Capillary constrictions prime cancer cell tumorigenicity through PIEZO1

Abstract

Metastasis is responsible for most cancer-related deaths. However, only a fraction of circulating cancer cells succeed in forming secondary tumours, indicating that adaptive mechanisms during circulation play a part in dissemination. Here, we report that constriction during microcapillary transit triggers reprogramming of melanoma cells to a tumorigenic cancer stem cell-like state. Using a microfluidic device mimicking physiological flow rates and gradual capillary narrowing, we show that compression through narrow channels causes cell and nuclear deformation, rapid chromatin remodelling and increased calcium signalling via mechanosensor PIEZO1. Within minutes, cells upregulate transcripts associated with metabolic reprogramming and metastatic processes. Over time, this results in the stable adoption of a cancer stem cell-like state. Squeezed cells express elevated melanoma stem cell markers, exhibit increased trans-endothelium invasion and display enhanced tumorigenicity in vitro and in vivo. Pharmacological inhibition of PIEZO1 blocks this transition, while activation with Yoda1 induces the stem cell-like state irrespective of constriction. Deletion of PIEZO1 completely abolishes the constriction-induced phenotype. Together, these findings demonstrate that compressive forces during circulation reprogram circulating cancer cells into tumorigenic, stem cell-like states, primed for extravasation and metastatic colonization.

Fig. 1. Compressive forces induce melanoma cell state changes in a model microcapillary device.

A Device schematic consisting of a series of parallel constrictive channels of 30 µm (i), 20 µm (ii), 10 µm (iii), 5 µm (iv) diameter. B Images of melanoma cells passing through the micro constrictions and relaxation chambers. (C) Quantification of viable cells in control (CTRL) and squeezed (SQZD) groups. The results are expressed as the mean ± SEM from three independent experiments. Statistical significance assessed with two-sided unpaired t-test: p  =  0.003 (***), 95% CI: [−38.52, −23.48]. D-F Plot of deformation index (DI) of melanoma cells transiting the microfluidic device, demonstration of inverse relationship between channel diameter and median deformation, and quantification of the % median deformation. n = 20 cells for each plotted condition. The observed trend was confirmed with 3x biological repeat experiments performed on different days. G Representative fluorescence images of nuclei (Hoechst) for CTRL and SQZD cells. H Graph of nuclear area, expressed as mean ± SEM from three independent experiments. Statistical significance assessed with two-sided unpaired t-test: p  =  0.0212 (*), 95% CI: [−56.12, −7.86]. I Graph of nuclear intensity (Hoechst), expressed as mean ± SEM from three independent experiments. Statistical significance assessed with two-sided unpaired t-test: p  =  0.0261(*), 95% CI: [6.688, 62.11]. L and M Immunofluorescence images and quantification of H3K9ac in melanoma cells, expressed as mean ± SEM from three independent experiments. Statistical significance assessed with two-sided unpaired t-test: p  =  0.00895 (ns), 95% CI: [−47.12, −5.131]. N and O Immunofluorescence images and quantification of H3K9me3 in melanoma cells, expressed as mean ± SEM from three independent experiments. Statistical significance assessed with two-sided unpaired t-test: p  =  0.0382 (**), 95% CI: [5.805, 126.2]. Source data are provided as a Source Data file.

Fig. 2. Transcriptomics and immunofluorescence after constriction indicate increased invasiveness and evidence for stem cell-like states.

A MDS of the two groups control (X1, X2, X3 CTRL) and squeezed (X4, X5, X3 SQZD) analysed in triplicates, where X and Y axes represent dimensions that capture the largest differences between samples based on their gene expression profiles. B Corresponding Venn diagram showing unique and overlapping genes between the two groups. C Volcano plot indicating significantly differentially expressed mRNAs between SQZD and CTRL. The grey horizontal line shows P-value cut-off (*p < 0.01) and the vertical dashed lines indicate up/down-regulated genes (<−1.5 and >1.5fold change). Statistical significance was determined using the edgeR exact test (two-sided), with Benjamini-Hochberg correction for multiple comparisons. D Heatmap of the differentially expressed genes. E Bar graph displaying the number of genes involved in the GO biological processes and KEGG pathways associated with metabolic reprogramming (orange), tumorigenicity and metastasis (pink), cell-matrix interactions (green) and cell cycle (purple). Ticks in the gene number x axis are every 5 genes. F Representative fluorescence images of melanoma cancer cells, seeded on glass before (CTRL) and 4 hours after been squeezed (SQZD) into the microfluidic device, for biomarkers relative to stemness characteristic, e.g., CD44, CD271, ABCB5, PRDM14. G Bar graph showing the fold change of fluorescence signal of stemness-related biomarkers for SQZD cells compared to CTRL cells. The results are expressed as the mean ± SEM from five independent experiments. Source data are provided as a Source Data file.

Fig. 3. Microcapillary-like constrictions enhance melanoma cell tumorigenicity.

A Representative images of control (CTRL) and squeezed (SQZD) melanoma cells forming a tumour sphere in serum-free media after 7 days of culture. B and C Graph representative of the mean tumour sphere area increasing over time for different conditions and the quantification of tumour sphere area at day 7. The result is expressed as the mean ± SEM from n = 4 CTRL and n = 3 SQZD independent experiments. Statistical significance was assessed using a two-sided unpaired t-test. Asterisks indicate p-values, p  =  0.0174 (*), 95% CI: [1060, 6975]. D IVIS Imaging of lung metastases that developed after 30 days from injecting CTRL and SQZD luciferase melanoma cells into the lateral tail vein of 7-week-old female Balb/c Nude mice. Representative H&E-stained lung sections from mice injected with CTRL cells (E) and SQZD cells (F) harvested 30 days post tail vein injection. These samples correspond to the mice with the highest luciferase signal in each group (n = 4 mice per group). Histological analysis was performed on these individual animals only. G Whole body luciferase signal of mice injected intra-cardially at week 4, the final timepoint at which all mice in both groups were accounted for (CTR, n = 8 mice; SQZ, n = 10 mice). Statistical significance was assessed using a two-sided unpaired t-test. Asterisks indicate p-values, p  = 0.0195 (*), 95% CI: [182243370, 5419316630]. Photon flux from individual metastatic sites revealed increased median signals in the H lung and significantly higher signals in (I) bone and L brain in the SQZ group compared to controls. Statistical significance was assessed using a two-sided unpaired t-test. Asterisks indicate p-values, Bone p  =  0.0031 (**), Brain p  =  0.0022 (**). Source data are provided as a Source Data file. Each dot represents an individual animal. M Metastatic burden and Kaplan–Meier survival curves following IC injection with control (grey) and squeezed (cyan) cells.

Fig. 4. PIEOZ1 activity directs stem cell-like states and tumorigenicity after microconstriction.

A Representative fluorescence images of PIEZO1 expression for CTRL and SQZD cells. B Quantification of PIEZO1 immunofluorescence intensity expressed as the mean ± SEM from five independent experiments. Statistical significance assessed using a two-sided unpaired t-test: p  = 0.0003 (***). p  =  0.0003(***), 95% CI: [8.521, 18.99]. C Representative fluorescence images showing melanoma cells as they squeeze through the constrictive channels with quantification of fold change intracellular calcium concentration using Calbryte 520 AM. The results are expressed as the mean ± SEM from 2 independent experiments. Statistical significance was assessed using one-way ANOVA (two-sided): 30 µm vs. 10 µm (CC) p  =  0.0185 (*) 95% CI: [−0.1084 to −0.01587]; 30 µm vs. 5 µm (CC) p = 0.0071 (**) 95% CI: [−0.1271 to −0.03463]. D Representative fluorescence images of cancer cells showing stemness biomarkers under CTRL condition and after treatment with 20 µM Yoda1 and 30 µM Ruthenium Red (RR). E Bar graph showing the fold change of fluorescence signal of stemness-related biomarkers for cells under different treatments. The result is expressed as the mean ± SEM from three independent experiments. Statistical significance was assessed using a two-sided unpaired t-test. Asterisks indicate p-values as follows: CD44 p  =  0.0044 (**) 95% CI: [−0.9418, −0.3318], PRDM14 p  =  0.0212 (*) 95% CI: [−1.228, −0.1722], CD271 p  =  0.0175 (*) 95% CI: [−2.537, −0.4282], ABCB5 p  =  0.048 (*) 95% CI: [−5.333, −0.02600]. F-H Images of CTRL and treated melanoma cells, forming tumorsphere in serum-free media after 7 days of culture, demonstration of increased mean tumorsphere area over time and quantification of tumorsphere area expressed as the mean ± SEM from four independent experiments, except for YODA (n = 3). Statistical significance was assessed using two-way ANOVA (two-sided). Asterisks indicate p-values as follows: CTRL vs. YODA p  =  0.0346 (*) 95% CI: [−2712, −114.6]; CTRL vs. RR p  =  0.0333 (*) 95% CI: [117.0, 2522]; YODA vs. RR p  =  0.0008 (***) 95% CI: [1434, 4032].

Fig. 5. Loss of PIEZO1 obviates emergence of stem cell-like phenotypes after microcapillary-induced deformation.

A Plot of deformation index (DI) for melanoma cells transiting the microfluidic device (n = 100–150 cells per condition). c B Cell viability expressed as the mean ± SEM from three independent experiments. Statistical significance assessed with two-sided unpaired t-test: p  = 0.0001 (***) 95% CI: [-42.01, -28.66]. C, D Immunofluorescence images of melanoma cells transiting constrictive channels and quantification of fold change intracellular calcium concentration (n = 7–40 cells per condition). The results are expressed as the mean ± SEM from 2 independent experiments. Statistical significance assessed using two-sided unpaired t-test: 10 µm p = 0.0038 (**) 95% CI: [0.03575 to 0.1028]; 5 µm p = 0.0459 (*) 95% CI: [0.0008904 to 0.06797]. E, F Immunofluorescence images and quantification of nuclei with results expressed as the mean ± SEM from three independent experiments. Statistical significance assessed using a two-sided unpaired t-test: p  =  0.97 (ns), 95% CI: [−26.46, 27.12. G Immunofluorescence images of stemness markers in melanoma cells and H quantification of fold change relative to control, expressed as the mean ± SEM from five independent experiments. Statistical significance assessed using two-sided unpaired t-tests: CD44 p  =  0.0079 (**), 95% CI: [−0.7189, −0.1491]; CD271 p  =  0.037 (*), 95% CI: [−1.081, −0.04346]; ABCB5 p  =  0.0135(*), 95% CI: [−1.709, −0.2568]. I, L,M Representative images of melanoma cells forming tumorsphere after 7 days of culture, demonstration of change in tumorsphere area over time, and quantification of tumorsphere area at day 7, expressed as the mean ± SEM from four independent experiments. Statistical significance assessed using two-sided unpaired t-test: WT SQZD vs. WT SQZD_RR p < 0.0001 (****), 95% CI: [2130, 093]; WT SQZD vs. KO SQZD_RR p < 0.0001 (****), 95% CI: [2051, 5013].

Fig. 6. Microconstrictions facilitate increased engagement of melanoma stem cell-like cells with endothelial cells.

A Representative image showing Crystal Violet dye signal from WT and KO melanoma cells migrating through the transwell membrane before and after been squeezed into the microfluidic chip. The image highlights the cells that have successfully migrated through the membrane pores and adhered to the bottom surface of the transwell insert. B Bar graph showing the fold change of WT and KO cell coverage area, compared to their relative controls. Results represent the mean ± SEM from 3 independent experiments. Statistical significance was assessed using two-sided unpaired t-tests. Asterisks indicate p-values, p = 0.0142 (*) 95% CI: [−1.682, −0.3339]. C A simplified schematic illustrating the co-culture of melanoma cancer cells with a functional monolayer of endothelial cells. D Bar graph displaying the fluorescence expression of VCAM protein from endothelial cells cultured under CTRL condition and with the presence of WT and KO cells, before or after being squeezed into the microfluidic device. The results are expressed as the mean ± SEM from 3 (Endothelium, WT_CTRL, WT_SQZD) and n = 2 (KO_CTRL and KO_SQZD) independent experiments. Statistical significance was assessed using one-way ANOVA (two-sided). Asterisks indicate p-values, as follow, Endothelium vs. WT (SQZD) p = 0.0019 (**) 95% CI: [−7.288, −2.037]; WT (CTRL) vs. WT(SQZD) p = 0.0006 (***) 95% CI: [−8.149, −2.898]; WT (SQZD) vs. KO (CTRL) p = 0.036 (*) 95% CI: [0.2070, 6.077]; WT (SQZD) vs. KO (SQZD) p = 0.0012 (**) 95% CI: [−2.628, 8.499]. Source data are provided as a Source Data file. E Representative fluorescence images of vascular endothelial proteins, e.g., Vascular endothelial cadherin junction (VE-Cadherin) and Vascular Cell Adhesion Protein (VCAM), when the endothelial monolayer is co-cultured with melanoma cells. Experimental numbers and associated statistical analysis for the full data set are shown in panel D. Asterisks are reported in monolayer’s area, where VE-Cadherin remodel and leave gap in between cells.

Fig. 7. Microconstriction primes melanoma cells to compromise endothelial barrier function.

A Schematic illustrating the microfluidic chip used for melanoma extravasation assay and composed of two independent vascular channels and a central tissue chamber, communicating with a membrane of pores. After a dynamic culture protocol of 3 days to obtain the functional biological barrier into the vascular channel, melanoma cells are injected and allowed to extravasate into the tissue chamber overnight. B Representative fluorescence images of the mature blood vessel. Inset showing the expression of VE-Cadherin at the cell-cell border, indicating a functional biological barrier. Images are representative of at least three independent experiments with similar results. C Representative fluorescence images of WT melanoma cells (red) after 24 hours of residence in the blood vessel under dynamic conditions. These images illustrate the heightened invasiveness of SQZD WT cells compared to CTRL cells, with most SQZD WT cells having penetrated through the membrane pores into the tissue chamber. D Bar graph showing permeability coefficient comparison for CTRL (cell-free device), Vessel, WT-CTRL (Vessel + WT melanoma cells non-treated), WT-SQZD (Vessel + WT squeezed melanoma cells), KO-SQZD (Vessel + KO squeezed melanoma cells). Results are mean ± SEM from 3 independent experiments, except for vessel and KO-SQZD (n = 2). Statistical significance was assessed using one-way ANOVA (two-sided). Asterisks indicate p-values as follow, CTRL Vs Vessel p = 0.0016 (**) 95% CI: [6.863e−006, 2.508e−005]; Vessel Vs WT_SQZD p = 0.0169 (*) 95% CI: [−2.024e−005, −2.030e−006]. Source data are provided as a Source Data file.