A2058 (CRL-11147) melanoma cells obtained from the American Type Culture Collection were cultured in Dulbecco’s modified Eagle’s medium (Wisent; 319–007 CL) supplemented 10% v/v with fetal bovine serum (Wisent; 090–150-FBS) in 5% CO2 according to the supplied guidelines. Mycoplasma contamination was tested biweekly.

Advanced Biomatrix CytoSoft® 6-well Plates– discovery kit #5190: Eppendorf 6-well cell culture plate (Eppendorf 0030.720.113) with a 0.5 mm layer of activated biocompatible silicone of defined elastic modulus were used for all live-cell imaging. Advanced Biomatrix CytoSoft® Imaging 24-well Plate 0.2 kPa (#5183) and 64 kPa (#5189): Eppendorf 24 well cell imaging plate (0030.741.021) #1.5 glass bottom with a ~0.03 mm layer of activated biocompatible silicone of defined elastic modulus were used for fixed cell imaging. Silicone surfaces were coated with a Fibronectin solution (bovine plasma, Sigma; F1141) at a final concentration of 10 µg/ml in DPBS (Wisent; 311–425 CL) for 1 h at room temperature. The fibronectin solution was removed, and the plates maintained in DPBS until the cells were added.

siRNA-mediated knockdown was performed as previously described [17] using Dharmafect1 (Horizon Discovery Ltd; T-2001-03) and the following siRNA duplexes: FMNL2 siRNA Duplex1 (IDT; hs.Ri.FMNL2.13.1); FMNL2 siRNA duplex2 (IDT; hs.Ri.FMNL2.13.2).

Live cell imaging was performed on an Incucyte® S3 Live-Cell Analysis System (Sartorius). Full technical specifications can be found here: https://www.sartorius.com/download/930502/incucyte-s3-technical-specification-sheet-8000-0527-c00-en%96s-1%96data.pdf. High-definition phase-contrast images were acquired using the 10X/NA 0.3 objective with an image resolution of 1.2 µm/pixel. 72 h after transfection with control or FMNL2 siRNA duplex, A2058 cells were seeded at a density of 60,000 cells per well in CytoSoft® 6-well Plates (Advanced Biomatrix – discovery kit #5190) with the indicated moduli ranging from 0.2 kPa to 64 kPa. Remaining cells were transferred to fresh 3.5 cm dishes and incubated at 37 °C, 5% CO2. Cells were imaged using the Incucyte® S3 Live-Cell Analysis System at 37 °C in a 5% CO2 incubator. Phase contrast images were collected every 20 min for 24 h, the first time point at 2.5 h post seeding. Cells were then fixed for 10 min with freshly prepared 4% paraformaldehyde in PHEM (PIPES, Hepes, EGTA, MgCl2) for additional analysis [35]. The parallel cell samples were harvested and boiled in 1X Laemmli buffer to assess knockdown efficiency by immunoblotting.

High resolution fixed cell images were acquired using a Zeiss Axio Observer 7 inverted microscope with linear encoded stage and HXP 120 V light source with built in power supply, shutter, lamp module and infra-red filter. Zeiss filter set 37 Ex. BP 450/50 FT:480 Em. BP 510/50, 63 × 1.4NA oil immersion Plan-apochromat objective, detection with a Hamamatsu ORCA-Flash LT 16bit camera. Z-stacks captured with Zeiss Zen 3.0 software. 55 h after transfection with control or FMNL2 siRNA duplex, A2058 cells were seeded at a density of 2, 250 cells per well in CytoSoft® Imaging 24-well plates 0.2 kPa and 64 kPa moduli and incubated at 37 °C, 5% CO2. Remaining cells were transferred to fresh 3.5 cm dishes and incubated at 37 °C, 5% CO2. 72 h post transfection, cells were fixed for 10 min with freshly prepared 4% paraformaldehyde in PHEM (PIPES, Hepes, EGTA, MgCl2) [35]. The parallel cell samples were harvested in 1X Laemmli buffer and assessed for knockdown efficiency by immunoblotting.

Immunofluorescence was performed as in [17]. Briefly, cells fixed in 4% formaldehyde/PHEM buffer were permeabilized and blocked for 20 min in 0.3% Triton X-100, 5% donkey serum in 1 × PBS, washed in 1 × PBS, and incubated with Alexa Fluor 488 Phalloidin (Molecular Probes; A12379) diluted 1:200 in 0.03% Triton X-100 and 5% donkey serum in 1 × PBS for 1 h at room temperature. Washed and stored in 1X PBS.

siRNA knockdown efficiency was assessed by immunoblotting. Cells were washed with 1XPBS and harvested in 1x Laemmli buffer. Lysates were subjected to SDS-PAGE and immunoblotted with the indicated antibodies. Chemiluminescence was used for detection using the Immobilon® Crescendo western HRP substrate reagent (Millipore Sigma). FMNL2 was detected using chicken anti-FMNL2 [18] and Peroxidase AffiniPure™ Donkey Anti-Chicken IgY (IgG) (H + L) (Jackson ImmunoResearch Laboratories); tubulin with mouse anti-α-tubulin (Sigma; T5168) and Peroxidase AffiniPure™ Donkey Anti-Mouse IgG (H + L) (Jackson ImmunoResearch Laboratories).

All experimental measurements were recorded, and the mean and standard deviation (SD) of these values were calculated. The uncertainty associated with each mean value was calculated using the standard error of the mean (SEM). For calculating the uncertainty on measurements using the calculated mean values, propagation of error was used to determine their associated error. To determine statistical significance between two measured values, Analysis of Variance (ANOVA) testing was performed to discern any statistically significant differences between the groups. Tukey’s post hoc analysis was then performed to identify significant results between the mean values of different groups. All statistical analyses were conducted with a pre-established alpha level of 0.05, denoting the threshold for statistical significance.

Using the Python package OpenCV, the tiff stacks were first thresholded, a process where the pixels are binarized based on their intensity, with pixels having a value less/greater than the defined threshold assigned to white (0)/black (1). The tiff stacks were subsequently filtered using a Gaussian blur to reduce noise in the image. To quantify the morphology of cells, the cells were first located using the OpenCV function “findContours”, which identifies the boundary of objects in binarized images by looking for sharp increases or decreases in adjacent pixel values. Following this, the function “fitEllipse” was then used to fit an ellipse to the contours identified in the previous step. Using the contours, a mask was then created by converting all the pixels populating the inside of the contour into a binary image. The major and minor axis of the ellipse were then extracted and exported to a csv file, as well as the area, perimeter, and spatial coordinates of the boundary of the mask. With these values, we calculated the roundness of each cell (ri) as the ratio of the minor to major axis length of the fitted ellipse. We then averaged over all N cells on the substrate to obtain the group average roundness,

$$R = \langle \rangle = \frac\mathop \sum \limits_^N $$

To calculate the Feret diameter, a custom Python script calculated the distance between all points around the boundary of the mask, and then extracted the maximum value.

To track the cells, they were located using OpenCV’s contour finding function, and then tracked for the entire tiff stack with the Discriminative Correlation Filter with Channel and Spatial Reliability (CSRT) tracker in OpenCV. This object tracking algorithm works by applying discriminative correlation filters to different feature channels of the image (color, texture, etc) to determine their reliability. Each channel is weighted independently of the others based on its assigned reliability, which the CSRT tracker assesses by measuring each channel’s consistency in response over time, focusing on signal-to-noise ratio (SNR) to emphasize stable features and suppress noise. The tracker also learns to discriminate between the object and its background to enhance accuracy when the background may contain distracting elements. At each time point, the position of each cell along the x and y axes were then recorded and exported as a csv file and analyzed to determine all motility measurements once tracking was complete. The net distance travelled by each cell was then calculated by summing the distance travelled between each time step Δt from an initial time t0 up to the total time T:

$$D = \mathop \sum \limits_}^T = }\sqrt \right) - x\left( t \right)} \right)}^2} + \right) - y\left( t \right)} \right)}^2}} $$

The speed of each cell between each time step was also calculated to determine if the cell was moving. If the speed of the cell was <5 µm/h, then it was classified as not moving, as speeds less than this were often due to morphological changes in the cell that changed its center of mass, and not true movement. The time spent stationary and the time spent moving were then calculated for each cell using this restriction. These were averaged over the cell population to determine an ensemble-averaged moving time tm. Likewise, the ensemble-averaged speed sm was determined by averaging the mean cell speed during a cell trajectory over the cell population.

For actin fiber alignment analysis, images were first preprocessed in ImageJ by enhancing the contrast of the image by 0.35%. Subsequently, the stack was then exported as a maximum projection intensity image. A binary mask was also created and exported by thresholding the image to separate the cell from the background. The analysis was then performed with a previously used method known as Alignment by Fourier Transform [36] with some modifications to their python scripts. This method segments the windows of a defined size, performs a fast Fourier transform on each window in the image, and then outputs a vector field representing the alignment of F-actin fibers in a cell. The orientation angle θ for each fibre vector with respect to a central reference vector oriented along the major axis of the cell is then calculated and used to obtain a local F-actin orientational order parameter in a given cell [37]

This value characterizes the orientation of a given F-actin fibre with respect to the major axis of the cell. These local fibre order parameters are then averaged over all fibres in their cell to obtain the orientational order parameter of the entire cell: \( = \langle }\rangle.\) We can then average over all the cells on a given substrate to obtain the global orientational order parameter: \(S = \langle \rangle. \)