Engineering cannabidiol synergistic carbon monoxide nanocomplexes to enhance cancer therapy via excessive autophagy

2.2. Synthesis of CO fluorescence probe (FL-CO-1)

FL-CO-1 was synthesized according to the previous report³⁰, and its structure was shown in Supporting Information Fig. S6. ¹H NMR (600 MHz, DMSO-d₆) δ = 8.08 (dt, J = 7.7, 1.0, 1H), 7.84 (td, J = 7.5, 1.2, 1H), 7.78 (td, J = 7.5, 1.0, 1H), 7.47–7.36 (m, 3H), 7.08 (dd, J = 8.7, 2.4, 2H), 6.93 (d, J = 8.7, 2H), 6.01 (ddt, J = 17.3, 10.5, 5.6, 2H), 5.43 (dq, J = 17.3, 1.6, 2H), 5.32 (dq, J = 10.5, 1.4, 2H), 4.76 (dt, J = 5.6, 1.4, 4H).

2.3. RBC-4T1 hybrid membrane preparation

After low-speed centrifugation (800 rpm, Eppendorf, 5418R, Germany) and PBS (pH 7.4) washing, the obtained RBCs were lysed on ice with 0.25 × PBS (pH 7.4). Then, RBC membrane was collected by high-speed centrifugation (12,000 rpm, Eppendorf). Finally, the sonicated RBC membrane was filtered sequentially with 0.45 and 0.2 μm filters. 4T1 cell membranes were prepared using membrane protein extraction kit (Beyotime, P0033). After determining the protein concentration, the two kinds of membrane were mixed with a ratio of 1:1 and sonicated for 15 min on the ice (50 W, 0.5 s/1 s a cycle), followed water bath stirring at 37 °C for 2 h and extrusion through 0.45 and 0.22 μm filters to obtain RBC-4T1 hybrid membrane.

2.4. Synthesis of HMPOC@M

K₃[Fe(CN)₆]·3H₂O (16 mg) and PVP (2 g) dissolved in the 40 mL HCl (0.01 mol/L) were stirred for 20 h at 80 °C. After centrifugation (12,000 rpm, Eppendorf) and washing, the acquired PB NPs (20 mg) were mixed with PVP (100 mg) in the HCl solution and placed in a reaction vessel at 80 °C for 6 h. The HMPB NPs was collected after water washing. MnCO (5 mg) and HMPB (5 mg) were dissolved in the methanol (10 mL). Then, the mixture was stirred in the dark for 6 h. Finally, the HMPB@CO (HMPO) NPs were collected by centrifugation (12,000 rpm, Eppendorf) and washing with methanol. The methanol mixture containing HMPO NPs (10 mg) and CBD (5 mg) was stirred for 30 min at 50 °C. Then, 80 μg GS-01 was added to the above solution. After stirring for 2 h, distilled water was slowly added. The product was centrifuged (12,000 rpm, Eppendorf) and washed to obtain HMPB@CO@CBD (HMPOC) NPs. For membrane coating, the hybrid membrane and HMPOC were mixed at a weight ratio of 1:2, respectively. After ultrasound on the ice and subsequent extrusion, HMPB@CO@CBD@M (HMPOC@M) NPs were collected by centrifugation.

2.5. The effects of biomembrane on HMPOC@M behaviors

Membrane protein characterization: Briefly, total protein concentrations of m4T1, mRBC, 4T1&RBC M or HMPOC@M were detected by the BCA kits. After that, all mixture of samples and loading buffer were heated at 95 °C for 10 min. The levels of CD44 and CD47 in samples were detected using western blot.

Hemolysis assay: RBCs suspension (4% in PBS, v/v) was incubated with water, PBS, CBD, HMPO, HMPOC and HMPOC@M for 6 h. After centrifugation at 3000 rpm (Eppendorf), the absorbance of the supernatant at 562 nm was used to calculate hemolysis rate. The morphology of RBCs was investigated under microscope (OLYMPUS, IX-73, Japan).

Platelet aggregation assay: Briefly, the platelet-rich plasma from whole blood was first mixed with thrombin (9:1, v/v), PBS, CBD, HMPO, HMPOC or HMPOC@M (200 μg/mL). Then, the mixtures were incubated at 37 °C for 2 h. After that, the absorbance values of the supernatants at 650 nm were measured for evaluating the turbidity.

Immunogenicity of HMPOC@M in vivo: 15 female Balb/c mice were randomly separated into 3 groups and administered with saline, HMPOC, and HMPOC@M, respectively, via intravenous injection (0.5 mg/kg PB). LPS (0.5 mg/kg) treated mice via intraperitoneal injection were used as the positive control. Then, blood samples were collected from the retro-orbital plexus of mice at 48 h, and centrifuged (3000 rpm, Eppendorf) to obtain plasma for TNF-α and IL-12 assay using ELISA kits.

Cell uptake assay: 4T1 cells in a 24-well plate were treated with HMPOC@M^Rhb (containing 30 μg/mL PB and 10 μmol/L CBD) for 2, 4, 6, and 8 h, respectively. The cell nuclei were stained with DAPI. After pre-treatment of methyl-β-cyclodextrin (caveolae-mediated endocytosis), colchicine (micropinocytosis) and chloroquine (lathrin-dependent endocytosis) for 1 h, 4T1 cells were then incubated with HMPOC@M^Rhb (containing 30 μg/mL PB and 10 μmol/L CBD) for 6 h. SMC cells, BGC-823 cells or 4T1 cells in a 12-well plate were treated with HMPOC@M^Rhb (containing 30 μg/mL PB and 10 μmol/L CBD) for 6 h, respectively. The nuclei of cells were stained with DAPI. Then, RAW264.7 cells in 12-well plates were incubated with HMPOC^Rhb and HMPOC@M^Rhb (containing 30 μg/mL PB and 10 μmol/L CBD) for 6 h. The nuclei were stained with Hoechst 33342. The intensity difference of red fluorescence emitted by Rhodamine B was investigated under CLSM (OLYMPUS, FV1200, Japan).

2.6. Drug release and CO/Mn production

Photothermal effect assay: A quartz curvet containing 0.5 mL aqueous dispersion of PBS (pH 7.4) or HMPOC@M (30 μg/mL) NPs were irradiated with 808 nm laser (0.75 W/cm², 5 min). Temperature changes and infrared images were obtained using infrared induction camera. To gauge the photothermal effect of HMPOC@M, 500 μL of HMPOC@M solutions with different concentration were irradiated with 808 nm laser (0.75 W/cm², 5 min). Meanwhile, the laser with different powers were applied to irradiate HMPOC@M solution (30 μg/mL). In addition, the UV‒Vis spectra (Shimadzu, UV-1800, Japan) of samples before and after the test cycles were measured to evaluate the photothermal stability of HMPOC@M.

CBD release behavior assay: Dialysis bag containing HMPOC@M solution were immersed into PBS solution (20 mL, pH 7.4 or pH 6.8), followed with 42 °C water bath (simulate 0.75 W/cm² irradiation). The release process of CBD was performed on a rotator with a speed of 1000 rpm. At each set time point, the CBD concentration in the PBS solution was monitored by UV‒Vis spectra (Shimadzu) measurement. Meanwhile, the release process with 37 °C water bath was performed as the control.

CO release assay: The standard curve of CO was constructed according to the fluorescent intensity at 513.5 nm (HITACHI, F2500, Japan) of FL-CO-1 at the presence of different MnCO concentrations. To evaluate the effect of 100 μmol/L H₂O₂ on CO release, 1 mg/mL HMPOC@M solution was irradiated by 808 nm laser with 0.75 W/cm², and then 10 μmol/L FL-CO-1 and 10 μmol/L PdCl₂ were added to the above solutions. At predetermined time points, the fluorescent intensity at 513.5 nm was measured to calculate the amount of CO.

Intracellular CO detection: Cells in a 12-well culture plate were treated with different NPs. After incubating for 6 h, a mixture of FL-CO-1 and PdCl₂ (2.5 μmol/L: 2.5 μmol/L) was added into the culture medium, and then the corresponding groups were irradiated with 808 nm irradiation. Finally, the cells were washed 3 times with PBS and the fluorescence was imaged with an inverted microscope (OLYMPUS).

Mn²⁺ release assay: The dialysis bag (MW = 3.5 kD) containing 100 μmol/L H₂O₂ and 1 mg/mL HMPOC@M solution (5 mL) was immersed in 50 mL PBS buffer (pH 7.4 or pH 6.8) at 37 or 42 °C (mimic laser stimulation). At desirable time points, 1 mL sample was collected and the same volume of fresh buffer was supplemented. The release profile of Mn²⁺ from HMPOC@M was determined using ICP-MS analysis (Agilent, 8900, USA).

In vitro ·OH generation: Using 5,5-dimethyl-1-pyrroline N-oxide (DMPO) as trapping agent, electron paramagnetic resonance spectroscopy was used to detect free radical signal (JEOL, JES-FA 200, Japan). In brief, 25 mmol/L NaHCO₃/H₂CO₃ buffer with HMPOC@M (200 μg/mL) was shaken at 37 or 42 °C for 12 h before adding H₂O₂ (final concentration of 1 mmol/L) and DMPO (final concentration of 20 mmol/L). After reaction for 20 min, the EPR spectra of samples were collected on the spectrometer.

Intracellular ·OH detection: 4T1 cells in a 12-well plate were incubated with different NPs for 6 h. After that, the cells were treated with or without irradiation (808 nm, 0.75 W/cm², 5 min). Then, APF (10 μmol/L, 5 μL) was added into the culture medium for 30 min. After washing with PBS, images were captured under confocal laser scanning microscope (OLYMPUS).

2.7. Cell lines

4T1 cells, HUVECs, SMC cells and RAW264.7 cells were obtained from the Xiangya Central Laboratory, Central-South University. The cells were cultured in the DMEM medium supplemented with 10% FBS and 1% PS (Invitrogen, Carlsbad, CA, USA).

2.8. In vitro therapeutic effect

SMC cells and HUVEC cells were inoculated into 96-well plates and grown for 24 h. After that, all cells were incubated with HMPB@M (0, 12.5, 25, 50, and 100 μg/mL) for 24 h. Finally, the cytotoxic effect was detected by MTT method.

Before verifying the cytotoxicity of HMPOC@M, the power density of NIR was optimized. In short, 4T1 cells seeded in 96-well plates were incubated with culture medium containing 30 μg/mL HMPB@M. After incubating for 6 h, cells were treated with 808 nm laser. After incubating for another 24 h. MTT method was adopted to evaluate the cytotoxic effect of laser power.

To detect the cytotoxicity of HMPOC@M, 4T1 cells (10,000 per well) seeded in 96-well plates were cultured overnight at 37 °C under 5% CO₂. Then, the cells were treated with different NPs (30 μg/mL PB; 10 μmol/L CBD), respectively. After incubating for 6 h, 4T1 wells were treated with laser irradiation. After incubating for another 24 h, the cytotoxic effect was detected by MTT method.

Cells in a 12-well culture plate were treated with different NPs. After incubating for 6 h, different wells were treated with 0.75 W/cm² irradiation. Another 20 h later, calcein-AM (10 μg/mL) and PI (5 μg/mL) was added to stain cells for 30 min, the fluorescence was detected with an inverted fluorescence microscope (OLYMPUS).

4T1 cells in a 12-well plate were incubated with different NPs for 6 h. After that, the cells were treated with or without irradiation (808 nm, 0.75 W/cm², 5 min). Then, DCFH-DA (1 μL) was added into the culture medium for 30 min. Fluorescence images of cells were captured using confocal laser scanning microscope (OLYMPUS).

The variation of mitochondrial membrane potential was detected with the JC-1 kit. 4T1 cells were incubated with different NPs for 6 h, and with laser irradiation for 5 min. JC-1 dye was added into the medium for 30 min. After washing with PBS, the fluorescence images were acquired under an inverted microscope (OLYMPUS).

4T1 cells in a 96-well plate were incubated with fresh medium containing different NPs for 4 h before laser irradiation for 5 min. After culturing for 20 h, the harvested cells were lysed to determine protein concentration, following with intracellular ATP assay using an ATP bioluminescent assay kit.

2.9. The evaluation of autophagy in vitro

Western blot analysis: 4T1 cells in 6-well plates were incubated with different NPs for 6 h, followed with laser irradiation or not. After culturing for another 24 h, cells were lysed in the buffer containing protease inhibitors and clarified by spinning at 14,000 rpm for 10 min at 4 °C. Samples were separated using 12% SDS-PAGE and then transferred to polyvinylidene difluoride (PVDF) membranes before incubating with antibody, then incubating with secondary antibody.

Transient expression of mRFP-GFP-LC3: The adenovirus vector containing the mRFP-GFP-LC3 reporters was purchased from Hanbio (HB-LP2100001). Ad-mRFP-GFP-LC3adenovirus was transfected into 4T1 cells at 37 °C with MOI of 20 for 24 h. The red fluorescent protein mRFP, green fluorescent protein (GFP) and LC3 fusion protein could be highly expressed in the 4T1 cells after infection. After different treatments for 16 h, LC3 spots were recorded by confocal laser scanning microscope (λex of mRFP: 580 nm; λex of GFP: 488 nm, OLYMPUS).

Autophagy image: 4T1 cells were incubated with different NPs for 4 h followed with laser irradiation (0.75 W/cm², 5 min) or not. After another 4 h of incubation, 4T1 cells were sequentially fixed with 2.5% glutaraldehyde for and osmium tetroxide for 2 h. After dehydration, the cells were embedded into wax for preparing ultrathin sections. The autolysosomes of cell sections were imaged under TEM (JEOL, JEM-2100Plus, Japan).

AO staining: 4T1 cells were incubated with different NPs for 6 h followed with laser irradiation or not. After culturing with new fresh medium for another 10 h, tumor cells were treated with 1 μL AO for 30 min. Then, the cells were washed with PBS and the red fluorescent dots of cells were monitored using fluorescent microscopy (OLYMPUS).

Lysosomal localization: 4T1 cells were incubated with rhodamine-labeled HMPOC@M NPs for 4, 8, 16 and 20 h. Then, tumor cells were treated with 50 nmol/L Lyso-Tracker Green for 30 min. After that, the fluorescence images of cells were acquired under an inverted microscope (OLYMPUS).

In vitro live/dead assay: 4T1 cells were pre-treated with the culture medium containing RAPA (200 nmol/L), 3-MA (1 μmol/L) or CQ (10 μmol/L) for 4 h. Then, the cells were incubated with different NPs (HMPO@M, HMPO@M+PAPA, HMPO@M+3-MA, HMPOC@M, HMPOC@M+PAPA or HMPOC@M+3-MA) for 6 h with the assistance of laser irradiation. After culturing for 20 h, calcein-AM and PI were added for live/dead cells staining.

2.10. Metastasis suppression in vitro

Would healing assay: the wound in 4T1 cells seeded into 6-well plates was prepared by using sterile 200 μL pipette tip. After starving for 6 h, the cells were incubated with the culture medium containing CBD, HMPO@M and HMPOC@M NPs for 6 h with the assistance of laser irradiation. After 48 h, the cells were photographed using a microscope (OLYMPUS).

Invasion assay: 4T1 cells seeded into 6-well plates were incubated with CBD, HMPO@M and HMPOC@M NPs for 48 h. Then, the cells were seeded in the upper chamber coated with 10 μg/mL of diluted Matrigel® at a density of 1 × 10⁵ cells/well. After 48 h incubation, the migrated cells were fixed with 4% paraformaldehyde and stained with 0.1% crystal violet for imaging with optical microscopy (OLYMPUS).

Endothelial tube formation assay: 4T1 cells in a 6-well plate were treated with CBD, HMPO@M+L and HMPOC@M+L. 24 h post-treatment, the medium was removed and replenished with 2% serum-fresh medium for 24 h to finally get conditioned medium. A 24-well plate was coated with reduced growth factor basement membrane matrix (Gibco, 50 μL per cm²) and incubated at 37 °C to allow gel solidification. The HUVECs were diluted in conditioned medium at a density of 5 × 10⁴ cells per 500 μL. Tube formation was monitored after 6 h and images were taken using an Olympus inverted microscope.

2.11. The transcriptome high throughput sequencing

Total RNAs of treated group (HMPOC+L, n = 3) and control group (PBS, n = 3) were extracted with TRIZOL according to the manufacturer’s instructions. The RNA extraction, purification, reverse transcription, library construction, and sequencing were performed at Lianchuan Biotechnology Co., Ltd. (Hangzhou, China) using Illumina HiSeq4000 (Illumina, San Diego, CA, USA). The gene expression level was determined by calculating the fragments per kilobase of exon per million mapped reads (FPKM) method, which represented fold change and P-value of differentially expressed mRNAs. The data were analyzed online with the Omicstudio Platform.

2.12. Animal experiment method

Balb/c mice (4–5 weeks) were purchased from Hunan SJA Laboratory Animal Co., Ltd. All animal experiments conform to the guiding principles of “Declaration of Helsinki” and have been approved by the Medical Ethics Committee of Hunan University.

2.13. In vivo biodistribution in main organs and pharmacokinetics

ICG-labeled HMPOC NPs were synthesized according to self-assemble process of CBD. ICG-labeled HMPOC, HMPOC@M^RBC, HMPOC@M^4T1 or HMPOC@M (2.5 mg/kg of ICG) were injected via tail vein for bio-distribution and tumor targeting efficiency assay. The tumors and major organs were isolated for ex vivo imaging (Caliper, Lumina xr, USA) after administration for 48 h. After intravenous administration with HMPOC, HMPOC@M^RBC, HMPOC@M^4T1 or HMPOC@M, 100 μL whole blood of mice were collected at preset time points for fluorescence imaging and quantification. Additionally, the obtained plasma samples were used for serum fluorescence quantification.

2.14. Antitumor effect of HMPOC@M in vivo

The orthotopic 4T1 breast cancer models were established by injecting 4T1 cells (1 × 10⁷ cells) into the mice. When the tumor volume reached about 100 mm³, the mice were randomly divided into 4 groups (6 mice/per group), as follows: I: PBS, II: CBD, III: HMPO@M, IV: HMPOC@M (CBD: 5 mg/kg; HMPO@M: 5 mg/kg; HMPOC@M: containing 5 mg/kg PB and 5 mg/kg CBD). After intravenous injection, laser treatments (group III and group IV) were treated with NIR irradiation on the second day (808 nm laser with 0.75 W/cm² for 5 min). Meanwhile, healthy mice were used as the normal group. Treatment was stopped at 20th day and only observation and documentation were maintained for the remaining 20 days. The body weight and tumor volume were measured and recorded every day. Moreover, all tumors were collected at the end of experiment and the prepared sections were stained with Ki67, H&E, and TUNEL. Meanwhile, other major organs were harvested for H&E staining and spleens were collected for weighing.

2.15. Immunofluorescence imaging

Briefly, antigen-repaired paraffin sections of tumor tissue were immersed in 0.3% Triton X-100 for 30 min. Then, the sections were sequentially sealed with milk and incubated with antibody (anti-MMP9, anti-VEGF, anti-LC3B, anti-SQSTM1 or anti-BECN1) at 4 °C overnight followed with secondary antibody (1:1000, goat anti-rabbit IgG conjugated to Alexa Fluor 488) incubation for 2 h. Finally, DAPI was used to label nuclei. After washing with PBS for 3 times, the fluorescence images were captured under laser scanning confocal microscope (OLYMPUS).

3.2. Biocompatibility and biodistribution assay of HMPOC@M NPs

It was reported that biomembrane camouflage can avoid thrombosis or hemolysis caused by bare NPs during blood circulation³³. Likewise, we found the ultra-low hemolysis rates (5%) of HMPOC@M with different concentrations in this study, accompanying the complete morphology of erythrocytes (Supporting Information Fig. S2A). Meanwhile, HMPOC@M did not cause platelet aggregation in the blood at all (Fig. S2B). Also, hematological analysis indicated negligible effect of several key indicators (RBC, WBC, HGB, PLT and MPV) in each group (Fig. S2C), compared with the normal group. Importantly, negligible effect was found for HMPOC@M on the TNF-α and IL-12 secretion (Fig. S2D), implying no allergic reaction.

Excellent nano-delivery systems should possess good biocompatibility and tumor-specific targeting so as to ensure drug efficacy. The “don’t eat me” protein (CD47) on the surface of biomembrane allows nanocomplexes to evade the surveillance of the immune system³⁴. The adhesion molecule (CD44), a another surface marker, is involved into the adherence of homologous cancer cells^35,36. Considering the co-existence of CD44 and CD47 in the hybrid membrane, we suspect that HMPOC@M can evade immune surveillance. As shown in Supporting Information Fig. S3A, the uptake efficiency of HMPOC@M decreased significantly compared to bare HMPOC in macrophages. Moreover, the circulation half-life for HMPOC@M was investigated to be 6.62 h, which was longer than that of bare HMPOC or HMPOC@M^4T1 (2.34 or 4.14 h), while comparable to that of HMPOC@M^RBC (6.85 h, Fig. S3B). This result suggests that RBC membrane can extend the circulation half-life of NPs due to their immune escape ability³⁷. Since the targeting ability of NPs was an important prerequisite for effective tumor elimination, we then investigated the uptake behavior of 4T1 cells on HMPOC@M NPs in vitro. As shown in Fig. S3C, the uptake efficiency of HMPOC@M showed a time-dependent manner with the saturate signal at 6 h. To further investigate the uptake mechanism, 4T1 cells were pretreated with small molecule endocytosis inhibitors before incubating with HMPOC@M NPs. Significantly, colchicine pretreatment reduced the uptake of HMPOC@M by more than 50% (Fig. S3D), suggesting that most HMPOC@M NPs entered 4T1 cells via micropinocytosis. Importantly, the phagocytosis of HMPOC@M by 4T1 cells was significantly higher than that of the other two non-homologous cells (over 2.5-fold, Fig. S3E). Then, we further observed the biodistribution of HMPOC@M in vivo by injecting fluorescent-labeled NPs coated with different biomembrane into mice via tail vein for 48 h. Fig. S3F indicated that the accumulation of HMPOC@M or HMPOC@M^4T1 was significantly higher than that of other NPs in tumor, due to the homing effect of 4T1 cell membrane³⁷. Quantitative assay indicated that the accumulation of HMPOC@M in the tumor region increased 7.6-fold after coating with hybrid membrane. The enrichment of NPs in the lung, an organ for nanomaterials deposition was observed, which is consistent with previous reports³⁸.

3.3. Drug release and CO/Mn²⁺ production

To determine the drug-carrying capacity of the NPs, drug loading and encapsulation efficiency were assessed by UV spectrometry. Loading rates of MnCO and CBD in HMPOC NPs were about 53.2 ± 3.23% (Supporting Information Fig. S4A), and 11.22 ± 1.12%, respectively (Fig. S4B). Moreover, the hydrodynamic size of HMPOC@M NPs with the polydispersity index (PDI) of less than 0.2 showed negligible change after incubating serum-containing buffer for 3 days (Fig. S4C). This result demonstrated that biomembrane could enhance the stability of the nano-delivery system by reducing the adsorption of naked HMPOC NPs to serum proteins and avoiding the formation of protein corona³⁹. Next, we monitored the temperature change under 0.75 W/cm² laser irradiation and found that the temperature of HMPOC@M solution dramatically increased (30 μg/mL, 10 °C) (Supporting Information Fig. S5A). The photothermal conversion efficiency (PCE) of HMPOC@M was calculated to be 35.9% (Fig. S5B) with concentration and laser power-dependent manners (Fig. S5C). Finally, the UV–Vis spectra were not affected after through 6 on/off irradiation cycles (Fig. S5D), implying the good photostability of HMPOC@M. These results provide considerable potential to accurately control drug release in tumor tissue and protect normal cells from thermal damage⁴⁰.

Meanwhile, we investigated the effect of pH value and the laser on the CBD release profile of HMPOC@M by UV‒Vis assay. Fig. 2A demonstrates that laser irradiation accelerated the leakage of CBD and a burst release of CBD (72%) at acidic condition within 48 h, which can attribute to enhanced hydrogen bonding of amphiphilic alkyl ethers to water in an acidic environment and to the accelerated diffusion of drug molecules by photothermal heating. Next, we synthesized FL-CO-1 probe (Supporting Information Fig. S6) according to the previous method for CO detection³⁰. Using H₂O₂ as substrate, we monitored the fluorescence intensity of different samples as previous reported method³⁰. As shown in Fig. 2B, the amount of CO generation dramatically elevated to 40 μmol/L within 30 h in the presence of 100 μmol/L H₂O₂ (present at concentrations ranging from 100 μmol/L to 1 mmol/L in tumor)⁴¹. Furthermore, we detected the production of intracellular CO and found distinct green fluorescence in HMPO@M+L or HMPOC@M+L group (Fig. 2C). Next, ICP-MS assay was used to detect the Mn²⁺ generation and found it was affected by the H₂O₂, laser irradiation and the pH value (Fig. 2D). Then, electron spin resonance (ESR) assay indicated that the strongest peak standing for ·OH appeared in the HMPOC@M + H₂O₂ + HCO₃^‒ + NIR group. In contrast, no obvious peak was detected in the NPs or H₂O₂ solution (Fig. 2E). Then, aminophenol fluorescein (APF) probe was used to visually illustrate the levels of ·OH in tumor cells as the interaction between ·OH and APF probe can emit green fluorescence. As we expected that the HMPOC@M+L-treated 4T1 cells showed the strongest green fluorescence signal (Fig. 2F) due to the most efficient Mn²⁺ generation and accelerated subsequent ·OH production under acidic environment (Fig. 2G).

3.4. Cytotoxicity induced by HMPOC@M

Previous studies have indicated the function of CBD on the regulation of ROS and autophagy²⁵. In this study, we also found the concentration-dependent inhibition of CBD on the levels of NRF2 and LAMP1, accompanied by the c-Jun N-terminal kinase (JNK) and BECN1 upregulation (Fig. 3A). This suggests that CBD could amplify oxidative stress and stimulate autophagy in 4T1 cells. Then, we investigated the synergistic effect of HMPOC@M on the cytotoxicity of 4T1 cells in vitro. Supporting Information Fig. S7A demonstrated the IC₅₀ of CBD alone was 9 μmol/L and the toxicity of HMPB to 4T1 cells and normal cells was negligible (Fig. S7B and C). Meanwhile, viability of cells with HMPO@M +L (30 μg/mL, 0.75 W/cm²) treatment was 62%, due to the contribution of CO/Mn²⁺ (Fig. 3B). However, the viability was less than 15% after with HMPOC@M+L (Fig. 3B), due to the synergistic effect of CBD and CO/Mn²⁺. Also, HMPOC@M+L treatment induced the overwhelming cell death in the live/dead staining assay (Fig. 3C), as reflected by the loss of green fluorescence (live cells) in the visual field. This result also supports the synergistic effect of HMPOC@M+L on the cell viability.

CO can induce mitochondrial collapse through generating ROS⁴. As DCFH-DA can be oxidized by ROS to DCF after deacetylation, accompanied by the green fluorescence increase, this probe was used to detect intracellular ROS level caused by CO. As we expected, the HMPO@M+L group presented a moderate green fluorescent signal. However, the HMPOC@M+L group exhibited synergistic effect (amplification of oxidative stress) after introducing CBD (Fig. 3D), which was reflected by the stronger green fluorescence in 4T1 cells. In addition, western blot assay indicated that the levels of NRF2 in 4T1 cells decreased more than 40% after with HMPOC@M+L treatment due to the existence of CBD, while no similar effect was found in tumor cells with HMPO@M+L treatment (Fig. 3H). Considering that NRF2 can cause ROS detoxification in cancer cells⁴², this result suggests that NRF2 downregulation could amplify oxidative stress by co-operating with CO/Mn²⁺. It is well known that the decrease of mitochondrial membrane potential is the important sign of mitochondrial damage⁹, JC-1 probe, which can exhibit brighter red fluorescence signal compared to mitochondria with lower membrane potential, was used to detect the effect of ROS on the mitochondrial damage. As shown in Fig. 3E, the red fluorescence in 4T1 cells was wholly replaced by the bright green signal after with HMPOC@M+L treatment. This result demonstrates the strongest damage of HMPOC@M+L treatment on the mitochondria, comparing with other groups. Meanwhile, the swelling of inner mitochondrial cristae gaps and cristae vesicles visually demonstrated the strong mitochondria damage in the 4T1 cell (Fig. 3F). Similarly, the levels of intracellular ATP decreased more than 80% in 4T1 cells after with HMPOC@M+L treatment (Fig. 3G). In addition, western blot assay indicated that p-AMPK/AMPK ratio, responding to intracellular decrease of ATP, was elevated 2.8-fold in 4T1 cells HMPOC@M+L treatment (Fig. 3H). These results comprehensively indicate the strongest damage in 4T1 cells with HMPOC@M+L treatment, which may result in the occurrence of excessive autophagy in 4T1 cells.

3.5. Mechanisms of autophagy in HMPOC@M therapy

By observing the autophagy under bio-transmission electron microscopy, we found a lot of autolysosome accumulation (indicated by red arrows) in the 4T1 cells with HMPOC@M+L treatment, demonstrated that HMPOC@M+L and HMPO@M+L treatments could evoke autophagy in 4T1 cells (Fig. 4A). As a molecular sensor for damaged mitochondria, PTEN-induced kinase 1 (PINK1) can trigger the signal of mitophagy initiation and recruit the “enhancer” of PARKIN to mitochondria to amplify the mitophagy signals⁴³. Western blot assay indicated the similar upregulation of HMPOC@M+L and HMPO@M+L on the PINK1 and PARKIN levels of tumor cells. The above results demonstrate that HMPOC@M+L can induce mitophagy through PINK1/PARKIN signal pathway (Fig. 4B). It was reported that CBD induced JNK activation can inhibit the interaction between BECN1 and BCL-2 in TNBC²⁷. Then, the formed class III PI3K/BECN1 complexes initiate autophagy for cell death²⁸. Consistent with previous report²⁵, we found that CBD can upregulate JNK and BECN1 expression in a concentration-dependent manner (Fig. 3A). Similarly, CBD-contained HMPOC@M+L group upregulated BECN1 about 1.5-fold. In addition, both of LC3B (classical initiation proteins of autophagy) and LC3B/LC3A ratio in the HMPOC@M+L treated cells were higher than those of HMPO@M+L-treated cells. These results demonstrate that HMPOC@M+L treatment efficiently induces autophagy through JNK/BECN1 signaling pathway.

Next, we assessed the effect of HMPOC@M+L on the entire process of autophagy. AO staining was first performed to investigate the effect of HMPOC@M+L on late autophagic vesicles (autolysosomes) as the entering of AO into autolysosomes can emit red fluorescence⁴⁴. As shown in Fig. 4C, the red fluorescence signal in the HMPOC@M+L-treated cells significantly decreased comparing with that of HMPO@M+L-treated cells. This result demonstrate that CBD-contained HMPOC@M strongly inhibits the formation of autolysosomes, which can result in the accumulation of autophagosomes (the elevated LC3B levels). The above result in Fig. 3A indicates the inhibiting function of CBD on the LAMP1, a key protein for regulating pH value of lysosomes. We then investigated the effect of HMPOC@M+L on the pH value of lysosomes using Lyso-tracker probe (green fluorescence probe). Fig. 4D demonstrates the gradual decrease of green fluorescence signal in treated cells with the time elongation. Meanwhile, the level of LAMP1 was significantly decreased in 4T1 cells with HMPOC@M treatment (Fig. 4F). These results directly reflected the HMPOC@M can inhibit the formation of autolysosomes through lysosomes alkalization⁴⁵. The mRFP-GFP-LC3 adenovirus transfection was further used to track the autophagy flux in 4T1 cells. Fig. 4E indicates obvious increase of yellow dots in the HMPOC@M+L-treated cells. In contrast, only red dot number increased in the HMPO@M+L-treated cells with same treatment. As yellow dots (the overlapping of red and green fluorescence) and red dots (the quenching of green fluorescence as the acidic environment) represent autophagosomes and autolysosomes⁴⁴, respectively, this result also demonstrate that HMPOC@M+L treatment could inhibit autophagy flux by blocking autolysosome formation. Similarly, SQSTM1 upregulation in the HMPOC@M+L-treated cells also reflected the inhibition of autolysosome formation⁴⁴. In contrast, this protein was degraded in the HMPO@M+L group (Fig. 4F), due to the normal fusion of autophagosomes and lysosomes⁴⁴. In summary, HMPOC@M+L-induced excessive autophagy was contributed to the accumulation of autophagosomes and the blockade of autolysosome formation.

Next, autophagy regulators (activator and inhibitors) were adopted to confirm the excessive autophagy induced by HMPOC@M+L could efficiently kill tumor cells. According to the result of Supporting Information Fig. S8, we determined the safe concentrations of 1 and 10 μmol/L for 3-MA (early autophagy inhibitor) and CQ (late autophagy inhibitor) without cytotoxicity on tumor cells. However, the presence of 3-MA and CQ with safe concentration increased viability of 4T1 cells with HMPOC@M+L treatment (Fig. 4G). In contrary, the presence of 200 nmol/L RAPA (early autophagy activator) further reduced the viability of 4T1 cells with HMPOC@M+L treatment. This data suggested that mitophagy induced CO/Mn²⁺ can facilitate cancer cell survival. These results, which were consistent with live/dead cell assay (Supporting Information Fig. S9), suggested that the high 4T1 cell killing ability efficiency of HMPOC@M was tightly related with excessive autophagy (Fig. 4H).

3.6. Transcriptome high-throughput sequencing analysis

To further illuminate the underlying mechanism of HMPOC@M against tumor cells, transcriptome sequencing was adopted to investigate the mRNA levels of genes in 4T1 cells. The principal component analysis (PCA) and heat map apparently demonstrated transcriptomics change of 4T1 cells caused by HMPOC@M+LNPs (Fig. 5A and B). The venn diagram indicated exclusive expression of 1302 genes in the treatment group, except 22594 genes with co-expression (Fig. 5C). Based on the screening criteria of FDR < 0.05 and |log₂ FC|=>1, 1938 genes were found to involve in the treatment process of HMPOC@M+L, including upregulation of 625 genes and down-regulation of 1313 genes (Fig. 5D and E).

Subsequently, we performed gene ontology (GO) analysis to investigate the biological function of genes with transcription change and found that integral components of mitochondrion were highly enriched in the GO analysis (Supporting Information Fig. S10). Meanwhile, KEGG pathway enrichment analysis revealed the important role of HMPOC@M+L on the regulation of apoptosis, autophagy, MAPK signal pathway, PI3K-Akt-mTOR signal pathway and p53 signal pathway (Fig. 5F). Moreover, gene set enrichment analysis (GSEA) indicated the significant enrichment of differentially expressed genes in the signal pathways of apoptosis and autophagy (Fig. 5G). It was reported that high ROS level in mitochondria could induce apoptosis by activating the MAPK signal pathway. Meanwhile, the PI3K-Akt-mTOR signal pathway was involved in the cellular autophagy^46,47. p53 also played an important role in apoptosis and lethal autophagy triggered by mitochondrial dysfunction⁴⁸. By combining with above findings of transcriptome analysis, we can suspect that regulation of autophagic function by HMPOC@M+L NPs is the main reason of efficient TNBC therapy.

Low level of ROS acts as signaling messenger for promoting cancer cell invasion⁴⁹. However, anti-cancer drug-initiated stress signal could induce excessive ROS to induce cancer cells death. Paradoxically, excessive ROS often induces DNA mutation and/or promotes tumor metastasis at various stages of cancer progression⁵⁰. However, the therapeutics based on ROS is cancer-type and stage-dependent. Considering the high ROS level and autophagy induced by HMPOC@M+L treatment, we investigated its effects on 4T1 cell metastasis. Scratch assay indicated the wound healing rates of tumor cells treated with HMPOC@M+L decreased 62.1%, compared to the control group (Supporting Information Fig. S11A). Transwell invasion assay revealed that both of CBD and HMPOC@M+L treatment significantly attenuated the invasive ability of 4T1 cells compared with the blank control. Quantitative analysis demonstrated the similar invasion rates for CBD- and HMPOC@M+L-treated cells (40.2% vs. 38.1%, Fig. S11B). Considering the negligible difference of transwell assay between CBD and HMPOC@M+L treatment, this result directly reflects the inhibitory function of CBD on tumor cell invasion. Western blot assay further demonstrated significant down-regulation of MMP9 and VEGF, the two key proteins of metastasis regulation in CBD- and HMPOC@M+L-treated 4T1 cells. However, no similar effect was found in HMPO@M+L group (Fig. S11C). These data suggest that CBD plays a key role on cancer cell invasion into the extracellular matrix (ECM) by reducing MMP9 secretion⁵¹. In addition, conditional medium of CBD- or HMPOC@M+L-treated 4T1 cells with low VEGF expression was used to induce endothelial tube formation of HUVECs. By monitoring the number of nodes, number of mesh, and total tube length, we found that all angiogenesis parameters of HUVECs decreased in the conditional medium, comparing with that of whole medium (Fig. S11D). These results suggest that HMPOC@M+L maybe efficiently inhibit distal metastasis of tumor cells by reducing cell infiltration into ECM and blocking angiogenesis.