Mouse lymphatic endothelial cells SVEC4-10 (mLECs) were obtained from the American Type Culture Collection. Mouse breast cancer cell 4T1 cells were purchased from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). Human lymphatic endothelial cells (hLECs) were obtained from the PromoCell (C-12216, Heidelberg, Germany). Mouse melanoma cancer cell B16F10 cells were purchased from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). MS1 mouse endothelial cells were obtained from the Cell Bank of the Type Culture Collection of the Chinese Academy of Sciences (Shanghai, China). SVEC4-10, B16F10, and MS1 cells were cultured in Dulbecco’s modified Eagle’s medium (BasalMedia, Shanghai, China) respectively supplemented with 10% FBS (BasalMedia, Shanghai, China), 100 U/mL penicillin and 100 mg/mL streptomycin. Human lymphatic endothelial cells were cultured in EGM-2 medium (CC-3202, Lonza, BSL, Switzerland), and cells within passage numbers 4 to 7 were used for all experiments. 4T1 cells were cultured in RPMI 1640 medium (BasalMedia, Shanghai, China) respectively supplemented with 10% FBS (BasalMedia, Shanghai, China), 100 U/mL penicillin and 100 mg/mL streptomycin. All cells were tested for mycoplasma and were mycoplasma-negative. All cells were cultured in a humidified atmosphere under 5% CO₂.

Polyacrylamide (PA) hydrogels with mechanically tunable stiffness were prepared as previously described (Tse and Engler, 2010). Briefly, coverslips were soaked in 3-aminopropyltriethoxysilane (ST1087, Beyotime) at room temperature for 10 min. The coverslips were then thoroughly washed with sterile water and immersed in 0.5% glutaraldehyde (G849973, Macklin) for 30 min, followed by another extensive wash in sterile water and air drying for later use. Glass slides were immersed in dimethyldichlorosilane (D806824, Macklin) for 5 min. PA hydrogels with soft and stiff stiffness were prepared by adjusting the relative concentrations of acrylamide and bis-acrylamide. Different concentrations of 40% acrylamide (A6299, Macklin) and 2% bis-acrylamide (M6024, Macklin), 1:100 (v/v) ammonium persulfate (AR1166, Boster), and 1:1,000 (v/v) TEMED (AR1165, Boster) were mixed for polymerization. The gel mixture was quickly pipetted onto the slides, and treated coverslips (treated side down) were carefully placed onto the gel droplets. The gels were allowed to polymerize for 1 h. After polymerization, the bottom glass slide was removed, and the top coverslip-gel composite was transferred to a tissue culture plate and washed with 50 mM HEPES (C0215, Beyotime) to remove unpolymerized gel. To promote cell adhesion, the surface of the hydrogel was functionalized by applying sulfosuccinimidyl-6-(4-azido-2-nitrophenylamino)-hexanoate (sulfo-SANPAH, ProteoChem) and irradiating under 365-nm UV light (10W) for 6 min. The gels were then washed with 50 mM HEPES buffer and incubated with 0.1 mg/mL rat-tail collagen I (354,236, Corning) solution at 4 °C overnight. Finally, the coverslips were rinsed twice with PBS and sterilized under UV light for 30 min before cell seeding.

This study was approved by the Ethics Committee of Shanghai Jiao Tong University Affiliated Sixth People’s Hospital. Patients’ privacy rights of human subjects have been observed and written informed consent was obtained from all patients. Normal and tumor breast tissue samples were collected from patients who underwent mastectomy at Shanghai Sixth People’s Hospital. Patients who had received chemotherapy or radiotherapy prior to surgery were excluded.

For animal studies, two distinct tumor models were established. In the first model, 5 × 10⁵ 4T1 cells in 100 μL PBS were injected into the left fourth mammary fat pad of randomly selected 6–8-week-old female BALB/c mice. Control mice were injected with PBS alone. Ten days later, when the primary tumor diameter reached 1 cm, tumor tissues in the 4T1 group and mammary fat pads in the PBS group were collected for further analysis. Parallelly, 3 × 10⁶ B16F10 melanoma cells in 25 μL PBS were implanted subcutaneously into the footpads of 6–8-week-old C57BL/6 J mice. Control mice were injected with PBS alone. Ten days later, tumor tissues in the B16F10 group and footpads in the PBS group were collected for further analysis. All animal experiments were approved by the Animal Care and Use Committee.

Cells seeded on hydrogels with different stiffness were fixed in 4% paraformaldehyde (PFA) for 15 min, permeabilized with 0.1% Triton X-100 for 10 min at room temperature, and blocked with 3% bovine serum albumin (BSA) for 1 h. Cells were then incubated overnight at 4 °C with primary antibodies against Ki67 (14–5,698, 1:300, Invitrogen), FAT1 (ab190242, 1:200, Abcam), YAP (sc-101199, 1:300, Santa Cruz), Vinculin (V9131, 1:300, Sigma), and β-catenin (ab32572, 1:200, Abcam), non-phospho (active) β-catenin (8,814, 1:200, Cell Signaling Technology). After washing, cells were incubated with Alexa Fluor 594- or 488-conjugated secondary antibodies (Abcam, 1:1,000) for 1 h at room temperature. F-actin was stained with DyLight 647 phalloidin (40762ES75, Yeasen), and nuclei were counterstained with DAPI (D9542, Sigma). Images were acquired using a confocal microscope (Nikon A1, Tokyo, Japan, six to ten sections were imaged on confocal microscope with the resulting Z-stacks of the outflow region merged into a maximum intensity projection within the NIS-elements software. Then the images were analyzed using ImageJ software. For cell morphology analysis, cells of each group were manually analyzed with ImageJ to determine the cell morphologies, which were characterized by cell area and circularity (4π(area/perimeter²)) (Zhang et al., 2015). For nuclear to cytoplasmic YAP ratio analysis, we followed previously reported literature protocols (La et al., 2022). In brief, nuclear outlines were delineated in the DAPI channel using the ROI manager tool, and both the total cellular YAP intensity and the nuclear integrated fluorescence intensity were measured per cell. The YAP expression in the cytoplasm was calculated as the difference between the total cellular YAP intensity and the nuclear YAP intensity. The YAP nuclear/cytoplasmic ratio was defined as the ratio of nuclear to cytoplasmic YAP fluorescence intensity. The area and number of focal adhesions were calculated based on the previously reported literature (Güler et al., 2021). The percentage of Ki67 positive cells was calculated by the number of Ki67 positive cells to the number of DAPI positive cells per image. The percentage of nuclear β-catenin-positive cells was calculated by counting nuclear β-catenin-stained cells and total cells (DAPI) in randomly selected microscope fields (Fonseca et al., 2015).

Tissues from patients or mice were fixed, embedded in paraffin, and sectioned at 5 μm thickness. Sections were dewaxed, rehydrated, subjected to antigen retrieval, permeabilized, and blocked. Sections were then incubated overnight at 4 °C with primary antibodies against podoplanin (ab10288, 1:100, Abcam; ab11936, 1:200, Abcam), FAT1 (ab190242, 1:100, Abcam). After washing, appropriate fluorescent secondary antibodies were applied for 1 h at room temperature. Finally, the sections were mounted with an antifade medium containing DAPI (Abcam) and imaged using a confocal microscope (Nikon, Japan), Images were acquired using a confocal microscope (Nikon A1, Tokyo, Japan), six to ten sections were imaged on confocal microscope with the resulting Z-stacks of the outflow region merged into a maximum intensity projection within the NIS-elements software. Images were analyzed using ImageJ software. Lymphatic vessel density (LVD) was defined as the ratio of total podoplanin-positive lymphatic vessel area to total tissue area (Burchill et al., 2021). Based on previously established methods (Bhakuni et al., 2024), the mean FAT1 fluorescence intensity was calculated by dividing the FAT1 intensity by the total podoplanin-positive area. Subsequently, these values were normalized to the control group and subjected to statistical analysis.

The small interference RNA (siRNA) constructs targeting mouse FAT1 expression were designed and synthesized by Sangon biotech company (Shanghai, China), and the transfection was performed with Ribo Transfection Reagent (GuangZhou, China) according to the manufacturer’s protocol. The siRNA sequences were listed as follows: Mouse FAT1 siRNA (5′-UACUUGGUGACUUCGGUCC-3′).

Cell viability was assessed using the EdU assay. The Click-iT EdU-488 Kit (G1601, Servicebio) was used according to the manufacturer’s instructions to detect cellular proliferation. EdU-stained cells were imaged under an inverted microscope (Olympus) equipped with a digital camera (Canon). Images were analyzed using ImageJ software.

LEC spheroids were made as previously described (Heiss et al., 2015). Briefly, 3 × 10³ LECs per well were suspended in the appropriate culture medium containing 0.375% (w/v) methylcellulose (M0512, Sigma) and seeded into nonadherent round-bottom 96-well plates (3,422, Corning). Spheroids were cultured overnight. Following formation, the spheroids were transferred onto collagen-coated PA gel plates and allowed to attach. Subsequently, 300 μL of 1 mg/mL collagen I (354,249, Corning) was overlaid to cover the cells on each coverslip. Spheroids were visualized using an inverted microscope (Olympus) equipped with a digital camera (Canon).

Inserts (Ibidi Culture-Insert, Germany) were placed in 24-well plates coated with PA hydrogels of varying stiffness as previously reported (Wang Z. et al., 2025). Cell suspensions (5 × 10⁴ cells in 70 μL of culture medium) were added to both sides of each insert. Once the cells reached confluence, the inserts were removed, and the wells were washed twice with PBS to remove cellular debris. Cells were then cultured in medium containing 0.1% FBS. Monolayer wound closure was imaged at the indicated time points using an inverted microscope (Olympus) equipped with a digital camera (Canon). Scratch areas were quantified using ImageJ software. Wound closure rates were calculated as follows: (initial wound area–wound area at the indicated time) divided by the initial wound area.

SVEC4-10 cells were seeded at a density of 5 × 10⁴ cells per well onto soft (1.5 kPa) or stiff (10 kPa) hydrogels in 24-well plates (n = 3 biological replicates per condition). After 48 h, cells were harvested and lysed using Trizol. Total RNA was extracted and sequenced by Xu Ran Biological (Shanghai, China). Differentially expressed genes (DEGs) were identified based on a fold change >1.3 and p value <0.05.

Total RNA was extracted from cells cultured on PA gels with different stiffness using Trizol reagent (Takara, Dalian, China). RNA was reverse transcribed into complementary DNA using a cDNA Synthesis Kit (Takara, Dalian, China). Quantitative real-time PCR was performed in triplicate using a SYBR Green PCR Kit (Takara, Dalian, China) on an ABI 7500 system. Glyceraldehyde-3-phosphate dehydrogenase (GAPDH) served as the internal control. The primer sequences used are listed in Supplementary Table S1.

Cells were washed with cold PBS and lysed in RIPA lysis buffer. Then proteins were separated via sodium dodecyl sulfate-polyacrylamidegel electrophoresis (7.5%) gels and transferred to polyvinylidene difluoride membranes. The primary antibodies used in the study were as follows: anti-FAT1 (ab190242, 1:1,000, Abcam), We used β-Tubulin (10068-1-AP, 1:2000, Proteintech) as internal controls. The membranes were incubated with anti-mouse (M21001S, Abmart, Shanghai, China) or anti-rabbit (M21002S, Abmart, Shanghai, China) secondary antibodies for 1 h. Immunoblotting was visualized using an enhanced chemiluminescence solution (Pierce, Pierce, MO, United States), and the ImageQuant LAS 4000 mini (General Electric Company, Boston, MA, United States) was used to detect protein expression.

All statistical analyses were performed using GraphPad Prism software (GraphPad Software Inc., San Diego, CA, United States). Statistical significance between two groups was determined using Student’s t-test and one-way analysis of variance (ANOVA) was used for multiple comparations. A p-value <0.05 was considered statistically significant. At least three independent experiments were performed for each assay.

In accordance with previous reports demonstrating that solid tumors display markedly increased matrix stiffness relative to normal tissues (Bordeleau et al., 2016), we used a tunable PA hydrogel system to examine the effects of tumor-associated ECM stiffness on LEC biology. As shown in Figure 1A, the PA hydrogel model that offers excellent biocompatibility and mechanical stability, is an ideal platform for matrix stiffness studies (Caliari and Burdick, 2016). Following established protocols, we generated hydrogels with precisely controlled stiffness by varying the concentrations of acrylamide and bis-acrylamide crosslinker. Based on previous reports (Filipe et al., 2024), we defined 1,500 Pa as representing normal tissue stiffness and 10,000 Pa as mimicking the stiffer tumor microenvironment, thereby creating a biologically relevant system to investigate LECs response to pathological matrix stiffness. As shown in Figures 1B–D and Supplementary Figure S1A-C, we observed that both SVEC 4–10 cells (mLECs) and human primary lymphatic endothelial cells (hLECs) exhibited distinct stiffness-dependent morphological alterations, including progressively increased cell area and significantly reduced cell circularity. These morphological changes reflected a complete phenotypic transition from the characteristic cobblestone morphology observed under physiological stiffness to an elongated, spindle-shaped appearance under tumor-like stiffness conditions. To confirm that the observed morphological alterations were specifically mediated by matrix stiffness-induced mechanotransduction in LECs, we examined the subcellular localization of Yes-associated protein (YAP), a canonical mechanosensitive transcriptional regulator (Dupont et al., 2011). As demonstrated in Figures 1E–G and Supplementary Figure S1D, E, tumor-associated matrix stiffness promoted nuclear translocation of YAP in both mLECs and hLECs, further indicating that tumor-associated ECM stiffness can functionally regulate LEC phenotype.

To further investigate the mechanoregulatory effects of matrix stiffening on LECs function, we initially evaluated LECs proliferative capacity using Ki67 immunostaining and EdU incorporation assays. Quantitative analysis demonstrated a significant increase in the proportion of Ki67-positive and EdU-positive mLECs under high-stiffness conditions (Figures 2A–D). A similar trend of enhanced proliferation was observed in hLECs cultured on stiff substrates (Supplementary Figure S1F, G), collectively indicating that tumor-mimetic matrix stiffness promotes the proliferation of LECs.

Subsequently, a wound healing assay was conducted to evaluate the effect of tumor-associated matrix stiffness on LEC migration. The results demonstrated that high-stiffness conditions significantly enhanced the migratory potential of mLECs (Figures 2E,F). This 2D observation was further validated using a 2.5D spheroid outgrowth model, where mLECs exhibited significantly increased spreading areas on tumor-mimetic stiffness versus physiological stiffness (Figures 2G,H). These consistent findings indicate that tumor-associated matrix stiffening serves as a potent biophysical cue that enhances LEC migratory capacity through mechanotransduction pathways.

To elucidate the specific mechanisms by which elevated matrix stiffness affects the biological function of LECs, we conducted comparative RNA-seq analysis of mLECs cultured on physiologically soft (1.5 kPa) versus tumor-mimetic stiff (10 kPa) hydrogels to delineate stiffness-dependent transcriptional alterations. As shown in Figure 3A, differential expression analysis (fold change >1.3, p < 0.05) revealed that mLECs exposed to tumor-mimetic matrix stiffness exhibited 441 upregulated genes and 607 downregulated genes compared to the physiological stiffness group. Functional annotation analysis demonstrated that significant enrichment of these DEG in pathways related to positive regulation of cell migration and vasculature development (Figure 3B). Cluster heatmaps in Figures 3C,D illustrate the expression patterns of representative genes involved in these pathways.

Based on established literature (Hultgren et al., 2020; Zuo et al., 2023; Nukala et al., 2023; Li et al., 2023; Inagaki et al., 2014; Fahmy et al., 2003), we identified six candidate genes involved in endothelial cell proliferation and migration for quantitative PCR (qPCR) validation (Figure 3E). Of particular significance, FAT1, a prototypical cadherin superfamily member, exhibited the most pronounced transcriptional downregulation in mLECs grown on tumor-mimetic stiff substrates (10 kPa) relative to physiological soft matrices (1.5 kPa), with subsequent confirmation at the protein level in both mLECs and hLECs (Figures 3F,G; Supplementary Figure S1H, I).

Interestingly, we similarly observed that tumor-mimetic stiffness modulated the phenotype and function of blood endothelial cells (BECs); however, immunofluorescent staining detected no significant alteration in FAT1 expression in BECs cultured on substrates of different stiffnesses (Supplementary Figure S2A-K). Furthermore, analysis of a public dataset demonstrated that substrate stiffness did not significantly affect FAT1 expression in vascular endothelial cells (Supplementary Figure S2L). Together, these findings suggest that FAT1 may function as a lymphatic-specific gene regulated by substrate stiffness.

To further validate FAT1’s role as a mechanosensitive regulator in LEC function, we achieved efficient FAT1 knockdown in SVEC4-10 cells using small interfering RNA (siRNA), with significant reduction at both the mRNA and protein levels (Figures 4A–C). FAT1-silenced and negative control (NC) mLECs were subsequently cultured on hydrogels mimicking physiological or tumor-associated stiffness. Immunofluorescence analysis of YAP subcellular localization (Figures 4D,E) demonstrated that FAT1 knockdown significantly increased the nuclear-to-cytoplasmic ratio of YAP in mLECs grown on both soft and stiff matrices compared to NC groups, suggesting that FAT1 loss potentiates YAP-mediated mechanotransduction pathways.

Our further investigations revealed that FAT1 knockdown significantly increased the proliferative activity of mLECs grown on matrices of physiological stiffness (Figures 4F,G), though no synergistic effect was observed between FAT1 silencing and matrix stiffening in promoting cell proliferation.

We next evaluated the consequences of FAT1 knockdown on mLEC migratory behavior. 2D migration assays demonstrated consistent enhancement across different mechanical environments, with FAT1 knockdown mLECs showing improved motility on both matrices with physiological stiffness and tumor-mimetic stiffness (Figures 4H,I). This mechanosensitive phenotype was further confirmed using 2.5D spheroid outgrowth assays. FAT1 knockdown elevated the migratory ability of mLEC spheroids on soft matrices to levels comparable to those of the rigid matrix control group (Figures 4J,K), quantitative analysis of spheroid spreading areas confirmed the absence of additive effects between FAT1 silencing and matrix stiffening.

Together, these results establish FAT1 as a possible signaling molecule important to the regulation of LEC mechanotransduction, affecting both proliferative and migratory responses to extracellular matrix stiffness.

Given Wnt/β-catenin’s crucial role in proliferation, we investigated its involvement in FAT1-regulated LEC proliferation. Immunofluorescence analysis revealed that both total and non-phosphorylated (active) β-catenin shifted from membrane localization (physiological stiffness) to nuclear/perinuclear accumulation (tumor-associated stiffness) in mLECs (Figures 5A–D). Notably, FAT1 knockdown mimicked this translocation even at physiological stiffness, indicating FAT1 normally suppresses Wnt signaling in compliant environments. QPCR confirmed Wnt activation, showing increased expression of downstream targets Myc and Cyclin D1 after either matrix stiffening or FAT1 depletion (Figures 5E,F), further demonstrating FAT1 mechanosensitively regulates LEC proliferation via Wnt/β-catenin signaling.

As vinculin serves as the central mechanotransducer coordinating migration through focal adhesion (FA) dynamics (Holland et al., 2024), we examined FAT1’s regulation of LEC migration via vinculin. Immunofluorescence analysis (Figure 5G) revealed that FAT1 knockdown increased both the number and area of FAs in mLECs on matrices with physiological stiffness, reaching levels comparable to those observed in the NC group cells grown on tumor-mimetic stiff matrices. Furthermore, under tumor-mimetic stiffness conditions, FAT1 silencing further enhanced focal adhesion formation, increasing both their quantity (Figure 5H) and size (Figure 5I). These findings suggest that FAT1 normally suppresses FA formation to modulate mechanical responsiveness.

Since cells sense extracellular mechanical cues by connecting the cytoskeleton to FAs (Wang et al., 2024), we further elucidated FAT1’s regulatory role in LEC cytoskeletal remodeling. Immunofluorescence analysis revealed a significant increase in filopodia-positive mLECs cultured on matrices with tumor-like stiffness compared to matrices with physiological stiffness (Figure 5J). Strikingly, genetic ablation of FAT1 recapitulated this pro-migratory cytoskeletal phenotype even in compliant microenvironments, suggesting that FAT1 suppresses LEC migratory potential through negative regulation of filopodia formation (Figure 5K).

Building upon our in vitro findings that tumor-mimetic extracellular matrix stiffness suppresses FAT1 expression in LECs, we next characterized the expression pattern of FAT1 in lymphatic vessels during tumor progression in vivo. In the murine 4T1 breast cancer model, immunofluorescence co-staining with the lymphatic-specific marker podoplanin (PDPN) and FAT1 revealed a significant reduction in FAT1 fluorescence intensity within tumor-associated lymphatic vessels compared to those in normal mammary tissue (Figures 6A,C). Moreover, we observed an increase in lymphatic vessel density (LVD) (Figure 6B). Subsequent correlation analysis revealed a significant negative correlation between LVD and relative FAT1 fluorescence intensity, suggesting that reduced FAT1 expression in tumor-associated lymphatic vessels is associated with higher LVD (Figure 6D).

Consistent with these preclinical findings, analysis of matched clinical specimens from human breast cancer patients indicated that a negative correlation between FAT1 expression and lymphatic vessel density was also observed in human tumor-associated lymphatic vessels, supporting the existence of this relationship in clinical samples (Figures 6E–H). To establish the broader relevance of this phenomenon, we extended our investigation to the B16F10 melanoma model, in which a similar negative correlation between FAT1 expression and lymphatic vessel density was also observed in tumor-associated lymphatic vessels and normal lymphatic vessels (Figures 6I–L).

Collectively, our results suggest that matrix mechanotransduction may serve as a key regulatory mechanism modulating tumor-associated lymphatic function through the regulation of FAT1 expression.