Homologous and Mitochondrial Targeting Synergistic Induction of Apoptosis and Ferroptosis Enhanced PDT Performance Against Osteosarcoma HOS

Background: There have been no prominent advancements in osteosarcoma (OS) treatment in the past 20 years. Although photodynamic therapy (PDT) is an emerging technique for cancer therapy, its lack of targeting in OS treatment severely limits its applications. Methods: In this study, we constructed a potential theranostic nanoplatform by using (poly (ethylene glycol) polylactic-co-Glycolic acid (PEG-PLGA) nanoparticles (NPs) wrapping IR780 into the core (PEG-PLGA-IR780 NPs), which was further camouaged with human OS cell membranes from the HOS cell line (MH-PEG-PLGA-IR780 NPs) to show homologous and mitochondrial targeting capacities. In addition, the potential underlying anticancer mechanisms of MH-PEG-PLGA-IR780 NPs-mediated PDT was investigated. Results: We demonstrated that the MH-PEG-PLGA-IR780 NPs had excellent tumor/mitochondrial targeting with the help of homologous targeting to HOS cell line. Moreover, the excellent photoacoustic (PA)/uorescence (FL) imaging ability of MH-PEG-PLGA-IR780 NPs laid a foundation for further applications. Under near-infrared (NIR) irradiation, we demonstrated that dual-targeting NPs-mediated PDT could signicantly induce HOS cell apoptosis and ferroptosis, and further explored apoptosis was triggered by cytochrome c-activated mitochondrial apoptosis (endogenous apoptosis), and the specic molecular mechanisms of ferroptosis is the activation of NCOA4-mediated ferritinophagy and the passivation of GPX4 in vitro, synergistically leading to the excessive accumulation of ROS. In addition, MH-PEG-PLGA-IR780 NPs-induced PDT also showed an obvious inhibitory effect on tumor growth in vivo. Conclusion: These results suggest the dual-targeting-based theranostic nanoplatform provides an effective method to improve PDT performance in OS and paves a new and promising way for OS therapy. Statistical cleaved caspase-9, cleaved caspase-7 cleaved caspase-3. excessive production LPO and Lipid-ROS HOS cells after targeted PDT treatment as by CLSM and FC. scale bars morphology of mitochondria after various treatments as observed by TEM. scale bars µm. apoptosis-related analyses cytochrome

emulsi cation was emulsi ed using an ultrasonic probe (Sonics & Materials, Inc., USA) (50 W, 3 minutes). Subsequently, 4% polyvinyl alcohol (PVA) (10 mL) was added to the emulsi ed solution, which was homogenized by a second sonication to form a W/O/W double emulsion (35 W, 3 minutes). Then, 2% isopropanol solution (10 mL) was added to the prepared emulsion which was mechanically stirred for 2 h to remove the CH 2 Cl 2 followed by centrifugation (12,000 rpm for 6 minutes) to obtain the PEG-PLGA-IR780 NPs. ii) To prepare the HOS cell membranes, HOS cells were grown in T-75 culture asks to full con uence and detached with 2 mM EDTA in phosphate-buffered saline (PBS), and washed three times in PBS by centrifuging at 800 rpm for 5 minutes. Subsequently, the cells were lysed with Membrane and Cytosol Protein Extraction Kit A containing PMSF (100:1), at -80°C with repeated freeze-thaw cycles followed by centrifugation (800 rpm for 5 minutes) to collect the cellular supernatant, nally, the precipitate after centrifugation at 12,000 rpm for 30 minutes, was the HOS cell membrane (MH). iii) MH-PEG-PLGA-IR780 NPs were nally synthesized by physically extruding the mixture of PEG-PLGA-IR780 NPs and the MH at the same concentration for 11 passes through a 400 nm polycarbonate porous membrane on a mini extruder (Avanti Polar Lipids, USA). The nal product was stored at 4 °C for later use.

Characterization of the MH-PEG-PLGA-IR780 NPs
The size distribution, and zeta-potentials of various NPs and their stability of the NPs in fetal blood serum (10%) and PBS over seven days were determined using dynamic light scattering (DLS, Malvern Instruments, UK). The morphologies and structures of the NPs were observed using transmission electron microscopy (TEM, Hitachi H-7600, Japan). The optical absorption of the different NPs and free IR780 was determined with an ultraviolet-visible (UV-vis) spectrophotometer (US-2550, Shimadzu, Japan). A standard concentration curve of free IR780 measured at a wavelength of 798 nm was constructed to calculate the amount of IR780 encapsulated into the MH-PEG-PLGA NPs. The encapsulation e ciency (EE) and encapsulation content (EC) of IR780 were calculated by Equations (1) and (2) (1) EE IR780 (%) = (mass of total IR780 -mass of unentrapped IR780/mass of total IR780) ×100% (2) EC IR780 (%) = (mass of total IR780 -mass of unentrapped IR780/mass of total PEG-PLGA NPs) ×100% Cellular uptake and deep penetration capability of the MH-PEG-PLGA-IR780 NPs Laser confocal scanning microscopic (CLSM; Nikon A1+R, Japan) and Flow Cytometry (FC; BD FACSvantage SE, USA) were used to detect the cellular uptake of different NPs. Typically, HOS cells (1×10 5 /dish) were seeded into a laser confocal cell-culture dish. After 24 h of incubation, the culture medium was replaced with the serum-free medium containing PEG-PLGA NPs, PEG-PLGA-IR780 NPs or MH-PEG-PLGA-IR780 NPs (stained with DiI; λ excitation/λ emission = 549nm/565nm) for 1, 2, 3 or 4 h (IR780, 20 μg/mL). Then, the cells were xed in 4% formaldehyde for 10 minutes and washed with PBS. After incubation for different time intervals with various NPs, the cell nuclei were stained blue with DAPI (λ excitation/λ emission = 364 nm/454 nm). The fluorescence images were directly recorded by CLSM.
Moreover, the quantitative cellular uptake of PEG-PLGA NPs, PEG-PLGA-IR780 NPs and MH-PEG-PLGA-IR780 NPs at different time intervals was quanti ed and analyzed by FC.
The penetration of the MH-PEG-PLGA-IR780 NPs was estimated in vitro using 3D tumor spheroid models.
HOS tumor spheres (1×10 5 /well) were cultured in 6-well ultra-low attachment plates (Corning, Tewksbury, MA) in stem cell medium that consisted of serum-free DMEM/F12 with 20 ng/mL EGF, 20 ng/mL bFGF and B27. The medium was changed every two days over ten days. Then, the medium was replaced by the DiI-labelled NPs dispersed in DMEM/F12 (1 mL, 20 μg/mL). After 4 h of coincubation, the 3D tumor spheroids were stained with DAPI for 10 minutes, and observed using CLSM.

Mitochondrial location and macrophage uptake assay
To verify the mitochondrial location of the MH-PEG-PLGA-IR780 NPs, MitoTracker was used to the label mitochondria. Typically, HOS cells (1×10 5 /dish) were seeded into a laser confocal cell-culture dish for 24 h of incubation. Before the test, the growth medium was replaced with DiI labeled NPs (PEG-PLGA NPs, PEG-PLGA-IR780 NPs or MH-PEG-PLGA-IR780 NPs) at a concentration of 20 μg/mL for a 4 h of coincubation. Then, the cells were xed in 4% formaldehyde for 10 minutes and washed with PBS. Nest, the HOS cell mitochondria were labelled with MitoTracker (λ excitation/λ emission = 644 nm/665 nm) for 30 minutes, followed by 10 minutes of staining with DAPI. Finally, the mitochondrial localization of the NPs was observed using CLSM, and the Pearson correlation (PC) coe cients of each image were measured. This same staining method was used for 4T1, A549, MG63, 143B, K7M2 and HFOB 1.19 cells to identify the homologous targeting capabilities of the MH-PEG-PLGA-IR780 NPs.
Next, to verify the immune escape ability of the MH-PEG-PLGA-IR780 NPs, RAW 264.7 cells (1×10 5 /dish) were seeded in a laser confocal cell-culture dish. After a 24 h of incubation, the growth medium was replaced with serum-free medium containing DiI labeled NPs (20 μg/mL) for 4 h of coincubation. Subsequently, the cells were xed in 4% formaldehyde for 10 minutes and washed with PBS, YF633-Phalloidin and DAPI were used to label the cytoskeleton and nuclei of the RAW 264.7 cells within 10 minutes and 30 minutes, respectively. The uptake of various NPs by macrophages was nally observed by CLSM.
FL imaging/biodistribution and PA imaging of the MH-PEG-PLGA-IR780 NPs In vivo HOS tumor-bearing mice were intravenously injected with MH-PEG-PLGA-IR780 NPs suspension (2 mg/mL, 200 μL). Subsequently, NIR FL images were collected, pre-injection and at 1, 2, 3 and 6 h postinjection, and the relative FL intensity of each tumor region was measured by IndiGo 2.0.5.0 (Berthold Technologies, Germany). Ex vivo imaging was performed on the major organs and tumor tissues at 6 h post-injection to detect the biological distribution of the MH-PEG-PLGA-IR780 NPs.
A Vevo LAZR Photoacoustic Imaging System (VisualSonics Inc., Toronto, Canada) was used to determine the PA performance of the MH-PEG-PLGA-IR780 NPs. It has been reported that the PA signal intensity of IR780 is the greatest at the activation wavelength of 798 nm [27]. HOS tumor-bearing mice were intravenously injected with a MH-PEG-PLGA-IR780 NPs suspension (2 mg/mL, 200 μL). PA images were collected at different time points (pre-injection, 1, 2, 3 and 6 h post-injection), and the corresponding PA signal intensities were measured with a Vevo LAZR System.

Cell viability assay
A CCK-8 assay was used to determine cell viability after the different treatments. Brie y, HOS cells were seeded into 96-well plates at a density of 5,000 cells/well and incubated overnight for adherence. To determine the safety of the different NPs in vitro, HOS cells were incubated with various concentrations of NPs (NPs: 0.0, 0.2, 0.4, 0.6, 0.8, 1.0 mg/mL, and IR780: 0.0, 6.5, 13.0, 19.5, 26.0, 32.5 μg/mL) for different length of time (0, 12, 24 and 48 h). To assess the change in cell viability after pretreatment with different inhibitors (z-VAD-FMK, Nec-1, Baf-A1, DFO, Fer-1, NAC), HOS cells were pre-incubated with the aforementioned inhibitors for 24 h before exposure to MH-PEG-PLGA-IR780 NPs-associated PDT.
Following different treatments, 10 μL of CCK-8 to added to each well followed by incubation for an additional 1 h. Finally, the plates with cells were placed in a microplate reader (MK3, Thermo Scienti c) to measure the absorbance at 450 nm. Cell viability was calculated using the following formula: cell viability (%) = experimental group absorbance value/control group absorbance value×100%.

Measurement of intracellular ROS
Intracellular ROS was detected using DCFH-DA (λ excitation/λ emission = 488 nm/530 nm). Typically, HOS cells (1×10 5 /dish) were seeded into a laser confocal cell-culture dish. After a 24 h of incubation, the medium was replaced by serum-free medium with or without NPs (PEG-PLGA-IR780 NPs or MH-PEG-PLGA-IR780 NPs (IR780: 20 μg/mL)) followed by incubation for another 4 h. Then, the cells in the laser group received irradiation with an 808 nm laser at a power density of 1.5 W/cm 2 for 2 minutes. Then, the cells were incubated in serum-free medium containing 10 μM DCFH-DA in the dark at 37°C for 30 minutes. Next, the cells were washed with PBS to remove excess DCFH-DA. Finally, the cells were immediately observed by CLSM to detect the intracellular ROS levels, treated with trypsin, collected in 200 µL PBS and detected by FC.

MMP assay
The MMP (Δψm) was detected using a JC-1 assay kit. HOS cells (1×10 5 /dish) were seeded into a laser confocal cell-culture dish for a 24 h of incubation. After adherence, the HOS cells were divided into different treatment groups (control, laser alone, PEG-PLGA-IR780 NPs, laser+PEG-PLGA-IR780 NPs, MH-PEG-PLGA-IR780 NPs, and laser+MH-PEG-PLGA-IR780 NPs (IR780: 20 μg/mL)). After 4 h of intervention, all cells were cultured in fresh medium containing 1 mL of JC-1 staining solution (10 μM) and incubated for 30 minutes in the dark at 37 °C under 5% CO 2 . Then, the cells were washed with ice-cold JC-1 buffer, and JC-1 aggregates (λ excitation/λ emission = 585 nm/590 nm) and JC-1 monomers (λ excitation/λ emission = 514 nm/529 nm) were observed by CLSM. To quantify the MMP, after exposure to different treatments, the cells were trypsinized, collected in medium containing JC-1 staining solution (10 μM), and incubated in the same environment for 30 minutes. After washing with ice-cold JC-1 buffer, the cells were suspended in 200 µL of PBS for FC analysis.
TEM TEM was used to detect the changes in mitochondrial morphology after different treatments to assess ferroptosis induction. HOS cells were seeded overnight into a 6-well plate at a density of 10 5 cells/well to allow adherence and then divided into groups for various different treatments (control, laser alone, PEG-PLGA-IR780 NPs, laser+PEG-PLGA-IR780 NPs, MH-PEG-PLGA-IR780 NPs, and laser+MH-PEG-PLGA-IR780 NPs (IR780: 20 μg/mL)). Next, the cells were trypsinized, collected, and xed with 2.5% glutaraldehyde and 1% osmic acid. Then, the cells were dehydrated with a graded ethanol series and acetone, embedded, sliced, and stained with 3% uranyl acetate-lead citrate. Finally, the cells were examined by TEM (JEM-1400 Plus, JEOL, Japan).

Fe 2+ detection
FerroOrange (λ excitation/λ emission = 561 nm/570 nm) was used to detect the level of intracellular Fe 2+ . Brie y, HOS cells were seeded into a laser confocal cell-culture dish at a density of 10 5 cells/well overnight to allow adherence and then treated with laser alone, PEG-PLGA-IR780 NPs, laser+PEG-PLGA-IR780 NPs, MH-PEG-PLGA-IR780 NPs, laser+MH-PEG-PLGA-IR780 NPs (IR780: 20 μg/mL). Then, the cells were stained with 1μM FerroOrange for 30 minutes at 37°C in the dark. Next, the cells were washed with PBS to remove the excess FerroOrange. Finally, the cells were observed by CLSM to detect the intracellular Fe 2+ levels, and then treated with trypsin, collected in 200 µL of PBS and detected by FC.

Cell apoptosis assay
Apoptosis was determined by FC using annexin V-FITC/PI staining. Brie y, cells were seeded into 6-well plates (1×10 5 cells/well) and incubated overnight for adhesion. Then, the cells were exposed to laser alone, PEG-PLGA-IR780 NPs, laser+PEG-PLGA-IR780 NPs, MH-PEG-PLGA-IR780 NPs, or laser+MH-PEG-PLGA-IR780 NPs (IR780: 20 μg/mL). After treatments, the cells were collected, washed twice with ice-cold PBS, and stained with annexin V-FITC and PI according to the manufacturer's instructions. Finally, the samples were suspended in 200 µL of PBS and then analysed by FC.

Western blot
After HOS cells were treated according to the different regimens, the cells were lysed with RIPA lysis buffer containing PMSF and phosphatase inhibitor within a speci ed period of time to extract the total protein in the cells. Protein samples (30 to 50 μg/lane) were separated on a 10%-12% gel by SDS-PAGE and transferred to polyvinylidene uoride (PVDF) membranes, which were blocked with 5% skim milk for 1.5 h at room temperature, and then incubated overnight with the corresponding primary antibody (1:1000) at 4℃. Next, the membranes were washed with Tris-buffered saline with Tween-20 (TBST) and incubated with the secondary antibody (1:8000) at 37°C for 1 h. Finally, the reactive protein bands on the membrane were detected with an enhanced chemiluminescence (ECL) detection system and developed on lm.

Intracellular ATP level
An ATP assay kit was used to determine the intracellular ATP levels. Brie y, HOS cells were grown in 6well plates at a density of 10 5 cells/well and incubated overnight for adhesion. After the cells were exposed to laser alone, PEG-PLGA-IR780 NPs, laser+PEG-PLGA-IR780 NPs, MH-PEG-PLGA-IR780 NPs, laser+MH-PEG-PLGA-IR780 NPs (IR780: 20 μg/mL), the cells were trypsinized and centrifugation (12,000 rpm for 5 minutes) at 4°C. Finally, the cell supernatant was collected for RLU detection with a luminometer equipped with a multimode reader (260-Bio, Thermo Fisher Scienti c, USA).

Xenograft tumor model
Thirty male BALB/c nude mice (4 weeks old) were supplied by the Experimental Animal Center of Chongqing Medical University. All animal studies were approved by the Ethics Committee of Chongqing Medical University. The mice were housed with free access to a commercial diet and water under speci c pathogen-free conditions. After the mice were acclimated for 1 week prior to initiation of the study, HOS tumor-bearing mice were established by subcutaneous injection of 200 μL of sterile PBS containing a HOS cell suspension at a density of 10 6 cells/mL. After the tumor volumes reached 50 mm 3 , different treatments were initiated. Thirty mice were randomized into the following 6 groups: (1) control, (2) laser only, (3) PEG-PLGA-IR780 NPs, (4) laser+PEG-PLGA-IR780 NPs, (5) MH-PEG-PLGA-IR780 NPs, (6) laser+MH-PEG-PLGA-IR780 NPs (laser power: 2 W/cm 2 , 5 minutes of irradiation) (PEG-PLGA concentration: 5 mg/mL, volume of 200 µL). The tumor volume sizes and mouse weights were measured every 4 days for 16 days after treatment, with the tumor volume calculated according to the following formula: 1/2×a 2 b (where a is the short axis and b is the long axis of the tumor). Mice were sacri ced under anaesthesia on day 14, and the xenograft tumors of each animal were weighed and analyzed.
In order to avoid photothermal effect (PTT) and to ensure only effects due to the PDT treatment, the temperature of the tumor region was monitored during irradiation with a Xenogen IVIS Spectrum imaging system (PerkinElmer, USA) so that the temperature was always below 42°C. Experiments in vitro was operated on ice.

Haematoxylin-eosin (H&E) and immunohistochemistry (IHC)
After sacri ce, mouse tissues from the xenograft tumors, hearts, livers, spleens, kidneys and lungs were dissected out and xed in 10% formalin for histopathological studies. After xation, the tissues were dehydrated in a graded series of ethanol and xylene, embedded in para n, cut into sections, and stained with H&E. The expression of PCNA and Ki67 in the xenograft tumor tissues was detected by IHC. Brie y, the para n-embedded specimens were separated, xed with 4% paraformaldehyde, and embedded in para n. After embedding, the specimens were discontinuously cut into 4 mm thick sections with a microslicer. Tumor sections were blocked and immunostained with antibodies targeting Ki67 (1:200) or PCNA (1:200). Finally, images were captured using a microscope, and PCNA and Ki67 expression was evaluated by counting the number of positive cells from 5 randomly selected elds in the residual viable tumor tissue among the necrotic areas under a light microscope at a magni cation of 200×. Data are presented as the percentage of positive cells.

Biosafety of the MH-PEG-PLGA-IR780 NPs and PDT
To determine the toxicity of the MH-PEG-PLGA-IR780 NPs, they were injected into BALB/c nude mice (PEG-PLGA concentration: 5 mg/mL, volume: 200 µL) that had been randomly divided into 5 groups. Twenty-ve mice were euthanized on days 0, 1, 7, 14, and 28 post-injection. The vital organs were collected for H&E staining, and blood samples were sent for the blood index (routine blood and biochemistry) analyses. In addition, the methods of determining the biosafety of MH-PEG-PLGA-IR780 NPs-mediated PDT and the other parallel groups were consistent with the aforementioned methods after the mice were sacri ced at 16 days post-irradiation.

Statistical analysis
All data are expressed as mean ± SD and were analyzed with SPSS 22.0 software. Single Student's t-test and one-way ANOVA were used to determine statistical signi cance between pairs of groups or three or more groups, respectively. Signi cance levels are shown as *p < 0.05, **p < 0.01, and ***p < 0.001.

Results
Preparation and characterization of the MH-PEG-PLGA-IR780 NPs MH-PEG-PLGA-IR780 NPs were prepared using a double-emulsion approach and the physical extrusion method (Scheme 1). The morphology of the obtained NPs was observed by TEM ( Figure 1A-B). We found that all of the NPs were uniform in size and monodispersed, and there was a lipid shell on the outside of the NPs wrapped by cell membrane. We further analyze the overall protein components in both the MH and MH-PEG-PLGA NPs by sodium dodecyl sulfate-polyacrylamide gel electrophoresis (SDS-PAGE). As shown in Figure 1C, the expression levels of proteins in the two groups were comparable, which further con rmed the successful encapsulation of the HOS cell membrane. The particle sizes and zeta potentials of these NPs were measured by DLS, and the average size of MH-PEG-PLGA-IR780 NPs was distributed at approximately 236.8 nm ( Figure 1D), which is consistent with the size obtained by TEM. DLS also showed that the average sizes of the PEG-PLGA NPs and PEG-PLGA-IR780 NPs were narrowly distributed and centered at 202.1 nm and 218.2 nm, respectively ( Figure S2), suggesting that the resulting average sizes of these NPs would allow them to pass through the tumor endothelial space [28]. Next, to examine the stability of the synthesized NPs, they were placed in either DMEM (10% serum) or PBS at 4°C for one week. Their mean size was measured every day did not change signi cantly within 7 days ( Figure S3), suggesting that the obtained NPs had good stability under physiological conditions. In addition, the zeta potential of the PEG-PLGA NPs and PEG-PLGA-IR780 NPs were -4.62±0.79 mV and -6.70±0.50 mV respectively, while that of the MH-PEG-PLGA-IR780 NPs decreased to -10.09±0.70 mV ( Figure 1E), which was possibly caused by the negative charge from cell membrane and the minimization of the electrostatic repulsion of IR780. The negative zeta potential of the MH-PEG-PLGA-IR780 NPs is bene cial for accurate tumor targeting due to a decrease in rapid elimination by the reticuloendothelial system (RES) and prolongation of the time in systemic circulation [29]. The color of the PEG-PLGA NPs and NPscoated IR780 changed from white to green ( Figure 1F). UV-vis spectrum showed that the PEG-PLGA-IR780 NPs and MH-PEG-PLGA-IR780 NPs had a characteristic absorption peak from IR780 at 798 nm, further indicating the successful loading of IR780 compared with the PEG-PLGA NPs, which did not show this characteristic absorption peak ( Figure 1G). The absorbance of IR780 changed in a concentrationdependent manner, as determined by UV-vis spectrum, and a calibration curve was constructed ( Figures  1H, S4). Finally, the EE and LC of IR780 were calculated to be 67.8% and 3.25 wt%, respectively.  Figure 2A-B, the PEG-PLGA NPs had no toxicity to HOS cell, whereas the NPs-wrapped IR780 exhibited mild cytotoxicity at a high NPs concentration (PEG-PLGA ≥1.0 mg/mL) which was caused by the weak cytotoxicity of IR780 [30]. Additionally, the MH-PEG-PLGA-IR780 NPs (PEG-PLGA: 4 mg/mL, 200 μL) were intravenously injected into BALB/c nude mice, and blood samples and the major organs (heart, liver, spleen, lung and kidney) were collected after 0, 1, 7, 15, and 28 days post-injection. The routine blood examination and serum biochemical index results showed no signi cant differences between the sample, and no obvious histopathological changes were observed in the aforementioned vital organs after treatment with the MH-PEG-PLGA-IR780 NPs, indicating acceptable biocompatibility and biosafety in vivo ( Figure 2C-D).

Intracellular uptake and deep penetration of the MH-PEG-PLGA-IR780 NPs
To determine whether the MH-PEG-PLGA-IR780 NPs possessed high a nity for HOS cells, the cellular uptake e ciency of the MH-PEG-PLGA-IR780 NPs (labelled with DiI) was explored by CLSM and FC. As shown in Figures 3A, S5, compared with no obvious change of red FL in PEG-PLGA NPs group and a mild increase of red FL in PEG-PLGA-IR780 NPs group, the red FL in the cells coincubated with MH-PEG-PLGA-IR780 NPs increased signi cantly with the prolonged incubation time, reaching greater than 90% uptake after 4 h of coincubation by the quantitative analyses of FC, indicating that encapsulating the PEG-PLGA-IR780 NPs with the tumor cell membrane could solve the problem of poor a nity of the PEG-PLGA-IR780 NPs for HOS cells. Moreover, hypertension of the tumor sites caused by tumor vascular heterogeneity and the increase in the interstitial pressure further limits the penetration depth and width of NPs in the tumor region [31]. It has been reported that IR780 contained within NPs can achieve surface-to-core penetration of tumor cells [32], but PEG-PLGA-IR780 NPs failed to reveal the anticipated penetration capability. Thus, the following experiments were carried out to evaluate whether the tumor cell membrane was desirable to improve the deep penetration ability of the PEG-PLGA-IR780 NPs, which was implemented by using 3D tumor sphere models to simulate the complex conditions of tumor sites. We cultured 3D tumor spheroids in vitro to examine the penetration capability of the MH-PEG-PLGA-IR780 NPs. As shown in Figure 3B, CLSM observations showed that the MH-PEG-PLGA-IR780 NPs (labelled with DiI) could penetrate into the centre of the 3D tumor spheres for uniform distribution throughout the tumor cells, suggesting that the superior penetration depth and width of IR780 in HOS 3D tumor spheroids can be achieved by the homologous targeting from tumor cell membrane.
Homologous targeting facilitated mitochondrial targeting Due to tumor heterogeneity, the expression levels of surface antigens in the cell membranes responsible for multicellular aggregation formation in tumors are relatively diversi ed. MH-PEG-PLGA-IR780 NPs exhibit a homologous targeting ability owing to their functionalization by adhesion proteins from the HOS cell membrane. In addition, the homologous targeting of the MH-PEG-PLGA-IR780 NPs to HOS cell line should also have a potential capability of mitochondrial targeting. As shown in Figure 4A, CLSM images showed that the red FL (MH-PEG-PLGA-IR780 NPs) overlapped with the green FL (mitochondria) stained by MitoTracker. The corresponding PC coe cient was determined to be 0.951 for the MH-PEG-PLGA-IR780 NPs, while an unsatis ed co-localization was observed in PLGA-IR780 NPs and PEG-PLGA-IR780 NPs groups, whose PC coe cients were calculated to be 0.231 and 0.303, respectively, suggesting that the MH-PEG-PLGA-IR780 NPs had a high a nity for the mitochondria in HOS cells, which would potentially promote PDT performance because mitochondria are susceptible to ROS-mediated cytotoxicity.
Next, we wanted to verify whether the mitochondrial targeting capacity of the MH-PEG-PLGA-IR780 NPs to HOS cell line was achieved by its homologous targeting. It has been reported that 4T1 and A549 cells are sensitive to the mitochondria targeting ability of NPs wrapping IR780 [3,33], so both of these cells were with the MH-PEG-PLGA-IR780 NPs for 4 h followed by CLSM imaging, and only a suboptimal overlap (PC = 0.426) was found between the green FL and red FL in 143B cells, which may be due to the 143B cell was also known as a HOS cell line with k-ras oncogene transformation [34]. In vivo biodistribution by FL and PA Imaging IR780 has been used as a uorescent probe to impart the MH-PEG-PLGA-IR780 NPs with NIR FL (λ excitation/λ emission = 745 nm/820 nm) and PA (wavelength = 780nm) imaging capabilities to evaluate the distribution of tumor accumulation and the biodistribution of MH-PEG-PLGA-IR780 NPs in vivo [27].
MH-PEG-PLGA-IR780 NPs were intravenously injected into HOS tumor-bearing mice to detect the distribution of the MH-PEG-PLGA-IR780 NPs in vivo after prolonged periods of times. As shown in Figures   5A, 5C, the FL signal began to aggregate within the tumor region in a time-dependent manner and reached a peak at 6 h post-injection. Thus, FL intensities of the harvested vital organs and tumors were measured after 6 h post-injection for ex vivo FL imaging to con rm the NPs biodistribution, and it was found that the FL intensities in tumors were higher than those of the heart, liver, spleen and kidney, but lower than that in the lung [35] (Figures 5B, 5D), suggesting that the superior targeting properties of MH-PEG-PLGA-IR780 NPs in vivo.
In biological tissues, PA imaging has superior contrast, resolution and penetration, enabling the integration of diagnosis and treatment by determining the precise location of the drugs in the targeted region of the tumor [36]. Therefore, PA imaging was recorded to further clarify the distribution of the MH-PEG-PLGA-IR780 NPs in vivo. Generally, an obvious PA signal intensity began to focus on the tumor region at 1 h post-injection, reached a peak at 6 h post-injection ( Figure 5E-F). These results were consistent with those from FL imaging. In conclusion, both the results of FL and PA imaging veri ed that the MH-PEG-PLGA-IR780 NPs may effectively accumulate in tumor region in vivo, laying a foundation for further tumor diagnostic imaging and targeting therapy.
Homologous and mitochondrial targeting enhanced PDT performance After tumor/mitochondrial targeting was achieved by homologous targeting, a CCK-8 assay was used to estimate the antitumor e cacy of PDT in vitro. As shown in Figure 6A, compared to no obvious change in the control, laser alone, and single NPs without NIR irradiation groups, the MH-PEG-PLGA-IR780 NPs exposed to laser (targeted PDT group) decreased the cell viability of HOS in a IR780 dose-dependent manner at power densities of 1 W/cm 2 and 1.5 W/cm 2 , while laser+PEG-PLGA-IR780 NPs (PDT without targeting) showed weak photoinduced cytotoxicity on HOS cells. Considering that the use of IR780 alone at a concentration greater than 30 μg/mL had mild cytotoxic effects on HOS cells after 4 h of coincubation, 20 μg/mL MH-PEG-PLGA-IR780 NPs was selected to examine the e cacy and speci c mechanism of MH-PEG-PLGA-IR780 NPs-mediated PDT in the following experiments.
First, the morphology of the mitochondria was observed by CLSM. As shown in Figure 6B, compared with the physiological morphology of mitochondria in the other ve groups, mitochondrial fragmentation was found in HOS cells upon treatment with the targeted PDT, which could further lead to mitochondrial dysfunction. In addition, PDT performance was detected by DCFH-DA, a probe that detects intracellular ROS generation by showing green FL. As shown in Figure 6C, compared with no signi cant alteration in the control, laser alone and two single NPs groups, PDT without targeting showed mild green FL, while the targeted PDT produced strong green FL in HOS cells under 808 nm NIR irradiation (1.5 W/cm 2 ), which was accompanied by typical morphological features of apoptotic cell, indicating the initiation of cell death. Moreover, we quantitatively analyzed the percentage of intracellular ROS in cells by FC, and compared with the parallel groups, the amount of ROS produced by PDT without homologous and mitochondrial targeting was 44.60%, while this value after treatment with targeted PDT was signi cantly increased to 98.97% ( Figure 6D). ATP, one of the most important energy supply substances, is mainly derived from mitochondrial metabolism [37], and plays a key role in tumor proliferation and DNA replication. It has been reported that ATP depletion causes tumor cells to be more sensitive to PDT [38].
To demonstrate the changes in ATP contents after different treatments, a calibration curve of ATP was constructed ( Figure S7). And as shown in Figure 6F, compared with the other ve groups, only PDT with homologous and mitochondrial targeting treatment led to a signi cant decrease in the intracellular ATP content. In conclusion, these results suggested that homologous and mitochondrial targeting could improve PDT performance by increasing intracellular ROS production and inhibiting ATP synthesis.

Homologous and mitochondrial targeting synergistically induced apoptosis and ferroptosis.
Cell death is a fundamental biological process generally including necrosis, apoptosis, autophagy and ferroptosis. The synergistic induction of multiple death modes in combination with tumor therapy is an effective strategy to improve PDT performance [18]. We next set out to investigate the main cell death modes of the dual-targeting nanoplatform-mediated PDT. Therefore, before NIR irradiation, HOS cells were pretreated with multiple inhibitors of cell death pathways for 24 h, including an apoptosis inhibitor (z-VAD-FMK), a necrosis inhibitor (Nec-1), an autophagy inhibitor (Baf-A1), a ferroptosis inhibitor (Fer-1) and a general ROS scavenger (NAC). As displayed in Figure S8, the CCK-8 assay showed that compared with the signi cant inhibitory effect of cell viability by targeted PDT, the viability of HOS cells increased after pretreatment with each inhibitors (excepting Baf-1) followed targeted PDT treatment to different degrees, but NAC, Fer-1 and z-VAD-FMK signi cantly protected HOS cells from photoinduced cytotoxicity caused by PDT, while Baf-1 promoted the inhibitory effects of PDT on HOS cell viability, suggesting that in addition to apoptosis, ferroptosis was another key cell death mode mediated by MH-PEG-PLGA-IR780 NPs-induced PDT, and autophagy acted as a presurgical role in PDT treatment.
Therefore, we hypothesized that mitochondrial dysfunction accomplished by homologous targeting by targeted PDT treatment was desirable to synergistically induce mitochondrial apoptosis (endogenous apoptosis) and ferroptosis to kill HOS cells. Then, we sought to further verify this assumption and explore underlying cell death mechanism. First, we measured cell apoptosis by FC using annexin V-FITC/PI double staining analysis. As shown in Figure 7A-B, the control, NIR irradiation alone, PEG-PLGA-IR780 NPs and MH-PEG-PLGA-IR780 NPs alone without irradiation groups did not induce obvious apoptosis in HOS cells. The apoptosis rate of HOS cells induced by PDT without targeting was 25.05%, while the targeted PDT treatment increased the cell apoptosis rate to 64.57%. Then, considering the important role of MMP (Δψm) in mitochondrial apoptosis, we measured the Δψm with the uorescent probe JC-1. During the mitochondrial apoptotic process, the mitochondrial membranes were disrupted with the depolarization of the Δψm, and the mitochondrial depolarization was veri ed by the decrease in the red to green FL intensity ratio in JC-1 assay. As shown in Figure 7C-E, HOS cells were stained with JC-1 after different treatments. Compared with no obvious change in the control, single NIR irradiation, and two NPs alone groups, we observed that laser+PEG-PLGA-IR780 NPs treatment exhibited a mild decrease in red FL and a weak increase in green FL, and these changes were more prominent after treatment with the laser+MH-PEG-PLGA-IR780 NPs. Furthermore, to further investigate the underlying mechanisms of the pro-apoptotic effects of the nanoplatform-mediated PDT, the expression levels of apoptosis-related proteins were assessed by western blot. It was found that laser+MH-PEG-PLGA-IR780 NPs treatment signi cantly increased the expression of cytochrome c, cleaved caspase-7, cleaved caspase-9, pro-apoptotic protein Bax, which was accompanied by obvious downregulation of the anti-apoptotic protein Bcl-2 which ultimately led to the marked upregulation of cleaved caspase-3. However, the expression levels of these apoptosis-related proteins revealed no prominent change in the control, single NIR irradiation, two NPs alone without irradiation groups, and PDT without targeting group, which produced unstable proapoptotic effects ( Figure 7F-L).
Intriguingly, the dual-targeting (MH-PEG-PLGA-IR780 NPs) constructed in our study have the innate advantage to induce ferroptosis under NIR irradiation. On the one hand, the increase in intracellular ROS and the large number of polyunsaturated fatty acids (PUFAs) in the cell membrane were found to be the key drivers of ferroptosis [39], which are susceptible to oxidation to form LPOs. On the other hand, mitochondrial targeting-mediated ferroptosis has been reported to be an emerging strategy for cancer therapy [40]. To further examine the level of ferroptosis, LPOs and Lipid-ROS have been recognized as crucial biomarkers of ferroptosis to impair cell structure and integrity, and both of which were detected by liper uo and C11-BODIPY, respectively [41]. As shown in Figure 7M-N, CLSM images showed that the contents of both LPOs and lipid-ROS of HOS cells were signi cantly increased, as indicated by the signi cant increase in green FL intensity, after laser+MH-PEG-PLGA-IR780 NPs treatment compared with no remarkable alteration in the control, laser alone and single two NPs without irradiation groups. In addition, the slight green FL was observed in laser+PEG-PLGA-IR780 NPs treatment which was consistent with the mild cell apoptosis induced by this treatment group. Moreover, quantitative analysis via FC illustrated the same phenomenon ( Figure 7O-R), indicating that this dual-targeting-based theranostic nanoplatform had superior advantages on inducing ferroptosis under NIR irradiation, and the inadequate ability of PEG-PLGA-IR780 NPs combined with NIR irradiation to induce ferroptosis may be due to the lack of mitochondrial targeting and PUFAs. Mitochondria shrinkage is another important character of death phenotype in ferroptosis [42,43]. The results of TEM also demonstrated that the morphology of mitochondria treated with targeted PDT became round, dwindled in size, reduced or disappeared mitochondrial cristae, and exhibited structural damage, while mitochondria in the other ve groups showed a normal physiological morphology ( Figure 7S). Therefore, we con rmed that the emerging nanoplatform-mediated PDT could signi cantly synergistically induce HOS cell apoptosis and ferroptosis by homologous and mitochondrial targeting.
MH-PEG-PLGA-IR780 NPs-mediated PDT promoted ferroptosis by inactivating GPX4 and accumulating Fe 2+ by ferritinophagy Three of the major factors that affect ferroptosis are tailored lipid metabolism, redox-active iron (Fe 2+ ) and the inactivation of GPX4, a lipid repair enzyme that is responsible for preventing LPO cytotoxicity and maintaining membrane lipid bilayer homeostasis [44]. To further investigate the speci c mechanism of ferroptosis induced by MH-PEG-PLGA-IR780 NPs-mediated PDT, we rst detected the protein expression of GPX4 by western blot, as shown in Figure 8A-D, GPX4 expression was signi cantly downregulated by the laser+MH-PEG-PLGA-IR780 NPs treatment, while the other ve treatments did not achieve the identical effect. SLC7A11 and SLC3A2 are subunits of system X C -, a glutamate/cysteine antiporter, which are responsible for maintaining the cellular antioxidant environment, and the low expression of both of these subunits would induce the passivation of GPX4, preventing ROS production during ferroptosiss [45].
Unsurprisingly, compared with the parallel treatment groups, PDT without targeting showed no in uence on either SLC7A11 or SLC3A2 protein expression. However, SLC3A2 protein expression was signi cantly downregulated by the laser+MH-PEG-PLGA-IR780 NPs treatment, and the protein expression of SLC7A11 was only mildly decreased in this group. In conclusion, the downregulation of both SLC3A2 and SLC7A11 resulted in the deactivation of GPX4.
In addition, excessive accumulation of redox-active iron (Fe 2+ ) in cells promotes the production of excessive hydroxyl radicals because of the Fenton reaction (Fe 2+ + H 2 O 2 → Fe 3+ + (OH)-+ ·OH). Hydroxyl radicals (·OH), another ROS, are able to oxidize PUFAs, producing LPOs. Thus, we measured the intracellular Fe 2+ generation capability of dual-targeting nanoplatform-associated PDT by FerroOrange (Japan, colleagues). As shown in Figure 8E, CLSM analysis showed that compared with the parallel groups, the treatment of targeted PDT signi cantly increased the intracellular Fe 2+ levels, as shown by the strong red FL intensity. The quantitative analysis of FC also con rmed that the percentage of intracellular Fe 2+ by the targeted PDT treatment raised to 65.11%, which was consistent with the above ndings level was associated with NCOA4-mediated ferritinophagy, the protein expression levels of NCOA4, FTH and FTL were determined by western blot. As shown in Figure 8H-K, accompanied by no obvious changes in the control, laser alone, and NPs without irradiation groups, PDT without targeting treatment showed a mild increase in NCOA4 expression and a weak reduction in FTH expression, while the expression of NCOA4 was signi cantly elevated and the expression of FTH and FTL were remarkably declined in the laser+MH-PEG-PLGA-IR780 NPs group. In addition, the levels of FTH and FTL expression were further con rmed by immuno uorescence (IF) assay. Generally, consistent with western blot results, the laser+MH-PEG-PLGA-IR780 NPs showed signi cant green FL intensity from NCOA4, which was brighter than that in the other ve groups, and an arresting feeble red FL intensities of FTH and FTL was discovered in laser+MH-PEG-PLGA-IR780 NPs treatment which were remarkably lower than those of the control, single NIR irradiation, and two NPs without irradiation groups, while PDT without targeting showed limited changes in the FL intensities of NCOA4, FTH and FTL ( Figure 8L). In conclusion, we demonstrated that this dual-targeting-associated theranostic nanoplatform can not only decrease the activity of GPX4 by the inhibiting system Xctransport, but also promote the accumulation of Fe 2+ by activating NCOA4-mediated ferritinophagy to degrade ferritin, synergistically inducing ferroptosis in HOS cells. imaging results. After different therapies, the representative images of the mice before sacri ce and tumor visualization were captured, and tumor volumes were measured every 4 days over 16 days to monitor different treatment outcomes ( Figure 9A-B, S9), we found the tumor growth in the PBS and laser alone groups were increased to closely 6.82-fold and 6.13-fold compared with the original tumor volumes, and the tumor volumes of single two NPs groups presented a 5.25-fold and 4.78-fold increase, respectively, and laser+PEG-PLGA-IR780 NPs group inhibited tumor growth suboptimally due to its lack of targeting (3.07-fold increase). In contrast, laser+MH-PEG-PLGA-IR780 NPs group almost completely inhibited the growth of tumors, demonstrating a signi cant tumor inhibition effect in vivo, which is consistent with the photoinduced cytotoxicity in vitro. Subsequently, the trend in tumor size and weight ex vivo were measured and shown in Figure 9C-D. No obvious change in mice weights between the control and treated groups were observed during the experimental period ( Figure S10).

Antitumor e cacy in vivo
Furthermore, we also evaluated the expression of proliferation markers, including PCNA and Ki67, to further validate the inhibitory effects of different treatments on tumor growth. The expression levels of PCNA and Ki67 in tumor specimens measured by IHC from the laser+PEG-PLGA-IR780 NPs group were lower than those in the control, laser alone, and single two NPs groups, while the laser+MH-PEG-PLGA-IR780 NPs group showed the lowest expression. Specimens from the xenograft tumors were stained with H&E after mouse sacri ce, as tumor necrosis (karyopyknosis, karyorrhexis, and karyolysis) is a crucial criterion for evaluating the response to various treatments in HOS cells [49]. We found that the specimens from the two PDT groups showed different degrees of necrosis, while the laser+MH-PEG-PLGA-IR780 NPs group were associated with the most severe necrosis (Figures 9E, S11). To detect the therapeutic biosafety of PDT in vivo, H&E staining of the main organs (heart, liver, spleen, lung and kidney) was performed at the end of the various treatments, and routine blood and biochemistry analysis also showed no signi cant differences among the various groups ( Figure S12). In addition, no apparent histopathological abnormalities were observed via H&E staining ( Figure S13). Taken together, these ndings indicated that MH-PEG-PLGA-IR780 NPs could enhance PDT performance in a xenograft model with high therapeutic potential and biosafety.

Ferroptosis-promoted PDT performance
As a new mode of RCD, ferroptosis is desirable to improve the production of free radicals via the Fenton reaction, resulting in an increase in ROS-mediated oxidative damage to cellular constituents and ensuring the high e ciency of PDT. To further determine whether ferroptosis induced by targeted PDT treatment could enhance the original killing characteristics of PDT in vivo, we administered PDT targeted therapy to HOS xenograft tumor-bearing mice after a combination injection of MH-PEG-PLGA-IR780 NPs and Deferoxamine (DFO), a ferroptosis inhibitor that functions by chelating iron. The tumor size was measured every 4 days over 16 days in the above treatment groups, and it was found that compared with the signi cant inhibitory effects of the targeted PDT treatment, DFO weakened tumor growth in the xenograft model by inhibiting ferroptosis. Visualization and the weights of the representative tumors ex vivo followed a similar trend, with no signi cant alterations in body weight ( Figure 10A-D). These results indicated that ferroptosis activation allowed the new nanoplatform-mediated PDT to be capable of promoting the antitumor effects of PDT in vivo.
In vitro, we further clari ed whether the inhibition of ferroptosis could protect HOS cells from the photoinduced cytotoxic effects of the targeted PDT treatment. HOS cells were pre-incubated with DFO for 24 h followed by targeted PDT treatment. Since this nanoplatform-mediated PDT induced ferroptosis was also dependent on tailored lipid peroxidation, HOS cells were pretreated with another ferroptosis inhibitor Fer-1 for 24 h before targeted PDT intervention. As shown in Figures 10E-M showing weaker green FL intensity by CLSM, which is in accordance with the results obtained by FC.
Considering that DFO inhibited ferroptosis by depleting intracellular Fe 2+ and indirectly blocking lipid peroxidation, the content of intracellular Fe 2+ was detected by FerroOrange, and as shown in Figure S14, both CLSM images and FC analyses showed the same conclusion. Compared with the laser+MH-PEG-PLGA-IR780 NPs group, DFO signi cantly inhibited the production of intracellular Fe 2+ during PDT. In summary, these results indicated that effective blockade of ferroptosis might protect HOS cells from the cytotoxic effects of intracellular ROS, Lipid-ROS and LPOs induced by targeted PDT. Additionally, the CCK-Page 20/38 8 assay showed that the pretreatment of Fer-1 or DFO followed targeted PDT could effectively increase the cell viability of HOS ( Figure 10N). Moreover, to further investigate the relationship and the potential signaling pathways between the targeted PDT-induced apoptosis and ferroptosis, as shown in Figure   10O-W, both Fer-1 and DFO attenuated targeted PDT-mediated apoptosis by inhibiting Cytochrome c and Bax protein levels, which is consistent with the upregulation of Bcl-2 protein expression, ultimately resulting in a decreased protein level of cleaved caspase-3. Finally, annexin V-FITC/PI double staining analysis showed the cells pretreated with Fer-1 or DFO followed by targeted PDT treatment exhibited marked inhibition of the total apoptosis rate of HOS cells compared with targeted PDT without ferroptosis inhibitors ( Figure 10X-Y). Taken together, these results indicated that the induction of ferroptosis by the theranostic nanoplatform-mediated PDT could contribute to the PDT-mediated apoptosis and inhibit tumor growth in mouse xenografts.

Discussion
OS is the most common malignant bone tumor in children and adolescents. As an emerging treatment showing noninvasiveness, high selectivity and few side effects, PDT has been used in the clinical treatment of super cial tumors such as those in skin cancer [50]. Studies have also reported that PLGAcoated photosensitizers can enhance PDT performance to different cancers, such as breast cancer and prostate cancer [27,38,51]. However, few studies have been conducted to examine the combination of nano-scale drug-delivery systems and PDT in OS treatment. The lack of PDT targeting to OS tumors and the single death mode featuring unstable therapeutic effects severely limits the PDT e ciency. As a potential method, ligand-receptor binding may solve the problem of poor tumor targeting caused by immune rejection of the mononuclear phagocyte system (MPS) [52], but no speci c and effective receptor has been found for clinical treatment of the OS [4,53]. In this study, tumor cell membranes were used to endow NPs with homologous targeting and immune escape abilities, which showed tumor targeting in vivo and vitro to OS through ligand receptors on the surface of the cell membranes.
Modi cation of NPs using bio lms such as erythrocyte and macrophage membranes could endow NPs with some of the characteristics of the original cells [54,55]; for example, coating NPs with red blood cell membranes could their extend blood circulation time [56]. However, compared with tumor cell membranes, positive tumor targeting cannot be implemented due to the lack of targeting molecules on the surface of other types of cell membranes [13,14,57].
In addition, mitochondrial dysfunction plays a key role in the regulation of tumor cell death. (i) The mitochondria of tumor cells have a higher membrane potential (Δψm) and produce less ATP, which makes the mitochondria of tumor cells more sensitive to mitochondrial targeted drugs [58]. (ii) Mitochondrial dysfunction can cause depolarization of the Δψm,promote the release of cytochrome c and activate the endogenous apoptosis pathway mediated by cleaved caspase-9, cleaved caspase-7 and cleaved caspase-3 [49,59]. (iii) ROS act as a "double-edged sword" to cause damage to proteins, lipids and oxidative phosphatases [60], which are produced by PDT and mainly come from mitochondria. As a photosensitive agent, IR780 can target tumor mitochondria due to the activation of organic aniontransporting polypeptides (OATPs) in tumor cells, which contribute to the transmembrane transport of IR780 during tumor phagocytosis, and nally allow diffusion of the NPs into the tumor core [9,61,62]. Therefore, we constructed the tumor/mitochondrial targeting NPs vector MH-PEG-PLGA-IR780 by combining homologous targeting capacity from the HOS cell membrane and mitochondrial targeting capacity from IR780. According to the obtained CLSM and FC results, MH-PEG-PLGA-IR780 NPs had superior tumor/mitochondrial targeting to HOS cells and excellent penetration depth into 3D tumor spheroids, laying a foundation for improving the performance of PDT. However, there are a large number of ligands and receptors on the surface of the tumor cell membranes which might cause biosafety risks in vivo, so we performed routine blood, biochemistry analysis and H&E staining of vital organs (heart, liver, spleen, lung and kidney) at 1, 7, 14, and 28 days after the injection of MH-PEG-PLGA-IR780 NPs in BALB/c nude mice. The results showed that the MH-PEG-PLGA-IR780 NPs had no toxicity or side effects in vivo, corroborating the results from the CCK-8 assay in vitro, suggesting that the MH-PEG-PLGA-IR780 NPs not only had excellent targeting properties to HOS cells but also showed perfect biosafety in vivo and in vitro.
Inducing a single cell death mode may fail to eradicate tumor cells due to the complex tumor microenvironment and tumor heterogeneity [16]. Fortunately, the increase in intracellular ROS caused by dual-targeting theranostic nanoplatform-mediated PDT resulted in a close relationship between cell apoptosis and ferroptosis. Our results demonstrated that MH-PEG-PLGA-IR780 NPs-mediated PDT could signi cantly induce HOS cells apoptosis and ferroptosis, and the speci c mechanism was further explored. The induction of apoptosis involves two classical pathways: the death receptor pathway (extrinsic pathway) and the mitochondrial pathway (intrinsic pathway) [63]. In the mitochondrial apoptotic pathway, a decrease in the Δψm leads to an increase in mitochondrial membrane permeability, which further results in the release of cytochrome c from the mitochondria into the cytoplasm. In the cytoplasm, cytochrome c initiates the activation of caspases and eventually induces cell apoptosis. The release of cytochrome c is prevented by the anti-apoptotic protein Bcl-2. Furthermore, we observed a decrease in Bcl-2 expression accompanied by an increase in the expression of the pro-apoptotic protein Bax, cytochrome c, cleaved caspase-9, cleaved caspase-7 and cleaved caspase-3 by western blot. These results demonstrated that it is the activation of the mitochondrial apoptotic pathway that resulted in the cell apoptosis by the dual-targeting theranostic nanoplatform-mediated PDT. In addition, the nanoplatform-mediated PDT induced ferroptosis mainly resulted from two molecular mechanism. On the one hand, the excessive accumulation of intracellular Fe 2+ derived from the activation of NCOA4mediated ferritinophagy and the degradation of FTH protein and FTL protein. On the other hand, the inactivation of GPX4 by inhibiting the X C -system (SLC7A11/ SLC3A2), indicating a compromised cellular antioxidant defense system. Finally, both of these aspects led to the excessive accumulation of intracellular LPOs and lipid-ROS to impair cell structure and integrity. Due to the distinct differences between ferroptosis and apoptosis, compared with the single cell death induction, the simultaneous induction of apoptosis and ferroptosis was expected to overcome the problems of tumor heterogeneity and improve PDT performance. We also demonstrated MH-PEG-PLGA-IR780 NPs-mediated PDT could signi cantly inhibit tumor growth in mice with HOS xenograft tumors without toxicity or side effects to the vital organs, which is consistent with the prominent photoinduced cytotoxicity in vitro.
Fortunately, the excessive accumulation of redox-active iron is attributed to ferroptosis induction, which has the unique advantage of producing oxygen free radicals to induce cell apoptosis to kill tumors through the Fenton reaction in turn [26,64]. Fe 2+ and Fe 3+ are the main oxidation-reduction states of redox-active free iron, and the former is more important in the redox environment in tumor cells [65,66].
CLSM images and FC analyses demonstrated that MH-PEG-PLGA-IR780 NPs-mediated PDT increased the level of intracellular Fe 2+ . To further investigate the promotional or antagonistic role of ferroptosis in mediating apoptosis in HOS cells after the treatment with the nanoplatform-mediated PDT, we assessed cell viability and apoptosis after blocking ferroptosis with Fer-1 and DFO. Our results showed after pretreatment with ferroptosis inhibitors Fer-1 and DFO in HOS cells followed targeted PDT, the cell apoptosis rate and cell viability were signi cantly inhibited. In addition, DFO and MH-PEG-PLGA-IR780 NPs were codelivered to HOS tumor-bearing mice followed NIR irradiation, and the inhibition of ferroptosis via iron chelation signi cantly reduced the performance of MH-PEG-PLGA-IR780 NPsmediated PDT in suppressing tumor growth compared to the group without DFO treatment, these results suggested that ferroptosis played a key role as a "death switch" in dual-targeting theranostic nanoplatform-mediated PDT.

Conclusion
In summary, we successfully constructed an emerging nanoplatform (MH-PEG-PLGA-IR780 NPs) with coupled tumor targeting with mitochondrial targeting, and combined this with an NIR irradiation to synergistically induce apoptosis and ferroptosis and improve PDT e ciency in vitro and in vivo. Because of the superior FL/PA imaging performance, this nanoplatform is desirable for precise tumor diagnosis and therapy. In addition, our results suggest that the underlying mechanisms of apoptosis and ferroptosis are initiated by the mitochondrial apoptotic pathway, excessive accumulation of Fe 2+ and inactivation of GPX4. Finally, blocking ferroptosis inhibited PDT-induced apoptosis by decreasing intracellular ROS and inhibiting the mitochondrial apoptotic pathway, suggesting that ferroptosis could promote the e ciency of PDT in a dual-targeting-associated theranostic nanoplatform.       (The data are presented as the mean ± SD, n = 3, *p < 0.05, **p < 0.01). F) The content of ATP in HOS cells after different treatments (The data are presented as the mean ± SD, n = 3, ***p < 0.001).