Supramolecular Nano‐Grid Platform to Load and Deliver Liposomes and Exosomes

In general, carriers are designed to deliver pharmaceutical (featuring small drug‐molecules) and biopharmaceutical (featuring large‐molecule drugs, e.g., antibodies, proteins, etc.) active ingredients and platform that is capable of transporting nanoparticles is rare. Herein, the fabrication of a supramolecular carrier with a nano‐grid structure for the entrapment of liposomes and exosomes is reported. The nano‐grid particles (NGP) are made of cyclodextrins as elementary units which are widely used to load drugs through host–guest inclusion complexes, whilst its secondary structure further formed the NGP tertiary architectures, enabling the entrapment of nanoparticles. The NGP features stimuli responsiveness, tunable surface charge, and good biocompatibility. The entrapped liposomes in NGP present enhanced distribution in lungs and extended duration of action in mice. And elongated release profiles for over 14 days is obtained for exosomes—making one‐of‐its‐kind unique platform to load and deliver nanoparticles.


Introduction
Advanced drug-delivery systems have shown to be effective in improving pharmacokinetic and pharmacodynamic profiles, targeted delivery, and stability of small/ macro molecules. [1]With the improvement of cognition, current therapeutic agents are not limited to small or macromolecules, and increasing evidence suggests that nano-scale particles also exert a therapeutic role.For example, exosomes, membranederived extracellular vesicles that play an important role in cell-cell communication, [2] have been exploited in the treatment of cancer, [3] cardiovascular diseases, [4] neurodegenerative diseases, [5] inflammatory arthritis, [6] etc. Microorganisms, such as bacteriophages [7] and oncolytic viruses [8] offer strategies for the treatment of bacterial infections and tumors.1c,9] Despite the broad application of nanoparticles, there are still some challenges.For example, liposomes and exosomes are in liquid form and it is difficult to maintain a certain concentration locally, which limits the therapeutic efficiency.Besides, liposomes and exosomes without special modification are cleared rapidly after systemic administration. [10]ncorporating the nanoparticles into a carrier to form superstructures may achieve superior functionalities, such as enhanced stability and prolonged duration of action. [11]he efforts to deliver small/macromolecules efficiently and effectively have been ongoing for decades, while the study on the delivery of nanoparticles is quite rare due to the lack of appropriate carriers.Attempts to deliver nanoparticles are using microbubbles, which are micro-sized gas bubbles made of lipids, polymers, or proteins.Functionalized liposomes and other nanoparticles can be loaded on the surface of microbubbles to achieve enhanced delivery to the targeted tissue. [12]The attachment of nanoparticles to microbubbles is through specific interactions, such as antibodies.Only nanoparticles that are specifically modified can be loaded on microbubbles, thus the loading efficiency could be limited by the number of antibodies on the surface of nanoparticles and microbubbles. [13]Carriers that can load nanoparticles based on the general properties of nanoparticles such as size and shape instead of specific interactions may achieve broader significance.Given the fact that clinically approved nanoparticles and exosomes are usually in the size range of 50-150 nm, [2b,14] it is extremely challenging to produce carriers with desired dimensions and characteristics that can deliver these large nanoparticles.
Cyclodextrins (CDs), a family of cyclic oligosaccharides constituted by D-glucose units, are widely used as pharmaceutical excipients and building blocks of functional materials. [15]A variety of CD-based nanoparticles and polymers have been constructed through cross-linking and modification to the hydroxyl groups on the CD molecule. [16]Metal-organic frameworks based on CD (CD-MOF), where γ-CDs are orderly arranged and conjugated through potassium ions, exhibit permanent porosity and excellent biocompatibility. [17]The hydroxyl groups on the γ-CD units facilitate the crosslinking of CD-MOF to form polymeric particles with well-defined framework structures. [18]And the use of functional crosslinking agents enables these particles to respond to the biological stimulus, such as glutathione or reactive oxygen species (ROS) [18c,d] in previous examples.
Herein, we present a novel design for nano-grid particles (NGP) as supramolecular carriers that are capable of loading large nanoparticles including liposomes and exosomes.This supramolecular boronated organic framework (BOF), prepared by crosslinking CD-MOF using a boronated crosslinking agent, exhibits either NGP featured (N-BOF) or spherical shape (S-BOF) with pH and H 2 O 2 dual responsiveness.N-BOF particles are highly porous (pore size in the range of 10-300 nm) with interconnected nanowires made of CDs, with surface charge adjustable from À35 to 15 mV via modification with a positively charged protein.The entrapment of liposomes in N-BOF significantly enhanced its distribution in lungs.Exosomes were captured and protected by N-BOF demonstrating much elongated release profiles and responsiveness.

Preparation and Characterization of BOF
The supramolecular architecture of BOF with pH and ROS dual responsiveness is constructed via crosslinking of CD-MOFs using a symmetric boronated compound (SBC) (Figure S1, Supporting Information).The reaction temperature was observed to be a key parameter controlling the shapes and structures of BOF.As shown in the transmission electron microscope (TEM) and scanning electron microscope (SEM) images in Figure 1a,b, a typical nano-grid structure was obtained when cross-linked at 50 °C.The diameter of the nanowires was recorded in a range of 20-100 nm and the size of the voids was from 10 to 300 nm.The overall size distribution of N-BOF was calculated 1268 AE 26.5 nm using dynamic light scattering (DLS).When reaction temperature was raised to 60 °C, spherical particles of 50-300 nm size was obtained (Figure 1c,  d).It is hypothesized that the reactivity discrepancy between hydroxyl groups of CDs [19] and the spatial orientation of CD units in MOF controlled the crosslinking in a selective patternresulting into formation of different particle architectures under certain reaction temperatures (Figure S2, Supporting Information).
The chemical structures of BOFs were confirmed by Fouriertransform infrared spectroscopy (FTIR) and nuclear magnetic resonance spectroscopy (NMR).The characteristic peaks at 1751 cm À1 for carbonyl group in ester bonds in both the samples, N-BOF and S-BOF confirmed the successful crosslinking of CD through SBC linker (Figure 1e).BOFs were treated with sodium deuteroxide (NaOD) to disintegrate the cross-linked structure for NMR characterization.The 1 H-NMR spectra showed the characteristic peaks of SBC (δ = 7.2, 7.5, 4.5, and 3.6 ppm) and γ-CD (δ = 5.0 and 3.3-3.8ppm), further conforming the chemical structure of BOF (Figure 1f ).SBC is a phenylboronic ester that could react with H 2 O 2 , [20] leading to responsiveness in BOF structure (Figure S3, Supporting Information).FTIR and 1 H-NMR spectra of both, N-BOF and S-BOF were found to be identical.

H 2 O 2 /pH Dual Responsive Behaviors and Degradation Mechanisms of BOF
BOF exhibited pH and H 2 O 2 dual responsiveness, which is relevant to the microenvironmental changes in pathological conditions.Many diseases undergo variations in physiological parameters such as reduced pH or elevated H 2 O 2 levels.For example, the normal pH range of plasma and healthy tissues is 7.38-7.42; [21]lower pH exists in tumor conditions (6.5-7.2), [22]and it can be reduced to 5.4 in inflamed tissues. [23]OS such as H 2 O 2 exists at low levels in healthy tissues, whereas a sharp rise can occur in tumor or inflamed conditions. [24]herefore, microenvironmental changes make it possible for disease-specific drug delivery in controlled manner.To examine the H 2 O 2 /pH-dependent degradation behavior of BOF, material was suspended in phosphate-buffered saline (PBS) with different pH levels or H 2 O 2 concentrations and the transmittance of the suspension was periodically measured at 500 nm.The obtained results showed that BOF was stable at pH 7.4 and the degradation half-life was calculated ≈88.7 h.However, the hydrolysis rate sharply increased when subjected to acidic environments.The degradation half-life was recorded ≈0.19 h at pH 5.0, ≈466 times faster than that at pH 7.4 (Figure 2a,b).A similar responsive behavior was observed in H 2 O 2 solution.Around 60% of BOF was hydrolyzed in 1 mM solution of H 2 O 2 in 1 h compared to only 20%, when treated without H 2 O 2 for 24 h (Figure 2c).4-hydroxybenzyl alcohol (HBA), a side product was generated in a H 2 O 2 concentration dependent manner during BOF degradation (Figure 2d).
A mechanism of the BOF degradation was shown in Figure 2e.γ-CDs and SBC are linked together via ester bonds in BOF.SBC, a phenylboronic ester, is cleavable by oxidants such as H 2 O 2 , which makes BOF responsive to H 2 O 2 . [20]The pH responsiveness of BOF is based on the reversible boronate esterification reaction.Briefly, the boronic ester bond is formed by the reaction of boronic acid and 1,2-cis-diols or 1,3-cis-diols; when the environmental pH is higher than the pKa of the boronic acid, boronate esterification is prevalent, so that the boronic ester is stable.When the pH is lower than the pKa, the reaction was reversed and the boronic ester hydrolysis to boronic acid and 1,2-cis-diols or 1,3-cis-diols. [25]Under physiological environment (i.e., pH 7.4 with very low level of H 2 O 2 ), slow degradation of BOF could be attributed to the hydrolysis of ester bonds.At higher concentration, H 2 O 2 reacts with the boronic ester structure, leads to SBC breakdown into HBA, pentaerythritol and boric acid.Under acidic environment, SBC hydrolyzes into phenylboronic acid and diol structures.Overall, the degradation (hydrolysis) process is likely to be faster in pathological conditions where environment is acidic and/or H 2 O 2 is present.

Surface Charge Modification and Biocompatibility of BOF
To address the challenge of nanoparticles with various charge conditions, lysozyme (LYS), a positively charged protein was used to modify the surface charge of BOF (Figure 3a).By controlling the LYS content, the surface charge of both N-BOF and S-BOF changed from negative to positive, reaching a similar level of 15 mV (Figure 3b).As shown in TEM images in Figure S4, Supporting Information, N-BOF maintained its NGP structure after modification with LYS.
The biocompatibility of BOF and LYS-modified BOF (LYS-BOF) was evaluated by cytotoxicity and hemolysis assays.The cytotoxicity was tested via cell counting kit-8 (CCK-8) assay.BOF and LYS-BOF showed no cytotoxicity up to 100 μg mL À1 concentration, tested on WI26-VA4 cells, a human lung fibroblast cell line, and RAW264.7 cells, a macrophage cell line (Figure 3c,d).The hemolytic activity of samples was tested on mouse red blood cells.At 25-1600 μg mL À1 of BOF and LYS-BOF concentration, supernatants were clear (Figure 3e).The optical density (OD) value of the supernatant was determined at 545 nm to quantify the degree of hemolysis (Figure 3f ).The obtained results showed that there was no hemolysis, indicating good biocompatibility of BOF and LYS-BOF in the blood.

Liposome Payload in BOF and Its Biodistribution
Indocyanine green liposomes (ICG.LP) were loaded into BOF with varied charge levels to evaluate the entrapment efficiency of frameworks (Figure 4a).Negatively charged ICG.LP (zeta potential of À37.0 AE 1.30 mV) with a diameter of 127.9 AE 2.100 nm was used in these experiments.And aliquots of negatively-charged, neutral (modified by LYS), and positivelycharged (modified by LYS) N-BOF and S-BOF were tested in parallel, the zeta potential of which was confirmed by DLS (Table S1, Supporting Information).After co-incubation, positively charged N-BOF showed maximum adsorption for ICG.LP, which was evidential from change in color of supernatant and sediment: the green color of ICG.LP almost disappeared in the supernatant; instead, apparent green color displayed in the sedimented BOF particles (Figure 4b).Determination of free ICG confirmed the integrity of liposomes (Figure S5, Supporting Information).The loading efficiency of different BOF samples for liposomes is presented in comparison as Figure 4c.Positively charged N-BOF showed higher loading efficiency when compared to S-BOF with same charge, and both outperforming the neutral and negatively charged samplesindicating the role of net-grid structure and electrostatic interactions.
The in-vivo distribution of liposomes was assessed to check the impact of BOF by intravenously administrating free ICG.LP, ICG.LP incorporated in LYS-modified N-BOF (ICG.LP@LYS-BOF), and injecting ICG.LP and LYS-modified N-BOF separately (ICG.LP þ LYS-BOF) to BALB/c mice.Animals were sacrificed at 0.5, 1, 2, 6, 12, and 24 h post injection and ICG content in various organs was assessed using ex vivo imaging.Free ICG.LP distributed primarily in the liver and was eliminated rapidly (Figure 4d-f ).ICG.LP@LYS-BOF altered the biodistribution of liposomes, yielding a 90-fold increase in lung accumulation (P < 0.0001, n = 6), whereas injection of ICG.LP and LYS-BOF separately did not show any such impact.The results indicated that incorporation in N-BOF enables the targeted delivery of liposome into lungs which could be exploited as a potential strategy for other critical diseases.

Entrapment and pH Responsive Release of Exosomes
2b] To confirm the ability of N-BOF to entrap bionanoparticles without charge modifications on it, slightly negative charged (À12.9AE 1.1 mV) milk exosomes (MEs) were used.TEM imaging revealed a bilayer structure of MEs with the size of 50-150 nm, which was in agreement with the voids size of N-BOF.After incubating with MEs, the pores in N-BOF were found to be occupied by bilayer structured MEs as indicated by red arrows and red dotted circles (Figure 5a).The entrapment of MEs in the voids of N-BOF was further confirmed using  confocal microscopy, where DiO-labelled MEs co-localized well with N-BOF (Figure S6, Supporting Information).Next, various concentrations of N-BOF were used to determine the entrapment efficiency toward MEs.For the mass ratio 6, 12, and 24 for N-BOF to the MEs (quantified by total protein amount), the encapsulation was 27.2 AE 2.64%, 41.5 AE 2.95%, and 59.5 AE 2.83%, respectively (Figure 5b).The extended co-incubation time contributed to a higher MEs encapsulation efficiency until 4 h, when a saturation was reached (Figure 5c).The entrapped MEs showed an elongated release from N-BOF in response to the environmental pH, with 14.0 AE 0.39% of MEs was released over 14 days at pH 7.7, while 21.5 AE 0.70% was released at pH 6.0 in the same period of time (Figure 5d).The release kinetics of MEs from N-BOF showed biphase behavior, a rapid release in the first day followed by a constant release afterwards.MEs maintained its shape and bilayer membrane structure after being released from N-BOF, which was observed on day 6 using TEM (Figure 5e).

Conclusion
In summary, a novel supramolecular NGP platform, where cyclodextrins are the elementary units is developed for the load and delivery of liposomes and exosomes.N-BOF features H 2 O 2 /pH dual responsiveness, tunable surface charge, and good biocompatibility.The porous nano-grid architecture enables the successful entrapment of large nanoparticles.Liposomes incorporated in N-BOF exhibited dramatically enhanced distribution in lungs and exosomes showed protection and sustained release characteristics, which was benefited from its nano-grid structure and dual responsiveness to pH and H 2 O 2 .This study presents a in N-BOF after co-incubation for 2 h (the mass ratio of N-BOF to the total protein amount of MEs is 6, 12, and 24, respectively).Data are presented as mean AE s.d.(n = 3).One-way analysis of variance (ANOVA) with Tukey's multiple comparisons test was used for multiple comparisons among different groups.6 vs 12, **P = 0.0054; 6 vs 24, ****P < 0.0001; 12 vs 24, **P = 0.0017.c) The effect of co-incubation duration on the amount of MEs trapped by 10 mg mL À1 N-BOF (1, 2, and 4 h).Data are presented as mean AE s.d.(n = 3).Different groups were compared using one-way ANOVA with Tukey's multiple comparisons test. 1 vs 2, *P = 0.0270; 1 vs 4, **P = 0.0019; 2 vs. 4, ns, not significant.d) Accumulated release profiles of MEs@N-BOF at pH 6.0 and 7.7.e) TEM images of MEs released from MEs@N-BOF on day 6 (indicated by red arrows).Samples were stained with uranyl acetate.
new avenue for nanostructure fabrication and provides a versatile platform with the potential to deliver both therapeutic molecules and nanoparticles.

Experimental Section
Animals and Materials: The animal protocols were approved by the Institutional Animal Care and Use Committee of Shanghai Institute of Materia Medica, Chinese Academy of Sciences (code numbers: 2022-01-ZJW-38).BALB/c mice (8 weeks in age, 20-23 g, half male and half female) were provided by Shanghai Lab Animal Research Center.
Preparation of CD-MOF: CD-MOF crystals were produced using γ-CD and KOH through an well-established approach. [27]In brief, γ-CD (6.48 g, 5 mmol) and 2.24 g of KOH (2.24 g, 40 mmol) were dissolved in 200 mL of deionized water, and then methanol was added (60 mL of methanol to every 100 mL of this solution).After incubation at 60 °C for 20 min, the reaction solution became clear and 0.64 g of PEG 20 000 was added to modulate the size of CD-MOF.After incubation at 60 °C for another 20 min, the system was cooled to room temperature to trigger the crystallization.Precipitated CD-MOF crystals were collected by centrifugation (4000 rpm, 5 min) and impurities were removed by rinsing with ethanol and methanol twice respectively.Finally, CD-MOF was dried at 60 °C for 3 h.

Synthesis of SBC:
The cross-linking agent, termed as SBC in this article, was synthesized based on the reaction of boronic acid with diol.Specifically, 2.0584 g of 4-(hydroxymethyl)-phenylboronic acid and 0.8967 g of pentaerythritol with a molar ratio of 2:1 were dispersed in 25 mL of tetrahydrofuran.The reaction was magnetically stirred at ambient temperature for 24 h.After the mixture became transparent and clear, 900 mg of anhydrous sodium sulfate was added and stirred for another 12 h to remove the water.The product SBC was collected after evaporating the solvent using a rotary evaporator at 50 °C.
Synthesis of BOF: First, SBC was activated with CDI.Briefly, 4.417 g of SBC and 7.783 g of CDI (molar ratio 1:4) were dispersed in 60 mL of anhydrous dichloromethane, and the reaction system was stirred at ambient temperature for 4 h.CDI-activated SBC (CDI-SBC) was obtained by centrifugation and rinsing with anhydrous tetrahydrofuran to remove unreacted CDI.The sample was dried at 50 °C to give a white powder.
Subsequently, CDI-SBC was reacted with CD-MOF.216.6 mg of CD-MOF and 500.5 mg of CDI-SBC (molar ratio 1:6) were placed in 4 mL of dry DMF and 140 μL of TEA was added as a catalyst.The reaction temperatures was varied to obtain BOF with different structures.N-BOF and S-BOF were produced when the temperature of the reaction was kept at 50 and 60 °C, respectively.After stirring for 24 h, 8 mL of 95% ethanol was added to the reaction to quench it.The system was centrifuged and the precipitate was washed with ethanol and water and then the products were freeze-dried to obtain a white power.
Characterizations of BOF: 1 H-NMR spectroscopy (Bruker Instrument, 600 MHz, Inc., Germany) and SR-FTIR spectroscopy (Shanghai Synchrotron Radiation Facility BL01B beamline, China) were used to confirm the chemical structure of different samples.The morphologies of BOF were investigated by TEM (FEI Talos Arctica G2, USA) and SEM (S3400, Hitachi, Japan).The size and zeta potential of BOF was measured by DLS (Malvern Zetasizer Nano ZS90, UK).The pore size distribution of N-BOF was analyzed by Image J saftware based on TEM images.
Surface Charge Modification of BOF by LYS: Aliquots of 10 mg N-BOF and S-BOF were added in 1 mL of 2.5, 5, 10, and 20 mg mL À1 LYS solution, respectively.The BOF suspensions were vortexed gently (at room temperature for 2 h) followed by centrifugation.The precipitate was washed three times with 1 mL of deionized water to remove free LYS.LYS loading on BOF was quantified by measuring unloaded LYS in the supernatant and the solution collected during the washing steps using a BCA protein assay kit.The zeta potential of LYS-modified BOF was studied using DLS.
Cytotoxicity Assays: Macrophage RAW 264.7 cells and human lung fibroblast WI26-VA4 cells were used to evaluate the cytotoxicity of BOF and LYS-BOF.RAW264.7 cells were supplied by Shanghai Cell Bank, Chinese Academy of Sciences (Shanghai, China) and cultured in DMEM medium with 10% v/v FBS.WI26-VA4 cells were obtained from ATCC (Manassas, VA 20 108 USA) and cultured in EMEM medium with 10% v/v FBS.For CCK-8 assay, RAW 264.7 cells and WI26-VA4 cells were seeded in 96-well plates and cultured for 24 h (the densities were 1 Â 10 4 and 2 Â 10 4 cells well À1 , respectively).Then, the cells were incubated with BOF and LYS-BOF suspensions at different concentrations (0, 6.25, 12.5, 25, 50, and 100 μg mL À1 ) for 24 h.Cell viability was quantified using CCK-8 method.After incubation for another 1.5 h, the OD values were detected at 450 nm using a microplate reader (Thermo Scientific, Multiscan Go, USA) to calculate the cell viability OD sample indicates the OD value of the experimental group.OD control is the OD value of the control group and OD blank indicates the OD value of the CCK-8 solution at the experimental concentration.
Hemolysis Assays: Fresh mouse blood samples were collected from BALB/c mice by cardiac puncture under deep anesthesia using isoflurane.The plasma was separated by centrifugation (3000 rpm, 5 min), and the red blood cells in the bottom layer were washed three times with PBS and then diluted with PBS to a final concentration of 4% v/v.BOF and LYS-BOF suspended in saline were mixed with the same volume of red blood cell suspension.The final concentrations of BOF and LYS-BOF suspension were 25, 50, 100, 200, 400, 800, and 1600 μg mL À1 .Saline was used as a negative control and deionized water was used as a positive control.The samples were incubated at 37 °C for 2 h.After centrifugation (3000 rpm, 5 min), 100 μL of the supernatant of each sample was added to a 96-well plate.The hemolysis degree was calculated by measuring the OD values at 545 nm using a microplate reader (Thermo Scientific, Multiscan Go, USA).

Hemolysis ð%Þ ¼
OD sample À OD negative OD positive À OD negative Â 100% (2) Surface Charge Modification of BOF for Liposome Capturing Evaluation: LYS was used to modify the surface charge of BOF.BOF, with inherent negative charge adsorbed positively-charged LYS in an aqueous solution.The LYS loading content could be tuned by controlling the concentration of LYS solution, and the surface charge of BOF could be adjusted from negative to positive with the increase of LYS loading content.In practice, negatively-charged N-BOF and S-BOF could be obtained without modification (labeled as "N-" and "S-", respectively).To prepare neutral N-BOF and S-BOF, 10 mg of N-BOF and S-BOF were incubated with 1 mL of LYS solution (2.5 mg mL À1 for N-BOF and 5 mg mL À1 for S-BOF) for 2 h.Following centrifugation, the precipitates were washed with deionized water for three times to produce neutral N-BOF and S-BOF (labeled as "N" and "S", respectively).Positively-charged N-BOF and S-BOF (labeled as "Nþ" and "Sþ", respectively) were obtained by mixing 10 mg of N-BOF and S-BOF with 1 mL of LYS solution (20 mg mL À1 ) and incubating for 2 h, followed by washing steps similar to that of neutral BOF.The zeta potential of BOFs were measured using DLS at the concentration of 1 mg mL À1 .
Preparation of ICG.LP: The prescribed amounts of cholesterol and soybean lecithin S100 were precisely weighed and placed separately into a flask.An appropriate amount of absolute ethanol was added and the mixtures were shaken at room temperature (20-25 °C) till they were dissolved.The above solutions were placed in a 40 °C water bath and rotated to evaporate under vacuum until the absolute ethanol is completely evaporated, forming a lipid film deposit.Preheated purified water containing indocyanine green at 30 °C was introduced and stirred in a 40 °C water bath until it completely hydrated and uniformly mixed.The liposome suspension was sonicated for 15 min in an ultrasonic cell homogenizer after hydration and filtered sequentially through 0.8, 0.45, and 0.22 μm organic microporous membranes.
ICG.LP Capturing: ICG.LP was diluted to a concentration of 0.1 mg mL À1 .Aliquots of 1 mL diluted ICG.LP were mixed with 5 mg of N-BOF and S-BOF with different charge, respectively.The mixtures were vortexed for 1 h to allow the complete interaction of BOF and liposomes.Then, free liposomes were separated from those entrapped inside BOF by centrifugation at 4000 rpm for 5 min.The absorbance at 800 nm of the supernatants were measured for ICG.LP quantification using UV-Vis spectroscopy (UH5300, Hitachi, Japan).
Liposome Integrity Test: To investigate the effect of interaction of BOF on the integrity (structural damage and leak) of ICG.LP membrane, free ICG was detected before and interacting with BOF.Positively-charged N-BOF was used for the test due to its highest ICG.LP loading capacity.In brief, positively-charged N-BOF (containing 5 mg of N-BOF) was added into 1 mL of diluted ICG.LP, vortexed for 1 h, and then transferred to a dialysis bag (8000 Da).ICG.LP at the same concentration without adding BOF was used as the control.The dialysis bags were immerged into 15 mL of distilled water and incubated at 37 °C for 24 h.Finally, the absorbance of the dialysate at 800 nm was measured to detect free ICG using UV-Vis spectroscopy (UH5300, Hitachi, Japan).
Biodistribution Studies of Free and Loaded Liposomes: Preparation of the Samples: Positively-charged N-BOF modified by LYS (LYS-BOF) was prepared as follows: 60 mg of N-BOF was incubated with 3 mL of LYS solution (20 mg mL À1 ) for 2 h.After centrifugation, the precipitates (LYS-BOF) were washed with distilled water for three times.ICG.LP loaded in LYS-BOF (ICG.LP@LYS-BOF) was prepared as follows: 20 mg of LYS-BOF was added into 8 mL of ICG.LP (≈0.1 mg mL À1 ), vortexed for 1 h, and centrifugated.The precipitates were washed with deionized water and free ICG.LP in the supernatant was quantified to calculate the loading content of ICG.LP.
Biodistribution Studies of Free and Loaded Liposomes: BALB/c mice (half male and half female) were intravenously administered with free ICG.LP, ICG.LP@LYS-BOF, or injected ICG.LP and LYS-BOF separately (ICG.LP þ LYS-BOF).The mice were injected 200 μL of 0.05 mg mL À1 ICG.LP for free ICG.LP group and 200 μL of ICG.LP@LYS-BOF (comprising 0.05 mg mL À1 ICG.LP) for ICG.LP@LYS-BOF group.For ICG.LP þ LYS-BOF group, the mice were administered 100 μL of 0.1 mg mL À1 ICG.LP, immediately followed by 100 μL of 5 mg mL À1 LYS-BOF.The final dose was 0.5 mg of ICG per kg.Blood samples were collected at 0.25, 0.5, 1, 2, 6, 12, and 24 h post-injection.The mice were sacrificed and the organs (heart, liver, spleen, lungs, and kidneys) were collected and observed using an in vivo imaging system (PerkinElmer, Waltham, Massachusetts, USA).The average radiant efficiency of ICG was quantified (excitation wavelength: 745 nm; emission wavelength: 840 nm).The pharmacokinetic parameters of ICG fluorescence in each organ were calculated using DAS 2.0 software.
Isolation of MEs: Skimmed milk was centrifuged (13 000 g, 30 min) and then the pH of the supernatant was adjusted to 6.0 using 10% acetic acid.Rennet was dissolved in a 1% NaCl solution and activated in a water bath (37 °C, 0.5 h) before adding to milk at approximately 0.035 mg mL À1 .The mixture was placed in a water bath at 37 °C for 30-35 min.After complete coagulation of the milk protein and precipitation of whey, the centrifuge tube was rapidly cooled and then centrifuged (13 000 g, 60 min) to remove casein.The supernatant was filtered through a 0.45 μm membrane and then centrifuged (100 000 g, 60 min) to remove larger vesicles.Next, the filtrate was passed through a 0.22 μm membrane and ultracentrifugation was performed at 135 000 g for 90 min (Optima XPN-90, Beckman Coulter, USA).Finally, the precipitate was resuspended in PBS, ultracentrifuged again (135 000 g, 60 min), the MEs precipitate was collected, resuspended in PBS and filtered through a 0.22 μm membrane, and the total protein concentration was measured using BCA kit (The above centrifugation was carried out at 4 °C).
Entrapment of MEs by N-BOF: The MEs were labeled by DiI for quantification.The standard curve of MEs concentration and DiI fluorescent intensity was established by a multi-mode microplate reader (SpectraMax M5e, Molecular Devices, USA).To assess the entrapment capability of N-BOF, N-BOF was mixed with MEs at mass ratios of 6, 12, and 24 (MEs were quantified by total protein concentration), and shaken at 150 rpm, 37 °C for 2 h.To evaluate the influence of the incubation time, N-BOF (10 mg mL À1 ) was mixed with MEs and the mixture was kept on a shaker at 37 °C for 1, 2, and 4 h.After incubation, samples were centrifuged (12 000 g, 5 min), and the fluorescence intensity of the supernatant was measured to quantify the MEs captured by N-BOF.For localization, MEs were labeled by DiO.Confocal microscopy (TCS SP8, Leica, Germany) and TEM (Talos L120C, FEI, USA) were used as tools to locate the MEs.
Release of MEs from MEs@N-BOF: Aliquots of MEs@N-BOF were treated with 1 mL of PBS solution at pH 7.7 and pH 6.0, respectively.The samples were then incubated in a shaker at 37 °C and 100 rpm for a predetermined time period before removing the supernatant (250 μL), which was subsequently replenished with the identical volume of same pH PBS.The fluorescence intensity of the supernatant was determined using a multi-mode microplate reader to calculate the cumulative release of MEs.The morphology of MEs was examined by TEM at 6 days after the release.
Statistical Analysis: Data were analyzed using GraphPad Prism 9 (v.9.0.0).Data are presented as mean AE s.d.; sample size (n) for each statistical analysis is shown in the figure legend.One-way analysis of variance (ANOVA) with Tukey's multiple comparisons test or unpaired two-tailed t-test was used to compare the results of different groups.P < 0.05 was considered significant.

Figure 1 .
Figure 1.Characterization of BOF.a) TEM image of N-BOF.Inserted box: a zoomed-in view of N-BOF (upper) and the pore size distribution (lower).b) SEM image of N-BOF.c, d) TEM (c) and SEM (d) images of S-BOF.e) FTIR spectra of CD-MOF, SBC, S-BOF, and N-BOF.f ) 1 H-NMR spectra of S-BOF and N-BOF in NaOD/D 2 O.

Figure 3 .
Figure 3. Surface charge modification strategy and the biocompatibility assays of BOF.a) Schematic illustration of surface charge modification of BOF.b) Zeta potential of BOF switched from negative to positive as the increase of LYS loading content.c,d) Cytotoxicity assays of BOF and LYS-BOF in c) WI26-VA6 and d) RAW264.7 cells.Data are presented as mean AE s.d.(n = 6).e) Centrifuged red blood cells after incubation with saline, water, BOF, and LYS-BOF.f ) Hemolysis quantification of red blood cells incubated with BOF and LYS-BOF, with water as a positive control and saline as a negative control (n = 3).

Figure 4 .
Figure 4. Liposome entrapment ability of BOF and in vivo kinetic behaviors of ICG.LP incorporated in N-BOF.a) Schematic illustration of liposome loading process into BOF and the in vivo study.b) ICG.LP was mixed with equivalent quantities of N-BOF and S-BOF with N-, N, and Nþ representing negative, neutral, and positive charge, respectively for N-BOF.Similarly, S-, S, and Sþ represent negative, neutral, and positive charge, respectively for S-BOF.ICG.LP appeared green in aqueous solution and BOF appeared as white powder.c) ICG.LP adsorption percentage by N-BOF and S-BOF with different surface charge.Data are presented as mean AE s.d.(n = 3).Comparisons between two groups were conducted using unpaired two-tailed t-tests.****P < 0.0001.d) Biodistribution of ICG.LP, ICG.LP@LYS-BOF, ICG.LP þ LYS-BOF after intravenous injection.ICG was used as the fluorescence probe."Blank" referred to untreated mice.e) Average radiant efficiency (ARE) profile of ICG fluorescence in each organ.Data are presented as mean AE s.d.(n = 6).f ) AUC (0-24) of different forms of ICG.LP in organs, indicating the bioavailability.Data are presented as mean AE s.d.(n = 6).Unpaired two-tailed t-tests were used to compare the differences between two groups.****P < 0.0001.

Figure 5 .
Figure 5. Entrapment and pH responsive release of MEs.a) TEM images of N-BOF, MEs, MEs@N-BOF and the zoomed in view of MEs@N-BOF.MEs are indicated by red arrows and red dotted circles.MEs and MEs@N-BOF were negatively stained with uranyl acetate.b) Quantification of MEs entrapped in N-BOF after co-incubation for 2 h (the mass ratio of N-BOF to the total protein amount of MEs is 6, 12, and 24, respectively).Data are presented as mean AE s.d.(n = 3).One-way analysis of variance (ANOVA) with Tukey's multiple comparisons test was used for multiple comparisons among different groups.6 vs 12, **P = 0.0054; 6 vs 24, ****P < 0.0001; 12 vs 24, **P = 0.0017.c) The effect of co-incubation duration on the amount of MEs trapped by 10 mg mL À1 N-BOF (1, 2, and 4 h).Data are presented as mean AE s.d.(n = 3).Different groups were compared using one-way ANOVA with Tukey's multiple comparisons test. 1 vs 2, *P = 0.0270; 1 vs 4, **P = 0.0019; 2 vs. 4, ns, not significant.d) Accumulated release profiles of MEs@N-BOF at pH 6.0 and 7.7.e) TEM images of MEs released from MEs@N-BOF on day 6 (indicated by red arrows).Samples were stained with uranyl acetate.