Imagescope software v 12

Melanoma invasion through fibroblast-modified collagen was assayed using a protocol adapted from Timpson⁷⁴ . Briefly, equal numbers of HFF, UV-HFF, shMMP1-HFF and shMMP1-UV-HFF fibroblasts were mixed with collagen I, rat tail (Corning, 354236) and cultured in 35 mm culture dishes. Collagen discs were allowed to contract until they fit in a 24-well plate. Cell suspensions of Sk-mel-28 and A375 at 4 × 10⁴ cells/ml were plated on top of each collagen disc in duplicate for each fibroblast condition. Cells and collagen were cultured as normal for approximately 5 days. Collagen discs were then transferred to Falcon 3.0 µm high density PET membrane (Corning, 353092) in Falcon six-well Deep Well TC-treated Polystyrene Plates (Corning, 355467) to create an air/liquid interface to drive melanoma invasion into collagen. After 10 days constructs were fixed in 4% PFA, and embedded in paraffin and stained with H&E and stained for Fibronectin with FN1 antibody (F3648, Sigma Aldrich). For each construct, the number of cells invading into the collagen was counted in at least five different fields of view under light microscopy, by two-independent scorers. Leica SCN400 was used for whole slide imaging alongside ImageScope software v12.3 (Leica Microsystems).

After fixation, strips were stepwise dehydrated (Leica TP1020, Semi-enclosed benchtop tissue processor, Germany) and embedded in paraffin wax blocks. Strips were either embedded to get 1) axial cross-sections – such that the lumen was oriented perpendicular to the face of the wax block or 2) with the luminal side of the strip flush with the face of the wax block to achieve cross-sections radially through the wall thickness. The integrity of the strip after mechanical testing dictated which of these options was most feasible; specifically, if it was not possible to orient the samples such that axial cross-sections could be obtained, the samples were oriented to get radial cross-sections. Samples were sectioned at 7 μm using Feather C35 microtome blades. Samples were stained with Haematoxylin & Eosin (H&E), picrosirius red (PSR) and Verhoeff’s elastin (Leica ST5010, Autostainer XL, Germany). Brightfield imaging of all stains was performed on an Aperio CS2 microscope with ImageScope software V12.3 (Leica Biosystems Imaging, Inc., Vista, California). Polarised light microscopy (PLM) was performed on the PSR-stained samples using an Olympus BX41 microscope with Ocular V2.0 software (Teledyne Photometrics, Tuscon, Arizona). Two PLM images were taken for each slice, 45° to each other, with the exposure time kept consistent across all samples.