A Novel NIR-II Theranostic Agent: A Win–win Strategy of Tracing and Blocking Tumor-associated Vessels for Oral Squamous Cell Carcinoma

Tumor vessel co-option is a crucial predictor for tumor invasiveness but is easy to ignore in tumor vascular targeting therapy. A high density of tumor vessel co-option located alongside the tumor frontier predicted a high tendency of tumor invasion and regional metastasis. Herein, a novel NIR-II nanoagent (CS NPs) was constructed with an organic COi8DFIC dye and sorafenib, which demonstrated high tissue penetration and low tissue autouorescence, enabling imaging of vessel co-option and tumor micrometastasis. This nanotherapeutic agent displayed considerably improved quantum yield of uorescence (0.89%), high ROS generation and fairly good biosafety in vivo. Compared with indocyanine green (ICG), CS NPs exhibited better photostability and photothermal conversion eciency. By tracking tumor-associated vessels and real-time tumor imaging, CS NPs could open/stop vascular targeting therapy by laser on/off. The combination of vessel disruption and imaging-guided photothermal therapy/photodynamic therapy provided a win–win strategy for oral squamous cell carcinoma (OSCC).


Introduction
Oral squamous cell carcinoma (OSCC) accounts for approximately 3% of cancers worldwide and is ranked as the 8th most fatal cancer 1,2 . To date, surgery remains the major treatment for OSCC patients. However, surgery alone will lead to substantial esthetic and functional damage, which will greatly traumatize the patients 3,4 . Thus, it is vital to acquire an early and accurate diagnosis and improve the therapeutic outcome for OSCC patients. Solid tumor growth largely depends on blood vessels 5 , which is a profound approach for nutrient transport and cellular metabolism 6 . It is well known that a corresponding supply of the tumor vascular system will be required as the tumor grows larger than 2 mm 3 7, 8 . Vessels promote tumor growth and are closely related to malignant biological behaviors such as tumor recurrence, metastasis, and drug resistance 8, 9 . Moreover, it has been demonstrated that patients with a high density of microvessels in OSCC have a higher degree of malignancy, a higher ratio of lymph node metastasis and a worse prognosis [10][11][12] . Therefore, strategies to block tumor-associated vessels by targeting angiogenesis will bring considerable excitement in cancer therapies.
Notwithstanding, antiangiogenic therapy has not met researchers' expectations over the past two decades, which should be attributed to antiangiogenic agents, such as bevacizumab, nintedanib, and sorafenib 13 , suffering from drug resistance or off-target challenges [13][14][15] . Recently, several studies have indicated that tumors can continue to progress by hijacking the pre-existing vasculature of nonmalignant tissues, which is called vessel co-option 16 . The presence of vessel co-option can be clari ed, to some extent, as the barrier of anti-angiogenic therapies, including unaccountable drug resistance and uncontrollable tumor metastasis 17,18 . Cancer vessel co-option is a nonangiogenic process in which tumor cells directly hire the pre-existing vessels from the surrounding nonmalignant tissue, where cancer cells will nally in ltrate or migrate 16,19 . Recently, a body of reports has revealed that neglecting vessel co-option might be the vital reason for anti-angiogenesis failure 17,18,20,21 . Studies have highlighted that increased microvascular invasion could be one of the most independent prognostic factors in predicting Sorafenib release behavior from CS NPs was detected in 3 pH environments (7.4, 6.5 and 4.5). Five milliliters of CS NPs was transfused into a dialysis bag (molecular weight = 1000 Da). Then, the dialysis bags were immersed in 500 mL of PBS at different pH values at room temperature with stirring. One hundred microliters of outer dialysate was collected at each detected time point. This releasing experiment was studied for 48 hours. Additionally, to simulate PTT-stimulated drug release in tumor tissue, we studied drug release at room temperature, 45 ℃ and 50 ℃ at a pH value of 6.5.

Stability Assessment
Aqueous solutions, PBS, DMEM and FBS solutions of CS NPs were placed at RT for 30 days at a concentration of 50 µg/mL (determined by sorafenib). Then, the hydrodynamic diameter was assessed on days 0, 7 , 14 and 30 .

Nir-ii Fluorescence Quantum Yield (Plqy) Detection
The organic dye IR26 (Ф IR26 = 0.05% 35,36 ) was selected as a reference to calculate the quantum yields of C NPs and CS NPs. Five various concentrations were studied, and the 1000-1400 nm region uorescence was plotted under an 808 nm laser. the PLQY was quanti ed followed the manner 37 : Slope sample Slope IR26 n sample n IR26 2 Ф IR26 = 0.05%. The slope is the linear tting line with the absorption as the X axis and the uorescence intensity as the Y axis. n sample and n IR26 are the refractive indices of water and dichloroethane (DCE), respectively.

Ros Detection
DPBF was utilized as the ROS probe of CS NPs under 808 nm laser irradiation. In this study, different concentrations of CS NPs in deionized water (2.97 mL, 20 µg/mL) were mixed with DPBF (30 µL, 10 mM) in a quartz cuvette. Subsequently, the mixture was excited under an 808 nm laser with a power of 0.5 W for a total of 10 s, in which the UV absorbance of DPBF at 415 nm was read at each set-time point.
To calculate the quantum yield of ROS produced by Coi8DFIC, MB was used as the reference dye. In this study, Coi8DFIC and MB were dissolved in dichloromethane (DCM). In addition, the procedure was the same as upper one. The quantum yield of ROS was calculated using the following formula: C: the speci c heat capacity of solvent and solute, C Water = 4.2 × 10 3 J/g.
Ordinarily, the speci c heat of water was much higher than that of other materials. Thus, the m and C of CS NPs were too small to neglect compared with the m and C of water.

Cellular Uptake
To detect cellular uptake, HUVECs and CAL 27 cells were incubated with 5 µg/mL CS NPs-RB. Time gradients were set to clarify the time point for maximum uptake, which could be represented by the maximum RB uorescence in cells. Subsequently, the intracellular localization of CS NPs was investigated using LysoTracker and MitoTracker. Method was followed the instructions. The colocalization was calculated by ImageJ with Pearson's correlation coe cient.

Intracellular Ros Assessment
To assess total ROS generation, CAL 27 and HUVECs were cultured with 10 µg/mL CS NPs at the maximum uptake time point and then treated with 10 µM DCFH-DA. Cells were treated with 808 nm irradiation (100 W/cm 2 ) for 10 min. ROS generation was analyzed by confocal microscopy.

Cytotoxicity in vitro
Cell proliferation curves and clone forming approaches were adopted to investigate the cytotoxicity of C NPs and CS NPs. To detect the proliferation ability, CAL 27, HUVECs and HACAT cells were seeded in 96well plates and precultured overnight, and then the medium was replaced by C NP and CS NP concentrations equivalent to 0.94, 1.88, 3.75, 7.5, 15, 30 and 60.0 µg/mL. Thereafter, the cells were cultured for 24 h. The cell viability was assessed by CCK8. In addition, a growth curve was generated with a similar method.
For clone formation observation, CAL 27 and HUVEC single cell suspensions were seeded into 6-well plates at a density of 2000 cells per well. Then, the cell clones, which were stained with crystal violet (0.5% w/v), were counted after a 2-week incubation.

Angiogenesis Assay
HUVECs formed tube-like structures on Matrigel (Corning, USA) with the introduction of vascular endothelial growth factor (VEGF 100 ng/mL). The same method was adopted with the previous study. After [12][13][14][15][16] h of incubation at 37°C, tubes could be observed. The medium was removed and subsequently changed to contain sorafenib, C NPs, and CS NPs at a concentration of 10 µg/mL. As a result, the tube-like structures broke up 4 hours after incubation. The results were captured with the microscope.
Cell Apoptosis Investigation CAL 27 and HUVECs were seeded into 6-well plates overnight. The medium was replaced with fresh DMEM containing sorafenib, C NPs and CS NPs. After incubation for 12 h, cells in the C NP and CS NP groups were irradiated with an 808 nm laser (100 mW/cm 2 ) for 10 min. Next, the cells were labeled by using an Annexin V-FITC/PI kit. Cells were analyzed by ow cytometry.

Hemolysis Assay
The hemolysis assay was performed to clarify whether the CS NPs were safe for intravenous administration. Fresh blood (1 mL) was collected from healthy mice and then centrifuged (1000 g, 10 min, 4 ℃). Red blood cells (RBCs) were washed with PBS 3 times. In this study, 200 µL RBCs were treated with DI water, PBS, and different concentrations of CS NPs, in which PBS and DI water were employed as the negative and positive controls, respectively. Then, the absorbance of the hemoglobin released in the supernatant was monitored by a UV spectrophotometer at 570 nm.

Nir-ii Fluorescence Imaging Of Microcapillary Tubes And Vessels
To evaluate the NIR-II uorescence penetration depth, 5 µg/mL CS NPs were put into a capillary tube with a diameter of 1 mm. Then, they were placed under fresh chicken breasts with different thicknesses of 2 mm, 4 mm, 6 mm, 8 mm and 10 mm. The uorescence intensity of these capillary tubes showed the penetration ability of CS NPs.
CAL 27 tumor-bearing mice were injected with 100 µL CS NPs under anesthetization by iso urane. NIR-II uorescence of CS NPs was rst evaluated for its capability to track microvessels.
The full width at half-maximum (FWHM) of capillary tubes (0.5 mm) was assessed to evaluate the spatial resolution. The gray values of all NIR-II images were normalized by ImageJ-win64. Then, gray value statics were tted with Gauss tting to estimate FWHM via Origin software. Then, we further performed bioimaging of the hindlimb and its vessels. The signal-to-background ratio (SBR) was calculated using the following equation 39 :

Tumor And Vessel Co-option Imaging
To clarify the ability to trace tumor and tumor-associated vessels, mice were injected intravenously with 100 µL CS NPs (concentration of COi8DFIC equivalent to 50 µg/mL). Tumor-bearing mice were observed with the NIR-II uorescence animal imaging system at designated time points. In the meanwhile, ICG was employed as a control. Finally, the biodistribution of NPs in the different parts of organs was analyzed.

Antitumor therapy in vivo
A total of twenty-ve CAL 27 tumor-bearing mice were randomly divided into ve groups (i. physiological saline, ii. Sorafenib, iii. CS NPs, iv. C NPs + 808 nm laser, v. CS NPs + 808 nm laser). Each mouse received ve therapies at intervals of 5 days, and 100 µL of solution was injected intravenously with COi8DFIC or sorafenib at a concentration of 50 µg/mL. The 808 nm laser at the power of 0.5 W/cm 2 was performed 2 h after injection, and the irradiation time was 10 min. Additionally, the temperature at the therapy site was recorded. Before every treatment, the weight and tumor size of each mouse were documented. Upon nishing treatment, the mice were sacri ced by cervical dislocation for necropsy, and tumors and their major organs were collected for histopathological analysis. The tumor tissues were evaluated with H&E, CD 31 and TUNEL staining. Major organs were stained with H&E. All histopathological analysis procedures strictly followed the instructions.
Investigation of the immune effect of tumor-associated vascular blocking therapy C3H mice were subcutaneously injected on the right back with SCC7 cells (1 × 10 6 ) on the insulation blanket under anesthesia. When the tumor reached 100 mm 3 , tumor-and tumor-associated vessels were located by NIR-II uorescence. Then, PDT therapies were conducted with the guidance of precise orientations, especially for vessel co-options around the tumors. The concentrations of drugs and laser power were the same as the treatment for BALB/C nude mice. After 5 treatments, 1 × 10 6 SCC7 cells were injected at the opposite site (the left back) of the mice. Then, the left tumor size and immune cells in the blood were investigated. Immune cells, including CD3 + , CD8 + CD3 + , and CD4 + T cells, were analyzed by ow cytometry.

Blood Circulation Analysis And Biosafety Assessment
Sprague-Dawley (SD) rats (female) were employed to assess the medical effect of NPs. Approximately 130 g SD rats were randomly divided into three groups (i. saline, ii. Sorafenib, iii. CS NPs). Each group contained three rats. For blood circulation, the rats were treated by gavage with various solutions (200 µL, at a dose of 2.5 mg/kg). Blood was taken from the tail artery and stored in heparinized tubes at each time point ( The half-life of the drug was calculated as follows:

Pathological Study
All pathological progression was assessed by two individual experienced pathologists. All pathological assessments were based on the latest edition of the Classi cation guidance. WPOI types were assessed on H&E-stained sections. Tumor-associated vessels were con rmed by CD 31 immunohistochemical staining.
The vessel co-option located at the interface of OSCC and adjacent nonmalignant tissue was also marked with CD 31. Microvessel density (MVD) was employed to evaluate vessel co-option 40 .
Surrounding the edge of OSCC, three zones (1.5 mm 2 ) were randomly circled, and vessel co-option was counted in those circles. The MVD of vessel co-option was calculated using the following formula:

Statistical analysis
The chi-square test and Fisher's exact test were used to estimate correlations among the vessel co-option mode, WPOI and survival. GraphPad, Origin, ImageJ and FlowJo were employed to analyze all data. For all statistical analyses, p < 0.05 was considered signi cant.

Tumor associated vessels in OSCC
It is widely known that cancer cells have close interactions with vascular systems. In particular, a large number of cancers rely on the blood supply to progress. Recently, a body of reports has indicated that the expansion of tumors depends not only on angiogenesis inside it but also on the pre-existing vascular system in nonmalignant tissues, which are termed vessel co-options. Tumor angiogenesis and vessel cooption could be generated as tumor-associated vessels ( Figure 1B, 1C).
Based on the report, there were two ways that blood vessels might be coopted: 1. Cancer cells directly replace normal epithelial cells; 2. Malignant cells invade the nonmalignant zone and employ pre-existing vessels 18 .
As shown in Figure 1, tumor angiogenesis and vessel co-options (indicated by the red arrow) existed in OSCC patient samples ( Figure 1E, 1F). However, numerous studies have indicated that targeting angiogenesis shows only modest e cacy in patients and does not meet the high expectation due to low e cacy or drug resistance. Therefore, preexisting vessels drew our attention and were considered another major driver of tumor progression.

Vessel Co-option And Tumor In ltration Patterns
Patterns of invasion indicate OSCC tumor invasiveness. In this study, we found that vessel co-option showed a close correlation with the invasive pattern. The worst pattern of invasion (WPOI) was assessed on H&E-stained slides. Both the CD 31 + zone and the zone located in the frontier of the tumor represented vessel co-options, as demonstrated in Figure 2A. A total of 40 patients were enrolled in this study, in which 5 types of WPOIs were included: 8 were type 1, 7 were type 2, 8 were type 3, 20 were type 4 and 7 were type 5. All data of 50 patients are shown in Table 1. The vessel co-options in all 5 types of WPOIs are shown in Figure 2B. Then, the relevance between the MVD of vessel co-option and WPOI was calculated ( Figure 2C, Supplementary Table 2), and there was a comparable linear relationship between them. The higher MVD, the higher level of WPOI. The results proved that the MVD of vessel co-option could perfectly predict the mode of WPOI, and vessel co-option might be an effective biomarker to predict the invasiveness of OSCC. With the same method, there was a revalent relationship between MVD and T stage (Supplementary Table 3). Therefore, vessel co-option could be labeled with tumor biological behavior, and insights into vessel co-option would bring new expectations for OSCC therapy. Additionally, a growing body of literature suggests that tumors employing vessel co-option are more resistant to antiangiogenic therapies. Thus, strategies targeting tumor-associated vasculatures should include vessel co-option and angiogenesis inside tumors. Therefore, a novel win-win NIR-II nanoagent was designed to track the tumor-associated vasculatures in real time and target them by employing chemical blocking, PTT and PDT.

Synthesis And Characterization
An A-D-A-type π-conjugated small molecule, COi8DFIC, was demonstrated to be an excellent phototheranostic material. Herein, COi8DFIC and sorafenib, a vascular blocking drug, were combined by nano coprecipitation to produce nanoparticles named CS NPs ( Figure S2), of which the embedding rate of sorafenib was 87.34 ± 0.83% and that of COi8DFIC was 80.28 ± 3.53%. In this study, we proved that the best ratio of COi8DFIC and sorafenib was 1:1, while the characteristics of ratios 1:2 and 2:1 were not as good ( Figure S2). Therefore, a ratio of 1:1 was employed for the subsequent experiments. With the same method, COi8DFIC NPs (C NPs) were also synthesized. From organic COi8DFIC solution to C NP hydrophilic solution, the maximal absorption displayed a large redshift of ~100 nm ( Figure 3A). Then, the absorption spectra of CS NPs proved the success of the combination ( Figure 3B). In addition to the longwavelength absorption, CS NPs also possessed several attractive characteristics. First, CS NPs showed better photostability than ICG and COi8DFIC molecules ( Figure 3C and Figure S2B). After 10 min irradiation at 808 nm (0.5 W/cm 2 ), the absorption spectrum of CS NPs exhibited no change, but ICG showed considerable decay. Next, characterized by TEM, the C NPs and CS NPs were ~ 100 nm and ~120 nm, which was consistent with the DLS results of ~ 90 nm and 120 nm. (Figure 3D, 3E). The fabricated CS NPs displayed long-term stability in simulated physiological environments (PBS, DMEM and FBS, Figure 3F) at 37 ℃ for 30 days ( Figure S3).
Given the A-D-A structure, the aggregation property of COi8DFIC in NPs was activated, which showed a longwavelength redshift of 100 nm, reaching the NIR-II region ( Figure 3G). π conjugated compounds are well known for their high performance in strong luminescence, of which the electron-donating "D" core delivers intense emission properties. Therefore, C NPs and CS NPs showed longwavelength NIR-II uorescence (1052 nm) and high quantum yields (PLQYs) of 1.19% (23.82-fold of IR26) and 0.89% (17.85-fold of IR26), respectively ( Figure 3H). The uorescence of CS was not signi cantly quenched after 7 days of nondark storage at RT. ( Figure 3I).
To further investigate the characteristics, the photothermal effect of CS NPs under 808 nm laser irradiation was examined. As shown in Figure 4A, the temperature rise depended on the concentration of CS NPs and the power of the 808 nm laser. The temperature of the CS NP solution (20 µg/mL) could rise from 24°C to 63°C upon 8 min of 808 nm laser (1 W/cm 2 ) irradiation ( Figure 4A). As displayed in Figure  S4A, there was no signi cant change in temperature after 3 heating-cooling cycles performed on CS NPs, indicating eminent photostability. The photothermal effect of C NPs was also studied and is shown in Figure S4B and 4C. Following the formula, the photothermal conversion e ciency of CS NPs was 46.94%(Figure4A). Wit a high photothermal conversion e ciency and favorable stability, CS NPs are an excellent candidate for photothermal therapy.
Then, in vitro ROS generation of C NPs and CS NPs was evaluated by DFBP and SOSG probes, where MB and ICG were employed as the reference, respectively. As shown in Figure 4B, in the presence of MB and COi8DFIC, the absorption at 450 nm decreased, which meant that the generated ROS degraded DPBF.
Based on the formula in the method, the total ROS quantum yield of COi8DFIC was 18.85 (33.07-fold of MB). Then, the SOS e ciency of CS NPs was evaluated with SOSG, which was determined by comparison with the commercial photosensitizer ICG. Under 808 nm laser irradiation, CS NPs induced decomposition of SOSG, which was shown as a reduction in the uorescence intensity at 525 nm. The SOS quantum yield of CS NPs could be calculated by the reference of ICG (0.2%). The results revealed that CS NPs exhibited an ROS quantum yield of 0.92%, which was 4.60-fold that of ICG ( Figure 4C). Compared with ICG, CS NPs possessed a higher capability of producing SOS and resistance to photobleaching, which made CS NPs a capable and durable PDT agent.

Cellular Uptake And Ros Generation
Given the above results, it was indicated that CS NPs performed well both in e cient NIR-II region uorescence and laser-irradiated PDT/PTT effects. These characteristics make it a potential candidate for antitumor applications, including OSCC treatment. In addition, it was revealed that both vessel cooption and tumor angiogenesis inside the tumor itself played a signi cant role in patient prognosis. Thus, both of them should be regarded as the focus of tumor vascular occlusion therapy. According to previous results, it was demonstrated that the combination of sorafenib with PDT did a remarkable job in OSCC treatment 41 . Herein, the NIR-II photosensitizers COi8DFIC and sorafenib were combined to realize a winwin strategy in tracking tumor-associated vessels with deep penetration and high resolution and e ciently blocking tumor vessels.
The release of Sorafenib in vitro was studied. With laser irradiation time, smaller particles were present ( Figure 5A), which could be due to the release of sorafenib. The release manner was rst evaluated in different pH environments (pH = 7.4, 6.5, 5.3). Low pH could trigger more sorafenib released. Since the temperature would rise to 40 ℃-50 ℃ when CS NPs (20 µg/mL) were irradiated with 0.3 W/cm 2 laser.
The release of sorafenib at different surrounding temperatures was then studied. Compared with RT, 5.5fold sorafenib could be delivered at 50 ℃ ( Figure 5C). In addition, drug release increased with increasing acidity and temperature ( Figure 5B and 5C). The results indicated that drug release could be controlled by the photothermal effect, which could be achieved at the tumor site (pH = 6.5) with laser irradiation on/off. Thus, this controlled release manner would signi cantly enhance biosafety in vivo.
To better investigate cellular uptake, CS NPs were labeled with rhodamine B-CS-RB NPs, which could be observed by confocal microscopy ( Figure 3B, 5D and 5E). CAL 27 cells or HUVECs were cocultured with 5 µg/mL CS-RB NPs (concentration equal to sorafenib), which was observed for 24 hours. As shown in Figure 5D, the highest red signal in CAL 27 cells could be observed at 12 h, while a high signal could be observed at 6 h -8 h, where there were no signi cant differences. Then, the uptake of CS-RB NPs in HUVECs was exhibited in a time-dependent manner at 6 h, which indicated that the maximum therapeutic effect would be achieved after at least 6 hours of coincubation ( Figure 5E).
Based on the uptake results, intracellular ROS generation was detected by using a DCF probe 8 h after incubation with 10 µg/mL CS NPs. Figure 5F and 5G show that with laser irradiation (808 nm, 0.5 W/cm 2 ), an intense green signal was observed in CAL27 cells and HUVECs. In contrast, a weaker ROS signal was emitted without laser or CS NPs ( Figure 5F and 5G). These results indicated that CS NPs were potent for PDT in OSCC.

Colocalization In Cells
To further study the location of CS-RB NPs, LysoTracker and MitoTracker probes were used to mark lysosomes and mitochondria, respectively. CAL 27 and HUVECs were incubated with CS-RB NPs for 8 h, after which they were observed by confocal microscopy. Additionally, colocalization was calculated via Pearson's coe cient. The results revealed that the red signal of CS -RB NPs was considerably merged with MitoTracker with 91.8% and 81.2% Pearson's coe cient in CAL 27 and HUVECs, respectively ( Figure  6C and 6D). In addition, the Pearson's coe cient values of CS-RB NPs located in lysosomes were 68.61% and 74.3% in CAL 27 and HUVECs, respectively, which were signi cantly lower than those in mitochondria ( Figure 6A and 6B). The intracellular location of CS NPs would be bene cial for the effect of PDT. With laser-controlled drug delivery, CS NPs can achieve fairly precise local treatment, which contributes to maximizing the protection of normal tissue and limiting side effects.

In vitro Therapeutic effect
Given the detailed evaluation of the physicochemical properties and functions of CS NPs, studies at the cellular level were organized to investigate the therapeutic e ciency and biosafety. Medium supplemented with 5 µg/mL CS NPs was cocultured with CAL 27, HUVECs and human normal keratinocytes (HACAT) cells for 72 h to investigate whether CS NPs would affect the growth of the three

types of cells. Cell viability of CAL 27 cells, HUVECs and HACAT cells treated with different concentrations
of C NPs and CS NPs. C NPs alone had low cytotoxicity for all three types of cells, with a cell viability higher than 90%, even at a concentration of 60 µg/mL ( Figure 7A). In contrast, with the assistance of 808 nm laser irradiation, the cytotoxicity obviously increased, leading to ~ 40% and more than 50% cell death at a 7.5 µg/mL concentration treated with C NPs and CS NPs, respectively. ( Figure 7B: CAL 27 and HUVECs). Moreover, the cytotoxicity was much lower in HACAT cells, which was probably attributed to less targeting of sorafenib ( Figure 7A and 7B in HACAT groups). This was a good way to protect healthy and normal tissues. Consequently, the combination of chemotherapy and PDT/PTT could signi cantly increase the killing effect and achieve satisfactory therapeutic outcomes.
It is widely recognized that sorafenib possesses dual antitumor activity, which can not only inhibit the proliferation of tumor cells directly by impeding the RAF/MER/ERK pathway but also interfere with angiogenesis by inhibiting the expression of vascular endothelial growth factor receptor (VEGFR).
Therefore, to better reveal the process of vascular restriction, vessel tube formation and antiangiogenesis were conducted in vitro. As shown in Figure 7C, vessel tube formation was completed 6 h after induction, and 10 µg/mL sorafenib and CS NPs were added to the medium. During the following 6 h, vascular structures gradually decomposed, yet more vascular fractures could be witnessed in the CS NP group than in the sorafenib group. Additionally, with the combined use of laser, less than 10% vessel structure could be observed, indicating that the laser plus CS NPs played important roles in tumor vascular therapy.
Owing to the high proliferation of tumor cells, cell growth curves, colony formation assays and cell apoptosis assays were conducted to evaluate the effect of CS NPs. As shown in Figure 8A, compared with the control, CS NPs signi cantly reduced the growth of CAL 27 and HUVECs. However, the growth of HACAT cells was not affected by CS NPs compared with CAL 27 and HUVECs. Then, 2000 cells were seeded into wells to form cell colonies. However, after treatment with 10 µg/mL sorafenib or CS NPs for 72 h, cell proliferation decreased as the quantity of crystal violet spots decreased signi cantly, in which CS NPs displayed stronger cell restraints. As expected, with 3 laser irradiation cycles (0.3 W/cm 2 , 10 min), few CAL 27 or HUVECs survived and subsequently formed cell clones ( Figure 8B). Moreover, an Annexin V -FITC/PI kit was applied to detect cell apoptosis with different treatments. As displayed in Figure 8C and 8D, sorafenib alone caused a small proportion of cell apoptosis, mainly at the early apoptotic stage, while in the CS NP groups, there was more apoptosis (>20% at the early stage and >12% at the late stage). With the combination of laser irradiation, chemotherapy, PDT and PTT were simultaneously achieved, which resulted in over 70% cell death. Consequently, these results indicated the effectiveness of the combination therapy strategy.

Nir-ii Fluorescence Of Cs Nps
To detect the depth penetration of NIR-II uorescence, fresh chicken breast meat was used to simulate different thicknesses of physiological tissues (2 mm, 4 mm, 6 mm, 8 mm and 10 mm). The spatial resolution was evaluated through the full width at half-maximum (FWHM) of capillary tubes (0.5 mm).
The gray values of all NIR-II images were normalized by ImageJ-win64. Then, gray value statics were tted with the Gauss function to estimate FWHM via Origin software. As shown in Figure 9A, the uorescence image of CS NPs could best simulate the shape and width of capillaries covered with 2 mm tissue (FWHM=0.75 mm, SBR=5.92). An increased FWHM could be observed along with an increased penetration depth, in which 8 mm was the ultimate thickness of penetration.
Furthermore, microvascular imaging was performed on the hindlimbs of nude mice in vivo. Before that, the Hemolysis assay was conducted. As shown in Figure S5, no hemolysis was observed. Therefore, CS NP solutions were regarded to be safe enough for nude mice. After injection of 10 µg/mL CS NP solution for 5 min, the blood vessels in the bindlimb could be observed, where the middle of the hindlimb blood vessel was picked to calculate the FWHM and SBR ( Figure 9C and 9D). Based on the SBR and FWHM, 10-20 min was considered the best observation time window for hindlimb vessels after the injection, of which the SBR was 2.42, 2.56 and 3.34, respectively. In addition, the FWHMs of the hindlimb vessels at the same position were 0.60 mm (10 min), 0.64 mm (15 min) and 0.57 mm (20 min), illustrating the better bioimaging performance between 10-20 min.
According to the above results, tumor imaging in vivo was conducted to clarify the image rendering capability of CS NPs. Three positions were captured ( Figure 9E). Then, long live NIR-II uorescence was monitored for 12 h. Tumor imaging of CS NPs showed good pharmacokinetics and high quantum yield to afford high tumor accumulation and signals ( Figure 9F). Compared with ICG-based NIR-I imaging, much more uorescent interference in ICG-based imaging was observed, which made it di cult to distinguish the tumor count. In addition, the liver showed a very strong uorescent signal, which further enhanced the di culty of identifying the tumor site ( Figure 9G). In this study, we observed the accumulation of CS NPs in the tumor site for 4 h. More interestingly, a signal of a small spot was observed for 12 h, which was along with the vessel co-option and close to the tumor ( Figure 9F). The spot was subsequently con rmed to have micrometastasis by pathological examination ( Figure 9H). These features were of great signi cance for clinical navigation during surgery. Surgeons could clearly evaluate the real boundary of tumors in real time during surgery. More importantly, they could also discover micrometastasis as early as possible by image-guided tracking of vessel co-option. In particular, the connection (the vessel cooption) appeared to be a tunnel between the adjacent normal tissues and tumors.

Antitumor therapy in vivo
Based on the satisfactory outcome at the cellular level, in vivo experiments were then performed to further investigate the therapeutic outcome of CS NPs. In this experiment, CAL 27 tumor-bearing mice were randomly divided into 5 groups receiving different treatments (Control, Sorafenib, CS NPs, C NPs + Laser and CS NPs + Laser). In addition, vessel co-options were tracked by NPs and included in therapies. During the 25-day therapy process, tumor-bearing BALB/c mice in all groups, except the sorafenib group, manifested an increase in body weight, demonstrating the good biosafety of NPs ( Figure 10A). Compared with the control group, limited suppression of tumor growth was observed in the sorafenib group, which could be ascribed to the limited drug accumulation inside the tumor and the clearance effect. However, mice treated with CS NPs exhibited a remarkable tumor inhibition effect, especially mice treated with CS NPs + 808 nm laser irradiation. This result revealed that CS NP-mediated vascular blockade combined with PDT and PTT could acquire an ideal therapeutic effect ( Figure 10B). Furthermore, the survival of mice in all groups was analyzed. As shown in Figure 10C, free sorafenib did not show a bene t in survival compared with the control, implying the high biological toxicity or low therapeutic effect of sorafenib.
However, loading NPs, CS NPs and CS NPs + laser treatment signi cantly prolonged survival, which was consistent with the results at the cellular level. Subsequently, intact tumor and main organ specimens were obtained. The H&E results of tumors are shown in Figure 10E, which shows poorly differentiated squamous cell carcinoma.
To further evaluate the therapeutic e cacy, the expression levels of CD31, VEGFR and TUNEL in the tumor site were calculated. As displayed in Figure 11A, different degrees of vascular fractures appeared alongside the vessel co-options, which were located at the frontier of the OSCC tumor. The frequency of vascular fractures (white arrowhead) in the sorafenib groups was obviously lower than that in the CS NPs and CS NPs + laser groups. The CS NPs + laser group showed the most elevated frequency of vascular fractures, suggesting that the combination of NPs and laser holds high pro ciency in the vessel blockade effect ( Figure 11A). In addition, CS NPs + laser treatment exhibited the best antitumor outcome regarding cell apoptosis induction and remarkable VEGFR inhibition ( Figure 11B and 11C). All these data indicated the outstanding performance of the combination of vessel blockade with PDT and PTT in OSCC tumor therapy.
The main organs of mice were obtained and stained with H&E. As shown in Figure S6, damage to the livers in the sorafenib group could be witnessed. Red arrows indicate vacular degeneration and necrosis of hepatocytes. Moreover, foam-like structures could also be observed in liver specimens. In contrast, no abnormalities or toxic representations were observed in the other groups, indicating the excellent biosafety of NPs.

Biocompatibility and blood pharmacokinetics evaluation in vivo
Owing to the side effects of sorafenib, the biocompatibility of CS NPs was explored. First, a blood pharmacokinetics experiment was conducted. Six rats (~ 130 g) were randomly divided into two groups. Rats received gavage administration of sorafenib or CS NP solution at a concentration of 2.5 mg/kg.
Before that, the standard curve of sorafenib solution was evaluated by UFLC-QTRAP-MS/MS ( Figure  12A). Then, blood samples from rat veins at different time points were examined. The results in Figure   13B and 13C indicated that the blood sorafenib concentration in the CS NP groups was signi cantly higher than that in the free sorafenib group. After 2 h of gavage, the accumulation amount of CS NPs reached a maximum in the blood. The half-life time of CS NPs (t 1 / 2 = 10.65h) was more than 10-fold As shown in Figure 12D to 12I, the function of the liver and kidney indicated that gavage administration of C NPs and CS NPs did not cause any obvious disturbance to the hematological indexes, further supporting the good biocompatibility of NPs. However, sorafenib caused signi cant damage to liver function, which was in accordance with the previous H&E result in Figure S6. To further investigate the biocompatibility of these drugs, acute toxicity tests and dermal sensitization responses were added to KM mice and Sprague-Dawley rats (SD rats), respectively. As shown in Figure S8, sorafenib resulted in cardiotoxicity, decreased skin thickness and skin structural disorder, while no signi cant damage was seen in the CS NP or C NP groups. Therefore, nanotechnology was an effective and uncomplicated method to reduce the toxicity of sorafenib.

Antitumor Immune Effects Induced By Combination Therapy
Given that CS NP-mediated vascular therapy and PDT/PTT could almost eliminate the tumors and inhibit metastasis, a dual-tumor model (primary and metastatic tumors) in C3H mice was established to further study the antitumor effect. The experimental procedure is shown in Figure 13A. After 5 primary therapies, a metastasis model was established in the opposite site. As expected, the growth speed of metastatic tumors was different in various groups. Notably, sorafenib alone and NPs not only suppressed the primary tumor but also inhibited the metastatic tumor ( Figure 13B and 13C). In addition, CS NPs + laser yielded the highest inhibitory effect on metastatic tumors. All data suggested that vascular blockade, PDT and PTT have the potential to stimulate antitumor metastasis immune effects to eliminate distant metastasis. Therefore, effector T cells of antitumor immunity were assessed. CD8 + T cells, which can directly attack tumor cells, and CD4 + T cells also play a crucial role in antitumor immunity. The populations of CD3 + and CD4 + T cells and CD3 + and CD8 + T cells in the blood were detected by ow cytometry. As shown in Figure 13D, all treatments induced a systemic immune response to cytotoxic T cells (CD8 + ) and helper T cells (CD4 + ). Furthermore, it was found that with the delivery of NPs, the antitumor immune effect increased signi cantly. In particular, in the CS NPs + laser group, CS NPs played an important role in triggering systematic immunity for cancer therapy. In the mouse model, combination therapy of blood blockade and PTT and PDT effectively increased the ratio of antitumor/protumor immune cells and signi cantly inhibited metastatic tumor growth, and the e cacy of combination therapy was superior to that of monotherapy.

Conclusion
In this work, we successfully developed novel NIR-II therapeutic agent CS NPs for chemotherapy/PDT/PTT combination treatment, which achieved a win-win outcome by tracing and blocking tumor-associated vessels for oral squamous cell carcinoma. The A-D-A structure of COi8DFIC contributed to strong luminescence, with the PLQY value of C NPs achieving 1.19%. In addition, the obtained CS NPs possessed a high penetration depth of 6 mm and a high SBR of 4.3, which was bene cial for microvessel imaging. In vivo optical imaging results revealed that CS NPs could track vessel co-option and accurately locate tumor micrometastasis in a timely manner, which was signi cant for clinical navigation. By tracking and blocking vessel co-option, CS NPs + laser exhibited effective tumor suppression with outstanding biosafety and organ protection. Our results demonstrated that the combination therapy not only pruned blood vessels, which are essential to cancer growth and metastasis but also stimulated the tumor immune response. Therefore, CS NPs would bring promise for clinical translation, and the proposed strategy may provide a win-win strategy for OSCC treatment.

Competing interests
There is no competing interests in the research. Availability of data and material The processed data required to reproduce these ndings are available to download from the o cial website of the published article.    The SOS production of ICG and CS NPs irradiated at 808 nm (0.5 W/cm 2 ). SOSG probe was used to capture SOS. (a, b) the uorescence intensity pro les of ICG and CS NPs irradiated with 808 nm laser, respectively. (c) The linear tting of SOSG uorescence intensity at 525 nm and irradiation time. The slope of these lines was adopted to calculated the quantum yield.