Piezo1 Activation Improves NSCLC Liver Metastasis Immunotherapy by Overriding Matrix Stiffness-Mediated Bimodal PD-L1/CXCL10 Regulation

Abstract

Immunotherapy efficacy in NSCLC is significantly reduced upon liver metastasis due to profound alterations in the tumor microenvironment, characterized by the absence of cyclic mechanical stretch and increased extracellular matrix (ECM) stiffness. However, the mechanisms underlying the synergistic regulation of these two mechanical cues on the immunotherapy response in NSCLC cells remain poorly understood. In this study, it is demonstrated that both mechanical and biochemical activation of the mechanosensitive ion channel Piezo 1 induces nuclear translocation of YAP, thereby promoting an immunotherapy-responsive tumor immune microenvironment (TIME) through enhanced expression of PD-L1 and secretion of chemokine C-X-C ligand 10 (CXCL10, chemokine recruiting CD8⁺ T cells) in NSCLC cells. The mathematical modeling further reveals that cyclic stretch modulates the PD-L1/CXCL10 response to ECM stiffness, shifting from a bimodal to a unimodal distribution. In a murine model of liver metastasis, the combination of Piezo 1 agonist with anti-PD-1 therapy significantly improves the immunotherapy response, as evidenced by elevated PD-L1 levels and increased CD8⁺ T cell infiltration. These findings underscore the critical role of Piezo 1 in enhancing the immunotherapeutic response in NSCLC liver metastasis and highlight its potential as a therapeutic target.

Keywords: NSCLC liver metastasis; mechanomedicine; organ‐specific heterogeneity; tumor immune microenvironment; tumor physical microenvironment.

Figure 1

Both TIME and TPME variances correlate with poor immunotherapy response of NSCLC liver metastasis. a) Comparison of the efficacy of four periods of anti‐PD‐1 immunotherapy in PL and LM tumor sites in the same patient (n = 35). b) Comparison of tumor regression rates in PL and LM sites from the same patient at different therapy periods (n = 7). c) CT images of PL and LM sites in the same patient at baseline and in different periods of anti‐PD‐1 therapy. The red arrows indicate the location of the tumor tissue. d) The PD‐L1, CD8⁺ T cell, and CXCL10 immunohistochemical staining of PL and LM tumor tissue. The scale bars indicate 20 µm (enlarged image in the upper left corner) and 60 µm. e) PD‐L1 expression status in PL and LM sites. f) percentage, and g) absolute number of CD8⁺ T cell in PL and LM tumor tissues. h) The expression level of CXCL10 in PL and LM tumor tissues. I) The Young's modulus of patients’ tissues measured by AFM (N ≥ 7; 3 different sites for each sample, n ≥ 21). j) CT images of the inspiratory and expiratory phase, and respiratory strain range of bronchial at all levels. k) The schematic diagram of varying TIME and TPME in PL and LM sites. Data are classified as PD versus non‐PD and compared using the chi‐square test (a). Data are compared using the chi‐square test (e). Data are compared by a two‐tailed Student's t‐test (f‐i). In (f‐i), all data are shown as mean ± S.E.M. NSCLC, non‐small cell lung cancer; PL, primary lung cancer; LM, liver metastasis; CXCL10, chemokine C‐X‐C ligand 10; AFM, atomic force microscopy; PD, progressive disease; TIME, tumor immune microenvironment; TPME, tumor physical microenvironment.

Figure 2

Cyclic stretch increases PD‐L1 expression via YAP‐mediated Piezo 1 activation in NSCLC cells. a) The diagram of cells cultured on static and cyclically stretched PDMS (frequency: 0.25 Hz, strain: 15%). b) Quantification of the fluorescence intensity of Ca²⁺ probes after 24‐h specific treatment on H1299 cells (N = 3). c) The western blotting of H1299 cells with certain treatments. PD‐L1 protein levels were normalized to GAPDH (N = 3). d) The western blotting of H1299 cells on static and cyclical stretch condition with/without GsMTx4 treatment. PD‐L1 protein levels were normalized to GAPDH (N = 3). The immunofluorescent images of e) YAP, and g) FAK^Tyr397. The scale bars indicate 10 µm (images in the upper left corner) and 20 µm. Quantification of f) YAP nuclear/cytoplasmic (n/c) ratio and h) length of FAK^Tyr397 (N ≥ 5, n ≥ 35 cells). i) The western blotting of H1299 cells cultured on static and cyclically stretched PDMS with the Verteporfin and Y15 treatment. Phosphorylation of YAP^Ser127/FAK^Tyr397 = gray value ratio of YAP^Ser127/FAK^Tyr397 over YAP/FAK. PD‐L1 protein levels were normalized to GAPDH (N = 3). j) Illustration of how cyclic stretch influences PD‐L1 expression. k) Differential gene volcano plot of environmental information processing‐related pathways. Data from RNA‐seq of H1299 cells with treatment of cyclic stretch and static condition l) The expression levels of CXCL10 in H1299 cells as determined by ELISA after treatment (N ≥ 5). Data are compared by a two‐tailed Student's t‐test. In (f), (h), and (i), all data are shown as mean ± S.E.M. NSCLC, non‐small cell lung cancer; PDMS, polydimethylsiloxane; RFU, relative fluorescence units; CXCL10, chemokine C‐X‐C ligand 10; ELISA, enzyme‐linked immunosorbnent assay. Yoda 1, Piezo1‐specific agonist; GsMTx4, Piezo 1 ion channel‐specific inhibitor; Verteporfin, YAP transcriptional function inhibitor; Y15, phosphorylation inhibitor of FAK^Tyr397.

Figure 3

The regulation of PD‐L1 expression in NSCLC cells has an optimal matrix stiffness. a) The diagram of cells cultured on 3 kPa, 14 kPa, and 52 kPa hydrogels. b) The western blotting of H1299 cells cultured on 3 kPa, 14 kPa, and 52 kPa hydrogels. Phosphorylation of YAP^Ser127/FAK^Tyr397 = gray value ratio of YAP^Ser127/FAK^Tyr397 over YAP/FAK. PD‐L1 protein levels were normalized to GAPDH (N = 3). c) Linear analysis was performed based on FAK^Tyr397 and PD‐L1 gray values from all western blotting results. The immunofluorescent images of d) YAP, and e) FAK^Tyr397. The scale bars indicate 10 µm (enlarged image in the upper left corner) and 20 µm. Quantification of the f) YAP n/c ratio and g) FAK^Tyr397 (N ≥ 5, n ≥ 35 cells). h) The western blotting of H1299 cells on 14 kPa hydrogel with Verteporfin and Y15 treatment (N = 3). i) Linear analysis of the average length of FAK^Tyr397 per cell and YAP n/c ratio (blue line). The scaling function is constructed by a mathematical model (red dotted line). j) The top diagram shows the molecular clutch model of force transmission, and the bottom one shows the simulation results of the mathematical model and evolution of the YAP n/c ratio along with stiffness change. k) Illustration of how ECM stiffness influences PD‐L1 expression. l) The expression levels of CXCL10 in H1299 cells (N ≥ 5). Data are compared by a two‐tailed Student's t‐test. In (f), (g), and (l), all data are shown as mean ± S.E.M. NSCLC, non‐small cell lung cancer; CXCL10, chemokine C‐X‐C ligand 10; Verteporfin, YAP transcriptional function inhibitor; Y15, phosphorylation inhibitor of FAK^Tyr397.

Figure 4

Piezo1 activation dominates TPME‐related upregulation of PD‐L1 in NSCLC liver metastasis. a) The diagram of cells cultured on the stretching platform with tunable matrix stiffness. b) The western blotting of H1299 cells cultured on 14 kPa and 52 kPa hydrogels treated with Yoda 1 or cyclic stretch. *, The comparison between the control group and groups treated with Yoda 1 or cyclic stretch on 14 kPa (black) and 52 kPa (dark blue). Phosphorylation of YAP^Ser127/FAK^Tyr397 = gray value ratio of YAP^Ser127/FAK^Tyr397 over YAP/FAK. PD‐L1 protein levels were normalized to GAPDH (N = 3). The immunofluorescent images of c) YAP, and d) FAK^Tyr397. The scale bars indicate 10 µm (enlarged image in the upper left corner) and 20 µm. e) Quantification of the YAP n/c ratio (N ≥ 5, n ≥ 35 cells). f) Quantification of the length of FAK^Tyr397 in H1299 cells (N ≥ 5, n ≥ 35 cells). g) The YAP n/c ratio alteration of cells cultured on various stiffness hydrogels with and without cyclic stretch (cells on glass slides were treated with Yoda 1). h) Schematic illustration of mathematical model and prediction of adhesion length change in cells cultured on various stiffness matrices coupled or uncoupled with cyclic stretch. i) The expression levels of CXCL10 in the supernatant of H1299 cells (N ≥ 5). Data are compared by a two‐tailed Student's t‐test. In (b), (e), (f), and (i), all data are shown as mean ± S.E.M. NSCLC, non‐small cell lung cancer; Yoda 1, Piezo1‐specific agonist; CXCL10, chemokine C‐X‐C ligand 10.

Figure 5

Piezo1 activation could be a promising strategy for improving immunotherapy efficacy in lung cancer liver metastasis. a) Diagram depicts the construction of the murine model with schedules of drug treatment. b) Image of liver metastatic tumors. c) Quantification of the tumor volume (N = 5). d) Multiplex immunohistochemistry. The scale bars indicate 20 µm (enlarged image) and 60 µm. e) The active YAP immunohistochemistry staining. The scales indicate 20 µm (enlarged image in the upper left corner) and 60 µm. f) Quantification of the fluorescence intensity of each marker (each treatment group, N = 3; 3 different sites for each sample, n = 9). g) Quantification of the cells with active YAP (each treatment group, N = 3; 3 different sites for each sample, n = 9). Data are compared by a two‐tailed Student's t‐test. In (c), (f), and (g) data are shown as mean ± S.E.M. Yoda 1, Piezo1‐specific agonist.

Figure 6

The graphic abstract of this study. PL, primary lung cancer; LM, liver metastasis; CXCL10, chemokine C‐X‐C ligand 10; ECM, extracellular matrix; Yoda 1, Piezo1‐specific agonist.