Image processing and analysis

ImageJ was used for initial image processing. Images were separated into appropriate color channels. The actin channel (543 nm) was used to create a mask outlining the shape of each cell. Actin images from each cell were first stacked together. A Gaussian blur with a 2-pixel radius was then applied to the stack, followed by

thresholding by eye to include the cells but not noise. Cells other than the cell of interest were cropped out of each frame. Any unfilled holes in the outline of each cell were manually filled in. Figure 5.5a shows resulting masks for several sample cells. Matlab was used for all subsequent analyses, with the custom script provided in the

Supplementary Information.

The Matlab built-in object recognition command regionprops was used to analyze circularity of cells. Cell masks were read by Matlab and analyzed for area and perimeter. For each image, only the maximum area and perimeters were analyzed in order to exclude small pixels of noise. Circularity C was calculated as

(5.1)

where A is the cell area and P is the cell perimeter. Circularity is 1 for a perfect circle and increasingly less than 1 for increasingly less circular objects.

87

To estimate the relative percentages of the cells inhabited by actin filaments, the actin channel images were masked in Matlab using the masks described above so as to only consider the area inside the cell. To avoid any intensity saturation, the 95th

percentile of intensity was set as an intensity ceiling for each image, and the number of pixels greater than 55% of this ceiling were counted as actin-containing pixels. This method detected different intensity distributions based on normalizing each cell to its own maximum intensity, so that differences in dye uptake had no effect. 55% was chosen based on visual confirmation that thresholding at this level consistently highlighted actin filaments and minimized noise selection. Figure 5.5a shows some examples of thresholded images. The number of pixels exceeding the threshold set for each cell was normalized to cell area and then expressed as a percentage.

Actinin thresholding was performed similarly to actin thresholding, using 50% rather than 55% of the 95th percentile of intensity based on trial-and-error. Positions of actin and actinin pixels exceeding these thresholds were recorded, and the Matlab function knnsearch was used to find the nearest neighboring actin-containing pixel for each actinin-containing pixel.

5.3.8 Atomic force microscopy

AFM experiments used 100,000 cells cultured on 22x40 mm coverslips (Fisher Scientific, Pittsburgh, PA) coated with 10 µg/ml fibronectin for 30-40 minutes and placed in 60-mm Falcon dishes (Corning). This work used tipless cantilevers with nominal resonant frequencies of 10-20 kHz and nominal spring constants of 0.03-0.09 N/m (NanoAndMore, Lady’s Island, SC, #CSC38/tipless/No Al). Silica colloids were attached to the cantilevers using a micromanipulator attached to a long-distance optical

88

microscope (40X, Alessi REL-4100A, NJ). Silicon dioxide microspheres (C-SIO-5.0, diameter ~4.86 µm, Microspheres-Nanospheres, Corpuscular, NY) were glued to the cantilevers using two-part epoxy glue (JB Weld, Sulphur Springs, TX) with the help of a sharp tungsten wire (TGW0325, World Precision Instruments, Sarasota, FL). The glue was allowed to cure overnight at room temperature. The cantilevers were cleaned in ethanol (Sigma) and were UV-ozone treated (UVO Cleaner model 42, Jelight Co. Inc., Irvine, CA, USA) for at least 15 minutes before use. The colloid radius was measured optically using the 100x oil immersion lens on the Nikon microscope (described below). A sample image of the colloid is shown in the inset in Figure 5.7a.

An Asylum MFP-3D AFM conjugated to a Nikon Total Internal Reflectance Fluorescence (TIRF) microscope was used in force spectroscopy mode to probe cellular mechanical properties. Colloid-attached cantilevers were calibrated in DI water prior to each experimental session using the thermal-noise method (Hutter and Bechhoefer, 1993). The measured spring constants used for force spectroscopy ranged 0.06-0.21 N/m. Deflection sensitivity of the cantilevers was measured from the slope of the force- distance (FD) curves obtained on a clean glass slide in recording HBSS before the start of the experiment. FD curves on cells were acquired at the perinuclear region of the cell with an approach rate of 500 nm/s. The perinuclear region of the cell was chosen so as to avoid both substrate effects, which can be substantial at the cell periphery (Mahaffy et al., 2004), and mechanical contributions from the nucleus, which may differ from the cytoplasm in its mechanical properties (Haga et al., 2000; Kuznetsova et al., 2007). Repeated cell indentation was performed every few seconds, which is likely infrequent enough to allow for relaxation in between indentations (Caporizzo et al., 2015).

89

Cells were kept in an incubator with 5% CO2 at 37˚C until experiments, which were performed at room temperature in HBSS solution. FD curves were taken from 6-10 cells per dish, and dishes were discarded 30-60 minutes after experimentation began. A representative FD curve measured on a cell is shown in Figure 5.7a. Asylum research software AR 12 version was used to fit the Hertz model to the approach curves. The Hertz model relates the force on the cantilever (F) to the depth of indentation in the sample () by

(5.2)

where R is the tip radius and E* is the combined elastic modulus of the contact and is given by

(5.3)

where υ is the Poisson ratio, which is 0.17 for the fused silica tip and assumed to be 0.37 for cells (Shin and Athanasiou, 1999). Combining equations 1 and 2 allows for the determination of Ecell, the cell elastic modulus. In this work, we extracted Ecell from the data spanning 25% to 75% of the maximum force used. The amount of indentation this corresponded to depended on Ecell, and usually was several hundred nm, which is considerable smaller than the thickness of the cell (~2-3 µm) in the perinuclear region. The depth of indentation was small in comparison to the radius of the colloid (δ/R <0.1- 0.2). Force curves were repeatedly obtained for a given cell until 4-6 consecutive curves gave consistent measures of Ecell (with the extreme measures not differing by more than ~15%). The resulting averaged Ecell was used for data analysis. Each cell line had data pooled from 34-57 cells total, measured over two separate days. For linear regression, the input stiffness value for each cell was the mean elastic modulus, and the input gene expression value was the mean gene expression level for that particular gene in that cell

90

line. Regressions were also performed by plotting the geometric means of elastic modulus by cell line against the mean expression level of a given gene for that cell line.

5.3.9 Statistics

Matlab and R were both used for statistical analysis. Linear regressions were performed using the corrcoef command in Matlab or the lm command in R. R was also used for adjusting p-values for multiple comparisons (p.adjust) using the Benjamini- Hochberg method and linear mixed effects modeling (lm) with ANOVA to test for effects of experiment number. Analysis of variance (ANOVA) was performed using the anova1

command in Matlab. The p-values are generally accepted as significant when p<0.05.

5.4 RESULTS