during this process vary with the readout time, as longer incubation times can enhance detection sensitivities at the cost of increased background noise²⁹. Proper quality control (QC) is essential throughout the assay workflow. This includes verification of liposome size and stability, both after synthesis and before use, and inclusion of positive and negative control samples to validate the sensitivity and specificity of the assay signal^9,41,66.

Experimental design

The liposome–EV fusion approach is highly modular and can be tuned for different NA targets or EV subpopulations. For example, detection can be shifted from viral RNA to host-derived miRNAs, mRNAs or DNA fragments simply by redesigning the RPA primers and gRNA sequences without altering the liposome formulation or EV capture strategy. Similarly, specific EV subpopulations can be targeted by substituting capture antibodies (for example, CD81 for pan-EV capture versus EGFR or CD45 for tumor- or immune cell-derived EVs) or by decorating liposomes with affinity ligands to favor fusion with selected EV subsets. We recommend that readers conduct pilot experiments to evaluate these modifications in their specific context. Pilot studies should include (1) testing multiple gRNA/primer sets to ensure sensitivity and specificity for the intended NA target, (2) comparing capture antibodies for enrichment efficiency and background and (3) verifying fusion efficiency under the chosen conditions using controls such as dual-dye-labeled EVs. These preliminary tests will help optimize the system for new applications and ensure reproducibility. In theory, the liposome–EV fusion approach for detecting NAs within EVs can be applied to EVs from any source.

Once the target EV and NA are identified, the next step of experimental planning involves designing the CRISPR gRNA. As this method uses Cas12a for detection, the sequence design must adhere to Cas12a’s preferences. Subsequently, liposome surface modifications targeting EV-specific markers, such as antibodies against EV surface proteins, can be selected. The choice of surface modification chemistry, if necessary, and the type of modification will determine the composition of the liposome membrane.

Following CRISPR design and determination of the liposome membrane composition, the reagents can be encapsulated. Liposomes are synthesized via membrane filtration while encapsulating the RPA–CRISPR components (Steps 9–15). Comprehensive characterization and QC of the newly synthesized liposomes are essential, including assessments of size distribution, yield, encapsulation efficiency and detection efficiency. Depending on the requirements, different detection schemes can be used, such as surface enrichment (for example, 96-well plates; Step 68) or direct detection in the liquid phase (Step 70). For surface enrichment, antibodies must first be coated onto the plate (Fig. 1; Steps 57–64), whereas liquid-phase detection typically requires liposome surface modifications with targeting molecules such as antibodies or ligands. To enhance subpopulation targeting after a common EV capture (for example, CD81), we use ligand-decorated fusogenic liposomes (antibody/peptide/aptamer display). Receptor-mediated adhesion (for example, RGD-integrin, EpCAM mAb) increases liposome–EV contact and promotes membrane fusion, improving selective reagent delivery to the intended EV subset without altering the capture workflow. This ‘AND-gate’ design has precedent in immunoliposomes and engineered fusogens that increase membrane fusion efficiency^56,67,68. The next steps involve triggering fusion and detecting the fluorescent signals, with a fluorescence plate reader being the most versatile method. As the analysis relies on fluorescence intensity, data interpretation is relatively straightforward. However, it is important to note that factors such as liposome surface modifications, membrane composition and CRISPR target differences can notably influence the kinetics of the fluorescent signal, making pilot experiments highly recommended.

Design of CRISPR detection system (before starting procedure)

CRISPR–Cas12a detection systems used in this protocol should be optimized to achieve ultrasensitive, sequence-specific identification of EV-associated NAs while still operating within the constraints of liposome-encapsulated reactions, which is critical for integration into its isolation-free workflow. Cas12a is selected over Cas9. While both exhibit gRNA-mediated DNA recognition and cleavage activity, only Cas12a exhibits strong collateral single-stranded deoxyribonuclease activity upon target recognition. This feature enables target-specific and quantitative signal production, as target amplicon recognition results in a concentration-dependent cleavage of the quenched assay reporter. Notably, minor sequence mismatches between specific gRNA sequence regions and the target sequence can be used to detect single-nucleotide differences between closely related sequences and alter the cleavage activity of the Cas12a–gRNA complex. Careful selection of the target sequence and optimization of the gRNA sequence can thus be critical in obtaining high sensitivity and specificity. In our SARS-CoV-2 assay, we use a gRNA complementary to a conserved region of the SARS-CoV-2 N gene, which is chosen for its high expression stability across viral variants and absence of homology with human transcripts to minimize off-target sequence recognition and false-positive signal production. A false-positive signal is not detected when this system is used to analyze phylogenetically diverse pathogens (for example, influenza, Middle East respiratory syndrome coronavirus (MERS-CoV)). No sequence modifications are required to distinguish between variants in this assay, although it is possible to distinguish between variants with single nucleotide polymorphism differences by careful selection of the gRNA target sequence and by introduction of single-nucleotide mismatches to the target that further reduce binding affinity for a single nucleotide polymorphism variant. NA target concentration in captured EVs is assessed by the fluorescent signal produced upon cleavage of the quenched 6-carboxyfluorescein (FAM)-labeled single-stranded DNA assay reporter by Cas12a collateral cleavage activity, which is induced in proportion to amplicon concentration (Figs. 3 and 1).

Fig. 3 |. A step-by-step protocol for liposome–EV fusion assay using CRISPR-based detection.

This workflow outlines the major steps involved in preparing and applying RT-RPA–CRISPR-loaded liposomes for fusion with EVs, enabling sensitive NA detection. Steps 1–3: standard EV production; EVs are produced from 293T cells using lentiviral packaging and expansion, followed by sequential centrifugation and ultracentrifugation to isolate EVs. Steps 4–8: liposome synthesis; liposomes are formed by mixing DMPC and cholesterol, followed by vortexing and air-drying to yield empty liposomes. Steps 9–26: liposome loading; air-dried liposomes are rehydrated with the RPA–CRISPR reagent mix, containing RPA primers, enzymes, gRNA, Cas12a, reverse transcriptase and a fluorescent probe, at varying concentrations. Steps 27–56: purified EV and liposome characterizations; EV size distribution, morphology and fusion efficacy. Steps 57–79: liposome–EV fusion assay; EVs from serum or plasma are captured using CD81 antibody-coated plates, incubated with RT-RPA–CRISPR-loaded liposomes, and subjected to PEG 8000-induced fusion. Fluorescent signals generated by CRISPR activity are then detected using a plate reader.

RT, RPA and CRISPR Cas12a reagents are co-encapsulated to generate reagent-loaded liposomes for this assay to account for the low abundance of SARS-CoV-2 NAs in blood specimens. RPA is selected for NA amplification as: (1) RPA has optimal activity at 37–42 °C (and thus does not affect liposome stability), (2) Cas12a is compatible with RPA reaction conditions and (3) RPA permits rapid (<30 min) target amplification in fused liposome–EV vesicles. Cas12a is selected for the assay readout as its collateral cleavage activity, which is induced upon binding of the target amplicon, can exhibit cleavage rates that exceed 1,000 turnovers per min to achieve attomolar detection sensitivity without specialized equipment.

This system is modular, so it is possible to modify the NA target specificity without affecting the EV specificity, and vice versa. Altering the specificity of the integrated NA amplification/Cas12a reaction is relatively straightforward as this only requires the selection of a new RPA primer set and gRNA that exhibit specificity for the new NA target if the annealing temperature range of the new primers matches that of the old primers. Such substitutions are also possible when switching to an alternate NA amplification system that targets a different NA type (for example, miRNA, RNA or DNA), assuming its activity conditions are compatible with those of Cas12a (and the RT system, if applicable) and do not destabilize the liposome delivery vehicle. This flexibility can substantially reduce re-optimization efforts for new applications.

Preparation of EV standard (Steps 1–3)

EV standards are generated from native and recombinant 293T cell lines to establish a reliable benchmark for assay performance and address the requirement for reproducible controls in EV research. 293T cells are selected for their robust EV secretion capacity and compatibility with lentiviral transduction, which is used to produce stable expression of the SARS-CoV-2 N protein and mimic EV uptake of viral RNA. 293T cells are transduced with a pLenti-CMV-puro vector encoding the full-length SARS-CoV-2 N gene or an empty expression cassette (empty vector control (EV-ctrl)) and cultured under puromycin selection to obtain cultures with >95% transduction efficiency. This method is chosen because the EVs it produces use the standard EV cargo loading process to load the viral RNA target and should thus mimic EVs isolated from individuals infected with SARS-CoV-2. By contrast, alternative EV loading methods that use membrane disruption approaches can artificially skew EV composition or exhibit nonspecific NA adsorption on the EV surface to complicate their evaluation by more direct means.

EVs are collected from the conditioned media of 293T cells 48 h after these cultures are switched to serum-free media to avoid contamination of these samples with bovine serum EVs, a common issue for cell-line-derived EV controls. Conditioned media derived from these cells is centrifuged at 4 °C and 2,000g for 30 min and then 10,000g for 45 min, passes through a 0.45 μm filter to remove cellular debris, and then is centrifuged at 4 °C and 100,000g for 3 h to pellet the sample EVs. PEG-based EV precipitation methods are not used to generate this standard to avoid potential interference with the subsequent liposome fusion assay. EV pellets resuspended in phosphate-buffered saline (PBS) are then analyzed by bicinchoninic acid assay and NTA to confirm that they meet the conditions for our EV standards (0.05–5 μg protein/mL, ~1 × 10¹⁰ EVs/mL, and 120 ± 20 nm mean diameter).

These EV standards are also analyzed by TEM to confirm that they demonstrate a characteristic cup-shaped EV morphology and by western blot to confirm that they express characteristic EV markers (CD81, CD9), but not markers of other cell compartments¹⁹. EV standards used as positive and negative controls for SARS-CoV-2 infection are also required to exhibit strong versus undetectable expression of the SARS-CoV-2 N protein, respectively. Control EVs are then spiked into healthy donor plasma at physiologically relevant concentrations (10⁶–10⁸/mL) to generate positive and negative standards that reflect the complexity of EV populations in plasma specimens, before being subjected to identical CD81 antibody capture and liposome fusion workflows as surrogate patient samples. Dynamic light scattering is used to confirm the particle size distribution after spiking these samples with the EV control to exclude samples with potential aggregation artifacts.

These controls serve dual purposes: (1) assay calibration, N gene-transfected EVs provided a positive reference for CRISPR signal thresholds, while EV-ctrl established baseline noise levels; (2) specificity validation (optional), cross-reactivity testing against EVs from cells expressing unrelated viral RNAs (for example, influenza, MERS-CoV) confirms the CRISPR system’s selectivity. By aligning control EV preparation with International Organization for Standardization 9001 standards for reference materials, including batch consistency checks via zeta potential (−30 ± 5 mV) and proteomic profiling (optional; Supplementary Information), this protocol ensures controls meet diagnostic-grade reproducibility requirements.

The rationale for avoiding serum-containing media and PEG-based isolation underscores the need to preserve EV surface epitopes and fusion competency, critical for mimicking endogenous liposome–EV interactions. We use 293T-derived EV controls to benchmark the performance of an EV assay recognizing a viral RNA target, but this approach can be adapted for assays directed at other EV biomarker types (for example, disease-associated miRNAs, bacterial RNAs). By integrating physiologically relevant cargo loading with rigorous characterization, these controls address the longstanding reproducibility crisis in EV research while enabling standardized inter-laboratory comparisons.

Liposome preparation (Steps 4–26)

Reagent-loaded liposomes used in this assay eliminate the liquid-handling procedures associated with EV NA extraction and purification, which can reduce NA degradation and contamination risks. Further, integrating the normally discrete NA amplification and detection procedures into a single reaction addresses the recognized sensitivity versus workflow complexity trade-off of EV-based diagnostics. Liposomes used in this method must demonstrate good reagent encapsulation efficiency, stability under physiological conditions and efficient EV fusion kinetics to meet the three criteria to permit isolation-free EV analyses.

For our SARS-CoV-2 assay, liposomes are formulated using DMPC and cholesterol at a 10:1 molar ratio, a composition that is selected to balance membrane fluidity and structural integrity during the subsequent EV fusion and RT-RPA/Cas12a reactions. DMPC is selected owing to its relatively low phase transition temperature (~23 °C), which is anticipated to permit efficient fusion with captured EVs under moderate experimental conditions but preserve stability under storage conditions. Cholesterol is also incorporated into these vesicles to enhance the rigidity of the lipid bilayer to further minimize the leakage of their encapsulated reagents. RT-RPA/Cas12a reagents are mixed with these lipids and extruded through polycarbonate membranes with progressively smaller pore sizes (0.8–0.1 μm) to generate reagent-loaded liposomes with ~100 nm diameters—a size chosen to avoid nonspecific aggregation and maximize fusion efficiency with EVs, which are of similar size (30–150 nm). This ethanol-based lipid dissolution and aqueous extrusion process facilitates large-scale production of reagent-loaded liposomes. Reagent-loaded liposomes are then subjected to size-exclusion chromatography to generate vesicles with a constrained size range and to remove nonencapsulated reagents, as free reporters could yield background if degraded by nonspecific nuclease contaminants derived from the assay specimen. The resulting liposome preparations are then characterized by NTA to confirm that they represent monodisperse vesicle populations of the expected size range, which helps to ensure minimal batch-to-batch variability, a prerequisite for clinical-grade applications.

Decoupling liposome synthesis from EV isolation, the protocol eliminates common bottlenecks in EV analysis, such as yield loss during ultracentrifugation or protease degradation during prolonged processing. The design’s emphasis on fusion-enabled, isolation-free workflows directly addresses the unmet need for rapid, sensitive EV analysis in complex biological fluids, positioning it as a versatile platform for both diagnostic and mechanistic studies.

CRISPR assay control (Step 8)

CRISPR assay controls are engineered to ensure specificity, sensitivity and reproducibility across heterogeneous clinical samples, and to address three core challenges: off-target activation, matrix interference and inter-batch variability. Positive controls comprise synthetic SARS-CoV-2 N gene RNA containing EVs spiked into healthy donor plasma at clinically relevant concentrations (10²–10⁵ copies/mL), validating assay linearity and lower limit of detection (LoD, 1.2 × 10³ EVs/mL). Negative controls, including EVs isolated from healthy donor plasma and 293T-EV-ctrl, are used to establish baseline fluorescence thresholds (mean + 3 × s.d.) and to distinguish true signal from background noise.

To mitigate false positives from cross-reactive pathogens, optional specificity controls incorporate genomic RNA or viral particles from phylogenetically diverse agents (MERS-CoV, human coronavirus 229E (HCoV-229E), influenza A/B, respiratory syncytial virus (RSV)), confirming negligible signal generation (<5% of SARS-CoV-2-positive controls). This rigorous validation is critical given Cas12a’s reported collateral activity, with gRNA designs excluding regions homologous to human transcripts or common respiratory pathogens via BLAST alignment. No-template controls (NTCs) that contain all assay components except the input EVs are interspersed among experimental samples across all batches to detect reagent contamination—a key quality metric for clinical-grade workflows.

Validation data revealed 100% concordance between CRISPR and RT-qPCR for positive/NTC controls, with cross-reactivity controls showing 0% false positivity. This framework adheres to Clinical and Laboratory Standards Institute EP17-A2 guidelines for LoD determination for molecular diagnostics, positioning the assay for translational deployment. By preemptively addressing matrix effects (for example, heparin inhibition) via heparinase I pretreatment of plasma, the protocol ensures robustness across anticoagulant types, a common pitfall in clinical EV studies.

Liposome characterizations (Steps 27–63)

Reagent-loaded liposomes generated during assay development are characterized to validate critical parameters required for efficient and reproducible assay activity: monodispersity, reagent retention, fusion competency and batch-to-batch consistency. Liposome preparations are also monitored for batch-to-batch consistency of their zeta potential measurements (−25 ± 3 mV) to ensure that they consistently demonstrate sufficient electrostatic repulsion to prevent their aggregation before fusion with EV specimens. Subsequent batches, however, are analyzed by a rapid and streamlined process that uses NTA/NanoSight to assess their size distribution, concentration and stability (Box 2). Liposomes generated by this method reveal a mean 100 ± 15 nm diameter and spherical morphology when assessed by TEM and NanoSight (Fig. 2). Dynamic light scattering confirms the colloidal stability of these preparations, which retained structural integrity for >30 d at 4 °C, a requirement for scalable production and storage.

BOX 2. Rapid characterization of subsequent batches using NTA.

Once initial EV and liposome preparations have been fully validated, subsequent batches can be assessed using a streamlined protocol centered on NTA. This rapid assay provides key metrics, including:

1. Particle size distribution (mean, mode and s.d.)

2. Particle concentration (particles/mL)

3. Stability over time (via repeated measurements or temperature stress)

Recommended timing

• After EV isolation and purification

• After liposome preparation

• After EV–liposome fusion (optional, for consistency checks)

Procedure summary

1. Dilute the EV–liposome-fused sample in filtered PBS to the optimal concentration range (typically 10⁷–10⁹ particles/mL).

2. Load the sample into the NTA instrument (for example, NanoSight).

3. Record three 60-s videos per sample under consistent settings.

4. Analyze using the software’s batch processing mode to extract size and concentration data.

5. Compare results to reference values from fully characterized batches.

Encapsulation efficiency of RT-RPA–CRISPR reagents, quantified by fluorescence-based assays, reveals that >85% of the reagents present in the liposome extrusion mixture are present in the vesicle fraction after passage through the size-exclusion column used to remove free reagents. This high packaging efficiency is achieved through iterative optimization of lipid hydration times and extrusion pressures to balance reagent entrapment and vesicle deformability.

FRET assays determine that a 5% (wt/vol) PEG concentration is sufficient to maximize the liposome–EV fusion rate (>70% fusion) within a 1 h incubation period without inducing substantial vesicle lysis. Adapted from viral fusion studies, this strategy reduces hydration barriers to lipid bilayer fusion while preserving access to EV membrane proteins, an essential feature for workflows that use affinity ligands on liposomes to target and fuse with EV subpopulations bearing surface markers linked to specific cell types, disease states or infections. This FRET-based fusion assay²⁴ uses EVs dual-labeled with two lipophilic fluorescent dyes (1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) and 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindodicarbocyanine (DiD)) so that liposome–EV fusion can be detected by FRET signal decay, which is proportional to the surface dilution of this dye upon liposome membrane uptake.

ELISA validation of EV (Steps 27–41)

EV QC procedures are designed to assess EV integrity, cargo specificity and purity to minimize false positives in downstream CRISPR workflows. NanoSight quantifies EV particle concentrations (≥1 × 10¹⁰/mL) and polydispersity indices (<0.2), ensuring homogeneity, which is essential for reproducible liposome fusion rates. EV size and morphology are assessed via TEM, confirming cup-shaped structures (100–150 nm) devoid of protein aggregates or cellular debris. These two characterizations (NanoSight and TEM) can be performed in parallel.

Surface marker profiling via CD81/CD9 ELISA validates EV identity, with absorbance thresholds (>0.4 at 450 nm) established using healthy donor plasma EVs as negative controls. Crucially, ELISA or western blot analysis confirms the absence of apolipoprotein B (ApoB) and albumin, which are common contaminants in plasma-derived EVs, while retaining EV-specific markers (CD81, TSG101), aligning with Minimal Information for Studies of Extracellular Vesicles 2023 guidelines⁶⁹. For SARS-CoV-2 RNA-positive controls (293T-N EVs), western blot with lysed EVs verifies N protein expression, while proteomic profiling with liquid chromatography–mass spectrometry/mass spectrometry (optional; Supplementary Information) confirms the absence of viral proteins in EV-ctrl samples, ensuring specificity.

In addition to ApoB, which is commonly used to evaluate contamination by low-density lipoprotein particles, we recommend including ApoA1 as a critical QC marker. ApoA1 is a major protein component of high-density lipoproteins, which can co-isolate with EVs in plasma-derived samples.

Functional QC includes a fusogenicity assay where dual-dye-labeled EVs (DiI/DiD) are incubated with liposomes, with FRET decay (>30% signal reduction within 1 h) confirming fusion competency. Batch consistency is rigorously monitored via zeta potential (−25 ± 5 mV) and RNA integrity number (RIN) (RIN >7.0 for EV RNA isolates), precluding RNase contamination. For clinical samples, EV recovery rates are normalized against spiked-in cel-miR-39, correcting for matrix effects (for example, hemolysis, lipemia) that artificially inflate or suppress EV yields.

This QC framework directly addresses the following key pitfalls in EV diagnostics: (1) aggregation artifacts, TEM/NanoSight exclusion of polydisperse samples prevents nonspecific liposome interactions; (2) cargo specificity, proteomic and western blot validation ensures target signals derive from EVs, not co-isolated free RNA or protein complexes (optional; Supplementary Information); and (3) functional relevance, fusogenicity assays confirm EVs retain membrane plasticity for liposome integration, a prerequisite for CRISPR signal amplification. By enforcing International Organization for Standardization-compliant acceptance criteria (for example, >85% CD81⁺ EVs), the protocol bridges research-grade characterization with clinical diagnostic standards, ensuring that results are both biologically meaningful and analytically robust.

EV capture (Steps 57–63)

The EV capture protocol is designed to maximize specificity and yield while preserving membrane integrity for downstream liposome fusion, addressing key challenges posed by complex biological matrices such as plasma. CD81 is selected as the primary biomarker target for EV capture owing to its ubiquitous surface expression on EVs secreted by most cells and tissues and its minimal co-isolation of non-EV contaminants (for example, lipoproteins) compared with CD9 or CD63, as validated by proteomic profiling (optional; Supplementary Information). When adapting this strategy to capture other EV subpopulations, such as tumor-derived EVs via EGFR or immune cell-derived EVs via CD45, the following factors should be considered: (1) the target’s enrichment on EVs of interest versus background EVs or nonvesicular particles; (2) epitope accessibility after EV release; and (3) minimal expression on undesired EV sources to reduce off-target capture.

Biotinylated capture antibodies should be validated for high affinity and specificity using orthogonal assays (for example, flow cytometry of bead-bound EVs, western blot of eluted fractions) before integration into the workflow. For all target substitutions, an isotype-matched IgG control is essential to quantify nonspecific bead–EV or bead–protein interactions, alongside “no antibody” bead controls to assess background binding from the matrix. These controls allow accurate determination of assay sensitivity and specificity in the context of the chosen target.

For CD81 capture in plates, 1 μg/mL label-free anti-CD81 antibodies are coated on the 96-well plate surface. For CD81 capture in solution, biotinylated anti-CD81 antibodies are conjugated to streptavidin-coated magnetic beads (Dynabeads MyOne) at a 1:500 ratio and optimized via titration assays to balance binding capacity (>80% EV recovery) and minimal nonspecific adsorption (<5% (wt/vol) albumin retention). CD81 is selected as the primary capture marker because it is broadly expressed across EVs derived from multiple cell and tissue types and has been shown to minimize co-isolation of lipoprotein contaminants compared with CD9 or CD63 (Extended Data Fig. 1).

Capture efficiency is enhanced by pretreating plasma with heparinase I (10 U/mL, 30 min) to dissociate heparin–EV complexes that impede antibody binding, followed by a 2-h incubation with beads in PBS containing 0.1% bovine serum albumin (BSA) (wt/vol, pH 7.4). This buffer formulation minimizes hydrophobic interactions with lipoproteins while stabilizing EV membranes. Post-capture, beads are washed thrice with high-stringency buffer (0.5% Tween-20/PBS (wt/vol)) to remove unbound proteins, a critical step for reducing background in subsequent CRISPR assays. EV elution is achieved using a low-pH glycine buffer (pH 2.5) with immediate neutralization, preserving EV structural integrity (TEM validation) and fusion competency (>70% FRET decay in liposome assays).

This capture strategy outperforms ultracentrifugation in EV purity (CD81⁺:ApoB ratio >50:1 versus 5:1) by ELISA and retains small RNA cargo (RIN >7.0), as verified by Bioanalyzer. The modular design enables straightforward adaptation to new targets while maintaining a standardized workflow, providing flexibility for extending the assay to other clinically relevant EV populations without compromising sensitivity or specificity. By decoupling EV isolation from liposome fusion, the protocol eliminated detergent-based lysis steps that degrade membrane proteins, ensuring intact EVs for functional CRISPR analysis. The modular design allows seamless substitution of capture antibodies (for example, EGFR for tumor EVs) by modifying the biotinylated antibody, while retaining standardized workflows, a strategic advantage for expanding to nonviral targets.

Liposome–EV fusion and signal detection (Steps 64–75)

Liposome–EV fusion and signal detection are engineered to enable isolation-free, ultrasensitive NA analysis by integrating membrane fusion, isothermal amplification and CRISPR–Cas12a collateral activity into a single reaction compartment. Fusion is initiated by co-incubating CD81-captured EVs with reagent-loaded liposomes and 5% PEG 8000 (wt/vol) at 37 °C for 1 h, leveraging the effect to destabilize the hydration shells of these vesicles and thus destabilize their lipid bilayers to promote their fusion—a strategy adapted from viral fusion mechanisms. This fusion allows the RT-RPA–CRISPR reagents to diffuse into the EV lumen to interact with a target RNA within the fused vesicles. This permits the introduced RT to convert a target RNA (for example, the SARS-CoV-2 N gene) into cDNA, which is then amplified by RPA to generate double-stranded DNA amplicons that are recognized by a Cas12a–gRNA complex to induce its collateral single-stranded deoxyribonuclease (ssDNase) activity. This activity rapidly cleaves a quenched fluorescent single-stranded DNA reporter (FAM-TTTTTTTTTTTT-BHQ) to generate a signal proportional to the initial target concentration. This integrated workflow is found to achieve single-copy sensitivity within 45 min and outperforms RT-qPCR in detecting this low-abundance EV RNA by eliminating RNA losses incurred during its NA extraction process.

The signal generated in this assay is quantified by a fluorescent plate reader, using fluorescence thresholds calibrated against synthetic standards generated by spiking healthy plasma with SARS-CoV-2 RNA or EVs isolated from a recombinant cell line engineered to express the SARS-CoV-2 N gene, as described in the CRISPR assay control section above, including positive control, negative control and NTC. The specificity of the RT-RPA–Cas12a assay encapsulated in these liposomes is determined by analyzing synthetic samples containing genomic RNA from other respiratory viruses that produce similar symptoms (MERS-CoV, influenza A/B), which reveals a <1% background signal. For clinical samples, results are normalized to internal controls (for example, human GAPDH) and reported as ΔF (sample – NTC), with ΔF >3 × s.d. of healthy donor baselines (n = 10) defined as positive.

By preserving EV integrity and compartmentalizing reactions, the protocol should achieve 91% accuracy in triple-negative breast cancer EV glycan profiling⁷⁰ and 97.4% accuracy in colorectal cancer miRNA detection⁵⁵, demonstrating broad applicability. Future adaptations could integrate machine learning for automated thresholding or single-EV analysis via droplet microfluidics, further advancing precision medicine. With more disease-derived EV mechanism studies, this protocol can also offer a potential transformative platform for early disease detection, therapeutic monitoring and functional EV characterization.

Analysis (Steps 76–79)

The analytical framework is designed to rigorously validate CRISPR performance against gold-standard methods while addressing clinical and technical variability inherent to EV-based diagnostics. Data normalization first applies ΔF values (sample fluorescence – NTC) to correct for background noise, with thresholds calibrated using receiver operating characteristic curves derived from 293T-N EV controls and healthy donor plasma (area under the curve (AUC) = 0.99). This approach balances sensitivity (94.1%) and specificity (98.4%) while minimizing false positives in complex matrices such as lipemic or hemolyzed plasma. Key analytical components of this approach include:

1. Cross-reactivity validation: CRISPR signals for phylogenetically related viruses (MERS-CoV, human coronavirus 229E, influenza A/B) have been compared with SARS-CoV-2-positive samples¹⁷ using Mann–Whitney U tests, confirming negligible cross-reactivity (P < 0.001).

2. LoD: probit regression of serially diluted synthetic SARS-CoV-2 RNA (BEI Resources NR-52285) established an LoD of 1.2 × 10³ EV/mL (95% CI), validated against 293T-N EV spike-ins¹⁷.

3. Clinical concordance: Cohen’s κ coefficient (κ = 0.89) quantified agreement between CRISPR and nasal swab RT-qPCR in 46 patients with suspected coronavirus disease 2019 (COVID-19)¹⁷, with discrepancies resolved via seroconversion (IgG⁺) and clinical outcomes. Logistic regression of clinical covariates (age, comorbidities) in the cohort revealed no significant confounding (P > 0.05), confirming assay robustness across demographics.

This analytical strategy adheres to US Food and Drug Administration benchmarks for molecular diagnostics, with ΔF thresholding and receiver operating characteristic curve analysis provided to standardize inter-laboratoy comparisons. By integrating statistical rigor with clinical correlation, the framework not only validates CRISPR but also establishes a blueprint for EV-based assay development, extensible to cancer, neurodegeneration and infectious disease monitoring.

Expertise needed to implement the protocol

Successful use of this protocol requires working familiarity with several basic molecular biology, nanotechnology and bioinformatic procedures. Basic cell biology skills may also be required to generate cell lines expressing specific EV biomarker targets and to isolate control EVs from appropriate native or recombinant cell lines. Liposome synthesis, modification, isolation, handling and storage procedures in this protocol require basic expertise with standard liquid and material handling techniques. Knowledge of surface chemistry is also valuable for optimizing liposome–EV fusion efficiency, particularly when modifying the liposome surface for affinity-based recognition of EV biomarkers to target specific EV subpopulations. The ELISA-based workflow of this protocol also requires familiarity with antibody-mediated capture strategies, while its CRISPR assay design requires some knowledge of RPA and fluorescence-based detection approaches, and gRNA design and selection strategies. Experience with expression vector construction and cell transfection is useful but not mandatory, depending on whether new recombinant cell lines must be generated as the source of assay EV standards. Some knowledge of bioinformatics may also be useful in RPA primer and gRNA selection, statistical evaluation and interpretation of assay results, and comparative data analysis studies. Critical evaluation may also be required to detect and correct assay problems and to adjust assay conditions to optimize assay sensitivity and reproducibility, which may require iterative adjustments to its liposome formulation and fusion conditions, and to RT + RPA/CRISPR reagents and conditions.

Materials

Biological materials

• 293F cells (Gibco, cat. no. R79007)

• Vero E6 cells (ATCC, cat. no. CRL-1586)

• pLenti-CMV-puro vector (Addgene, cat. no. 17452)

• psPAX2 (Addgene, cat. no. 12260)

• pMD2.G (Addgene, cat. no. 12259)

• Human serum (healthy controls)

• Human serum (SARS-CoV-2 detection)

• Virus and viral RNA (Table 2)

• Pathogens for liposome–EV detection (see table below)

  Virus	Vendor	Catalog no.	Type
  SARS	BEI Resources	NR-52346	Genomic RNA
  MERS	BEI Resources	NR-45843	Genomic RNA
  HCoV 229E	BEI Resources	NR-52726	Viral particle
  HCoV OC43	BEI Resources	NR-52725	Viral particle
  HCoV HKU1	ATCC	VR-3262SD	Synthetic RNA
  HCoV NL63	BEI Resources	NR-470	Genomic RNA
  RSV	BEI Resources	NR-43976	Genomic RNA
  Influenza A	BEI Resources	NR-2760	Genomic RNA
  Influenza B	BEI Resources	NR-10048	Genomic RNA
  SARS-CoV-2	BEI Resources	NR-52285	Genomic RNA

Table 2 |.

Troubleshooting

Step	Problem	Possible reason	Possible solution
8	Thin white film of lipids not forming after air-drying	Air-drying occurs too quickly	Lower the air pressure during air-drying by raising the pipette attached to the air hose
9	RPA not amplifying amplicons	Primers not amplifying during RPA	Test alternative primers and ensure amplification occurs with agarose gel
9	CRISPR–Cas12a not producing fluorescence	Cas12a cleavage rate is too high	Consider using alternative gRNA sequence or a protospacer adjacent motif-free gRNA design
15	Lipid solution leaks in when passed through extrusion membrane	Insecure contact between filter support and apparatus	Disassemble extrusion apparatus and build again with new filter support
15	Lipid solution takes too much force to push through extrusion membrane	The pores in the extrusion are very small, and it is difficult to pass liquid through	Consider alternative liposome synthesis approaches such as sonication or freeze–thaw
20	Size exclusion column matrix dries	There is no liquid in the matrix	Continue to reapply PBS if elution of liquid stops
25	Tyndall effect not observed	Liposomes have already eluted	Re-add the eluate to the size exclusion column and check eluate with laser pointer
57	No EVs bind to CD81 antibodies coated on plate	CD81 antibodies do not coat the bottom of the plate or binding with EVs is disrupted	Consider antibodies targeting alternative EV markers such as CD63 or CD9
79	No fusion of EVs and liposomes	PEG concentration is too low or direct liposome–EV interaction is required	Add more PEG or attach an antibody targeting an EV membrane marker to the liposome membrane

Reagents

General reagents

• PBS (Thermo Fisher Scientific, cat. no. BP24384)

• BSA (Thermo Fisher Scientific, cat. no. BP9700100)

• Tween20 (Thermo Fisher Scientific, cat. no. PI85113)

• PEG 8000 (Thermo Fisher Scientific, cat. no. AA4344322)

• Ethanol (Pharmco, cat. no. 111000200)

Liposome synthesis chemicals

• 1,2-Dimyristoyl-sn-glycero-3-phosphorylcholine (DMPC; Avanti Research, cat. no. 850345)

• Cholesterol (Thermo Fisher Scientific, cat. no. A11470.18)

• Polycarbonate membranes (Avanti Research, cat. no.:100 nm 610005, 200 nm 610006, 400 nm 610007, 800 nm 610009)

• Ethanol (Pharmco, cat. no. 111000200)

• Sephadex G-25 Fine (Cytiva, cat no. 17003201)

• Peptide desalting StageTip (CDS Empore SDB-RPS Extraction Disks, cat. no. 13-110-023)

• EASY-Spray HPLC C18 analytical column (Thermo Fisher Scientific, cat. no. ES902)

EV characterization reagents

• Tris(2-carboxyethyl)phosphine (TCEP; Thermo Fisher Scientific, cat. no. 77720)

• Iodoacetamide (Sigma-Aldrich, cat. no. I1149)

• Phosphotungstic acid (PTA; Sigma-Aldrich, cat. no. P4006)

• Vybrant DiI (Molecular Probes, cat no. V-22885)

• DiD (Molecular Probes, cat. no. V-22887)

RPA and CRISPR reagents

• RPA Kit (TwistDx, cat. no. TALQBAS01)

• gRNA (Integrated DNA Technologies Custom oligo Alt-R L.b. Cas12a crRNA)

• EnGen Lba Cas12a/Cpf1 (enzyme to cut fluorescent reporter) (New England Biolabs, cat. no. M0653T)

• ProtoScript II Reverse Transcriptase (New England Biolabs, cat no. M0368L)

• NEBuffer 2.1 (New England Biolabs, cat. no. B7202 (previously B6002S))

• Zymo Quick-DNA/RNA Viral Kit (Zymo Research, cat. no. D7020)

• DNA/RNA Shield (Zymo Research, cat. no. R1200)

• Nuclease-free water (Thermo Fisher Scientific, cat. no. 10-977-015)

• Magnesium acetate (MgOAc; Sigma-Aldrich, cat. no. M5661)

• Deoxynucleotide triphosphate (10 mM) (dNTP; Thermo Fisher Scientific, cat. no. R0192)

• Custom DNA oligos (fluorescent probe and primers) used in SARS-CoV-2 N gene RNA detection and standard (see table below) (Integrated DNA Technologies)

  Name	Sequence	Function
  RT-RPA/CRISPR
  RT-RPA-N-F	GGGGAACTTCTCCTGCTAGAAT	N-gene RPA primer
  RT-RPA-N-R	AGACATTTTGCTCTCAAGCTG	N-gene RPA primer
  gRNA-N	UAAUUUCUACUCUUGUAGAUCUGCUGC UUGACAGAUUGAAC	N-gene gRNA
  Fluorescent reporter	FAM-TTTTTTTTTTTT-BHQ	Fluorescent reporter for CRISPR–Cas12a
  Cloning
  2019-nCovN.FOR	TAGTCCAGTGTGGTGGAATTCTGCAGAT ATCAACAAGTTTAT GAGTGACAATGGAC CACAGAACCAGAG	Full-length N gene forward primer for Gibson assembly
  2019-nCovN.REV	TTGTCGAGCGGCCGCCACTGTGCTGGAT ATCAACCACTTTTT AATGGTGATGGTGAT GGTGAGCCTGGGTGCTGTCAGCAGAAC	Full-length N gene reverse primer for Gibson assembly
  RT-qPCR
  USCDC-N2-F	TTACAAACATTGGCCGCAAA	N-gene forward primer from the Centers for Disease Control and Prevention (CDC) Emergency Use Authorization (EUA) kit
  USCDC-N2-R	GCGCGACATTCCGAAGAA	N-gene reverse primer from the CDC EUA kit
  USCDC-N2-Probe	FAM-ACAATTTGCCCCCAGCGCTTCAG-BHP1	N-gene probe from the CDC EUA kit

ELISA reagents (western blotting reagents; Supplementary Information)

• H₂SO₄ (Sigma-Aldrich, cat. no. 7664-93-9)

• Tetramethylbenzidine (TMB; Thermo Fisher Scientific, cat. no. 34022)

• Goat anti-mouse horseradish peroxidase secondary antibody (HRP; Abcam, cat. no. ab6789)

• Recombinant SARS-CoV-2 N protein (SinoBiological, cat. no. 40588-V08B)

• 4–20% gradient sodium dodecyl-sulfate polyacrylamide gel electrophoresis gels (Bio-Rad, cat. no. 4561094)

• Nitrocellulose membranes (Thermo Fisher Scientific, cat. no. 88018)

• Ammonium bicarbonate (Sigma-Aldrich, cat. no. 09830)

EV isolation

• Penicillin–streptomycin (10,000 U/mL) (Thermo Fisher Scientific, cat. no. 15140122)

• Dulbecco’s modified Eagle medium with 10% FBS (% vol/vol) (DMEM; Gibco, cat. no. 11965092)

• FBS (Thermo Fisher Scientific, cat. no. A5670701)

• EV-depleted FBS (Thermo Fisher Scientific, cat. no. A2720801)

• Puromycin (Gibco, cat. no. A1113803)

• Trypsin (cell culture)

• 0.45 μm filters (Millipore, cat. no. SLHV033RB)

• Protease Inhibitor Cocktail (Roche, cat. no. 4693159001)

• Serum-free DMEM (Gibco, cat. no. 11965092)

  ▲ CRITICAL For EV collection from 293F cells.

• Heparinase I (New England Biolabs, cat. no. P0735S)

  ▲ CAUTION Protease; use gloves.

Antibodies

• Anti-CD81 murine monoclonal antibody (Invitrogen, cat. no. MD5-13548)

• Anti-CD9-biotin rabbit polyclonal antibody (Invitrogen, cat. no. MA119485)

• Anti-SARS-CoV-2 N protein antibody (Sino Biological, cat. no. 40143-MM05)

• Goat anti-mouse-HRP secondary antibody (Jackson Immuno Research, cat. no. 115-035-003)

Equipment

• Extruder kit (Avanti Research, model no. 610000), includes 0.8 μm, 0.4 μm and 0.2 μm polycarbonate membranes (Avanti Research, model no. 610009/610007/610006)

• Fluorescent plate reader (Tecan, model no. Infinite 200 Pro)

• Glass size exclusion column (The Lab Depot, model no. CG-1187-03)

• Ultracentrifuge (Beckman Coulter, model no. Optima XPN-100)

• Vortex mixer (Thermo Fisher Scientific, model no. 02-215-450)

• Transmission electron microscope (JEOL, model no. JEM-1400)

• Nanoparticle tracking analyzer (NanoSight Malvern Panalytical, model no. NS300)

• UV transilluminator (UVP, model no. 95-0442-01)

• Transmission electron microscope (JEOL, model no. JEM-1400)

• Laser pointer

• Liquid chromatography–mass spectroscopy equipment (Supplementary Information)

• GraphPad 10.2.1 software

• Fiji 1.54k 15 software

• Design and Analysis 2.50 (real-time qPCR) software (Thermo Fisher Scientific)

• Western blot gel image (BioRad)

• ELISA and liposome fusion plate: 96-well high binding standard ELISA microplates (Greiner, cat. no. 655081)

• 15 mL 100 kDa centrifugal units (Amicon, cat. no. UFC910008)

• Au-flat 300 Mesh EM grid (EMS, cat. no. AUFT306-05)

Reagent setup

• PBS–Tween (PBST): make PBS with 0.5% Tween-20 (% vol/vol); used for washing ELISA plates to reduce nonspecific binding. PBST can be stored at room temperature for up to a month

• ELISA blocking buffer: make 1% (wt/vol) BSA in PBST; used to block nonspecific binding sites on ELISA plates. The blocking buffer needs to be stored at 4 °C for up to a month

• cryo-TEM staining solution (freshly prepared): make 2% PTA, pH 7.0, % wt/vol); use for staining samples for TEM analysis

Equipment setup

Extruder assembly

• Obtain Avanti Research Extruder set with holder/heating block (model no. 610000) and 0.8 μm, 0.4 μm and 0.2 μm polycarbonate membranes

• Place the two white internal membrane supports of the extruder upright so the O-rings are facing upward

• Wet one filter support by dipping briefly into deionized water. Use tweezers to handle the filter supports and polycarbonate membranes

• Place the wet filter support within the diameter of the O-ring on one of the white internal membrane supports

• Repeat the previous two steps with another filter support and the other internal membrane support

• Place one of the internal membrane supports with filter support into the extruder outer casing with the O-ring facing upward

• Insert the polycarbonate extrusion membrane onto the O-ring of the internal membrane support in the extruder outer casing

• Insert the other internal membrane support into the extruder outer casing with the O-ring facing downward. Do not twist when inserting the second internal membrane support; this will disrupt the polycarbonate membrane

• Add a Teflon bearing to the retainer nut. Twist the retainer nut onto the threaded end of the extruder outer casing

Extrusion and heat block apparatus assembly

• Place the Avanti heating block onto the hot plate, set to 37 °C

• Load the rehydrated lipid sample into one of the Avanti gas-tight syringes

• Insert the tip of the gas-tight syringe with lipid solution into one end of the extruder

• Insert the tip of the empty gas-tight syringe into the other end of the extruder

• Add the entire extruder and connected syringes onto the Avanti heating block on the hot plate

• Wait 5 min for lipid sample to reach 37 °C

• Gently push the plunger of the syringe with the lipid sample completely to the other syringe

• Gently push the plunger of the other syringe (now filled with lipid solution) until all solution has passed to the other syringe

• Repeat the previous two steps 10 times so that the lipid solution passes through the membrane 20 times

• Remove the filled syringe and save the solution in an Eppendorf tube for size exclusion

Procedure

Standard EV isolation from plasma or cell culture

● TIMING 5 d

• 1Transfect 293T cells with the pLenti-CMV-N plasmid using Lipofectamine 3000 (Supplementary Information).
• 2Harvest EVs 72 h post-transfection.
• 3
  To isolate EVs from cell culture, follow option A. To isolate EVs from plasma, follow option B.

  1. Cell culture EVs

     1. Seed 293T cells in serum-free DMEM for 48 h.
     2. Collect and centrifuge conditioned medium at 2,000g (30 min) and 10,000g (45 min), and filter the resulting supernatant through a 0.45 μm filter to remove remaining non-EV particulates.
     3. Centrifuge this filtrate at 100,000g (4 °C, 3 h) and resuspend its pellet in PBS.
     4. Quantify the particle numbers using NanoSight (Steps 42–46), and dilute EVs to 10⁴–10⁷ EVs/mL for the standard curve.
  2. Plasma EVs

     1. Dilute 1 mL plasma with 9 mL PBS and centrifuge this sample at 20,000g (45 min).
     2. Filter this supernatant through a 0.45 μm filter and then centrifuge this filtrate at 100,000g (4 °C, for 3 h).

Liposome lipid preparation

● TIMING 1 h 15 min

• 4Add 15.9 mg DMPC to 1 mL ethanol and vortex to mix.
• 5Add 9.1 mg cholesterol to 2 mL ethanol and vortex to mix.
• 6Pipette 200 μL of the cholesterol solution into the 1 mL DMPC solution and vortex to mix.
• 7Air-dry the DMPC–cholesterol solution until a thin, white film forms on the tube.

RT-RPA–CRISPR reagent preparation

● TIMING 20 min

• 8
  Add the following reagents (volumes needed for 20 samples) to a chilled 1.5 mL Eppendorf tube to generate the RPA–CRISPR or RT-RPA–CRISPR reaction mix and keep the reagents on ice:

  Reagent	Volume (μL)
  2× reaction buffer	150
  dNTP (10 mM)	12.8
  10× Basic E-mix	30
  Forward primer (10 μM)	7.2
  Reverse primer (10 μM)	7.2
  20× Core Reaction Mix	15
  MgOAc	15
  ProtoScript II Reverse Transcriptase (RT-RPA–CRISPR only)	6
  NEBuffer 2.1	300
  Fluorescent Probe (100 μM)	15
  gRNA (100 μM)	1
  EnGen Lba Cas12a (Cpf1)	1
  Total	RPA–CRISPR: 554.2
  	RT-RPA–CRISPR: 560.2

  ▲ CRITICAL STEP Reagent volumes for 20 samples are the minimum amount. However, ifmore sample testing is needed, the reagent amount can be increased (up to a total of 50 samples). Reagents can be used as a CRISPR assay without loading to liposomes, and the assay can be started with 23 μL mixed reagents with 2 μL (1~10 ng) template.

  ◆ TROUBLESHOOTING

Liposome loading

● TIMING 1 h

• 9Add 560 μL RPA–CRISPR mix and 1.44 mL PBS to the previously prepared air-dried lipid tube with a thin, white film.

  ◆ TROUBLESHOOTING

• 10Tape the tube to a vortex and vortex for 30 min.
• 11While vortexing this sample, set up the extruder kit apparatus (‘Equipment setup’).
• 12Use the syringe to take up 1 mL of RPA–CRISPR–PBS solution and lipid mixture.
• 13Place the syringe onto the extruder kit apparatus with an empty syringe on the opposite side.
• 14Slowly push the syringe to force the solution through the three different sizes of membrane within the extruder kit apparatus, from the largest to the smallest pore size (800 nm membrane, then 400 nm membrane and then 200 nm membrane), pushing the solution 20 times through each size ofmembrane (60 times in total).

  ▲ CRITICAL STEP Using a membrane pore size <200 nm is not recommended; it will produce more fragments rather than intact liposomes.

• 15Repeat Steps 13–14 until all the RPA–CRISPR–PBS solution is processed.

  ◆ TROUBLESHOOTING

Liposome purification

● TIMING 30 min

• 16Obtain a 100 mL clean glass gel filtration column.
• 17Mix 7 g Sephadex G-25 Fine and 70 mL PBS. More G-25 and PBS slurry can be made and added if needed.

  ■ PAUSE POINT Isolated liposomes in PBS can be stored at −20 °C for up to 3 months without freeze–thaw cycles.

• 18Allow Sephadex to swell in PBS.
• 19Pour this mixture into the size exclusion column and allow the Sephadex to settle as the PBS drains from the column.
• 20Add PBS to the packed column and allow it to pass through the column.

  ▲ CRITICAL NOTE Add PBS as needed before use to ensure that the column does not dry out. All liquid additions should be performed slowly and against the side ofthe column to avoid disturbing the gel surface.

  ◆ TROUBLESHOOTING

• 21Add the entire volume from Step 20 to the column after the PBS has drained to the top of the packed gel surface. No PBS layer should be visible above the gel when adding the liposome sample.
• 22Allow the column to drain to the gel surface, then add PBS to the column.
• 23Collect the eluate as sequential 200 μL fractions in 1.5 mL Eppendorf tubes.
• 24Shine the laser pointer through each eluate fraction as it is collected to detect the Tyndall effect. Collect all fractions displaying this effect in a 5 mL Eppendorf tube. This tube contains your liposomes. There should be no substantial particle loss (<10%) after washing three times with PBS (Extended Data Fig. 2). The particle numbers can be assessed by NanoSight in Steps 42–46.
• 25Add all eluted fractions (4–5 mL) into a 15 mL 100 kDa Amicon Ultra Centrifugal Filter, and spin the filter at 4,000g for 15 min.

  ◆ TROUBLESHOOTING

• 26Insert a 1 mL pipettor into the bottom of the filter and withdraw the remaining liquid using a side-to-side sweeping motion to ensure total recovery.

ELISA to check for EV-specific surface markers

• 27Add 50 μL/well of diluted anti-CD81 (1 μg/mL in PBS).
• 28Incubate overnight at 4 °C.

  ■ PAUSE POINT Coated plates can be stored dry at 4 °C for up to 1 week.

• 29The next day, wash three times with PBST (300 μL/well).

  ▲ CRITICAL STEP Ensure thorough washing after each step to reduce background.

• 30Add 200 μL/well of 1% BSA/PBS (wt/vol). Incubate for 1 h at room temperature (22–25 °C) or overnight at 4 °C.
• 31Wash three times with PBST (300 μL/well).
• 32Add 50 μL/well of purified EVs. Incubate for 1 h at room temperature.
• 33Wash three times with PBST (300 μL/well).
• 34Add 50 μL/well of biotinylated anti-CD63 (1:1,000) secondary antibody. Incubate for 1 h at room temperature. Secondary antibodies can target other EV markers (for example, CD9 or CD81) or cell-specific proteins (for example, EpCAM for pancreatic cancer cells).
• 35Wash three times with PBST (300 μL/well).
• 36Add 50 μL/well of poly-HRP (1:5,000). Incubate for 30 min in the dark.
• 37Wash three times with PBST (300 μL/well).
• 38Prepare TMB by dissolving 1 tablet in 1 mL DMSO + 9 mL Buffer-A + 2 μL H₂O₂. Prepare TMB fresh and use within 1 h.
• 39Add 100 μL of TMB to each well. Monitor blue color development (5–15 min).
• 40To stop the reaction, add 50 μL/well 1.25 mol/L H₂SO₄ (solution turns yellow).
• 41Measure OD₄₅₀ using a plate reader.

Quantification of particle numbers and size distribution with NanoSight

• 42Dilute EV or liposome samples in PBS to a final volume of 1 ml and mix gently.
• 43Load each sample into a 1 mL syringe.
• 44Wash the NanoSight channel with filtered PBS (20 nm), then blow the solution out by inputting 1 mL/s air flow.
• 45Connect the sample syringe to the machine and import it into the detection channel until the solution reaches 200 μL.
• 46Set the detection settings according to the manufacturer’s software manual (NanoSight NS300 User Manual). For our experiments, the sample is captured three times, and each capture duration is 60 s.

Characterization ofEV or liposome morphology with TEM

• 47Dilute the liposome or EV samples to a final concentration of ~8.45 × 10⁹ vesicles per μL in 2% PTA (wt/vol, pH 7.0).
• 48Spot 20 μL of the sample onto a carbon-coated grid and allow it to adhere for 20 min.
• 49Rinse the grid with distilled water and stain with 2% PTA for 1 min.
• 50Dry the grid at room temperature (22–25 °C).
• 51Use an FEI TECNAI F30 transmission electron microscope operating at 300 kV to capture images of the liposomes, EVs and fusion products.

(Optional) cryo-TEM imaging of EVs

• 52Prepare Au-flat 300 Mesh EM grid with lacey carbon.
• 53Pipette 10 μL of each EV sample on the grid.
• 54Blot the excess sample once between 1 s and 2 s with filter paper.
• 55Plunge the grid into liquid ethane kept in equilibrium with solid ethane.

  ■ PAUSE POINT Store the grid after vitrification at room temperature (22–25 °C) for up to 1 week until use.

• 56Image the grid using an FEI Tecnai electron microscope operated at 120 kV. Images are recorded at ×52,000 magnification.

CRISPR liposome fusion assay

● TIMING 8 h

• 57Pipette 100 μL 1 μg/mL anti-CD81 antibody into 96-well ELISA plate wells and incubate at 37 °C for 2 h.

  ◆ TROUBLESHOOTING

• 58Pipette 200 μL PBST three times to wash wells.
• 59Pipette 100 μL blocking buffer (1% BSA in PBST, wt/vol) into wells and incubate at 37 °C for 1 h.
• 60Pipette 200 μL PBST three times to wash wells.
• 61Pipette 100 μL plasma into wells and incubate at 37 °C for 2 h.
• 62Pipette 200 μL PBST three times to wash wells.
• 63Use a pipette to remove the wash solution from the well. The pipette tip can be added to a plastic tube attached to a vacuum to make this step faster.
• 64Mix 500 μL of liposome mixture with 500 μL PBS.
• 65Add 50 μL of diluted liposomes into each well.
• 66Add 50 μL of PEG 8000 into each well.
• 67Wrap plate in foil and incubate at 37 °C for 2 h.
• 68Read fluorescence intensity using a plate reader with an excitation wavelength of 480 nm and emission wavelength of530 nm.

FRET analysis for fused liposome–EV

• 69Resuspend 2 × 10⁸ EVs in 1 mL PBS containing 5 μL Vybrant DiI (donor) and 5 μL DiD (acceptor).
• 70Incubate at room temperature (22–25 °C) for 20 min.
• 71Filter the DiI-labeled (1 μM) EVs three times through a 100 kDa centrifugal filter unit to remove free dyes.
• 72Mix the DiI-labeled EVs with liposomes (2 × 10⁸ or 2 × 10⁹ liposomes) and perform the fusion reaction as described in Steps 71–75.
• 73Measure the fluorescent signals using a SpectraMax iD5 plate reader with excitation at 480 nm and emission spectra recorded from 525 to 750 nm.
• 74Label liposomes with DiD (1 μM).
• 75Incubate at 37 °C for 1 h. Measure FRET decay (excitation wavelength 549 nm/emission wavelength 665 nm).

Fluorescence detection and data analysis

• 76Use a SpectraMax iD5 plate reader to measure fluorescence at excitation/emission wavelengths of 485/530 nm.
• 77
  Perform the following four readings for each sample:

  1. Before liposome addition
  2. After liposome addition
  3. Baseline (immediately after PEG addition)
  4. After 2 h of incubation
• 78Compare the fluorescence intensity of positive and negative samples to determine the presence of SARS-CoV-2 RNA (Fig. 5j).
• 79Calculate the fold change in fluorescence intensity for each sample relative to the baseline.

  ◆ TROUBLESHOOTING

Fig. 5 |. Characterization of liposomes and representative fusion assay results.

a–d, Liposomes prepared using a membrane extruder. e–h, Liposomes prepared via sonication. Visualization of the Tyndall effect indicating the presence of liposomes: under ambient light (a,e) and under laser illumination (b,f), showing light scattering by colloidal particles. Liposomes in a–d are prepared using a membrane extruder, and those in e–h are generated via sonication. cryo-TEM images of synthesized liposomes, displaying uniform spherical vesicles with an average size of 100–200 nm (c, g). Size distribution of liposome particles measured by NTA, indicating a monodisperse population centered at 150–200 nm (d, h). i, Standard curve for EV–liposome fusion quantification. j, Representative fluorescence signals from liposome fusion CRISPR assays. Fusion was evaluated under various conditions, with an enhanced signal observed in the presence of PEG over time (0 min versus 120 min). Bar graphs display mean ± s.d. values. a.u., arbitrary unit.

Troubleshooting

Troubleshooting advice can be found in Table 2.

Timing

Step 1–3, standard EV production (optional): 4~5 d

Steps 4–7, liposome synthesis: 1 h 15 min

Step 8, cargo (RPA–CRISPR reagents) mix: 20 min

Steps 9–15, cargo loading: 1 h

Steps 16–26, liposome purification: 30 min

Steps 27–56, EV characterizations: 4~5 h

Steps 27–41, ELISA: 4~5 h

Steps 42–46, NanoSight: 1~2 h

Steps 47–56, TEM/SEM/cryo-EM: 1~2 h

Steps 57–79, CRISPR liposome–EV fusion assay: 8~10 h

Steps 57–68, fusion assay: 8 h

Steps 69–75, FRET: 1~2 h

Steps 76–79, data collection and analysis: 0.5~1 h

Anticipated results

The synthesis of liposomes using a formulation of 15.9 mg DMPC and 9.1 mg cholesterol in 1 mL ethanol, followed by extrusion through 0.1 μm polycarbonate membranes, is expected to yield ~8.5 × 10⁹ liposomes per mL after purification via gel filtration and ultracentrifugation. This provides sufficient material for subsequent fusion with EVs and CRISPR-based RNA detection for 20 assays. However, liposome aggregation or incomplete extrusion during processing may reduce yields and necessitate adjustments to lipid ratios or extrusion cycles. Liposomes generated in this process are expected to have a mean diameter of100~200 nm (Fig. 2a–d). Other ways to synthesize liposomes, such as sonication, can produce a similar size of liposomes⁴¹. Although the Tyndall effect is similar to that of membrane extruders (Fig. 2e,f), the morphology (Fig. 2g) and yield (Fig. 2h) of sonication are inferior to membrane extruders. More importantly, a 200 nm membrane extruder provides better stability of synthesized liposome, which is stable over 10 d in 4 °C after production (Fig. 2d); however, liposomes produced by sonication or extrusion through a 100 nm membrane are almost undetectable after storage (Fig. 2h). Reproducible liposome size distributions are critical for consistent fusion activity with their target EV populations, and deviations from this size range could indicate suboptimal lipid hydration or extrusion conditions.

Anticipated results for high-quality EV preparations include:

• NTA: monodisperse particle population with mean size 100–150 nm and concentration ≥1 × 10¹⁰ particles/mL

• TEM: characteristic cup-shaped morphology without substantial protein aggregates or debris

• ELISA for EV markers (CD81, CD9, TSG101): positive signal above background, confirming EV identity

• ELISA/western blot for ApoB and ApoA1: negative or near-background signals, indicating minimal lipoprotein contamination

Liposome loading efficiency for the RT, RPA and Cas12a reagents is projected to exceed >85% and should be validated by fluorescence intensity measurements of the FAM probe after liposome–EV fusion. High loading efficiency ensures robust amplification and detection of target RNA. Low fluorescence signals detected with positive controls may indicate poor reagent encapsulation efficiency or inactivation or degradation of enzyme reagents during liposome synthesis, requiring optimization of the liposome loading method, purification protocol or storage conditions. Liposome morphology, as confirmed by TEM, should reveal spherical vesicles with intact membranes and minimal aggregation. Irregular shapes or membrane disruptions observed in TEM images may suggest lipid oxidation or improper extrusion parameters. Critical benchmarks when troubleshooting liposome performance issues include >85% packaging efficiency, <5% leakage rates during storage and >80% high EV fusion efficiency, as quantified via FRET dequenching assays using DiI/DiD-labeled EVs¹⁷. Failure to meet these benchmarks, even after subsequent optimization, may necessitate re-evaluating the quality of the lipid reagents.

Sample wells coated with CD81-specific antibodies are expected to capture >95% of CD81-positive EVs, which covers 70% of total EVs from plasma samples, as determined by EV reductions in postcapture supernatants as measured by NTA results¹⁷. This high capture efficiency is necessary to maximize the number of EVs available for liposome fusion. EV capture rates below 80% indicate potential issues with antibody activity or the surface conjugation step. LoDs for RNA targets in the general EV population are anticipated to reach ~10 copies/μL based on our SARS-CoV-2 results, but this should be validated for new RNA targets via serial dilution experiments and fluorescence calibration curves. High LoD values may indicate inefficient EV and liposome fusion, suboptimal primer design or reagent instability.

Cas12a activity is expected to generate detectable fluorescence signals within 10–15 min at 37 °C, with signal intensity plateauing after 2 h for low-abundance targets. Delayed signal detection or reduced signal intensity at the reaction plateau can indicate attenuated RT, RPA or Cas12a activity, primer or gRNA integrity or concentration issues, or reaction temperature fluctuations.

The procedure is specific to RNA in EV, not surface or cell-free RNA (Extended Data Fig. 3). Successful liposome–EV fusion assays should yield fluorescent signal in proportion to target abundance, as we observed by linear regression analysis of spiked SARS-CoV-2 RNA samples (Fig. 2i). Plasma samples from individuals with a confirmed disease or conditions are expected to exhibit significantly higher fluorescence signals than those of healthy controls, with minimal overlap between groups (Fig. 2j). Specificity should be confirmed by analyzing samples from disease controls who exhibit similar symptoms but are diagnosed with alternate conditions, or by analyzing EVs engineered to contain or spiked with high levels of related targets. These off-target control samples should yield fluorescence levels comparable to negative controls, and observed cross-reactivity may require bioinformatic analysis and redesign of the assay primers or gRNAs to reduce their potential recognition of related sequences. Nonspecific fluorescence signals in negative controls should be reproducibly <10% of positive control values for low-concentration samples, and failure to achieve this may require re-evaluating the quenching efficiency or integrity of the reporter and potential autofluorescence from contaminants in the analyzed specimens, which may require the use of an alternate reporter dye-quencher system.

If liposome–EV fusion fails to generate detectable signals, potential issues include incomplete antibody-mediated EV capture, liposome aggregation or reagent degradation during storage. Re-optimization of PEG 8000 concentrations during fusion or extension of incubation times may improve results. For assays targeting low-abundance NA targets, affinity-based enrichment may enhance sensitivity. For example, complementary single-stranded DNA capturing is broadly used for low-abundance NA enrichment, following NGS or direct detections^71,72.

Overall, the protocol is expected to provide a reproducible and sensitive means to detect an NA target in plasma EVs (for example, SARS-CoV-2 RNA), where liposome characteristics, reagent loading efficiency and antibody capture performance serve as critical determinants of success. Iterative refinement of the reagent–liposome synthesis procedure and reaction conditions may be required to refine the performance and robustness of new assays, but adherence to QC and troubleshooting procedures should allow rapid resolution oftechnical challenges. Future research should focus on scaling up clinical applications, improving stability and reproducibility, and integrating multiomics approaches to uncover novel disease mechanisms.

Extended Data

Extended Data Fig. 1 |. EV surface marker and isolation comparisons.

(a) 100 μL of serum from a healthy donor was processed using three approaches, ultracentrifugation, anti-CD81 antibody-conjugated magnetic beads (CD81 capture), or anti-CD9 antibody-conjugated magnetic beads (CD9 capture) to isolate EVs, and protein yield was quantified by Bradford assay. (b–d) ELISA-based semi-quantification of EVs from 100 μL serum of SARS-CoV-2 swab-positive individuals (COVID-19) or healthy donors (Healthy) was performed under different capture/detection conditions: (b) no capture antibody with detection by anti-CD81 antibody, (c) capture with anti-CD9 antibody and detection by anti-CD81 antibody, and (d) capture with anti-CD81 and anti-CD9 antibodies.

Extended Data Fig. 2 |. Evaluation of washing step for liposome purification.

Bar graph showing the concentration of particles per milliliter measured after different liposome purification steps: no wash, after one wash with PBS (Wash 1), and after two washes with PBS (Wash 2). Liposome concentration was determined by NanoSight nanoparticle tracking analysis (NTA). Each bar represents the mean of five replicates, with individual data points overlaid. Statistical significance was determined by one-way ANOVA followed by Tukey’s multiple comparisons test. A significant reduction in particle concentration was observed after the first wash (p = 0.0046), with a further significant reduction after the second wash (p < 0.0001), indicating the effectiveness of sequential washes in removing unincorporated materials or loosely associated particles.

Extended Data Fig. 3 |. Comparison of Lysed EV and intact EV RNA detection using Liposome-EV fusion.

CRISPR liposome assay kinetics detected upon analysis of plasma aliquots (50 μL) from an individual with COVID-19 diagnoses based on positive nasal swab RT-qPCR results. EVs from the subject with positive nasal swab results were at 90 °C for 30 min then incubated with reagent-loaded liposomes (Lysed EVs + liposome), CRISPR-FDS reagents not packed into liposomes (Lysed EVs + free reagents), EVs without heating were incubated with reagent-loaded liposomes in the absence of PEG (Intact EVs + liposomes (no PEG)) or CRISPR-FDS reagents not packed into liposomes (Intact EVs + free reagents). Plasma aliquots from three individuals with and without COVID-19 also revealed CRISPR-FDS signal differences one hour after initiating the EV-liposome fusion reaction, which steadily increased until the two-hour reaction endpoint, while signal did not differ without fusion.

Supplementary Material

The online version contains supplementary material available at https://doi.org/10.1038/s41596-025-01317-7.

Fig. 4 |. A step-by-step decision-making guide for the liposome–EV fusion assay.

The assay begins by assessing EV abundance and quality in the sample, verified through ELISA (CD81 capture and CD63 detection). Upon confirming EV presence, liposome synthesis (DMPC and cholesterol) proceeds, with troubleshooting tips for air-drying, extrusion and size exclusion column handling. The workflow includes checkpoints for verifying RPA amplification and CRISPR–Cas12a fluorescence generation. Downstream troubleshooting addresses EV capture, fusion efficiency (PEG concentration or antibody-mediated targeting) and signal threshold determination using positive and negative EV controls. This schematic ensures a systematic approach to identifying and resolving common issues in liposome–EV fusion experiments. PAM, protospacer adjacent motif.

Key points.

• By fusing reagent-loaded liposomes with extracellular vesicles in patient samples, CRISPR-based amplification and detection occur inside the vesicles, allowing sensitive identification of low-abundance biomarkers while preserving extracellular vesicle integrity.

• The modular design supports targeting of specific extracellular vesicle subpopulations via surface modifications, enabling disease-specific biomarker analysis and drug delivery with minimal background noise and enhanced diagnostic and therapeutic precision.

Data availability

The data supporting the results in this study are available within the paper and its Supplementary Information. Source data are provided with this paper. The raw and analyzed datasets generated during the study are too large to be publicly shared, yet they are available for research purposes from the corresponding author on reasonable request.