Angiopep-2-functionalized nanoparticles enhance transport of protein drugs across intestinal epithelia by self-regulation of targeted receptors

Abstract

Ligand-modified nanoparticles (NPs) have been widely used in oral drug delivery systems to promote endocytosis on intestinal epithelia. However, their transcytosis across the intestinal epithelia is still limited. Except for complex intracellular trafficking, recycling again from the apical sides into the intestinal lumen of the endocytosed NPs cannot be ignored. In this study, we modified NP surfaces with angiopep-2 (ANG) that targeted the low-density lipoprotein receptor-related protein 1 (LRP-1) expressed on the intestine to increase both the apical endocytosis and basolateral transcytosis of NPs. Notably, our finding revealed that ANG NPs could increase the apical expression and further basolateral redistribution of LRP-1 on Caco-2 cells, thus generating an apical-to-basolateral absorption pattern. Because of the enhanced transcytosis, insulin loaded ANG NPs possessed much stronger absorption efficiency and induced maximal blood glucose reduction to 61.46% in diabetic rats. Self-regulating the distribution of receptors on polarized intestine cells to promote basolateral transcytosis will provide promising insights for the rational design of oral delivery systems of protein/peptide drugs.

---

1. Introduction

Numerous strategies have been attempted to conquer the epithelial absorption barriers for oral administration of protein/peptide drugs.^1–6 In particular, ligand-modified nano-delivery systems attract great attention. As intestinal epithelia express various receptors, appropriate surface modification with targeting ligands on nanoparticles (NPs) could facilitate cellular internalization via ligand–receptor interactions.^7–9 Nevertheless, the majority of studies only focused on enhancing the apical uptake in intestinal epithelia, which did not guarantee efficient transport of NPs across the epithelia to enter the blood (basolateral exocytosis).¹⁰ In fact, enzymatic degradation in lysosomes and excessive apical exocytosis back into the intestinal lumen pose great challenges and severely limit the transport efficiency.^11–13 Therefore, bypassing/escaping the endo–lysosome pathways and achieving basolateral exocytosis are crucial to improve the absorption of drugs loaded in ligand modified NPs.^14,15

In our previous work, the addition of HA₂ peptide (lysosomal escaping agent) and metformin (LDLR signaling pathway regulator) could significantly promote the transport of P22 peptide modified NPs.¹⁴ In another study, butyrate modified NPs that targeted MCT-1 receptors could enhance the transepithelial transport of NPs to some degree, but in an in vitro intestinal Caco-2 cell monolayer model, a great proportion of the internalized NPs were still observed to be exocytosed back into the apical side. Consequently, leptin, a butyric acid absorption regulator could increase the basolateral distribution of MCT-1 and boost the basolateral exocytosis.^16,17 These studies indicated that altering the intracellular trafficking routes and (or) regulating the ligand–receptor interactions could be a promising direction to facilitate the epithelial transport of ligand-modified NPs. Nevertheless, the combination regimen, administration sequences and time between the ligand-modified NPs and regulating agents can be great challenges for in vivo delivery, as well as the stability of the regulating agents in the GI tract. Therefore, the combined administration regimen seems too complicated for oral delivery systems. Hence, more simple and practical strategies need to be proposed.

As reported, EGP peptide (KRKKKGKGLGKKRDPCLRKYK) modified NPs, targeting the heparan sulfate proteoglycans (HSPGs) expressed on Caco-2 cells, were more involved in caveolae-mediated transport and could bypass the endo–lysosomal pathways. Compared to unmodified NPs, EGP NPs achieved efficient apical-to-basolateral transcytosis.¹⁸ The results highlighted that introducing suitable ligands would also realize an identical effect for the design and development of effective oral delivery systems.

The low-density lipoprotein receptor-related protein 1 (LRP-1), a member of the low-density lipoprotein receptor family, can interact with over 40 ligands and is widely used as a target in the blood–brain barrier (BBB).^19–21 According to previous reports, the LRP-1 receptor is also highly expressed on intestinal cells (Caco-2 cells and HT29 cells) and hepatocytes, which are responsible for the shuttling of dietary lipid and free fatty acids.^22–24 Angiopep-2 (ANG), a 19 amino acid peptide, has high affinity toward LRP-1 receptor.^25–27 In this study, on the basis of 1,2-distearoyl-sn-glycero-3-phosphatidyl-ethanolamine-polyethylene glycol (DSPE-PEG) modified poly(lactic-co-glycolic acid) (PLGA) NPs, we construct ANG-modified NPs by replacing the DSPE-PEG with DSPE-PEG-ANG aiming to improve oral drug delivery efficiency. Surprisingly, compared with unmodified NPs (PEG NPs), the ANG modification did not influence the endocytosis and intracellular trafficking pathways, but still promoted basolateral transcytosis of NPs. A detailed study revealed that ANG NPs were able to increase the expression of LRP-1 on the apical side of epithelial cells and further induce their redistribution to the basolateral side, thus inducing increased epithelial transportation. As a result, insulin loaded ANG NPs possessed much stronger absorption efficiency and generated maximal blood glucose reduction to 61.46% in diabetic rats. Collectively, the ANG NPs we developed were not only able to target LRP-1, but further increased the transport efficiency by self-regulating the distribution of LRP-1 without additional agents, which provided insights into the rational and simple design for an oral delivery system of protein/peptide drugs.

2. Materials and methods

2.1. Materials

DSPE-PEG₂₀₀₀-Mal and DSPE-PEG₂₀₀₀-COOH were purchased from Ponsure Biotechnology (Shanghai, China). Poly(lactic-co-glycolic acid) (PLGA, 50/50; viscosity, 0.15–0.25 dL g⁻¹) with one methoxyl end group was purchased from Lactel Absorbable Polymers (Birmingham, UK). Soybean phospholipid was bought from Taiwei Pharmaceutical Co. (Shanghai, China). Angiopeptide-2 with cysteine (Cys-ANG) at the N-terminal (KRKKKGKGLGKKRDPCLRKYKC) was chemically synthesized by Ontores Biotechnology Co., Ltd (Hangzhou, China). Porcine insulin (INS, 28.2 IU mg⁻¹) was purchased from Wanbang Bio-Chemical Co., Ltd (Jiangsu, China). 1,2-Distearoyl-sn-glycero-3-phosphatidylethanolamine, 3,3′-dioctadecyloxacarbocyanineperchlorate (DiO), 1,1′-dioctadecyl-3,3,3′,3′-tetramethylindocarbocyanine perchlorate (DiI) and Alamar Blue were purchased from Invitrogen (Carlsbad, CA, USA). NaN₃, M-β-CD, genistin, lovastatin, filipin, chlorpromazine, sucrose, chloroquine and amiloride were all purchased from Sigma-Aldrich (St Louis, MO). BrefedinA, Monecin, LY294002, BCA protein assay kits, Golgi tracker, DAPI and LDH assay kit were acquired from Beyotime (Haimen, Jiangsu, China). Lysosome-tracker green and endoplasmic reticulum tracker green were obtained from KeyGEN Biotech Corp. (Nanjing, China). Rabbit anti-tubulin was purchased from Abcam (Cambridge, MA, USA). Rabbit anti-LRP1 was bought from Affinity Bioscience (Beijing, China). Alexa fluor 647-labeled goat anti-rabbit IgG was purchased from Solarbio Science and Technology Co., Ltd (Beijing, China). All other chemical reagents in the study were of analytical grade.

Animals. Male Sprague-Dawley (SD) rats (weight: 180–220 g), Institute for Cancer Research (ICR) mice (weight: 18–22 g) and Balb/c mice (weight: 18–22 g) were bought from Dashuo Biological Technology (Chengdu, China). All animal procedures were performed in accordance with the Guidelines for the Care and Use of Laboratory Animals of Sichuan University, and approved by the Animal Ethics Committee of Sichuan University.

2.2. Synthesis of DSPE-PEG₂₀₀₀-ANG

DSPE-PEG₂₀₀₀-ANG was synthesized by conjugating Cys-ANG to DSPE-PEG₂₀₀₀-Mal via the specific addition reaction between thiol (–SH) group and maleimide (–Mal) group.²⁸ DSPE-PEG-Mal dissolved in dimethyl formamide (DMF) was added to a solution of Cys-ANG peptide dissolved in phosphate-buffered saline (PBS, pH 7.4) at a molar ratio of 1 : 2 (Cys-ANG peptide : DSPE-PEG₂₀₀₀-Mal). The mixture was allowed to react at room temperature with gentle stirring under nitrogen protection for 24 h. Next, the resultant product was purified using dialysis method for 72 h to remove the free Cys-ANG and DSPE-PEG-Mal. After lyophilization, the successful conjugation was confirmed using ¹H nuclear magnetic resonance (¹H NMR) spectroscopy (Bruker Varian INOVA 400, Billerica, MA, USA).

2.3. Preparation and characterization of nanoparticles (NPs)

ANG modified NPs (ANG-NPs) and unmodified NPs (PEG-NPs) were prepared via self-assembled nano-precipitation methods.²⁹ In brief, PLGA and DSPE-PEG₂₀₀₀-OMe dissolved in dimethyl sulfoxide (DMSO) were mixed with soybean phospholipid dissolved in methanol at required amounts. Then, the resulting organic mixture was added dropwise into deionized water with the volume ratio of 1 : 20 under magnetic stirring (900 rpm) at room temperature, and PEG-NPs were obtained. Then the organic solvent and free compounds were removed by ultrafiltration in Amicon tubes (MWCO 100 kDa; Millipore). ANG-NPs were prepared as described above, using various proportions of DSPE-PEG₂₀₀₀-ANG to replace DSPE-PEG₂₀₀₀ and the fabricated NPs were named 0%, 25%, 50%, 75% and 100% ANG-NPs according to the feed molar ratio of DSPE-PEG₂₀₀₀-ANG, respectively. DiO or DiI loaded NPs were also prepared by the same method.

After re-dispersion in deionized water, the dynamic light scattering (DLS) size and zeta potential of NPs were measured by dynamic light spectrum (Malvern Zeta-size NanoZS90). And the morphological features of NPs were observed via a transmission electron microscope as previously described (TEM, Tecnai G2 F20). Besides, Nanosight LM10 (Malvern Instruments Ltd, UK) was used to determine the size of the NPs.

2.4. In vitro stability study of the NPs

To test the colloidal stability of the NPs, the above-mentioned prepared NPs were suspended in simulated gastric fluid (SGF, pH 1.2), simulated intestinal fluid (SIF, pH 6.8) and phosphate buffered saline (PBS, pH 7.4), respectively, and incubated in a shaker at 100 rpm for 8 h at 37 °C. Then the sizes of particles were measured at predetermined times.³⁰

In addition, dye release behaviors of all NPs were evaluated in PBS (pH 7.4). Briefly, 1 mL of the NP suspension was placed in a dialysis tube (MWCO: 100 kDa) against 40 mL of PBS for 8 h. An aliquot of the exterior medium was withdrawn and supplemented by fresh medium at predetermined time intervals, and measured by Varioskan Flash Multimode Reader.

Moreover, the Förster resonance energy transfer (FRET) assay and dye release study were carried out to detect the in vitro stability of the NPs. In brief, DiO and DiI, as a FRET pair, were co-encapsulated into the same NPs (DiO/DiI-NPs). Meanwhile, DiO loaded NPs (DiO-NPs) and DiI loaded NPs (DiI-NPs) were prepared as control groups. Next, NPs were incubated in SGF, SIF, PBS (pH 5.0) and PBS (pH 7.4). At determined time points, the fluorescence intensity of dual-loaded and single-loaded NPs was detected by Varioskan Flash Multimode Reader (Thermo Fisher Scientific, San Jose, CA, USA) with an excitation wavelength of 460 nm and the emission spectrum was recorded from 490 to 620 nm. The energy transfer efficiency (E) was calculated by the following formula:

E = 1 − F_DA/F_D

where F_D and F_DA represent the fluorescence intensity of the donor alone (DiO-loaded NPs) and with the presence of the acceptor (DiO/DiI-loaded NPs), respectively.

2.5. In vitro cellular absorption study

2.5.1. LRP-1 receptor expression on Caco-2 cells. Caco-2 cells were cultured on cover-slips at the density of 1 × 10⁴ per well for 4 days. Prior to the study, the medium was removed and cells were washed with PBS twice. Subsequently, cells were fixed with 4% paraformaldehyde for 15 min, and blocked by 5% normal goat serum in PBS for 60 min at 37 °C. Then, the cells were incubated with rabbit anti-LRP-1 antibody at 4 °C overnight. Next, the cells were incubated with Alexa fluor 674-labled goat anti-rabbit IgG for 2 h at 37 °C in dark, followed by staining with DAPI for nucleus identification. The fluorescent images of cells were acquired using a CLSM.

2.5.2. Cellular uptake. Before the experiment, we evaluated the toxicity of NPs by Alamar Blue assay and LDH release level (in Method S1 and S2†). Caco-2 cells were seeded in 96-well plates at the density of 5 × 10⁴ cells per mL and cultured for 4 days. After removal of DMEM and washing with PBS, Caco-2 cells were incubated with DiI-labeled NPs for 3 h. Then, the cells were washed twice with cold PBS to separate extracellular NPs. Finally, 100 μL of DMSO was added to destroy the cells. The cell-associated fluorescence intensity of NPs was determined via a Varioskan Flash Multimode Reader and the cell amounts were calculated by Alamar Blue assay.

In order to visualize the uptake of NPs, cells were seeded and cultured on a cover slip in a 12-well plate. After being treated with different Dil-labeled NPs for 3 h, the cells were washed with PBS three times and then fixed with 4% paraformaldehyde. DAPI was applied to stain the nuclei for 3 min before imaging. Cell fluorescence imaging was performed using a confocal laser-scanning microscope (CLSM).

2.5.3. Receptor–ligand interaction study. To verify the specific interaction between ANG and LRP-1 receptor, free ANG and aprotinin³¹ (one of the LRP-1 ligand) were pre-incubated with cells in 96-well plates for 30 min before NP addition, then intracellular fluorescence intensity was measured as reported above. Moreover, the internalized NPs were observed with 30 min pre-incubation of LRP-1 antibody by CLSM.

2.5.4. Endocytosis mechanism investigation. To explore the cellular internalization mechanism, different endocytic inhibitors (summarized in Table S3†) were chosen according to previous reports.¹¹ In short, cells were first incubated with specific inhibitors alone for 30 min, followed by the administration of DiI-labeled NPs and inhibitors together for 3 h. Meanwhile, the NP treatment in the absence of inhibitors served as the control. The fluorescence intensity was measured as mentioned above.

2.5.5. Transcellular transport investigation. Caco-2 cells were seeded and cultured on transwell inserts at the density of 5 × 10⁴ per well for 16–21 days. A cell monolayer with a transepithelial electric resistance (TEER) value higher than 300 Ω cm² was selected for the experiment.³² Prior to the study, the medium in the apical and basolateral chambers was replaced with pre-warmed Hank's Balanced Salt Solution (HBSS) and equilibrated for 30 min. Then, the apical medium was replaced by 200 μL of DiI-labeled NP suspension at 37 °C. At certain intervals, 50 μL of samples were withdrawn from the basolateral chambers and an equal volume of fresh buffer was supplemented. Finally, the fluorescence intensity was measured after the addition of 50 μL of DMSO as described above. And the apparent permeability coefficient (Papp) was calculated according to the following equation:³³
where dQ/dt is the flux rate of DiI from donor side to acceptor side, C₀ is the initial concentration of DiI in the donor compartment, and A is the membrane area (cm²).

The transepithelial electrical resistance (TEER) value of each well was monitored at different time points (0, 1, 2, 4, 6 and 8 h) to test the influence of the NPs on cell monolayer integrity.

2.6. Exocytosis detection from both sides

Firstly, NP suspension and fresh medium were added into the apical and basolateral chambers respectively for 2 h. Then both sides were replaced by fresh medium. At predetermined time points, 50 μL each of apical and basolateral medium were withdrawn, respectively, and the same volume of fresh medium was supplemented. Finally, all the samples were destroyed by 50 μL of DMSO and were measured by Varioskan Flash Multimode Reader. The exocytosis rate was calculated as according to the following equation:
where dQ/dt is the accumulative transport amount of DiI into the apical (or basal) chamber per second.

After bidirectional exocytosis study, the transwell membrane was captured and nuclei of the cell monolayer were stained by DAPI. Subsequently, cells of the monolayer fixed on the glass slides were detected by CLSM.

2.7. Intracellular trafficking pathways of NPs

In order to investigate the fate of internalized NPs, several chemical agents were used to block relevant intracellular trafficking routes.¹⁶ In brief, cells in 96-well plates were incubated with NPs for 2 h, then the non-internalized NPs were removed thoroughly, followed by the addition of corresponding inhibitors for another 2 h. The intracellular fluorescence intensity was also measured by Varioskan Flash Multimode Reader and the cell amounts were calculated by Alamar Blue assay.

To directly observe the intracellular trafficking of NPs, the cells were primarily cultured with DiI-labeled NPs for 2 h, and then were replaced by fresh medium containing different organelle trackers according to previous reports. Finally, the cells were imaged by CLSM.

2.8. Receptor distribution and expression on both sides after NP incubation

Caco-2 cell monolayer on transwell was acquired and incubated with DiO-labeled NPs for 4 h as described above. Afterwards, the cell monolayer was stained with LRP-1 polyclonal antibodies and imaged by CLSM. Besides, cells without NP treatments were regarded as the control group. For the quantitative determination of LRP-1 expression, cell monolayer was digested to form cell suspensions after a 4 h incubation of the NPs. Subsequently, the cell suspensions were incubated with LRP-1 antibody according to the immune-fluorescence labelling methods. Eventually, the samples were analyzed by flow cytometry.

2.9. In situ intestinal absorption study

Firstly, the expressions of LRP-1 on different intestine sections were characterized on male Sprague-Dawley (SD) rats, which were fasted overnight with free access to water. The rats were sacrificed and ligation sections were cut out, gently washed with 0.9% saline solution, fixed with 4% paraformaldehyde for 6 h and dehydrated in 30% sucrose overnight. Then, the samples were embedded in optimal cutting temperature compound and cryo-sectioned into 10 μm slices. Subsequently, the LRP-1 in intestinal cell slices was labeled by immunofluorescence staining methods mentioned above, followed by staining with DAPI for nucleus identification. Finally, slices were visualized by CLSM.

The intestinal absorption study was conducted via an in situ intestinal loop model, as described previously.³⁴ Specifically, male Sprague-Dawley (SD) rats were fasted overnight with free access to water. Rats were anesthetized with chloral hydrate. Next, approximately 2 cm sections of duodenum, jejunum and ileum were ligated at both ends to form intestinal loops. Subsequently, the suspension of 200 μl of DiI-labeled PEG NPs and ANG NPs was injected into all the loops, respectively. After 3 h of administration, rats were sacrificed and ligation sections were cut out, gently washed with 0.9% saline solution, fixed with 4% paraformaldehyde for 6 h and dehydrated in 30% sucrose overnight. Then, samples were embedded in optimal cutting temperature compound and cryo-sectioned into 10 μm slices. Cell nuclei were stained with DAPI, and slices were visualized by CLSM.

2.10. In vivo pharmacokinetic test of dye-labeled NPs

BALB/c mice (male, 25 ± 2 g) were fasted overnight before the experiments. Mice were orally administered DiR-labeled PEG NPs and ANG NPs, and then sacrificed at 2 h, 4 h and 6 h. In the meantime, the stomach, intestine and major organs (heart, liver, spleen, lungs, kidneys) were separated, washed with 0.9% saline solution and detected by an in vivo imaging system.²⁹

Thereafter, coumarin 6-loaded NPs (coumarin 6: 1 mg kg⁻¹) were administered to the mice via oral gavage, blood samples (20 μL) were collected from the orbit, and the fluorescence intensity of coumarin 6 was measured using a microplate reader, as previously described.^35,36

The blood biocompatibility of NPs was indicated by hemolysis ratio.³⁷ Freshly collected erythrocytes were incubated with different concentrations of NPs for 2 h (37 °C, 60 rpm). Finally, the supernatant was taken out after centrifugation and the hemolysis ratio was measured via UV–vis spectrum at 545 nm. The erythrocytes mixed with saline and 0.1% Triton solution served as the negative control and positive control, respectively. The hemolysis ratio was calculated from the following equation:

2.11. Preparation and characterization of insulin-loaded NPs

Firstly, soybean phospholipid was mixed with porcine insulin (INS) to form phospholipid complexes, then mixed with PLGA and DSPE-PEG₂₀₀₀/DSPE-PEG₂₀₀₀-ANG completely. Thereafter, the mixture was transferred to deionized water under stirring via the self-assembly nanoprecipitation method. The INS-loaded NPs were collected by ultrafiltration centrifugation as described above and re-dispersed into deionized water to detect size and zeta potential.

For INS encapsulation efficiency and loading capacity determination, NP suspension was transferred into an ultrafiltration tube to separate free INS and INS-loaded NPs. Then the amounts of INS encapsulated in NPs and total INS were quantitatively detected via reverse high performance liquid chromatography (RP-HPLC) respectively. INS encapsulation efficiency (EE%) was calculated by the following formula:

To obtain INS loading capacity, a certain volume of INS-loaded NPs had to undergo freeze-drying process after ultrafiltration. Then the drying products were weighed as mass of INS-loaded NPs. Drug loading capacity (DL%) was calculated by the following formula:

2.12. Drug release and enzymatic stability

To evaluate drug release behavior, insulin-loaded NPs were transferred into dialysis tubes (MWCO: 100 kDa), which were immersed in SGF (pH 1.2) for 2 h and in SIF (pH 6.8) for 6 h more. At certain time intervals, 50 μL of samples was withdrawn and was measured by RP-HPLC.³⁸

For the enzymatic stability assessment, insulin-loaded NPs and free insulin were dispersed into SIF containing trypsin respectively. Aliquots of the samples were withdrawn at determined times, and 100 μL of 0.1 M ice-cold trifluoroacetic acid was immediately added to terminate the enzymatic interaction and destroy the NPs as well. Thereafter, the remaining insulin was measured using RP-HPLC.³⁹

2.13. Study of bioactivity of insulin-loaded NPs

Insulin released from the insulin-loaded NPs and original free insulin under the same concentration were subcutaneously administered to normal SD rats respectively. And the blood glucose level was determined with the JPS-6 blood glucose monitoring system at certain time points.

2.14. Hypoglycemic effects in streptozotocin induced diabetic rats

The diabetic rats were induced by injecting male Sprague-Dawley rats with streptozotocin (70 mg kg⁻¹) as previously described.⁴⁰ Prior to the experiment, diabetic rats were fasted overnight but had access to water. The rats were randomly divided into 5 groups for different treatments. Blood samples were collected from the tail vein at determined time points and blood glucose levels were measured with a glucose meter. Then the curves of the blood glucose level of all the groups with time were drawn. Thereafter, pharmacological availability (PA%) was calculated according to the following formula:
where AAC(Oral) and AAC(Saline) represented area above the curve of oral INS-PEG NPs (INS-ANG NPs) and saline groups, respectively. Dose(SC) and Dose(Oral) were the administered INS dose of oral INS-PEG NPs (INS-ANG NPs) and saline groups, respectively.

2.15. Statistical analysis

All differences between groups were calculated by SPSS program, a two-tailed Student's t-test and one-way ANOVA analysis. All data were presented as the mean ± SD and summarized by GraphPad Prism 8.0. Significant differences (p < 0.05) were defined as * or #, and ** or ## meant highly significant difference (p < 0.01). The results are presented as mean ± SD.

3. Results and discussions

3.1. Synthesis of DSPE-PEG₂₀₀₀-ANG

DSPE-PEG₂₀₀₀-ANG was synthesized by the Michael addition reaction between the maleimide on DSPE-PEG₂₀₀₀-Mal and the -SH on ANG peptide shown in Fig. S1,† which was confirmed through ¹H NMR spectra (Fig. S2†). Specifically, the disappearance of the featured proton peaks from the Mal groups after reacting (showed in the red circle) suggested the successful conjugation of ANG peptide to the DSPE-PEG₂₀₀₀-Mal.

3.2. Preparation and characterization of NPs

ANG modified and unmodified NPs were prepared by nano-precipitation methods and their schematic structures are shown in Fig. 1A. PEG NPs (unmodified NPs), 25%, 50%, 75% and 100% ANG-NPs were obtained with various feed ratios of DSPE-PEG₂₀₀₀-ANG (0%, 25%, 50%, 75% and 100%, mol/mol). The DLS sizes of all NPs dispersed in deionized water ranged from 97 nm to 118 nm. Meanwhile, the zeta potential exhibited an increasing tendency from negative charge to electroneutrality as the ANG modification augmented, which demonstrated the successful ANG decoration (Fig. 1B). Moreover, uniform size distributions and obvious core–shell structures were confirmed by nanoparticle-tracking analysis (NTA) in Fig. 1C and transmission electron microscopy (TEM) in Fig. 1D, respectively. All dye-loaded NPs were prepared by the same method and their sizes and zeta potentials were also characterized and are shown in Table S1.†

Fig. 1 (A) The schemes of PEG NPs and 100% ANG NPs. (B) The DLS sizes and zeta potentials of all NPs. Error bars represent SD (n = 3). (C) Uniform size distribution of PEG NPs and 100% ANG NPs measured by the Malvern Nano Sight LM10 system. (D) Transmission electron microscopy (TEM) images of PEG NPs and 100% ANG NPs. Scale bars: 50 nm. (E) Size change of all NPs after incubation with SGF (pH 1.2), SIF (pH 6.8) and PBS (pH 7.4) within 8 h. (F) FRET efficiency changes of fluorescence-labeled NPs after incubation with SGF (pH 1.2), SIF (pH 6.8), PBS (pH 5.0) and PBS (pH 7.4). Error bars represent SD (n = 3). (G) The percentage of DiI released from NPs in PBS.

In general, it is very vital for orally administered NPs to maintain colloidal stability in the severe gastrointestinal environment, which can provide essential protections to proteins/peptides.⁴¹ The simulated gastric fluids (SGF), simulated intestinal fluids (SIF) and phosphate buffer solution (PBS) were used to mimic different physiological media. As shown in Fig. 1E, no significant size change of NPs was observed after incubation with SGF, SIF and PBS (pH 7.4) during 8 h. Besides, the DiI-DiO co-loaded NPs could generate significant FRET effect and could be maintained stable in various media for 8 h, indicating the integrity of the NP structures in the GI tract (Fig. 1F). In addition, for the DiI loaded-NPs, no obvious dye release was detected in PBS (Fig. 1G), which means the florescence of the encapsulated DiI can represent the behaviors of NPs in the following in vitro quantitative and qualitative studies.⁴² Collectively, we successfully prepared PEG NPs and ANG-modified NPs with uniform size distributions and good colloidal stability in the GI tract.

3.3. ANG modification increased the transepithelial transport of NPs via ligand–receptor interaction

The tight intestinal epithelial layer is an obstacle for NPs to enter blood circulation. Caco-2 human colon carcinoma cell line was selected as an in vitro intestinal cell model to evaluate the absorption efficiency of ANG modified-NPs.⁴³ Firstly, the high expression of LRP-1 on Caco-2 cells was confirmed by CLSM (Fig. 2A). Both results of Alamar Blue assay and LDH release level indicated no cell toxicity was observed under the incubation concentrations of NPs (Fig. S4A and Fig. S4B†). Fig. 2B showed that the internalization of NPs gradually increased with the augmenting of the ANG ratio, and 100% ANG NPs showed the highest cellular uptake which was 4.1-fold higher than that for PEG-NPs. Besides, the 100% ANG-NP-treated group exhibited stronger green fluorescence than PEG-NPs on CLSM images in Fig. 2C, indicating a larger amount of ANG NPs were internalized. This ANG content-dependent internalization revealed the superiority of ANG modification in promoting the cellular uptake. Since 100% ANG-NPs possessed the highest internalization, it was chosen for the following investigation.

Fig. 2 (A) Characterization of LRP-1 receptor expressions on Caco-2 cells by CLSM. (B) Cellular uptake of NPs with different amounts of ANG surface modification on Caco-2 cells, *p < 0.05, **p < 0.01, versus PEG NPs. #p < 0.05, ##p < 0.01. (C) Uptake of PEG NPs and ANG NPs in Caco-2 cells by CLSM. Scale bar: 50 μm. (D) CLSM images of Caco-2 cells exposed to NPs with or without LRP-1 antibody. DiO (green) represented NPs, DAPI (blue) represented nucleus. Scale bar: 20 μm. (E) Cellular uptake of PEG NPs and ANG NPs with free ANG and free Aprotinin. Error bars represent SD (n = 3). **p < 0.01 and ##p < 0.01 versus the control groups. (F) Study of endocytic of NPs on Caco-2 cells in the presence of various inhibitors. Error bars represent SD (n = 3). **p < 0.01 and *p < 0.05 versus the PEG NPs group, #p < 0.05 and ##p < 0.01 versus the ANG NPs group. (G) Papp value and transported amounts of PEG NPs and ANG NPs across Caco-2 cell monolayer for 6 h.

As the ANG peptide was reported as a ligand of LRP-1, we seek to understand the role LRP-1 played in the transcellular transport of the NPs. After pre-incubating with specific LRP-1 antibody to block LRP-1 receptor, the green fluorescence intensity of the ANG NPs group was apparently weaker than that of the control group, while the internalization of PEG NPs was not affected (Fig. 2D). Meanwhile, the addition of free ANG peptide (400 μg ml⁻¹) significantly reduced the internalization of ANG-NPs, and had no effect on PEG-NPs as well (Fig. 2E). As free ANG may produce competitive inhibition by binding with the receptors, these results indicated that the specific interactions between ANG and the receptor may help with the internalization of ANG NPs. Besides, aprotinin, as a classic LRP-1 ligand, also gave a similar inhibition result, which could further validate this point of view.

In addition, the endocytosis mechanisms of both NPs were investigated by using specific inhibitors.⁴⁴ As shown in Fig. 2F, cellular uptake of both NPs had a significant decline after NaN₃ treatment, implying energy-dependent endocytosis. Meanwhile, the internalization of both NPs involved caveolae-mediated, clathrin-mediated and lipid raft pathways, as well as micropinocytosis, which indicated that ANG modification had little influence on the endocytosis pathways of related NPs.

To further explore the potential of ANG modification on transepithelial transport of NPs, transport efficiency of PEG-NPs and ANG-NPs was investigated on Caco-2 cell monolayer model (Fig. 2G). The TEER values had no change during 8 h exposure to PEG NPs and ANG NPs (Fig. S4C†). After 6 h incubation, the Papp value and transported amounts of ANG-NPs had a 2.1-fold and a 1.74-fold increase, respectively, as compared with that of PEG-NPs. Therefore, ANG modification can enhance the transepithelial transport of NPs.

3.4. Elevated basolateral exocytosis favored transepithelial transport

Evidence showed that most ligand-modified NPs tended to get into the dilemma of “easy endocytosis but hard transcytosis”. One reason for this dilemma is that a certain proportion of endocytosed NPs might be transported back into the intestinal lumen owing to the distinct transporter expression on the apical and basolateral membranes, which greatly limited the final transcytosis.^44,45 Thus, exploring the exocytosis from both sides of epithelial cells is quite essential to study the behavior of NPs. A polarized Caco-2 cell monolayer on transwells was used and the specific operations were conducted according to Fig. 3A.

Fig. 3 (A) Schematic illustrations of bidirectional NPs’ exocytosis studies in Caco-2 monolayer cells grown on transwell devices. (B) Relative exocytosis rates of NPs from apical and basolateral sides. The exocytosis rate of NPs from basolateral to apical side were taken as 100%, *p < 0.05 versus the PEG NPs control group, error bars represent SD (n = 3). (C) CLSM images (z-axis) of cell monolayer grown on transwell taken in the membrane pore layer after incubation with NPs for 2 h and fresh medium for 4 h at 37 °C. Scale bar: 50 μm.

As shown in Fig. 3B, the internalized PEG NPs showed more exocytosis from the apical side of Caco-2 cells, and the amount of NPs detected from the basolateral side was about 79% compared with that from the apical side. Interestingly, equivalent exocytosis rates of both sides were observed in the ANG NP-treated group. Similarly, the results of CLSM in Fig. 3C also demonstrated the different distribution patterns of the two kinds of NPs on the cell monolayer. In detail, after 4 h of exocytosis, PEG NPs mainly distributed near the apical side of the Caco-2 cell monolayer, while ANG NPs had a more symmetrical distribution on both sides of cells, implying that ANG modification could facilitate more basolateral exocytosis of NPs.

3.5. ANG functionalization did not alter the intracellular trafficking routes of NPs

In general, NPs have to undergo complex intracellular trafficking, including Golgi/plasma membrane (PM), endoplasmic reticulum (ER)/Golgi pathway, endo–lysosomal pathway and microtubule-dependent intracellular movements.^12,46 Among them, more participation in the endo–lysosomal pathway could result in more degradations and poor transcytosis due to the enzymes and acidic environment. In contrast, NPs are believed to have more chances to be transported outside cells when entering the Golgi/ER related retrograde pathway. As ligand modification could sometimes profoundly change the trafficking routes of the targeted NPs inside cells which may have a huge impact on the subsequent exocytosis or transepithelial efficiency, we then compared the intracellular trafficking pathways of NPs before and after ANG modification.

Monecin and brefeldin A were used to inhibit Golgi/plasma membrane (PM) and endoplasmic reticulum (ER)/Golgi pathway, respectively. As the results show in Fig. 4A and C, exocytosis of both the ANG NPs and PEG NPs was significantly inhibited by the two chemicals, indicating that both the NPs were transported via Golgi- and ER-associated pathways. Consistently, significant colocalizations of both NPs with Golgi and ER could be observed in Fig. 4B and D.

Fig. 4 Intracellular trafficking investigations of NPs in different organelles. (A, C, E and G) The influence of different inhibitors on intracellular trafficking and exocytosis of NPs, error bars represent SD (n = 3). **p < 0.01 and *p < 0.05 versus the PEG NPs control group, #p < 0.05 and ##p < 0.01 versus the ANG NPs control group. (B, D, F and H) The colocalization of NPs with Golgi, ER, microtubules and lysosomes in Caco-2 cells. Scale bar: 20 μm.

Then, exocytosis inhibition by nocodazole and colchicine indicated the microtubule-dependent intracellular movements of both NPs (Fig. 4E), which was consistent with qualitative results of the CLSM study shown in Fig. 4F.

Next, LY294002 and Chloroquine were used to inhibit the endo–lysosomal pathway (Fig. 4G). In addition, the NPs also showed obvious co-localization (R_r > 0.5) with lysosomes (Fig. 4H), which further confirmed the endo/lysosomal trafficking of both NPs.

Taken together, ANG modification did not change the intracellular trafficking routes of NPs as compared with non-targeted PEG-NPs. Therefore, more detailed studies are still needed to further explore the secrets of ANG modification.

3.6. ANG redistributes LRP-1 to basolateral side of intestinal epithelia

Because the ANG ligand modification had limited impact on the intracellular trafficking pathways across the epithelial monolayer, we attempted to seek other mechanisms to elucidate the enhanced basolateral exocytosis of ANG NPs. Since the trafficking of ligand-functionalized NPs may follow a gradient of targeted receptors inside the cells, we next investigated and compared the expression of LRP-1 receptors on both apical and basolateral sides of intestinal epithelia (Fig. 5A).

Fig. 5 (A) Schematic illustrations about exploring the interactions of NPs and LRP-1 in Caco-2 cell monolayer grown on transwell devices. (B) CLSM images (x − z) of LRP-1 distributions in PBS treated group, CLSM images (x − z, y − z) of LRP-1 distributions exposed to PEG NPs (C) and ANG NPs (D). Scale bar: 50 μm. (E) The characterization of LRP-1 expression in Caco-2 cell monolayer after incubation with NPs for 4 h by cyto-flowmetry, error bars represent SD (n = 3). **p < 0.01 and *p < 0.05. Orthogonal CLSM images and relative intensity of LRP-1 and NPs on Caco-2 cell monolayer after 4 h transcytosis of PEG NPs (F) and ANG NPs (G). Scale bar: 50 μm.

Results in Fig. 5B show that LRP-1 normally accumulate on the apical membrane of the untreated Caco-2 cell monolayer. Treatment with PEG NPs did not change the overall distribution of LRP-1 with significant higher LRP-1 expression on the apical side than on the basolateral side of the epithelial monolayer (Fig. 5C). Interestingly, considerable expression of LRP-1 appeared on the basolateral side of intestinal epithelia after the treatment with ANG-NPs (Fig. 5D). In addition, ANG NPs increased the overall LRP-1 expression by 1.2-fold as compared with the untreated control and PEG NPs (Fig. 5E). Consequently, while PEG NPs mainly locate on the apical side of the monolayer (Fig. 5F), ANG NPs were more prone to accumulate at the basolateral side after the interaction with the redistributed LRP-1 (Fig. 5G).

Collectively, these results revealed a unique feature of ANG ligand and well explained its ability to promote apical-to-basolateral transport of NPs: initially, LRP-1 receptors were primarily present on the apical side of intestinal epithelia, representing a promising target that was easily accessible by ANG NPs for endocytosis; upon the specific binding between ANG and LRP-1, the redistribution of LRP-1 to basolateral side was triggered, thereby transporting substantial amounts of ANG NPs also to the basolateral side for exocytosis.

3.7. ANG NPs have superior absorption efficiency in in vivo distribution and pharmacokinetics

We proved the great transepithelial efficiency and found a special apical-to-basolateral transport mechanism of ANG NPs. Then, in vivo oral potential of ANG NPs was further evaluated on animals. Firstly, we verified the LRP-1 expressions on intestinal regions slices of fasted rats by immunofluorescence staining. As shown in Fig. 6A, duodenum, jejunum and ileum all had significant distributions of LRP-1, while the colon showed no fluorescence. Therefore, duodenum, jejunum and ileum regions were chosen as in situ intestinal loops to visualize the oral absorption efficiency of NPs. In Fig. 6B, obvious red fluorescence was seen at the villi tops, indicating that the nanoparticles can be absorbed by the intestinal villi. Meanwhile, the red fluorescence on jejunum was significantly stronger than that on the duodenum and ileum, which may be related to the annular folds and densely distributed villi of jejunum, as well as the rich blood vessels. Specifically, the ANG NP-treated group exhibited stronger red fluorescence in the lamina propria near the villi bottom than the PEG NP-treated group, indicating that ANG NPs are more likely to be absorbed. This may be due to the interaction between ANG ligands and LRP-1 receptors distributed on the intestinal surface which may enhance the effective absorption of nanoparticles.

Fig. 6 (A) Characterization of LRP-1 receptor expressions on SD rat intestine sections. (B) Localization of absorbed PEG NPs and ANG NPs in in situ intestinal loop model on fasted rats at 2 h by CLSM. (C) In vivo distribution of DiR in mice detected by in vivo imaging system after oral administration of PEG NPs or ANG NPs at 2 h, 4 h and 6 h. The organs from the top to bottom were heart, liver, spleen, lungs and kidneys. (D) Concentration of courmarin-6 in blood after oral administration of courmarin-6 loaded NPs. Error bars represent SD (n = 5), *p < 0.05, **p < 0.01, versus PEG NPs.

Next, we performed intragastric administration of DiR-labeled NPs to fasted mice and in vivo imaging system were employed to visualize the bio-distribution of NPs in the GI tract and major organs. As shown in Fig. 6C, fluorescence intensities of NPs in the stomach and intestine decreased over time with obvious distribution in organs. It is worth noting that remaining amounts of ANG NPs in the intestine were much less than that of PEG NPs, while ANG NPs showed more distribution in the liver. Meanwhile, the blood concentration–time curves of Courmarin-6 loaded NPs via oral administration were obtained (Fig. 6D) and pharmacokinetic parameters are listed in Table S3.† Apparently, ANG NPs were quickly absorbed in the first 30 min, reaching peak concentration at 1 h and remained at a higher level than that of PEG NPs for the following duration. In other words, ANG NPs can be more efficiently absorbed into blood circulation compared with PEG NPs, which revealed a promising ligand-modified system for oral delivery.

Moreover, we also conducted the blood erythrocyte-induced hemolysis to investigate the blood biocompatibility of NPs. The hemolysis ratios of PEG NPs and ANG NPs under different concentrations were obviously lower than 5.0% of the international standard⁴⁷ in Fig. S4D,† suggesting good blood biocompatibility of ANG NPs for drug delivery application.

3.8. ANG modified INS-loaded NPs produced preferable hypoglycemic effect in diabetic rats

To further investigate the in vivo potential of NPs, insulin (INS) was chosen as the model protein drug to evaluate in vivo hypoglycemic effects of ANG NPs. For INS-loaded NPs, the characterization including DLS size, zeta potential, PDI, EE% and DL% were summed in Table 1. The NTA exhibited uniform size distributions for both NPs (Fig. 7A).

Fig. 7 (A) Schematic images of INS-loaded NPs. Uniform size distributions of INS-PEG NPs and INS-ANG NPs by NTA. (B) Cumulative amounts of INS released from INS-PEG NPs and INS-ANG NPs in SGF (pH 1.2, without enzymes) and SIF (pH 6.8, without enzyme). Error bars represent SD (n = 3). (C) The percentage of the remaining insulin undegraded at different time points in PBS (pH 6.8) with trypsin. Error bars represent SD (n = 3). (D) The profile of the blood glucose levels in mice after subcutaneous injection of free insulin and insulin released from NPs. Error bars represent SD (n = 5).(E) Blood glucose level in fasted diabetic rats via oral administration of the INS-loaded NPs and free insulin (50 IU kg⁻¹), subcutaneous injection of free insulin (5 IU kg⁻¹), or oral administration of saline. Error bars represent SD (n = 5). **p < 0.01 and *p < 0.05 versus the oral free insulin group, #p < 0.05 and ##p < 0.01 versus the INS-PEG NPs group.

Table 1 Characterization of Ins-PEG NPs and Ins-ANG NPs (n = 3)

Samples	Size (nm)	PDI	Zeta potential (mV)	EE%	DL%
PDI: Polydispersity index; EE%: INS encapsulation efficiency; DL%: drug loading capacity.
INS-PEG NPs	107.7 ± 1.124	0.255 ± 0.013	−17.8 ± 0.96	50.61 ± 2.3	10.1± 2.68
INS-ANG NPs	121.4 ± 0.896	0.184 ± 0.028	−18.2 ± 0.45	47.04 ± 0.99	10.92 ± 0.662

---

Meanwhile, both NPs possessed similar sustained release behaviors (Fig. 7B) and the protection of loaded insulin against the enzyme (Fig. 7C). The results suggested that ANG modification had little influence on the encapsulation and release of INS. Moreover, the blood glucose levels in mice after the subcutaneous injection of free insulin and insulin released from NPs were investigated. It revealed organic solvent introduced into the system and the preparation process did not impair the bioactivity of INS from the Fig. 7D.

Subsequently, the hypoglycemic effects of INS-loaded NPs were investigated in streptozotocin (STZ)-induced diabetic rats (Fig. 7E). Consistent with previous reports, free INS solution (5 IU kg⁻¹) by subcutaneous injection elicited the strongest reduction in blood glucose level, even <12% of the initial glucose values, which also reflected the potential hypoglycemia risks brought about by INS subcutaneous injection. Besides, oral free INS solution (50 IU kg⁻¹) did not generate any hypoglycemic activity owing to its degradation and poor absorption in the GI tract. For PEG NPs-treated groups, the blood glucose level of INS-PEG NPs showed a maximum decline to 87.76%. Notably, INS-ANG NPs (50 IU kg⁻¹) achieved significant and persistent hypoglycemic effects with the maximum reduction to 61.46% at 2 h. Besides, the pharmacological availability (PA%) of the INS-ANG NPs group was 6.07%, which was 2.27-fold higher than that of INS-PEG NPs (Table 2). These results are consistent with the in vitro results. In conclusion, ANG modification could enhance hypoglycemic efficiency of INS-loaded NPs.

Table 2 Pharmacological availability of INS loaded formulations in diabetic rats (n = 5)

Samples	Dose (IU kg^−1)	PA%
Each value represents the mean ± SD (n = 5). PA%: pharmacological availability. Sc: subcutaneous injection.
Free INS solution(oral)	50	0.34 ± 3.71
INS-PEG NPs(oral)	50	2.67 ± 1.80
INS-ANG NPs(oral)	50	6.07 ± 0.81

---

4. Conclusion

In this study, ANG ligand was applied for the first time to construct orally delivered NPs. The oral efficiency and transport mechanisms of ANG modified NPs were systematically investigated comparing to PEG NPs. Results showed that ANG modification can absolutely enhance the cellular uptake and transcytosis efficiency of NPs. Mechanistic studies revealed that ANG functionalization did not alter the intracellular transport pathway, but still promoted more basolateral exocytosis of NPs. A major reason for the enhanced basolateral exocytosis was demonstrated as ANG NPs could increase the expression and basal accumulation of LRP-1. Thus, by this self-favorable adjustment, ANG NPs had greater ability to go through an apical-to-basal transport, thus leading to improved oral absorption. Moreover, ANG NPs achieved increased in vivo absorption promotion both in normal mice and diabetic rats.⁴⁸ This study may support ANG as a potential ligand for the design of active targeting NPs for oral drug delivery and may also provide some helpful references for polarized cells on transcytosis.

Conflicts of interest

The authors declare no competing financial interest.

Acknowledgements

We gratefully acknowledge financial support from the National Science Foundation for Distinguished Young Scholars (81625023), the National Natural Science Foundation of China (81872818), and the Major Research Plan of National Natural Science Foundation of China (81690261).

References

1. W. Shan, X. Zhu, M. Liu, L. Li, J. Zhong, W. Sun, Z. Zhang and Y. Huang, ACS Nano, 2015, 9, 2345–2356.
2. X. Zhu, J. Wu, W. Shan, W. Tao, L. Zhao, J. M. Lim, M. D'Ortenzio, R. Karnik, Y. Huang, J. Shi and O. C. Farokhzad, Angew. Chem., Int. Ed., 2016, 55, 3309–3312.
3. M. Liu, J. Zhang, X. Zhu, W. Shan, L. Li, J. Zhong, Z. Zhang and Y. Huang, J. Controlled Release, 2016, 222, 67–77.
4. J. Wu, Y. Zheng, M. Liu, W. Shan, Z. Zhang and Y. Huang, ACS Appl. Mater. Interfaces, 2018, 10, 9916–9928.
5. W. Shan, X. Zhu, W. Tao, Y. Cui, M. Liu, L. Wu, L. Li, Y. Zheng and Y. Huang, ACS Appl. Mater. Interfaces, 2016, 8, 25444–25453.
6. P. Lundquist and P. Artursson, Adv. Drug Delivery Rev., 2016, 256.
7. Y. Yun, Y. W. Cho and K. Park, Adv. Drug Delivery Rev., 2013, 65, 822–832.
8. X. Song, L. Rui, H. Deng, L. Ye, Y. Cui, Z. Hua, W. Dai, H. Bing, Z. Ying and X. Wang, Biomaterials, 2018, 180, 78–90.
9. E. M. Pridgen, F. Alexis, T. T. Kuo, E. Levy-Nissenbaum, R. Karnik, R. S. Blumberg, R. Langer and O. C. Farokhzad, Sci. Transl. Med., 2013, 5, 167.
10. W. Fan, D. Xia, Q. Zhu, X. Li and Y. Gan, Biomaterials, 2018, 151, 13–23.
11. Y. Xu, J. Xu, W. Shan, M. Liu, Y. Cui, L. Li, C. Liu and Y. Huang, Int. J. Pharm., 2016, 500, 42–53.
12. W. Fan, D. Xia, Q. Zhu, L. Hu and Y. Gan, Drug Discovery Today, 2016, 21, 856–863.
13. W. Du, Y. Fan, N. Zheng, B. He, L. Yuan, H. Zhang, X. Wang, J. Wang, X. Zhang and Q. Zhang, Biomaterials, 2013, 34, 794–806.
14. Y. Cui, W. Shan, R. Zhou, M. Liu, L. Wu, Q. Guo, Y. Zheng, J. Wu and Y. Huang, Nanoscale, 2018, 10, 11443–11456.
15. Y. Xu, Y. Zheng, L. Wu, X. Zhu and Z. Zhang, ACS Appl. Mater. Interfaces, 2018, 10, 9315–9324.
16. W. Lei, B. Yuli, L. Min, L. Lian, S. Wei, Z. Zhirong and H. Yuan, Mol. Pharm., 2018, 15, 4273–4283.
17. D. Yang, D. Liu, M. Qin, B. Chen, S. Song, W. Dai, H. Zhang, X. Wang, Y. Wang and B. He, ACS Appl. Mater. Interfaces, 2018, 10, 11443–11456.
18. Z. Yaxian, W. Jiawei, S. Wei, W. Lei, Z. Rui, L. Min, C. Yi, Z. Minglu, Z. Zhirong and H. Yuan, ACS Appl. Mater. Interfaces, 2018, 10, 34039–34049.
19. Y. Zhao, D. Li, J. Zhao, J. Song and Y. Zhao, Rev. Neurosci., 2016, 27, 623–634.
20. Y. Jiang, J. Zhang, F. Meng and Z. Zhong, ACS Nano, 2018, 12, 11070–11079.
21. Y. Zou, Y. Liu, Z. Yang, D. Zhang and Y. Lu, Adv. Mater., 2018, 30, 1803717.
22. T. Abolghasem, G. Rebecca, R. Anna and K. Terry, Toxins, 2017, 9, 237.
23. D. Schttelndreier, K. Seeger, G. A. Grassl, M. R. Winny, R. Lindner and H. Genth, Front. Microbiol., 2018, 9, 1483.
24. G. W. Go and A. Mani, Yale J. Biol. Med., 2012, 85, 19–28.
25. Y. Li, X. Zheng, M. Gong and J. Zhang, Oncotarget, 2016, 7, 79401–79407.
26. K. Parrish, J. Sarkaria and W. Elmquist, Clin. Pharmacol. Ther., 2015, 97, 336.
27. Q. Guo, Q. Zhu, T. Miao, J. Tao, X. Ju, Z. Sun, H. Li, G. Xu, H. Chen and L. Han, J. Controlled Release, 2019, 303, 117–129.
28. W. Han, G. Yin, X. Pu, X. Chen, X. Liao and Z. Huang, J. Biomater. Sci., Polym. Ed., 2017, 28, 1695–1712.
29. L. Wu, M. Liu, W. Shan, X. Zhu, L. Li, Z. Zhang and Y. Huang, J. Controlled Release, 2017, 262, 273–283.
30. Y. Bai, R. Zhou, L. Wu, Y. Zheng, X. Liu, R. Wu, X. Li and Y. Huang, J. Mater. Chem. B, 2020, 8, 2636–2649.
31. H. Xin, X. Sha, X. Jiang, L. Chen, K. Law, J. Gu, Y. Chen, X. Wang and X. Fang, Biomaterials, 2012, 33, 1673–1681.
32. J. Zhang, X. Zhu, Y. Jin, W. Shan and Y. Huang, Mol. Pharmaceutics, 2014, 11, 1520–1532.
33. M. Guo, M. Wei, W. Li, M. Guo, C. Guo, M. Ma, Y. Wang, Z. Yang, M. Li and Q. Fu, J. Controlled Release, 2019, 307, 64–75.
34. K. Maisel, L. Ensign, M. Reddy, R. Cone and J. Hanes, J. Controlled Release, 2015, 197, 48–57.
35. L. Wu, Y. Bai, L. Wang, X. Liu, R. Zhou, L. Li, R. Wu, Z. Zhang, X. Zhu and Y. Huang, J. Controlled Release, 2020, 323, 151–160.
36. L. Li, W. Sun, L. Li, Y. Liu, L. Wu, F. Wang, Z. Zhou, Z. Zhang and Y. Huang, Nanoscale, 2017, 9, 314–325.
37. W. Shan, Y. Cui, M. Liu, L. Wu, Y. Xiang, Q. Guo, Z. Zhang and Y. Huang, Int. J. Pharm., 2017, 530, 89–98.
38. X. Zhu, J. Wu, W. Shan, Z. Zhou, M. Liu and Y. Huang, Adv. Funct. Mater., 2016, 26, 2728–2738.
39. W. Shan, X. Zhu, W. Tao, Y. Cui, M. Liu, L. Wu, L. Li, Y. Zheng and Y. Huang, ACS Appl. Mater. Interfaces, 2016, 8, 25444–25453.
40. B. Sarmento, S. Martins, D. Ferreira and E. B. Souto, Int. J. Nanomed., 2007, 2, 743.
41. M. Liu, L. Wu, W. Shan, Y. Cui and Y. Huang, J. Mater. Chem. B, 2018, 6, 593–601.
42. F. Deng, H. Zhang, X. Wang, Y. Zhang, H. Hu, S. Song, W. Dai, B. He, Y. Zheng and X. Wang, ACS Appl. Mater. Interfaces, 2017, 9, 5803–5816.
43. A. R. Hilgers, R. A. Conradi and P. S. Burton, Pharm. Res., 1990, 7, 902–910.
44. G.-H. Chai, Y. Xu, S.-Q. Chen, B. Cheng, F.-Q. Hu, J. You, Y.-Z. Du and H. Yuan, ACS Appl. Mater. Interfaces, 2016, 8, 5929–5940.
45. P. L. Tuma and A. L. Hubbard, Physiol. Rev., 2003, 83, 871–932.
46. R. Sakhtianchi, R. F. Minchin, K. B. Lee, A. M. Alkilany, V. Serpooshan and M. Mahmoudi, Adv. Colloid Interface Sci., 2013, 201–202, 18–29.
47. X. Li, X. Kong, Z. Zhang, K. Nan, L. Li, X. Wang and H. Chen, Int. J. Biol. Macromol., 2012, 50, 1299–1305.
48. D. Yang, D. Liu, H. Deng, J. Zhang, M. Qin, L. Yuan, X. Zou, B. Shao, H. Li and W. Dai, ACS Nano, 2019, 13, 5058–5076.

---

Footnote

† Electronic supplementary information (ESI) available. See DOI: 10.1039/d1bm00020a

---