Collagen 1

P3 spheroids were prepared 3 days respectively before inclusion by seeding of 10⁴ cells in neurobasal medium with 0.4% methylcellulose (Sigma) in a U-bottom 96 wells plate (Falcon). A solution of 1 mg/ml of collagen I (Fisher Scientific) was prepared in PBS with 7.2 mM NaOH. After 30 min of incubation on ice, spheroids were individually picked, washed in PBS and included in the collagen solution. After 45 min at 37 °C in a cell incubator, neurobasal medium with the different treatments is added. P3 spheroid invasion areas are measured after 24 h with FIJI software. The total area was delineated by two independent investigators as well as the central spheroid core. An invasive index was calculated by the ratio of the total spheroid surface and the spheroid core.

The 3D colonoids were harvested from Matrigel after 7 days of culture by addition of EDTA solution (0.5 mM; Alfa Aesar) on ice, then transferred in 15 mL tubes and centrifuged (100× g, 4°C, 5 min). The organoid pellet was incubated in 1 mL TrypLE Express (Gibco) for 10 min while shaking at 37°C in a water bath. The centrifuged (100×g, 4°C, 5 min) organoid fragments were resuspended in complete medium [11 (link)] and further dissociated by repeated pipetting and subsequent filtering of the cell suspension through a cell strainer (cut-off size, 40 μm, Corning) to obtain a single-cell suspension. Transwell inserts (0.4 μm pores, Corning) were pre-coated with Matrigel (100 μg/mL; Corning) and collagen I (30 μg/mL; Fisher Scientific) in PBS or basal medium at 37°C for 1 h. Dissociated cells were counted manually using a cell counter (Hemocytometer; Hausser Scientific) and seeded at 10⁶ cells/mL in pre-coated Transwell inserts. After 3 days of incubation in a humidified incubator at 37°C with 5% CO₂, the cell monolayer was established. The morphology of a cell monolayer was intermittently monitored for up to two weeks by phase-contrast microscopy (Axiovert 40CFL, Zeiss).

Human HCC cell lines HepG2 and SMMC-7721 were cultured in DMEM and 1640 medium supplemented with 10% FBS, 100U/mL penicillin, and 100 mg/L streptomycin. To compare the effects of different ECM on the proliferation of hepatocellular carcinoma cells, 10⁵ HepG2 and SMMC-7721 cells were plated on 6-well culture plate pre-coated with 5 ug/cm² of either collagen I, collagen IV (BD Biosciences, Canada) or fibronectin (Thermo Fisher scientific, USA) for 24 h. After reaching confluence, protein was extracted from these cells for western blotting. In order to further analyze the effects of different concentrations of collagen I on the proliferation, cells were also plated on 6-well culture plate pre-coated with 25ug/cm² of collagen I or adding recombinant human epidermal growth factor (EGF, Peprotech, USA) to the medium at a concentration of 100 ng/mL.

Evaluation of CMVs migration was performed using a 24‐well transwell chamber of 3 µm pore size (Costar, USA). 5 µg DiI‐stained CMVs and CXCR4‐CMVs were mixed with 100 µL collagen I (Corning) and seeded into the upper chamber. After the formation of collagen I for 30 min, 100 ng mL⁻¹ CXCL12 (Peprotech, USA) dispersed in 300 µL PBS was placed into the lower chamber, while PBS without CXCL12 served as controls (n = 3). Then the chambers were incubated for 1 h at 37 °C within a humidified atmosphere with 5% CO₂. PBS in the lower chamber was collected, mixed well, and placed into a glass bottom dish (Cellvis, USA) for laser scanning confocal microscopy analysis. The 3D images of CMVs distribution were observed. The counting of CMVs at every layer of the scanning field was analyzed via Image J‐2.0.0.

DCs were cultured as stated in motility assay and then added to fluidic channels (μ-Slide VI^0.4; Ibidi) with confluent bEnd.3 cell monolayers or pre-coated with collagen I (1 mg/ml, Life Technologies), and allowed to adhere for 10 min. Phase-contrast and fluorescence images were first captured in static condition. Fluidic shear stress was then applied by flowing CM at 0.2 dyn/cm² through the channels. Live cell imaging was immediately initiated and images acquired every 5 s for up to 10 min, at 10× magnification. Fluidic shear stress was then increased to 1 dyn/cm² and live cell imaging immediately initiated. Time-lapse images were consolidated into stacks and motility and path-length data obtained from 20 cells (Manual Tracking, ImageJ) yielding mean velocities and pathlengths (Chemotaxis and migration tool, v 0.2.0). The percentage of remaining adherent cells was calculated by dividing the number of cells per frame following shear stress, by the number of cells in the same frame in static condition. Representative track plots were created in Python 3.6.9 (https://www.python.org/) with pandas 1.1.5 and matplotlib 3.1.3.