5-FU (Beijing Bailingwei Technology Co., LTD.); Cetyl trimethyl ammonium bromide (Jining Sanshi Biotechnology Co., LTD.); Ethyl orthosilicate (Tianjin Kermel Chemical Reagent Co., LTD.); (Tianjin Guangfu Fine Chemical Research Institute); Dimethyl sulfoxide (Jiangsu Haolong Chemical Co., LTD.); and Phosphate buffer salt solution (PBS, Hyclone Corporation, USA) were utilized. All other reagents were domestic analytical pure reagents.

The synthesis steps of mesoporous silica carrier were shown in Figure 1 . 0.535g cetyltrimethylammonium bromide (CTAB) was dissolved in 240mL sterilized ultrapure water, and ultrasonic dispersion was performed for 15min. Then, 1.25mL of 2mol/L NaOH solution was added, followed by ultrasonic dispersion for 5min. The liquid was stirred continuously at 80°C for 30 minutes. Ethyl orthosilicate (TEOS) was added dropwise every 4s, totaling 5mL. After 2-hour reaction, the liquid was left at room temperature for 30min. The lower layer solution was taken for centrifugation at 10,000rpm for 3min to obtain the precipitate. The mesoporous silica nanoparticles (MSN) were then immersed in ethanol solution, followed by ultrasonic dispersion and centrifugation at 10,000 rpm for 3 minutes. The above steps were repeated three times. Next, the sample was transferred in a vacuum drying oven and dried overnight to obtain the purified MSN sample.

3-aminopropyltriethoxysilane (AMEO) was mixed with MSN at a ratio of 2:3, and an appropriate amount of toluene was added, followed by reflux under nitrogen protection for 12h at 110°C. Then, the solution was centrifuged at 10,000rpm for 3 mins, and the precipitate was washed with ethanol solution. The above steps were repeated three times. Finally, the surface amination treatment of MSN was carried out under vacuum drying conditions. The synthesis steps were shown in Figure 2 .

MSN-5-FU nanoparticles were prepared by reverse phase microemulsification. 7.5 mL cyclohexane, 1.6 mL hexanol, and 1.8 mL Triton-100 (surfactant) were mixed evenly, and 5-Fu solution was added to form an inverting microemulsion. Magnetic stirring was performed at room temperature for 5 min. 150 μL ethyl orthosilicate and 100 μL 25% ammonia were added to the mixture, and the reaction was stirred continuously at room temperature for 24 h. 2 mL acetone was added for demulsification and centrifugation for 20 min. The products were collected and dispersed with ethanol and water respectively, followed by centrifugation to remove unreacted 5-Fu and solvent. The final nanoparticles were vacuum-dried.

Luria-Bertani solid medium was used to cultivate Escherichia coli. After culturing at 37°C for 36 hours, a single colony was inoculated in Luria-Bertani liquid medium and continued to be cultured for 24 hours. 2mL of bacterial solution was inoculated in blank Luria-Bertani liquid medium, set as the blank control group. The culture was terminated when the OD600 was 1. Then, centrifugation was performed at 12,000 rpm for 20 min at 4°C. The supernatant was then passed through the 0.45μm disposable filter. The filtrate was transferred to 20mL ultrafiltration centrifuge tube, followed by centrifugation at 8000rpm at 4°C for 15min. The centrifugal liquid was concentrated to 1.5mL, and PBS was used for resuspension to obtain OMVs.

High-pressure nitrogen was used to extrude the OMVs to uniform their particle sizes. The above steps were repeated 3 times. OMVs and MSN-5-FU were mixed at a ratio of 1:5, and then extruded 6 times. The extruded effluent was centrifuged at 8000 rpm at 4°C for 20 minutes, and the supernatant was discarded. The bottom precipitate was OMVs-MSN-5-FU.

The release amount of OMVs-MSN-5-FU drug delivery system was determined through the immersion test of artificial gastric juice and artificial intestinal juice. The first step was to prepare artificial gastric juice and artificial intestinal juice. Preparation of artificial gastric juice: 1mol/mL dilute hydrochloric acid with pH=1.5 was mixed with 0.001g/mL pepsin. The mixture was then passed through 0.2μm disposable sterile filter membrane for filter sterilization. Preparation of artificial intestinal juice: 6.8g KH2PO₄ was dissolved in 500mL sterile ultrapure water. With the pH of the solution set to 6.8, 0.01g/mL trypsin was then added. Next, the mixture was passed through a 0.2μm disposable sterile filter membrane for filter sterilization. Subsequently, 3mg of MSN-5-FU drug delivery system was added to 20mL of artificial gastric juice and 3mg OMVs-MSN-5-FU drug delivery system was added to another 20mL of artificial gastric juice, followed by shaking at 135rpm at 37°C. The artificial intestinal juice was treated in the same way. 2mL release solution was taken after 0.5h, 1.0h, 2.0h, 4.0h, 8.0h, 12.0h, 24.0h, 48.0h, and 72.0h of shaking, respectively, followed by centrifugation at 2000rpm for 5min at room temperature. Then, the supernatant was taken to determine the content of 5-FU.

KOSC-2 cl3-43 cells were inoculated in a 96-well plate. 0, 0.1, 0.25, 0.5, 1, 2.5, 5, and 10μmol/L 5-FU, MSN-5-FU, and OMVs-MSN-5-FU were inoculated, respectively. At 24h, 48h, and 72h of culturing, 10μL of MTT solution was added to each well. Then, 100μL of dimethylsulfoxide solution was added to each well, followed by shaking for 10 minutes. Finally, the absorbance was measured at a wavelength of 570nm. Cell inhibition rate was calculated. The calculation method of cell inhibition rate is shown in equation (1), where RI was the cell inhibition rate, AS was the absorbance value of the experimental well, AK was the absorbance value of the blank well, and AY was the absorbance value of the negative well.

The clean-grade BALB/c-nu nude mice were used, aged about 5 weeks old, weighing 16-21g, regardless of the gender. They were provided by the XXX animal laboratory. Animal License No.:SCXK(Hebei)2019-0027. In XXX animal laboratory, the day and night cycle was 12 hours at 22~26°C and 45~50% humidity. After three days of adaptive breeding, they were used in formal experiments.

Human tongue squamous cell carcinoma cells Tca8113 were cultured first, and PBS was used to prepare a single cell suspension at a concentration of 1×10⁷ during subculture. 10% chloral hydrate solution was intraperitoneally injected to anesthetize BALB/c-nu nude mice, and 0.2mL of Tca8113 cell suspension was injected into the buccal mucosa. The growth status of the tumor was observed regularly.

When the diameter of the tumor was approximately 0.5 cm, BALB/c-nu nude mice were anesthetized by intraperitoneal injection of 10% chloral hydrate solution, and 100μg/mL OMVs-MSN-5-FU suspension was injected into the buccal mucosa around the tumor in the oral cavity.

Blood was drawn from the tail vein of the tumor-bearing animal model, and enzyme-linked immunosorbent assay was used to detect the level of Naa10 in the peripheral blood. The serum sample was transferred in an enzyme-labeled plate, and incubated at 37°C for 2 hours. After the supernatant was discarded, 100μL of test reagent A was added to each well. After 1 hour, the supernatant was discarded and the cells were washed with sterile ultrapure water 3 times. Then, 100μL of test reagent B was added to each well. After 1 hour, 90μL of enzyme-labeled reagent was added to each well, and 50μL of stop solution was added after incubation for 20min. Finally, the absorbance was measured using a microplate reader at 450nm.

Flow cytometry was used to detect the changes of T lymphocyte subsets (CD3⁺, CD⁴⁺ and CD⁸⁺), B lymphocytes (CD¹⁹⁺) and NK cells (CD⁵⁶⁺) in peripheral blood.

Three days after the injection of OMVs-MSN-5-FU suspension, the animal model was sacrificed by cervical dislocation, and the tissue around the carcinoma and ipsilateral cervical lymph nodes were harvested. The tissue was washed with PBS, and then put in the embedding box. Then, 70%, 90%, 95%, 95%, 100%, 100%, and 100% ethanol solution was used in turn to dehydrate the tissue, followed by immersion in xylene solution for 20 minutes. This step was repeated 3 times. The treated tissue was embedded in paraffin solution to made paraffin tissue sections with a thickness of 3μm. After being dried, the slide was immersed in xylene solution for 10 minutes, twice. Next, 100%, 100%, 95%, 95%, 80% ethanol solution and distilled water were used in turn for tissue rehydration. Subsequently, hematoxylin and eosin staining solutions were used to stain the tissue, followed by immersion in xylene. Finally, the slide was sealed using neutral gum, and visualized under an optical microscope.

SPSS19.0 was used to process the data. The data were all expressed by the mean ± standard deviation (x(-) ± s).One-way analysis of variance was used for statistical analysis of the differences between multiple groups, and P<0.05 was the threshold for significance.

The morphology of MSN, MSN-5-FU, OMVs, and OMVs-MSN-5-FU was visualized under scanning electron microscope and transmission electron microscope. It was found that, all MSN, MSN-5-FU, OMVs, and OMVs-MSN-5-FU were spherical in shape, with uniform distribution. The particle diameters were approximately 60nm, 80nm, 90nm, and 140nm, respectively. Among them, OMVs had an oriented core-shell structure. The scanning electron microscope and transmission electron microscope results were shown in Figures 3 , 4 .

The average Zeta potentials of MSN, MSN-5-FU, OMVs, and OMVs-MSN-5-FU were -20.6 ± 2.3mV, -28.7 ± 2.2mV, -18.2 ± 3.1mV, and -17.4 ± 1.7mV, respectively. The average zeta potential of OMVs was close to that of OMVs-MSN-5-FU. The results of Zeta potential detection were shown in Figure 5 .

The in vitro release rate of MSN-5-FU and OMVs-MSN-5-FU drug delivery systems was analyzed. It was noted that, the cumulative release rate of MSN-5-FU and OMVs-MSN-5-FU drug delivery system gradually increased in artificial gastric juice and artificial intestinal juice. In addition, the 48-hour cumulative drug release rate in artificial gastric juice was 61.2 ± 2.3% and 26.5 ± 3.1%, respectively, and the 48-hour cumulative drug release rate in artificial intestinal juice was 70.5 ± 6.3% and 32.1 ± 3.8%, respectively. The cumulative release of MSN-5-FU was always higher than that of OMVs-MSN-5-FU. The cumulative release rate results were shown in Figure 6 .

The results of cell viability detected by MTT showed that, with the increase of the concentration of 5-FU, MSN-5-FU and OMVs-MSN-5-FU drug delivery systems, the proliferation activity of KOSC-2 cl3-43 cells showed a gradually decreasing trend. At the same time, under the same dosage, the inhibitory rate of OMVs-MSN-5-FU drug delivery system on the proliferation activity of KOSC-2 cl3-43 cells was higher than that of 5-FU and MSN-5-FU. The specific results were shown in Figure 7 .

HE staining was used to analyze the pathological changes of tumor-bearing animal models. It was found that, the cancer tissue showed enlarged nuclei, darkened staining, and irregularly shaped cancer cells. The adjacent tissue mainly consisted of striated muscle, and there were oval nuclei on the edge. Observation of cervical lymph node slices revealed that, the lymphocytes in the cortex were very dense and were divided by cancer cells. The shape of the nucleus of lymphocytes was approximately round and there was no cytoplasm; while the nucleus of cancer cells was enlarged, and the cytoplasm was connected to each other into a sheet. It suggested that, the cancer cells have gradually metastasized to the lymph nodes in the neck, accompanied by cell proliferation. The specific staining results were shown in Figure 8 .

The test results showed that, the model group showed increased levels of Naa10, CD8⁺, and CD56⁺ in peripheral blood, while decreased levels of CD3⁺, CD4⁺, CD4⁺/CD8⁺ and CD19⁺ (P<0.05) versus the normal control group; and that compared with the model group, the OMVs-MSN-5-FU group showed decreased levels of Naa10, CD8⁺, and CD56⁺ in the peripheral blood, while increased levels of CD3⁺, CD4⁺, CD⁺/CD8⁺ and CD19⁺ (P<0.05). However, there was no significant difference between the control group and the OMVs-MSN-5-FU group in the levels of Naa10 and lymphocyte subgroups (P>0.05). The detection results of Naa10 and lymphocyte subgroups in peripheral blood were shown in Figure 9 .

The HE staining results showed that, the aligned nuclei were noted in the striated muscle of the peri-carcinoma tissue, and the injection of OMVs-MSN-5-FU reduced the number of stained cells. In the cortex of the neck lymphocytes, dense lymphocytes were clearly observed. The specific staining results were shown in Figure 10 .