A method for constructing a tongue squamous cell carcinoma immune rejection ternary co-culture model

Abstract

The present application relates to the field of biological medical technology, and in particular to a method for constructing a tongue squamous cell carcinoma immune rejection ternary co-culture model. The tongue squamous cell carcinoma immune rejection ternary co-culture model comprises a tongue squamous cell carcinoma patient-derived organoid region (PDO core region) in a central region, an LRRN4-positive cancer-associated fibroblast barrier region (LRRN4+CAF outer ring region) surrounding the PDO core region, and an NK cell migration region (NK cell peripheral region) outside the LRRN4+CAF outer ring region. The tongue squamous cell carcinoma immune rejection ternary co-culture model not only can better simulate the spatial relationship between tumor nests, stromal barriers and immune cells in tongue squamous cell carcinoma tissue, but also provides a standardized and repeatable in vitro evaluation platform for comparing the effects of different interventions on each link of immune rejection.

Description

Technical Field

[0001] This invention relates to the field of biomedical technology, specifically to a method for constructing a ternary co-culture model of tongue squamous cell carcinoma with immune rejection. Background Technology

[0002] Tongue squamous cell carcinoma exhibits significant individual heterogeneity. In clinical practice, even when receiving the same chemotherapy, immunotherapy, or combination therapy regimens, different patients may still present drastically different treatment responses. Therefore, establishing a reliable in vitro model is crucial. This model must not only preserve the heterogeneity of the patient's tumor cells but also be able to reproduce key microenvironmental features such as the tumor stromal barrier and restricted immune cell infiltration.

[0003] Currently, two-dimensional tumor cell culture models lack three-dimensional tissue structures, making it difficult to realistically simulate the spatial interactions between tumor cells and the extracellular matrix, fibroblasts, and immune cells. While patient-derived organoids (PDOs) can better preserve the morphology and some molecular characteristics of the primary tumor, conventional PDO culture systems are mainly composed of tumor epithelial cells and lack microenvironmental components such as cancer-associated fibroblasts (CAFs) and natural killer cells (NK cells). Therefore, it is difficult to effectively evaluate the migration of immune cells, their crossing of the stromal barrier, and the killing process after entering the tumor nest.

[0004] Existing organoid immunotherapy co-culture methods mostly involve direct mixing and simultaneous inoculation of tumor organoids with immune cells, or simple suspension co-culture. While these systems can evaluate endpoint killing, they disrupt the original spatial arrangement of different cell populations and cannot stably reconstruct the partitioned structure of "tumor nest-CAF barrier-peripheral immune cells." Consequently, it is difficult to distinguish whether candidate interventions promote NK cell migration and weaken the CAF barrier, or simply enhance the killing function of NK cells that have already entered the tumor nest. Summary of the Invention

[0005] Based on this, the present invention provides a method for constructing a ternary co-culture model of tongue squamous cell carcinoma immune rejection, which solves at least one problem in the prior art.

[0006] In a first aspect, the present invention provides a method for constructing a ternary co-culture model of tongue squamous cell carcinoma immune rejection, comprising the following steps: The PDO of the tongue squamous cell carcinoma was suspended in the first three-dimensional matrix and formed into the PDO core region; The LRRN4 positive CAF is suspended in a second three-dimensional matrix, forming an LRRN4+CAF outer ring region surrounding the PDO core region; NK cell suspension was added to the periphery of the outer ring region of the LRRN4+CAF, or NK cell suspension was added to the core region of the PDO and / or the outer ring region of the LRRN4+CAF to obtain a tongue squamous cell carcinoma immune rejection ternary co-culture model.

[0007] The tongue squamous cell carcinoma immune rejection ternary co-culture model comprises a patient-derived organoid region (PDO core region) located in the central region, an LRRN4-positive cancer-associated fibroblast barrier region surrounding the PDO core region (LRRN4+CAF outer ring region), and an NK cell migration region (NK cell peripheral region) located outside the LRRN4+CAF outer ring region. The model does not involve the simultaneous random mixing of PDO, LRRN4-positive CAF, and NK cells. Instead, it is constructed using a spatial partitioning and sequential seeding method (partitioned sequential construction method) that involves first forming a central PDO tumor nest, then constructing an LRRN4-positive CAF barrier around it, and finally adding NK cells from outside the CAF barrier.

[0008] The core advantage of the aforementioned sequential partitioning construction method lies in the fact that by pre-defining the initial spatial locations and addition order of PDO, CAF, and NK cells, a relatively stable and reproducible spatial structure of "central tumor nest - peripheral stromal barrier - peripheral immune cells" can be formed, reducing the spatial relationship disorder and batch-to-batch variability caused by direct random mixing of the three cell types. Simultaneously, as NK cells migrate from the periphery to the center, they sequentially experience approaching the CAF barrier, crossing the CAF barrier, and entering the PDO tumor nest. This allows for relatively independent observation and quantitative evaluation of CAF barrier formation, NK cell migration / nesting, and the killing effect after entering the tumor nest. Therefore, the aforementioned tongue squamous cell carcinoma immune rejection ternary co-culture model not only better simulates the spatial relationships between tumor nests, stromal barriers, and immune cells in tongue squamous cell carcinoma tissue, but also provides a standardized and reproducible in vitro evaluation platform for comparing the effects of different interventions on various aspects of immune rejection.

[0009] The progenitor cells (PDOs) of tongue squamous cell carcinoma can be derived from fresh, excised tongue squamous cell carcinoma tissue, biopsy tissue, or an established patient-derived organoid bank. The obtained tongue squamous cell carcinoma PDOs are preferably identified by bright-field morphological observation, hematoxylin andeosin staining (H&E), pan-CK immunostaining, and Ki-67 immunostaining, combined with at least one method for evaluating the consistency between the PDO and the corresponding primary tumor, to confirm that the established PDO can retain the main histological and / or biological characteristics of the primary tumor. The diameter of the PDO used to construct the ternary co-culture model of this invention is preferably 50-400 μm, more preferably 80-250 μm.

[0010] LRRN4-positive cell fibroblasts (CAFs) can originate from tumor tissue of the same patient with tongue squamous cell carcinoma PDO, or from primary CAFs, stably cultured CAFs, or LRRN4-overexpressing fibroblasts obtained through gene manipulation from other patients. CAFs preferably show positivity for at least one of FAP, α-SMA, PDGFRβ, or COL1A1, and are negative for pan-CK and CD45. LRRN4-positive cells can be obtained by flow cytometry sorting, magnetic bead sorting, immunofluorescence screening, or gene expression detection; preferably, the LRRN4 positivity rate after sorting is not less than 70%.

[0011] NK cells can be selected from patient's autologous peripheral blood NK cells, healthy donor NK cells, in vitro expanded NK cells, NK-92 cells, or engineered NK cells. Primary NK cells are preferably confirmed by CD3-negative and CD56-positive phenotypes, with a purity preferably not less than 80%, and more preferably not less than 90%.

[0012] Without altering the core partition structure of the tongue squamous cell carcinoma immune rejection ternary co-culture model, tumor cell-derived exosomes, CAF-derived exosomes, cytokines, chemokines, extracellular matrix proteins, or other microenvironmental components can be further added to enhance specific immunosuppressive phenotypes.

[0013] In some optional embodiments, the method for constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model includes the following steps: Tongue squamous cell carcinoma PDO was suspended in a first three-dimensional matrix and placed in a culture container for 6-96 h to form the PDO core region; LRRN4-positive CAFs were suspended in a second three-dimensional matrix and placed around the PDO core region. After culturing for 6-96 hours, an LRRN4+CAF outer ring region was formed. NK cell suspension was added to the periphery of the outer ring region of the LRRN4+CAF, or NK cell suspension was added to the core region of the PDO and / or the outer ring region of the LRRN4+CAF to obtain a tongue squamous cell carcinoma immune rejection ternary co-culture model.

[0014] In some optional embodiments, the inoculation density of LRRN4-positive CAFs in the outer ring region of the LRRN4+CAF can be 0.1 × 10⁻⁶. 6 -10×10 6 Cells / mL, preferably 0.5 × 10⁻⁶. 6 -5×10 6 per mL.

[0015] In some optional embodiments, the seeding density of the NK cells can be 0.1 × 10⁻⁶. 6 -10×10 6The effector-to-target ratio, calculated based on the estimated number of NK cells to tongue squamous cell carcinoma PDO tumor cells, can be 0.5:1-20:1, preferably 1:1-10:1.

[0016] In some optional embodiments, the PDO kernel region is formed after 12-48 h of culture, and the LRRN4+CAF outer ring region is formed after 12-72 h of culture.

[0017] In some optional embodiments, the first three-dimensional matrix and the second three-dimensional matrix are at least one of basement membrane extract, Matrigel, collagen gel, fibrin gel, alginate gel, hyaluronic acid gel, gelatin methacrylamide gel, and polyethylene glycol synthetic hydrogel. The PDO core region is formed by embedding the first three-dimensional matrix, the function of which is to maintain PDO survival, expansion, and tumor nesting structure. The second three-dimensional matrix may be the same as or different from the first three-dimensional matrix. Preferably, the second three-dimensional matrix is ​​collagen I, fibrin, basement membrane extract, or a combination of at least two of these. The LRRN4+CAF outer ring region is formed by the second three-dimensional matrix containing LRRN4-positive CAF and at least partially surrounds the PDO core region, wherein the function of the second three-dimensional matrix is ​​to support CAF expansion, contraction, and the formation of continuous or semi-continuous barrier structures.

[0018] In some optional embodiments, the volume of the PDO core region can be 3-100 μL, preferably 5-50 μL; the volume of the peripheral LRRN4+CAF outer ring region can be 10-500 μL, preferably 20-200 μL. The width of the peripheral barrier region can be 0.1-5 mm, preferably 0.2-2 mm. The PDO core region and the LRRN4+CAF outer ring region can be in direct contact, or a 0-1 mm interface gap can be maintained.

[0019] In some alternative embodiments, a detachable annular limiting element, a concentric positioning template, a dual-zone culture insert, or a culture plate with a central positioning mark is used for positioning to define the PDO core region and the LRRN4+CAF outer ring region. The annular limiting element can be made of polystyrene, polycarbonate, polydimethylsiloxane, or other cell culture compatible materials and is sterilized. This positioning step improves batch-to-batch consistency.

[0020] In some alternative embodiments, the NK cells are replaced with CD8-positive T cells, γδ T cells, CAR-NK cells, or other immune cells with tumor-killing capabilities.

[0021] In some optional embodiments, the method for constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model includes the following steps: The tongue squamous cell carcinoma PDO was suspended in a first three-dimensional matrix and placed in the central tumor channel of a microfluidic device to form the PDO core region; LRRN4-positive CAF is suspended in a second three-dimensional matrix and placed in a matrix channel of the microfluidic device adjacent to the central tumor channel to form an outer ring region of LRRN4+CAF surrounding the PDO core region. NK cell suspension was added through the outer perfusion channel of the microfluidic device to obtain a ternary co-culture model of tongue squamous cell carcinoma immune rejection.

[0022] In some optional embodiments, the method for constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model includes the following steps: The PDO of tongue squamous cell carcinoma was suspended in a first three-dimensional matrix, and the PDO core region was formed by three-dimensional printing. The LRRN4 positive CAF was suspended in a second three-dimensional matrix, and the outer ring region of LRRN4+CAF surrounding the PDO core region was formed by three-dimensional printing. NK cell suspension was added to the periphery of the outer ring region of the LRRN4+CAF, or NK cell suspension was added to the core region of the PDO and / or the outer ring region of the LRRN4+CAF to obtain a tongue squamous cell carcinoma immune rejection ternary co-culture model.

[0023] In some optional embodiments, the method for constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model includes the following steps: The tongue squamous cell carcinoma PDO was suspended in the first three-dimensional matrix and placed in a central circular region enclosed by an isolation structure to form the PDO core region; The LRRN4 positive CAF is suspended in a second three-dimensional matrix and placed in an annular region enclosed by an isolation structure, wherein the annular region is located outside the central circular region, forming an outer ring region of LRRN4+CAF surrounding the PDO core region. Add NK cell suspension to the periphery of the outer ring region of the LRRN4+CAF, or add NK cell suspension to the core region of the PDO and / or the outer ring region of the LRRN4+CAF, to obtain a tongue squamous cell carcinoma immune rejection ternary co-culture model. The isolation structure is made of a soluble or extractable material, and the isolation structure is dissolved or extracted after the PDO core region and LRRN4+CAF outer ring region are stabilized.

[0024] Secondly, the present invention provides the application of the tongue squamous cell carcinoma immune rejection ternary co-culture model obtained by the method of constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model in screening drugs for the treatment of tongue squamous cell carcinoma.

[0025] Thirdly, the present invention provides a kit for constructing a ternary co-culture model of tongue squamous cell carcinoma with immune rejection.

[0026] In some alternative embodiments, the kit includes a model building core component, which includes PDO culture medium or a concentrated portion thereof, a first three-dimensional matrix, a second three-dimensional matrix, LRRN4-positive CAF or a cryopreservation formulation thereof, and NK cell culture or activation components.

[0027] In some optional embodiments, the kit further includes one or more of the following: LRRN4 detection / sorting reagent, NK cell tracing reagent, organoid activity or death detection reagent, concentric positioning template or detachable ring-shaped limiting element, multi-well culture plate, image analysis program, and instruction manual.

[0028] The first three-dimensional matrix is ​​used to form the PDO core region; the second three-dimensional matrix is ​​used to load LRRN4-positive CAF and form an LRRN4+CAF outer ring region around the PDO core region; the PDO culture medium or its concentrated components are used to maintain tongue squamous cell carcinoma PDO culture; the NK cell culture or activation components are used to maintain the survival and / or activity of NK cells added to the model. The auxiliary components are used for the detection or sorting of LRRN4-positive CAF, NK cell tracing, PDO activity or death detection, spatial localization, and image acquisition and quantitative analysis of the model results.

[0029] Due to the adoption of the above technical solutions, the embodiments of the present invention have at least the following beneficial effects: (1) By clearly dividing the PDO core region, the peripheral LRRN4+CAF outer ring region and the peripheral NK cell peripheral region, the model is closer to the spatial relationship between tumor nests, stromal barrier and immune cells in tongue squamous cell carcinoma tissue; (2) Sequential seeding instead of direct mixing of the three cell types can control the formation of PDO, the establishment of CAF barrier and the addition time of NK cells, reducing batch-to-batch differences caused by random mixing; (3) It can separate and quantify the three processes of “NK cell migration and nesting”, “CAF barrier function” and “nesting killing”, which is helpful to identify the specific action links of candidate interventions; (4) Introducing LRRN4-positive CAF as a functional barrier component can evaluate the effect of specific CAF subsets on immune rejection, and can set up mechanism controls through LRRN4 low expression, LRRN4 knockdown or no CAF models. (5) The model is compatible with patient autologous cells, healthy donor immune cells, cell lines and engineered immune cells, and has a wide range of applications. It can be used for drug screening, cell therapy evaluation and mechanism research. (6) The model can be established in conventional culture plates and can also be extended to microfluidics and high-throughput platforms, with the potential to be standardized, reproducible and commercialized into kits; (7) All operations are performed in ex vivo samples and in vitro systems, which can reduce the reliance on animal experiments for early candidate screening. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the spatial structure of the tongue squamous cell carcinoma immune rejection ternary co-culture model in Embodiment 2 of the present invention, wherein the left figure is a top view and the right figure is a cross-sectional view.

[0031] Figure 2 This is a bright-field image of live cells from the ternary co-culture model of tongue squamous cell carcinoma immune rejection in Example 3 of the present invention.

[0032] Figure 3 This is a parallel endpoint confocal image of the tongue squamous cell carcinoma immune rejection ternary co-culture model after 0-72 h of culture in Example 3 of the present invention.

[0033] Figure 4 This is a graph showing the quantitative analysis results of the tongue squamous cell carcinoma immune rejection ternary co-culture model in Example 3 of the present invention.

[0034] Figure 5 This is a representative bright-field image of live cells after 48 h of co-culture of tongue squamous cell carcinoma with PDO under different CAF conditions in Example 4 of this invention.

[0035] Figure 6 The images show the endpoint confocal images and representative three-dimensional reconstructions and orthogonal sections of the tongue squamous cell carcinoma PDO co-culture model under different CAF conditions after 72 h of culture in Example 4 of this invention.

[0036] Figure 7 This is a graph showing the quantitative analysis results of the culture endpoints of each group of models in Example 4 of the present invention.

[0037] Figure 8 These are representative confocal images of the partitioned sequential model and the direct hybrid model after culturing for 24 h, 48 h, 72 h and 96 h in Example 4 of this invention.

[0038] Figure 9 These are representative confocal plots of three independent batches of the partitioned sequential model and the direct mixed model in Embodiment 4 of the present invention.

[0039] Figure 10 This is a graph showing the combined quantitative analysis results of the partitioned sequential model and the direct mixed model in Embodiment 4 of the present invention.

[0040] Figure 11 These are the three-dimensional maximum projection diagrams and XZ and YZ orthogonal sectional diagrams of each processing group in Embodiment 5 of the present invention.

[0041] Figure 12 This is a graph showing the quantitative analysis results of each indicator in each treatment group in Embodiment 5 of the present invention.

[0042] Figure 13 This is a diagram showing the gating strategy and representative results of NK cell flow cytometry detection in Embodiment 5 of the present invention.

[0043] Figure 14 This is a graph showing the quantitative analysis results of NK cell functional indicators in each treatment group in Example 5 of the present invention.

[0044] Figure 15 This is a graph showing the PDO mortality rate analysis results of each treatment group under NK cell-free conditions in Example 5 of the present invention.

[0045] Figure 16 This is a graph showing the number of exosome-like particles and marker protein detection results in the culture supernatant of each treatment group in Example 5 of the present invention.

[0046] Figure 17 This is a graph showing the detection results of TGF-β / Smad signals in the CAF of each processing group in Embodiment 5 of the present invention.

[0047] Figure 18 This is a graph showing the cell viability analysis results after 72 h of culture in each treatment group in Example 5 of the present invention.

[0048] Figure 19 These are representative three-dimensional maximum projection images and XZ and YZ orthogonal cross-sectional images of the ternary co-culture model derived from 6 patients in Embodiment 6 of the present invention.

[0049] Figure 20 This is a patient-specific technical repeatability CV distribution map of five indicators in the ternary co-culture model derived from six patients in Embodiment 6 of the present invention.

[0050] Figure 21 This is a multiplex immunofluorescence assay result of a ternary co-culture model derived from three patients (P1, P3, and P6) in Example 6 of this invention.

[0051] Figure 22 This is a graph showing the correlation analysis results in Embodiment 6 of the present invention.

[0052] Figure 23 This is a graph showing the Bland-Altman consistency analysis results of the four spatial indicators in Embodiment 6 of the present invention. Detailed Implementation

[0053] The following will provide a clear and complete description of the concept and technical effects of the present invention, so as to fully explain the purpose, solution and effects of the present invention.

[0054] Based on one of the following technical objectives, this embodiment of the invention constructs a ternary co-culture model for immune rejection of tongue squamous cell carcinoma: (1) To solve the problem that conventional tongue squamous cell carcinoma PDO models lack CAF and NK cells, and cannot reconstruct the tumor stromal barrier and the process of immune cell homing; (2) To solve the problem that the initial spatial relationship of the existing mixed co-culture system is unstable and different cell components are mixed together, making it impossible to evaluate the killing effect of NK migration, crossing the CAF barrier and entering PDO separately; (3) To address the problem that the geometric structure and cell ratio of co-culture models vary greatly among different experimental personnel and batches, making it difficult to achieve standardization, reproducibility evaluation, and cross-comparison of drugs; (4) Provide a model and kit that can be performed entirely in vitro and is compatible with live cell imaging, confocal imaging, flow cytometry and molecular detection, for evaluating candidate interventions that improve immune cell homing and anti-tumor effects in tongue squamous cell carcinoma.

[0055] In the following embodiments, a reproducible in vitro tongue squamous cell carcinoma immune rejection model was established by spatial partitioning and sequential seeding of different cell components. Accompanying kits, construction methods and quantitative evaluation indicators were developed for in vitro screening and functional evaluation of candidate drugs, nucleic acid preparations, antibodies, cell therapy products and other immunomodulatory methods.

[0056] In the following embodiments, all materials used were commercially available. The tongue squamous cell carcinoma tissue sample was obtained from a patient who underwent surgery for tongue squamous cell carcinoma at the Department of Stomatology, First Affiliated Hospital of Nanchang University in July 2025. GW4869 (BeyoEx) TM Exosome inhibitors were purchased from Shanghai Beyotime Biotechnology Co., Ltd., catalog number S1971. Anti-TGF-β was purchased from R&D Systems, catalog number MAB1835, Clone 1D11.

[0057] Example 1: Preparation of PDO, LRRN4-positive CAF and NK cells from tongue squamous cell carcinoma Fresh squamous cell carcinoma tissue of the tongue was harvested from vitro. Significant necrotic areas and adipose tissue were removed, and the tissue was mechanically minced and digested using a tissue digestion solution containing collagenase, dispersant enzyme, and DNase I. The digested tumor tissue was filtered through a screen to obtain tumor cell clusters. These tumor cell clusters were then embedded in a three-dimensional matrix and cultured using a tongue squamous cell carcinoma PDO culture system.

[0058] PDOs with a diameter of 80-250 μm, intact boundaries, and no obvious central necrosis were selected for subsequent model construction. Bright-field observation, H&E, pan-CK immunostaining, and Ki-67 immunostaining were used to identify the PDOs, and they were compared with the corresponding primary tumor tissues of the patients to confirm that the PDOs retained the main histological features of the primary tumor.

[0059] Cells were isolated from the digestate of tongue squamous cell carcinoma tissue from the same patient and primary cancer-associated fibroblasts (CAFs) were obtained through adherent culture. The expression of FAP, α-SMA, PDGFRβ, COL1A1, and LRRN4 in the obtained CAFs was detected, and their negative expression of pan-CK and CD45 was confirmed. Subsequently, flow cytometry was used to enrich LRRN4-positive CAFs, obtaining a CAF population with an LRRN4 positivity rate ≥70%, which was used for the subsequent construction of a ternary co-culture model.

[0060] Peripheral blood was collected from the same patient, and peripheral blood mononuclear cells (PBMCs) were isolated. CD3-negative and CD56-positive NK cells were isolated using a combination of magnetic bead negative and positive selection. Flow cytometry was used to detect the expression of CD3 and CD56 in the obtained NK cells. After confirming that the purity of the NK cells met the experimental requirements, they were used to construct a ternary co-culture model of immune rejection in tongue squamous cell carcinoma.

[0061] Example 2: Construction of a ternary co-culture model of immune rejection in tongue squamous cell carcinoma A ternary co-culture model of tongue squamous cell carcinoma with immune rejection was constructed using 24-well glass-bottomed culture plates. Positioning markers were placed at the center of the bottom of each culture well. Twenty tongue squamous cell carcinoma PDOs prepared in Example 1 were suspended in 20 μL of Matrigel matrix and added dropwise to the center of the bottom of each culture well. The mixture was then gelled at 37 °C for 20 min. After gelation, tongue squamous cell carcinoma PDO culture medium was added, and the mixture was incubated at 37 °C in a 5% CO2 incubator for 24 h to allow the PDOs to form a stable PDO core region in the center of each culture well.

[0062] The LRRN4-positive CAF obtained in Example 1 was processed at a concentration of 1.0 × 10⁻⁶. 6Cells were suspended at a density of 1 cell / mL in type I collagen gel. 80 μL of type I collagen gel containing LRRN4-positive CAFs was concentrically added around the core region of the PDO cell, forming a ring-shaped CAF region approximately 1.0 mm wide continuously surrounding the PDO core region. The gel was incubated at 37 °C for 30 min to allow complete solidification, followed by incubation in co-culture medium for 24 h, allowing the LRRN4-positive CAFs to extend around the PDO and form a stable LRRN4+ CAF outer ring region. The co-culture medium was prepared by mixing tongue squamous cell carcinoma PDO basal culture medium and NK cell maintenance culture medium at a 1:1 volume ratio, with IL-15 added to a final concentration of 10 ng / mL to maintain NK cell survival and basal activity during co-culture.

[0063] The NK cells prepared in Example 1 were labeled with a cell-tracing fluorescent dye. Following an effector-to-target ratio of 2:1 (estimated number of NK cells to PDO tumor cells), the labeled NK cells were added to the culture medium surrounding the outer ring of the LRRN4+CAF, allowing the NK cells to initially distribute outside the CAF barrier. This resulted in a tongue squamous cell carcinoma immune rejection ternary co-culture model consisting of the PDO nuclear region, the outer LRRN4+CAF ring, and the peripheral NK cell migration zone.

[0064] In the ternary co-culture model of tongue squamous cell carcinoma with immune rejection, LRRN4-positive CAF forms a continuous or semi-continuous ring barrier around PDO. In the early stage of NK cell introduction, NK cells are mainly located outside the outer ring of LRRN4+CAF, and then gradually migrate to the inner core region of PDO. This can be used to observe and quantitatively analyze the process of NK cells crossing the CAF barrier and entering the PDO tumor nest.

[0065] Figure 1 The spatial structure of the above-described ternary co-culture model of tongue squamous cell carcinoma with immune rejection is shown. The central region of the figure is the PDO core region (tumor nest-like structure) formed by PDO in tongue squamous cell carcinoma. The LRRN4+CAF outer ring region (LRRN4-positive CAF) is located around the PDO and forms a ring barrier. NK cells migrate from the outer region to the PDO core region. This structure is used to demonstrate that the embodiments of the present invention can form a ternary co-culture model with a clear spatial relationship of "PDO tumor nest-CAF barrier-peripheral NK cells" through partitioning and sequential seeding.

[0066] Example 3: Quantitative Analysis of the Model A ternary co-culture model of tongue squamous cell carcinoma with immune rejection was constructed according to the method in Example 2. The model was then cultured in a 37 ℃, 5% CO2 incubator for 48 h. After NK cells were added to the model, live-cell bright-field and fluorescence imaging were performed at 0 h, 12 h, 24 h and 48 h, respectively. At 48 h, a parallel culture model was set up for endpoint fixation and confocal imaging.

[0067] The PDO tumor region was delineated and segmented using pan-CK fluorescence signals; LRRN4-positive CAFs were identified using LRRN4 and FAP double-positive fluorescence signals, and the outer ring region of LRRN4+CAFs located around the PDO was defined; NK cells were identified using NK cell signals that had been pre-labeled with fluorescence tracers as described in Example 2.

[0068] The outer edge of the pan-CK positive PDO region was used as the PDO boundary, and a 200 μm outer analysis area was set outside the PDO boundary. The number of NK cells inside the PDO and the number of NK cells within the 200 μm area outside the PDO boundary were counted separately, and the shortest distance between each NK cell and the nearest PDO boundary was calculated. The NK cell homing ratio was calculated using the following formula: NK cell homing rate (%) = number of NK cells inside PDO ÷ (number of NK cells inside PDO + number of NK cells in the 200 μm region outside the PDO boundary) × 100%.

[0069] The mean boundary distance of NK cells is the average of the shortest distances from each NK cell in the analysis area to the nearest PDO boundary, which is used to evaluate the degree of migration and approach of NK cells to PDO tumor nests.

[0070] The periphery of the PDO was further determined using the pan-CK signal, and a ring-shaped analysis region was established within 50 μm outside this boundary. The length of continuous coverage of the PDO boundary by LRRN4 and FAP double-positive CAFs within this region was statistically analyzed, and the CAF barrier index was calculated using the following formula: CAF barrier index (%) = length of PDO boundary of continuous coverage by LRRN4 and FAP double-positive CAF ÷ total length of PDO boundary × 100%.

[0071] The CAF barrier index is used to reflect the continuity of the barrier structure formed by LRRN4-positive CAFs around PDO.

[0072] After co-culturing for 48 h, the viability and death of PDO cells were evaluated by live / dead dual fluorescence staining. The PDO mortality rate was calculated based on the proportion of fluorescent signal from dead cells in the PDO region to the total fluorescent signal of cells, in order to evaluate the tumor-killing effect produced by NK cells after entering the PDO.

[0073] After culture, NK cells were recovered, and the expression levels of CD107a, GZMB, PRF1, and IFN-γ were detected by flow cytometry to evaluate the degranulation ability and cytotoxic effect of NK cells in the model.

[0074] Figure 2This is a bright-field image of live cells in a ternary co-culture model of tongue squamous cell carcinoma with immune rejection. This image was used to observe the morphology and integrity of PDOs during the model culture process, as well as the structural stability of the surrounding CAF region. It demonstrates that after adopting the partitioned sequential culture method, PDOs can maintain a relatively stable tumor nest-like structure during co-culture, and no obvious structural disintegration occurred in the surrounding CAF region, providing a stable spatial basis for subsequent NK cell migration and immune rejection evaluation.

[0075] Figure 3 This is a confocal image showing the parallel endpoints of a ternary co-culture model of tongue squamous cell carcinoma with immune rejection after 0-72 h of culture. Blue represents cell nuclei (DAPI), cyan represents PDO tumor cells (pan-CK), red represents LRRN4 and FAP double-positive CAF cells, and green represents fluorescently labeled NK cells. This image illustrates the spatial relationships between PDO cells, LRRN4-positive CAF cells, and NK cells, demonstrating that the model can form a partitioned structure consisting of a central PDO nucleus region, a peripheral LRRN4+CAF outer ring region, and a peripheral NK cell migration zone. It also allows direct observation of the migration and homing of NK cells from the peripheral region into the CAF barrier and PDO interior.

[0076] Figure 4 This figure shows the quantitative analysis results of NK cell spatial migration and homing in the above-mentioned ternary co-culture model of tongue squamous cell carcinoma with immune rejection, including the homing rate of NK cells and the average distance of NK cells from the PDO boundary. The results show that with prolonged culture time, the homing rate of NK cells gradually increases, while the average distance of NK cells from the PDO boundary gradually decreases, indicating that NK cells can gradually migrate from the periphery of the model towards the inner core of the PDO and enter the PDO tumor region. These results further demonstrate quantitatively that the ternary co-culture model constructed in this invention can dynamically reflect the spatial process of NK cell migration and homing from the periphery to the tumor nest, providing a quantitative basis for subsequent evaluation of the effects of different CAF conditions or candidate interventions on NK cell migration and homing ability.

[0077] Example 4: Effects of partitioned sequential culture structure and LRRN4-positive CAF on immune rejection phenotype To verify the effect of sequential culture on the formation of model spatial structure and to evaluate the influence of LRRN4-positive CAF on NK cell migration and nesting, tongue squamous cell carcinoma PDO, LRRN4-positive CAF and NK cells prepared in Example 1 were used to construct different co-culture models according to the conditions described in Example 2, and imaging and quantitative analysis were performed using the method described in Example 3.

[0078] 1. Experimental grouping of CAF models with different LRRN4 expression states Set up the following 4 groups: Group A (LRRN4 high expression CAF group): The PDO core region was formed according to the method described in Example 2; CAFs with an LRRN4 positive rate of ≥70% were used and suspended in the second three-dimensional matrix at the same seeding density as in Example 2, and a ring-shaped CAF region was formed around the PDO core region. After culturing for 24 h, the CAFs extended around the PDO and formed an LRRN4+CAF outer ring region; fluorescently labeled NK cells were added from the outside of the LRRN4+CAF outer ring region.

[0079] Group B (Naturally Low-Expression CAF Group): Primary CAFs without gene intervention and whose LRRN4 expression level was confirmed to be lower than that of Group A by LRRN4 expression detection were selected. All other PDO seeding conditions, CAF seeding density, second and third-dimensional matrix types and volumes, CAF pre-culture time, NK cell addition location and culture conditions were kept the same as those of Group A.

[0080] Group C (siLRRN4-CAF group): A CAF region surrounding the PDO was constructed using CAF cells with LRRN4 knockdown treatment. Following the same partitioning sequence as Group A, a CAF region was formed around the PDO. After 24 h of culture, fluorescently labeled NK cells were added from the outside of the CAF region. Specifically, LRRN4 knockdown was performed using siRNA-mediated CAF cells with high LRRN4 expression, and the reduction in LRRN4 expression compared to pre-treatment was confirmed using appropriate detection methods before model construction.

[0081] Group D (no CAF group): The PDO core region was formed using the same method as the other groups, but without the addition of CAF. To minimize the influence of the second three-dimensional matrix itself, an equal volume of CAF-free second three-dimensional matrix was added around the PDO as a matrix control. After the same pre-culture time as the other groups, the same number of fluorescently labeled NK cells were added to the corresponding peripheral region.

[0082] Except for the presence status of CAF and the expression level of LRRN4, the quantity and size range of PDO used in each group, the types and volumes of the first and second three-dimensional matrices, the CAF seeding density (except for the group without CAF), the time of NK cell addition, the effector-target ratio of NK cells to PDO tumor cells, the composition of the culture medium and the culture conditions were kept as consistent as possible.

[0083] After adding NK cells at an estimated effector-to-target ratio of 2:1 (NK cells to PDO tumor cells), the cells were cultured for 48 h, and live cell imaging was performed at 0 h, 12 h, 24 h, and 48 h. A parallel culture model was also set up, and confocal imaging and spatial quantitative analysis were performed after the culture endpoint was fixed.

[0084] 2. Comparison of live cell morphology and spatial structure in different models Figure 5 Bright field images of representative live cells after 48 h of co-culture of tongue squamous cell carcinoma with PDO under different CAF conditions.

[0085] In the LRRN4-high expression CAF group, PDO maintained a relatively complete organoid morphology. Numerous spindle-shaped or elongated spindle-shaped CAFs were observed around the PDO in a relatively dense, continuous ring or semi-ring arrangement, forming a distinct peripheral enveloping structure. In the naturally low expression CAF group, CAF distribution was still observed around the PDO, but its arrangement was relatively loose, and the degree of continuous enveloping was reduced compared to the LRRN4-high expression CAF group. In the siLRRN4-CAF group, the aggregation and continuous enveloping of CAFs around the PDO were further weakened, and the ring barrier-like structure was less pronounced than in the LRRN4-high expression CAF group. In the CAF-free group, no continuous enveloping structure composed of spindle-shaped fibroblasts was observed around the PDO.

[0086] The above results indicate that, under the same culture conditions, the presence of CAF can form a certain enveloping spatial structure around PDO. Among them, CAF with high LRRN4 expression exhibits more obvious peripheral aggregation and continuous enveloping characteristics. After reducing LRRN4 expression in CAF, the barrier-like structure weakens accordingly. Figure 5 This study supports the role of LRRN4-positive CAF in the formation of the PDO peripheral barrier structure at the live cell morphology level, and provides a morphological basis for subsequent evaluation of its effects on NK cell migration and homing.

[0087] 3. Effects of different LRRN4 expression states on the spatial distribution of NK cells by CAF. Figure 6 Endpoint confocal images and representative 3D reconstructions and orthogonal sections of tongue squamous cell carcinoma PDO co-culture models after 72 h of culture under different CAF conditions. Blue represents cell nuclei (DAPI), cyan represents pan-CK positive PDO tumor regions, red represents FAP positive CAF, and green represents fluorescently labeled NK cells.

[0088] In the LRRN4-overexpressing CAF group, the CAF mainly forms a relatively dense, continuous or semi-continuous ring-shaped barrier-like structure around the periphery of PDO; most NK cells are distributed outside the CAF region, and only a small number of NK cells can be seen inside the CAF barrier and within the PDO region, showing obvious spatial repulsion characteristics.

[0089] In the naturally low CAF expression group, CAF distribution can still be observed around PDO, but its continuity and density are weakened compared to the LRRN4 high expression CAF group. At the same time, more NK cells can be seen approaching the PDO boundary, and some NK cells enter the PDO region.

[0090] In the siLRRN4-CAF group, the CAF surrounding the PDO became further looser and discontinuous, and the distribution of green NK cells around and inside the PDO significantly increased. Three-dimensional reconstruction and XZ and YZ orthogonal sections further showed that, compared with the LRRN4-high expression CAF group, NK cells in the siLRRN4-CAF group were more likely to cross the outer CAF region and enter the PDO interior.

[0091] In the group without CAF, no CAF barrier-like structure was formed around the PDO, and NK cells could approach the PDO more directly and were more distributed around and inside the PDO, further suggesting that the peripheral CAF structure has a certain restrictive effect on NK cell entry into the PDO.

[0092] The above results indicate that the spatial barrier formed by CAF around PDO is related to the restriction of NK cell migration to PDO and nesting. The peripheral barrier structure formed by CAF with high LRRN4 expression is more obvious and accompanied by strong NK cell spatial repulsion. After naturally low expression of LRRN4 or knockdown of LRRN4 expression in CAF, the continuity of CAF barrier is weakened, and the migration of NK cells to PDO region and entry into PDO increases.

[0093] therefore, Figure 6 The findings support the correlation between LRRN4 expression status and CAF-mediated spatial repulsion of NK cells, supported by both two-dimensional confocal imaging and three-dimensional spatial reconstruction. Reducing LRRN4 expression in the CAF weakens the peripheral barrier-like structure and promotes NK cell access to and entry into the PDO tumor region. These results provide spatial imaging evidence for subsequent quantitative evaluation using NK cell homing ratio, average distance from NK cells to the PDO boundary, and the CAF barrier index.

[0094] 4. Quantitative analysis of immune rejection phenotypes in CAF models with different LRRN4 expression states Following the method described in Example 3, quantitative analysis was performed on the CAF coverage area, NK cell nesting ratio, average distance of NK cells to PDO boundary, and PDO mortality rate of each model group.

[0095] Figure 7 The quantitative analysis results of the culture endpoints for each model group are as follows: The average CAF coverage area in the LRRN4 high-expression CAF group, the naturally low-expression CAF group, the siLRRN4-CAF group, and the CAF-free group were 50%, 48%, 40%, and 7%, respectively; the average NK cell homing rate was 10%, 20%, 38%, and 60%, respectively; the average distance from NK cells to the PDO boundary was 115 μm, 82 μm, 52 μm, and 18 μm, respectively; and the average PDO mortality rate was 10%, 18%, 36%, and 60%, respectively.

[0096] The analysis showed that as LRRN4 expression decreased in CAF, the proportion of NK cells entering the nest gradually increased, the average distance from NK cells to the PDO boundary gradually shortened, and the PDO mortality rate increased accordingly. The differences between the relevant groups reached statistical significance (*P<0.05, **P<0.01, ***P<0.001). The differences in CAF coverage area between groups were relatively small.

[0097] The above results indicate that high LRRN4 expression in CAF is associated with a strong NK cell spatial repulsion phenotype; after reducing LRRN4 expression, NK cells are more likely to access and enter the PDO region, accompanied by an enhanced PDO killing effect. Figure 7 This further supports the view that LRRN4-positive CAFs are involved in the formation of the functional barrier surrounding PDO and restrict NK cell nesting.

[0098] 5. Dynamic comparison between the partitioned sequential culture model and the direct mixed culture model To further evaluate the effect of partitioned sequential culture on the maintenance of the model's spatial structure, the same number of PDO, LRRN4-overexpressing CAF and NK cells were used to construct partitioned sequential models and direct mixing models for parallel comparison.

[0099] The two models maintained the same PDO quantity, CAF and NK cell seeding amounts, 3D matrix usage, culture medium composition, and culture conditions, differing only in the spatial construction method of the cells. The direct mixing model directly mixed PDO, CAF, and NK cells into the same 3D culture system at the start of culture; the partitioned sequential model was constructed in the following order: first, the PDO core region was formed; then, the LRRN4+CAF outer ring region was constructed around it; and finally, NK cells were added from outside the CAF barrier.

[0100] Figure 8 Representative confocal images of the two models after 24 h, 48 h, 72 h, and 96 h of culture are shown. In the partitioned sequential model, the PDO core region and the surrounding CAF region can be clearly identified at each time point. The CAF maintains a ring or semi-ring distribution around the PDO. NK cells mainly migrate gradually from the periphery towards the CAF barrier and PDO. With prolonged culture time, some NK cells can be observed to cross the CAF region and enter the PDO, but the overall spatial hierarchy remains relatively clear. In contrast, the direct mixing model shows PDO, CAF, and NK cells mixed together from the early stages of culture. The CAF does not form a stable outer surrounding structure, and there is a lack of clear spatial boundaries between the PDO and CAF regions. Furthermore, the mixed distribution remains quite obvious with prolonged culture time.

[0101] The above results indicate that, compared with direct mixed culture, sequential partitioning culture can maintain a relatively stable spatial structure of "PDO core region - peripheral LRRN4+ CAF outer ring region - peripheral NK cells" over a longer culture period, thus being more conducive to time-series evaluation of the process of NK cells approaching, crossing, and entering the CAF barrier. Therefore, Figure 8 This further supports the fact that the partitioned sequential culture method used in the embodiments of the present invention can reduce the spatial relationship disorder caused by direct mixed culture and improve the observability and evaluability of the spatial process of immune rejection.

[0102] 6. Evaluation of inter-batch repeatability of the partitioned sequential culture model To evaluate the batch-to-batch reproducibility of different model construction methods, three batches of sequential partitioned models and three batches of direct mixed models were constructed independently. Each batch used PDO, LRRN4-overexpressing CAF and NK cells of the same source and quantity, and maintained consistent three-dimensional matrix volume, cell seeding conditions, culture medium composition, and culture time.

[0103] Figure 9 Representative confocal images of three independent batches from both models are shown. The results indicate that in the partitioned sequential model, all three independent batches formed a relatively consistent PDO core region and surrounding CAF structure. The spatial hierarchy among PDO, CAF, and NK cells was relatively clear, and the overall morphology and spatial distribution characteristics of each batch were relatively consistent. In contrast, the direct mixing model showed significant differences in cell spatial distribution between different batches. PDO, CAF, and NK cells exhibited varying degrees of random mixing, and the distribution position and continuity of CAF around PDO varied considerably, resulting in lower batch-to-batch consistency in the overall spatial structure.

[0104] The above results indicate that by pre-defining the initial spatial locations of PDO, CAF, and NK cells and adopting a sequential seeding method, the spatial structure fluctuations caused by direct random mixing can be reduced. This allows the model to maintain a relatively stable spatial relationship of "PDO core region - peripheral CAF region - peripheral NK cells" in different independent experimental batches, thereby improving the reproducibility and standardization of model construction. Therefore, Figure 9 The partitioned sequential culture method supported by this invention has better batch-to-batch spatial structure repeatability than direct mixed culture.

[0105] 7. Comprehensive quantitative comparison of the partitioned sequential model and the direct mixed model Figure 10 The results present the combined quantitative analysis of the partitioned sequential model and the direct mixture model. The results show that the average spatial partition identification rate of the partitioned sequential model is 81%, significantly higher than the 40% of the direct mixture model; the average CAF boundary coverage rates are 74% and 35%, respectively, with the partitioned sequential model showing a significantly higher rate.

[0106] In terms of model stability, the average 72-hour model integrity rate of the partitioned sequential model was 79%, which was higher than the 48% of the direct mixed model; the average 96-hour model integrity rates were 78% and 35%, respectively, and the differences were statistically significant.

[0107] In terms of batch reproducibility, the average coefficient of variation of PDO area in the partitioned sequential model was 9%, which was significantly lower than 28% in the direct mixing model; the average coefficients of variation of NK cell homing proportion were 14% and 29%, respectively, which were also significantly lower in the partitioned sequential model.

[0108] The above results indicate that, compared with direct mixed culture, partitioned sequential culture can form clearer spatial partitions and more stable CAF outer boundaries, and maintain high model integrity during 72 h and 96 h of culture. At the same time, the inter-batch variation of PDO area and NK cell homing ratio is lower, indicating that this culture method has better stability and reproducibility.

[0109] Example 5: Application of a ternary co-culture model of immune rejection in tongue squamous cell carcinoma for drug screening To verify whether the tongue squamous cell carcinoma immune rejection ternary co-culture model constructed in this embodiment of the invention can be used to evaluate interventions that improve NK cell migration and nesting and enhance anti-tumor effects, exosome release inhibitor GW4869 and TGF-β neutralizing antibody (anti-TGF-β) were selected as actual interventions to verify the function of the model.

[0110] 1. Establishment of the ternary co-culture model and experimental grouping PDO cells derived from three patients with tongue squamous cell carcinoma were selected. PDO, LRRN4-positive CAF, and NK cells were prepared according to the method described in Example 1, and a ternary co-culture model was constructed according to the partitioned sequential culture method described in Example 2. First, the PDO core region was formed, and then an LRRN4+CAF outer ring region was constructed around the PDO. After the CAF barrier structure was formed, the corresponding intervention was added, and fluorescently labeled NK cells were added from the outside of the CAF barrier.

[0111] The experiment used a 2×2 factorial design, with the following 4 groups: Group A (vector control group): The corresponding Vehicle and isotype control antibody were added to the experimental group, but GW4869 and anti-TGF-β were not added. Group B (GW4869 group): GW4869 was added to a final concentration of 10 μmol / L; Group C (anti-TGF-β group): anti-TGF-β was added at a final concentration of 10 μg / mL; Group D (Combined Intervention Group): GW4869 and anti-TGF-β were added simultaneously at a final concentration of 10 μmol / L.

[0112] Except for the intervention factors, the number of PDOs, the type and volume of the three-dimensional matrix, the seeding density of LRRN4-positive CAFs, the NK cell effector-to-target ratio, the composition of the culture medium, and the culture conditions were kept consistent across all groups. Each patient-derived model was independently constructed and replicated, using PDOs from three patients. NK cells and the corresponding interventions were added and cultured for another 72 h.

[0113] 2. Effects of GW4869 and anti-TGF-β on NK cell spatial migration and homing Confocal imaging was performed on each group of models after 72 h of culture. Figure 11 The images show the maximum three-dimensional projection and XZ and YZ orthogonal sections for each treatment group. Cyan represents PDO tumor regions, red represents CAF, and green represents fluorescently labeled NK cells.

[0114] In the vector control group, the CAF mainly formed a relatively obvious barrier-like structure around the PDO. Most NK cells were distributed outside the CAF region, and only a small number of NK cells were visible in the CAF region and inside the PDO, showing a relatively obvious spatial repulsion feature.

[0115] Compared with the vector control group, both the GW4869 group and the anti-TGF-β group showed that more NK cells approached and entered the outer ring region of LRRN4+CAF and the core region of PDO. Orthogonal XZ and YZ sections further showed that the spatial distribution of NK cells towards the PDO core increased after individual intervention.

[0116] The changes were more pronounced in the combined intervention group. The number of NK cells outside the CAF barrier was relatively reduced, while more NK cells were observed in the CAF region and inside the PDO region, suggesting that the combined intervention of GW4869 and anti-TGF-β can further promote NK cells to cross the outer CAF barrier and enter the PDO region.

[0117] The above results indicate that the ternary co-culture model constructed in this embodiment of the invention can intuitively display the spatial changes of NK cell migration from the periphery to the CAF barrier and PDO interior under different intervention conditions through three-dimensional maximum projection and orthogonal section, and can be used to evaluate the effect of candidate intervention measures on improving NK cell migration and homing.

[0118] 3. Quantitative analysis of immune rejection phenotype after intervention Following the method of Example 3, the PDO region was determined by the pan-CK positive signal, the peripheral CAF region was determined by the CAF-related fluorescence signal, and NK cells were identified by the fluorescence tracer signal. The NK cell nesting ratio, the average distance from NK cells to the PDO boundary, the PDO mortality rate, the continuity of the CAF layer, and the CAF coverage density were calculated for each group.

[0119] Figure 12 The results of the quantitative analysis of the above indicators for each treatment group are shown below. Each point represents an independently established model, and the data are expressed as mean ± SEM. In the statistical markers, * indicates P < 0.05, ** indicates P < 0.01, *** indicates P < 0.001, and **** indicates P < 0.0001.

[0120] The results showed that the average NK cell homing rates in the vector control group, GW4869 group, anti-TGF-β group, and combined intervention group were 16%, 27%, 43%, and 63%, respectively; the average distances from NK cells to the PDO boundary were 112 μm, 82 μm, 53 μm, and 35 μm, respectively; and the average PDO mortality rates were 10%, 24%, 42%, and 61%, respectively.

[0121] Meanwhile, the average CAF layer continuity was 56%, 50%, 30%, and 20% in each group, and the average CAF coverage density was 52%, 47%, 23%, and 15%, respectively. Overall, GW4869 or anti-TGF-β treatment alone could attenuate CAF barrier-related phenotypes and promote NK cell migration to the PDO region to varying degrees, with anti-TGF-β showing a more significant effect. The combined intervention resulted in the most pronounced changes, manifested as a further increase in the proportion of NK cells homing, a further shortening of the distance to the PDO boundary, a decrease in CAF layer continuity and CAF coverage density, and an increase in PDO mortality. The differences between the relevant groups reached [data missing]. Figure 12 The statistical significance level is shown.

[0122] The above results indicate that GW4869 and anti-TGF-β intervention can improve the spatial exclusion phenotype of NK cells in the model to varying degrees, with combined intervention showing a more significant improvement trend. Therefore, Figure 12 Furthermore, it is demonstrated that the model in this embodiment of the invention can simultaneously quantitatively evaluate the intervention effect from three aspects: CAF barrier structure, NK cell migration into nests, and PDO killing effect, rather than using only the endpoint tumor cell death as the evaluation indicator. Therefore, it can be used for functional screening and comparison of candidate immune microenvironment intervention measures.

[0123] 4. Evaluation of NK cell function after intervention After 72 h of culture, NK cells from each group were recovered, and flow cytometry was used to detect NK cell function-related indicators CD107a, GZMB, and IFN-γ. Among them, CD107a was used to reflect the level of NK cell degranulation, GZMB was used to evaluate the expression of NK cell cytotoxic effector molecules, and IFN-γ was used to evaluate the effector function status of NK cells.

[0124] Figure 13 This section presents the gating strategy and representative results for NK cell flow cytometry detection. The gating was performed sequentially using single-cell and live-cell methods, and NK cells were identified based on markers such as CD45, CD3, and CD56. The proportions of CD107a, GZMB, and IFN-γ-positive cells were further detected.

[0125] Figure 14 The results show the quantitative results of NK cell function indicators in each treatment group. The average proportions of CD107a-positive NK cells in the vector control group, GW4869 group, anti-TGF-β group, and combined intervention group were 18%, 28%, 38%, and 55%, respectively; the average proportions of GZMB-positive NK cells were 7%, 12%, 16%, and 21%, respectively; and the average proportions of IFN-γ-positive NK cells were 7%, 12%, 17%, and 22%, respectively.

[0126] The results showed that, compared with the vector control group, intervention with GW4869 or anti-TGF-β alone could increase the positive rates of CD107a, GZMB, and IFN-γ in NK cells to varying degrees, with a more significant increase in the combined intervention group. The differences between the relevant groups reached [number missing]. Figure 14 The statistical significance level is shown.

[0127] The above results indicate that GW4869 and anti-TGF-β intervention not only improve NK cell migration and homing to the PDO region, but also enhance NK cell degranulation, expression of cytotoxic effector molecules, and secretion of effector factors, with the combined intervention showing a more significant effect. Therefore, Figure 13 and Figure 14 Furthermore, it is demonstrated that the ternary co-culture model of the present invention can simultaneously evaluate the effects of candidate interventions on NK cell spatial migration and anti-tumor effector function, thereby achieving a multi-dimensional evaluation of the improvement effect of immune rejection.

[0128] 5. Validation of direct cytotoxicity under NK cell-free conditions To clarify whether the increased mortality rate of PDO after intervention is mainly mediated by NK cells, rather than by the direct cytotoxic effect of GW4869 or anti-TGF-β on PDO, a control experiment without NK cells was also set up.

[0129] A PDO-LRRN4-positive CAF culture system was established according to the above method, but without the addition of NK cells. After the outer structure of the CAF was formed, a vector control group, a GW4869 group, an anti-TGF-β group, and a combined intervention group were set up. The final concentration of GW4869 was 10 μmol / L, and the final concentration of anti-TGF-β was 10 μg / mL. The culture conditions and intervention time of each group were consistent with those of the experiment containing NK cells.

[0130] After 72 h of culture, the PDO mortality rate was detected according to the method described in Example 3.

[0131] Figure 15 The results showed that the PDO mortality rate in each treatment group under NK cell-free conditions was low. The results indicated that the PDO mortality rate remained at a low level in all groups; no significant increase in PDO mortality was observed with GW4869 alone, anti-TGF-β alone, or in combination. There was no statistically significant difference between the combined intervention group and the vector control group (ns, P > 0.05).

[0132] The above results indicate that, at the concentrations and treatment times used in the embodiments of this invention, GW4869 and anti-TGF-β did not exhibit significant direct cytotoxic effects against PDO. Combined with the significantly increased PDO mortality rate in the ternary co-culture model containing NK cells, this supports the view that the tumor-killing effect after intervention is mainly related to NK cell involvement, rather than being caused by the intervention directly killing PDO. Therefore, Figure 15 This is used to eliminate confounding effects caused by the direct cytotoxic effects of GW4869 and anti-TGF-β, and further supports the ability of the model of this invention to evaluate NK cell-dependent antitumor effects.

[0133] 6. Effect of GW4869 on extracellular vesicle release in culture system To verify the role of GW4869 in the model, culture supernatant from each treatment group was collected after 72 h of culture. Extracellular vesicles were isolated, and nanoparticle tracking analysis (NTA) was used to detect particle concentration. At the same time, Western blot was used to detect exosome-related marker proteins CD63, CD81, and TSG101.

[0134] Figure 16 The number of exosome-like particles and marker proteins in the culture supernatant of each treatment group were detected. The average supernatant particle concentrations of the vector control group, GW4869 group, anti-TGF-β group, and combined intervention group were 7.5 × 10⁻⁶. 8 4.9×10 8 3.6×10 8 and 2.0×10 8particles / mL. Compared with the vector control group, the supernatant particle concentration was significantly reduced after GW4869 treatment; the particle concentration was further reduced in the combined intervention group. Western blot results showed that CD63, CD81, and TSG101 were detected in the isolated products of each group, supporting the exosome-related characteristics of the detected particles.

[0135] The above results indicate that GW4869 intervention can reduce the release level of exosome-like particles in this co-culture system, consistent with the aforementioned results of increased NK cell homing and improved immune rejection phenotype. Therefore, Figure 16 The actual intervention effect of GW4869 in this model was verified at the exosome release level, and it supports the use of the model of this invention to evaluate candidate interventions targeting CAF-related intercellular communication pathways.

[0136] 7. Effect of anti-TGF-β on TGF-β / Smad signal in CAF To verify the actual role of anti-TGF-β in this model, CAFs from each group were recovered after 72 h of culture. The expression of p-Smad2 / 3 (S465 / 467), total Smad2 / 3, and GAPDH was detected by Western blot, and the relative ratio of p-Smad2 / 3 to total Smad2 / 3 was calculated to evaluate the degree of TGF-β / Smad signal activation.

[0137] Figure 17 The results show the detection of TGF-β / Smad signals in the CAF of each treatment group. The average p-Smad2 / 3 and Smad2 / 3 values ​​in the vector control group, GW4869 group, anti-TGF-β group, and combined intervention group were 0.92, 0.55, 0.31, and 0.20, respectively. Western blot results showed that the total Smad2 / 3 and GAPDH levels remained stable in all groups, while the p-Smad2 / 3 level gradually decreased with intervention.

[0138] Compared with the vector control group, the p-Smad2 / 3 and Smad2 / 3 ratios were decreased in the GW4869 group (*P<0.05); the decrease was more significant in the combined intervention group. The anti-TGF-β treatment group also showed a significant decreasing trend in p-Smad2 / 3.

[0139] The above results indicate that anti-TGF-β can inhibit the activation of TGF-β / Smad signaling in the CAF of this model, manifested as a decrease in Smad2 / 3 phosphorylation levels; this signal inhibition is more pronounced after combining with GW4869. Therefore, Figure 17The study validated the actual biological role of anti-TGF-β in the ternary co-culture system at the molecular signaling level, and provided corresponding mechanistic support for the aforementioned weakening of the CAF barrier, increased NK cell homing, and enhanced PDO killing.

[0140] 8. Cell viability evaluation under intervention conditions To further rule out the widespread non-specific cytotoxic effects caused by GW4869 and anti-TGF-β, the ternary co-culture systems of each group were recovered after 72 h of culture. Single-cell suspensions were prepared by mild digestion, and the cell viability of PDO, CAF, and NK cells was detected by flow cytometry combined with fixable live / dead cell dye (FVD). PDO, CAF, and NK cells were identified by their respective surface markers, and the proportion of FVD-negative cells in each cell population was used as the cell viability.

[0141] Figure 18 The cell viability after 72 h of culture in each treatment group was statistically analyzed. The results showed that the average viability of PDO cells in the vector control group, GW4869 group, anti-TGF-β group, and combined intervention group were 73%, 74%, 74%, and 72%, respectively; the average viability of CAF cells were 72%, 69%, 71%, and 69%, respectively; and the average viability of NK cells were 70%, 72%, 69%, and 69%, respectively. The overall cell viability was similar across groups, and there was no statistically significant difference between the vector control group and the combined intervention group.

[0142] The above results indicate that, at the concentrations and treatment times used in the embodiments of this invention, GW4869 and anti-TGF-β did not induce significant non-specific cell death in PDO, CAF, or NK cells. Therefore, the aforementioned changes in the CAF barrier, NK cell migration into nests, and effector function cannot be simply attributed to the generalized cytotoxicity caused by the intervention.

[0143] Example 6: Validation of the reproducibility and consistency with primary tumor tissue of a ternary co-culture model of tongue squamous cell carcinoma with immune rejection from different patients. To verify the reproducible construction capability of the tongue squamous cell carcinoma immune rejection ternary co-culture model in samples from different patients, and to evaluate whether the model can reflect biological differences among patients while maintaining basic spatial structure stability, PDO, CAF and NK cells from multiple tongue squamous cell carcinoma patients were used for independent modeling, and further compared with the relevant spatial features of the corresponding primary tumor tissues.

[0144] 1. Patient source and model establishment Six patients with squamous cell carcinoma of the tongue (P1-P6) were selected, and fresh squamous cell carcinoma tissue and peripheral blood of the corresponding patients were obtained.

[0145] Following the method in Example 1, PDO (partially oxidized fibroblastic ovarian cells) were established from the tumor tissues of each patient, and CAF (corporeal cell line) was isolated and cultured. LRRN4-positive CAF was obtained after LRRN4 expression detection and enrichment. NK cells were also isolated from the peripheral blood of the corresponding patients. Therefore, the PDO, LRRN4-positive CAF, and NK cells in the model constructed for each patient all originated from the corresponding sample of that patient.

[0146] A ternary co-culture model was established according to the partitioned sequential culture method in Example 2. All patient-derived models used the same culture plates, three-dimensional matrix, PDO seeding conditions, CAF seeding density, NK cell addition method, and culture time. PDO, CAF, and NK cells were constructed at an estimated cell ratio of 1:1:2.

[0147] Three parallel models were independently built for each patient, designated Rep1, Rep2, and Rep3, for a total of 18 independent models. Each parallel model underwent independent culture, imaging, and subsequent quantitative analysis; different duplicate models were not combined for testing.

[0148] 2. Spatial structure and repeatability observation of models from different patient sources Confocal imaging was performed on each model after 72 hours of culture. Figure 19 Representative three-dimensional maximum projection images and XZ and YZ orthogonal cross-sectional images of ternary co-culture models from six patients are shown. Three independent replicate models (Rep1-Rep3) are displayed for each patient. In the figures, cyan represents the PDO tumor region, red represents the surrounding CAF region, and green represents NK cells. The results show that the P1-P6 derived models can all form a basic spatial structure with the PDO as the center, CAF mainly distributed around the PDO, and NK cells migrating from the periphery inward. Among Rep1-Rep3 from the same patient, there is good consistency in the overall morphology of the PDO, the surrounding pattern of the CAF, and the spatial distribution of NK cells, indicating that the partitioned sequential culture method of this invention can stably reconstruct the above spatial relationships in independent replicate experiments.

[0149] At the same time, some differences can still be observed between models from different patients, including the morphology and size of PDO, the degree of CAF encirclement, and the degree to which NK cells approach and enter the PDO region. This suggests that while maintaining the stability of the basic construction pattern, the models can still exhibit a certain degree of inter-patient heterogeneity.

[0150] therefore, Figure 19This indicates that the ternary co-culture model of the present invention can stably establish the basic spatial structure of "PDO core region - peripheral CAF region - NK cell migration region" in samples from different patients and in independent replicate experiments, while preserving the spatial phenotypic differences between different patient-derived models, and has good reproducibility and patient-derived feature retention capabilities.

[0151] 3. Intra-patient technical repeatability evaluation To evaluate the reproducibility of models from the same patient, the following five parameters were measured for each patient in three independent models: PDO volume, CAF encapsulation continuity, NK cell homing rate, PDO viability, and PDO mortality. For each parameter for each patient, the coefficient of variation (CV) was calculated based on the measurements from Rep1, Rep2, and Rep3. CV (%) = Standard deviation ÷ Mean × 100%.

[0152] Then, the CVs for the corresponding indicators of the 6 patients were summarized. Further, the intra-patient CVs for each indicator of the 6 patients (P1–P6) were summarized to evaluate the technical repeatability of the model in samples from different patient sources.

[0153] Figure 20 The coefficient of variation (CV) distribution for intra-patient technical reproducibility across five parameters in models from six patients is shown. The x-axis, P1-P6, represents six different patients; scatter plots for the same patient represent the CV values ​​calculated by different independent evaluation units based on Rep1, Rep2, and Rep3. The central horizontal line and error bars represent the summative statistics and dispersion of the corresponding CVs. Table 1 presents the statistical results of the CV for intra-patient technical reproducibility. The results show that the average CV for intra-patient technical reproducibility of PDO volume, CAF encapsulation continuity, NK cell homing rate, PDO viability, and PDO mortality is all below 20%, indicating that independent models from the same patient have good reproducibility.

[0154] Table 1. Intra-patient reproducibility of ternary co-culture models from different patient sources 4. Individual differences in models from different patient sources Combination Figure 19 and Figure 20 It can be seen that, while maintaining good intra-patient technical reproducibility, the ternary co-culture models from different patient sources can still show certain phenotypic differences, including differences in indicators such as CAF encapsulation continuity, NK cell homing degree, PDO viability, and PDO mortality.

[0155] Figure 20Further evidence indicates that the overall variability arising from technical replication of models from the same patient source is relatively small, with the average intra-patient CV for all indicators below 20%. Therefore, after excluding or minimizing the impact of technical fluctuations during model establishment and testing, Figure 19 The differences in spatial structure and functional phenotypes observed between models from different patients are more likely to reflect the differences in the biological characteristics and interactions of tumor cells, CAFs, and immune cells from different patients, rather than simply being caused by random fluctuations in the experiment.

[0156] The above results demonstrate that the patient-derived ternary co-culture model constructed in this embodiment of the invention has good technical repeatability and stability, and can retain biological heterogeneity among samples from different patients to a certain extent. Therefore, it can be used to evaluate the characteristics of the tumor microenvironment and individualized functional phenotypes from patient sources.

[0157] 5. Comparison of the spatial structure of the ternary co-culture model and the corresponding primary tumor tissue To evaluate whether the ternary co-culture model constructed in this embodiment of the invention can retain the spatial distribution characteristics of tumor nests, LRRN4-positive CAFs, and NK cells in the primary tumor tissue of the corresponding patients, parallel analysis was performed on the models from the above 6 patients and their corresponding primary tumor tissues.

[0158] Primary tumor tissue was analyzed using multiplex immunofluorescence assays, with pan-CK used to identify the tumor epithelial region, LRRN4 and FAP used to identify CAF, and CD56 used to identify NK cells. Spatial imaging of PDO, CAF, and NK cells was performed on the model side using the corresponding markers. Three independent models were established for each patient, and the average of the three independent models was used as the model measurement values ​​for that patient in subsequent quantitative analyses.

[0159] Figure 21 Representative cases P1, P3, and P6 were selected for demonstration, representing low, medium, and high NK cell homing phenotypes, respectively. The spatial characteristics of the primary tumor tissues and the PDO ternary co-culture model were compared after 72 h of culture. Results showed that in the primary tumor tissues of P1 patients, CAF formed a relatively obvious surrounding structure around the tumor nests, with fewer NK cells entering the tumor area; the corresponding PDO model also showed significant peripheral CAF encirclement and a low level of NK cell homing. The primary tissues and corresponding model of P3 patients showed moderate CAF encirclement and NK cell homing. In the primary tissues of P6 patients, NK cells were more distributed within the tumor area, and the corresponding PDO model also showed a high level of NK cell homing.

[0160] Differences in the degree of CAF encapsulation and NK cell spatial infiltration were observed among different patients, while models from the same patient and their corresponding primary tumor tissues generally showed similar spatial phenotypic trends. Therefore, Figure 21 This indicates that the ternary co-culture model of the present invention can reproduce the spatial relationship between "tumor nest-peripheral CAF-NK cells" in the primary tumor tissue of the corresponding patient to a certain extent, and preserve the differences in immune rejection phenotypes among different patients.

[0161] 6. Quantitative comparison of CAF parcel continuity Tumor nest regions in the PDO model or primary tumor tissue were identified using pan-CK positive signals, and the boundaries of the tumor nests were delineated. A ring-shaped analysis region was established within 50 μm outside the tumor nest boundary. LRRN4-positive CAFs were identified based on the double-positive signals of LRRN4 and FAP, and the length of continuous CAF wrapping along the tumor nest boundary was calculated. The continuity of CAF wrapping was calculated using the following formula: CAF encapsulation continuity (%) = length of tumor nest boundary continuously encapsulated by LRRN4 and FAP double-positive CAF ÷ total perimeter of tumor nest × 100%.

[0162] The continuity of CAF encapsulation in the primary tumor tissue and the corresponding ternary co-culture model was calculated for each patient. The average of the three independent models for each patient was used as the model measurement value for that patient.

[0163] 7. Quantitative comparison of the proportion of NK cell spatial infiltration The tumor region in the PDO model or primary tumor tissue was determined based on the pan-CK positive signal, and NK cells were identified based on the CD56 positive signal. Based on confocal Z-stack images, the NK cell signal volume within the tumor region and the corresponding total tumor volume were calculated. The NK cell spatial infiltration ratio was calculated using the following formula: NK cell spatial infiltration rate (%) = Volume of CD56 positive NK cell signal within the tumor region ÷ Total volume of the corresponding tumor region × 100%.

[0164] The spatial infiltration ratio of NK cells in the primary tumor tissue and the corresponding ternary co-culture model was calculated for each patient. The average of the three independent models for each patient was also used as the model measurement value for that patient.

[0165] 8. Correlation analysis between the model and spatial parameters of primary tumor tissue The CAF encapsulation continuity and NK cell spatial infiltration ratio of the corresponding primary tumor tissues were obtained for six patients in a ternary co-culture model. The average of the measurement results of the three independent models for each patient was taken as the model measurement value for that patient. Therefore, the independent sample size for the correlation analysis was n=6.

[0166] Spearman rank correlation analysis was used to evaluate the correlation of the above spatial indices between the ternary co-culture model and the corresponding primary tumor tissue.

[0167] Figure 22 For correlation analysis results, each data point represents one patient. The results showed that the proportion of NK cell spatial infiltration in the ternary co-culture model was significantly positively correlated with the proportion of NK cell spatial infiltration in the corresponding primary tumor tissue (Spearman r=0.87, P=0.018); the continuity of CAF encapsulation in the model was also significantly positively correlated with the continuity of CAF encapsulation in the corresponding primary tumor tissue (Spearman r=0.85, P=0.023).

[0168] The above results indicate that the differences in NK cell spatial infiltration and CAF encapsulation characteristics exhibited by the ternary co-culture models from different patients show a good correlation with the corresponding spatial characteristics in the primary tumor tissues of the respective patients. Therefore, Figure 22 From a quantitative perspective, the ternary co-culture model constructed in this embodiment of the invention can, to a certain extent, preserve the individual differences in CAF encapsulation and NK cell spatial infiltration in the primary tumor tissues of different patients, providing a basis for evaluating patient-specific immune rejection phenotypes using this model.

[0169] 9. Consistency analysis between the model and the primary tumor tissue Based on correlation analysis, the Bland-Altman method was further used to evaluate the consistency of spatial index measurements between the ternary co-culture model and the corresponding primary tumor tissue. The analytical indicators included the average distance from the CAF to the tumor nest, the average distance from NK cells to the tumor nest, the proportion of NK cells entering the nest, and the continuity of CAF encapsulation.

[0170] Figure 23 The Bland-Altman concordance analysis results for the four spatial indices are shown above, with each point representing one patient. The horizontal axis represents the mean of the model measurements and the corresponding primary tumor tissue measurements, and the vertical axis represents the difference between the two. The results show that the mean difference between the model and the primary tumor tissue for each spatial indices is close to 0, and all measurements are within the 95% concordance limit, indicating good concordance between the two.

[0171] Combination Figure 21 Representative spatial structures shown and Figure 22 The correlation analysis results further support that the model in this embodiment of the invention can, to a certain extent, preserve the CAF encapsulation characteristics and NK cell spatial distribution differences in the primary tumor tissues of different patients, and has a certain ability to maintain patient-origin characteristics. Therefore, Figure 23 From the perspective of consistency analysis, this invention's ternary co-culture model further supports the feasibility of using it for evaluating patient-derived immune rejection phenotypes and comparing the functions of candidate interventions.

[0172] The above description is merely a preferred embodiment of the present invention. The present invention is not limited to the above-described embodiments. Any embodiment that achieves the technical effects of the present invention by the same or equivalent means should fall within the protection scope of the present invention. Within the protection scope of the present invention, various modifications and variations can be made to the technical solutions and / or implementation methods.

Claims

1. A method for constructing a ternary co-culture model of tongue squamous cell carcinoma with immune rejection, characterized in that, Includes the following steps: The PDO of the tongue squamous cell carcinoma was suspended in the first three-dimensional matrix and formed into the PDO core region; The LRRN4 positive CAF is suspended in a second three-dimensional matrix, forming an LRRN4+CAF outer ring region surrounding the PDO core region; NK cell suspension was added to the periphery of the outer ring region of the LRRN4+CAF, or NK cell suspension was added to the core region of the PDO and / or the outer ring region of the LRRN4+CAF to obtain a tongue squamous cell carcinoma immune rejection ternary co-culture model.

2. The method according to claim 1, characterized in that, The method for constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model includes the following steps: Tongue squamous cell carcinoma PDO was suspended in a first three-dimensional matrix and placed in a culture container for 6-96 h to form the PDO core region; LRRN4-positive CAFs were suspended in a second three-dimensional matrix and placed around the PDO core region. After culturing for 6-96 h, an LRRN4+CAF outer ring region was formed. NK cell suspension was added to the periphery of the outer ring region of the LRRN4+CAF, or NK cell suspension was added to the core region of the PDO and / or the outer ring region of the LRRN4+CAF to obtain a tongue squamous cell carcinoma immune rejection ternary co-culture model.

3. The method according to claim 2, characterized in that, The inoculation density of LRRN4-positive CAFs in the outer ring region of the LRRN4+ CAF can be 0.1 × 10⁻⁶. 6 -10×10 6 The NK cell seeding density can be 0.1 × 10⁶ cells / mL. 6 -10×10 6 per mL.

4. The method according to claim 2, characterized in that, The first three-dimensional matrix and the second three-dimensional matrix are at least one of the following: basement membrane extract, Matrigel, collagen gel, fibrin gel, alginate gel, hyaluronic acid gel, gelatin methacrylamide gel, and polyethylene glycol synthetic hydrogel.

5. The method according to claim 2, characterized in that, The NK cells are replaced with CD8-positive T cells, γδ T cells, CAR-NK cells, or other immune cells with tumor-killing capabilities.

6. The method according to claim 1, characterized in that, The method for constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model includes the following steps: The tongue squamous cell carcinoma PDO was suspended in a first three-dimensional matrix and placed in the central tumor channel of a microfluidic device to form the PDO core region; LRRN4-positive CAF is suspended in a second three-dimensional matrix and placed in a matrix channel of the microfluidic device adjacent to the central tumor channel to form an outer ring region of LRRN4+CAF surrounding the PDO core region. NK cell suspension was added through the outer perfusion channel of the microfluidic device to obtain a ternary co-culture model of tongue squamous cell carcinoma immune rejection.

7. The method according to claim 1, characterized in that, The method for constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model includes the following steps: The PDO of tongue squamous cell carcinoma was suspended in a first three-dimensional matrix, and the PDO core region was formed by three-dimensional printing. The LRRN4 positive CAF was suspended in a second three-dimensional matrix, and the outer ring region of LRRN4+CAF surrounding the PDO core region was formed by three-dimensional printing. NK cell suspension was added to the periphery of the outer ring region of the LRRN4+CAF, or NK cell suspension was added to the core region of the PDO and / or the outer ring region of the LRRN4+CAF to obtain a tongue squamous cell carcinoma immune rejection ternary co-culture model.

8. The method according to claim 1, characterized in that, The method for constructing the tongue squamous cell carcinoma immune rejection ternary co-culture model includes the following steps: The tongue squamous cell carcinoma PDO was suspended in the first three-dimensional matrix and placed in a central circular region enclosed by an isolation structure to form the PDO core region; The LRRN4 positive CAF is suspended in a second three-dimensional matrix and placed in an annular region enclosed by an isolation structure, wherein the annular region is located outside the central circular region, forming an outer ring region of LRRN4+CAF surrounding the PDO core region. Add NK cell suspension to the periphery of the outer ring region of the LRRN4+CAF, or add NK cell suspension to the core region of the PDO and / or the outer ring region of the LRRN4+CAF, to obtain a tongue squamous cell carcinoma immune rejection ternary co-culture model. The isolation structure is made of a soluble or extractable material, and the isolation structure is dissolved or extracted after the PDO core region and LRRN4+CAF outer ring region are stabilized.

9. The application of the tongue squamous cell carcinoma immune rejection ternary co-culture model obtained by the construction method of any one of claims 1-8 in screening drugs for the treatment of tongue squamous cell carcinoma.

10. A kit for constructing a ternary co-culture model of immune rejection in tongue squamous cell carcinoma, characterized in that, The model includes core components for model construction, which include PDO culture medium or its concentrated components, a first three-dimensional matrix, a second three-dimensional matrix, LRRN4-positive CAF or its cryopreservation preparation, and NK cell culture or activation components.

Images