Dual-responsive nanoparticles loading bevacizumab and gefitinib for molecular targeted therapy against non-small cell lung cancer

Abstract

The combination of vascular endothelial growth factor (VEGF) inhibitors and tyrosine kinase inhibitors (TKIs) is newly available for molecular targeted therapy against non-small cell lung cancer (NSCLC) in clinic. However, the therapeutic benefits remain unsatisfying due to the poor drug delivery to targets of interest. In this study, we developed bevacizumab-coated gefitinib-loaded nanoparticles (BCGN) with dual-responsive drug release for inhibiting tumor angiogenesis and phosphorylation of epidermal growth factor receptor (EGFR). Through an exogenous corona strategy, bevacizumab is easily coated on gefitinib-loaded nanoparticles via electrostatic interaction. After intravenous injection, BCGN are efficiently accumulated in NSCLC tumors as confirmed by dual-model imaging. Bevacizumab is released from BCGN upon oxidation in tumor microenvironment, whereas gefitinib is released after being internalized by tumor cells and disassembled in reduction cytoplasm. The dual-responsive release of bevacizumab and gefitinib significantly inhibits tumor growth in both A549 and HCC827 human NSCLC models. Our approach provides a promising strategy to improve combinational molecular targeted therapy of NSCLC with precisely controlled drug release.

Introduction

Lung cancer is the leading cause of cancer-related death worldwide and non-small cell lung cancer (NSCLC) accounts for 84% of all lung cancer diagnoses, with only 20%–30% population of 5-year-survival [1–4]. Molecular targeted therapies including tyrosine kinase inhibitors (TKIs) and anti-vascular endothelial growth factors (VEGFs) antibodies play important role in treating NSCLC [5–10]. Epidermal growth factor receptor (EGFR) is overexpressed in 60%–80% of NSCLC, thus inhibition of EGFR TK has become an important target in treating NSCLC [11]. EGFR TKIs such as gefitinib, erlotinib, afatinib, and osimertinib are commonly available in practice [12–14]. Gefitinib is the first-generation EGFR TKIs with an anilinoquinazoline structure that reversibly binds to the ATP-binding site of the EGFR kinase domain. Gefitinib inhibits EGFR autophosphorylation and blocks downstream signal transduction, which has a higher affinity for EGFR with activating mutations compared to the wild-type [15]. Bevacizumab, an anti-VEGF monoclonal antibody, has been approved by FDA for inhibiting tumor angiogenesis [16–18]. Bevacizumab neutralizes VEGF-A and reduces the binding of VEGF-A to vascular endothelial cell receptors VEGFR-1 and VEGFR-2 [19]. By blocking VEGF, bevacizumab inhibits angiogenesis and normalizes tumor blood vessels, preventing “leaky” vasculature and theoretically also improving tumor-targeted drug delivery [6]. Although great advances are made, acquired resistance of EGFR TKIs and inadequate progression-free survival of anti-VEGF antibodies remain major challenges in treating NSCLC [20–22]. In EGFR-mutant NSCLC, upregulated EGFR signaling increases the concentration of VEGF through a hypoxia-independent mechanism, which in turn leads to resistance to EGFR TKIs [23]. EGFR has been indicated as a predominant hypoxia-independent driver of hypoxia-inducible factor (HIF) expression [24]. Expression of EGFR mutations is associated with increased HIF-1α levels in NSCLC even with normoxic condition. HIF-1α drives the transcription of genes involved in angiogenesis, especially VEGF. Thus, over expression of EGFR increases the level of VEGF [25]. Furthermore, VEGF involved in the regulation of mesenchymal epithelial transition factor (MET). MET activation has been considered to mediate primary and acquired resistance to EGFR TKIs in EGFR-mutant NSCLC [26]. Inhibition of the EGFR-VEGF dual pathways has shown potential in delaying acquired resistance to EGFR TKIs [27]. Multiple clinical trials, including two large randomized phase III studies, have identified therapeutic benefits of EGFR-VEGF dual inhibition, including prolonging progression-free survival (PFS), delaying the emergence of drug resistance in treatment of naive patients with EGFR-mutant NSCLC, and significantly improving clinical outcomes [28–31]. But there are still unmet needs toward effective delivery and precise release of drugs in combinational therapy. On the one hand, the pharmacokinetic gap between EGFR TKIs and anti-VEGF antibodies complicates the dosage regimen. The good circulation of anti-VEGF antibodies increases risks of hypertension and vascular thromboembolic events [32]. On the other hand, lack of precise drug delivery and effective drug release may decrease drug efficacy. Therefore, it remains challenging to achieve co-delivery of anti-VEGF antibodies and EGFR TKIs with controllable release behaviors for improving combinational molecular targeted therapy of NSCLC.

Nanoparticles can improve the delivery efficacy of payloads and enable co-delivery of different kinds of drugs [33–37]. For example, small molecule drugs were encapsulated in the core of nanoparticles and antibodies could also be loaded in the core or linked on the surface of nanoparticles [38–40]. Nevertheless, most of nanoparticles are still far from achieving satisfying performance for co-delivering macromolecules and small molecules. Tumor cells prefer to internalize nanoparticles loading macromolecules and decrease the drug efficiency. Besides, surface modification of antibodies is complicated and difficult for scale-up production. Moreover, most approaches can barely meet the demand for releasing antibodies in tumor microenvironment while releasing cytotoxic small molecules in tumor cells. Nanoparticles co-delivering anti-VEGF antibodies and EGFR TKIs in controlled release properties are still urgently required to improve molecular targeted therapy against NSCLC.

Herein, we developed bevacizumab-coated gefitinib-loaded nanoparticles (BCGN) with dual-responsive drug release for enhancing molecular targeted therapy of NSCLC. Gefitinib-loaded nanoparticles (GN) are self-assembled by 9-fluorenyl methoxycarbonyl (Fmoc)-KCRGDK-phenylboronic acid (FK-PBA), polyvinyl alcohol (PVA), cationic polyethyleneimine-derived pyropheophorbide-a (PEI-PPa) and gefitinib. After the assembly of GN, bevacizumab is coated onto the cationic surface of GN via an exogenous corona strategy to obtain BCGN (Scheme 1a). Phenylboronic ester bonds are formed between FK-PBA and PVA to stabilize BCGN. The presence of oxygen anion in tumor microenvironment will lead to cleavage of the phenylboronic ester bonds, thus reducing electrostatic interaction to trigger bevacizumab release. Afterward, the GN are internalized by tumor cells and high level of intracellular glutathione (GSH) break the disulfide bond in FK-PBA to disassemble GN (Scheme 1b). The accumulation of BCGN in tumors is monitored by PPa-based dual-model imaging. BCGN inhibits angiogenesis, suppresses phosphorylation of EGFR and delays the growth of human NSCLC tumors in mouse models (Scheme 1c). The approach reported here provides a controlled-release strategy for co-delivering anti-VEGF antibodies and EGFR TKIs, which will enhance molecular targeted therapy of NSCLC.

Scheme 1. Schematic illustration of the preparation of BCGN and its mechanisms for dual-responsive targeted therapy against NSCLC.

a BCGN was prepared with FK-PBA peptides, PVA and PEI-PPa polymers. Gefitinib was loaded via hydrophobic interaction and bevacizumab was covered on the surface of nanoparticles through a protein corona process. The BCGN was responsive to oxidation and reduction for triggering the release of payloads at different circumstances. b Chemical mechanisms of dual-responsive property. The red circles indicated the reactive sites. c BCGN was distributed into tumor tissues after intravenous injection and released upon oxidation condition. Then, GN was taken up into cytoplasm and released at reduction circumstance. The combination of anti-VEGF and EGFR TKIs inhibited the progression of NSCLC.

Materials and methods

Reagents

Polyethyleneimine (M_W = 1800 Da) was purchased from Sinopharm Group Chemical Reagent Co., Ltd. (XW90029862, Beijing, China). Fmoc-K (Fmoc) CRGDK was purchased from Hefei Bank Peptide Biotech Co., Ltd (Hefei, China). 4-Mercaptophenyl boronic acid (PBA) was purchased from J&K Chemical Co., Ltd (116021, Shanghai, China). 2,2ʹ-Dipyridyl disulfide was purchased from J&K Chemical Co., Ltd (228648, Shanghai, China). 1-Hydroxybenzotriazole (HOBT) was purchased from J&K Chemical Co., Ltd (191943, Shanghai, China). 1-(3-Dimethylaminopropyl)-3-ethylcarbodiimide Hydrochloride (EDCI) was purchased from J&K Chemical Co., Ltd (495017, Shanghai, China). Pyropheophorbide-a (PPa) was obtained from Shanghai D&B Chemical Technology Co., Ltd (H124006, Shanghai, China). Polyvinyl alcohol (PVA) polymer (M_W = 67 kDa) was purchased from Sigma-Aldrich (81383, Shanghai, China). Wheat germ agglutinin-FITC conjugate (WGA-FITC) was purchased from Sigma-Aldrich (L4895, Shanghai, China). Bevacizumab was obtained from Geneway Bio-Technology Co., Ltd (FC08DS101902, Shanghai, China). Gefitinib was obtained from Tokyo Chemical Industry Co., Ltd (G0546, Tokyo, Japan). Phalloidin-TRITC was purchased from Shanghai Yeasen Biotech Co., Ltd (40734ES75, Shanghai, China). Bicinchoninic acid (BCA) quantification kit was purchased from Shanghai Yeasen Biotech Co., Ltd (20201ES76, Shanghai, China). Phenylmethanesulfonyl fluoride was purchased from Shanghai Yeasen Biotech Co., Ltd (20104ES03, Shanghai, China). Alexa Fluor® 488 AffiniPure goat anti-rabbit IgG (H + L) was purchased from Shanghai Yeasen Biotech Co., Ltd (34206ES60, Shanghai, China). Triton X-100 was purchased from Shanghai Beyotime Biotech Co., Ltd (ST797, Shanghai, China). 4′,6-Diamidino-2-phenylindole (DAPI) was purchased from Shanghai Beyotime Biotech Co., Ltd (C1006, Shanghai, China). Antifade solution was purchased from Shanghai Beyotime Biotech Co., Ltd (P0126, Shanghai, China). Omni-PAGE^TM Hepes-Tris Gels was obtained from Epizyme Biomedical Technology Co., Ltd (LK202, Shanghai, China). Western blot marker was purchased from Dakewe Biotech Co., Ltd (8011041, Shanghai, China). SuperLimit ECL chemiluminescent substrate was purchased from Dakewe Biotech Co., Ltd (8061011, Shanghai, China). SpectBlue protein staining reagent was purchased from Dakewe Biotech Co., Ltd (8019011, Shanghai, China). SDS-PAGE protein loading buffer (5×) was purchased from Dakewe Biotech Co., Ltd (8015011, Shanghai, China). Phospho-EGFR (Tyr1068) antibody (AF3045), ERK1/2 antibody (AF0155), phospho-ERK1/2 (Thr202/Tyr204) antibody (AF1015), pan-AKT1/2/3 antibody (AF6261), phosphor-pan-AKT1/2/3 (Ser473) antibody (AF0016) were purchased from Affinity Biosciences Co., Ltd (Changzhou, China). Human IgG ELISA kit was purchased from Neobioscience Technology Co., Ltd (EHC124.48, Shenzhen, China).

Cell lines and animals

Human NSCLC cell lines A549 and HCC827 cells were obtained from the Cell Bank of the Chinese Academy of Sciences (Shanghai, China). Both kinds of cell lines were cultured in complete RPMI-1640 cell culture medium containing 10% fetal bovine serum (FBS), 2.5 g/L of glucose, 0.11 g/L of sodium pyruvate, 100 U/mL of penicillin G sodium and 100 μg/mL of streptomycin sulfate. Cell cultures were maintained at 37 °C in 5.0% CO₂ atmosphere. Four-week-old female BALB/c nude mice were obtained from the Shanghai Experimental Animal Center (Shanghai, China). Animal procedures were carried out under the guidelines approved by the Institutional Animal Care and Use Committee (IACUC) of the Shanghai Institute of Material Medica, Chinese Academy of Sciences. All the animal procedures followed the guidelines of Animal Research Advisory Committee (ARAC). The tumor burden was not greater than 10% body weight. The tumor was not exceeded 20 mm in any one dimension, and the tumor volume was less than 2000 mm³. Carbon dioxide (CO₂) inhalation is applied for mice euthanasia.

Synthesis of FK-PBA and PEI-PPa

To synthesize FK-PBA, 154 mg of 4-mercaptophenylboronic acid and 264 mg of 2,2ʹ-dipyridyl disulfide were reacted in methanol for 4 h at room temperature and then precipitated in n-hexane to obtain 4-(dipyridyl disulfide)-PBA. Then, 57.9 mg of the product was reacted with 230 mg of FK in methanol at 70 °C for 4 h. The FK-PBA was purified in diethyl ether and vacuum-dried for use.

For PEI-PPa synthesis, 60.8 mg of HOBT was dissolved in tetrahydrofuran. 96.3 mg of PPa, 86 mg of EDCI were dissolved in anhydrous dichloromethane. The molecules were mixed together and reacted in dark for 2 h. Then the solution was added into anhydrous dichloromethane solution containing 81 mg of PEI (M_W = 1800 Da) and 11.4 mg of TEA. Then, the product was dried with rotatory evaporation to remove organic reagents and dissolved in water sequentially. Last, the solution was lyophilized for further use.

Preparation and characterization of GN and BCGN

To optimize the formulation of nanoparticle, FK-PBA was firstly dissolved in DMF and incubated with PVA (M_W = 67 kDa) at weight ratio of 15:1 for 24 h. The product was precipitated in PEI-PPa aqueous solution at weight ratio of 10:1, 10:3, 10:5, 10:7 and 10:10 under vortex and examined by Zetasizer (Malvern, UK). The weight ratio of FK-PBA and PEI-PPa was fixed at 10:5 after the optimization. To further investigate the loading of gefitinib, FK-PBA was reacted with PVA again as mentioned above, and incubated with gefitinib in DMF. The ratio among FK-PBA, PEI-PPa and gefitibin was 4:2:1. To prepare gefitinib-loaded nanoparticles (termed as GN), the mixture was added dropwise into PEI-PPa under vortex and the weight ratio of PEI-PPa to gefitinib was 8:1, 4:1, 2:1 and 1:1, respectively. In detail, FK-PBA was stirred with PVA (M_W = 67 kDa) in DMF at the final concentration of 25 mg/mL and 1.7 mg/mL for 24 h, respectively. Then the solution was incubated with 6.3 mg/mL of gefitinib for 2 h. The mixture was added into 1.8 mg/mL of PEI-PPa aqueous solution under vortex, where the weight ratio of PEI-PPa to gefitinib was 2:1. Excess gefitinib was removed by centrifugation (2000 rpm, 2 min) and DMF was removed by centrifugal filtration (molecular weight cut-off, 100 kDa, Millipore). The diameter and polydispersity index (PDI) of GN were characterized by Zetasizer. The morphology of GN was obtained by 120 kV TEM (FEI, USA).

To prepare nanoparticles with bevacizumab corona and gefitinib (termed as BCGN), bevacizumab was firstly dispersed in a fixed final concentration of 100 μg/mL and cultured with various amount of GN (8.4, 33.6, 168, 336 μg/mL) for 1 h. Then the mixture was centrifugated under 15,000 × g for 15 min to remove free mAb. mAb absorbed on GN was in the precipitation as the protein corona of BCGN [41]. Then mAb in supernatant was denatured and detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Free mAb in supernatant was confirmed by stripes of SDS-PAGE. Finally, to prepare BCGN, GN (5 mg/mL of FK-PBA) was incubated with 3 mg/mL of bevacizumab for 1 h and free mAb was removed by ultrafiltration. The hydrodynamic size and PDI of BCGN was characterized by dynamic light scattering (DLS). The morphology of BCGN was obtained from 120 kV TEM.

To detect the encapsulation efficiency and drug loading efficiency of BCGN, the standard curve of bevacizumab was firstly constructed by ELISA. Uncoated bevacizumab was separated via high-speed centrifugation (15,000 × g, 15 min) and measured by ELISA. The standard curve of gefitinib was firstly constructed by HPLC (Thermo Fisher, USA). Based on the standard curve, encapsulation efficiency and drug loading efficiency were detected after the ultrasonic decomposition of BCGN by methanol.

H₂O₂ sensitivity of GN and BCGN

To detect the H₂O₂ sensitivity of GN, GN was prepared and incubated in 10 mM H₂O₂ for 1 h. Then particle size and zeta potential were measured by Zetasizer. To further investigate the release of bevacizumab from BCGN after H₂O₂ treatment, BCGN were prepared and incubated in 10 mM H₂O₂ for 0, 1, 6 h, respectively. The BCGN was under high-speed centrifugation (15,000× g, 15 min) for free mAb separation. The released mAb was measured by ELISA. Meanwhile, bevacizumab was treated with H₂O₂ and separated by the same way to evaluate the influence of H₂O₂ on the viability of mAb.

Reduction sensitivity of BCGN

BCGN was incubated in 10 mM of GSH for 1 h and TEM was introduced to clarify the morphological change. Gefitinib, BCGN were added into dialysis tubes (MWCO = 3 kDa) loaded 20 mL of PBS in different medium: PBS or 10 mM GSH. The tubes were shaken at 100 rpm at 37 °C. At predetermined time intervals, the solutions were centrifugated at 2000 rpm for 2 min and the gefitinib sedimentation was collected, dissolved in methanol and detected by HPLC.

In vitro cellular uptake of BCGN

To examine cellular uptake of BCGN, HCC827 cells were seeded on cell slides for 24 h and then treated with BCGN (0.4 μg/mL of PEI-PPa) for different time durations. The quantitative analysis of the cellular PEI-PPa fluorescence intensity was examined by flow cytometry. Also, confocal laser scanning microscopy (CLSM) was performed. The cells were stained with WGA-FITC and DAPI and then the fluorescence of PPa (Ex/Em: 640 nm/680 nm) was examined.

In vitro cytotoxicity

The in vitro cytotoxicity of BCGN against HCC827 cells was evaluated using the Cell Counting Kit-8 (CCK-8). HCC827 cells were seeded in 96-well plates and cultured for 24 h. Then, cells were incubated with 0.2 μg/mL of gefitinib, IgG-NP, GN, and BCGN for 48 h, followed by adding 10% of CCK-8 solution. The absorbance was measured at 450 nm by a multimode plate reader. IgG-NP were IgG coated nanoparticles, prepared by using FK-PBA, PVA, PEI-PPa, which were of the same weight ratio as BCGN and covered by total IgG extracted from human serum.

In vitro inhibition of p-EGFR and relative pathways

HCC827 cells were seeded on cell slides for 24 h and treated with BCGN or gefitinib (2 μg/mL) for 6 h. Then cells were fixed with 4% paraformaldehyde, permeabilized with 0.1% Triton X-100 and blocked with 10% serum for 45 min at room temperature. Next, cells were labeled with Phospho-EGFR (Tyr1068) Rabbit Polyclonal Antibody at room temperature for 1.5 h and labeled with Alexa Fluor® 488 AffiniPure goat anti-rabbit IgG (H + L) for 1 h at room temperature. F-actin was stained with phalloidin-tetramethylrhodamine (TRITC) (Ex/Em: 540 nm/560 nm) and nuclei were stained with DAPI.

For Western blot assay, HCC827 cells were seeded on cell slides for 24 h and treated with BCGN, gefitinib (2 μg/mL) or PBS for 6 h. Then cells were extracted by adding RIPA Lysis solution and phenylmethanesulfonyl fluoride (PMSF). Then proteins were quantified by BCA quantification kit and denatured. Samples containing same concentration of total proteins were detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). Electrophoresed protein samples were transferred to polyvinylidene difluoride membranes (Bio-Rad). After washing three times, the membranes were blocked by non-fat dry milk for 1 h at room temperature and incubated overnight at 4 °C with primary antibodies to mark p-EGFR (Tyr1068), t-EGFR, ERK1/2, p-ERK1/2 (Thr202/Tyr204), AKT1/2/3, p-AKT1/2/3 (Ser473) (1:1,000 dilution). After washing three times, the membranes were incubated for 1 h at room temperature with HRP-conjugated species-specific secondary antibody and then pictures were captured by electrochemiluminescence (ECL) system (Bio-Rad).

Tumor models

A549 human NSCLC model and HCC827 human NSCLC model with EGFR mutant were used. For A549 tumor model, 1 × 10⁷ of A549 tumor cells were subcutaneously injected in lower rear flank of 4-week-old female BALB/c nude mice. For HCC827 model, 1 × 10⁷ of HCC827 tumor cells were subcutaneously injected in lower rear flank of 4-week-old female BALB/c nude mice. When tumors reached about 500 mm³, they were surgically resected and divided into small pieces (3 mm) and implanted into the subcutaneous lower flank tissue of BALB/c nude mice [42].

In vivo biodistribution of BCGN

HCC827 models were established and PEI-PPa, GN, BCGN were i.v. injected at same dose of PEI-PPa (5 mg/kg), when tumors reached about 300 mm³. Fluorescence images were carried out by IVIS (Perkinelmer, USA). After the injection (4 h and 24 h), tumors and major organs were collected for determining ex vivo fluorescence (n = 3). Also, the tumors were collected and sectioned into 8 μm slices. The nuclei were stained with DAPI (blue) and PPa was represented in red. To detect the photoacoustic imaging ability of BCGN, PEI-PPa, GN, BCGN were administered in HCC827 models as mentioned above and tumors were detected at an excitation wavelength of 715 nm at 0, 2, 4, 8 and 24 h. Ultrasound images of tumor regions were represented in gray and PEI-PPa was in red.

Tumor models and antitumor treatments

A549 cells (1.0 × 10⁷) were s.c. injected in the lower backside of BALB/c nude mice. When tumor volume reached about 100 mm³, mice were randomly divided into five groups: PBS, GN, BCN, Gefitinib+Bevacizumab and BCGN. Different formulations were i.v. injected at equal gefitinib dose of 4.2 mg/kg or bevacizumab dose of 5.0 mg/kg on Day 0, 4, 8, 12. On Day 23, mice were euthanized and the tumors were collected and weighed. For HCC827 model, HCC827 tumor cells were s.c. injected into the lower rear flank of BALB/c nude mice. Tumor was collected and cut into 1 mm³ blocks. Then small tumor sections were s.c. inoculated in BALB/c nude mice. When tumor volume reached about 300 mm³, mice were randomly divided into two groups: PBS and BCGN (4.2 mg/kg gefitinib and 5.0 mg/kg bevacizumab). The growth of tumors and body weight of mice were recorded every 5 days. The tumor volume was calculated according to the following equation:

$$\mathrm{Volume}=\mathrm{Longest}\mathrm{dimension}\times \mathrm{Shortest}\mathrm{dimension}\times \mathrm{Shortest}\mathrm{dimension}/2$$

Immunohistochemical (IHC) analysis and blood routine detection

HCC827 tumor-bearing model was established. PBS, GN, BCN, Gefitinib+Bevacizumab, and BCGN were i.v. injected at equal gefitinib at dose of 4.2 mg/kg or bevacizumab at dose of 5.0 mg/kg twice at a time interval of 4 d. Twenty-four hours after the second administration, mice were euthanized. HCC827 tumors were collected and cut into slides for IHC staining of p-EGFR, VEGF-A, and CD31. Healthy BALB/c nude mice were i.v. injected with PBS, Gefitinib+Bevacizumab and BCGN (4.2 mg/kg of gefitinib, 5.0 mg/kg of bevacizumab). Two days after administration, blood serum was collected and detected.

Statistical analysis

All the data were presented as mean ± SD. One-way analysis of variance (ANOVA) and two-sided unpaired Student’s t-test were used to determine the significance of the difference. Statistical significance was set at *P < 0.05, **P < 0.01, ***P < 0.001.

Results

Preparation and characterization of BCGN

To prepare BCGN, we firstly synthesized FK-PBA and PEI-PPa according to our previous reports [43]. FK-PBA was synthesized by conjugating 4-mercaptophenylboronic acid with Fmoc-KCRGDK peptides via disulfide bond (Supplementary Fig. S1). The amino groups of PEI and carboxyl groups of PPa reacted via an amide condensation reaction to obtain PEI-PPa (Supplementary Fig. S2). We prepared nanoparticles and tested the weight ratios among FK-PBA, PEI-PPa and gefitinib on the formation of nanoparticles. Accordingly, FK-PBA was firstly incubated with PVA (M_W = 67 kDa) at weight ratio of 15:1 for 24 h. Then, the FK-PBA/PVA solution was precipitated in PEI-PPa solutions at FK-PBA to PEI-PPa weight ratios of 2:1. It was found that the hydrodynamic size and PDI of the nanoparticles was 211.4 nm and 0.3, respectively (Supplementary Fig. S3a). Next, we fixed the FK-PBA to PEI-PPa weight ratio at 2:1 and changed the amount of gefitinib. To make the nanoparticles uniform and high drug-loading, we finally verified the proper ratio among FK-PBA, PEI-PPa and gefitinib was 4:2:1. The average hydrodynamic size was 213.1 nm and PDI was 0.29 (Supplementary Fig. S3b). Transmission electron microscopy (TEM) and dynamic light scattering (DLS) showed the morphology of GN was spherical and uniform, with diameter of 213.7 ± 4.0 nm and zeta potential of 32.2 ± 1.9 mV (Fig. 1a). Next, the inner interactions of GN were evaluated in various medium. Hydrodynamic size of GN was stable in 80 mM of NaCl or urea, while declined with increasing concentrations of Tween-20 (Fig. 1b).

Fig. 1. Preparation and characterization of GN and BCGN.

a Hydrodynamic size of GN examined by DLS. Inset: TEM images of GN. Scale bar, 1 μm. b Hydrodynamic size of GN cultured in different medium of NaCl, urea or Tween 20 (n = 3). c Hydrodynamic size of BCGN examined by DLS. Inset: TEM images of BCGN. Scale bar, 1 μm. d SDS-PAGE electrophoresis of free mAb in the supernatant separated from BCGN via high-speed centrifugation (15,000 × g, 15 min). Fixed amount of mAb was cultured with different concentrations of GN for BCGN preparation. mAb represents bevacizumab. e Zeta potential of GN treated with 10 mM H₂O₂ for 1 h (n = 3). f Relative loading efficiency of mAb on BCGN after treated BCGN with 10 mM H₂O₂ for 0, 1 or 6 h, respectively (n = 3). g TEM images of BCGN after cultured in 10 mM GSH for 1 h. h In vitro gefitinib release profiles of BCGN in response to 10 mM GSH (n = 3). Data are shown as mean ± SD.

Cationic nanoparticles could readily interact with serum proteins which are mostly negatively charged in physiological environments and then form protein corona [44]. We used this strategy to coat negatively charged bevacizumab on the surface of positively charged GN for establishing BCGN. BCGN was spherical and uniform. Due to the surface coating of proteins, BCGN was larger in diameter (239.4 ± 3.6 nm) and lower in zeta potential (4.9 ± 0.4 mV) than GN (Fig. 1c). The gefitinib loading efficacy and encapsulation efficacy in BCGN was 10.1% ± 0.6% (w/w) and 67.7% ± 4.4%, respectively. To test the coating of bevacizumab on GN, high speed centrifugation was used to separate free monoclonal antibodies (mAb) from BCGN [41]. In the experiment, mAb was fixed at 100 μg/mL and cultured with different amount of GN for 1 h. Then the mixture was centrifuged under 15,000 × g for 15 min. mAb was precipitated following the formation of BCGN and unloaded mAb would be dispersed in the supernatant. Then mAb in supernatant was detected by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). With increased addition of GN, less mAb was found in supernatant as confirmed by lighter stripes of SDS-PAGE (Fig. 1d). The results proved that mAb was effectively loaded on GN to form BCGN. The loading efficacy and encapsulation efficiency of bevacizumab was 12.2% ± 2.4% (w/w) and 34.8% ± 7.9%, respectively. After entering blood stream, endogenous proteins would compete with the exogenous corona of BCGN and might result in undesirable release of bevacizumab. To investigate the stability of BCGN in the presence of serum proteins, BCGN was cultured in 10 mg/mL of human serum albumin (HSA) or PBS at same concentration of BCGN for 1 h. Next, the suspensions were centrifuged and SDS-PAGE gel shift analysis was applied. No obvious visible increase of free mAb stripes was found in HSA culturing groups compared to corresponding PBS groups, indicating the mAb was seldom exchanged from BCGN by HSA (Supplementary Fig. S4). The stability of BCGN in 10% FBS and PBS was also determined (Supplementary Fig. S5).

Next, the stimuli-responsive characteristics of BCGN were further tested. Bevacizumab was coated on GN by electrostatic interaction. Oxidation dramatically decreased the zeta potential of GN from 32.2 ± 1.9 mV to 7.4 ± 1.4 mV (Fig. 1e) while without significant change in hydrodynamic size (Supplementary Fig. S6). After treated by H₂O₂, the size of BCGN was decreased from ~237.9 nm to ~201.1 nm (Supplementary Fig. S7a). The borate ester between PBA and PVA was sensitive to oxidation, but oxidation did not trigger disassembly of BCGN. The decrease of particle size was probably due to the partially release of bevacizumab from BCGN. The ζ-potential of BCGN was merely changed after the treatment, from 6.2 mV to 5.4 mV (Supplementary Fig. S7b). The negligible change of surface charge could attribute to the maintenance of nanoscale diameter and partial release of bevacizumab. We investigated whether the change of surface property could influence the release of bevacizumab. After incubating BCGN with H₂O₂ for different time periods, released mAb was collected via centrifugation and detected by enzyme linked immunosorbent assay (ELISA). The results showed H₂O₂ triggered the shedding of bevacizumab from BCGN and ~40.8% of bevacizumab was released after 6 h of incubation (Fig. 1f). Meanwhile, 6 h incubation with H₂O₂ was of little influence on the viability of bevacizumab (Supplementary Fig. S8). Except for the oxidation-responsive property, the disulfide bond in FK-PBA was sensitive to high level of GSH in tumor cells [45–47]. TEM images of BCGN treated by 10 mM GSH showed a GSH-induced disassembly of nanoparticles (Fig. 1g). The cumulative release of gefitinib from BCGN was only 31.5% within 48 h, while increased to 75.3% with GSH treatment (Fig. 1h).

In vitro inhibition of p-EGFR and relative pathways by BCGN

The cellular uptake of BCGN was measured by flow cytometry and CLSM in HCC827 cells, an EGFR-mutant NSCLC model [48, 49]. BCGN was detected by the fluorescence of PPa and presented a time-dependent cellular internalization behavior (Fig. 2a). Fluorescence conjugated wheat germ agglutinin (WGA) is commonly used to combine glycoproteins for labeling the plasma membrane in cells. BCGN effectively localized in cytoplasm after 4 h of culturing (Fig. 2b). To investigate the cellular toxicity of BCGN, different formulations were cultured with HCC827 cells for 48 h. The results showed that BCGN had a strong killing efficacy even at a low gefitinib concentration of 0.2 μg/mL (Fig. 2c). IgG-NP was negative control, which was prepared with total human IgG and did not load gefitinib. It showed low cytotoxicity to HCC827 cells compared to other groups. Thus, BCGN has great potential in anti-NSCLC therapy. We further verified the mechanism and pathway of EGFR post gefitinib treatment. EGFR is a transmembrane protein with endogenous tyrosine kinase activity that transduces important growth factor signals from the extracellular environment into the cell [50]. EGFR consists of three distinct structural regions, namely, the extracellular ligand-binding region, the transmembrane region, and the intracellular tyrosine kinase region [51]. Activation of tyrosine kinases in the intracellular domain leads to autophosphorylation of the intracellular domain, which in turn triggers a series of intracellular signal transduction. The major signaling cascades induced by activation of EGFR include mitogen-activated protein kinase (MAPK) pathways and the phosphatidylinositol 3-kinase (PI3K) pathway. MAPK pathways can be activated in a Ras/Raf-dependent or Ras/Raf-independent manner, converging on extracellular signal-regulated kinase (ERK1/2) and ERK5. PI3K activation leads to phosphorylation and activation of the AKT kinase [52]. The activation of RAS/MAPK/ERK and PIK3/AKT pathways promotes tumor cell proliferation, metastasis, anti-apoptosis and other processes [53]. CLSM results showed that PBS-treated HCC827 cells were of strong green fluorescence and maintained high level of phosphorylated-EGFR (p-EGFR). After 6 h of BCGN treatment, green fluorescence of p-EGFR in HCC827 cells remarkably declined with the increasing concentrations of gefitinib (Fig. 2d). It was also confirmed by Western blot assays with a pronounced down-regulation in p-EGFR of BCGN group (Fig. 2e). These results confirmed that BCGN could effectively suppress the phosphorylation of EGFR. We also examined the phosphorylation status of AKT and ERK. We found that the phosphorylation levels of AKT and ERK were significantly suppressed after 6 h treatment with BCGN, without obvious change in total AKT and ERK (Fig. 2e). Therefore, BCGN could be readily internalized by cancer cells, facilitate the p-EGFR inhibition and induce death of NSCLC cells.

Fig. 2. BCGN induced in vitro inhibition of p-EGFR.

a Uptake of BCGN when HCC827 cells were treated with BCGN for different time periods and examined by PPa fluorescence with flow cytometry. b CLSM images of HCC827 cells with BCGN treatments. Cell nuclei were labeled with DAPI (blue). membrane was labeled with WGA-FITC (green) and PPa was represented in red. Scale bars, 25 μm. c Cytotoxicity of gefitinib, IgG-NP, GN, BCGN after incubation with HCC827 tumor cells for 48 h (n = 5). IgG-NP were human IgG coated nanoparticles without loading gefitinib. d Immunofluorescence analysis of p-EGFR levels in HCC827 cells after treated by BCGN or gefitinib for 6 h. Cytoskeleton was labeled with Phalloidin-TRITC (red) and p-EGFR was stained with FITC linked antibodies (green). Scale bars, 20 μm. e Western blotting detecting the efficiency of BCGN for inhibiting phosphorylation of EGFR, AKT and ERK in HCC827 cells. Data are shown as mean ± SD.

Biodistribution and penetration of BCGN in NSCLC mouse model

We next investigated in vivo distribution of BCGN. The biodistribution of PEI-PPa, GN and BCGN was determined in HCC827-bearing mice with intravenous injection at an equal dose of PPa. Compared to PEI-PPa, the GN and BCGN were effectively distributed in tumors since 2 h (Fig. 3a). In addition, BCGN showed strongest fluorescence in ex vivo tumors after 4 h of administration, suggesting that more BCGN was accumulated in tumors than PEI-PPa and GN (Fig. 3b, c). After treated by different suspensions for 24 h, the accumulation of both GN and BCGN in tumor declined compared with 4 h (Supplementary Figs. S9, S10). The declined intensity might because the positive surface charge induced rapid clearance of nanoparticles from blood and major organs. It has been reported that cationic nanoparticles easily interact with negatively charged serum components and cause opsonization. Those nanoparticles are rapidly cleared from circulation and easily entrapped by lung [54, 55]. Plasma proteins such as serum albumin, complement proteins, and immunoglobulins could be adsorbed on the surface of positively charged nanoparticles after intravenous administration and form protein corona [56]. Protein corona promotes recognition and clearance by the reticuloendothelial system (RES), which is dominated by monocytes-macrophages [57]. At the same time, the adsorption of complement proteins activates the complement system, which leads to the binding of phagocytosis or NK cells [58, 59]. Both the process results in the clearance of cationic nanoparticles [60]. Here, owing to the surface absorption of bevacizumab, BCGN has relatively low surface zeta potential and decreased lung accumulation compared with GN (Supplementary Fig. S11). Therefore, BCGN showed a reliable tumor distribution behavior for drug delivery. Furthermore, PPa, derivative of chlorophyll-a could act as an imaging molecule for monitoring the distribution and penetration of BCGN in tumor [61]. We intravenously injected PEI-PPa, GN and BCGN in HCC827-bearing mice and imaged the signals using the photoacoustic (PA) imaging system at 0, 2, 4, 8 and 24 h, respectively. The PA signals from BCGN showed most significant enhancement after administration and performed a greater tumor penetration than PEI-PPa and GN (Fig. 3d). The improvement of tumor accumulation was also detected in tumor sections by CLSM. BCGN showed strongest PPa fluorescent signals than PEI-PPa and GN (Fig. 3e). Therefore, BCGN was promising for tracking distribution, realizing both fluorescence and PA dual-model imaging in one platform.

Fig. 3. Biodistribution of BCGN in NSCLC mouse model.

a In vivo biodistribution of PEI-PPa, GN and BCGN in HCC827 tumor-bearing mice (n = 3). b The ex vivo fluorescence of PEI-PPa, GN and BCGN in major organs of HCC827 tumor-bearing mice and (c) tumor fluorescence was quantified (n = 3). d photoacoustic imaging of tumors when the mice were intravenously injected with PEI-PPa, GN or BCGN and detected at an excitation wavelength of 715 nm. e Fluorescence imaging of ex vivo tumor sections after 4 h of administration was performed on HCC827 tumor-bearing mice. PEI-PPa, GN, BCGN were injected at same PEI-PPa dose (5 mg/kg) and represented in red. Cell nuclei were stained with DAPI (blue). Scale bar, 30 μm. Data are shown as mean ± SD.

In vivo antitumor studies of BCGN in A549 and HCC827 NSCLC mouse models

To evaluate the antitumor effects of BCGN, we firstly performed antitumor studies in A549-bearing mice, A549, derived from a 58-year-old male patient, known to be EGFR wild type is a commonly used NSCLC model [62, 63]. Mice were administered different formulations for 4 times at a time interval of 4 days. GN and bevacizumab-coated nanoparticles without gefitinib (BCN) delayed the growth of A549 in contrast to PBS group. Compared with monotherapy of GN or BCN, the tumor volume was significantly inhibited in gefitinib + bevacizumab group (Fig. 4a). Furthermore, compared with gefitinib + bevacizumab group, BCGN nanoparticles significantly enhanced the efficacy of growth inhibition. BCGN showed strongest tumor inhibition and the mean volume was depressed by 63.9% compared with PBS group on day 23 (Fig. 4a). No obvious body weight loss was found in BCGN group (Fig. 4b). The hematoxylin and eosin (H&E) staining results of the heart, liver, spleen, lung, and kidney harvested from healthy mice did not show obvious histopathological changes with BCGN treatment (Supplementary Fig. S12). However, GN group revealed thickening of alveolar walls, infiltration of inflammatory cells in lungs and visible liver injury (Supplementary Fig. S12). Additionally, mice were euthanized and the tumors were collected and weighed on day 23. The average tumor weight of PBS, GN, BCN, gefitinib + bevacizumab and BCGN groups were 1671.2, 1221.4, 1149.4, 832.0 and 550.1 mg, respectively. Average tumor weight in BCGN group was only 32.9% of PBS group and 67.1% of gefitinib + bevacizumab group, respectively (Fig. 4c, d). Consistent with tumor growth trend, BCGN showed strongest tumor suppression among all groups. In order to further evaluate the efficacy of BCGN, HCC827-bearing mice model was established which has a deletion in the EGFR tyrosine kinase domain (ΔE746-A750, exon 19) [64]. EGFR exon 19 deletion (19del) was very common and accounts for 44% in EGFR mutations [65]. BCGN showed a significant antitumor effect as the average tumor volume only increased by 7.8% on day 45, from 336.3 mm³ to 362.4 mm³. By comparison, PBS group showed rapid tumor progression, with the tumor volume increased by 118.0% from 368.7 mm³ to 803.6 mm³ (Fig. 4e). As an indicator of the systemic toxicity, the body weights were measured. There is no significant change in body weight of HCC827-bearing mice between BCGN and PBS (Fig. 4f). In another study, the immunohistochemical (IHC) staining on HCC827 tumors after different treatments was performed. It showed that GN, gefitinib + bevacizumab and BCGN effectively suppressed the phosphorylation of EGFR as lower p-EGFR was stained (Fig. 4g). Bevacizumab could bind to VEGF-A and inhibit angiogenesis of NSCLC. BCN, gefitinib + bevacizumab and BCGN groups had decreased level of VEGF-A in tumor microenvironment (Fig. 4g). Since VEGF signaling is highly associated with angiogenesis, the blood vessels in tumor sections were stained with anti-CD31 antibody. As a result, the number of micro-vessels (CD31 stained) in BCN, gefitinib + bevacizumab and BCGN groups greatly declined in HCC827 xenografts (Fig. 4g). Besides, the blood routine results from healthy mice treated with gefitinib + bevacizumab or BCGN showed no significant difference with the PBS group (Fig. 4h). These results showed that BCGN could effectively suppress the growth of human NSCLC tumors in mouse models.

Fig. 4. In vivo antitumor activity of BCGN in A549 and HCC827 NSCLC models.

a–d BALB/c nude mice were inoculated with A549 cells (1.0 × 10⁷) in the lower backside. The nude mice were intravenously treated with PBS, GN, BCN, Gefitinib+Bevacizumab or BCGN at an equal dose of gefitinib (4.2 mg/kg) and bevacizumab (5.0 mg/kg) when the tumor volume reached 100 mm³. Mice were euthanized on day 23. Then the tumors were obtained and weighed (n = 6). BCN represented bevacizumab coated nanoparticles without loading gefitinib. a Average tumor volume of A549 tumors. Arrows represented the time points of injection (Day 0, 4, 8 and 12). b Body weight of mice. c Tumor weights in control and treated groups. d Photographs of tumors after 23 days of treatment. e, f HCC827-bearing mice were treated with PBS or BCGN at equal dose of gefitinib (4.2 mg/kg) and bevacizumab (5.0 mg/kg) when the tumor volume reached about 300 mm³ (n = 5). e Average growth kinetics of HCC827 tumors. Arrows represented time points (Day 0, 4, 8 and 12) for administration. f Body weight of mice. g IHC images of p-EGFR, VEGF-A and CD31 in HCC827 tumor sections after two-round of treatments. Scale bars, 50 μm. h Blood routine analysis of healthy mice was performed after treatment with PBS, gefitinib + bevacizumab and BCGN (n = 4). Data are shown as mean ± SD. Statistical significance was calculated via two-tailed Student’s t test (a, c, e) and one-way ANOVA test (h). *P < 0.05, **P < 0.01, ***P < 0.001, ****P < 0.0001.

Discussion

Combined molecular targeted therapies with free EGFR TKIs and anti-VEGF antibodies are limited by poor administration. Precise delivery of these drugs to specific spatial state in tumor tissues is urgently needed. In BCGN, both gefitinib and bevacizumab were highly loaded, which facilitated the dosage in vitro and in vivo. Gefitinib as a quinazoline derivative could interact with aromatic rings via π–π stacking and was assembled into nanoparticles with FK-PBA and PEI-PPa [66–68]. The interactions among gefitinib, PVA-grafted FK-PBA and PEI-PPa were evaluated. Surfactant resulted in the dissociation of GN, confirming the assembly of GN was majorly dependent on hydrophobic forces. Although bevacizumab was highly loaded, the surface charge of BCGN could not be adjusted to negative. This might because the steric effects of covered antibodies prevented loading more antibodies. The dual-responsive properties were also investigated. Based on the chemical structure of GN, boronic ester bonds formed between FK-PBA and PVA were sensitive to oxidation condition [69]. Upon oxidation, the surface charge of GN was reduced, thus weakening the electrostatic interaction with bevacizumab to trigger release. This facilitated the release of bevacizumab in oxidized tumor microenvironment. Besides, the GN was totally disassembled in reduction condition, inducing rapid release of gefitinib from the nanoparticles. We successfully prepared BCGN with dual-responsive property upon oxidation and reduction, which facilitated the needs toward controlled release of bevacizumab and gefitinib in NSCLC tumor.

EGFR tightly regulates tumor cell growth and gefitinib could inhibit the phosphorylation of EGFR [70, 71]. In vitro studies on HCC827 cells indicated that phosphorylation of EGFR was effectively reduced post BCGN treatment. The expression of p-EGFR decreased following increased amount of gefitinib from 0.2 μg/mL to 5 μg/mL. Moreover, inhibition of EGFR phosphorylation constrains the AKT/ERK pathway, which in turn inhibits cell proliferation and cell survival [72]. Changes in these molecular pathways would provide important clues for the mechanisms of BCGN. In biodistribution studies, BCGN showed superior accumulation in tumor sites in contrast to control groups. On the one hand, BCGN enabled passive targeting to tumors by leveraging enhanced permeability and retention (EPR) effect. On the other hand, the low surface charge of BCGN suppressed opsonization and reduced entrapment by lung. Besides, PA imaging showed that BCGN enabled effective tumor penetration. The fluorescence/PA dual-model imaging would facilitate the monitor of in vivo fate of BCGN.

BCGN showed effective antitumor effects in both A549 tumor model and HCC827 tumor model. The improved antitumor effects could be contributed to the combination of two drugs and nanoparticle-based delivery. Theoretically, gefitinib has stronger affinity with EGFR mutants than wild type (non-mutated EGFR) cells like A549 [73]. The EGFR mutant tumors, HCC827 tumors, did not show progression in a duration over 40 days, suggesting the good antitumor efficiency of BCGN on EGFR mutant tumors. VEGFs play a pivotal role in controlling the formation of blood vessels. The subtype VEGF-A, especially, acts as a therapeutic target for angiogenesis inhibition and tumor vasculature normalization [6]. Given the effects of bevacizumab to block VEGFs, the angiogenesis of tumors was obviously suppressed. Compared with GN, BCGN did not induce inflammatory cells in lungs as well as visible liver injury, suggesting that the cover of bevacizumab on GN also reduced the lung and liver toxicity. This might be associated with the lower surface charge of BCGN.

In summary, we have developed a dual-responsive nanoparticle loading bevacizumab and gefitinib for combined molecular targeted therapy of NSCLC. BCGN provides a promising controlled release option for enhancing therapeutic efficiency of co-delivered drugs. The easy fabrication and stable drug loading features of BCGN will facilitate its translation to scale-up manufacture. BCGN enables efficient delivery and controlled release of bevacizumab and gefitinib in tumor. Effectively, BCGN suppresses tumor angiogenesis and inhibits the proliferation of tumor cells. Meanwhile, PPa molecules facilitate the monitor of in vivo fate of BCGN. Overall, this work provides an alternative strategy to co-deliver anti-VEGF antibodies and EGFR inhibitors for targeted therapy against NSCLC.

Author contributions

DGW and ZTZ designed the project. ZTZ did the experiment and data analysis. JW and LF helped with material synthesis. XDQ and YC helped with Western blot and antitumor studies in vivo. MLH, HQC, and GRW provided technical helps for the experiment. YYJ, DGW, and YPL supervised, wrote and reviewed the final manuscript. All of the authors have read and approved the final manuscript.

Competing interests

The authors declare no competing interests.

Supplementary information

The online version contains supplementary material available at 10.1038/s41401-022-00930-6.