Multifunctional biohybrid systems of plant-derived nanovesicles with inorganic nanomaterials

1 Introduction

Plant-derived nanovesicles (PDNVs) are nanoscale lipid-bilayer vesicles that may contain varying combinations of lipids, proteins, nucleic acids and plant metabolites. Their lipid-bilayer architecture and complex biomolecular composition have attracted increasing interest in recent years for construction of drug delivery systems and biohybrid materials (Zhang et al., 2026). However, terminology in this field remains inconsistent, and non-uniform nomenclature or overlapping terms are often applied to various PDNVs obtained through biologically and experimentally distinct routes, despite potential differences in their cellular origins, composition and formation mechanisms (Pinedo et al., 2021; Welsh et al., 2024; Zhang et al., 2026). To distinguish these categories, the term “plant extracellular vesicles (PEVs)” is reserved here for lipid-bilayer vesicles released by living plant cells into extracellular compartments, including the apoplast (Pinedo et al., 2021; Welsh et al., 2024; Zhang et al., 2026). By contrast, major PDNVs used in biomedical studies and nanomaterial-based biohybrid systems are recovered from plant tissues subjected to disruptive processing, such as grinding, pressing, juicing or homogenization. Such processing may yield heterogeneous samples containing naturally released extracellular vesicles, intracellular vesicles, organelle-derived membranes, membrane fragments and other membranous particles formed during processing. Consequently, the precise cellular origins or biogenesis pathways of the resulting PDNVs generally cannot be determined with certainty (Rutter and Innes, 2020; Pinedo et al., 2021; Zhang et al., 2026). To maintain terminological consistency, this review uses the term PDNVs collectively for the plant-derived nanoscale vesicles discussed herein. This nomenclature is adopted for the purpose of the present review and does not imply that all such vesicles share the same cellular origin or biogenesis pathway. The more specific term “PEVs” is used only when extracellular release from living plant cells is supported by experimental evidence and is relevant to the discussion. Vesicles reconstructed from extracted plant lipids are identified separately when their preparation route is important to the interpretation.

PDNVs may contain lipids, proteins, nucleic acids and plant metabolites, which together influence their membrane organization, surface chemistry, biological interactions and intrinsic bioactivity. These constituents also provide the molecular basis for interactions between PDNVs and recipient cells or microorganisms. Experimental studies have implicated PDNVs in the cross-kingdom transfer of small RNAs, mRNAs and defense-related proteins, through which they may contribute to plant-pathogen interactions. Likewise, membrane lipids, nucleic acids and metabolites associated with tissue-derived PDNVs have been reported to influence cellular uptake, gene expression and metabolic responses in recipient cells or microorganisms (Cai et al., 2018; Teng et al., 2018; He et al., 2021; Wang S. et al., 2024; Huang et al., 2025). These compositional and biological features support the development of PDNVs as delivery vehicles. However, when used alone, PDNVs may be constrained by limited loading capacity, insufficient structural stability, restricted functional tunability, and limited responsiveness to external stimuli (Cong et al., 2022; Fang and Liu, 2022; Jung et al., 2025). Hybridization with functional inorganic nanomaterials offers a strategy to address these limitations while preserving key features of the PDNVs and their associated bioactivity. Such integration may enhance cargo-loading capability and introduce catalytic, optical, magnetic and energy-conversion functions. In addition, fluorescent nanomaterials (e.g., carbon dots) can be incorporated into vesicles to facilitate vesicle tracking and characterization (Kowkabany and Bao, 2024). The performance of these biohybrids therefore depends not only on the properties of the individual components but also on material-membrane interaction, spatial organization, interfacial stability and assembly conditions. Accordingly, establishing stable and experimentally verifiable PDNVs–nanomaterial interfaces and defining the relationships between hybrid architecture and functional performance remain central priorities in this emerging field.

Most previous reviews of PDNVs have focused primarily on their biogenesis, isolation and purification, molecular composition, pharmacological activity, oral delivery and clinical translation (Cong et al., 2022; Fang and Liu, 2022; Jung et al., 2025). Up to date, only a few of original research and reviews have concerned about hybrid systems combining plant-derived or other extracellular vesicles with synthetic nanostructures (Ducrot et al., 2023; Mondal et al., 2023; Kong et al., 2024). Although these studies provide an important foundation for the engineering of PDNVs-based hybrid systems, this area remains comparatively underexplored. Beyond the terminological inconsistency discussed above, substantial variation in sample preparation, isolation, purification, structural characterization and reporting continues to hinder direct comparisons across studies. However, the formation and stability of hybrid interfaces are often insufficiently the material–membrane interface has rarely been used as a unifying analytical framework for examining how PDNVs interact with inorganic nanomaterials and how these interactions influence the biohybrid system performance.

Accordingly, this review focuses on biohybrid systems integrating PDNVs with functional inorganic nanomaterials. It first clarifies the terminology used for PEVs, PDNVs extracted from tissues and nanovesicles reconstructed from extracted plant lipids, before summarizing the isolation and purification methods, structural characterization techniques, membrane composition and surface properties relevant to biohybrid assembly. The principal engineering strategies used to regulate material–membrane interactions—including surface regulation, interfacial coupling, membrane hybridization and cargo loading—are then discussed. Subsequently, several representative biohybrid systems of PDNVs incorporated with metal-organic frameworks (MOFs), metallic nanoparticles, carbon dots, silicon-based and other inorganic nanomaterials are reviewed, with particular attention to how spatial organization, material–membrane interactions and interfacial stability influence functional performance and biological outcomes. Figure 1 illustrates the principal engineering strategies and four representative biohybrid systems of PDNVs with inorganic nanomaterials covered in this review. Finally, the major challenges facing the field are considered, together with priorities for future research and translation.

2 Isolation, purification and characterization of PDNVs

2.1 Isolation and purification methods of PDNVs

Consistent with the terminology outlined above, the sample preparation workflows including isolation and purification process used for naturally released PEVs and PDNVs extracted from tissues should reflect their distinct biological origins. PEVs are commonly recovered from apoplectic wash fluids, phloem sap, root exudates or plant-cell-conditioned media under conditions that minimize tissue disruption. By contrast, PDNVs extracted from tissues are typically obtained after washing and mechanically disrupting plant tissues by grinding, pressing, juicing or homogenization. Even when the resulting nanovesicles exhibit overlapping size distributions or similar zeta potential values, the two sample types should not be assumed to be compositionally or biologically equivalent (Welsh et al., 2024). As for Arabidopsis thaliana (A. thaliana), PDNVs isolated from disrupted leaves and PEVs recovered from apoplastic fluid differed in vesicle yield, protein composition and cellular uptake behavior (Liu Y. et al., 2020).

Once source-specific samples have been collected, most workflows proceed through two successive stages: preliminary clarification and concentration, followed by further separation to improve relative purity. During the first stage, stepwise differential centrifugation is commonly used to remove intact cells, coarse tissue debris and larger membranous structures (Rutter and Innes, 2017; Kocholata et al., 2022). The clarified supernatant is then concentrated by pelleting ultracentrifugation, ultrafiltration or another method selected according to sample volume and downstream requirements. Rotor type, relative centrifugal force and centrifugation time can substantially affect nanovesicles recovery and the composition of the resulting sample preparation. Although pelleting ultracentrifugation provides a direct way of concentrating nanoparticles, it does not selectively recover nanovesicles and may also co-sediment protein aggregates and other non-vesicular materials (Kocholata et al., 2022; Welsh et al., 2024). Ultrafiltration can rapidly concentrate large sample volumes, while tangential-flow filtration offers additional potential for scale-up (You et al., 2021; Kang et al., 2026). However, filter-membrane composition and molecular weight cut-off can affect particle recovery through non-specific adsorption and size-dependent retention (Vergauwen et al., 2017).

After concentration, various methods with greater separation selectivity are often employed to increase the relative purity of vesicle-enriched fractions. Density-gradient centrifugation separates particles according to buoyant density and can improve the selective enrichment of vesicular populations, although the procedure is time-consuming and requires fraction collection and subsequent sample re-concentration (Rutter and Innes, 2017; Welsh et al., 2024). Size-exclusion chromatography (Welsh et al.), by contrast, provides a relatively gentle size-based separation that can reduce soluble-protein contamination and preserve vesicle-associated biological activity (You et al., 2021; Sanchez-Lopez et al., 2024). However, non-vesicular particles with similar hydrodynamic sizes may still co-elute with vesicles, and the collected fractions are usually diluted (Welsh et al., 2024). Polymer precipitation is operationally simple and often produces high apparent particle recovery, but its limited separation specificity may result in protein contamination, while residual polymer can interfere with downstream characterization (Kalarikkal et al., 2020; Kocholata et al., 2022). It is therefore no single isolation or characterization method is sufficient on its own. Generally, no single isolation or purification method can simultaneously optimize recovery, selectivity, and preparation purity; and that sequential or complementary workflows are necessary. The main advantages and disadvantages of the above-mentioned methods are summarized in Table 1.

TABLE 1. Commonly used isolation and purification methods for PDNVs.

Practical applications of SEC to PDNVs further illustrate why relative enrichment should not be equated with absolute purification. In isolating cabbage-derived nanovesicles, SEC increased the particle-to-protein ratio, while PEV-enriched SEC fractions obtained from melon phloem sap retained proteasomal activity (You et al., 2021; Sanchez-Lopez et al., 2024). These findings show that SEC can increase the relative abundance of vesicular particles while preserving vesicle-associated biological activity, but neither outcome demonstrates the complete removal of non-vesicular contaminants. Accordingly, isolation, enrichment and increased relative purity are more appropriate descriptors than claims of absolute purity. In addition to the particle-to-protein ratio, potential co-isolated contaminants—including polysaccharides, pigments, soluble proteins, lipid aggregates and organelle-derived membranes—should be assessed according to their botanical sources and preparation methods (Rutter and Innes, 2020; Pinedo et al., 2021; Welsh et al., 2024).

2.2 Characterization techniques of PDNVs

Because the isolation procedures described above yield vesicle-enriched rather than chemically homogeneous preparations, isolation outcomes alone cannot establish membrane integrity, confirm vesicular identity or, in the case of PEVs, demonstrate extracellular release from living plant cells. Reliable characterization therefore requires complementary analyses of vesicle concentration and size distribution, membrane architecture, surface properties, molecular composition and potential co-isolated contaminants (Rutter and Innes, 2020; Pinedo et al., 2021; Welsh et al., 2024)

Particle-sizing methods provide an initial description of the recovered population but do not establish vesicular identity. Nanoparticle tracking analysis (NTA) estimates particle concentration and a number-based size distribution, whereas dynamic light scattering (DLS) reports an intensity-weighted hydrodynamic diameter and polydispersity index. Because neither technique is specific to membrane-enclosed vesicles and both may detect non-vesicular particles, their results should be interpreted together with direct imaging of membrane structure rather than used alone as evidence of vesicular identity (Bachurski et al., 2019; Phan et al., 2021).

Structural imaging can then be used to assess vesicle morphology and membrane architecture. Conventional transmission electron microscopy (TEM) reveals particle outlines and membrane-like boundaries, but fixation, dehydration and negative staining may collapse vesicles and generate shrinkage or cup-shaped artefacts. Cryogenic transmission electron microscopy (cryo-TEM) and cryo-electron tomography (cryo-ET) better preserve hydrated morphology and are more suitable for resolving membrane organization in two and three dimensions. Cryogenic imaging of apoplast-derived PEVs from Sorghum bicolor (S. bicolor) and A. thaliana revealed vesicles with single or multiple membrane layers and variable internal electron density, indicating substantial structural heterogeneity even among naturally released PEVs (Chaya et al., 2024). Atomic force microscopy (AFM) provides complementary information on surface topography and nanomechanical behavior, although substrate adhesion and probe–sample forces may deform soft vesicles (Welsh et al., 2024).

Surface characterization provides an additional, but strongly condition-dependent, dimension. Zeta potential is commonly used as an indicator of apparent interfacial charge and colloidal behavior. However, the measured value depends on pH, ionic strength, buffer composition, residual plant-derived solutes, particle concentration and temperature, and should therefore not be treated as a fixed property of a botanical source. Meaningful comparisons across studies require complete reporting of the dispersion medium, pH value, conductivity, particle concentration and measurement temperature (Welsh et al., 2024).

Omics analyses can identify lipids, proteins, and RNAs associated with PDNVs but cannot independently determine whether these molecules are in the lumen, embedded in the membrane, or exposed to the outer surface. RNA enclosure should be assessed using nuclease-protection assays combined with membrane disruption, whereas protein surface exposure may be examined through limited proteolysis or membrane-impermeable labelling. Fluorescence-tracing experiments should also exclude signals from free dyes, dye micelles, and dye aggregates before cellular fluorescence is interpreted as uptake of intact PDNVs or delivery of intra-vesicular cargo (Rutter and Innes, 2020; Welsh et al., 2024). Reliable characterization therefore depends on the integration of particle sizing, structural imaging, surface analysis, molecular profiling, and appropriate controls. Reliable characterization therefore depends on the integration of particle sizing, structural imaging, surface analysis, molecular profiling, and appropriate controls. No single technique can establish vesicle identity, membrane integrity, topology, composition, and the absence of relevant contaminants; complementary methods and controls are therefore necessary. The detailed information provided by commonly used characterization techniques and their technical limitations is summarized in Table 2.

TABLE 2. Commonly used characterization techniques for PDNVs.

2.3 Surface characteristics of PDNVs

Following assessment of membrane integrity and molecular composition, the interfacial behavior of PDNVs can be considered in relation to their membrane composition and surface properties. The lipid bilayer provides the structural boundary of the vesicle, while its composition influences membrane packing, fluidity, permeability and stability. These properties are shaped by lipid-headgroup chemistry, acyl-chain length and saturation, and the relative abundance of sphingolipids and sterols (Liu N. J. et al., 2020; Frallicciardi et al., 2022; Wang F. et al., 2025). Consequently, differences in membrane composition and physical state may affect whether exogenous lipids, amphiphilic ligands or nanomaterials adsorbed to the membrane, insert into the bilayer or induce membrane reorganization (Mendozza et al., 2019).

PDNVs from different botanical sources can exhibit distinct particle-size distributions and predominantly negative zeta potential values even when processed and analyzed under comparable conditions (Wang F. et al., 2025). These values are not fixed, however, and can change with the surrounding environment and storage history. As for PDNVs from ginger, changes in pH altered both particle size and zeta potential, including a shift from negative to weakly positive values under stomach-like acidic conditions (Zhang et al., 2016). More recent studies of PDNVs have further shown that storage conditions and buffer composition can influence their physicochemical and molecular stability (Derkus et al., 2026). Accordingly, PDNVs intended for construction of biohybrid systems should be re-characterized under the actual dispersion conditions used for integration of inorganic nanomaterials. Particle-size distribution, polydispersity, zeta potential and membrane integrity should be assessed together to determine whether the vesicles remain colloidally and structurally stable before subsequent interfacial engineering.

Membrane-associated proteins add a further layer of chemical and spatial heterogeneity to the PDNVs interface. When exposed to the outer surface, ionizable, polar and hydrophobic amino-acid residues can contribute to electrostatic interactions, hydrogen bonding and hydrophobic association, while accessible amino, carboxyl and thiol groups may provide potential sites for covalent coupling. However, bulk proteomic identification does not establish whether an individual protein is membrane-embedded, luminal, externally accessible or merely co-isolated. In Arabidopsis PEVs, density-gradient fractionation combined with protease-protection assays distinguished vesicle-associated proteins from extravesicular proteins, illustrating the need for direct topological validation rather than inference from proteomic abundance alone (Koch et al., 2025). Depending on membrane composition, surface charge, protein accessibility, material surface chemistry and assembly conditions, external components may associate with the vesicle surface, insert into the lipid bilayer, or promote membrane fusion and reorganization. These interactions provide the physicochemical basis for the interfacial anchoring and membrane hybridization strategies discussed in Chapter 3 (Huang et al., 2023; Moon et al., 2023; Yang et al., 2025).

Taken together, botanical source, isolation and purification processing history, membrane composition, structural heterogeneity, surface charge and membrane-associated proteins shape the interfacial properties of PDNVs. These features provide the biological and physicochemical basis for the surface engineering strategies discussed in the following chapter, including surface regulation, interfacial anchoring, membrane fusion and hybridization, targeting functionalization and cargo loading.

3 Surface engineering of PDNVs

During the construction of PDNVs–nanomaterial biohybrids, surface charge, steric environment, accessible functional groups, membrane composition and properties can be adjusted to regulate interfacial compatibility and assembly stability. However, such modifications may also alter membrane integrity, cargo retention and the surface presentation of native biomolecules; surface engineering should therefore aim to establish controllable material–membrane interactions while preserving the membrane properties required for the intended biological function (Li et al., 2026). Accordingly, this chapter generalizes surface engineering strategies of PDNVs.

3.1 Surface charge regulation

As discussed in Section 2.3, tissue-derived PDNVs commonly exhibit negative zeta potentials. However, their particle size, surface charge, abundance, and molecular stability vary with botanical source, isolation processing, pH, buffer composition, storage history, and surrounding medium. PDNVs intended for hybridization with inorganic nanomaterials should therefore be characterized under the actual dispersion conditions rather than only in water. Particle-size distribution, polydispersity, particle concentration, and zeta potential should be monitored over relevant time intervals to assess colloidal stability before and during hybrid construction.

Post-insertion of amphiphilic lipid–polymer conjugates provide a relatively mild approach to regulating the PDNVs interface. The amphiphilic lipid anchor partitions into the bilayer of nanovesicle, while the hydrophilic polymer chain extends into the aqueous phase and provides steric shielding. As for Asparagus cochinchinensis-derived nanovesicles, distearoyl phosphoethanolamine (DSPE)–polyethylene glycol (PEG) modification prolonged blood circulation and increased tumor accumulation without substantially altering vesicle size or morphology (Zhang et al., 2021). Actually, PEGylated mammalian extracellular vesicles have been previously reported displaying similar effect such as enhanced cell specificity and circulation time (Kooijmans et al., 2016). More recently, hydrophobic-insertion strategy has also been applied to grapefruit-derived nanovesicles, in which DSPE–PEG2000–maleimide was incorporated into the vesicle membrane and subsequently coupled to a thiolate targeting aptamer (Moon et al., 2023).

The extent of surface modification should be sufficient to improve colloidal stability without excessively masking membrane-associated biomolecules. In grapefruit-derived nanovesicles, insertion of DSPE–PEG2000 reduced the magnitude of the negative zeta potential, while subsequent aptamer conjugation produced further changes in surface charge without remarkable changes in particle size (Moon et al., 2023). Accordingly, modified PDNVs should be characterized under relevant actual conditions by particle-size distribution, polydispersity, zeta potential, morphology, membrane integrity and time-dependent stability. Where charged ligands, inorganic salts or nanomaterials are introduced, pH and ionic strength should also be reported because they can apparently influence surface charge and colloidal behavior. Measurements in water or other low-ionic-strength media should therefore not be used alone to infer stability under electrolyte- or protein-containing conditions (Li et al., 2026). Time-resolved measurements are particularly important because reversible association, irreversible aggregation and stable material attachment may produce similar changes in ensemble particle-size measurements.

3.2 Interfacial anchoring

External ligands or nanomaterials can be anchored to PDNV membranes through hydrophobic insertion, non-covalent interactions or covalent coupling. As above mentioned, DSPE–PEG2000–maleimide inserted into the grapefruit-derived nanovesicle membrane and subsequently coupled the terminal maleimide group to a 3′-thiol-modified aptamer through maleimide–thiol chemistry (Moon et al., 2023). This anchoring strategy separated membrane insertion from ligand conjugation and presented the reactive group at the distal end of a hydrophilic PEG spacer. Moreover, accessible amino and carboxyl groups can provide active sites for carbodiimide-mediated coupling, while thiol–maleimide and azide–alkyne reactions offer more selective conjugation routes; however, their applicability to PDNVs depends on the availability and accessibility of the corresponding surface functional groups (Li et al., 2024; Hong et al., 2026).

Smyth et al. introduced terminal alkyne groups onto surface-accessible amines of mammalian exosomes by carbodiimide-mediated coupling with 4-pentynoic acid and subsequently conjugated an azide-bearing fluorophore through copper-catalyzed azide–alkyne cycloaddition. Under the reported conditions, neither particle-size distribution nor association with recipient cells was measurably altered (Smyth et al., 2014). Although this strategy has not been directly demonstrated in PDNVs, related covalent reactions could in principle be adapted to display peptides, chelators, imaging probes or nanomaterial-binding ligands on their surfaces (Li et al., 2024; Hong et al., 2026). Such modification should be optimized with respect to reagent concentration, reaction time and grafting density, because excessive or poorly controlled functionalization may promote crosslinking or aggregation and may alter the accessibility of native membrane components. Coupling efficiency should therefore be evaluated together with particle size, membrane integrity and colloidal stability after removal of unreacted reagents.

Non-covalent association is operationally simple but can be sensitive to dilution, ionic screening and competitive adsorption. Changes in average particle size or zeta potential alone are therefore insufficient to demonstrate the formation of a defined material–membrane interface. After removal of unbound components, biohybrid systems should be evaluated using complementary approaches such as electron or fluorescence microscopy, elemental or spectroscopic mapping, repeated washing or competitive-displacement tests, and membrane-integrity measurements (Welsh et al., 2024). PDNVs-inorganic hybrid systems provide direct examples of the need for such structural verification; for instance, cucumber-derived nanovesicles combined with Cu-MOF nanodots were characterized by TEM and zeta potential measurements to support nanodots incorporation into the vesicular structure (Kong et al., 2024). Single-particle analysis can provide additional evidence by determining whether signals from the vesicular and introduced components are detected on the same individual particle, as demonstrated for extracellular vesicle (EV)–liposome hybrids (Ishikawa et al., 2022). Vesicle-only, material-only, simple mixture and disrupted-membrane controls should further be included to distinguish engineered association from aggregation or non-specific co-sedimentation.

3.3 Membrane fusion and hybrid vesicle formation

Membrane fusion alters the composition of the vesicular boundary rather than simply attaching an external component to the surface. For example, Pidnarulex (CX-5461)-loaded grapefruit PDNVs were fused with CCR6-expressing nanovesicles derived from engineered gingival mesenchymal stem cells by repeated co-extrusion. The resulting fusion vesicles retained bioactive features of the parental components and showed preferential accumulation in CCL20-rich inflamed skin through CCR6-mediated chemotactic targeting (Huang et al., 2023). Moreover, Lemon-derived nanovesicles have likewise been hybridized with 4T1 breast-cancer-cell membranes by sonication. Fluorescence Resonance Energy Transfer (FRET)-based lipid mixing, fluorescence colocalization, sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS–PAGE) and western blotting for E-cadherin, CD44 and CD47 supported membrane integration and retention of 4T1-derived membrane proteins, while subsequent cellular and in vivo studies demonstrated enhanced homotypic tumor targeting (Yang et al., 2025).

EV–liposome studies provide complementary fusion methods. PEG 8000-mediated fusion introduced both lipophilic membrane cargoes and hydrophilic aqueous cargoes into hybrid EVs (Piffoux et al., 2018). Repeated freeze-thaw treatment produced hybrid exosomes without a statistically significant change in mean size or size distribution, while the precursor lipid composition substantially altered cellular uptake (Sato et al., 2016). The viral mucogenic protein GP64 enabled pH-responsive fusion and single-particle analysis (Ishikawa et al., 2022). Cell-penetrating peptide Tat-PEG-lipid-mediated fusion depended on anchor length: C9 and C12 conjugates promoted fusion, whereas C14 mainly increased membrane association and aggregation (Sato et al., 2024).

Membrane fusion should not be inferred from a single fluorescence assay, because apparent lipid mixing may also arise from close membrane contact, probe transfer or hemi-fusion. Piffoux et al. combined a FRET-based lipid-mixing assay with complementary analyses of membrane and soluble-content transfer, cryo-TEM, imaging flow cytometry and Nycodenz density-gradient purification, and demonstrated PEG-mediated full fusion under optimized conditions while separating hybrid EVs from unfused liposomes (Piffoux et al., 2018). Protein profiling, structural imaging, removal of unfused precursors and dual-component single-particle analysis can provide additional evidence for membrane integration. Importantly, bulk fusion measurements may conceal substantial heterogeneity within the resulting population; recent single-particle analyses of EV–liposome hybrids showed marked differences between average fusion metrics and the proportion of individual particles containing both components (Mizenko et al., 2026). Functional comparisons should therefore include each parental vesicle population and an equivalent unfused mixture to distinguish fusion-dependent effects from those arising from simple co-administration. Processing variables should also be reported and standardized. In freeze–thaw-mediated EV–liposome fusion, both the number of freeze–thaw cycles and the lipid composition of the fusion partner influence membrane mixing and the properties of the resulting hybrids (Sato et al., 2016), while buffer and handling conditions should be reported because they can further affect EV integrity and comparability (Sato et al., 2016; Welsh et al., 2024).

Membrane wrapping or coating can extend vesicle engineering from membrane–membrane hybridization to the integration of inorganic nanomaterials. Kong et al. combined ultra-small Cu-MOF nanodots with cucumber-derived nanovesicles by sonication followed by incubation. TEM images confirmed nanodots (NDs) distributed within the vesicular structure, after which RGD peptide (Arg-Gly-Asp) was introduced through 1-(3-Dimethylaminopropyl)-3-ethylcarbodiimide (EDC)/ N-Hydroxy succinimide (NHS)-mediated amide coupling (Kong et al., 2024). The resulting NDs@EV-RGD platform combined Cu-MOF-mediated reactive oxygen species (ROSs) generation and glutathione depletion with the deformability and transdermal delivery capability of the PDNVs. In a non-plant system, Piao et al. coated gold nanocages with red-blood-cell membranes, yielding membrane-coated particles with improved colloidal stability, prolonged blood circulation and increased tumour accumulation (Piao et al., 2014). For both membrane encapsulation and coating, membrane continuity, orientation, residual free membrane and uncoated or unencapsulated inorganic particles should be assessed before biological effects are attributed to hybrid architecture. The material-specific functions of these inorganic components are discussed in Chapter 4.

3.4 Targeting ligands and cargo integration

Targeting functionalization can further tune the cellular and tissue interactions of engineered PDNVs. In ginger-derived nanovesicles, membrane insertion of DSPE–PEG–iRGD increased uptake by B16F10 melanoma cells and enhanced tumor accumulation in vivo (Wang F. et al., 2025). Grapefruit-derived nanovesicles were functionalized by inserting DSPE–PEG2000–maleimide into the membrane and subsequently coupling a 3′-thiol-modified aptamer, which increased uptake by brain endothelial cells (Moon et al., 2023). Similarly, Folic acid–PEG2000–cholesterol was inserted into ginger-derived nanovesicles to generate folate-functionalized vesicles that showed enhanced uptake by M1 macrophages through folate-receptor-mediated endocytosis and greater accumulation in inflamed joints in vivo (Han et al., 2025).

Ligand functionalization can be combined with cargo loading to generate multifunctional PDNVs. In lemon-derived EVs, heparin–cRGD was introduced onto the vesicle surface through EDC/NHS-mediated coupling before doxorubicin loading, producing PDNVs nanodrugs with activity against multidrug-resistant ovarian cancer (Xiao et al., 2022). Beyond such combined targeting and loading strategies, the choice of loading method depends strongly on the physicochemical properties of the cargo. For example, miR-184 was introduced into cabbage PDNVs using a lipid-based transfection reagent, whereas doxorubicin was loaded into red-cabbage-derived nanovesicles by passive incubation (You et al., 2021). Membrane fusion offers an alternative route by enabling the incorporation of both membrane-associated and aqueous cargoes (Piffoux et al., 2018). Loading performance should therefore be evaluated not only by loading or encapsulation efficiency, but also by cargo release, membrane integrity and colloidal stability (Li et al., 2024). In multifunctional systems, targeting ligands, therapeutic cargoes and nanomaterials may further compete for membrane area or accessible reactive sites. A staged assembly strategy, accompanied by physicochemical and functional characterization after each modification step, can help distinguish the contribution of individual engineering processes. Modification density should therefore be optimized to achieve the intended function while minimizing unnecessary perturbation of the PDNVs interface.

Surface and membrane engineering can expand the functional scope of PDNVs while retaining selected characteristics of the plant-derived membrane interface. In general, surface engineering strategies should provide sufficient interfacial stability and functionalization while minimizing unnecessary changes to membrane organization, integrity and biological activity. These principles provide a common framework for evaluating the interactions of PDNVs with MOFs, metallic nanoparticles, silicon-based nanomaterials, carbon-based nanomaterials and other inorganic components discussed in Chapter 4.

4 Biohybrid systems integrating PDNVs with inorganic nanomaterials

On the basis of the engineering principles outlined in Chapter 3, this chapter examines how PDNVs are integrated with functional nanomaterials to form biohybrid systems. Previous reviews of EV–nanomaterial hybrids have described several common approaches, including membrane coating, nanoparticle loading and surface coupling (Chan et al., 2022; Gui et al., 2025). In these biohybrid systems, inorganic nanomaterials can introduce multi-functions such as cargo loading or protection, catalytic activity, photothermal conversion, ROSs generation, magnetic responsiveness and optical imaging, whereas PDNVs provide a hydrophilic membrane interface and plant-derived biomolecular components that can influence colloidal behavior, material association and biological interactions (Gui et al., 2025). Importantly, these PDNVs can adopt different spatial organizations, ranging from direct surface or membrane association and membrane coating to co-integration within larger matrices. These configurations should not be regarded as equivalent, because material localization, membrane integrity and interfacial stability can influence cargo release, catalytic accessibility and biological performance. Accordingly, the following sections evaluate four representative PDNVs hybrid systems by material classification, with particular attention to the relationship between PDNVs and inorganic nanomaterials, material–membrane interactions and thus different functional performance. Representative PDNV–inorganic nanomaterial hybrid systems with different materials classes, component functions, and applications are summarized in Table 3.

TABLE 3. Representative PDNV–inorganic nanomaterial hybrid systems with different materials classes, component functions, and applications.

4.1 Integrating PDNVs with MOFs

MOFs are crystalline coordination materials constructed from metal-containing nodes and organic linkers. Their modular structures allow pore architecture, surface chemistry, coordination environment and degradability to be tuned for different biomedical functions. Nanoscale MOFs have therefore been widely investigated as nanocarriers for small molecules and biomacromolecules and as functional nanoplatforms for catalytic, photodynamic, sonodynamic and imaging applications (Abánades Lázaro et al., 2024; Wang et al., 2026). Their biological performance, however, also depends on colloidal and chemical stability, framework degradation, release of metal ions or organic linkers, and composition-dependent toxicity (Ettlinger et al., 2022). In PDNV-MOF hybrids, MOFs can serve as porous or degradable cargo carriers and as catalytic or energy-responsive functional domains, while PDNVs provide a hydrophilic biological interface that can influence the material dispersion and interaction with biological environment.

Wang et al. coated a zirconium (Zr)-porphyrin MOF with ginger-derived EVs to construct electrodynamic MOF@EV nano-sponges, as illustrated in Figure 2 (Wang Y. et al., 2024). The MOF, constructed from Zr-containing nodes and meso-tetra(4-carboxyphenyl) porphin (TCPP), acted as an electro-sensitizer that generated ROSs under an alternating electric field. PDNVs encapsulation shifted the surface charge of MOFs towards negative values, maintained particle stability in phosphate-buffered saline and prolonged blood circulation, thereby increasing accumulation at infected sites. In addition, the PDNV-derived membrane conferred toxin-neutralizing activity: the hybrid reduced the hemolytic and cytotoxic effects of proteins secreted by Staphylococcus aureus, while the MOF retained its field-responsive ROS-generating activity. The resulting nanoplatform therefore combined two spatially distinct functions, with the MOF providing electrodynamic antibacterial activity and the surrounding PDNV membrane contributing toxin sequestration and biological interfacing. In a subcutaneous S. aureus abscess model, MOF@EV-assisted electrodynamic therapy reduced bacterial burden and inflammatory responses and promoted tissue recovery. Comparisons among MOF, MOF@EV and the corresponding treatments with or without the alternating electric field further helped distinguish the contributions of membrane coating and electrodynamic activation.

Cui et al. coated tumor necrosis factor-α (TNF-α) siRNA-loaded zeolitic imidazolate framework-8 (ZIF-8) nanoparticles with ginger PDNVs to construct an oral delivery system for ulcerative colitis (Cui et al., 2024). Unlike the electrodynamic Zr-porphyrin MOF described above, ZIF-8 primarily functioned as a porous siRNA carrier, providing high loading capacity and facilitating lysosomal escape after cellular uptake. The ginger-PDNV coating contributed resistance to gastric acidity and preferential targeting of the colon and macrophages, while the endogenous 6-shogaol provided an additional anti-inflammatory component. In vivo, the hybrid reduced colonic TNF-α expression and promoted intestinal barrier repair, accompanied by partial restoration of gut microbial balance. The therapeutic response therefore reflected complementary contributions from the siRNA cargo, the ZIF-8 carrier and the ginger-derived nanovesicle rather than gene silencing alone.

ZIF-8-based platforms have also been adapted by varying the botanical source, therapeutic cargo and targeting ligand. Jiang et al. coated ZIF-8 nanoparticles carrying TGF-β1-targeting siRNA with turmeric-derived PDNVs for the treatment of liver fibrosis (Jiang et al., 2026). The resulting system improved siRNA stability and cellular uptake enabled pH-responsive release and lysosomal escape and reduced hepatic stellate-cell activation and collagen deposition. In another design, Saffar et al. coated DOX-loaded ZIF-8 with Cucurbita moschata-derived PDNVs and subsequently conjugated an epithelial cell adhesion molecule (EpCAM) aptamer to the outer surface, enhancing uptake by EpCAM-positive PC-3 prostate cancer cells (Saffar et al., 2025). These examples show how a ZIF-8 hybrid can be adapted to different pathological settings by changing the cargo and biological interface. However, as additional ligands and functional components are introduced, changes in cellular uptake should be interpreted together with alterations in surface charge, particle size and colloidal stability, rather than attributed to receptor recognition alone.

MOF–PDNV architectures are not restricted to conventional ZIF-8 core–shell carriers. Kong et al. integrated ultra-small Cu-MOF nanodots with cucumber PDNVs and subsequently introduced RGD surface functionalization (Kong et al., 2024). In such a MOF–PDNV hybrid, the Cu-MOF nanodots were distributed within the vesicular structure of PDNVs, providing photosensitizing ROSs generation and glutathione depletion, whereas the deformable vesicular structure of PDNVs facilitated transdermal delivery for hypertrophic-scar treatment. More recently, Gao et al. developed a reconstructed architecture in which a lycopene-loaded ZIF-8 core was surrounded by a ROSs-responsive shell containing unsaturated phospholipids, cholesterol and garlic PDNV components, with polymyxin B incorporated into the outer shell (Gao et al., 2026). Oxidative stress first de-stabilized the outer shell and promoted rapid polymyxin B release, whereas subsequent acidification degraded the ZIF-8 core and released lycopene and Zn²⁺, supporting later antioxidant, immunoregulatory and tissue-repair responses. Together, these biohybrid systems illustrate that PDNV can function either as a vesicular coating around a nanomaterial component or as one component of an artificially reconstructed multicomponent shell.

In addition, it is known that mammalian EVs also provide useful methodological precedents for the construction and evaluation of membrane-coated MOFs. Illes et al. coated drug-loaded MOF nanoparticles with exosome membranes through lipid fusion, demonstrating that a preformed porous framework could be combined with a cell-derived membrane while retaining intracellular cargo delivery (Illes et al., 2017). In a related design, macrophage-derived macrovesicles were used to coat methotrexate-loaded ZIF-8 and were further functionalized with a folate-containing lipid–PEG conjugate, resulting in prolonged circulation and enhanced accumulation in inflamed joints (Wang et al., 2022). These studies provide useful references for membrane coating, post-assembly characterization and component-control design. However, they should be regarded as methodological precedents rather than direct surrogates for PDNVs-based systems, because differences in membrane composition and biological origin may alter material association and interactions with recipient tissues.

Validation of reported PDNV–MOF systems have relied primarily on conventional TEM, changes in hydrodynamic size and shifts in zeta potential to support membrane-associated assembly. Although informative, these data are insufficient to characterize the fine structure and compositions of assembled hybrids or exclude the presence of free PDNVs, membrane fragments or uncoated MOFs. Future studies should therefore quantify assembly efficiency, remove and assess residual unbound components, and use complementary structural or single-particle analyses to determine the localization of the MOF relative to the vesicle membrane. The effects of framework degradation should also be considered, particularly whether released metal-containing species alter membrane integrity, lipid oxidation or the function of membrane-associated biomolecules. Component controls should include the MOF, PDNV, cargo where applicable, an equivalent unassembled mixture and the purified hybrid. Finally, stability should be evaluated in the media used for administration and functional testing, because phosphate-containing buffers, serum proteins and other coordinating species can substantially alter MOF degradation and cargo release (Luzuriaga et al., 2019).

4.2 Integrating PDNVs with metallic nanoparticles

Metallic nanoparticles can introduce optical, photothermal, catalytic and antimicrobial functions into PDNV-based biohybrids. Noble-metal nanostructures are particularly attractive because their physicochemical properties can be tuned through composition, size, morphology and surface chemistry. Gold (Au) nanostructures are widely investigated for plasmonic imaging and photothermal conversion, whereas silver (Ag) nanomaterials can exert antimicrobial effects through Ag⁺ release and surface-mediated interactions. Platinum (Pt)- and palladium (Pd)-containing nanomaterials provide additional catalytic and redox functions that can be exploited in therapeutic systems (Guo et al., 2024). These metallic nanoparticles also can influence the biological interface properties of PDNVs, including biomolecular adsorption, cellular uptake, dissolution or transformation, biodistribution and toxicity. In PDNV-based hybrids, metallic components may be associated with the vesicle surface or membrane, incorporated as preformed nanostructures, or formed through local metal-ion enrichment and in situ nucleation. The resulting spatial organization is therefore an important determinant of both material accessibility and membrane-associated biological function.

Qiao et al. reported Pd–Pt nanosheets conjugated to the surface of ginger-derived EVs to construct the anti-infective EV–Pd–Pt platform shown in Figure 3 (Qiao et al., 2022). Carboxyl-functionalized Pd–Pt nanosheets were coupled to accessible amino groups on the PDNV surface through amide-bond formation, while elemental mapping and X-ray Photoelectron Spectroscopy (XPS) provided additional evidence of successful conjugation. The Pd–Pt nanosheets exhibited electro-driven catalytic activity, generating ROSs under an alternating electric field, and their near-infrared photothermal response provided an additional therapeutic function. The PDNV component prolonged blood circulation and increased accumulation at infected sites. Importantly, lipid-reconstitution experiments showed that depletion of phosphatidic acid from ginger-derived lipids markedly reduced bacterial uptake, supporting a role for membrane lipids in the interaction of EV–Pd–Pt with Staphylococcus aureus and Escherichia coli. The combined electrodynamic and photothermal treatment reduced bacterial viability and disrupted S. aureus biofilms. Because the Pd–Pt nanosheets were synthesized before conjugation and subsequently attached to the PDNV surface, this system represents post-synthetic surface anchoring of a preformed metallic nanomaterial rather than in situ metal nucleation. This well-defined surface-conjugation strategy also illustrates how a PDNV membrane can enhance the biological accessibility of an otherwise preformed catalytic nanomaterial.

Huang et al. used ginger PDNVs as a biological scaffold for the in situ formation of Ag nanoparticles (AgNPs) and subsequently incorporated the resulting Ag–PDNV hybrid into a dissolvable microneedle patch for infected diabetic wounds (Huang et al., 2026). The PDNV component promoted the stabilization and uniform dispersion of AgNPs, while the AgNPs provided the principal antibacterial and antibiofilm activity. At the same time, ginger-derived biomolecular components retained antioxidant, anti-inflammatory and pro-angiogenic functions, providing a regenerative contribution complementary to the antimicrobial activity of AgNPs. The resulting Ag–PDNV microneedle reduced methicillin-resistant Staphylococcus aureus (MRSA) burden, disrupted mature biofilms and promoted neovascularization and collagen deposition in vivo. However, the reported imaging did not unambiguously establish whether individual AgNPs were located within the vesicle lumen, embedded in membrane-derived structures or associated with the outer surface. The metallic phase was therefore more cautiously described as forming in situ in close association with PDNVs. Further fractionation and single-particle elemental analysis would be useful not only for distinguishing intact Ag–PDNV hybrids from free AgNPs and residual vesicular material, but also for investigating how PDNV-associated biomolecules contribute to metal-ion binding, particle formation and subsequent release of Ag⁺.

In fact, previous studies using mammalian EVs illustrate several additional routes for integrating metallic nanomaterials with vesicular systems. Zhang et al. used EVs as templates for the surface growth of Au nanoparticles, producing nanopopcorn structures that combined doxorubicin delivery with near-infrared photothermal treatment (Zhang et al., 2019). Sancho-Albero et al. instead exploited the exosome biogenesis pathway by pre-incubating donor cells with PEGylated hollow Au nanoparticles, which were subsequently recovered within secreted exosomes at substantially higher loading efficiencies than those achieved by direct physicochemical loading (Sancho-Albero et al., 2019a). Hydrophobic Au nanoclusters have also been co-assembled with EVs into paraparticles for high-resolution three-dimensional uptake visualization (Kauscher et al., 2020). Other metals illustrate distinct loading chemistries: ultra-small Pt nanoparticles were generated in situ within isolated exosomes by reduction of a Pt²⁺ precursor, providing an alternative Pt-based therapeutic formulation (Sancho-Albero et al., 2022), whereas ultrathin Pd nanosheets were assembled directly inside cancer-derived exosomes and used for targeted biorthogonal catalysis (Sancho-Albero et al., 2019b). Together, these previous studies provide methodological precedents for surface-directed metal growth, biogenesis-assisted loading, paraparticle assembly and intravascular metal formation, although their application to PDNVs requires direct experimental validation.

The evaluation of metal–PDNVs biohybrid systems should consider both the chemical fate of the metallic component and its effects on the vesicle interface. Beyond morphological characterization, studies should quantify metal content and oxidation state and examine metal release and transformation under storage and biologically relevant conditions. These measurements should be complemented by assessment of membrane integrity, including lipid oxidation, membrane leakage and changes in the function or accessibility of membrane-associated proteins, to determine whether the intended optical, catalytic or antimicrobial activity is accompanied by substantial perturbation of the PDNV membrane. For systems involved in situ mineralization or reduction, rigorous purification and component-resolved controls are particularly important for distinguishing the contributions of vesicle-associated biomolecules from those of co-isolated plant metabolites to metal binding, nucleation and particle growth (Welsh et al., 2024).

4.3 Integrating PDNVs with silicon-based nanomaterials

Silicon-based nanomaterials encompass chemically and structurally distinct platforms, including elemental silicon, silica (SiO₂), organosilicon and related porous or hollow architectures (Chen et al., 2024). Among these materials, mesoporous silica nanoparticles (MSNs) are particularly attractive as cargo-bearing components because their pore size, surface chemistry, particle geometry and internal pore volume can be tailored for drug loading and controlled release of small molecules and biomacromolecules. Their structural versatility has supported applications in drug delivery, stimuli-responsive release, imaging and tissue engineering (Manzano and Vallet-Regi, 2025). However, these advantages should not be generalized across all silicon-based materials, because elemental silicon and SiO₂ differ in surface chemistry, dissolution and degradation pathways (Chen et al., 2024). In PDNV-based hybrids, the silicon-containing component can therefore serve different material functions depending on its composition and architecture, ranging from porous cargo storage to structurally defined support or protection, whereas the PDNV membrane forms the biological interface with the surrounding environment. The resulting behavior is consequently determined by both the intrinsic properties of the silicon-based material and its spatial organization relative to the PDNV membrane.

Mao et al. used customized large mesoporous silicon nanoparticles (LMSNs) as the inorganic core of a ginger PDNV-coated oral antibody-delivery system, as shown in Figure 4 (Mao et al., 2021). The enlarged pores enabled antibody infliximab (INF) loading of up to 61.3 wt% while reducing antibody aggregation. Ginger-derived exosomes were subsequently coated onto the INF-loaded LMSNs to form a core–shell nanocomposite. In this architecture, the LMSNs provided a high-capacity porous compartment for INF loading, whereas the PDNV coating improved gastrointestinal stability, colon-targeted delivery and intestinal epithelial permeability. The ginger-derived component also exerted an intrinsic anti-inflammatory effect associated with blockade of the NLRP3 inflammasome. Thus, such a hybrid system combined a structurally tailored silicon-based carrier with a biologically active plant-derived interface, allowing the inorganic core and PDNV coating to provide distinct but complementary functions.

A related oral design used hollow MSNs for a different material function. Wang et al. encapsulated ammonia borane within hollow MSNs and coated the resulting hydrogen (H₂)-generating core with ginger PDNVs to construct a biohybrid system (Wang X. et al., 2025). The hollow silica compartment provided space for the H₂ precursor and enabled acid-responsive H₂ generation in the gastrointestinal tract, whereas the ginger PDNV coating improved biocompatibility and contributed to modulation of the intestinal microbiota. In T2DM mice, MSN-PDNV-based hydrogen therapy reduced oxidative stress and inflammatory responses, ameliorated insulin resistance and liver steatosis, and inhibited pancreatic β-cell dedifferentiation. This nanoplatform also increased the abundance of Lactobacillus spp. and altered tryptophan metabolites, including indole and indole-3-acetic acid, which were associated with activation of the AhR/IL-22 pathway and improved intestinal-barrier function. Thus, unlike the antibody-loading LMSN system as above-mentioned, here the hollow MSNs functioned primarily as a reaction compartment for local H₂ generation, while the PDNV component contributed to the biological and microbiota-related effects of the oral nanoplatform.

Non-plant EVs and synthetic lipid-bilayer systems provide methodological precedents for constructing silica–membrane interfaces, rather than direct evidence for PDNV–silica hybrids. Dumontel et al. used tumor-derived EVs to coat amino-functionalized MSNs, producing an EV-associated outer layer and preferential internalization by cells corresponding to the EV source (Dumontel et al., 2023). In a synthetic lipid-bilayer system, MSNs were coated with a fusogenic bilayer bearing one coiled-coil lipopeptide, whereas the complementary lipopeptide was introduced into the target-cell membrane. This configuration promoted membrane-fusion-mediated cytosolic delivery of cytochrome c, with inhibitor experiments supporting membrane fusion as a dominant uptake route (Yang et al., 2017). Regarding the hybrid system combining INF-loaded large MSNs with nanovesicles isolated from ginger tissue, it was reported showing an improved gastrointestinal stability, colonic localization and epithelial permeability after oral administration in a murine colitis model (Mao et al., 2021). However, because the ginger-derived nanovesicles were obtained after tissue disruption, they should not be classified as naturally released PEVs, and the precise membrane–silica topology and interfacial stability were not fully established. Collectively, these studies indicate that MSNs can serve as a porous core for vesicular or synthetic membrane association, fusion-mediated uptake and antibody delivery, although the organization and release behavior of PDNV–silica hybrids require direct validation.

Future studies should establish whether MSNs are completely and continuously coated and whether the membrane layer occludes pore openings or alters cargo loading and release. For silica-containing nanomaterials, surface silanol groups may interact with proteins and membrane constituents, potentially affecting their conformation, accessibility or interfacial organization. The effects of sonication and extrusion on PDNV membrane integrity should therefore be evaluated together with loading efficiency and release behavior. Material transformation, degradation and clearance should also be assessed separately for elemental silicon, SiO₂ and organosilicon, because these materials follow distinct chemical and biological pathways. Regarding oral formulations, stability should be evaluated sequentially under simulated gastric and intestinal conditions and subsequently in mucus-containing media, with protein-containing media included relevant to subsequent biological exposure.

4.4 Integrating PDNVs with carbon-based nanomaterials

Carbon-based nanomaterials encompass graphene and its derivatives, carbon nanotubes, nanodiamonds, carbon dots, fullerene and other carbon-based nanostructures. Their physicochemical functions vary substantially with dimensionality, graphitized degree, defect structure, heteroatom doping and surface chemistry. Graphene oxide (GO), for example, provides a large two-dimensional nanosheet containing aromatic domains and oxygen-containing functional groups that facilitate adsorption and association with polymers and nucleic acids. Carbon dots, by contrast, are particularly attractive for their ultra-small size, quasi-spherical morphology, abundant oxygen- or nitrogen-containing surface groups, tunable photoluminescence in the range of ultraviolet-visible-near infrared and, depending on their carbon core and surface states, can also exhibit photodynamic, photothermal, redox-regulating and enzyme-mimicking activities (Azam et al., 2021; Krasley et al., 2024; Zhu et al., 2025). These properties support diverse applications ranging from bioimaging, phototherapy to antimicrobial treatment and oxidative-stress modulation (Magesh et al., 2022; Jiang et al., 2023; Qin et al., 2024; Zhu et al., 2026). It is noteworthy that currently reported configurations of PDNV-nanocarbon hybrids include polymer-mediated PDNV–GO assemblies as well as the co-integration of PDNVs and carbon dots within hydrogels, which are different from commonly seen MOF–PDNVs or metal–PDNVs biohybrids with direct interfaces. These architectures should therefore be distinguished according to the actual spatial relationship between the carbon nanomaterial, the PDNV membrane and any additional matrix.

Saroj et al. conjugated PDNVs extracted from neem, mint and curry leaves respectively with chitosan (CS) and PEGylated-GO (GP) to construct multicomponent vectors for delivering estrogen receptor α (ERα)-targeting siRNA to MCF-7 breast cancer cells (Saroj et al., 2024). The resulting EV–CS–GP conjugates combined the high-surface-area carbonaceous support of GO with the cationic character of CS and the biological interface provided by the PDNVs. Among the three botanical sources, neem-derived EV–CS–GP showed the highest cellular siRNA-delivery efficiency, which was attributed in part to hyaluronic-acid-mediated recognition by CD44 on MCF-7 cells. Because CS and GP can simultaneously influence nucleic-acid association, surface charge and colloidal dispersion, this system is more appropriately regarded as a PDNV–CS–GP multicomponent assembly than as a simple PDNV–GO nanohybrid. Further purification and single-particle analyses would be useful for determining component stoichiometry, spatial organization and the fraction of unbound or incompletely assembled components.

Hu et al. constructed a self-repair platform integrating orange PDNVs, magnesium-doped carbon dots (Mg-CDs) and an egg-white protein hydrogel, as shown in Figure 5 (Hu et al., 2026). The Mg-CDs were synthesized hydrothermally from orange peel and MgO and served both as antioxidant nanozymes and as crosslinking components of the hydrogel. Orange PDNVs were subsequently incorporated by rehydrating the prefabricated Mg-CDs-crosslinked hydrogel in a vesicle-containing solution. The resulting platform therefore combined two orange-derived nanocomponents with distinct origins and complementary functions. Mg-CDs scavenged ROSs and suppressed neuronal ferroptosis through GPX4/ACSL4-associated regulation, while both Mg-CDs and PDNVs contributed to inflammatory modulation. The PDNVs additionally supplied antioxidant and anti-inflammatory biomolecules and were associated with transcriptional responses related to tissue remodeling and neural repair. The hydrogel prolonged local retention and enabled sustained release of both components. Because Mg-CDs were incorporated into the hydrogel network rather than loaded into or coated by PDNVs, this formulation is more likely the co-integration of PDNVs and Mg-CDs within a macroscopic hydrogel matrix. Unfortunately, in this work, the in vivo treatment groups did not include a PDNV-containing egg-white hydrogel without Mg-CDs as a control sample, a more careful experimental design would therefore strengthen component-specific functional analysis.

Other reported PDNV–carbon dot hybrid systems have adopted a similar matrix-based integration strategy. Ding et al. incorporated Flos Sophorae-derived carbon quantum dots and cactus PDNVs into a rapidly gelling sprayable hydrogel (Ding et al., 2025). The resulting multicomponent platform combined the functional carbon dots and PDNVs with a tannic acid-Fe³⁺-polyacrylamide network, exhibiting hemostatic, photothermal, immunomodulatory and ferroptosis-related functions. The cactus PDNVs contributed to antioxidant, anti-inflammatory, pro-angiogenic and cell-migration-promoting activities, while the overall formulation accelerated wound healing, angiogenesis and hair-follicle regeneration. Zaffar et al. similarly incorporated rose-petal PDNVs and fluorescent carbon dots into an injectable hydrogel formed from oxidized sodium alginate and branched polyethyleneimine for antibacterial treatment and diabetic-wound repair (Zaffar et al., 2025). The biohybrid provided local retention and sustained release of both nanocomponents and showed antibacterial activity against Escherichia coli in a diabetic wound model. These hybrid systems are therefore better regarded as multicomponent nanocomposites in which PDNVs, and carbon dots provide complementary functions, rather than as direct carbon dot–PDNV hybrids.

Nevertheless, mammalian EV studies provide additional methodologies of carbon nanomaterials–vesicle integration. When mesenchymal stem cell-derived EVs covalently conjugated to alkyne-functionalized GO through Cu-catalyzed azide–alkyne cycloaddition, a directly surface-linked EV–GO hybrid was constructed (Driscoll et al., 2021). In another example, ¹⁰B-containing carbon dots were encapsulated within macrophage-derived exosomes for boron neutron capture therapy. The carbon dots served as both boron carriers and fluorescent tracers, whereas exosome encapsulation altered their biodistribution, facilitating blood-brain barrier crossing and accumulation in orthotopic glioma tissue (Li et al., 2021). A recent review of vesicle-mediated carbon-dot delivery further surveyed structurally distinct configurations involving vesicular encapsulation, membrane- or surface-associated carbon dots, and hydrogel-based co-delivery (Xu et al., 2026). These distinctions are equally relevant to PDNV-containing systems because they prevent direct material–membrane hybrids, intravascular loading and matrix-based co-integration from being grouped under a single structural designation.

Current PDNV–carbon nanomaterial hybrid systems require structure-specific evaluation. Consistent with the validation framework outlined above, direct PDNV–carbon nanomaterial assemblies should be assessed using complementary evidence, including single-particle colocalization of the PDNV and carbon nanomaterials, material-specific elemental or spectroscopic mapping, and membrane-integrity assays (Welsh et al., 2024; Mizenko et al., 2026). Where relevant, protease-protection experiments and post-purification assessment of residual unbound components should be combined with appropriate vesicle-only, material-only, and equivalent unassembled-mixture controls; matrix controls should be included for hydrogel-based systems to distinguish stable material–membrane association from nonspecific adsorption, aggregation, or co-sedimentation. As for PDNV–GO assemblies, attachment stability, membrane integrity, and colloidal behavior should be assessed after removal of unbound components. In hydrogel-based systems, PDNVs and carbon dots should be tracked separately to determine their retention and release kinetics in vivo. Future studies should therefore relate functional outcomes to the experimentally verified spatial organization of each component.

4.5 Integrating PDNVs with other inorganic nanomaterials

Beyond the major biohybrid systems discussed above, PDNVs may also be integrated with other inorganic nanomaterials, including metal oxides, metal–drug coordination nanoparticles, Prussian blue, black phosphorus, up-conversion rare-earth nanoparticles and semiconductor quantum dots. These nanomaterials can introduce magnetic, catalytic, redox, photothermal, mineralization-related and optical functions, while their biological behavior and fate depend strongly on material composition, structure and surface chemistry (Yang et al., 2019; Lee et al., 2025). Because direct evidence of PDNV-based hybrid system remains unclear, the following discussion distinguishes PDNV-based systems from methodological examples established using other EVs.

Ye et al. developed a ginger-derived EV encapsulated cerium-carbenicillin coordination nanoplatform (Ce–Car@EV) to overcome carbenicillin resistance in Pseudomonas aeruginosa (Ye et al., 2026). Ce (IV) coordinated with the carboxyl groups of carbenicillin and formed Ce–Car nanoparticles that were subsequently incorporated into PDNVs by ultrasonication. TEM images indicated that the Ce-Car nanoparticles were predominantly encapsulated within the vesicles. Under acidic conditions, protonation of the carboxyl groups destabilized the coordination structure and promoted carbenicillin release. Mechanistic analyses showed that the Ce component depleted intracellular ATP, impaired oxidative phosphorylation and respiratory-chain activity, and inhibited energy-dependent efflux, thereby restoring bacterial susceptibility to carbenicillin. The PDNV coating reduced macrophage uptake and increased accumulation and retention in infected lung tissue. Importantly, the inorganic component is a Ce (IV)-drug coordination nanoparticle rather than elemental Ce or CeO₂. Although the study showed almost elemental Ce clearance within 7 days, the chemical speciation of Ce during degradation and clearance remains insufficiently resolved and would benefit from speciation-sensitive analysis.

Another PDNV-based hybrid example was reported by Zhao et al., who incorporated selenium nanoparticles (SeNPs) into Cyperus PDNVs for selenium delivery (Zhao et al., 2025). Among the loading approaches evaluated, ultrasonication produced comparatively high Se incorporation, while the resulting SeNP–PDNV hybrid system retained stability during storage and under simulated gastrointestinal conditions. This study extends PDNV–inorganic nanomaterials integration beyond coordination chemistry method, although the spatial distribution of SeNPs within individual vesicles and the effects of loading on membrane integrity remain to be established.

Other mammalian EV–inorganic nanomaterial systems provide complementary methodological precedents. For example, ZnO nanocrystals have been loaded into mammalian EVs and combined with antibody-mediated surface targeting for acoustically activated cytotoxicity (Dumontel et al., 2022), whereas glioblastoma-derived exosomes have been used to coat Prussian blue nanoparticles for tumor targeting and photothermal treatment (Hill et al., 2024). Black phosphorus quantum dots and up-conversion rare-earth nanoparticles have also been incorporated into vesicle-based platforms for biomineralization or targeted neurological applications (Wang et al., 2019; Jin et al., 2025). These studies broaden the available hybridization and assembly strategies, but their application extended to PDNVs requires direct validation of loading or coating stability, membrane integrity, targeting behavior and the intracellular fate of both components.

In addition, both the stability of the inorganic phase and its effects on the PDNV membrane require attention. Metal oxides may dissolve or change oxidation state, black phosphorus is susceptible to oxidation, and degradation of some up-conversion rare-earth or semiconductor nanoparticles may release potentially toxic metal ions. Conversely, inorganic components may perturb membrane lipids or proteins. Material transformation, ion release, membrane integrity and biological activity should therefore be evaluated in parallel, as magnetic, optical or catalytic signals alone do not establish a stable vesicle–material architecture. Across current studies, MOFs primarily contribute porous loading and stimulus-responsive coordination chemistry, metallic nanoparticles provide catalytic, photothermal and antibacterial functions, silicon-based nanomaterials offer tunable porous storage, and carbon nanomaterials introduce optical, photodynamic, redox and nanozyme activities. Other inorganic nanomaterials further extend these functions to magnetic guidance, mineralization and photon conversion, while PDNVs mainly provide the biological interface and plant-derived bioactivity.

Increasing compositional complexity does not necessarily improve performance of the hybrid system. As the same validation criteria discussed earlier, a well-defined PDNV–nanomaterial biohybrid requires concurrent characterization of the PDNV membrane, the physicochemical behavior of the inorganic nanomaterial components, and their association after purification, storage, and administration, supported by complementary evidence such as single-particle colocalization of the PDNV and inorganic components, material-specific elemental or spectroscopic mapping, and membrane-integrity assays (Welsh et al., 2024; Mizenko et al., 2026). Where relevant, protease-protection experiments and post-purification assessment of residual unbound components should be combined with comparisons involving the nanomaterials alone, PDNVs alone, and an equivalent unassembled mixture to distinguish intact hybrids from free components, membrane fragments, or co-sedimented mixtures. The term “synergistic” should be reserved for interaction-dependent effects supported by appropriate comparisons; otherwise, complementary functions or combined effects are more precise. Establishing traceable structure–interface–function relationships will be essential for developing PDNV–nanomaterial hybrid systems with defined compositions and experimentally supported mechanisms of their biomedical functionalities. The preceding sections discuss mammalian extracellular vesicles and plant-derived vesicles in different contexts, including their biological origin, molecular composition, isolation and characterization, engineering strategies, and biological functions. To consolidate these dispersed comparisons and provide an integrated overview, Table 4 summarizes the principal similarities and differences between the two vesicle systems.

TABLE 4. Comparison of mammalian extracellular vesicles and plant-derived vesicles.

5 Conclusion and future perspective

This review has examined PDNV-based biohybrid systems that integrate PDNVs with various functional inorganic nanomaterials, e.g., MOFs, MSNs, metal NPs and carbon dots. It first considered the biological sources, composition, isolation, enrichment and characterization of PDNVs, before discussing surface regulation, interfacial anchoring, membrane fusion, and targeting and cargo integration as principal routes for regulating material–membrane interactions (Hong et al., 2026). Several representative biohybrid systems integrating PDNVs with MOFs, metallic and metal-oxide NPs, silicon-based nanomaterials, carbon-based nanomaterials and other inorganic components were then critically evaluated. Across these systems, inorganic nanomaterials can provide cargo protection, catalytic activity, energy-conversion capability and imaging functions, whereas PDNVs-derived membranes can influence colloidal behavior, biological interactions and drug delivery dynamics (Gui et al., 2025; Jung et al., 2025). Hybrid performance therefore depends not only on the functions of the individual components, but also on their spatial organization, membrane integrity, interfacial stability and dynamic behavior under biological conditions.

Despite this progress, the reproducible construction of these biohybrids remains limited by the heterogeneity and poor comparability of starting materials. Naturally released PEVs, tissue-isolated PDNVs and nanovesicles reconstructed from extracted plant lipids differ in their origins, molecular compositions and potential co-isolated constituents. Botanical source, tissue type, growth and harvesting conditions, storage and downstream processing can alter the composition and stability of PDNVs and may consequently affect hybrid material association, nucleation, membrane fusion, cargo loading and colloidal stability. These factors should therefore be treated as critical design variables rather than incidental preparative details. Improved reproducibility and cross-study comparison will require integrated reporting of sample provenance, processing history, compositional profiles and measures of relative enrichment (Li et al., 2026).

More importantly, the actual architecture of the hybrid interface and the attribution of its functions often remain insufficiently resolved. Changes in hydrodynamic size or zeta potential values, or the appearance of a peripheral layer in conventional TEM images, cannot alone establish stable material–membrane association, because similar observations may result from particle aggregation, free membrane fragments or residual unbound material. Interface validation should therefore combine post-purification compositional analysis, dual-component colocalization, elemental or spectroscopic mapping, membrane-integrity assays and, where feasible, single-particle analysis (Kapoor et al., 2024; Welsh et al., 2024; Lundy et al., 2026). These approaches should determine the localization of the inorganic component relative to the vesicle membrane and distinguish intact hybrids from free vesicles, free nanomaterials and membrane fragments. Interface dissociation, material transformation and cargo release should also be examined during storage and in relevant biological environments. Comparisons with the individual components, an equivalent unassembled mixture and, where appropriate, disrupted interface controls are needed to establish whether the observed effects arise from the intact hybrid architecture or from components released after disassembly. Direct material–membrane hybrids should also be distinguished from hierarchical composites and co-delivery systems in which PDNVs, and nanomaterials are merely co-confined within an external matrix.

Future research should develop beyond empirical mixing toward interface engineering, governed by explicit design rules. Plant cultivar, geographical origin, harvesting period, storage conditions, and pretreatment procedures should be treated as critical raw-material attributes. In parallel, PDNV membrane composition and physicochemical properties should be evaluated together with inorganic nanomaterial surface chemistry and assembly conditions to establish traceable structure–interface–function relationships.

A staged translational pathway should then be used to guide further development. First, critical quality attributes should be defined for the PDNVs, the inorganic component, and the assembled interface, including sample provenance, composition, particle size and concentration, membrane integrity, inorganic identity and loading, residual free-material fraction, component localization, interfacial stability, and release behavior. Second, scalable and reproducible manufacturing should be combined with validated storage conditions and batch-level quality control to ensure consistent composition and performance. Third, in vivo studies should track PDNV and inorganic components separately and evaluate interface dissociation, material transformation, ion release, biodistribution, clearance, and dose-normalized exposure, with comparisons against the individual components and an equivalent unassembled mixture (Li et al., 2026; Lundy et al., 2026). Finally, biosafety assessment should encompass co-isolated plant constituents, microbial contamination, potentially immunogenic plant proteins and glycoconjugates, route-dependent effects, long-term accumulation, and clearance. This staged framework will help determine whether observed functions arise from a defined and stable hybrid architecture and will provide a practical basis for advancing PDNV–nanomaterial hybrid systems with predictable performance and translational potential.

Generative AI statement

The author(s) declared that generative AI was not used in the creation of this manuscript.

Any alternative text (alt text) provided alongside figures in this article has been generated by Frontiers with the support of artificial intelligence and reasonable efforts have been made to ensure accuracy, including review by the authors wherever possible. If you identify any issues, please contact us.

Publisher’s note

All claims expressed in this article are solely those of the authors and do not necessarily represent those of their affiliated organizations, or those of the publisher, the editors and the reviewers. Any product that may be evaluated in this article, or claim that may be made by its manufacturer, is not guaranteed or endorsed by the publisher.

Footnotes

Edited by: Enbo Zhu, University of California, Los Angeles, United States

Reviewed by: Peng Zhao, University of California, Los Angeles, United States

Yichen Zhu, University of California, Los Angeles, United States