Copper (II) chloride dihydrate (CuCl₂·2H₂O, analytical grade), sodium citrate dihydrate (Na₃C₆H₅O₇·2H₂O, analytical grade), sodium sulfide nonahydrate (Na₂S·9H₂O, analytical grade), and polyvinylpyrrolidone K30 (PVP K30, analytical grade) were obtained from Sinopharm Chemical Reagent Co., Ltd. (Shanghai, China). Zinc acetate dihydrate [Zn (CH₃CO₂) · 2H₂O, 98%], 2-methylimidazole (98%), catalase (CAT) were purchased from Aladdin Chemistry Co., Ltd. (Shanghai, China). Tris (4,7-diphenyl-1,10-phenanthroline)-ruthenium (II) dichloride [Ru (dpp)₃Cl₂, 98%] was obtained from Rhawn Reagent Co., Ltd. (Shanghai, China). Doxorubicin hydrochloride (DOX·HCl, 98%) was supplied from Aladdin Chemistry Co., Ltd. (Shanghai, China). BCA Protein Assay Kit, 3-[4,5-dimethylthiazol-2-yl]-2,5-diphenyltetrazoliumbromide (MTT, 98%) and hoechst 33258 were obtained from Beyotime Biotechnology Co., Ltd. (Shanghai, China). Cell Viability/Cytotoxicity Detec (Calcein AM/PI) was obtained from Yuanye Bio-Technology Co., Ltd (Shanghai, China).

Mouse breast cancer cells (4T1) and leukemia cells in mouse macrophage (RAW264.7) were obtained from the Chinese Academy of Sciences (Shanghai, China). The 4T1 cells were incubated with DMEM/f12 medium containing 10% FBS, 1% streptomycin (50 U·mL⁻¹), and penicillin (50 U·mL⁻¹) in a 5% CO₂ atmosphere at 37°C. The RAW264.7 cells were cultured in DMEM medium supplemented with 10% FBS, 1% streptomycin (50 U·mL⁻¹), and penicillin (50 U·mL⁻¹) in a 5% CO₂ atmosphere at 37°C.

BALB/c female mice (20 ± 2 g) were purchased from the Laboratory Animal Center of Xuzhou Medical University (Xuzhou, China), and maintained in a sterile environment, and allowed free access to food and water, and all experiment procedures were approved by the Experimental Animal Ethics Committee of Xuzhou Medical University and carried out by the guidelines of the National Act on the Use of Experimental Animals (People’s Republic of China).

The above prepared DC@ZIF8 and CAT were dispersed in 10 ml of RBC membrane solution and then ultrasonicated with ice bath for 6 min. The precipitation of DC@ZIF8-MEM_C was collected by washing and centrifugation, and placed at 4°C for later use. DC@ZIF8-MEM was prepared by the same method above except for the absence of CAT.

The loading content (LC) was calculated using the following formula: LC = weight of loaded drug/weight of NPs*100%. DOX·HCl was quantitatively analyzed using fluorescence spectrum (excitation wavelength = 488 nm, emission wavelength = 560 nm). The quantification of CAT was determined by ultraviolet spectrophotometer based on ammonium molybdate colorimetric method. The NPs were digested by nitric acid, and Cu²⁺ content was detected by an inductively coupled plasma spectrometer (PQ9000, German), and then the LC of CuS was analyzed.

Infrared spectra and ultraviolet (UV)-visible (UV-vis) spectra were obtained by using a Fourier transform infrared spectrometer (FTIR, BRUKER) and an Evolution 220 UV-Visible Spectrophotometer (Mapada Instruments, Shanghai), respectively.

DC@ZIF8-MEM_C was incubated with 10 mM PBS (pH 7.4 or pH 5.0) at 37°C for 8 h, and the morphology or size was detected.

DC@ZIF8-MEM_C was suspended in PBS (pH 7.4, 10 mM) or 10% fetal bovine serum (FBS), and measurements were performed in triplicate at different intervals.

The fluorescence of O₂-sensitive fluorescent probe Tris (4,7-diphenyl-1,10-phenanthroline) ruthenium (II) dichloride [Ru (DPP)₃Cl₂] will be quenched to a certain extent when O₂ concentration is sufficient. H₂O₂ solution (10 mM), Ru (DPP)₃Cl₂ solution (1 mM), and DC@ZIF8-MEM_C (125 μg·mL⁻¹) were mixed and sealed. The emission spectra were scanned at a predetermined time (0, 10, 20, 30, 45, 60 min).

Additionally, the portable dissolved oxygen meter was used to measure the in situ O₂ generation. Free CAT, DC@ZIF8-MEM_C, DC@ZIF8-MEM, and PBS were mixed with H₂O₂ solution (10 mM), and the O₂ production was dynamically detected by the portable dissolved oxygen meter (JPBJ-609L, INESA Scientific Instrument Co., Ltd., China) every 30 s for 600 s.

Catalase activity was evaluated by the Goth method (Goth, 1991). Free CAT, DC@ZIF8-MEM_C [(CAT) = 2.5 μg·mL⁻¹] and DC@ZIF8-MEM were incubated with H₂O₂ solution (50 mM) at 37°C, then ammonium molybdate was added to terminate the reaction, and the absorbance was determined at 400 nm according to a standard curve of H₂O_2. CAT was reliable to degrade by the protease K under complicated physiological conditions (Phua et al., 2019). Free CAT and DC@ZIF8-MEM_C was incubated with protease K (100 μg·mL⁻¹) at 37°C, and then the absorbance was determined according to the method above.

4T1 cells were inoculated and incubated overnight at a density of 1 × 10⁵ cells/dish in laser confocal Petri dishes. Fresh medium containing DC@ZIF8-MEM_C [(DOX·HCl) = 1 μg·mL⁻¹] was added and incubated for 0.5, 2, 6 h, respectively. The samples were washed with PBS and then stained with Lysotracker for 20 min and Heochst-33258 for 10 min. Intracellular location was observed by confocal laser microscopy (CLMS) and analyzed by ImageJ software (Zhu et al., 2018).

RAW 264.7 cells were selected as a macrophage model to demonstrate that the composite vector coated by blood cells can reduce the probability of being phagocytosed by macrophages. RAW 264.7 cells were seeded at density of 1 × 10⁴ cells/well in 96-well plates and incubated overnight. Different samples (DOX·HCl, DC@ZIF8, and DC@ZIF8-MEM_C, with DOX·HCl concentration 1 μg·mL⁻¹) were added, and incubated for 4 h. The medium was discarded, and Heochst-33258 was used as a stain for 10 min. Cell uptake was observed by inverted fluorescence microscopy. The quantitative uptake was investigated similarly as described above.

4T1 cells were seeded at a density of 4 × 10⁴ cells/well into 24-well plates, and 10 µM H₂O₂ was added for 2 h with a hypoxic incubator. Different samples were added and incubated for 4 h. Ru (DPP)₃Cl₂ (0.5 µM) was added for 1 h and observed by an inverted fluorescence microscopy.

The animal model was built by subcutaneous injection with 100 µl of 4T1 cells (1 × 10⁶ cells) in the right hind limb of female BALB/c mice. The tumor-bearing mice were randomly divided into seven groups (each group had five mice with similar tumor volume about 200 ± 20 mm³): 1) PBS group, 2) DOX·HCl group, 3) C@ZIF8-MEM group; 4) D@ZIF8-MEM group; 5) DC@ZIF8-MEM + NIR group; 6) DC@ZIF8-MEM_C group; 7) DC@ZIF8-MEM_C + NIR group. The mice were treated with PBS (group 1), NPs (groups 2 to 7, with a dosage of DOX·HCl 5 mg·kg⁻¹) via tail vein injection at day 0, 3, 6, and 9 for an overall of four times. For the NIR groups, the mice were treated with NIR 808 nm irradiation (2 W·cm⁻²) at the tumor site for 5 min after 8 h post-injection. Tumor volume was calculated by the formula (a × b²)/2, where a and b are the long and short diameters of the tumor, respectively, which were measured every other day by a vernier caliper (Wu et al., 2018).

On day 14, the mice were sacrificed, and major organs, including the heart, liver, spleen, lungs, kidneys, and tumor were collected for pathological evaluation.

The tumor-bearing mice were tail vein injected with PBS, DC@ZIF8, DC@ZIF8-MEM, and DC@ZIF8-MEM_C. After an 8 h-injection, the mice were anesthetized with 40 μl of 10% chloral hydrate by intraperitoneal injection. The tumor site of the mice was irradiated by 808 nm of NIR (2 W·cm⁻²) for 5 min. Simultaneously, the photothermal imaging of the tumor site was recorded by the near-infrared imager. The temperature change in the tumor site with time was observed, and the temperature curve was plotted.

The hemocompatibility was evaluated by hemolysis assay (Yang et al., 2019). First, fresh blood extracted from rabbit heart was centrifuged at 3,000 rpm for 15 min and purified with normal saline (NS) to obtain erythrocytes. DC@ZIF8-MEM_C with various concentrations were mixed with the 2% erythrocytes (v/v) and incubated for 3 h. Afterward, the mixtures were centrifuged at 3,500 rpm for 15 min, and the absorbance (A) of the sample supernatant at 540 nm was measured by UV-vis spectrophotometer. PBS (0.01 M, pH 7.4) and deionized water were regarded as the negative and positive controls, respectively. The hemolysis ratio (HR) was calculated as follows: HR (%) = (A_sample − A_{negative control})/(A_{positive control} − A_{negative control)} * 100%.

The data were expressed as the mean ± standard deviation (SD). Statistical analysis was performed using a two-tailed Student’s t-test and analysis of variance (ANOVA) with the software SPSS 23.0. All analyses were shown compared with the control, and the significance of the difference was indicated as *p < 0.05, **p < 0.01, or ***p < 0.001.

CuS NPs appeared dark green (Supplementary Figure S1). Both DC@ZIF8 and DC@ZIF8-MEM_C showed purplish red due to the loaded DOX·HCl (Supplementary Figure S1). CuS NPs were spherical, while DC@ZIF8 was polyhedral due to the frame of ZIF8 by TEM (Supplementary Figure S2). After coating with MEM-embedded CAT, DC@ZIF8-MEM_C maintained a polyhedral structure by TEM (Figure 2A). The size of CuS NPs was only 7 nm, and C@ZIF8 increased to 230 nm due to the framework of ZIF8 (Figure 2B). After drug loading, the size of DC@ZIF8 was 232 nm, and the size of DC@ZIF8-MEM_C increased to 254 nm for the coating of cell membrane (Figure 2B). Both C@ZIF8 and DC@ZIF8 were positive zeta potential (Figure 2C). However, the zeta potential of DC@ZIF8-MEM_C reversed to be negative (−9.3 mV), demonstrating a successful membrane coating (Figure 2C). DOX·HCl was quantitatively analyzed by fluorescence spectrum (Supplementary Figure S3). The CAT was quantitatively analyzed by ultraviolet spectrophotometer (Supplementary Figure S4). The LC of DOX, CAT, and CuS was 4.9%, 6.2%, and 2.5%, separately. Moreover, DC@ZIF8-MEM_C exhibit good colloidal stability in both PBS and 10% FBS, as indicated by the negligible increase in hydrodynamic diameter (Supplementary Figure S5).

UV-Vis analysis showed that DC@ZIF8-MEM_C, DC@ZIF8, and CuS NPs all had a very wide absorption band in the near infrared region from 800 to 1,100 nm. Therefore, a near-infrared laser of 808 nm can be used as a heat source for photothermal effect. DC@ZIF8-MEM_C had characteristic absorbance peaks from DOX (480 nm) and CAT (400 nm), confirming the successful loading of these two materials (Figure 2D).

By recording Fourier transform infrared (FTIR) spectra, the prepared DC@ZIF8-MEM_C was further characterized. As depicted in Figure 2E, DC@ZIF8-MEM_C exhibited a characteristic peak of 1,700 cm⁻¹, which was the C=O stretching vibration of sodium citrate in CuS NP_S, and a –CH stretching vibration peak at 2,800 cm⁻¹, which was consistent with the characteristic peak of DOX·HCl. DC@ZIF8-MEM_C had a characteristic peak of 3,500–3,700 cm⁻¹, which was abundant –OH peaks on the surface of CAT and heterozygous membrane. The results of FTIR spectrum analysis and UV analysis were consistent, which showed that DC@ZIF8-MEM_C was successfully constructed.

Drug release in vitro showed pH-dependent manners (Figure 2F). The cumulative release of DOX·HCl in pH 7.4, pH 6.5, and pH 5.0 media at 24 h was 3.1% 11.9%, and 44.7%, respectively. There was little drug release in pH 7.4, indicating that the developed NPs were stable under physiological conditions. Compared with pH 5.0 group, release from pH 5.0 + NIR group was higher, indicating that NIR accelerated the drug release. Interestingly, pH sensitivity of DC@ZIF8-MEM_C was evident from Figures 2G, H. When DC@ZIF8-MEM_C was in a neutral condition (pH 7.4 PBS), the morphology and size changed little for 8 h (Figure 2H). It degraded under acidic conditions (pH 5.0), and the size increased to 2,000 nm, 10-fold bigger than that of PBS (pH 7.4) (Figure 2G). Under acidic conditions, the release amount of the drug was significantly increased, which may be related to the degradation of the material at lower pH value (Wu et al., 2018). The imidazole groups from ZIF8 were protonated in an acidic environment, and the unstable coordination bonds between zinc ions and imidazole groups became unstable, leading to the disintegration of the skeleton and acceleration of the drug release.

As shown in Figures 3A, B, the temperature rise followed a concentration-dependent manner. The temperature of the water increased by only 2.1°C. At concentrations of 60, 125, 250 μg·mL⁻¹, DC@ZIF8-MEM_C increased by 9.5°C, 17.7°C, and 28.4°C, respectively (Figures 3A, B). Furthermore, the temperature increased with power. The temperature increased by 17.4°C when the concentration was 125 μg·mL⁻¹, and the power density of the 808 nm laser was 2.0 W·cm⁻² (Figures 3C, D).

Moreover, by measuring the rising/falling temperature, it was found that the photothermal conversion was stable. The photothermal conversion efficiency was calculated to be 51.5%, consistent with a report (Shi et al., 2018), indicating the excellent photothermal performance of DC@ZIF8-MEM_C (Figures 3E, F). Intriguingly, the photothermal conversion efficiency of DC@ZIF8 was 54%, similar to that of DC@ZIF8-MEM_C, which indicated that there was no influence of membrane coating on photothermal property (Supplementary Figure S6).

The in situ O₂ production was measured by using a portable dissolved oxygen meter. As seen from Figure 3G, the DC@ZIF8-MEM and PBS groups did not generate any O₂, while the DC@ZIF8-MEM_C and CAT groups generated time-dependent O₂ production due to the activity of CAT. Bubbles were visible in the EP tubes of free CAT and DC@ZIF8-MEM_C, while no bubbles were found with DC@ZIF8-MEM and PBS (Figure 3G). As an oxygen-sensitive dye, the fluorescence of Ru (DPP)₃Cl₂ was quenched when O₂ was present in the solution. With the prolonging of reaction time, the fluorescence intensity of Ru (DPP)₃Cl₂ decreased gradually (Figure 3H). When the reaction time was 30 min, the fluorescence intensity of Ru (DPP)₃Cl₂ decreased to the lowest.

CAT is an enzyme that catalyzes the decomposition of H₂O₂ to generate O₂. At 2 h, 68% of H₂O₂ was consumed with 2.5 μg·mL⁻¹ of free CAT, 64% for DC@ZIF8-MEM_C with the same CAT concentration, indicating the same catalytic efficiency as that of free CAT (Figure 3I, Supplementary Figure S7). Comparatively, DC@ZIF8-MEM consumed barely H₂O_2. Thus, the catalytic activity of DC@ZIF8-MEM_C indeed was derived from CAT. CAT is unstable under physiological conditions, due to enzymic digestion by protein K. At 12 h, the catalytic activity decreased to 34% for free CAT, while 62% was maintained for DC@ZIF8-MEM_C (Figure 3J). DC@ZIF8-MEM_C is capable of protecting CAT from degradation, which is useful for CAT delivery to maintain catalytic activity in vivo.

To localize the cells, the cell nuclei were stained blue by Heochst-33258. The traffic of DOX·HCl inside the cells could be visualized by its intrinsic red fluorescence (Figure 4A). The fluorescence intensity of DC@ZIF8-MEM_C was stronger than that of other treatment groups, presenting the improved cellular uptake through endocytosis (Figure 4A). The relative fluorescence intensities were quantified, which suggested a higher uptake efficiency of DC@ZIF8-MEM_C, consistent with the result of qualitative analysis (Figure 4B, Supplementary Figure S8). It was attributed from MEM coating, which was helpful for accumulation at the tumor sites by passive targeting (Jiang et al., 2017; Ye et al., 2019).

The in vitro tumor ablation was measured by MTT assays. The cell killing efficiency was dose dependent (Figure 4C). When concentration was at 5 μg·mL⁻¹, the cell viability was 59%, 19% for free DOX·HCl, DC@ZIF8-MEM_C, respectively. Under laser irradiation, the cell viability of groups treated with DC@ZIF8-MEM_C decreased to 11%. DC@ZIF8-MEM_C indicated that a combination of CT and PTT as well as CAT was better than monotherapy alone.

As seen in Figure 4E, both DC@ZIF8-MEM_C and DC@ZIF8-MEM exhibited dose-dependent cytotoxicity. Under hypoxia condition (1% O₂), compared with DC@ZIF8-MEM, the cell viability of DC@ZIF8-MEM_C decreased significantly, demonstrating that the O₂ produced by CAT could alleviate hypoxia-induced resistance to DOX (12 versus 51% at 0.5 μg·mL⁻¹). These results could explain, to some extent, why the treatment effect of DC@ZIF8-MEM_C was much better than DC@ZIF8-MEM in vivo experiment.

In addition, when the concentration of NPs was 120 μg·mL⁻¹, the cell survival rate of C@ZIF8-MEM was 80.0%, and the cell survival rate decreased to 35.0% after NIR (Figure 7A). The results showed photo-responsive cytotoxicity of NPs.

To further evaluate the therapeutic outcome of the NPs, the treated cells were co-stained by calcein-AM and propidium iodide (PI) for live (green) and dead/late apoptotic cells (red), respectively (Liu et al., 2019). Both the control group and NIR group displayed strong green fluorescence (Figure 4D). However, compared with DOX·HCl and DC@ZIF8-MEM_C groups, the red fluorescence was predominant for the DC@ZIF8-MEM_C + NIR group, displaying a significant number of dead cells, which was in line with the results of cytotoxicity in vitro (Figure 4D).

As shown in Figure 5A, there were orange spots that emerged from the red and green channels, an evidence that red fluorescence of the majority of DC@ZIF8-MEM_C was in colocalization with green fluorescence (endosomes/lysosomes) at 0.5 h. Over time, the amount of orange spots was weakened, and red color was evenly distributed throughout the cytoplasm, which showed minimal colocalization with the green fluorescence of the endo/lysosomes. The decreased colocalization signal of green and red fluorescence was visualized after 2 h due to the ongoing intracellular transport and rapid endosomal escape. Meaningfully, most of the overlapped signals disappeared after 6 h. The results presented that DC@ZIF8-MEM_C entered the lysosome and then escaped into the cytoplasm after uptake into the cell. To semiquantitatively evaluate the capacity of endosomal escape for DC@ZIF8-MEM_C, the index of colocalization and line scanning profiles of fluorescent intensity of the selected 4T1 cells were calculated by ImageJ software, as given in Figures 5B, C.

These results jointly corroborated the efficient endosomal/lysosomal escape behaviors of DC@ZIF8-MEM_C, which was attributed to the ZIF8 frame. Under the lysosome pH (pH 4.5–5.5), the ZIF8 moiety could release the 2-methylimidazole ligand, which initiated the “proton-sponge” effect due to the protonation of imidazole rings and resulted in the rupture of lysosome membrane. (Liu Z et al., 2021). The rapid lysosomal escape capability of DC@ZIF8-MEM_C was beneficial to maintain the enzymatic activity of CAT and the therapeutic activity of DOX from lysosome to cytoplasm (Varkouhi et al., 2011).

Red fluorescence was strong for DOX·HCl group, while weak it was for DC@ZIF8-MEM_C group (Figure 5D). As shown in Figure 5E, the uptake rate of DC@ZIF8-MEM_C by RAW 264.7 cells were significantly lower than those of the other two groups, displaying that NPs camouflaged by blood cell membranes could avoid the phagocytosis of immune cells. As natural stealth coating, the cell membrane could alleviate immunogenicity and enhance biocompatibility (Graham and Unger, 2018).

Oxygen probe Ru(DPP)₃Cl₂ was applied for intracellular determination. The fluorescence of Ru(DPP)₃Cl₂ would be weakened in the presence of abundant O₂. As from Figure 5F, compared with the control group, the red fluorescence of C@ZIF8-MEM was strong and predominant, while that of the C@ZIF8-MEM_C group was weak and marginal. For C@ZIF8-MEM without embedded CAT, the fluorescence of Ru (DPP)₃Cl₂ barely changed. However, the fluorescence change was weak due to the O₂ generation for the C@ZIF8-MEM_C group. Moreover, the fluorescence tended to be weaker with the increase in CAT concentration of C@ZIF8-MEM_C. The results verified that the intracellular O₂ generation of C@ZIF8-MEM_C originated from embedding the CAT.

The experimentation for the therapeutic model is shown in Figure 6A. The antitumor effect was monitored by measuring the tumor size using a caliper every other day. With PBS or C@ZIF8-MEM treatment, the tumor grew rapidly over time (Figure 6D). Without irradiation, the DC@ZIF8-MEM_C showed moderate tumor inhibition due to the CT of DOX·HCl and hypoxic improvement of CAT. Combined with irradiation, the tumor inhibition effect was significantly elevated (Figure 6D). The tumor weight was only 0.12 g for the DC@ZIF8-MEM_C + NIR group, while the tumor weights were 1.19, 1.13, 0.57, and 0.57 g for DOX·HCl, DC@ZIF8-MEM, DC@ZIF8-MEM + NIR, and DC@ZIF8-MEM_C, respectively (Figure 6E). It could be clearly seen from the photographs of the extracted tumor tissues after treatment that the tumors treated with DC@ZIF8-MEM_C plus irradiation were ablated significantly (Figure 6F). Interestingly, compared with the DC@ZIF8-MEM + NIR, the DC@ZIF8-MEM_C + NIR group exhibited higher tumor inhibition effect, which may be attributed to amelioration of CAT for tumor hypoxia (Figures 6D–F). The results were consistent with cytotoxicity in vitro. Upon irradiation, DC@ZIF8-MEM_C displayed the most effectivity in tumor inhibition, superior to other groups, indicating the advantages of combined cancer therapy. Tumor slices had a large area of necrosis for the DC@ZIF8-MEM_C + NIR group, consistent with the therapeutic effect (Figure 6G).

After continuous irradiation of the tumor site with the 808 nm near-infrared laser (2 W·cm⁻²) for 5 min, the near-infrared imaging image of the tumor site is shown in Figure 6B. The infrared light intensity of the tumor site for the DC@ZIF8-MEM_C group was significantly stronger than that of the other groups. Figure 6C shows the temperature rise curves of each group. The temperature of the DC@ZIF8-MEM group and the DC@ZIF8-MEM_C group changed rapidly within 1 min. After 5 min, the temperature of the tumor site increased to 40.1°C, 46.6°C, 52°C, and 55.5°C for PBS, DC@ZIF8, DC@ZIF8-MEM, and DC@ZIF8-MEM_C, respectively (Figure 6C). It presented that DC@ZIF8-MEM_C played an effective photothermal effect.

The biocompatibility of the NPs was evaluated. The cell viability of C@ZIF8-MEM displayed more than 80%, exhibiting good biocompatibility (Figure 7A, Supplementary Figure S9). The positive control (deionized water) showed obvious hemolysis, with HR as high as 100%. There was no hemolysis found for DC@ZIF8-MEM_C and PBS. HR of DC@ZIF8-MEM_C was less than 2% at a concentration as high as 300 μg·mL⁻¹ (Figure 7B). Therefore, DC@ZIF8-MEM_C was highly biocompatible and could be directly administered by intravenous injection. There was no obvious decrease in body weight for all treatment groups except the DOX·HCl group (Figure 7C). Additionally, no significant histopathology changes were noticed for the main organs such as the heart, liver, spleen, lungs, and kidneys, suggesting invisible short-term toxicity during treatments (Figure 7D). These results demonstrated DC@ZIF8-MEM_C as promising for biomedicine application with high bio-safety.